Social Stratification

FOURTH EDITION

Social Stratification
Class, Race, and Gender in
Sociological Perspective

EDITED BY

David B. Grusky
and

Katherine R. Weisshaar

New York London

First published 2014 by Westview Press

Published 2018 by Routledge

711 Third Avenue, New York, NY 10017, USA
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN

Routledge is an imprint of the Taylor & Francis Group, an informa business

Copyright © 2014 Taylor & Francis

All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or
by any electronic, mechanical, or other means, now known or hereafter invented, including
photocopying and recording, or in any information storage or retrieval system, without
permission in writing from the publishers.

Notice:
Product or corporate names may be trademarks or registered trademarks, and are used only for
identification and explanation without intent to infringe.

Every effort has been made to secure required permissions for all text, images, maps, and other art
reprinted in this volume.

Designed by Timm Bryson

Library of Congress Cataloging-in-Publication Data

Social stratification : class, race, and gender in sociological perspective / edited by David B. Grusky
in collaboration with Katherine R. Weisshaar. — Fourth edition.
pages cm
Includes bibliographical references and index.
ISBN 978-0-8133-4671-7 (pbk.) 1. Social stratification. 2. Equality. 3. Social status. 4. Social
classes. I. Grusky, David B. II. Weisshaar, Katherine R.
HM821.S65 2013
305—dc23
2013029844

ISBN 13: 978-0-8133-4671-7 (pbk)

For the 99 percent
 CONTENTS

Preface and Acknowledgments xix

Resources for the Study of Poverty and Inequality xxii
1. Introduction
David B. Grusky and Katherine R. Weisshaar 1
The Questions We Ask About Inequality
David B. Grusky and Szonja Szelényi 17
The Stories We Tell About Inequality

PART I: THE FUNCTIONS AND DYSFUNCTIONS

OF INEQUALITY

2. Kingsley Davis and Wilbert E. Moore 28

Some Principles of Stratification
3. Melvin M. Tumin 31
Some Principles of Stratification: A Critical Analysis
4. Claude S. Fischer, Michael Hout, Martín Sánchez Jankowski,
Samuel R. Lucas, Ann Swidler, and Kim Voss 39
Inequality by Design

PART II: INEQUALITY IN COMPARATIVE PERSPECTIVE

CROSS-SOCIETAL DIFFERENCES
5. David B. Grusky and Katherine R. Weisshaar 44
A Compressed History of Inequality
6. Gøsta Esping-Andersen and John Myles 52
The Welfare State and Redistribution
TRENDS IN ECONOMIC INEQUALITY
7. Anthony B. Atkinson, Thomas Piketty, and Emmanuel Saez 59
Top Incomes in the Long Run of History

vii
viii • Contents

8. Claudia Goldin and Lawrence F. Katz 73

The Race Between Education and Technology
9. Bruce Western and Jake Rosenfeld 80
Unions, Norms, and the Rise in U.S. Wage Inequality
10. Robert Frank 86
Why Is Income Inequality Growing?
11. Jacob S. Hacker and Paul Pierson 91
Winner-Take-All Politics: Public Policy, Political Organization,
and the Precipitous Rise of Top Incomes in the United States
12. Thomas A. DiPrete, Gregory M. Eirich, and Matthew Pittinsky 104
Leapfrogs and the Surge in Executive Pay
13. Yujia Liu and David B. Grusky 113
The Winners of the Third Industrial Revolution

PART III: THE STRUCTURE OF INEQUALITY

MARXIAN THEORIES OF CLASS

14. Karl Marx 127
Alienation and Social Classes
Classes in Capitalism and Pre-Capitalism
Ideology and Class
15. Ralf Dahrendorf 143
Class and Class Conflict in Industrial Society
16. Erik Olin Wright 149
A General Framework for the Analysis of Class Structure
17. Immanuel Wallerstein 162
Class Conflict in the Capitalist World Economy
WEBERIAN THEORIES OF CLASS
18. Max Weber 165
Class, Status, Party
Status Groups and Classes
Open and Closed Relationships
19. Anthony Giddens 183
The Class Structure of the Advanced Societies
20. Frank Parkin 193
Marxism and Class Theory: A Bourgeois Critique
 Contents • ix

21. Tak Wing Chan and John H. Goldthorpe 202

Is There a Status Order in Contemporary British Society?
DURKHEIMIAN THEORIES OF CLASS
22. Emile Durkheim 217
The Division of Labor in Society
23. Kim A. Weeden and David B. Grusky 222
The Changing Form of Inequality
CLASSIC GRADATIONALISM
24. Donald J. Treiman 233
Occupational Prestige in Comparative Perspective
25. John H. Goldthorpe and Keith Hope 237
Occupational Grading and Occupational Prestige
26. David L. Featherman and Robert M. Hauser 244
Prestige or Socioeconomic Scales in the Study of
Occupational Achievement?
27. Robert M. Hauser and John Robert Warren 246
Socioeconomic Indexes for Occupations: A Review,
Update, and Critique
THE NEW GRADATIONALISM?
28. Aage B. Sørensen 251
Foundations of a Rent-Based Class Analysis
29. Amartya K. Sen 260
From Income Inequality to Economic Inequality

PART IV: THE RULING CLASS, ELITES, AND THE UPPER CLASS

CLASSIC STATEMENTS
30. Gaetano Mosca 276
The Ruling Class
31. C. Wright Mills 282
The Power Elite
32. Anthony Giddens 292
Elites and Power
CONTEMPORARY STATEMENTS
33. G. William Domhoff 297
Who Rules America?
x • Contents

34. Alvin W. Gouldner 302

The Future of Intellectuals and the Rise of the New Class
35. David Brooks 310
Bobos in Paradise: The New Upper Class and How They Got There
36. Gil Eyal, Iván Szelényi, and Eleanor Townsley 316
Post-Communist Managerialism
37. Andrew G. Walder 322
China’s Evolving Oligarchy

PART V: POVERTY AND THE UNDERCLASS

THE EXPERIENCE OF POVERTY

38. Barbara Ehrenreich 330
Nickel and Dimed
39. Kathryn Edin, Timothy Nelson, and Joanna Miranda Reed 339
Low-Income Urban Fathers and the “Package Deal” of Family Life
POVERTY AND THE ECONOMY
40. William Julius Wilson 347
Being Poor, Black, and American
41. Sheldon Danziger, Koji Chavez, and Erin Cumberworth 357
Poverty and the Great Recession
THE EFFECTS OF POLITICS AND INSTITUTIONS
42. Lane Kenworthy 365
How Rich Countries Lift Up the Poor
43. Katherine S. Newman and Rourke L. O’Brien 369
Taxing the Poor: How Some States Make Poverty Worse
NEIGHBORHOODS AND SEGREGATION
44. Douglas S. Massey and Nancy A. Denton 376
American Apartheid: Segregation and the Making
of the Underclass
45. Robert Sampson 386
Legacies of Inequality
46. Stefanie DeLuca and James E. Rosenbaum 395
Does Changing Neighborhoods Change Lives?
47. Patrick Sharkey and Felix Elwert 403
The Legacy of Multigenerational Disadvantage
 Contents • xi

HOW IMPORTANT IS EARLY CHILDHOOD?

48. James J. Heckman 412
Skill Formation and the Economics of Investing in
Disadvantaged Children
49. Greg J. Duncan and Katherine Magnuson 417
The Long Reach of Early Childhood Poverty
50. Gary W. Evans, Jeanne Brooks-Gunn,
and Pamela Kato Klebanov 423
Stressing Out the Poor
INCARCERATION AND POVERTY
51. Bruce Western and Becky Pettit 431
Incarceration and Social Inequality

PART VI: WHO GETS AHEAD?

CLASS MOBILITY
52. David L. Featherman and Robert M. Hauser 443
A Refined Model of Occupational Mobility
53. Robert Erikson and John H. Goldthorpe 453
Trends in Class Mobility: The Post-War European Experience
54. Richard Breen 464
Social Mobility in Europe
55. Jan O. Jonsson, David B. Grusky, Matthew Di Carlo,
and Reinhard Pollak 480
It’s a Decent Bet That Our Children Will Be Professors Too
INCOME MOBILITY
56. Gary Solon 496
Intergenerational Income Mobility
57. John Ermisch, Markus Jäntti, Timothy Smeeding,
and James A. Wilson 501
Advantage in Comparative Perspective
CLASSIC MODELS OF STATUS ATTAINMENT
58. Peter M. Blau and Otis Dudley Duncan, with the
collaboration of Andrea Tyree 506
The Process of Stratification
59. Christopher Jencks, Marshall Smith, Henry Acland, Mary Jo Bane,
David Cohen, Herbert Gintis, Barbara Heyns, and Stephan Michelson 517
Inequality: A Reassessment of the Effect of Family and
Schooling in America
xii • Contents

EDUCATION AND REPRODUCTION

60. Richard Breen and John H. Goldthorpe 524
Explaining Educational Differentials
61. Sean F. Reardon 536
The Widening Academic Achievement Gap Between
the Rich and the Poor
62. Richard Breen, Ruud Luijkx, Walter Müller,
and Reinhard Pollak 551
Nonpersistent Inequality in Educational Attainment
63. Michelle Jackson 562
Determined to Succeed
64. Sigal Alon 569
Towards a Theory of Inequality in Higher Education
65. Florencia Torche 578
Does College Still Have Equalizing Effects?
66. Jennie E. Brand and Yu Xie 587
Who Benefits Most from College?
SOCIAL PSYCHOLOGICAL MODELS
67. William H. Sewell, Archibald O. Haller,
and Alejandro Portes 596
The Educational and Early Occupational Attainment Process
68. Jay MacLeod 608
Ain’t No Makin’ It: Leveled Aspirations in a
Low-Income Neighborhood
69. Stephen L. Morgan 621
A New Social Psychological Model of Educational Attainment
LABOR MARKETS
70. Michael J. Piore 629
The Dual Labor Market: Theory and Implications
71. Aage B. Sørensen and Arne L. Kalleberg 632
An Outline of a Theory of the Matching of Persons to Jobs
72. Arne L. Kalleberg 640
The Rise of Precarious Work
73. Jake Rosenfeld 645
Little Labor: How Union Decline Is Changing
the American Landscape
 Contents • xiii

SOCIAL CAPITAL, NETWORKS, AND ATTAINMENT

74. Mark S. Granovetter 653
The Strength of Weak Ties
75. Nan Lin 657
Social Networks and Status Attainment
76. Ronald S. Burt 659
Structural Holes
77. Roberto M. Fernandez and Isabel Fernandez-Mateo 663
Networks, Race, and Hiring
78. Paul DiMaggio and Filiz Garip 671
When Do Social Networks Increase Inequality?

PART VII: RACE AND ETHNICITY

CONSTRUCTING RACIAL CATEGORIES

79. Michael Omi and Howard Winant 682
Racial Formation in the United States
80. Aliya Saperstein and Andrew M. Penner 687
The Dynamics of Racial Fluidity and Inequality
CLASSIC MODES OF INCORPORATION
81. Edna Bonacich 696
A Theory of Ethnic Antagonism: The Split Labor Market
82. Alejandro Portes and Robert D. Manning 710
The Immigrant Enclave: Theory and Empirical Examples
83. Richard D. Alba and Victor Nee 721
Assimilation Theory for an Era of Unprecedented Diversity
NEW MODES OF INCORPORATION
84. Alejandro Portes and Min Zhou 729
The New Second Generation: Segmented Assimilation
and Its Variants
85. Tomás R. Jiménez 740
Why Replenishment Strengthens Racial and Ethnic Boundaries
DISCRIMINATION
86. Marianne Bertrand and Sendhil Mullainathan 747
Are Emily and Greg More Employable Than Lakisha and Jamal?
A Field Experiment on Labor Market Discrimination
xiv • Contents

87. Claude Steele 752

Stereotype Threat and African-American Student Achievement
88. Devah Pager 757
Marked: Race, Crime, and Finding Work in an
Era of Mass Incarceration
ARE RACIAL AND ETHNIC DISTINCTIONS DECLINING IN SIGNIFICANCE?
89. William Julius Wilson 765
The Declining Significance of Race: Blacks and
Changing American Institutions
The Declining Significance of Race: Revisited and Revised
90. Reanne Frank, Ilana Redstone Akresh, and Bo Lu 780
How Do Latino Immigrants Fit into the Racial Order?
91. Zhenchao Qian and Daniel T. Lichter 788
Are Recent Trends in Intermarriage Consistent with Assimilation Theory?

PART VIII: GENDER INEQUALITY

LABOR FORCE PARTICIPATION

92. Arlie Russell Hochschild 803
The Time Bind: When Work Becomes Home and
Home Becomes Work
93. Lisa Belkin 807
The Opt-Out Revolution
94. Christine Percheski 811
Opting Out?
DISCRIMINATION
95. Claudia Goldin and Cecilia Rouse 820
Orchestrating Impartiality: The Impact of “Blind” Auditions
on Female Musicians
96. Shelley J. Correll, Stephen Benard, and In Paik 831
Getting a Job: Is There a Motherhood Penalty?
97. András Tilcsik 842
Do Openly Gay Men Experience Employment Discrimination?
98. Barbara F. Reskin 849
Rethinking Employment Discrimination and Its Remedies
99. Trond Petersen 858
Discrimination: Conscious or Nonconscious?
 Contents • xv

SEX SEGREGATION
100. William T. Bielby 865
The Structure and Process of Sex Segregation
101. Jerry A. Jacobs 876
Revolving Doors: Sex Segregation and Women’s Careers
102. Barbara F. Reskin 881
Labor Markets as Queues: A Structural Approach to
Changing Occupational Sex Composition
103. Elizabeth H. Gorman and Julie A. Kmec 890
Glass Ceilings in Corporate Law Firms
104. Maria Charles and David B. Grusky 902
Egalitarianism and Gender Inequality
GENDER GAP IN WAGES
105. Trond Petersen and Laurie A. Morgan 912
The Within Gender Wage Gap
106. Paula England 919
Devaluation and the Pay of Comparable Male
and Female Occupations
107. Tony Tam 924
Why Do Female Occupations Pay Less?
108. Francine Blau 929
The Sources of the Gender Pay Gap
HOW GENDER INTERSECTS
109. Margaret L. Andersen and Patricia Hill Collins 942
Why Race, Class, and Gender Matter
110. Emily Greenman and Yu Xie 943
Double Jeopardy
A STALLING OUT?
111. Paula England 955
The Gender Revolution: Uneven and Stalled
112. David Cotter, Joan M. Hermsen, and Reeve Vanneman 965
The Anti-Feminist Backlash and Recent Trends
in Gender Attitudes
113. Cecilia L. Ridgeway 973
The Persistence of Gender Inequality
xvi • Contents

PART IX: THE CONSEQUENCES OF INEQUALITY

LIFESTYLES
114. Pierre Bourdieu 982
Distinction: A Social Critique of the Judgment of Taste
115. Tak Wing Chan and John H. Goldthorpe 1004
The Social Stratification of Theater, Dance,
and Cinema Attendance
116. Annette Lareau 1013
Unequal Childhoods: Class, Race, and Family Life
117. Sean F. Reardon and Kendra Bischoff 1023
Income Inequality and Income Segregation
POLITICS AND ATTITUDES
118. Thomas Frank 1035
What’s the Matter with Kansas?
119. Michael Hout and Daniel Laurison 1037
The Realignment of U.S. Presidential Voting
HEALTH
120. Johannes Siegrist and Michael Marmot 1046
Health Inequalities and the Psychosocial Environment:
Two Scientific Challenges
121. Karen Lutfey and Jeremy Freese 1050
The Fundamentals of Fundamental Causality
122. Richard Miech, Fred Pampel, Jinyoung Kim,
and Richard G. Rogers 1056
Why Do Mortality Disparities Persist?

PART X: THE FUTURE OF INEQUALITY

INDUSTRIALISM AND POST-INDUSTRIALISM

123. Daniel Bell 1066
The Coming of Post-Industrial Society
124. Gøsta Esping-Andersen 1078
Social Foundations of Post-Industrial Economies
125. Michael Hout and Erin Cumberworth 1091
The Labor Force and the Great Recession
 Contents • xvii

POST-SOCIALISM
126. Victor Nee 1098
Post-Socialist Stratification
127. Gil Eyal, Iván Szelényi, and Eleanor Townsley 1104
Making Capitalism Without Capitalists
128. Andrew G. Walder 1110
Elite Opportunity in Transitions from State Socialism
POST-MODERNITY AND HIGH MODERNITY
129. John W. Meyer 1116
The Evolution of Modern Stratification Systems
130. Anthony Giddens 1125
Social Justice and Social Divisions
GLOBALIZATION AND INEQUALITY
131. Joseph E. Stiglitz 1132
Globalism’s Discontents
132. Glenn Firebaugh 1139
The New Geography of Global Income Inequality

About the Editors, 1151

Names Index, 1153
Subject Index, 1157
 PREFACE AND ACKNOWLEDGMENTS

Not so long ago, the standard lament of the inequality scholar was that no matter how many books
on rising inequality were published, the public seemed neither to notice nor to care all that much.
But certainly no one is issuing that lament now. For the first time in years, there is ongoing public
discussion of income inequality and the legitimacy of taxing the rich, indeed the 2012 presidential
election was partly a mandate on inequality issues. The Occupy Wall Street movement, despite its
current dormancy, is the first explicitly anti-inequality movement in recent US history. We also have
a new term, the “one percent,” to denote the elite within the United States, a term with a valence that
is not as straightforwardly worshipful as the elite have perhaps come to expect. The study of poverty
and inequality is no longer a sleepy little enterprise confined to the halls of academia.
How did all of this happen? It is possible that we have finally crossed some inequality threshold
beyond which the public simply cannot help but take notice. The recent financial crisis also may have
served to call into question the long-standing presumption that the rich deserve their rewards because
they work hard or are especially talented. Although some commentators speculated that the downturn
would be severe enough to usher in anti-inequality reforms of the New Deal variety, so far there have
been no such labor market reforms, and there is no evidence of a decrease in income inequality. As of
early 2013, the recovery remained a largely jobless one, and poverty and long-term unemployment
remained high.
These developments have generated as much interest within academia as outside it. Because the
field of stratification is recasting itself so quickly, we had no choice but to make this revision of Social
Stratification an especially substantial one, perhaps the most thoroughgoing revision ever. The classics
from the third edition have been retained in almost all cases, but many of the contemporary pieces
had become dated and had to be replaced. Of the 132 selections in the fourth edition, more than 40
percent (i.e., 57) are new.
This rapid change in the field has been driven by the erosion of old methodological and disciplinary
barriers and the rise of a more open and competitive social science landscape. We discuss below the
three most important developments of the last several decades.
The study of inequality has become deeply multidisciplinary. For better or worse, there is no longer
a consensual, interdisciplinary division of labor in which economists study the total output of goods
and sociologists study how goods and resources are unequally distributed. Although such a simple
division never existed even in the past, clearly economists have turned to the study of poverty and
inequality, once the special province of sociologists, with unprecedented vigor. Likewise, psychologists
and political scientists are making key contributions to the field, contributions that sociologists ignore
at their peril. We do not serve our students well by producing head-in-the-sand anthologies that ignore
the fact that contributions from political science, psychology, and economics now routinely enter into
sociological discourse.

xix
xx • Preface and Acknowledgments

The survey-based study is no longer so dominant. The standard survey instrument has come to be
appreciated as just one useful approach to studying poverty and inequality, and other approaches, espe-
cially qualitative and experimental ones, have clearly become as prominent. The rise of mixed-method
studies is also very striking. In many cases experiments and qualitative studies have attracted much
attention and become widely celebrated, and not just among a narrow academic audience.
The structure of social inequality is changing in rapid, complicated, and sometimes mystifying ways.
Although some types of inequality are under attack and appear to be weakening (e.g., suppression of
civil liberties for gays), we are also seeing unprecedented increases in income inequality, wealth in-
equality, and a deepening spatial divide between rich and poor. The fields of race and gender inequality
likewise reveal a tension between scholars who emphasize the rise of formally egalitarian practices and
those who stress the persistence of discrimination and segregation or the emergence of anti-egalitarian
backlashes that may undo gains of the past. Even as these backlashes play out, the relentless march of
egalitarian sensibilities exposes and delegitimates “new” inequalities that, not so long ago, were taken
for granted, rarely discussed, or barely seen (e.g., inequalities based on disability). We have included
many new contributions that address this constellation of trends and ask whether they allow for any
unified understanding of the forces making for equality and inequality.
As the foregoing discussion suggests, the research literature has become so large and complex that
the task of reducing it to manageable form poses many difficulties, not the least of which is simply
defining defensible boundaries for a field that at times seems indistinguishable from social science
at large. What types of commitments do we bring to the task of imposing order on such confusion?
Perhaps most obviously, we seek to represent the diversity of research traditions on offer, while at the
same time giving precedence to those traditions that have so far borne the greatest fruit. As is often the
case, the pool of disciplinary knowledge has developed in uneven and ramshackle fashion, so much
so that any attempt to cover all subjects equally would grossly misrepresent the strengths and weak-
nesses of contemporary stratification research. This sensitivity to disciplinary fashion reveals itself, for
example, in our decision to include many selections that address issues of race, ethnicity, and gender.
These subfields rose to a special prominence in the 1970s and continue to be popular even after forty
years of intensive and productive research. If the concepts of class, status, and power formed the holy
trinity of postwar stratification theorizing, then the partly overlapping concepts of class, race, and
gender play analogous roles now.
We have sought, secondarily, to give precedence to contemporary research, while also appreciating
that there is value in exposing students not just to the standard classics, but also to near-classics that
were written well after the foundational contributions of Karl Marx or Max Weber. The latter body of
work is often ignored by editors of anthologies, because it cannot be stamped with an exclusively Marx-
ian or Weberian imprimatur. Because we have often selected readings that best anticipate or motivate
current research fields and interests, the novice reader might be left with the impression that the disci-
pline is more straight-line cumulative than it truly is, an impression that should naturally be resisted.
The final, and most difficult, task faced by editors of anthologies is to chart an optimal course
between the Scylla of overly aggressive excerpting and the Charybdis of excessive editorial timidity. By
the usual standards of anthologies, we have charted a very average course, as our objective was to elim-
inate all inessential material while still preserving the integrity of the contributions. We have excised
many clarifying and qualifying footnotes, almost all decorative theorizing and literature reviews, and
much analysis that was not crucial in advancing the argument. Understandably, some of our readers
and contributors would no doubt oppose all excerpting, yet the high cost of implementing such a
radical stance would be a substantial reduction in the number of readings that could be reproduced.
We apologize to our authors for being unable to present the selections in their entirety and encourage
our readers to consult the original and full versions of our excerpted pieces.
 Preface and Acknowledgments • xxi

In some cases we have asked the authors themselves to provide trimmed versions of articles that
were originally published elsewhere, an approach that can yield more cohesive pieces when the ex-
cerpting would otherwise have to be very heavy. We have included many such new essays in this
edition. Although these pieces are crafted to be stand-alone contributions, we nonetheless encourage
our readers to consult the original and full versions for a more complete presentation of the research.
The editing rules adopted throughout this anthology were in most cases conventional. For exam-
ple, brackets were used to mark off a passage that was inserted for the purpose of clarifying meaning,
whereas ellipses were used whenever a passage appearing in the original contribution was excised
altogether. The latter convention was violated, however, if the excised text was a footnote, a minor
reference to a table or passage (e.g., “see table 1”), or otherwise inconsequential. When necessary, tables
and footnotes were renumbered without so indicating in the text, and all articles that were cited in
excised passages were likewise omitted from the list of references at the end of each chapter.
The spelling, grammar, and stylistic conventions of the original contributions were otherwise pre-
served. In this respect, the reader should be forewarned that some of the terms appearing in the original
contributions would now be regarded as inappropriate or offensive, whereas others have passed out
of common usage and will possibly be unfamiliar. Although a strong argument could be made for
eliminating all language that is no longer acceptable, this type of sanitizing would not only exceed
usual editorial license but would also generate a final text that contained inconsistent and possibly
confusing temporal cues.
Among the various functions that an anthology performs, one of the more obvious is to define
and celebrate what a field has achieved, and in so doing to pay tribute to those who have made such
achievement possible. We are grateful to the many dozens of scholars who allowed their work to be
reproduced for this anthology or who agreed to write one of the many commissioned essays. This
book provides a well-deserved occasion to recognize the many successes of a field that is perhaps better
known for its contentiousness and controversy.
We follow convention and close by acknowledging the village that it took to write and publish
what has come to be hopefully (lovingly) termed “the brick.” Early in the process we sought to carry
out a formal market analysis, and the anonymous reviewers who agreed to assist in this task provided
unusually constructive criticism. In selecting the new contributions, we also relied on our own trusted
advisors, including Michelle Jackson, Kim Weeden, Michael Hout, and Tim Smeeding. Over the
course of a yearlong production process, we relied extensively on the excellent Westview staff, espe-
cially Priscilla McGeehon, Cathleen Tetro, Brooke Smith, Annie Lenth, and Cisca Schreefel. Finally,
we must thank our Westview Press senior editor, Leanne Silverman, for bearing with us so patiently,
for choreographing the entire process so expertly, and for such excellent advice on . . . everything.
David B. Grusky and Katherine R. Weisshaar
Stanford, California, 2013
 RESOURCES FOR THE STUDY
OF POVERTY AND INEQUALITY

In addition to the three-volume series of readers—one of which you hold in your hands—we offer
multiple resources to further support your interest in and study of poverty and inequality:

The Three-Volume Series

An advanced reader: The original book in our three-part series, Social Stratification, combines a full
roster of classic readings with the most advanced scholarship on poverty and inequality. Now fully
revised and updated, the fourth edition of Social Stratification remains the industry standard, the
most complete introduction available in the field.
An introductory reader: The Inequality Reader is intended for those requiring a more streamlined,
less technical introduction to the field. This volume draws broadly from classic theoretical and
conceptual pieces, qualitative and ethnographic studies, and quantitative analyses of the sources
of poverty and inequality.
A compilation of the classics: Published in 2006, Classic Readings in Race, Class, and Gender is a yet
slimmer volume of essential readings in poverty and inequality, readings that define the concep-
tual and empirical foundations of the field. This book, which represents the classical core alone,
may be supplemented with contemporary articles that are tailored to particular topics or interests.

Online Sources
Online Course: A new online course, offered via OpenEdX, draws on readings in Social Stratification
and features dozens of leading scholars introducing the key research that has transformed the field.
Website: Our website, www.inequality.com, provides recent data, news, and research on poverty and
inequality.
Facebook: The latest updates on poverty and inequality issues are available at our Facebook page (go
to The Stanford Center on Poverty and Inequality).
Twitter: For alerts on poverty and inequality news, please follow us (@CenterPovIneq).

xxii
 Grusky and Weisshaar—The Questions We Ask About Inequality • 1

Introduction

1 • David B. Grusky and Katherine R. Weisshaar

The Questions We Ask About Inequality

The dispassionate observer of the last half century of poverty and inequality research might easily
characterize the field as one of the great success stories of modern social science. The breakthroughs
are many and include (1) building a sophisticated infrastructure for monitoring key trends in poverty,
segregation, income inequality, racial and gender inequality, and other labor market outcomes; (2)
resolving long-standing debates about how much discrimination there is, whether residential segrega-
tion matters, how much social mobility there is, and whether social classes are disappearing; and (3)
making substantial headway in understanding the deeper causes of poverty, mobility, and many types
of inequality. If poverty and inequality are still very much with us, it is not because they are any longer
“puzzles” or enigmas driven by forces we cannot fathom. It instead is because we have collectively
decided not to undertake the well-known institutional reforms necessary to reduce them.
Although we have come quite far, our optimistic stance has to be tempered with some appreciation
of the field’s equally visible shortcomings. The most prominent failures have occurred when scholars
have turned to the idiographic task of understanding the dynamics of change. The spectacular takeoff
in income inequality, one of the most consequential developments of our time, was largely unpre-
dicted. The resilience of school segregation was something of a surprise to most social scientists. The
sudden “stalling out” of long-standing declines in gender inequality remains poorly understood. The
equally sudden evaporation of opposition to gay marriage and other types of gay rights was unimag-
inable just five years ago. The lightning-fast rise and fall of one of the first anti-inequality movements
in contemporary US history (the Occupy Wall Street movement) is likewise a puzzle. Although the
field has by this accounting performed less impressively in understanding the dynamics of change, we
must concede that this is hardly a distinctive feature of the poverty and inequality field itself. Across all
fields of social science, the idiographic task has never fully yielded to science, an obvious implication
of having but one world and one case to explain.
This chapter introduces the types of questions that have been and continue to be taken on in the
field of poverty and inequality. We consider questions pertaining to the lawlike behavior of frequently
repeated events (i.e., “nomothetic” phenomena) as well as those pertaining to the way in which singu-
lar entities, such as the US stratification system, are evolving (i.e., “idiographic” phenomena). Because
the often-complicated answers to these questions are laid out in the readings themselves, it would be
unwise to attempt here an encyclopedic introduction to all that is known in the sprawling poverty and
inequality field. Rather, our objective is to discuss the types of questions that have been taken on, with
a special emphasis on understanding how those questions have changed and evolved.

Original article prepared for this fourth edition of Social Stratification: Class, Race, and Gender in Sociological
Perspective, edited by David B. Grusky. Copyright © 2014 Westview Press.
2 • Introduction

The second half of this introductory chapter shifts the focus from the questions that we ask to the
stories that we tell. That is, whereas our purpose here is to provide a roadmap to the book and the
types of questions that our contributors have taken on, the second half introduces readers to the main
narratives about the trajectory of inequality that show up throughout the chapters of this book.

Basic Concepts
Before introducing the ten main lines of inquiry that form the intellectual backbone of the poverty
and inequality field, we discuss some of the basic concepts used to describe the amount and structure
of inequality, its genesis and persistence, and its effects on social behavior. The key components of
stratification systems are (1) the institutional processes that define certain types of goods as valuable
and desirable; (2) the rules of allocation that distribute these goods across various jobs or occupations
in the division of labor (e.g., doctor, farmer, “housewife”); and (3) the mobility mechanisms that link
individuals to jobs, thereby generating unequal control over valued resources. It follows that inequality
is produced by two types of matching processes: the social roles in society are first matched to “reward
packages” of unequal value, and individual members of society are then allocated to the positions so
defined and rewarded.1 In all societies, there is a constant flux of occupational incumbents as new-
comers enter the labor force and replace dying, retiring, or out-migrating workers, yet the positions
themselves and the reward packages attached to them typically change only gradually. As Schumpeter
(1953) puts it, the occupational structure can be seen as “a hotel . . . which is always occupied, but
always by different persons” (171).
What types of rewards do these two matching processes distribute? It is increasingly fashionable
to recognize that inequality is “multidimensional,” that income is accordingly only one of many im-
portant resources, and that income redistribution in and of itself would not eliminate inequality (e.g.,
Sen, Chapter 29). In taking a multidimensionalist approach, one might usefully distinguish among
the eight types of assets listed in Table 1.1, each understood as valuable in its own right rather than as a
mere investment item.2 These assets may of course serve investment as well as consumption functions.
For example, most economists regard schooling as an investment that generates future streams of
income, and some sociologists likewise regard social networks as forms of capital that can be parlayed
into educational credentials, income, and other valued goods.3 Although most of the assets listed in
Table 1.1 are clearly convertible in this fashion, the individuals involved do not necessarily regard them
as investments. In fact, many valuable assets can be secured at birth or through childhood socialization
(e.g., the “good manners” of the aristocracy), and they are therefore acquired without the beneficiaries
explicitly weighing the costs of acquisition against the benefits of future returns.4
The implicit claim underlying Table 1.1 is that the listed assets exhaust all consumption goods and
as such constitute the raw materials of stratification systems. The stratification field has developed a
vocabulary that describes how these raw materials are distributed among the population. The main
parameters of interest are (1) the overall amount of inequality, (2) the extent to which individuals are
locked permanently into certain positions (i.e., rigidity), (3) the extent to which ascriptive traits fixed
at birth are used for purposes of allocation, and (4) the degree to which the various dimensions of
inequality cohere (or are “crystallized”).
The overall amount of inequality in any given resource (e.g., income) may be defined as its dis-
persion or concentration among the individuals in the population. Although many scholars seek to
characterize the overall level of societal inequality with a single parameter, such attempts will obviously
be compromised insofar as some types of rewards are distributed more equally than others. This com-
plexity clearly arises in the case of contemporary inequality regimes. The recent diffusion of “citizen-
ship rights,” for example, implies that civil goods are now widely dispersed among all citizens, whereas
economic and political goods continue to be controlled by a comparatively small elite.
 Grusky and Weisshaar—The Questions We Ask About Inequality • 3

Table 1.1. Types of Assets and Examples of Advantaged and Disadvantaged Groups

Assets Examples
Asset group Examples of types Advantaged Disadvantaged
1. Economic Wealth Billionaire Bankrupt worker
Income Professional Laborer
Ownership Capitalist Worker (i.e., employed)
2. Power Political power Prime minister Disenfranchised person
Workplace authority Manager Subordinate worker
Household authority Head of household Child
3. Cultural Knowledge Intelligentsia Uneducated persons
Digital culture Silicon Valley residents Residents of other places
“Good” manners Aristocracy Commoners
4. Social Social clubs Country club member Nonmember
Workplace associations Union member Nonmember
Informal networks Washington A-list Social unknown
5. Honorific Occupational Judge Garbage collector
Religious Saint Excommunicate
Merit-based Nobel Prize winner Nonwinner
6. Civil Right to work Citizen Unauthorized immigrant
Due process Citizen Suspected terrorist
Franchise Citizen Felon
7. Human On-the-job training Experienced worker Inexperienced worker
General schooling College graduate High school dropout
Vocational training Law school graduate Unskilled worker
8. Physical Mortality Person with long life A “premature” death
Physical disease Healthy person Person with AIDS, asthma
Mental health Healthy person Depressed, alienated

The rigidity of a stratification system is indexed by continuity in the social standing of its members.
We consider a stratification system to be rigid, for example, insofar as the current wealth, power, or
prestige of individuals can be accurately predicted on the basis of their prior statuses or those of their
parents. It should again be emphasized that such rigidity (or “social closure”) will typically vary in
degree across the different types of assets listed in Table 1.1.
The stratification system rests on ascriptive processes to the extent that conditions present at birth
(e.g., parental wealth, sex, race) influence the subsequent social standing of individuals. If ascriptive
processes of this sort are very prominent, it is possible (but by no means guaranteed) that the un-
derlying traits will become bases for group formation and collective action (e.g., race riots, feminist
movements). In contemporary societies, ascription of all kinds is usually seen as undesirable or dis-
criminatory, and much governmental policy (e.g., antidiscrimination legislation) is therefore oriented
toward fashioning a stratification system in which individuals acquire resources solely by virtue of
their achievements.5
The degree of crystallization is indexed by the correlations among the various resources (e.g., in-
come, wealth, education) that are socially valued. If these correlations are strong, then the same in-
dividuals (i.e., the “upper class”) will consistently appear at the top of all hierarchies, while other
individuals (i.e., the “lower class”) will consistently appear at the bottom of the stratification system.
By contrast, various types of status inconsistencies (e.g., a poorly educated millionaire) will emerge
in stratification systems with weakly correlated hierarchies, and it is correspondingly difficult in such
systems to define a unitary set of classes that have predictive power with respect to all resources.
4 • Introduction

Although the preceding terms are fundamental to the field, there are of course a great many others,
some of which are introduced in the following sections. We turn next to introducing the ten subfields
around which the field of inequality—and this book—are organized.

Core Fields of Inquiry

The main task of stratification scholarship, describing the contours of inequality and explaining its
causes, is pursued in well-defined subfields that have become so large that it is not unusual for scholars
to spend their entire careers working within just one of them. It is accordingly possible to provide only
brief summaries of each subfield.

The Functions and Dysfunctions of Inequality

In economics and sociology alike there is a long theoretical tradition emphasizing that some forms of
inequality have the positive effect of (1) inducing the most qualified workers to take on the impor-
tant jobs and (2) incentivizing people to work hard and innovate. The famous trade-off thesis implies
that a taste for equality is exercised at the cost of reducing output and potentially leaving everyone
worse off. We cannot, as Okun (1975, 2) so cleverly put it, “have our cake of market efficiency and
share it equally.” In Part I we introduce some of the key protagonists in this debate, with Davis and
Moore (Chapter 2) and Tumin (Chapter 3) presenting the classic sociological version of this debate,
and Fischer and his colleagues (Chapter 4) providing the requisite update. Although some commen-
tators and ideologues continue to take the trade-off thesis as an article of faith, most social scientists
now treat it as testable by asking whether groups or societies with higher inequality are indeed more
productive.
The trade-off thesis of course implies that inequality is functional only insofar as it delivers the
proper incentives. There are many ways in which high-inequality regimes may not be incentivizing all
people to work hard in a manner that then increases total output. If, for example, inequality takes a
simple winner-take-all form, it could be counterproductive and reduce output. The relevant thought
experiment here is that of a golf tournament in which all prize money is allocated exclusively to the
winner (e.g., Freeman 2012). In this setup, mediocre golfers who never have a chance of winning lack
any incentive to exert themselves, with the implication that total output (as measured, say, by total
strokes) will diminish. It follows that certain forms of extreme inequality may not succeed in creat-
ing the incentives that the trade-off thesis presumes. Likewise, insofar as inequality arises because of
corruption and rent-seeking behavior, it is not contributing to overall output and cannot be justified
in functionalist terms. There is presumably no collective benefit, for example, in allowing a chief ex-
ecutive officer (CEO) to pack the board with cronies simply because they will then vote him or her a
generous compensation package (see DiPrete, Eirich, and Pittinsky, Chapter 12). The current debate
about the importance of rent and corruption in generating the takeoff in income inequality (e.g.,
Grusky 2012) is motivated by precisely this interest in assessing whether extreme inequality of the sort
observed today can be understood as efficient or functional.

Inequality in Comparative Perspective

If inequality is a ubiquitous feature of human society, it nonetheless has taken on many different
forms. The purpose of Part II is to examine the historical and cross-national variability in the amount
of inequality, the types of assets that underlie inequality (e.g., power, prestige, economic resources),
the extent to which opportunities vary with social origins and other ascriptive traits, and the ideologies
by which prevailing inequalities are legitimated. The differences among tribal, feudal, caste, socialist,
 Grusky and Weisshaar—The Questions We Ask About Inequality • 5

and capitalist forms of inequality are especially stark on these and other dimensions (see Grusky and
Weisshaar, Chapter 5), but even within contemporary capitalist societies the institutional variability
is arguably impressive (see Esping-Andersen and Myles, Chapter 6). Although popular commentators
often assume that market economies are associated with a single characteristic form of inequality, in
fact the evidence clearly reveals the wide variety of ways that inequality has been and continues to be
organized within market economies.
The same lesson is readily gleaned by examining how income inequality has evolved over the last
century in market societies (see Chapters 7–12). As Atkinson, Piketty, and Saez (Chapter 7) show,
the trend in income inequality follows a U-shaped form in the English-speaking countries (i.e., the
United States, Canada, Ireland, the United Kingdom, Australia, and New Zealand), all of which have
so-called liberal market economies. In each of these countries, inequality dropped precipitously in the
late 1920s and during World War II, but then stabilized at a comparatively low level over the next
thirty years. During the 1970s, inequality then began its infamous takeoff, ultimately returning to lev-
els approaching or surpassing those that prevailed in the 1920s. By contrast, a rather different pattern
obtains in coordinated market societies, such as France and Germany. Although these countries also
experienced a real decline in inequality during the first half of the twentieth century, that decline was
not followed by any subsequent rebound in inequality of the sort found in the United States and the
other English-speaking countries. Instead, we see a rough stability in the amount of inequality or, in
a few cases, even a continuing slight decline. The key task of Part II is to explain why there has been
such a precipitous takeoff in inequality in some countries but not others (see Chapters 7–13).

The Structure of Inequality

Because so much social science research now focuses on the takeoff in inequality, we have a growing
body of knowledge about what is driving that takeoff (e.g., Goldin and Katz, Chapter 8), who has
benefited most from it (e.g., Hacker and Pierson, Chapter 11), and which countries have experienced
it in especially virulent form (e.g., Atkinson, Piketty, and Saez, Chapter 7). We know less, however,
about the form that inequality takes. This is surprising because some of the most contentious debates
within sociology are about the form rather than the extent of inequality. Although most sociologists
assume that inequality is organized around big social classes (e.g., capitalists, managers, craft workers)
that affect the politics, interests, and life chances of their members, there is much debate about (1)
exactly how big such classes are, (2) how best to draw the boundaries between them, and (3) what types
of assets underlie them. The chapters in Part III examine the many brands of class models as well as
some of the most prominent alternatives to them.
We discuss class models in fuller detail in the section “Measuring Inequality.” The main point of
interest here is that debates about the form of inequality do speak to issues of some consequence. If the
objective, for example, is to understand the experience of poverty, one should care whether those with
low income also tend to have very little in the way of other resources, such as wealth or education. We
should presumably be more troubled by a form of poverty in which many types of disadvantage (e.g.,
income, wealth, education) come together. The form that inequality takes will also probably affect how
it is experienced and whether it is well tolerated. In a highly crystallized system, an economic crisis
might generate a resurgence of class-based rhetoric, with the sharpest vitriol aimed at a culturally ho-
mogenous upper class with distinctive lifestyles, attitudes, and behaviors. In a less crystallized system,
the same economic crisis might generate a more fragmented political discourse, perhaps equally con-
tentious in its own way but with little coherence across issues and a weaker “us versus them” rhetoric.
The chapters in Part III consider arguments of this sort while also paying special attention to
Marxian, Weberian, and Durkheimian formulations of class inequality (see Chapters 14–21). We then
turn to gradational models that represent inequality either as (1) a hierarchy of individuals differing in
6 • Introduction

their income, wealth, or education (see Chapters 28–29) or (2) a hierarchy of occupations differing in
their prestige or socioeconomic status (see Chapters 24–27). Although the usefulness of class models
is currently a matter of debate (see Weeden and Grusky, Chapter 23), gradational models have quietly
disappeared from the day-to-day practice of inequality research without much fanfare. It appears that
this development can be attributed as much to the inevitable half-life of any research tradition as to
any hard empirical evidence that gradationalism is declining over time. The last two chapters of Part
III lay out various ways in which the gradationalist approach might be resuscitated within sociology
(Sørensen, Chapter 28; Sen, Chapter 29).

The Ruling Class, Elites, and the Upper Class

In recent years “the one percent” has become the modern-day label for the elite, but it is likely that
few commentators seriously believe that an arbitrary income threshold of that sort captures very well
the structure of power and advantage at the top. The chapters in Part IV develop a more rigorous
vocabulary for understanding how power and advantage can be organized (see Chapters 30–32). This
conceptual apparatus is then used to understand the structure of elites and the upper class in capitalist
(e.g., Chapters 33–35) and formerly socialist (Chapters 36–37) societies.
The key question that cuts across the readings of Part IV is whether late industrial societies gen-
erate a powerful and solidary upper class. The “power elite” narrative (see Mills, Chapter 31) holds
that economic, political, and military elites form a unitary class by virtue of residing in the same
neighborhoods, belonging to the same clubs, sharing a coherent political ideology, and interacting
as approximate equals. This small elite occupies positions of considerable power and can therefore
“mightily affect the everyday worlds of ordinary men and women” (Mills, Chapter 31, p. 283). The
most prominent opposing view is that capitalists, by virtue of their economic power, have become
the de facto “ruling intellectual force” (Marx, Chapter 14, p. 141). This view is perhaps more widely
diffused than its Marxian heritage might suggest. Although key leadership positions are of course for-
mally filled by political or military elites, it is no longer deemed all that radical to suggest that they are
acting at least partly at the behest of capitalists and corporations (who wield influence via lobbying,
campaign funding, and the like).
More recently, Gouldner (Chapter 34) has suggested that a third claimant, the “knowledge class,”
is on the road to power as expert knowledge becomes a source of value and a prerequisite for respon-
sible and successful leadership (see also Brooks, Chapter 35). For such “new class” theorists, the main
struggle of the contemporary period is not between capitalists and workers (because workers long ago
lost out), but rather takes the form of the ongoing “civil war” (Gouldner, Chapter 34, p. 306) between
capitalists with the power to buy influence and intellectuals who claim to have the knowledge to suc-
cessfully wield it. The outcome of this civil war is far from obvious. If capitalists bring ever-growing
economic power to the table, intellectuals rest their case on the efficiency of relying on results-driven
scientific and expert knowledge. The readings in Part IV also address the more complicated struggles
that arise when capitalists and intellectuals confront, as they do in formerly socialist societies, an un-
usually powerful political elite (see Eyal, Szelényi, and Townsley, Chapter 36; Walder, Chapter 37).

Poverty and the Underclass

In Part V our attention turns to the bottom of the class structure, where a rather different set of
questions comes to the fore. Whereas scholars of elites have typically emphasized the power struggles
among competing factions (e.g., capitalists, intellectuals, political elites), scholars of poverty have
focused on the various external forces that have affected the amount and experience of poverty. These
forces take the form of (1) impersonal economic factors (e.g., globalization) that have eliminated
 Grusky and Weisshaar—The Questions We Ask About Inequality • 7

manufacturing jobs for low-skill workers (Wilson, Chapter 40); (2) the relatively anemic redistributive
programs within the United States as a whole (Kenworthy, Chapter 42) and the South more specifi-
cally (Newman and O’Brien, Chapter 43); (3) the extremely high rates of incarceration among blacks
(Western and Pettit, Chapter 51); (4) the role of concentrated poverty in reducing access to jobs, good
schools, and other neighborhood assets (Chapters 45–49); and (5) the lasting effects of early child-
hood stress on cognitive development (Chapters 48–50). This literature on the institutional sources
of poverty is joined, of course, by much research on individual-level “risk factors” that predispose one
to poverty. It is well known, for example, that poverty rates are especially high among children, young
adults, the less educated, blacks and Hispanics, and female heads of households (Danziger, Chavez,
and Cumberworth, Chapter 41).
The foregoing institutional factors combine to produce an unusually high poverty rate in the United
States. In 1973, the official poverty measure dropped to a historic low of 11.1 percent, but that level
was never again achieved during the following four decades. The recent recession, like most prior ones,
caused a surge in poverty. The official measure stood at 15.1 percent in early 2013, the highest level in
the last half century, and most scholars expect that poverty will remain high for many more years (Dan-
ziger, Chavez, and Cumberworth, Chapter 41). Because the forces behind this unusually high poverty
rate are well understood, one could readily effect institutional reforms that would drastically reduce
poverty, as indeed some countries have done (see Grusky and Weeden 2013). It bears reemphasizing
that poverty is not some deeply complicated puzzle that social scientists are laboring to “crack.” We have
therefore included several readings that ask why the United States has been so reluctant to undertake
well-known institutional reforms that would reduce poverty (see especially Kenworthy, Chapter 42).

Who Gets Ahead?

The language of stratification theory makes a sharp distinction between the distribution of social re-
wards (e.g., the income distribution) and the distribution of opportunities for securing these rewards.
As sociologists have frequently noted, the typical American is quite willing to tolerate substantial in-
equalities in power, wealth, or prestige provided that the opportunities for securing these social goods
are distributed equally. If they can be assured that the competition has been fair, most Americans are
willing to reward the winners and punish the losers. This line of reasoning has motivated the long
tradition of research on social mobility introduced in Part VI.
We lead off this section with an introduction to the analysis of conventional “mobility tables” that
represent the intergenerational flows between origin and destination class categories. These tables tell
us, for example, how frequently children born into the craft category move up to the professional
or managerial sector, remain within the craft category, or move down to less-skilled manual work.
It is of course not possible on the basis of such raw observational data to make any straightforward
assessment of the opportunities available to children born into each social class. This is partly because
standard outcome-based measures of inequality are tainted by the confounding effect of differential
tastes. For example, a worker with a well-developed taste for leisure may opt to work for relatively few
hours, leading to low earnings but nonetheless optimal utility (by virtue of the high valuation placed
on leisure). Whenever differential outcomes are generated through the operation of differential tastes,
some scholars argue that the resulting variations in outcomes are unproblematic, because they simply
express trade-offs that are being made among different types of valued goods (e.g., leisure and income).
This line of reasoning implies that inequality scholars should measure the prevailing distribution of op-
portunities before differential tastes can express themselves. The key objective under this formulation
is to determine whether “capabilities” (i.e., opportunities to secure rewards) are equally distributed,
not whether rewards themselves, which reflect the operation of tastes, are equally distributed. It bears
noting that, insofar as tastes themselves are endogenous (by virtue of parental socialization), it is yet
8 • Introduction

more difficult to tease out the way in which social origins constrain destinations on the basis of con-
ventional observational data.
This is all to suggest that one cannot read the mobility table as straightforwardly expressing the
extent to which opportunities are unequally distributed. Even so, most sociologists and social scientists
tend to set aside all such complexities, and analyses of mobility tables are thus understood as exposing,
if only imperfectly, the extent to which the liberal ideal of an open system has been realized. This
objective underlies analyses of (1) the gross association between occupational origins and destinations
(Chapters 52–55); (2) the extent to which children garner a return from advantaged economic origins
(Chapters 56–57); (3) the extent to which educational investments are a key mediating variable be-
tween origins and destinations (Chapters 58–59); (4) the role of social psychological variables, such as
aspirations and expectations, in affecting the process of attainment (Chapters 67–69); (5) the effects
of networks on occupational and other outcomes (Chapters 74–78); and (6) the effects of labor mar-
ket institutions (e.g., unions) on remuneration and bargaining power (see Chapters 70–73). We have
also devoted an entire subsection to examining how educational opportunities are allocated (Chapters
60–64) and how educational investments are rewarded (Chapters 65–66).

Race and Ethnicity

In Part VII our attention shifts to racial and ethnic inequalities, which are arguably the most profound
forms of ascription in the US case (and in those of many other countries). The obvious starting point
in studying race and ethnicity is to understand how the classification schemes themselves (e.g., “black,”
“white,” “Asian”) are constructed and reconstructed over time. We have included a classic reading on
the institutional processes by which such schemes come into being and are reproduced (see Omi and
Winant, Chapter 79) as well as a newer study revealing how individuals are sorted (and sort them-
selves) into schemes that are already in place (Saperstein and Penner, Chapter 80).
The next readings in Part VII examine how newly entering groups, not just particular individuals,
are sorted into the racial and ethnic system in the host country. The chapters in this subsection ex-
plore whether “straight-line” assimilation is in play (see Alba and Nee, Chapter 83; Portes and Zhou,
Chapter 84) or is instead interrupted by the rise of ethnic enclaves (Portes and Manning, Chapter 82)
or the constant replenishment of the Hispanic population (see Jiménez, Chapter 85). We then turn to
the role of discrimination in perpetuating racial and ethnic disadvantage. Although statistical analyses
of discrimination have been performed for some time, they have been less influential than recent ex-
perimental studies in documenting the extent of discriminatory processes (Bertrand and Mullainathan,
Chapter 86; Pager, Chapter 88) and showing how discrimination and stereotyping can affect behavior
in self-fulfilling ways (Steele, Chapter 87).
The final subsection of Part VII explores the forces of change and stability in racial and ethnic
inequality. Because our various institutions have all been profoundly racialized, there are accordingly
many institutional arenas in which change might occur, a point that Wilson (Chapter 89) makes
effectively. We have included new research on (1) the rise of interracial assortative mating (Qian and
Lichter, Chapter 91), (2) the mechanisms through which Latino immigrants are affected by and affect
our binary understanding of race (Frank, Akresh, and Lu, Chapter 90), and (3) the continuing role of
urban joblessness in reproducing racial inequality (Wilson, Chapter 89).

Gender Inequality
In sections VI through VIII we cover in turn the types of inequality that have been built up around
social origins, race, and gender, each a key form of ascription in the contemporary context. It was
once fashionable to argue that all forms of ascription are fated to wither away because (1) they rest on
 Grusky and Weisshaar—The Questions We Ask About Inequality • 9

a caste-like allocation of labor that ignores considerations of merit and talent, and (2) they are incon-
sistent with our moral commitment to running an equal-opportunity society (Parsons 1994). By this
logic, the ascription that persists to this day is but a residual holdover, with the twin forces of efficiency
and egalitarianism slowly but gradually wearing it away.
But are all forms of ascription indeed created equal? Although each type has its own peculiarities,
the gender form is especially distinctive by virtue of playing out so prominently within families.
Because we watch “up close and personal” the gender division of labor in our families, we are under
special pressure to believe essentialist stories justifying it, stories that represent women as especially
suited for nurturing or service and men as especially suited for cognitive or manual labor. These stories,
some have argued (see Charles and Grusky, Chapter 104), take on a powerful life of their own, tend to
legitimate and perpetuate an essentialist division of labor, and thereby impart a special staying power
to gender-based ascription.
For this and many other reasons, we might expect gender inequality to evolve under a logic that
differs from that governing racial, ethnic, or class inequalities. We examine the structure of this logic
in Part VIII. We ask whether women are “opting out” of the workplace (Chapters 92–94); whether
there is much discrimination against women, mothers, and gay men (Chapters 95–99); whether there
are “glass ceilings” and other forms of workplace hypersegregation (Chapters 100–104); whether the
gender gap in wages can be explained away by occupational and other variables (Chapters 105–108);
whether race, class, and gender interact and intersect in complicated ways (Chapters 109–110); and
whether there is a recent “stalling out” in the historic decline in gender inequality (Chapters 111–113).
As with many other subfields, we find that the ideographic questions on this list (i.e., those featuring
the dynamics of change) have proven difficult to resolve, whereas the nomothetic ones have been more
successfully addressed.

The Consequences of Inequality

We have focused to this point on the twofold task of describing various types of inequality and then
examining how they are generated and reproduced. In Part IX, our attention shifts to the consequences
of inequality, with a particular emphasis on the many types of microlevel consequences. The microlevel
agenda, which examines the effects of an individual’s social position on individual outcomes (e.g.,
lifestyles, attitudes, health), is of course a characteristically modern obsession. The classical literature
on inequality, especially that of Marx (Chapter 14), focused instead on the macrolevel effects of in-
equality on grander systemic change (e.g., revolution). The macrolevel tradition is not well represented
in Part IX, but it does live on in the form of contemporary cross-national studies of the effects of in-
equality on total economic output, ethnic unrest, democracy, and much more. It also continues in the
form of speculative commentary, much of which is included in Part X, on the various ways in which
present-day inequality regimes contain the kernel of future inequality regimes.
The purpose of Part IX is to address the more modest question of how an individual’s position
within an inequality regime affects the course of her or his life. We have included chapters that examine
such microlevel effects on (1) lifestyles, tastes, and consumption practices (Bourdieu, Chapter 114;
Chan and Goldthorpe, Chapter 115); (2) childrearing and socialization practices (Laureau, Chapter
116); (3) the type of neighborhood to which one is exposed (Reardon and Bischoff, Chapter 117); (4)
political attitudes and voting behavior (Frank, Chapter 118; Hout and Laurison, Chapter 119); and
(5) opportunities for both a healthy (Siegrist and Marmot, Chapter 120; Lutfey and Freese, Chapter
121) and long life (Miech et al., Chapter 122). It should come as no surprise that microlevel effects of
this sort are ubiquitous. In contemporary societies, almost everything is marketized and commodified,
meaning that one’s participation in a wide variety of domains is directly affected by the capacity to
pay for that participation. Because of this broad marketization, the cynic might argue that the harder
10 • Introduction

intellectual challenge would be to find a good, service, or outcome that is somehow untouched by
“class,” one that is perfectly and equally distributed to all.

The Future of Inequality

We led off this introduction by arguing that the field has made much headway in addressing standard
nomothetic questions about the causes of poverty, mobility, and inequality (bearing in mind the in-
evitable complications in making any causal inferences). The field’s record has been less impressive,
however, when it comes to predicting such developments as the takeoff in income inequality, the
recent stalling of prominent gender-equalizing trends, or the seemingly sudden embrace of a straight-
forwardly egalitarian position on matters of sexual orientation.
This relatively poor record gives us pause in evaluating the wide range of predictive narratives found
in Part X. For purposes of comparison and historical perspective, we have led off this section with
one of the venerable essays on postindustrialism (Bell, Chapter 123) as well as a now-classic updating
of that argument (Esping-Andersen, Chapter 124; also see Hout and Cumberworth, Chapter 125).
We turn thereafter to contemporary narratives about the postsocialist transition (Chapters 126–128),
the rise of postmodernity and high modernity (Chapters 129–130), and the forces of globalization
(Chapters 131–132). The role of such narratives within the field is discussed in more detail in the
following chapter.

Measuring Inequality
We turn now to the way in which inequality has characteristically been measured. Because the mea-
surement of inequality has been so controversial, it is important to provide at least some introduction
to the key debates. We do so in Table 1.1 by listing of the eight types of “raw materials” from which
inequality regimes are crafted. Given the seeming complexity of contemporary inequality, one might
have expected that most scholars would adopt a multidimensional approach in which each individual
is located in terms of the eight scores of Table 1.1, thus yielding an eight-dimensional “inequality
space.” This instinct, however reasonable it might seem, has not in fact been dominant in the field. The
convention among sociologists has been to describe stratification systems in terms of discrete classes
or strata whose members are presumed to have similar levels or types of assets. In the most extreme
versions of this approach, the resulting classes are assumed to be real entities that exist before the dis-
tribution of assets, and many scholars therefore refer to the “effects” of class location on the assets that
their incumbents control. We present here a stylized history of such class models as well as competing
approaches to characterizing the structure of inequality.
The claim that inequality takes on a “class form” is one of the distinctively sociological contribu-
tions to inequality measurement and is the main alternative to approaches that either focus exclusively
on income inequality or analyze the many dimensions of inequality independently and separately. The
main advantage of class-based measurement, as argued by some sociologists, is that conventional class
categories (e.g., professional, manager, clerk, craft worker, laborer, farmer) are institutionalized within
the labor market and are accordingly more than purely nominal or statistical constructions. The labor
market, far from being a seamless and continuous distribution of incomes, is instead understood as
a deeply lumpy entity, with such lumpiness mainly taking the form of institutionalized groups (i.e.,
“classes”) that constitute prepackaged combinations of the valued goods listed in Table 1.1.
Within sociology the implicit critique of income-based approaches rests not so much on the argu-
ment that income is just one of many distributions of interest (i.e., multidimensionalism), but rather
on the argument that measurement strategies based on income alone impose an excessively abstract,
analytical, and statistical lens on a social world that has much institutionalized structure to it. This
 Grusky and Weisshaar—The Questions We Ask About Inequality • 11

structure takes the tripartite form of (1) a set of social classes that are privileged under capitalist labor
markets (e.g., capitalists, professionals, managers), (2) a set of social classes that are less privileged under
advanced capitalism (e.g., routine nonmanuals, craft workers, operatives), and (3) an “underclass” that
stands largely outside the labor market and is accordingly deeply disadvantaged in market systems.
The rise of class models should therefore be understood as a distinctively sociological reaction to the
individualism of the income paradigm and other unidimensional approaches to measuring inequality.
The foregoing account, which is a largely consensual rendition of the rationale for social class mea-
surement, nonetheless conceals much internal debate within the field over how best to identify and
characterize the boundaries dividing the population into classes. These debates have played out in the
three phases described in the following sections.

Structuralist Phase (ca. 1945–1985)

The class models of the postwar period rested implicitly or explicitly on the assumption that classes
are bundles of endowments (e.g., education levels), working conditions (e.g., amount of autonomy),
and reward packages (e.g., income) that tend to cohere (esp. Giddens, Chapter 19; Parkin, Chapter
20). The middle class of “craft workers,” for example, comprises individuals with moderate educa-
tional investments (i.e., secondary school credentials), considerable occupation-specific investments in
human capital (i.e., on-the-job training), average income coupled with substantial job security (at least
until deindustrialization), middling social honor and prestige, quite limited authority and autonomy
on the job, and comparatively good health outcomes (by virtue of union-sponsored health benefits
and regulation of working conditions). By contrast, the underclass may be understood as comprising
a rather different package of conditions, one that combines minimal educational investments (i.e.,
secondary school dropouts), limited opportunities for on-the-job training, consequently intermittent
labor force participation and low income, virtually no opportunities for authority or autonomy on
the job (during those brief bouts of employment), relatively poor health (by virtue of lifestyle choices
and inadequate health care), and much social denigration and exclusion. The other classes appearing
in conventional class schemes (e.g., professional, managerial, routine nonmanual) may likewise be
understood as particular combinations of scores on the dimensions of Table 1.1.
For purposes of illustration, consider a simplified case in which the multidimensional “inequality
space” comprises only three individual-level variables (e.g., education, autonomy, income), thus allow-
ing the class hypothesis to be readily graphed. Also, assume that the class structure can be represented
by six classes (e.g., professional, managerial, sales and clerical, craft, laborer, farm), signified in Figure
1.1 by six symbols (dark squares, light squares, dark circles, etc.). As shown in this figure, the two main
claims underlying the class hypothesis are that (1) the structural conditions of interest tend to cluster
together into characteristic packages, and (2) these packages of conditions correspond to occupational
or employment groupings. For structuralists, the inequality space is presumed to have a relatively
low dimensionality, which is precisely why the class formulation is attractive. The individuals falling
within the classes constituting this scheme will accordingly have endowments, working conditions,
and reward packages that are close to the averages prevailing for their classes. Moreover, even when
individual scores deviate from class averages, the conventional class-analytic assumption (albeit wholly
untested) is that the contextual effect of the class is dominant and overcomes any such individual-level
deviations. For example, a full professor who lacks a PhD is presumably just as marketable as a fully
credentialed (but otherwise comparable) full professor, precisely because membership in the professo-
rial class is a master status that tends to dominate all others.
The postwar period was just as notable for a flourishing of gradational measurement approaches
that again treated occupations as the fundamental units of analysis, but now assumed that such occu-
pations may be ordered into a unidimensional socioeconomic or prestige scale (e.g., Treiman, Chapter
12 • Introduction

Figure 1.1. Class Regime Figure 1.2. Gradational Regime

24; Goldthorpe and Hope, Chapter 25). In Figure 1.1, we assumed that the class structure cannot be
understood in simple gradational terms, meaning that at least some classes were formed by combining
high values on one dimension with low values on another. It is possible, however, that the structural
conditions of interest tend to covary linearly, thus generating a class structure of the very simple type
represented in Figure 1.2. In a regime of this sort, inequality becomes rather stark (i.e., crystallized),
as privilege on one dimension very reliably implies privilege on another. There should accordingly be
much interest in determining whether inequality indeed takes this form. Unfortunately, inequality
scholars of the postwar period did not typically test the crystallization assumption, but rather simply
assumed that it held and proceeded to develop socioeconomic scales that treated education and in-
come as the main dimensions of interest (and ranked occupations by averaging scores on these two
dimensions).
It may be noted that many neo-Marxian scholars during this period also deviated from a strict
multidimensional stance by nominating particular dimensions within Table 1.1 as theoretically crucial
and hence the appropriate basis upon which social classes might be defined. There are nearly as many
claims of this sort as there are dimensions in Table 1.1. To be sure, Marx is most commonly criticized
for placing “almost exclusive emphasis on economic factors as determinants of social class” (Lipset
1968, 300), but in fact much of what passed for stratification theorizing during this period amounted
to reductionism of some kind, albeit often an expanded version in which two or three dimensions
were nominated as especially crucial. When a reductionist position is adopted, the rationale for a
class model is not typically that classes are coherent packages of conditions (as represented in Figure
1.1), but rather that the nominated dimension or dimensions are crucial in defining interests and will
accordingly come to be the main sources of social action. The classic Marxian model, for example,
assumes that workers will ultimately appreciate that their status as workers (i.e., nonowners) defines
their interests.

Culturalist Phase (ca. 1985–1995)

In the mid-1980s Bourdieu (Chapter 114) and other sociologists (esp. Wilson, Chapter 89) sought
to develop a culturalist rationale for class models that rested on the claim that classes are not merely
constellations of structural conditions (e.g., working conditions, rewards), but are also socially closed
groupings in which distinctive cultures emerge and come to influence attitudes, behaviors, or even
preferences of class members. Throughout this period, many sociologists continued to work with more
 Grusky and Weisshaar—The Questions We Ask About Inequality • 13

narrowly structuralist definitions of class (e.g., Wright, Chapter 16; Erikson and Goldthorpe, Chapter
53), but Bourdieu (Chapter 114) and Wilson (Chapter 89) were instrumental in pressing the claim
that class-specific cultures are a defining feature of inequality systems (also, Lareau, Chapter 116).
The two main forms of closure that serve to generate class-specific cultures are residential segre-
gation (e.g., urban ghettos) and workplace segregation (e.g., occupational associations). As Wilson
notes, members of the underclass live in urban ghettos that are spatially isolated from mainstream
culture, thus allowing a distinctively oppositional culture to emerge and reproduce itself. The effects
of residential segregation for other social classes operate, by contrast, in more attenuated form; after all,
residential communities map only imperfectly onto class categories (e.g., the demise of the “company
town”), and social interaction within contemporary residential communities is in any event quite su-
perficial and cannot be counted upon to generate much in the way of meaningful culture. If distinctive
cultures emerge outside the underclass, they do so principally through the tendency of members of
the same occupation to interact disproportionately with one another in the workplace and through
leisure activities. In accounting, for example, for the humanist and left-leaning culture and lifestyle
of sociologists, class analysts would stress the forces of social closure within the workplace, especially
the liberalizing effects of (1) lengthy professional training and socialization into the “sociological
worldview” and (2) subsequent interaction in the workplace with predominantly liberal colleagues.
When occupations or classes are allowed to have cultures in this fashion, one naturally wishes to
understand how the content of those cultures is related to the structural conditions (i.e., endowments,
outcomes, institutional setting) of a class or occupational situation. At one extreme, class cultures may
be understood as nothing more than “rules of thumb” that encode optimizing behavioral responses
to prevailing institutional conditions, rules that allow class members to forgo optimizing calculations
themselves and rely instead on cultural prescriptions that provide reliable and economical shortcuts
to the right decision. For example, Breen and Goldthorpe (Chapter 60) argue that working-class
culture disparages educational investments not because of some maladaptive oppositional culture, but
because such investments expose the working class (more so than other classes) to a real risk of down-
ward mobility. In most cases, working-class children lack insurance in the form of substantial family
income or wealth, meaning that they cannot easily recover from an educational investment gone awry
(i.e., dropping out). If such an investment is nonetheless undertaken, there is a real danger of it going
awry and leading to substantial downward mobility. The emergence of a working-class culture that
regards educational investments as frivolous may be understood as encoding that conclusion and thus
allowing working-class children to engage in optimizing behaviors without explicitly making decision
tree calculations. The behaviors that a “rule of thumb” culture encourages are deeply adaptive because
they take into account the endowments and institutional realities encompassed by class situations (see
Morgan, Chapter 69).
The foregoing example may be understood as one in which a class-specific culture instructs re-
cipients about appropriate (i.e., optimizing) means for achieving ends that are widely pursued by all
classes. Indeed, the prior “rule-of-thumb” account assumes that members of the working class share
the conventional interest in maximizing labor market outcomes, with their class-specific culture merely
instructing them about the best approach for achieving that conventional objective. At the other
extreme, one finds class-analytic formulations that represent class cultures as deeper worldviews that
instruct not merely about the proper means to achieve ends, but also about the proper valuation of
the ends themselves. For example, some class cultures (e.g., aristocratic ones) place an especially high
valuation on leisure, with market work disparaged as “common” or “polluting” (see Veblen 2008). This
orientation presumably translates into a high reservation wage within the aristocratic class. Similarly,
“oppositional cultures” within the underclass may be understood as worldviews that place an espe-
cially high valuation on preserving respect and dignity for class members, with of course the further
14 • Introduction

prescription that these ends are best achieved by (1) withdrawing from and opposing conventional
mainstream pursuits; (2) representing conventional mobility mechanisms (e.g., higher education) as
tailor-made for the middle class and, by contrast, unworkable for the underclass; and (3) pursuing
dignity and respect through other means, most notably total withdrawal from and disparagement of
mainstream pursuits. This is a culture that advocates that respect and dignity deserve an especially
prominent place in the utility function and further specifies how those ends might be achieved (of
course under massive constraints).
It should be clear by now that sociologists operating within the class-analytic tradition make very
strong assumptions about how inequality and poverty are structured. The class concept embodies
such claims as that (1) the space of outcomes and capabilities has a (low) dimensionality equaling the
number of social classes, (2) the class locations of individuals become master statuses that dominate
(or at least supplement) the effects of individual-level endowments, and (3) such class locations are
socially closed and come to be associated with adaptive or maladaptive cultures. The prior claims have
been unstated articles of faith among class analysts in particular and sociologists more generally. In
this sense, class analysts have behaved rather like stereotypical economists, the latter frequently being
parodied for their willingness to assume almost anything provided that it leads to an elegant model.

Postmodernist Phase (ca. 1995–Present Day)

The third phase of conceptual work in sociology has been marked, however, by an increased willing-
ness to challenge the assumptions underlying the class-analytic status quo. In recent years, such criti-
cisms of the class-analytic enterprise have escalated, with many postmodernist scholars now sufficiently
emboldened to argue that the concept of class should be abandoned altogether. Although the post-
modern literature is notoriously fragmented, the variant of postmodernism that is most relevant here
proceeds from the assumption that the labor movement is rooted in the old and increasingly irrelevant
conflicts of industrial capitalism, that political parties have abandoned class-based platforms in favor of
those oriented toward values and lifestyles, and that class-based identities accordingly have become ever
weaker and more attenuated (see Hout and Laurison, Chapter 119). The resulting “individualization
of inequality” (e.g., see Pakulski and Waters 2008) implies that lifestyles and consumption practices are
becoming decoupled from work identities as well as other status group memberships. The stratification
system may be regarded, then, as a “status bazaar” (Pakulski and Waters 2008) in which identities are
actively constructed as individuals select and are shaped by their multiple statuses.6
This hypothesis, which is represented in extreme form by Figure 1.3, has not yet been subjected to
convincing empirical testing and may well prove to be premature (but see Jonsson et al., Chapter 55;
Weeden and Grusky, Chapter 23). Moreover, even if lifestyles and life chances are truly decoupling
from economic class, this should not be misunderstood as a more general decline in inequality per
se. The brute facts of inequality will still be with us even if social classes of the conventional form
are weakening. Most obviously, income inequality is clearly on the rise (e.g., Atkinson, Piketty, and
Saez, Chapter 7), while other forms of inequality show no signs of withering away. The postmodernist
hypothesis speaks, then, to the way in which inequality is organized as opposed to the overall amount
of such inequality.

Conclusions
We have sought to provide a brief overview of the raw materials of inequality, the basic concepts used
to analyze inequality, the types of questions that have occupied scholars of inequality and with which
our contributors are engaged, and the ways in which inequality is typically measured by sociologists
 Grusky and Weisshaar—The Questions We Ask About Inequality • 15

and others. We have stressed throughout that the Figure 1.3. Disorganized Inequality
nomothetic tradition has been quite successful and
that the knowledge base with which we are now op-
erating is accordingly substantial. This book may be
understood as a testimony to that extraordinary base.
Is there any theme running through the many
questions that sociologists of inequality have taken
on? The field is of course sprawling and the ques-
tions correspondingly diverse. If there is any theme
to them, it is simply that they problematize inequal-
ity by asking how it came about.
This problematizing orientation is distinctively
modern in its underpinnings. For the greater part of
human history, the stratification order was regarded
as an immutable feature of society, and commenta-
tors mainly sought to explain or justify this order
in terms of religious or quasi-religious doctrines (e.g., Bottomore 1965; Tawney 1931). It was only
with the Enlightenment that a critical rhetoric of equality emerged in opposition to the civil and
legal advantages of the aristocracy and other privileged status groupings. In its most radical form,
this egalitarianism led to Marxist interpretations of human history, and it ultimately provided the
intellectual underpinnings for socialist stratification systems. For the most part, stratification theory
has been formulated in reaction and opposition to these early forms of Marxist scholarship,7 yet the
field nonetheless shares with Marxism the distinctively Enlightenment premise that individuals are
to be equally valued and respected regardless of differences in their economic or social standing (see
Meyer, Chapter 129). In part, the questions laid out here reflect this tension between the brute facts
of extreme inequality and the modern commitment to at least some forms of equality, especially an
equality of opportunity and respect.

NOTES
1. In some stratification systems, the distribution of rewards can be described with a single matching algo-
rithm, because individuals receive rewards directly rather than by virtue of the social positions that they occupy.
The limiting case here would be the tribal economies of Melanesia in which “Big Men” (Oliver 1955) secured
prestige and power through personal influence rather than through incumbency of any well-defined roles (see
also Granovetter 1981, 12–14).
2. It goes without saying that the assets listed in Table 1.1 are institutionalized in very diverse ways. For
example, some assets are legally recognized by the state or by professional associations (e.g., civil rights, property
ownership, educational credentials), others are reserved for incumbents of specified work roles (e.g., workplace
authority), and yet others have no formal legal or institutional standing and are revealed probabilistically
through patterns of behavior and action (e.g., high-status consumption practices, deference, derogation).
3. It is sometimes claimed that educational credentials are entirely investment goods and should therefore be
excluded from any listing of the primitive dimensions underlying stratification systems (e.g., Runciman 1968,
33). In evaluating this claim, it is worth noting that an investment rhetoric for schooling became fashionable
only quite recently (e.g., Becker 1975), whereas intellectuals and humanists have long viewed education as a
simple consumption good.
4. This is not to gainsay the equally important point that parents often encourage their children to acquire
such goods because of their putative benefits.
5. Although “native ability” is by definition established at birth, it is sometimes seen as a legitimate basis
for allocating rewards (because it is presumed to be relevant to judgments of merit).
16 • Introduction

6. Although Pakulski and Waters (2008) use the label postmodern in their analyses, other scholars have
invented such alternative terms as late modernity, high modernity, or reflexive modernization (see Giddens,
Chapter 130). We frequently use the term postmodern in this introduction without intending to disadvantage
the analyses of those who prefer other labels.
7. In fact, the term stratification has itself been seen as anti-Marxist by some commentators (e.g., Duncan
1968), because it emphasizes the vertical ranking of classes rather than the exploitative relations between them.
The geological metaphor implied by this term does indeed call attention to issues of hierarchy. Nonetheless,
whenever it is used in the present introduction, the intention is to refer generically to inequality of all forms.

REFERENCES
Becker, Gary S. 1975. Human Capital. Chicago: University of Chicago Press.
Bottomore, Thomas B. 1965. Classes in Modern Society. London: Allen and Unwin.
Duncan, Otis Dudley. 1968. “Social Stratification and Mobility: Problems in the Measurement of Trend.”
In Indicators of Social Change, edited by Eleanor B. Sheldon and Wilbert E. Moore, 675–719. New York:
Russell Sage Foundation.
Freeman, Richard. 2012. “(Some) Inequality is Good for You.” In The New Gilded Age, edited by David B.
Grusky and Tamar Kricheli-Katz, 63–87. Stanford, CA: Stanford University Press.
Granovetter, Mark. 1981. “Toward a Sociological Theory of Income Differences.” In Sociological Perspectives
on Labor Markets, edited by Ivar Berg, 11–47. New York: Academic Press.
Grusky, David B. 2012. “What to Do About Inequality?” Boston Review (March/April). http://www.bostonre-
view.net/BR37.2/ndf_inequality.php.
Grusky, David B., and Kim A. Weeden. 2013. “Why Is There So Much Poverty?” In Occupy the Future, edited
by David B. Grusky, Douglas McAdam, Robert Reich, and Debra Satz. Cambridge: MIT Press.
Lipset, Seymour M. 1968. “Social Class.” In International Encyclopedia of the Social Sciences, edited by D. L.
Sills, 296–316. New York: Macmillan.
Okun, Arthur M. 1975. Equality and Efficiency: The Big Tradeoff. Washington, DC: Brookings Institution
Press.
Oliver, Douglas L. 1955. A Solomon Island Society. Cambridge, MA: Harvard University Press.
Pakulski, Jan, and Malcolm Waters. 2008. “The Death of Class.” In Social Stratification, Class, Race, and Gender
in Sociological Perspective, 3rd ed., edited by David B. Grusky, 1022–1029. Boulder, CO: Westview Press.
Parsons, Talcott. 1994. “Equality or Inequality in Modern Society, or Social Stratification Revisited.” In So-
cial Stratification, Class, Race, and Gender in Sociological Perspective, 1st ed., edited by David B. Grusky,
670–685. Boulder, CO: Westview Press.
Runciman, Walter G. 1968. “Class, Status, and Power?” In Social Stratification, edited by J. A. Jackson, 25–61.
New York: Cambridge University Press.
Schumpeter, Joseph. 1953. Aufsätze zur Soziologie. Tübingen: Mohr/Siebeck.
Tawney, R. H. 1931. Equality. London: George Allen and Unwin.
Veblen, Thorstein. 2008. “The Theory of the Leisure Class.” In Social Stratification, Class, Race, and Gender
in Sociological Perspective, 3rd ed., edited by David B. Grusky, 862–869. Boulder, CO: Westview Press.
 Grusky and Szelényi—The Stories We Tell About Inequality • 17

David B. Grusky and Szonja Szelényi

The Stories We Tell About Inequality

There is a growing consensus among academics, policy makers, and even politicians that poverty and
inequality should no longer be treated as “soft” social issues that can safely be subordinated to more
fundamental interests in maximizing total economic output. The most important sources of this ris-
ing concern with poverty and inequality are (1) the spectacular increase in economic inequality and
other forms of disadvantage in many late-industrial countries (the takeoff account); (2) the striking
persistence of many noneconomic forms of inequality despite decades of quite aggressive egalitarian
reform (the persistence account); (3) an emerging concern that poverty and inequality may have neg-
ative macrolevel effects on terrorism, total economic production, and ethnic unrest (the macrolevel
externalities account); (4) a growing awareness of the negative individual-level effects of poverty on
health, political participation, and a host of other life conditions (the microlevel externalities account);
(5) the rise of a “global village” in which regional disparities in the standard of living have become
more widely visible and hence increasingly difficult to ignore (the visibility account); (6) the ongoing
tendency to expose and delegitimate new types of inequalities (based on sexual orientation, disability,
or citizenship) that not long ago were taken for granted, rarely discussed, and barely seen (the new
inequalities account); (7) a confluence of recent events (e.g., hurricane Katrina, the Great Recession,
high pay for poorly performing CEOs) that have brought poverty and inequality into especially sharp
relief (the idiographic account); and (8) a growing commitment to a conception of human entitlements
that includes, at a minimum, the right to seek or secure employment and thereby be spared extreme
deprivation (the social inclusion account).
The foregoing list is remarkable in two ways. First, just three of the eight reasons for this increasing
interest are about the brute empirics of inequality (i.e., its growth or intransigence), and all the others
are about changes in how we have come to view, study, and evaluate those empirics. When scholars
now argue, for example, that inequality has multifarious effects (i.e., an externalities account), they pre-
sumably don’t mean to suggest that such effects have suddenly multiplied in the contemporary period.
Rather, we are to understand that inequality was always rife with externalities, however inadequately
we may have appreciated them in the past. Although changes in empirics hardly exhaust the sources
of our growing concern with inequality, this is not to gainsay the equally important point that such
changes, especially the takeoff in income inequality and the Great Recession, are probably a core reason
that inequality has come to be understood as a fundamental social problem of our time.
It is equally notable that normative concerns are not very prominent in our list. To be sure, there
appears to be a growing sentiment that, at a minimum, contemporary social systems should guarantee
all citizens the opportunity to participate in economic life and hence avoid the most extreme forms of
social and economic exclusion (i.e., the social inclusion account). It would nonetheless be a mistake to
understand the rising interest in poverty and inequality as primarily fueled by some sudden realization

Original article prepared for this fourth edition of Social Stratification: Class, Race, and Gender in Sociological
Perspective, edited by David B. Grusky. Copyright © 2014 Westview Press.
18 • Introduction

that social inclusion is a fundamental social good. Indeed, far from treating inequality as a moral prob-
lem in itself, the contemporary tendency is to emphasize its profound consequences and threats for the
world community as a whole (i.e., the macrolevel externalities account). The rhetoric of “sustainability,”
although more frequently featured in discussions of environmental problems, is increasingly consid-
ered relevant to discussions of inequality as well. In adopting this rhetoric, the claim is that extreme
inequality is counterproductive not just because it reduces total economic output, but also because
other very legitimate objectives, such as reducing mortality rates, might also be compromised. By this
logic, social policy must simultaneously be oriented to increasing economic output and restraining the
rise of debilitating and counterproductive forms of inequality, a rather more complicated maximization
problem than those conventionally taken on in economics (see Fischer et al., Chapter 4).

The Role of Benign Narratives in Past Scholarship

The foregoing orientation to poverty and inequality should be understood as a sea change relative to
the sensibilities that prevailed after World War II and even into the 1960s and 1970s. To be sure, the
standard-issue sociologist of the past also embraced the view that inequality was an important social prob-
lem, but overlaid on this sensibility was an appreciation of various “logics of history” that operated in the
main to reduce inequality, if only gradually and fitfully. Under such a formulation, inequality becomes
a tractable moral problem, indeed nothing more than an unfortunate side effect of capitalism (and even
socialism) that would become more manageable with the transition into the increasingly affluent forms
of advanced industrialism. This orientation to inequality, which we characterize as “benign,” is expressed
in the standard postwar narratives about three types of outcomes: (1) the distribution of income, power,
and other valued resources; (2) the distribution of opportunities for securing income, power, and other
valued resources; and (3) the formation of social classes and other institutionalized groups (e.g., racial
groups, gender groups). We review each of these types of benign narratives in the following discussion.

Long-Term Trend in Inequality

The dominant inequality narrative of the postwar period featured the emergence of egalitarian ideol-
ogies and the consequent delegitimation of the extreme forms of inequality found in agrarian systems
(Bell, Chapter 123; Meyer, Chapter 129). The Enlightenment was understood to foster a critical rhet-
oric of equality that unleashed one of the most profound revolutions in human history. The resulting
decline in inequality can be seen, for example, in (1) the European revolutions of the eighteenth and
nineteenth centuries against the privileges of rank (honorific equality); (2) the gradual elimination of
inequalities in the right to vote, own property, and speak and assemble (civil equality); (3) the abolition
of slavery and the establishment of the radically egalitarian principle of self-ownership (equality of
human assets); and (4) the equalization of economic assets via the rise of socialism, welfare capitalism,
and their many institutions (economic equality).1
As is well known, the commitment to equalizing economic assets was weaker than the commit-
ment to other forms of equalization, with the result that economic inequalities remained extreme in
all market economies. There was nonetheless a gradual decline in economic inequality throughout the
postwar period in the United States (Atkinson, Piketty, and Saez, Chapter 7). According to the classic
Kuznets curve (Kuznets 1955), the initial stages of capitalist development bring about an increase in
income inequality as capital is increasingly concentrated among a small number of investors, whereas
more advanced forms of capitalism entail some growth in the size of the middle class and a consequent
reversal of the upward trend. The causal dynamics behind the resulting inverted-U pattern remain
unclear (see Acemoglu and Robinson 2002), but most sociologists attribute the late-industrial decline
 Grusky and Szelényi—The Stories We Tell About Inequality • 19

in inequality to the increasingly crucial role that the skilled working class played in production, the
associated growth in working-class productivity, and the leverage that this growth in skills and pro-
ductivity gave skilled workers.
How does cultural egalitarianism of this sort diffuse and take hold? The conventional view is that
a series of crucial historical events after the Enlightenment (e.g., the defeat of Nazism, the civil rights
movement) defined equality as one of society’s core cultural commitments. Absent some revolutionary
event that changes this cultural trajectory, the course of human history then becomes the “working
out” of this commitment, a task that involves shedding subsidiary values that sometimes conflict with
our deeper commitment to egalitarianism. The core mechanism that drives this cultural diffusion may
therefore be understood as the gradual reconciling of competing values to a new value, that of equality,
that has been elevated by one or more historical events to a position of prominence.

Long-Term Trend in Inequality of Opportunity

The second narrative of interest rests on a sharp distinction between the distribution of social rewards
(e.g., income) and the distribution of opportunities for securing these rewards. In liberal welfare re-
gimes, extreme inequalities in rewards may be tolerated, but only insofar as opportunities for attaining
these rewards are understood to be equally distributed. The long-standing liberal formula thus treats
inequalities of opportunity as especially illegitimate (see Erikson and Goldthorpe, Chapter 53).
The dominant narratives of the postwar period portray these inequalities of opportunity as gradu-
ally weakening. We have characterized the narratives of this period as “benign” because they describe
the withering away of precisely those types of inequalities (i.e., inequalities of opportunity) that are
regarded as problematic or illegitimate. The trademark of the benign narrative is this simple corre-
spondence between what we want and what we think will happen. We describe here four benign
subnarratives that characterize some of the processes that weaken inequalities of opportunity.
The most famous such subnarrative addresses the discrimination-reducing effects of competitive
market economies. In his original formulation of the “taste for discrimination” model, Becker (1957)
argued that discrimination would be eroded by competitive market forces because it requires employ-
ers to pay a premium to hire members of the preferred class of labor, whether they be males, whites, or
any other ascriptively defined class. This taste is “discriminatory” because it rests on exogenous pref-
erences for a certain category of labor that cannot be understood as arising from some larger concern
for maximizing profitability or market share. When managers make hiring decisions in accord with
such tastes, their firms will not be competitive with nondiscriminating firms, because they must pay
extra to secure labor from the preferred class (without any compensating increase in productivity). In
standard renditions of this account, it is presumed that discriminating firms will gradually be selected
out by the market, although it is also possible that some discriminating firms will change their hiring
practices to remain competitive.
This economic subnarrative works in tandem with a second, “organizational” one that emphasizes
the diffusion of modern personnel practices in the form of universalistic hiring practices (e.g., open
hiring, credentialism) and bureaucratized pay scales and promotion procedures (see Weber, Chapter
18). The essence of such bureaucratic personnel practices is a formal commitment to universalism (i.e.,
treating all workers equally) and to meritocratic hiring and promotion (i.e., hiring and promoting on
the basis of credentials). In its ideal-typical form, the spread of bureaucracy is an organizational process
with its own dynamic, a process of diffusion that rests not on actual efficiencies, as in the economic
subnarrative, but simply on the presumption that bureaucratic practices are efficient and that “modern
firms” must therefore adopt them. This subnarrative, like the economic one, implies that firms will
gradually come to embrace organizational procedures that reduce inequalities of opportunity.
20 • Introduction

The third subnarrative features political forces. Whereas the economic and organizational subnarra-
tives treat change in inequality as an unintended by-product of macrolevel forces (i.e., competition and
bureaucratization), the political subnarrative is about instrumental action explicitly oriented toward
effecting a decline in inequality. In theory, such political action could be oriented toward reducing
inequalities of either opportunity or outcome, but historically a main emphasis within liberal welfare
regimes has been legislation aimed at reducing inequality of opportunity (e.g., antidiscrimination
legislation, early education programs, educational loans). The distinctive assumption of the political
subnarrative is that straightforward “social engineering” is an important source of change and the
unintended or unanticipated consequences of such engineering are too often overemphasized.
The final subnarrative, a simple cultural one, rests on the argument that Western ideals of justice
and equality continue to be endogenously worked out through a logic that diffuses independently of
the economic efficiency of such ideals. The cultural subnarrative can be straightforwardly distinguished
from the economic one because the growing “taste” for equality is presumed to be an exogenous shift
rather than some accommodation to the rising economic cost of exercising discriminatory tastes.
Likewise, the cultural subnarrative is distinct from the organizational subnarrative because it focuses
on the spread of tastes for equality and equality-enhancing practices, not the spread of organizational
forms (e.g., bureaucratization) that are deemed efficient, normatively desirable, or both. Finally, the
cultural and political subnarratives are closely related because political commitments to equal oppor-
tunity, antidiscrimination legislation, and school reform may be partly or even largely motivated by
these newfound tastes for equal opportunity. At the same time, the cultural commitment to equal
opportunity is not expressed exclusively in such political terms, but is also expressed in the attitudes,
behaviors, and personnel practices of employers. Most obviously, employers may gradually shed their
preferences for certain categories of labor and instead develop tastes for equality in hiring, firing, and
promotion, tastes that might at the limit be exercised in the labor market even with some loss in profits
or efficiency.
The spread of such tastes for equal opportunity may again be viewed as part of our Enlightenment
legacy, albeit a particular “liberal” variant that emphasizes equalizing opportunities, not outcomes.
This commitment is expressed not only at the individual level (e.g., changes in attitudes) but also at
the collective level through various types of political reform (e.g., antidiscrimination legislation) and
the diffusion of bureaucratic personnel policies (e.g., open hiring).

Long-Term Trend in Class Formation

The final benign narrative of interest describes the gradual transition from “lumpy” class-based labor
markets to more purely gradational ones (see Treiman, Chapter 24; Goldthorpe and Hope, Chapter
25; Chan and Goldthorpe, Chapter 115). Within this narrative, the early-industrial economy is rep-
resented as deeply balkanized into partly independent labor markets defined by detailed occupations
(e.g., economist, carpenter), big social classes (e.g., manager, farmer), or yet more aggregated factors of
production (e.g., worker, capitalist). For our purposes, what is principally of interest is our collective
fascination with arguments describing how these classes, however they may be defined, gradually dis-
sipate and leave us with gradational labor markets that increasingly approximate the seamless neoclas-
sical ideal. The first step in this transition, as described most famously by Dahrendorf (Chapter 15),
is the gradual “institutionalization” of class conflict, a regularization of labor-capital relations achieved
through the establishment of unions, collective bargaining agreements, and other laws defining how
labor and capital should negotiate (see also Parsons 1994). The second step in this transition involves
the dismantling of unions and other institutionalized residues of classes as the liberal welfare ideals of
“deregulation” and flexibility are increasingly pursued (see Rosenfeld, Chapter 73).
 Grusky and Szelényi—The Stories We Tell About Inequality • 21

This line of argument is somewhat differently expressed in more recent postmodernist narratives
(see Weeden and Grusky, Chapter 23). Although the postmodern literature is itself notoriously frag-
mented, most variants have proceeded from the assumption that class identities, ideologies, and orga-
nization are attenuating and that “new theories, perhaps more cultural than structural, [are] in order”
(Davis 1982, 585). The core claim is that politics, lifestyles, and consumption practices are no longer
class determined and increasingly become a “function of individual taste, choice, and commitment”
(Crook, Pakulski, and Waters 1992, 222; Pakulski 2005).
In more ambitious variants of postmodernism, the focus shifts away from simply claiming that
attitudes and practices are less class determined, and the older class-analytic objective of understand-
ing macrolevel stratificational change is resuscitated. This ambition underlies, for example, all forms
of postmodernism that seek to represent “new social movements” (e.g., environmentalism) as the
vanguard force behind future stratificatory change. As argued by Eyerman (1992) and others (e.g.,
Touraine 1981), the labor movement can be seen as a fading enterprise rooted in the old conflicts of
the workplace and industrial capitalism, whereas new social movements provide a more appealing call
for collective action by emphasizing issues of lifestyle, personal identity, and normative change. With
this formulation, the proletariat is stripped of its privileged status as a universal class, and new social
movements emerge as an alternative and far more benign force “shaping the future of modern societies”
(Haferkamp and Smelser 1992, 17).

New Approaches to Studying Inequality

The foregoing narratives, all of which were fixtures of the postwar intellectual landscape, describe
the emergence of a world in which inequalities are less profound, opportunities are more equally
distributed, and class conflicts and interclass differences become attenuated. These narratives are be-
nign in the sense that they push us toward equilibria that might appeal to most commentators, even
neo-Marxian ones. The benign narrative is accordingly built on the happy correspondence between
what should be and what will be.
If there is any theme to contemporary analyses of inequality, it is that the benign narrative has
largely fallen out of fashion. We have nevertheless laid out these narratives in some detail because they
(1) still motivate some contemporary research (e.g., Breen, Chapter 54), (2) often provide an impor-
tant backdrop to current theorizing, and (3) are sometimes used as foils by contemporary scholars
seeking to motivate their own analyses. The benign narrative is in this sense lurking in the background
of contemporary discussions of inequality. We turn now to a closer discussion of how contemporary
analyses of inequality have developed partly in reaction to these benign narratives.

Multidimensionalism and New Inequalities

As a natural starting point for this discussion, we note that contemporary inequality scholarship is
increasingly concerned with new forms of inequality that were ignored in the past, have been spawned
by new technologies, or have emerged within new institutions or sectors of the economy. This growing
emphasis on new inequalities is consistent with the now fashionable view that inequality is “multidi-
mensional” and that conventional studies of economic, socioeconomic, or cultural inequality do not
exhaust its many forms (e.g., Sen, Chapter 29).
If a multidimensional approach were pursued aggressively, it might feature the eight forms of in-
equality listed in Table 1.1 (p. 3), each form pertaining to a type of good that is intrinsically valuable
(as well as possibly an investment). The multidimensional space formed by these variables may be
labeled the “inequality space.” We can characterize the social location of an individual within this
22 • Introduction

inequality space by specifying her or his constellation of scores on each of the eight types of variables
in this table.
This framework allows scholars to examine how individuals are distributed among the less conven-
tionally studied dimensions of the inequality space. Are “new” assets as unequally distributed as old
ones? Is inequality becoming more of an “all or nothing” condition in which upper-class workers are
advantaged on all dimensions of interest and lower-class workers are disadvantaged on all dimensions
of interest? Are new assets sometimes distributed in ways that compensate for shortfalls in older ones?
In the present volume, multidimensionalist questions of this kind are posed for such “new” outcomes
as health (Chapters 120–122), imprisonment or capital punishment (Western and Pettit, Chapter 51),
and networks and social capital (Chapters 74–77). These new types of inequality may be understood
in some cases as truly new divides generated by new technologies (e.g., the digital divide) or new social
institutions (e.g., modern mass prisons). More typically, the “new” outcomes are merely increasingly
popular topics of study among academics, not truly new forms (e.g., health inequalities).

The Intransigence of Poverty and Inequality

The foregoing line of research typically takes the form of an exposé of the extent to which seemingly
basic human entitlements, such as living outside of prison, freely participating in “digital” culture, or
living a long and healthy life, are unequally distributed in ways that sometimes amplify well-known
differentials of income or education. The continuing attraction of such exposés (at least among aca-
demics) may be attributed to our collective discomfort with an economic system that generates more
inequality than is palatable under contemporary cultural standards. Although the equalizing reforms
of social democracy have historically been a main solution to this tension, their declining legitimacy
(especially in Europe and the United States) leaves the tension increasingly unresolved.
Whereas the old narratives focused on the forces contributing to a decline in inequality, a more pes-
simistic assessment of the trajectory of late industrialism has now taken hold, and much scholarship ac-
cordingly focuses on documenting that inequality has persisted at higher levels than were anticipated.
This sensibility underlies contemporary research showing that (1) residential segregation in the United
States is so extreme as to constitute a modern form of “apartheid” (Massey and Denton, Chapter
44); (2) racial discrimination in labor markets likewise remains extreme (Bertrand and Mullainathan,
Chapter 86; Pager, Chapter 88); (3) the occupational structure is “hypersegregated” by gender (Chap-
ters 100–104); (4) income inequality has increased markedly in many countries over the last thirty-five
years (Chapters 7–13); (5) poverty rates in the United States remain strikingly high (Danziger, Chavez,
and Cumberworth, Chapter 41); (6) African Americans are routinely harassed, slighted, and insulted
in public places (Omi and Winant, Chapter 79); (7) working-class and middle-class children are raised
in profoundly different ways (Lareau, Chapter 116); and (8) massive class disparities in access to health
services persist (Lutfey and Freese, Chapter 121).
It is worth asking whether this new muckraking tradition is really all that necessary. Is there truly
a large public that does not already appreciate the persistence of many of these inequalities? After all,
we live in a market society in which virtually everything is commodified, meaning that almost all
goods and services (e.g., health care, housing) are allocated on the basis of our ability to pay for them.
It might be argued that extreme inequality is the natural—and obvious—consequence of extreme
commodification and marketization.
Although such processes of commodification and marketization can of course work to increase the
reach of inequality, the muckraking tradition establishes that at least some of the inequality we observe
is attributable to additional institutional processes that layer in yet more inequality. It is possible to
imagine market societies that do not entail racial or gender discrimination of the magnitude that we
 Grusky and Szelényi—The Stories We Tell About Inequality • 23

currently experience. It is likewise possible to imagine market societies in which class differences in
politics, culture, and childrearing practices would have by now abated. The benign narratives of the
postwar period laid out precisely such imaginings. If it is now clear that these imagined futures have
not been realized, surely we need to document that conclusion with all the rigor that we can muster.
To be sure, many of us know well that the world is a massively unequal one, but even so the force
of the known can be readily lost when we live with profound inequality on a day-to-day basis. This
commitment to remind ourselves of what comes to be taken for granted is the cornerstone, we suspect,
of the renewed interest in the inequality-documenting function.

The Rise of New and Less Benign Narratives

The rise of this muckraking exposé of inequality has been coupled with increasing interest in devel-
oping narratives that explain why inequality has persisted or grown more extreme. These narratives
are typically less grand than the encompassing narratives of the postwar period; that is, specialized
narratives have recently developed around many of the various trends of interest (e.g., the expansion
of income inequality, the stalling decline in the gender wage gap), and little attention has been paid to
developing some grand metanarrative that links these specialized narratives together. The signature of
the contemporary narrative is this highly delimited focus, a commitment to developing a rigorously
empirical foundation, and a special interest in identifying those more insidious social forces that un-
dermine the benign stories of the past.
By way of example, consider the historic rise in income inequality, a development that has spawned
one of the most sustained efforts at narrative building of our time. As noted previously, the classic
Kuznets curve aligns nicely with the facts of inequality up to the early 1970s, but then a dramatic,
unprecedented upswing in inequality in the post-1970 period made it clear that inequality history had
not ended. We have since witnessed one of the most massive research efforts in the history of social
science as scholars sought to identify the “smoking gun” that accounted for this dramatic increase in
inequality (see Chapters 7–13).
Initially, the dominant hypothesis was that deindustrialization (i.e., the relocation of manufacturing
jobs to offshore labor markets) brought about a decline in demand for less-educated manufacturing
workers, a decline that generated increases in inequality by hollowing out the middle class and send-
ing manufacturing workers into unemployment or the ranks of poorly paid service work. Although
this argument still has its advocates, it cannot easily be reconciled with evidence suggesting that the
computerization of the workplace and related technological change has been at least one force behind
a heightened demand for highly educated workers. Because of this result (and other supporting evi-
dence), the deindustrialization story has now been largely supplanted by the converse hypothesis that
“skill-biased technological change” has increased the demand for high-skill workers beyond the in-
crease in supply, thus inducing a short-term disequilibrium and a correspondingly increased payoff for
high-skill labor. At the same time, most scholars acknowledge that this story is at best an incomplete
one and that other accounts, especially more narrowly political ones, must also be entertained (e.g.,
Hacker and Pierson, Chapter 11). Most notably, some of the rise in income inequality in the United
States was clearly attributable to the declining minimum wage (in real dollars), a development that in
turn has to be understood as the outcome of political processes that increasingly favored pro-inequality
forces. The effects of politics are even more obviously revealed in the regressive tax policy of the Bush
presidency and the progressive tax policy of the Obama presidency.
The future of income inequality depends on which of these underlying mechanisms is principally
at work. The silver lining of the deindustrialization story is that within-country increases in inequal-
ity should be offset by between-country declines (as poor countries profit from new manufacturing
24 • Introduction

jobs), whereas the silver lining of skill-biased technological change is that the heightened demand
for high-skill workers is presumably a one-time, short-term disequilibrium that will, by virtue of the
higher payoff for high-skill jobs, trigger a compensating growth in the supply of high-skill workers.
Unfortunately there is no shortage of competing stories that imply more disturbing futures, even some
consistent with a classical Marxian account in which wages for low-skill workers are reduced because
of a globalization-induced “race to the bottom.”
We have focused on the rise of income inequality and the various narratives it has generated only
because this literature is especially well known and central to the field. The larger point that we seek
to make is that, no matter the subfield, there appears to be much interest in developing narratives
to explain why long-standing declines in inequality have slowed down, stalled altogether, or even re-
versed themselves. We are referring, for example, to (1) globalization narratives that describe how the
liberalization of financial and capital markets has harmed poor countries (Stiglitz, Chapter 131); (2)
deindustrialization narratives that describe the loss of inner-city jobs and the associated rise of an urban
underclass (Wilson, Chapter 89); (3) “segmented assimilation” narratives that describe the relatively
bleak prospects for at least some new immigrant groups (Portes and Zhou, Chapter 84); (4) “opting
out” narratives that describe highly trained women eschewing stressful careers in favor of recommitting
to their children, spouses, and domestic responsibilities (Belkin, Chapter 93; Percheski, Chapter 94);
and (5) “essentialist segregation” narratives that describe how sex-typed occupational ghettos continue
to be built around presumed differences in male and female aptitudes (Charles and Grusky, Chapter
104). Although counternarratives of the more optimistic sort are also being developed (e.g., Alba and
Nee, Chapter 83; Firebaugh, Chapter 132; Breen, Chapter 54), these seem not to be as frequently
generated or as readily embraced, and their proponents find themselves beleaguered, outnumbered,
and on the defensive.

The Future of Narratives About Inequality

The second part of our introductory comments has focused on revolutionary changes in the narratives
that sociologists and other social scientists have applied over the past half century to make sense of
trends in inequality. The postwar narratives were strikingly benign in the sense that they treated the
dominant logics of history as operating mainly to reduce inequality. These benign narratives, which
now seem naive and quaint, have been supplanted by a host of new narratives that give far greater
weight to the forces making for inequality.
Has the pendulum swung too far? It is of course child’s play to posit any number of nonempirical
sources of our fascination, some might say obsession, with the pessimistic narrative. It is useful to lay
out a few of these forces that could be leading us to be unduly pessimistic.
The newsworthiness account. If the postwar era of the benign narrative was in fact characterized by
across-the-board declines in inequality, then a special incentive presumably emerged to ferret out
results that were inconsistent with the prevailing trends and hence would be regarded as news-
worthy. It is surely difficult to market one’s analyses and to forge a celebrated career on the basis
of business-as-usual evidence. We might ask, for example, whether the “digital divide” emerged as
a newsworthy topic precisely because the early evidence revealed a substantial divide. If access to
computing had instead been found to be equal, would we have quickly discarded the topic and set
out to find some other, more unequal outcome? This type of selection on the dependent variable
(i.e., inequality) will create a research literature that exaggerates how unequal the world is.
The “moral credentials” account. We should bear in mind that the contemporary academic, far from
taking a vow of poverty, is now firmly ensconced in the middle class, often the upper middle
class. When relatively privileged scholars study poverty and disadvantage, they often feel a special
 Grusky and Szelényi—The Stories We Tell About Inequality • 25

obligation to demonstrate a strong commitment to amelioration and to display liberal, progres-

sive, or even radical political sentiments. In some cases this pressure may motivate them to down-
play any evidence of decline in inequality or disadvantage, presumably out of concern that undue
emphasis on the progress achieved so far will make it appear that the remaining disadvantage is
acceptable or unproblematic.
The “obsession with small differences” account. The continuing diffusion of egalitarian values renders
any departures from equality, no matter how small, problematic and newsworthy. By this logic,
even increasingly small intergroup differences will attract much attention, especially because ever
more powerful models and statistical methods now make it possible to tease them out (see Nisbet
1959, 12).

If these nonempirical interpretations should not be dismissed out of hand, they also should not be
taken too seriously. It is relevant that many of the pessimistic narratives involve indisputably dramatic
changes in outcomes (e.g., income) that were frequently studied well before any reversal in the trend
line was detected. Although some of the pessimistic narratives featured in this volume are still in in-
cipient form and are not yet well researched, it will of course be difficult to continue to maintain these
narratives should strong contradictory evidence turn up. The implication, then, is that our exaggerated
taste for pessimism might conceivably lead us to cycle through a great many pessimistic stories (and fail
to develop enough benign ones), but at least the usual rules of evidence will have us excise egregiously
flawed narratives, no matter how benign or pessimistic they may be.

NOTES
1. This is not to suggest that all postwar sociologists and social scientists emphasized forces causing a decline
in inequality. However, even when a benign narrative was not adopted, there was usually some effort to engage
with it and to explain or defend the decision not to take it up. This defensiveness was especially apparent in
neo-Marxian analyses of the postwar era. Although such analyses were based on deeply pessimistic subnarra-
tives about the trajectory of capitalism, these subnarratives were typically attached to larger and more benign
narratives about the postcapitalist trajectory (see Marx, Chapter 14; Wright, Chapter 16).

REFERENCES
Acemoglu, Daron, and James A. Robinson. 2002. “The Political Economy of the Kuznets Curve.” Review of
Development Economics 6: 183–203.
Becker, Gary S. 1957. The Economics of Discrimination. Chicago: University of Chicago Press.
Crook, Stephen, Jan Pakulski, and Malcolm Waters. 1992. Postmodernization. London: Sage Publications.
Davis, James. 1982. “Achievement Variables and Class Cultures: Family, Schooling, Job, and Forty-nine De-
pendent Variables in the Cumulative GSS.” American Sociological Review 47: 569–586.
Eyerman, Ron. 1992. “Modernity and Social Movements.” In Social Change and Modernity, edited by Hans
Haferkamp and Neil J. Smelser. Berkeley: University of California Press.
Haferkamp, Hans, and Neil J. Smelser. 1992. “Introduction.” In Social Change and Modernity, edited by Hans
Haferkamp and Neil J. Smelser. Berkeley: University of California Press.
Kuznets, Simon. 1955. “Economic Growth and Income Inequality (Presidential Address).” American Economic
Review 45: 1–28.
Nisbet, Robert A. 1959. “The Decline and Fall of Social Class.” Pacific Sociological Review 2: 11–17.
Pakulski, J. 2005. “Foundations of a Post-Class Analysis.” In Approaches to Class Analysis, edited by E. Wright.
Cambridge, UK: Cambridge University Press.
Parsons, Talcott. 1994. “Equality or Inequality in Modern Society, or Social Stratification Revisited.” In So-
cial Stratification, Class, Race, and Gender in Sociological Perspective, 1st ed., edited by David B. Grusky,
670–685. Boulder, CO: Westview Press.
Touraine, Alain. 1981. The Voice and the Eye: An Analysis of Social Movements. Cambridge, UK: Cambridge
University Press.
 PART I

The Functions and

Dysfunctions of Inequality
2 Some Principles of Stratification 28
3 Some Principles of Stratification: A Critical Analysis 31
4 Inequality by Design 39
28 • Part I: The Functions and Dysfunctions of Inequality

2 • Kingsley Davis and Wilbert E. Moore

Some Principles of Stratikcation

In a previous paper some concepts for handling placing and motivating individuals in the social
the phenomena of social inequality were pre- structure. As a functioning mechanism a society
sented.1 In the present paper a further step in must somehow distribute its members in social
stratification theory is undertaken—an attempt positions and induce them to perform the duties
to show the relationship between stratification of these positions. It must thus concern itself
and the rest of the social order.2 Starting from with motivation at two different levels: to instill
the proposition that no society is “classless,” or in the proper individuals the desire to fill cer-
unstratified, an effort is made to explain, in func- tain positions, and, once in these positions, the
tional terms, the universal necessity which calls desire to perform the duties attached to them.
forth stratification in any social system. Even though the social order may be relatively
Throughout, it will be necessary to keep in static in form, there is a continuous process of
mind one thing—namely, that the discussion metabolism as new individuals are born into
relates to the system of positions, not to the it, shift with age, and die off. Their absorption
individuals occupying those positions. It is one into the positional system must somehow be
thing to ask why different positions carry differ- arranged and motivated. This is true whether
ent degrees of prestige, and quite another to ask the system is competitive or non-competitive.
how certain individuals get into those positions. A competitive system gives greater importance
Although, as the argument will try to show, both to the motivation to achieve positions, whereas
questions are related, it is essential to keep them a non-competitive system gives perhaps greater
separate in our thinking. importance to the motivation to perform the
Most of the literature on stratification has duties of the positions; but in any system both
tried to answer the second question (particularly types of motivation are required.
with regard to the ease or difficulty of mobility If the duties associated with the various po-
between strata) without tackling the first. The sitions were all equally pleasant to the human
first question, however, is logically prior and, in organism, all equally important to societal sur-
the case of any particular individual or group, vival, and all equally in need of the same ability
factually prior. or talent, it would make no difference who got
into which positions, and the problem of social
placement would be greatly reduced. But actu-
The Functional Necessity of ally it does make a great deal of difference who
Stratification gets into which positions, not only because some
Curiously the main functional necessity explain- positions are inherently more agreeable than
ing the universal presence of stratification is pre- others, but also because some require special tal-
cisely the requirement faced by any society of ents or training and some are functionally more

Kingsley Davis and Wilbert E. Moore. “Some Principles of Stratification,” American Sociological Review 10
(April 1945), pp. 242–249. Published by SAGE in 1945.
 Davis and Moore—Some Principles of Stratikcation • 29

important than others. Also, it is essential that and must therefore possess a certain amount of
the duties of the positions be performed with the institutionalized inequality.
diligence that their importance requires. Inevi- It does not follow that the amount or type
tably, then, a society must have, first, some kind of inequality need be the same in all societies.
of rewards that it can use as inducements, and, This is largely a function of factors that will be
second, some way of distributing these rewards discussed presently.
differentially according to positions. The rewards
and their distribution become a part of the social
order, and thus give rise to stratification. The Two Determinants of
One may ask what kind of rewards a society Positional Rank
has at its disposal in distributing its personnel Granting the general function that inequality
and securing essential services. It has, first of subserves, one can specify the two factors that
all, the things that contribute to sustenance and determine the relative rank of different posi-
comfort. It has, second, the things that contrib- tions. In general those positions convey the best
ute to humor and diversion. And it has, finally, reward, and hence have the highest rank, which
the things that contribute to self-respect and ego (a) have the greatest importance for the society
expansion. The last, because of the peculiarly so- and (b) require the greatest training or talent.
cial character of the self, is largely a function of The first factor concerns function and is a mat-
the opinion of others, but it nonetheless ranks in ter of relative significance; the second concerns
importance with the first two. In any social sys- means and is a matter of scarcity.
tem all three kinds of rewards must be dispensed
differentially according to positions. Differential Functional Importance. Actually
In a sense the rewards are “built into” the a society does not need to reward positions in
position. They consist in the “rights” associ- proportion to their functional importance. It
ated with the position, plus what may be called merely needs to give sufficient reward to them
its accompaniments or perquisites. Often the to insure that they will be filled competently. In
rights, and sometimes the accompaniments, are other words, it must see that less essential po-
functionally related to the duties of the position. sitions do not compete successfully with more
(Rights as viewed by the incumbent are usually essential ones. If a position is easily filled, it need
duties as viewed by other members of the com- not be heavily rewarded, even though important.
munity.) However, there may be a host of subsid- On the other hand, if it is important but hard
iary rights and perquisites that are not essential to fill, the reward must be high enough to get
to the function of the position and have only an it filled anyway. Functional importance is there-
indirect and symbolic connection with its duties, fore a necessary but not a sufficient cause of high
but which still may be of considerable impor- rank being assigned to a position.3
tance in inducing people to seek the positions
and fulfill the essential duties. Differential Scarcity of Personnel. Practically all
If the rights and perquisites of different po- positions, no matter how acquired, require some
sitions in a society must be unequal, then the form of skill or capacity for performance. This
society must be stratified, because that is pre- is implicit in the very notion of position, which
cisely what stratification means. Social inequal- implies that the incumbent must, by virtue of his
ity is thus an unconsciously evolved device by incumbency, accomplish certain things.
which societies insure that the most important There are, ultimately, only two ways in which
positions are conscientiously filled by the most a person’s qualifications come about: through in-
qualified persons. Hence every society, no mat- herent capacity or through training. Obviously,
ter how simple or complex, must differentiate in concrete activities both are always necessary,
persons in terms of both prestige and esteem, but from a practical standpoint the scarcity may
30 • Part I: The Functions and Dysfunctions of Inequality

lie primarily in one or the other, as well as in the external situation, may wholly obviate the
both. Some positions require innate talents of necessity of certain kinds of skill or talent. Any
such high degree that the persons who fill them particular system of stratification, then, can be
are bound to be rare. In many cases, however, understood as a product of the special conditions
talent is fairly abundant in the population but affecting the two aforementioned grounds of dif-
the training process is so long, costly, and elabo- ferential reward.
rate that relatively few can qualify. Modern med-
icine, for example, is within the mental capacity NOTES
of most individuals, but a medical education is 1. Kingsley Davis, “A Conceptual Analysis of
Stratification,” American Sociological Review. 7:309–
so burdensome and expensive that virtually none
321, June, 1942.
would undertake it if the position of the M.D.
2. The writers regret (and beg indulgence) that
did not carry a reward commensurate with the the present essay, a condensation of a longer study,
sacrifice. covers so much in such short space that adequate
If the talents required for a position are abun- evidence and qualification cannot be given and that
dant and the training easy, the method of ac- as a result what is actually very tentative is presented
quiring the position may have little to do with in an unfortunately dogmatic manner.
its duties. There may be, in fact, a virtually ac- 3. Unfortunately, functional importance is diffi-
cidental relationship. But if the skills required cult to establish. To use the position’s prestige to es-
are scarce by reason of the rarity of talent or the tablish it, as is often unconsciously done, constitutes
circular reasoning from our point of view. There are,
costliness of training, the position, if function-
however, two independent clues: (a) the degree to
ally important, must have an attractive power which a position is functionally unique, there being
that will draw the necessary skills in competition no other positions that can perform the same func-
with other positions. This means, in effect, that tion satisfactorily; (b) the degree to which other po-
the position must be high in the social scale— sitions are dependent on the one in question. Both
must command great prestige, high salary, ample clues are best exemplified in organized systems of
leisure, and the like. positions built around one major function. Thus,
in most complex societies the religious, political,
How Variations Are to Be Understood. In so far economic, and educational functions are handled by
distinct structures not easily interchangeable. In ad-
as there is a difference between one system of
dition, each structure possesses many different posi-
stratification and another, it is attributable to
tions, some clearly dependent on, if not subordinate
whatever factors affect the two determinants to, others. In sum, when an institutional nucleus
of differential reward—namely, functional im- becomes differentiated around one main function,
portance and scarcity of personnel. Positions and at the same time organizes a large portion of the
important in one society may not be important population into its relationships, the key positions
in another, because the conditions faced by the in it are of the highest functional importance. The
societies, or their degree of internal develop- absence of such specialization does not prove func-
ment, may be different. The same conditions, tional unimportance, for the whole society may be
relatively unspecialized; but it is safe to assume that
in turn, may affect the question of scarcity; for
the more important functions receive the first and
in some societies the stage of development, or
clearest structural differentiation.
 Tumin—Some Principles of Stratikcation • 31

3 • Melvin M. Tumin

Some Principles of Stratikcation

A Critical Analysis

The fact of social inequality in human society is 1. Certain positions in any society are func-
marked by its ubiquity and its antiquity. Every tionally more important than others, and
known society, past and present, distributes its require special skills for their performance.
scarce and demanded goods and services un- 2. Only a limited number of individuals in
equally. And there are attached to the positions any society have the talents which can be
which command unequal amounts of such trained into the skills appropriate to these
goods and services certain highly morally toned positions.
evaluations of their importance for the society. 3. The conversion of talents into skills involves
The ubiquity and the antiquity of such in- a training period during which sacrifices of
equality has given rise to the assumption that one kind or another are made by those un-
there must be something both inevitable and pos- dergoing the training.
itively functional about such social arrangements. 4. In order to induce the talented persons to
Clearly, the truth or falsity of such an as- undergo these sacrifices and acquire the
sumption is a strategic question for any general training, their future positions must carry
theory of social organization. It is therefore most an inducement value in the form of differ-
curious that the basic premises and implications ential, i.e., privileged and disproportionate
of the assumption have only been most casually access to the scarce and desired rewards
explored by American sociologists. which the society has to offer.2
The most systematic treatment is to be found 5. These scarce and desired goods consist of
in the well-known article by Kingsley Davis and the rights and perquisites attached to, or
Wilbert Moore, entitled “Some Principles of built into, the positions, and can be classi-
Stratification.”1 More than twelve years have fied into those things which contribute to
passed since its publication, and though it is one (a) sustenance and comfort, (b) humor and
of the very few treatments of stratification on a diversion, (c) self-respect and ego expansion.
high level of generalization, it is difficult to lo- 6. This differential access to the basic rewards
cate a single systematic analysis of its reasoning. of the society has as a consequence the
It will be the principal concern of this paper to differentiation of the prestige and esteem
present the beginnings of such an analysis. which various strata acquire. This may be
The central argument advanced by Davis and said, along with the rights and perquisites,
Moore can be stated in a number of sequential to constitute institutionalized social in-
propositions, as follows: equality, i.e., stratification.

Melvin M. Tumin. “Some Principles of Stratification: A Critical Analysis,” American Sociological Review 18
(August 1953), pp. 387–394. Published by SAGE in 1953.
32 • Part I: The Functions and Dysfunctions of Inequality

7. Therefore, social inequality among different of choice with infinite time dimensions. For
strata in the amounts of scarce and desired at some point along the line one must face the
goods, and the amounts of prestige and es- problem of adequate motivation for all workers
teem which they receive, is both positively at all levels of skill in the factory. In the long
functional and inevitable in any society. run, some labor force of unskilled workmen is
as important and as indispensable to the factory
Let us take these propositions and examine as some labor force of engineers. Often enough,
them seriatim.3 the labor force situation is such that this fact is
brought home sharply to the entrepreneur in the
(1) Certain positions in any society are more func- short run rather than in the long run.
tionally important than others, and require special Moreover, the judgment as to the relative in-
skills for their performance. dispensability and replaceability of a particular
The key term here is “functionally impor- segment of skills in the population involves a
tant.” The functionalist theory of social organi- prior judgment about the bargaining-power of
zation is by no means clear and explicit about that segment. But this power is itself a cultur-
this term. The minimum common referent is ally shaped consequence of the existing system of
to something known as the “survival value” of a rating, rather than something inevitable in the
social structure.4 This concept immediately in- nature of social organization. At least the con-
volves a number of perplexing questions. Among trary of this has never been demonstrated, but
these are: (a) the issue of minimum vs. maximum only assumed.
survival, and the possible empirical referents A generalized theory of social stratification
which can be given to those terms; (b) whether must recognize that the prevailing system of
such a proposition is a useless tautology since any inducements and rewards is only one of many
status quo at any given moment is nothing more variants in the whole range of possible systems
and nothing less than everything present in the of motivation which, at least theoretically, are
status quo. In these terms, all acts and structures capable of working in human society. It is quite
must be judged positively functional in that they conceivable, of course, that a system of norms
constitute essential portions of the status quo; could be institutionalized in which the idea of
(c) what kind of calculus of functionality exists threatened withdrawal of services, except under
which will enable us, at this point in our develop- the most extreme circumstances, would be con-
ment, to add and subtract long- and short-range sidered as absolute moral anathema. In such a
consequences, with their mixed qualities, and case, the whole notion of relative functionality,
arrive at some summative judgment regarding as advanced by Davis and Moore, would have to
the rating an act or structure should receive on be radically revised.
a scale of greater or lesser functionality? At best,
we tend to make primarily intuitive judgments. (2) Only a limited number of individuals in any
Often enough, these judgments involve the use society have the talents which can be trained into
of value-laden criteria, or, at least, criteria which the skills appropriate to these positions (i.e., the
are chosen in preference to others not for any more functionally important positions).
sociologically systematic reasons but by reason of The truth of this proposition depends at least
certain implicit value preferences. in part on the truth of proposition 1 above. It is,
Thus, to judge that the engineers in a factory therefore, subject to all the limitations indicated
are functionally more important to the factory above. But for the moment, let us assume the va-
than the unskilled workmen involves a notion lidity of the first proposition and concentrate on
regarding the dispensability of the unskilled the question of the rarity of appropriate talent.
workmen, or their replaceability, relative to If all that is meant is that in every society
that of the engineers. But this is not a process there is a range of talent, and that some members
 Tumin—Some Principles of Stratikcation • 33

of any society are by nature more talented than restrict further access to their privileged posi-
others, no sensible contradiction can be offered, tions, once they have sufficient power to enforce
but a question must be raised here regarding such restrictions. This is especially true in a cul-
the amount of sound knowledge present in any ture where it is possible for an elite to contrive
society concerning the presence of talent in the a high demand and a proportionately higher
population. reward for its work by restricting the numbers
For in every society there is some demon- of the elite available to do the work. The recruit-
strable ignorance regarding the amount of talent ment and training of doctors in the modern
present in the population. And the more rigidly United States is at least partly a case in point.
stratified a society is, the less chance does that society Here, then, are three ways, among others
have of discovering any new facts about the talents of which could be cited, in which stratification
its members. Smoothly working and stable systems systems, once operative, tend to reduce the sur-
of stratification, wherever found, tend to build-in vival value of a society by limiting the search,
obstacles to the further exploration of the range recruitment and training of functionally impor-
of available talent. This is especially true in those tant personnel far more sharply than the facts of
societies where the opportunity to discover talent available talent would appear to justify. It is only
in any one generation varies with the differential when there is genuinely equal access to recruit-
resources of the parent generation. Where, for ment and training for all potentially talented
instance, access to education depends upon the persons that differential rewards can conceivably
wealth of one’s parents, and where wealth is dif- be justified as functional. And stratification sys-
ferentially distributed, large segments of the pop- tems are apparently inherently antagonistic to the
ulation are likely to be deprived of the chance development of such full equality of opportunity.
even to discover what are their talents.
Whether or not differential rewards and op- (3) The conversion of talents into skills involves
portunities are functional in any one generation, a training period during which sacrifices of one
it is clear that if those differentials are allowed kind or another are made by those undergoing the
to be socially inherited by the next generation, training.
then the stratification system is specifically dys- Davis and Moore introduce here a concept,
functional for the discovery of talents in the next “sacrifice,” which comes closer than any of the
generation. In this fashion, systems of social rest of their vocabulary of analysis to being a di-
stratification tend to limit the chances available rect reflection of the rationalizations, offered by
to maximize the efficiency of discovery, recruit- the more fortunate members of a society, of the
ment and training of “functionally important rightness of their occupancy of privileged posi-
talent.”5 tions. It is the least critically thought-out con-
Additionally, the unequal distribution of re- cept in the repertoire, and can also be shown to
wards in one generation tends to result in the be least supported by the actual facts.
unequal distribution of motivation in the suc- In our present society, for example, what are
ceeding generation. Since motivation to succeed the sacrifices which talented persons undergo in
is clearly an important element in the entire the training period? The possibly serious losses
process of education, the unequal distribution involve the surrender of earning power and the
of motivation tends to set limits on the possi- cost of the training. The latter is generally borne
ble extensions of the educational system, and by the parents of the talented youth undergoing
hence, upon the efficient recruitment and train- training, and not by the trainees themselves. But
ing of the widest body of skills available in the this cost tends to be paid out of income which
population.6 the parents were able to earn generally by virtue
Lastly, in this context, it may be asserted that of their privileged positions in the hierarchy of
there is some noticeable tendency for elites to stratification. That is to say, the parents’ ability
34 • Part I: The Functions and Dysfunctions of Inequality

to pay for the training of their children is part of of the trainee. Even assuming, for the moment,
the differential reward they, the parents, received that there is a difference, the amount is by no
for their privileged positions in the society. And means sufficient to justify a lifetime of continu-
to charge this sum up against sacrifices made by ing differentials.
the youth is falsely to perpetrate a bill or a debt What tends to be completely overlooked,
already paid by the society to the parents. in addition, are the psychic and spiritual re-
So far as the sacrifice of earning power by the wards which are available to the elite train-
trainees themselves is concerned, the loss may ees by comparison with their age peers in the
be measured relative to what they might have labor force. There is, first, the much higher
earned had they gone into the labor market prestige enjoyed by the college student and the
instead of into advanced training for the “im- professional-school student as compared with
portant” skills. There are several ways to judge persons in shops and offices. There is, second,
this. One way is to take all the average earnings the extremely highly valued privilege of having
of age peers who did go into the labor market greater opportunity for self-development. There
for a period equal to the average length of the is, third, all the psychic gain involved in being al-
training period. The total income, so calculated, lowed to delay the assumption of adult responsi-
roughly equals an amount which the elite can, bilities such as earning a living and supporting a
on the average, earn back in the first decade of family. There is, fourth, the access to leisure and
professional work, over and above the earnings freedom of a kind not likely to be experienced by
of his age peers who are not trained. Ten years the persons already at work.
is probably the maximum amount needed to If these are never taken into account as re-
equalize the differential.7 There remains, on the wards of the training period it is not because
average, twenty years of work during each of they are not concretely present, but because the
which the skilled person then goes on to earn emphasis in American concepts of reward is al-
far more than his unskilled age peers. And, what most exclusively placed on the material returns
is often forgotten, there is then still another ten- of positions. The emphases on enjoyment, en-
or fifteen-year period during which the skilled tertainment, ego enhancement, prestige and es-
person continues to work and earn when his un- teem are introduced only when the differentials
skilled age peer is either totally or partially out of in these which accrue to the skilled positions
the labor market by virtue of the attrition of his need to be justified. If these other rewards were
strength and capabilities. taken into account, it would be much more dif-
One might say that the first ten years of dif- ficult to demonstrate that the training period, as
ferential pay is perhaps justified, in order to re- presently operative, is really sacrificial. Indeed,
gain for the trained person what he lost during it might turn out to be the case that even at
his training period. But it is difficult to imagine this point in their careers, the elite trainees were
what would justify continuing such differential being differentially rewarded relative to their age
rewards beyond that period. peers in the labor force.
Another and probably sounder way to mea- All of the foregoing concerns the quality of
sure how much is lost during the training period the training period under our present system
is to compare the per capita income available to of motivation and rewards. Whatever may turn
the trainee with the per capita income of the age out to be the factual case about the present sys-
peer on the untrained labor market during the tem—and the factual case is moot—the more
so-called sacrificial period. If one takes into ac- important theoretical question concerns the
count the earlier marriage of untrained persons, assumption that the training period under any
and the earlier acquisition of family dependents, system must be sacrificial.
it is highly dubious that the per capita income of There seem to be no good theoretical grounds
the wage worker is significantly larger than that for insisting on this assumption. For while under
 Tumin—Some Principles of Stratikcation • 35

any system certain costs will be involved in job satisfaction” and be most readily identifiable
training persons for skilled positions, these costs as socially serviceable. Is it indeed impossible
could easily be assumed by the society-at-large. then to build these motivations into the social-
Under these circumstances, there would be no ization pattern to which we expose our talented
need to compensate anyone in terms of differ- youth?
ential rewards once the skilled positions were To deny that such motivations could be in-
staffed. In short, there would be no need or jus- stitutionalized would be to overclaim our pres-
tification for stratifying social positions on these ent knowledge. In part, also, such a claim would
grounds. seem to derive from an assumption that what has
not been institutionalized yet in human affairs
(4) In order to induce the talented persons to un- is incapable of institutionalization. Admittedly,
dergo these sacrifices and acquire the training, their historical experience affords us evidence we can-
future positions must carry an inducement value in not afford to ignore. But such evidence cannot
the form of differential, i.e., privileged and dispro- legitimately be used to deny absolutely the pos-
portionate access to the scarce and desired rewards sibility of heretofore untried alternatives. Social
which the society has to offer. innovation is as important a feature of human
Let us assume, for the purposes of the discus- societies as social stability.
sion, that the training period is sacrificial and the On the basis of these observations, it seems
talent is rare in every conceivable human society. that Davis and Moore have stated the case much
There is still the basic problem as to whether the too strongly when they insist that a “functionally
allocation of differential rewards in scarce and important position” which requires skills that are
desired goods and services is the only or the most scarce, “must command great prestige, high sal-
efficient way of recruiting the appropriate talent ary, ample leisure, and the like,” if the appro-
to these positions. priate talents are to be attracted to the position.
For there are a number of alternative moti- Here, clearly, the authors are postulating the un-
vational schemes whose efficiency and adequacy avoidability of very specific types of rewards and,
ought at least to be considered in this context. by implication, denying the possibility of others.
What can be said, for instance, on behalf of the
motivation which De Man called “joy in work,” (5) These scarce and desired goods consist of the
Veblen termed “instinct for workmanship” and rights and perquisites attached to, or built into,
which we latterly have come to identify as “in- the positions, and can be classified into those things
trinsic work satisfaction”? Or, to what extent which contribute to (a) sustenance and comfort,
could the motivation of “social duty” be insti- (b) humor and diversion, (c) self-respect and ego
tutionalized in such a fashion that self-interest expansion.
and social interest come closely to coincide?
Or, how much prospective confidence can be (6) This differential access to the basic rewards of
placed in the possibilities of institutionalizing the society has as a consequence the differentiation
“social service” as a widespread motivation for of the prestige and esteem which various strata ac-
seeking one’s appropriate position and fulfilling quire. This may be said, along with the rights and
it conscientiously? perquisites, to constitute institutionalized social in-
Are not these types of motivations, we may equality, i.e., stratification.
ask, likely to prove most appropriate for precisely With the classification of the rewards offered
the “most functionally important positions”? Es- by Davis and Moore there need be little argu-
pecially in a mass industrial society, where the ment. Some question must be raised, however,
vast majority of positions become standardized as to whether any reward system, built into a
and routinized, it is the skilled jobs which are general stratification system, must allocate equal
likely to retain most of the quality of “intrinsic amounts of all three types of reward in order to
36 • Part I: The Functions and Dysfunctions of Inequality

function effectively, or whether one type of re- rewards and prestige between social strata, is
ward may be emphasized to the virtual neglect of what Davis and Moore, and most sociologists,
others. This raises the further question regarding consider the structure of a stratification system.
which type of emphasis is likely to prove most The former distinctions have nothing necessarily
effective as a differential inducer. Nothing in the to do with the workings of such a system nor
known facts about human motivation impels us with the efficiency of motivation and recruit-
to favor one type of reward over the other, or to ment of functionally important personnel.
insist that all three types of reward must be built Nor does the differentiation of power be-
into the positions in comparable amounts if the tween young and old necessarily create differen-
position is to have an inducement value. tially valued strata. For no society rates its young
It is well known, of course, that societies as less morally worthy than its older persons, no
differ considerably in the kinds of rewards they matter how much differential power the older
emphasize in their efforts to maintain a reason- ones may temporarily enjoy.
able balance between responsibility and reward.
There are, for instance, numerous societies in (7) Therefore, social inequality among different
which the conspicuous display of differential strata in the amounts of scarce and desired goods,
economic advantage is considered extremely bad and the amounts of prestige and esteem which they
taste. In short, our present knowledge commends receive, is both positively functional and inevitable
to us the possibility of considerable plasticity in in any society.
the way in which different types of rewards can If the objections which have heretofore been
be structured into a functioning society. This is raised are taken as reasonable, then it may be
to say, it cannot yet be demonstrated that it is stated that the only items which any society
unavoidable that differential prestige and esteem must distribute unequally are the power and
shall accrue to positions which command differ- property necessary for the performance of dif-
ential rewards in power and property. ferent tasks. If such differential power and prop-
What does seem to be unavoidable is that dif- erty are viewed by all as commensurate with the
ferential prestige shall be given to those in any differential responsibilities, and if they are cul-
society who conform to the normative order as turally defined as resources and not as rewards,
against those who deviate from that order in a then, no differentials in prestige and esteem
way judged immoral and detrimental. On the need follow.
assumption that the continuity of a society de- Historically, the evidence seems to be that
pends on the continuity and stability of its nor- every time power and property are distributed
mative order, some such distinction between unequally, no matter what the cultural defini-
conformists and deviants seems inescapable. tion, prestige and esteem differentiations have
It also seems to be unavoidable that in any tended to result as well. Historically, however, no
society, no matter how literate its tradition, the systematic effort has ever been made, under pro-
older, wiser and more experienced individuals pitious circumstances, to develop the tradition
who are charged with the enculturation and so- that each man is as socially worthy as all other
cialization of the young must have more power men so long as he performs his appropriate tasks
than the young, on the assumption that the task conscientiously. While such a tradition seems ut-
of effective socialization demands such differen- terly utopian, no known facts in psychological or
tial power. social science have yet demonstrated its impossi-
But this differentiation in prestige between bility or its dysfunctionality for the continuity of
the conformist and the deviant is by no means a society. The achievement of a full institutional-
the same distinction as that between strata of ization of such a tradition seems far too remote
individuals each of which operates within the to contemplate. Some successive approximations
normative order, and is composed of adults. The at such a tradition, however, are not out of the
latter distinction, in the form of differentiated range of prospective social innovation.
 Tumin—Some Principles of Stratikcation • 37

What, then, of the “positive functionality” social stratification systems function to dis-
of social stratification? Are there other, negative, tribute unequally the sense of significant
functions of institutionalized social inequal- membership in the population.
ity which can be identified, if only tentatively? 7. To the extent that loyalty to a society de-
Some such dysfunctions of stratification have pends on a sense of significant membership
already been suggested in the body of this paper. in the society, social stratification systems
Along with others they may now be stated, in the function to distribute loyalty unequally in
form of provisional assertions, as follows: the population.
1. Social stratification systems function to 8. To the extent that participation and apa-
limit the possibility of discovery of the full thy depend upon the sense of significant
range of talent available in a society. This membership in the society, social stratifi-
results from the fact of unequal access to cation systems function to distribute the
appropriate motivation, channels of recruit- motivation to participate unequally in a
ment and centers of training. population.
2. In foreshortening the range of available tal-
ent, social stratification systems function to Each of the eight foregoing propositions
set limits upon the possibility of expand- contains implicit hypotheses regarding the con-
ing the productive resources of the society, sequences of unequal distribution of rewards
at least relative to what might be the case in a society in accordance with some notion of
under conditions of greater equality of the functional importance of various positions.
opportunity. These are empirical hypotheses, subject to test.
3. Social stratification systems function to They are offered here only as exemplary of the
provide the elite with the political power kinds of consequences of social stratification
necessary to procure acceptance and domi- which are not often taken into account in deal-
nance of an ideology which rationalizes the ing with the problem. They should also serve to
status quo, whatever it may be, as “logical,” reinforce the doubt that social inequality is a de-
“natural” and “morally right.” In this man- vice which is uniformly functional for the role of
ner, social stratification systems function as guaranteeing that the most important tasks in a
essentially conservative influences in the so- society will be performed conscientiously by the
cieties in which they are found. most competent persons.
4. Social stratification systems function to The obviously mixed character of the func-
distribute favorable self-images unequally tions of social inequality should come as no sur-
throughout a population. To the extent prise to anyone. If sociology is sophisticated in
that such favorable self-images are requisite any sense, it is certainly with regard to its aware-
to the development of the creative potential ness of the mixed nature of any social arrange-
inherent in men, to that extent stratification ment, when the observer takes into account
systems function to limit the development long- as well as short-range consequences and
of this creative potential. latent as well as manifest dimensions.
5. To the extent that inequalities in social re-
wards cannot be made fully acceptable to the
less privileged in a society, social stratifica- Summary
tion systems function to encourage hostility, In this paper, an effort has been made to raise
suspicion and distrust among the various questions regarding the inevitability and posi-
segments of a society and thus to limit the tive functionality of stratification, or institution-
possibilities of extensive social integration. alized social inequality in rewards, allocated in
6. To the extent that the sense of significant accordance with some notion of the greater and
membership in a society depends on one’s lesser functional importance of various positions.
place on the prestige ladder of the society, The possible alternative meanings of the concept
38 • Part I: The Functions and Dysfunctions of Inequality

“functional importance” has been shown to be Journal of Sociology, XLV (November, 1940), pp.
one difficulty. The question of the scarcity or 849–862, approaches the problem in terms of why
abundance of available talent has been indicated “differential ranking is considered a really funda-
as a principal source of possible variation. The mental phenomenon of social systems and what are
the respects in which such ranking is important.”
extent to which the period of training for skilled
The principal line of integration asserted by Parsons
positions may reasonably be viewed as sacrificial is with the fact of the normative orientation of any
has been called into question. The possibility has society. Certain crucial lines of connection are left
been suggested that very different types of mo- unexplained, however, in this article, and in the
tivational schemes might conceivably be made Davis and Moore article of 1945 only some of these
to function. The separability of differentials in lines are made explicit.
power and property considered as resources ap- 2. The “scarcity and demand” qualities of goods
propriate to a task from such differentials consid- and services are never explicitly mentioned by Davis
ered as rewards for the performance of a task has and Moore. But it seems to the writer that the argu-
ment makes no sense unless the goods and services
also been suggested. It has also been maintained
are so characterized. For if rewards are to function as
that differentials in prestige and esteem do not differential inducements they must not only be dif-
necessarily follow upon differentials in power ferentially distributed but they must be both scarce
and property when the latter are considered as and demanded as well. Neither the scarcity of an
appropriate resources rather than rewards. Fi- item by itself nor the fact of its being in demand
nally, some negative functions, or dysfunctions, is sufficient to allow it to function as a differential
of institutionalized social inequality have been inducement in a system of unequal rewards. Leprosy
tentatively identified, revealing the mixed char- is scarce and oxygen is highly demanded.
acter of the outcome of social stratification, and 3. The arguments to be advanced here are con-
densed versions of a much longer analysis entitled,
casting doubt on the contention that
An Essay on Social Stratification. Perforce, all the
reasoning necessary to support some of the conten-
Social inequality is thus an unconsciously tions cannot be offered within the space limits of
evolved device by which societies insure that this article.
the most important positions are conscien- 4. Davis and Moore are explicitly aware of the
tiously filled by the most qualified persons.8 difficulties involved here and suggest two “indepen-
dent clues” other than survival value. See footnote 3
NOTES on p. 244 of their article.
The writer has had the benefit of a most help- 5. Davis and Moore state this point briefly on p.
ful criticism of the main portions of this paper by 248 but do not elaborate it.
Professor W. J. Goode of Columbia University. In 6. In the United States, for instance, we are only
addition, he has had the opportunity to expose this now becoming aware of the amount of productivity
paper to criticism by the Staff Seminar of the Sociol- we, as a society, lose by allocating inferior oppor-
ogy Section at Princeton. In deference to a possible tunities and rewards, and hence, inferior motiva-
rejoinder by Professors Moore and Davis, the writer tion, to our Negro population. The actual amount
has not revised the paper to meet the criticisms of loss is difficult to specify precisely. Some rough
which Moore has already offered personally. estimate can be made, however, on the assumption
1. American Sociological Review, 10 (April, 1945), that there is present in the Negro population about
pp. 242–249. An earlier article by Kingsley Davis, the same range of talent that is found in the White
entitled, “A Conceptual Analysis of Stratification,” population.
American Sociological Review, VII (June, 1942), pp. 7. These are only very rough estimates, of course,
309–321, is devoted primarily to setting forth a and it is certain that there is considerable income
vocabulary for stratification analysis. A still earlier variation within the so-called elite group, so that the
article by Talcott Parsons, “An Analytical Approach proposition holds only relatively more or less.
to the Theory of Social Stratification,” American 8. Davis and Moore, op. cit., p. 243.
 Fischer et al.—Inequality by Design • 39

4 • Claude S. Fischer, Michael Hout, Martín Sánchez Jankowski,

Samuel R. Lucas, Ann Swidler, and Kim Voss

Inequality by Design

Why do some Americans have a lot more than people inherit from their parents, the resources
others? Perhaps, inequality follows inevitably that their friends can share with them, the quan-
from human nature. Some people are born with tity and quality of their schooling, and even the
more talent than others; the first succeed while historical era into which they are born boost
the others fail in life’s competition. Many people some up and hold others down. The children
accept this explanation, but it will not suffice. of professors, our own children, have substantial
Inequality is not fated by nature, nor even by the head starts over children of, say, factory workers.
“invisible hand” of the market; it is a social con- Young men who graduated from high school in
struction, a result of our historical acts. Ameri- the booming 1950s had greater opportunities
cans have created the extent and type of inequality than the ones who graduated during the Depres-
we have, and Americans maintain it. sion. Context matters tremendously.
To answer the question of what explains in- The answer to the question of why societies
equality in America, we must divide it in two. vary in their structure of rewards is more polit-
First, who gets ahead and who falls behind in ical. In significant measure, societies choose the
the competition for success? Second, what de- height and breadth of their “ladders.” By loos-
termines how much people get for being ahead ening markets or regulating them, by providing
or behind? To see more clearly that the two services to all citizens or rationing them accord-
questions are different, think of a ladder that ing to income, by subsidizing some groups more
represents the ranking of affluence in a society. than others, societies, through their politics,
Question one asks why this person rather than build their ladders. To be sure, historical and ex-
that person ended up on a higher or lower rung. ternal constraints deny full freedom of action,
Question two asks why some societies have tall but a substantial freedom of action remains. In a
and narrowing ladders—ladders that have huge democracy, this means that the inequality Amer-
distances between top and bottom rungs and icans have is, in significant measure, the histor-
that taper off at the top so that there is room ical result of policy choices Americans—or, at
for only a few people—while other societies have least, Americans’ representatives—have made. In
short and broad ladders—ladders with little dis- the United States, the result is a society that is
tance between top and bottom and with lots of distinctively unequal. Our ladder is, by the stan-
room for many people all the way to the top. dards of affluent democracies and even by the
The answer to the question of who ends up standards of recent American history, unusually
where is that people’s social environments largely extended and narrow—and becoming more so.
influence what rung of the ladder they end up To see how policies shape the structure of re-
on.1 The advantages and disadvantages that wards (i.e., the equality of outcomes), consider

Claude S. Fischer, Michael Hout, Martín Sánchez Jankowski, Samuel R. Lucas, Ann Swidler, and Kim Voss.
Inequality by Design, pp. 7–10, 126–128, 241, 260–261, 279–280, 284, 286, 290, 292, 295, 299, 301. Copy-
right © 1996 by Princeton University Press. Reprinted by permission of Princeton University Press.
40 • Part I: The Functions and Dysfunctions of Inequality

these examples: Laws provide the ground rules concerns differences in equality among societies.
for the marketplace—rules covering incorpo- Theories of natural inequality cannot tell us why
ration, patents, wages, working conditions, countries with such similar genetic stocks (and
unionization, security transactions, taxes, and economic markets) as the United States, Can-
so on. Some laws widen differences in income ada, England, and Sweden can vary so much in
and earnings among people in the market; others the degree of economic inequality their citizens
narrow differences. Also, many government pro- experience. The answer lies in deliberate policies.
grams affect inequality more directly through, Appeals to nature also cannot satisfactorily
for example, tax deductions, food stamps, social answer even the first question: Why do some
security, Medicare, and corporate subsidies. individuals get ahead and some fall behind? Cer-
To see how policies also affect which particu- tainly, genetic endowment helps. Being tall, slen-
lar individuals get to the top and which fall to the der, good-looking, healthy, male, and white helps
bottom of our ladder (i.e., the equality of oppor- in the race for success, and these traits are totally
tunity), consider these examples: The amount of or partly determined genetically. But these traits
schooling young Americans receive heavily de- matter to the degree that society makes them
termines the jobs they get and the income they matter—determining how much, for example,
make. In turn, educational policies—what sorts good looks or white skin are rewarded. More im-
of schools are provided, the way school resources portant yet than these traits are the social milieux
are distributed (usually according to the commu- in which people grow up and live.
nity in which children live), teaching methods Realizing that intentional policies account for
such as tracking, and so on—strongly affect how much of our expanding inequality is not only
much schooling children receive. Similarly, local more accurate than theories of natural inequal-
employment opportunities constrain how well ity; it is also more optimistic. We are today more
people can do economically. Whether and where unequal than we have been in seventy years. We
governments promote jobs or fail to do so will, are more unequal than any other affluent Western
in turn, influence who is poised for well-paid nation. Intentional policies could change those
employment and who is not. conditions, could reduce and reverse our rush to a
Claiming that intentional policies have sig- polarized society, could bring us closer to the av-
nificantly constructed the inequalities we have erage inequality in the West, could expand both
and that other policies could change those in- equality of opportunity and equality of result.
equalities may seem a novel idea in the current Still, the “natural inequality” viewpoint is a
ideological climate. So many voices tell us that popular one. Unequal outcomes, the bestselling
inequality is the result of individuals’ “natural” Bell Curve argues, are the returns from a fair pro-
talents in a “natural” market. Nature defeats any cess that sorts people out according to how intel-
sentimental efforts by society to reduce inequal- ligent they are.3 But The Bell Curve’s explanation
ity, they say; such efforts should therefore be of inequality is inadequate. The authors err in
dropped as futile and wasteful. Appeals to nature assuming that human talents can be reduced to
are common and comforting. As Kenneth Bock a single, fixed, and essentially innate skill they
wrote in his study of social philosophy, “We have label intelligence. They err in asserting that this
been quick to seek explanations of our problems trait largely determines how people end up in
and failures in what we are instead of what we life. And they err in imagining that individual
do. We seem wedded to the belief that our situ- competition explains the structure of inequality
ation is a consequence of our nature rather than in society. . . .
of our historical acts.”2 In this case, appeals to Disparities in income and wealth, [other]
nature are shortsighted. analysts argue, encourage hard work and saving.
Arguments from nature are useless for an- The rich, in particular, can invest their capital in
swering the question of what determines the production and thus create jobs for all.4 This was
structure of rewards because that question the argument of “supply-side” economics in the
 Fischer et al.—Inequality by Design • 41

1980s, that rewarding the wealthy—for exam- and to provide opportunities for everyone. Lin-
ple, by reducing income taxes on returns from coln once said “that some would be rich shows
their investments—would stimulate growth to that others may become rich.”8 Many, if not
the benefit of all. The 1980s did not work out most, Americans believe that inequality is
that way, but the theory is still influential. We needed to encourage people to work hard.9 But,
could force more equal outcomes, these analysts if so, how much inequality is needed?
say, but doing so would reduce living standards For decades, sociologists have been compar-
for all Americans. ing the patterns of social mobility across socie-
Must we have so much inequality for overall ties, asking: In which countries are people most
growth? The latest economic research concludes likely to overcome the disadvantages of birth and
not; it even suggests that inequality may retard move up the ladder? In particular, does more or
economic growth. In a detailed statistical anal- less equality encourage such an “open” society?
ysis, economists Torsten Persson and Guido The answer is that Western societies vary lit-
Tabellini reported finding that, historically, soci- tle in the degree to which children’s economic
eties that had more inequality of earnings tended successes are constrained by their parents’ class
to have lower, not higher, subsequent economic positions. America, the most unequal Western
growth. Replications by other scholars substan- society, has somewhat more fluid intergener-
tiated the finding: More unequal nations grew ational mobility than do other nations, but so
less quickly than did more equal societies.5 . . . does Sweden, the most equal Western society.10
This recent research has not demonstrated There is no case for encouraging inequality in
precisely how greater equality helps economic this evidence, either.
growth,6 but we can consider a few possibilities. In sum, the assumption that considerable
Increasing resources for those of lower income inequality is needed for, or even encourages,
might, by raising health, educational attain- economic growth appears to be false. We do not
ment, and hope, increase people’s abilities to need to make a morally wrenching choice be-
be productive and entrepreneurial. Reducing tween more affluence and more equality; we can
the income of those at the top might reduce have both. But even if such a choice were neces-
unproductive and speculative spending. Take, sary, both sides of the debate, the “altruists” who
as a concrete example, the way American cor- favor intervention for equalizing and the sup-
porations are run compared with German and posed “realists” who resist it, agree that inequal-
Japanese ones. The American companies are run ity can be shaped by policy decisions: Wittingly
by largely autonomous managers whose main or unwittingly, we choose our level of inequality.
responsibility is to return short-term profits and
high stock prices to shareholders and—because NOTES
they are often paid in stock options—to them- 1. We know that in statistical models of indi-
vidual status attainment much, if not most, of the
selves as well. Japanese and German managers
variance is unaccounted for. Of the explained vari-
are more like top employees whose goals largely
ance, however, the bulk is due to social environment
focus on keeping the company a thriving enter- broadly construed. Also, we believe that much of
prise. The latter is more conducive to reinvesting the residual, unexplained variance is attributable to
profits and thus to long-term growth.7 Whatever unmeasured social rather than personal factors.
the mechanisms may be, inequality appears to 2. Kenneth Bock, Human Nature Mythology (Ur-
undermine growth. Americans certainly need bana 1994), p. 9.
not feel that they must accept the high levels of 3. Richard J. Herrnstein and Charles Murray,
inequality we currently endure in order to have The Bell Curve: Intelligence and Class Structure in
a robust economy. American Life (New York 1994).
4. See, for example, Rich Thomas, “Rising Tide
A related concern for Americans is whether
Lifts the Yachts: The Gap Between Rich and Poor
“leveling” stifles the drive to get ahead. Ameri- Has Widened, but There Are Some Comforting
cans prefer to encourage Horatio Alger striving Twists,” Newsweek, May 1, 1995. See also George
42 • Part I: The Functions and Dysfunctions of Inequality

Will, “What’s Behind Income Disparity,” San Fran- event, the statistical results suggest that government
cisco Chronicle, April 24, 1995. intervention on behalf of equality in the market,
5. Torsten Persson and Guido Tabellini, “Is In- rather than after the market, would be beneficial.
equality Harmful for Growth?,” American Economic 7. See, for example, Michael Porter, Capital
Review 84, 1994; Roberto Chang, “Income Inequal- Choices: Changing the Way America Invests in Industry
ity and Economic Growth: Evidence and Recent (Washington 1992).
Theories,” Economic Review 79, 1994; George R. G. 8. Quoted by Alan Trachtenberg, The Incorpora-
Clarke, “More Evidence on Income Distribution and tion of America: Culture and Society in the Gilded Age
Growth,” Journal of Development Economics 47, 1995. (New York 1982), p. 75.
See also Peter H. Lindert, “The Rise of Social Spend- 9. See, for example, Lee Rainwater, What Money
ing,” Explorations in Economic History 31, 1994. Buys: Inequality and the Social Meanings of Income
6. Persson and Tabellini’s explanation (“Is In- (New York 1974); James R. Kluegel and E. R. Smith,
equality Harmful?”) for their results is that in so- “Beliefs About Stratification,” Annual Review of So-
cieties with greater earnings inequality, there is less ciology 7, 1981.
political pressure for government redistribution; 10. Harry B. G. Ganzeboom, Donald J. Treiman,
such redistribution impairs growth. However, their and Wout C. Ultee, “Comparative Intergenerational
evidence for the explanation is thin, and Clarke’s re- Stratification Research,” Annual Review of Sociology
sults (“More Evidence”) are inconsistent with that 17, 1991.
argument. Chang (“Income Inequality”) suggests
that with more equality, lower-income families could
make longer-term investment decisions. In any
 PART II

Inequality in Comparative
Perspective

Cross-Societal Differences
5 A Compressed History of Inequality 44
6 The Welfare State and Redistribution 52

Trends in Economic Inequality

7 Top Incomes in the Long Run of History 59
8 The Race Between Education and Technology 73
9 Unions, Norms, and the Rise in U.S. Wage Inequality 80
10 Why Is Income Inequality Growing? 86
11 Winner-Take-All Politics: Public Policy, Political Organization,
and the Precipitous Rise of Top Incomes in the United States 91
12 Leapfrogs and the Surge in Executive Pay 104
13 The Winners of the Third Industrial Revolution 113
44 • Part II: Inequality in Comparative Perspective

Cross-Societal Differences

5 • David B. Grusky and Katherine R. Weisshaar

A Compressed History of Inequality

The task of identifying the dynamics behind so- are largely conservative, but they can sometimes
cial change is fundamental to all of sociology, serve as forces for change as well as stability. If,
but nowhere is this agenda more important than for example, the facts of labor market processes
in the field of inequality. This chapter lays out are inconsistent with the prevailing ideology
some of the most prominent narratives about the (e.g., racial discrimination in advanced indus-
dynamics of changing inequality over the long trialism), then a variety of ameliorative actions
run of human history. We begin by developing might be anticipated.
a conventional typology of past inequality forms The categories of Table 5.1 have been drawn
and then conclude very briefly with the question from various brands of sociology. The staple of
of how the late modern form has come to be modern classification efforts has been the tripar-
understood. tite distinction among class, caste, and estate, but
The simple typology of Table 5.1 distin- there is also a long and illustrious Marxian tradi-
guishes societies by the amount of inequality, the tion that introduces the additional categories of
extent to which access to social classes is closed primitive communism, slave society, and social-
(i.e., “rigidity”), the extent to which the various ism. As shown in Table 5.1, these conventional
dimensions of inequality (e.g., money, prestige, approaches are largely complementary, and it
power) come together to form a consistently ad- is therefore possible to fashion a hybrid classi-
vantaged upper class (i.e., “crystallization”), and fication that incorporates most of the standard
the type of ideologies that elites and others use to distinctions. The resulting typology should be
justify inequality. For each of the eight regimes understood as a highly stylized and compressed
listed in Table 5.1, a particular class of assets is history of inequality forms over the premodern,
assumed to be especially crucial in defining the modern, and late modern periods.
structure of inequality, thus making it possible The resulting forms also should be taken as
to specify the dominant social classes in terms ideal types. In constructing these categories, our
of those assets.1 intention is not to make empirical claims about
The final column of Table 5.1 rests on the how existing and past systems operated in prac-
further assumption that inequality regimes have tice, but rather to capture and distill the accu-
(reasonably) coherent ideologies that serve to mulated wisdom about how these systems might
legitimate the rules and criteria by which in- operate in their purest form. These ideal-typical
dividuals are allocated to positions in the class models can nonetheless assist us in understand-
structure. In most cases, ideologies of this kind ing empirical systems. Indeed, insofar as societies

Original article prepared for this fourth edition of Social Stratification: Class, Race, and Gender in Sociological
Perspective, edited by David B. Grusky. Copyright © 2014 Westview Press.
 Grusky and Weisshaar—A Compressed History of Inequality • 45

Table 5.1. Parameters of Stratification for Eight Ideal-Typical Systems

evolve through the gradual overlaying of new the overall level of economic inequality (but not
stratification forms on older (and partly super- necessarily on other forms of inequality).
seded) ones, it is possible to interpret contempo- In early tribal societies the economy was
rary systems as a complex mixture of several of based on hunting and gathering, and a nomadic
the ideal types presented in Table 5.1. lifestyle was the norm. Given this economic con-
text, the typical food share was barely above the
subsistence minimum, and any substantial accu-
Hunting and Gathering Societies mulation of material possessions was “generally
The first section of Table 5.1 pertains to the impossible” (Kerbo 2012, 54). As Milanovic,
tribal systems that dominated human society Lindert, and Williamson (2007, 5) have noted,
from the very beginning of human evolution the small surplus and population size reduced
until the Neolithic revolution of some ten thou- the opportunity for inequality, and customs such
sand years ago. The characterizations of columns as gift exchange, food sharing, and the like were
2–7 necessarily conceal much variability in the accordingly common. In fact, some observers
structure of these systems. As Anderson (1974, (e.g., Marx [1939] 1971) have treated these so-
549) so eloquently put it, it is “merely in the cieties as examples of “primitive communism,”
night of our ignorance [that] all alien shapes because the means of production (e.g., tools,
take on the same hue.” Although such variable land) were owned collectively, and other types of
features of tribal societies are clearly of interest, property typically were distributed evenly among
for our purposes the important similarities are tribal members. The archaeological evidence
that (1) the total size of the distributable surplus thus suggests that “humans have existed for well
was in all cases quite limited, and (2) this cap over 2 million years in a state of relative equality”
on the surplus placed corresponding limits on (Hayden 2007, 1).
46 • Part II: Inequality in Comparative Perspective

This is obviously not to suggest that a perfect class (Midlarsky 1999, 6). As agrarian societies
equality prevailed. The more powerful medicine developed, this class engaged in various types of
men (i.e., shamans) within tribal societies often predatory behavior, especially the “harvesting”
secured a disproportionate share of resources, and of agricultural surpluses (Milanovic, Lindert,
the tribal chief could exert considerable influence and Williamson 2007; Boix 2010; Ensminger
on the political decisions of the day. When arti- and Knight 1997). The result was early agrar-
facts from burial grounds are examined, it is typ- ian empires in Egypt, China, and Europe with
ically found that some individuals garnered more political and religious ruling elites that had “vast
lavish symbolic offerings upon death, suggesting economic as well as political control” (Kerbo
the existence of a prestige hierarchy (Hayden 2012, 71).
2007, 9). The available evidence also reveals that Because the “Asiatic mode” is often treated
some individuals or families lived in larger units as an intermediate formation in this transition
(Kerbo 2012). Furthermore, ancient relics sug- to advanced agrarian society (e.g., feudalism),
gest a gender division of labor, with men more we start our typology with the Asiatic case (see
likely to participate in hunting and politics. At row B2).3 The explicit evolutionary theories of
the same time, the evidence is obviously partial Godelier (1978) and others have not always
and difficult to interpret, and what is available been well received, yet many scholars still take
points to some variability among tribal societies the fallback position that Asiaticism is an im-
in the extent of inequality (see Mushinski and portant analytical category in the development
Pickering 2000; Schröder 2003, 452). of class society (e.g., Hobsbawm 1965). As
In most cases these residual forms of power agricultural technologies improved, the state
and privilege were not directly inherited, nor elite began to organize large-scale agricultural
were they typically allocated in accordance with projects, which allowed it to accumulate power
well-defined ascriptive traits (e.g., racial traits).2 based on its political position rather than formal
The evidence suggests instead that tribal mem- ownership of private property (Kerbo 2012, 69).
bers often secured political office or acquired The main features of this formation were (1) a
status and prestige by demonstrating superior large peasant class residing in agricultural vil-
skills in hunting, magic, or leadership. Although lages that were “almost autarkic” (O’Leary 1989,
meritocratic forms of allocation are usually seen 17); (2) the absence of strong legal institutions
as prototypically modern, in fact they were pres- recognizing private property rights; (3) a state
ent in incipient form at the very earliest stages of elite that extracted surplus agricultural produc-
societal development (Kerbo 2012, 56). tion through rents or taxes and expended it on
“defense, opulent living, and the construction
of public works” (Shaw 1978, 127);4 and (4) a
Horticultural and Agrarian Society constant flux in elite personnel due to “wars of
Some ten thousand years ago a gradual transition dynastic succession and wars of conquest by no-
from hunting and gathering to agrarian society madic warrior tribes” (O’Leary 1989, 18).
began, although there was considerable variabil- Beyond this skeletal outline, all else is open to
ity among societies in the timing of the shift dispute. There are long-standing debates, for ex-
(Hayden 2007). This transformation ultimately ample, about how widespread the Asiatic mode
led to major changes in daily life. When the was and about the appropriateness of reducing
economic surplus became large enough, a com- all forms of Asian development to a “uniform
plex division of labor could be supported, and residual category” (Anderson 1974, 548–549).
specialists in science, religion, arts, and military These issues are clearly worth pursuing, but for
technology emerged (Kerbo 2012, 54). our purposes it suffices that the Asiatic mode
The key development during this period was provides a conventional example of how a “dic-
the rise of a segregated and partly closed ruling tatorship of officialdom” can flourish in the
 Grusky and Weisshaar—A Compressed History of Inequality • 47

absence of private property and a well-developed feudal stratification system, because the means
proprietary class (Gouldner 1980, 327–328). of production (i.e., land, labor) were controlled
Under this reading of Asiaticism, the parallel by a proprietary class that emerged quite inde-
with modern socialism looms large (at least in pendently of the state.8
some quarters), so much so that various scholars The historical record makes it clear that
have suggested that Marx downplayed the Asian agrarian stratification systems were not always
case for fear of exposing it as a “parable for so- based on strictly hereditary forms of social clo-
cialism” (see Gouldner 1980, 324–352). sure (see Table 5.1, section B, column 5). The
Whereas the institution of private property case of European feudalism is especially instruc-
was underdeveloped in the East, the ruling class tive in this regard, because it suggests that strat-
under Western feudalism was, by contrast, very ification systems often become more rigid as the
much a propertied one.5 The distinctive fea- underlying institutional forms mature and take
ture of feudalism was that the nobility not only shape. Although it is well known that the era of
owned large estates or manors, but also held classical feudalism was characterized by a “rigid
legal title to the labor power of its serfs (see row stratification of social classes” (Bloch 1961,
B3).6 If a serf fled to the city, this was consid- 325),9 there was greater permeability during the
ered a form of theft: the serf was stealing that period prior to the institutionalization of the
portion of his or her labor power owned by the manorial system and the associated transforma-
lord (see Wright, Chapter 16; Sharma 1983). tion of the nobility into a legal class. In this tran-
Under this interpretation, the status of serf and sitional period, access to the nobility was not yet
slave differed only in degree, and slavery thereby legally restricted to the offspring of nobility, nor
constituted a limiting case in which workers lost was marriage across classes or estates formally
all control over their own labor power (see row prohibited (see Bloch 1961, 320–331). The case
B4). At the same time, it would obviously be a of ancient Greece provides a complementary ex-
mistake to reify this distinction, given that the ample of a (relatively) open agrarian society. As
history of agrarian Europe reveals “almost in- Finley (1960) and others have noted, the condi-
finite gradations of subordination” (Bloch 1961, tion of slavery was indeed heritable under Greek
256) that confuse and blur the conventional di- law, yet manumission was so common that the
viding lines among slavery, serfdom, and free- slave class had to be constantly replenished with
dom. The slavery of Roman society provides new captives secured through war or piracy. The
the best example of complete subordination, possibility of servitude was thus something that
whereas some of the slaves of the early feudal “no man, woman, or child, regardless of status
period had rights of real consequence (e.g., the or wealth, could be sure to escape” (Finley 1960,
right to sell surplus product), and some of the 161).
(nominally) free men were in fact obliged to At the same time, hereditary forms of closure
provide rents or services to the manorial lord were more fully developed in some slave systems,
(Bloch 1961, 255–274).7 most notably the American one. As Sio (1965)
The classes that emerged under European notes, slavery in the antebellum South was “he-
agrarianism were thus structured in quite com- reditary, endogamous, and permanent” (303),
plicated ways. There were nonetheless three with the annual manumission rate apparently as
broad types of estates: the priestly class, the low as 0.04 percent by 1850 (see Patterson 1982,
nobility, and commoners. In all cases, property 273). The slave societies of Jamaica, South Af-
ownership was firmly established, and the life rica, Brazil, and rural Iraq were likewise based on
chances of individuals were defined, in large largely permanent slave populations (see Sum-
part, by their control over property in its vari- merhill 2010). Although the degree of inequal-
ous forms. Unlike the ideal-typical Asiatic case, ity between slave and owner varied somewhat
the nation-state was largely peripheral to the across societies (Kerbo 2012, 57), this form of
48 • Part II: Inequality in Comparative Perspective

stratification was typically characterized by ex- delegitimation of the extreme forms of stratifi-
treme inequality and rigidity. cation found in caste, feudal, and slave systems.
The most extreme examples of hereditary This can be seen, for example, in the European
closure are of course found in caste societies revolutions of the eighteenth and nineteenth
(see row B5). In some respects, American slav- centuries, which pitted the egalitarian ideals of
ery might be seen as having “caste-like features” the Enlightenment against the privileges of rank
(see Berreman 1981), but Hindu India clearly and the political power of the nobility. In the
provides the defining case of caste organiza- end, these struggles eliminated the last residue
tion.10 The Indian caste system is based on (1) of feudal privilege, but they also made new types
a hierarchy of status groupings (i.e., castes) that of inequality and stratification possible. Under
are ranked by ethnic purity, wealth, and access the class system that ultimately emerged (see row
to goods or services; (2) a corresponding set of C6), the estates of the feudal era were replaced
“closure rules” that restrict all forms of intercaste by purely economic groups (i.e., “classes”), and
marriage or mobility and thereby make caste closure rules based on heredity were likewise
membership both hereditary and permanent; (3) supplanted by (formally) meritocratic processes.
a high degree of physical and occupational seg- The resulting classes were neither legal entities
regation enforced by elaborate rules and rituals nor closed status groupings, and the associated
governing intercaste contact; and (4) a justifying class-based inequalities could therefore be rep-
ideology that induces the population to regard resented and justified as the natural outcome of
such extreme forms of inequality as legitimate competition among individuals with differing
and appropriate (Smaje 2000; Dumont 1970; abilities, motivation, or moral character (i.e.,
Srinivas 1962). What makes this system so dis- “classical liberalism”). As indicated in row C6 of
tinctive is not merely its well-developed closure Table 5.1, the class structure of early industrial-
rules, but also the fundamentally honorific (and ism had a clear “economic base” (Kerbo 1991,
noneconomic) character of the underlying social 23), so much so that Marx (Chapter 14) de-
hierarchy. As indicated in Table 5.1, the castes fined classes by their relationship to the means
of India were ranked on a continuum of ethnic of economic production. The precise contours
and ritual purity, with the highest positions in of the industrial class structure are nonetheless
the system reserved for castes that prohibit be- a matter of continuing debate. For example, a
haviors seen as dishonorable or polluting. The simple Marxian model focuses on the cleavage
resulting system was based on strict rules gov- between capitalists and workers, while more
erning how people from different castes could elaborate Marxian and neo-Marxian models
came into contact (Kerbo 2012, 58). Under identify additional intervening or “contradic-
some circumstances, castes that acquired polit- tory” classes (e.g., Wright, Chapter 16), and
ical and economic power eventually advanced yet other (non-Marxian) approaches represent
in the status hierarchy, yet they typically did so the class structure as a continuous gradation of
only after mimicking the behaviors and lifestyles wealth or income.
of higher castes. The descendants of lower-status Whatever the relative merits of these models
castes continue even today to face discrimina- might be, the ideology underlying the socialist
tion, higher poverty rates, and lower chances of revolutions of the nineteenth and twentieth cen-
upward economic mobility (Kijima 2006, 30). turies was of course explicitly Marxist. The in-
tellectual heritage of these revolutions and their
legitimating ideologies can again be traced to the
Industrial Society Enlightenment, but the rhetoric of equality that
The defining feature of the industrial era (see emerged in this period was directed against the
Table 5.1, section C) has been the emergence economic power of the capitalist class rather than
of egalitarian ideologies and the consequent the status and honorific privileges of the nobility.
 Grusky and Weisshaar—A Compressed History of Inequality • 49

The evidence from Eastern Europe and else- of long-standing capitalist ones (see Heyns 2005;
where suggests that these egalitarian ideals were Wu and Treiman 2007; Bian 2002).
only partially realized. In the immediate post-
revolutionary period, factories and farms were
indeed collectivized or socialized, and various Where Are We Going?
fiscal and economic reforms were instituted for We do not attempt to address in any detail on-
the express purpose of reducing income inequal- going debates about how best to characterize late
ity and wage differentials among manual and industrialism. It is important, however, to point
nonmanual workers. Although these egalitarian out that the typology of Table 5.1 tells a story
policies were subsequently weakened through the about advanced industrialism that, although still
reform efforts of Stalin and others, inequality on dominant, is increasingly subject to challenge.
the scale of prerevolutionary society was never We close by reviewing first this conventional wis-
reestablished among rank-and-file workers. dom and then presenting the emerging challenge
There nonetheless remained substantial in- to it.
equalities in power and authority. Most notably, As shown in section C of Table 5.1, we have
the socialization of productive forces did not characterized early industrialism as a “class sys-
have the intended effect of empowering workers, tem” featuring the divide between capitalists
as the capitalist class was replaced by a “new class” and workers, whereas we have characterized
of party officials and managers who continued to advanced industrialism as an “occupational sys-
control the means of production and to allocate tem” featuring the divide between skilled and
the resulting social surplus (see Eyal, Szelényi, unskilled workers. The conventional wisdom,
and Townsley, Chapters 36 and 127). This class at least until recently, was that the propertied
has been variously identified with intellectuals class has been weakened by ongoing structural
or intelligentsia, bureaucrats or managers, and changes, with the most important of these being
party officials or appointees (Gouldner, Chapter (1) the rise of a service economy and the grow-
34). Regardless of the formulation adopted, the ing power of the “service class” (Bell, Chapter
presumption is that the working class ultimately 123; Esping-Andersen, Chapter 124); (2) the
lost out in contemporary socialist revolutions, increasing centrality of theoretical knowledge in
just as it did in the so-called bourgeois revolu- the transition to a new “information age” (Bell,
tions of the eighteenth and nineteenth centuries. Chapter 123); and (3) the consequent emer-
Whereas the means of production were so- gence of technical expertise, educational de-
cialized in the revolutions of Eastern Europe grees, and training certificates as “new forms of
and the former Soviet Union, the capitalist class property” (Berg 1973, 183; Parkin, Chapter 20;
remained intact throughout the process of in- Gouldner, Chapter 34). The foregoing develop-
dustrialization in the West. Although income ments all suggest that human and cultural capital
inequality increased with early industrialization, are displacing economic capital as the principal
it then declined as various equalizing institu- stratifying forces in advanced industrial society
tions (e.g., unions) diffused. In the 1970s, in- (see Table 5.1, row C8). By this formulation, a
come inequality began to increase again, and in new class of cultural elites may be emerging in
some countries it has now reached levels rivaling the West (see Lamont and Molnár 2002), much
those of the first Gilded Age (late nineteenth as the transition to state socialism (allegedly)
and early twentieth centuries). Likewise, as generated a new class of intellectuals in the East.
Central-Eastern European nations transitioned It is nonetheless unclear whether cultural
to market-based economies, support for inequal- and intellectual elites will ultimately triumph
ity increased (Kelley and Zagorski 2004), and against the growing economic power of capital-
many formerly socialist societies now have levels ists. Although scholars have recently been fixated
of income inequality that equal or exceed those on rising income inequality among laborers, a
50 • Part II: Inequality in Comparative Perspective

quieter takeoff has also played out in the form always vary in tandem with persistence between gen-
of a long-term increase in the share of national erations (i.e., intergenerational inheritance). We have
income that accrues to capital rather than labor. nonetheless assumed that each of our ideal-typical
Over the last forty years of US history, labor’s stratification systems can be characterized by a single
“rigidity parameter” (see column 5).
share has fallen from 66 to 60 percent of total
2. This claim does not hold for gender. That is,
national income, with the corresponding share women and men were typically assigned to different
going to capitalists rising from 34 to 40 percent roles, which led to consequent differences in the dis-
(see Kristal 2010). There is some debate about tribution of rewards.
what underlies this development. Although 3. It should again be stressed that our typology
some scholars argue that computer and infor- by no means exhausts the variability of agrarian
mation technologies have increased the produc- stratification forms (see Kerbo 2000 for an extended
tivity of capital and reduced labor’s share of total review).
national income, others suggest that capitalists 4. The state elite was charged with constructing
and maintaining the massive irrigation systems that
have newfound leverage that allows them to ap-
made agriculture possible in regions such as China,
propriate income that workers were once able India, and the Middle East (cf. Anderson 1974,
to secure. The power of capitalists is increasing 490–492).
because (1) unions have weakened and no longer 5. This is not to suggest that feudalism was only
protect workers against competition from other found in the West or that the so-called Asiatic mode
domestic workers, and (2) international barriers was limited to the East. Indeed, the social struc-
to the flow of capital have also weakened and ture of Japan was essentially feudalistic until the
no longer protect domestic workers against com- mid-nineteenth century (with the rise of the Meiji
petition from foreign workers. The bargaining state), and the Asiatic mode has been discovered
in areas as diverse as Africa, pre-Columbian Amer-
power of capital has increased insofar as workers
ica, and even Mediterranean Europe (see Godelier
must now negotiate without the protection once
1978). The latter “discoveries” were of course pred-
afforded by unions or by “territorial closure” in icated on a broad and ahistorical definition of the
the form of an implicit national monopoly on underlying ideal type.
the provision of labor. 6. This economic interpretation of feudalism
The upshot is that late industrialism may yet is clearly not favored by all scholars. For example,
turn against the cultural and educational elite Bloch (1961, 288–289) argues that the defining fea-
and prove ultimately to be as favorable for cap- ture of feudalism is the monopolization of author-
italists as was early industrialism. Although the ity by a small group of nobles, with the economic
concomitants of this authority (e.g., land ownership)
immediate postwar period was indeed an era of
thus being reduced to a position of secondary impor-
proworker institutions (e.g., unions) and capi-
tance. The “authority classes” that emerge under his
talist restraint, we cannot rule out the possibility specification might be seen as feudal analogues to the
that it was an idiosyncratic moment in history social classes that Dahrendorf (Chapter 15) posits for
rather than, as some have suggested, its glorious the capitalist case.
end. 7. In the so-called secondary stage of feudalism
(Bloch 1961), the obligations of serfs and free men
NOTES became somewhat more formalized and standard-
1. The assumptions embedded in columns 4–6 of ized, yet regional variations persisted.
Table 5.1 are clearly far-reaching. Unless a stratifica- 8. It was not until the early fourteenth century
tion system is perfectly crystallized, its parameters for that states of the modern sort appeared in Europe.
inequality and rigidity cannot be represented as sca- 9. In describing this period of classical feudal-
lar quantities, nor can the intercorrelations among ism, Bloch (1961) noted that “access to the circle of
the multiple stratification dimensions be easily sum- knights . . . was not absolutely closed, [yet] the door
marized in a single parameter. Moreover, even in was nevertheless only very slightly ajar” (325).
stratification systems that are perfectly crystallized, 10. The Indian caste system flourished during the
there is no reason to believe that persistence over the agrarian period, yet it persists in attenuated form in
life course (i.e., intragenerational persistence) will modern industrialized India.
 Grusky and Weisshaar—A Compressed History of Inequality • 51

REFERENCES by Jack Cohen and edited by Eric J. Hobsbawm,

Acemoglu, Daron, and James A. Robinson. 2000. 9–65. New York: International Publishers.
“Why Did the West Extend the Franchise? De- Kelley, J., and K. Zagorski. 2004. “Economic
mocracy, Inequality, and Growth in Historical Change and the Legitimation of Inequality: the
Perspective.” Quarterly Journal of Economics 115 Transition from Socialism to the Free Market in
(4): 1–33. Central-East Europe.” Research in Social Stratifi-
Anderson, Perry. 1974. Lineages of the Absolutist cation and Mobility 22: 319–364.
State. London: Verso. Kerbo, Harold R. 1991. Social Stratification and In-
Berg, Ivar. 1973. Education and Jobs: The Great equality: Class Conflict in Historical and Compar-
Training Robbery. Harmondsworth, UK: Pen- ative Perspective. New York: McGraw-Hill.
guin Books. ———. 2000. Social Stratification and Inequality:
Berreman, Gerald. 1981. Caste and Other Inequities. Class Conflict in Historical, Comparative, and
Delhi, India: Manohar. Global Perspective. New York: McGraw-Hill.
Bian, Yanjie. 2002. “Chinese Social Stratification ———. 2012. Social Stratification and Inequality:
and Social Mobility.” Annual Review of Sociology Class Conflict in Historical and Comparative Per-
28: 91–116. spective. New York: McGraw-Hill.
Bloch, Marc. 1961. Feudal Society. London: Rout- Kijima, Y. 2006. “Caste and Tribe Inequality: Ev-
ledge and Kegan Paul. idence from India, 1983–1999.” Economic De-
Boix, Carles. 2010. “Origins and Persistence of velopment and Cultural Change 54 (2): 369–404.
Economic Inequality.” Annual Review of Political Kristal, Tali. 2010. “Good Times, Bad Times: Post-
Science 13 (1): 489–516. war Labor’s Share of National Income in Capi-
Borooah, V. K. 2005. “Caste, Inequality, and Pov- talist Democracies.” American Sociological Review
erty in India.” Review of Development Economics 75 (5): 729–763.
9 (3): 399–414. Lamont, Michèle, and Virág Molnár. 2002. “The
Crowley, Stephen. 2004. “Explaining Labor Weak- Study of Boundaries in the Social Sciences.” An-
ness in Post-Communist Europe: Historical nual Review of Sociology 28: 167–195.
Legacies and Comparative Perspective.” Anthro- Marx, Karl. [1939] 1971. The Grundrisse. New York:
pological Theory 18 (3): 394–429. Harper & Row.
Dumont, Louis. 1970. Homo Hierarchicus. Chicago: Midlarsky, Manus. 1999. The Evolution of Inequality:
University of Chicago Press. War, State Survival, and Democracy in Compara-
Ensminger, Jean, and Jack Knight. 1997. “Changing tive Perspective. Stanford, CA: Stanford Univer-
Social Norms: Common Property, Bridewealth, sity Press.
and Clan Exogamy.” Current Anthropology 38: 1. Milanovic, B., Peter H. Lindert, and Jeffrey G. Wil-
Finley, Moses J. 1960. Slavery in Classical Antiquity. liamson. 2007. “Measuring Ancient Inequality.”
Cambridge, UK: W. Heffer and Sons. National Bureau of Economic Research, Working
Godelier, Maurice. 1978. “The Concept of the ‘Asi- Paper 13550, 1–86.
atic Mode of Production’ and Marxist Models Mushinski, D. W., and K. Pickering. 2000. “Inequal-
of Social Evolution.” In Relations of Production, ity in Income Distributions: Does Culture Matter?
Marxist Approaches to Economic Anthropology, An Analysis of Western Native American Tribes.”
edited by D. Seddon, 209–257. London: Cass. Journal of Economic Issues 34 (2): 403–412.
Gouldner, Alvin. 1980. The Two Marxisms: Con- O’Leary, Brendan. 1989. The Asiatic Mode of Produc-
tradictions and Anomalies in the Development of tion. Oxford: Basil Blackwell.
Theory. New York: Seabury Press. Patterson, Orlando. 1982. Slavery and Social Death.
Hayden, Brian. 2007. “Richman, Poorman, Beggar- Cambridge, MA: Harvard University Press.
man, Chief: The Dynamics of Social Inequal- Schröder, Ingo W. 2003. “The Political Economy of
ity.” In Archeology at the Millennium, edited Tribalism in North America: Neotribal Capital-
by G. Feinman and T. Price, 1–42. New York: ism?” Anthropological Theory 3 (4): 435–456.
Springer. Sharma, K. L. 1983. “Agrarian Stratification: Old
Heyns, Barbara. 2005. “Emerging Inequalities in Issues, New Explanations and New Issues, Old
Central and Eastern Europe.” Annual Review of Explanations.” Economic and Political Weekly 18
Sociology 31 (1): 163–197. (42): 1796–1802.
Hobsbawm, Eric J. 1965. “Introduction.” In Shaw, William H. 1978. Marx’s Theory of History.
Pre-Capitalist Economic Formations, translated Stanford, CA: Stanford University Press.
52 • Part II: Inequality in Comparative Perspective

Sio, Arnold A. 1965. “Interpretations of Slavery: The Summerhill, W. 2010. “Colonial Institutions, Slav-
Slave Status in the Americas.” Comparative Stud- ery, Inequality, and Development: Evidence
ies in Society and History 7: 289–308. from São Paulo, Brazil.” Social Science Research
Smaje, Chris. 2000. Natural Hierarchies: The His- Network, Working Paper 1591449.
torical Sociology of Race and Caste. Oxford: Basil Wu, X., and D. J. Treiman. 2007. “Inequality and
Blackwell. Equality under Chinese Socialism: The Hukou Sys-
Srinivas, Mysore N. 1962. Caste in Modern India. tem and Intergenerational Occupational Mobility.”
London: Asia Publishing House. American Journal of Sociology 113 (2): 415–445.

6 • Gøsta Esping-Andersen and John Myles

The Welfare State and Redistribution

Citizens obtain welfare from three basic sources: where the value of tax expenditures for social
markets, family, and government. The market purposes exceeds 2 percent of GDP, and where
provides income and sells commercial welfare private social spending, at 9 percent of GDP, ac-
inputs such as child minding or medical insur- counts for 37 percent of total public and private
ance. For most people, during most of their lives, outlays (Adema and Ladaique, 2005: Table 6).
the market is undoubtedly the chief source of A second hallmark of the liberal regime is
well-being. Families, too, play a pivotal role in its preference for targeting public benefits to
welfare packaging, in part by providing services the neediest, traditionally via means-testing,
and care for kin, in part via income transfers. In- but recent years have seen a shift towards
come pooling in families is the norm and income work-conditional, negative income tax type
transfers between the generations is substan- policies. This, again, mirrors the principle
tial—in particular from the elderly to the young that government’s role should be restricted to
(Albertini et.al, 2006). To fully understand wel- a demonstrable inability to uphold one’s own
fare states we need to situate them in the full livelihood.
context of welfare production and consumption. In line with its support for market solutions,
I term this welfare regime. this regime favours unregulated labor markets
With some simplification, we can distin- under the assumption that this bolsters employ-
guish three distinct regimes.1 The Anglo-Saxon ment growth. But it also promotes greater labor
countries represent the ‘liberal’ regime, one that turnover, which heightens social insecurity, and
favors minimal public intervention under the as- greater wage inequality which, in turn, increases
sumption that the majority of citizens can obtain the risks of poverty.
adequate welfare from the market. The role of The combination of a pro-market bias and
government is, in part, to nurture rather than re- targeting should, in principle, lead to contra-
place, market transactions and this explains why dictory distributional outcomes. Private welfare
these countries favor subsidizing private welfare reliance, bolstered by tax subsidies, should have
via tax deductions. The US is the extreme case a clear adverse effect on equality while we should

Gøsta Esping-Andersen and John Myles. 2008. “The Welfare State and Redistribution,” pp. 6–12, 15–18,
20–25. Used by permission of the authors.
 Esping-Andersen and Myles—The Welfare State and Redistribution • 53

expect that targeting in favor of the neediest a major equalizing effect. But since they are [al-
sponsors equality. The latter may be weakened most] all taxed, the net outcome should be more
by the ‘paradox of redistribution’ effect: social vertical redistribution (high-income retirees are
benefits to the poor is comparably ungenerous taxed more heavily than their low-income equiv-
in the Anglo countries, and this helps explain alents). And returning to the ‘paradox of redistri-
why poverty remains exceptionally high. bution’, minimal targeting should promote more
The Nordic countries represent a second, equality of outcome. Overall, to the extent that
‘social-democratic’, regime that is characterized the model delivers maximum employment, per-
by its emphasis on universal inclusion and its haps the most decisive redistribution occurs via
comprehensive definition of social entitlements. the equalization of primary, pre-redistribution in-
It has been vocally committed to equalize living comes. The de facto universality of employment
conditions across the citizenry. For this reason, should translate into exceptionally low poverty
policy has deliberately sought to marginalize the risks.
role of private welfare markets and of targeted The third, and somewhat more heteroge-
social assistance. The model is internationally neous, regime embraces the majority of Con-
unique also in its emphasis on ‘de-familializing’ tinental European countries: Austria, Belgium,
welfare responsibilities, especially with regard to France, Germany, Italy, the Netherlands, and
care for children and the elderly. Denmark and Spain. These welfare states all have conservative
Sweden now boast de facto complete coverage origins, built around social insurance, often along
both for child and elderly care. The upshot is a narrowly defined occupational distinctions. This
boost to female labor supply and an unusually implies that entitlements depend primarily on
service-intensive welfare state. Excluding health life-long employment which has, historically,
care, social services now account for more than helped cement the male-breadwinner logic of
20 percent of all social expenditure in Denmark social protection. With the partial exception
and Sweden. of Belgium and France, this regime is strongly
Another hallmark of the Nordic model is familialistic, assuming that primary welfare re-
its unique synchronization of social and labor sponsibilities lie with family members. Policies
market policies. While labor markets are fairly that help reconcile motherhood and careers are
regulated, both in terms of hiring and firing relatively undeveloped. Hence, these welfare
(with the exception of Denmark) and in terms states are transfer-heavy and service-lean.
of wage setting, social policy has played a major The role of private welfare is marginal, as in
role in promoting maximum employment. Fe- the Nordic countries, primarily because social
male employment rates are now basically similar insurance offers generous benefits and broad
to men’s and increasingly on a full-time basis. coverage of the employed population, but also
This has potentially huge consequences for the due to high prices for private services—brought
income distribution. Female employment will about by high fixed labour costs. Strong labor
promote more income equality the more it is market regulation and high labor costs have
embraced also in the bottom quintile house- influenced patterns of employment in several
holds. Active labor market policies are a second ways. Firstly, while prime age workers risk little
crucial instrument for minimizing unemploy- unemployment, entry into first employment is
ment. By guaranteeing generous benefits to made difficult. Youth unemployment is, except
participants, the policy gives huge incentives to in Germany, extraordinarily high. Secondly, the
participate in training, retraining and job relo- lack of support for working mothers means that
cation programs. female employment has remained low, especially
The net distributional impact of these fea- among less educated women. And thirdly, early
tures is difficult to predict—and measure. Since retirement has been used intensely as a vehicle
most benefits are universal, they may not have for clearing the labor market of older workers,
54 • Part II: Inequality in Comparative Perspective

thus lowering overall employment rates and rais- to a Senegalese, but will have severe difficulties
ing pension expenditure. making ends meet in the US. A second issue is
Its redistributive profile is therefore likely to where to draw the line. The latter can have mas-
diverge from any of the other two. The dom- sive repercussions, all depending on where the
inance of insurance implies an accent on hor- poor are situated.
izontal redistribution. Low employment rates, Drawing the line will, in all cases, provoke
particularly among older workers, imply how- some artificiality of assessment. There may, for
ever that redistribution will appear strong, the example, be a large population that falls just
reason being that extensive early retirement helps immediately below the line. They would be
bloat the number of zero-earners whose main in- classified as poor whereas those with just a few
come derives from pensions. Low female partici- additional Euros would not. The standard ap-
pation, especially within the lower deciles of the proach that has been adopted in most studies is
income distribution, implies also greater house- the 50-percent of median line, although the EU
hold vulnerability—which, of course, is partially has officially chosen a 60-percent line. Studies
offset by strong employment protection. But low that wish to capture the ‘poverty gap’, i.e. how
female employment means that groups like lone far below the line the poor find themselves, cal-
mothers are vulnerable. culate either a gap measure or use several poverty
There is no clear connection between these lines (Mitchell, 1991; Kenworthy, 1999).
institutional features and welfare state size (see A first step towards ascertaining welfare state
Table 6.1). The liberal regime is, unsurprisingly, effects would be to compare the level of nations’
a low spender. But notwithstanding their insti- income inequality (measured, for example, by
tutional differences, the Nordic and Continental the Gini coefficient) between ‘original’ market
regimes basically converge in terms of spending. incomes and post-redistribution, disposable
Regime distinctions are more accentuated when incomes. Few studies actually adopt this ap-
we examine the role of private welfare, social proach due primarily to the huge problem of
servicing intensity, and degree of targeting. We zero market income populations among the el-
can also capture differences by examining the derly. Those that do find routinely that the large
generosity and distribution of transfer incomes welfare states are more redistributive (Forster,
to households. In the Nordic countries, aver- 2000). Following the welfare regime approach,
age transfer income (for active age households) it is clear that the Anglo-Saxon group is sys-
equals 18 percent of median gross income, com- tematically far less redistributive than the other
pared to 10 percent in Germany and 14 percent two which, in turn, appear surprisingly similar
in Britain. The targeted profile of Britain’s wel- within such aggregate comparisons. But when
fare state emerges clearly when we compare the one excludes the elderly population, the conver-
ratio of transfer income in the lowest and highest gence disappears and the Continental European
income quintile (4.6:1). The Swedish ratio, in welfare states begin to look more ‘liberal’. This
comparison, is 2.5:1 (Fritzell, 2000). suggests an important difference in welfare state
A huge amount of research has focused on design. The Continental European countries’
welfare states’ effectiveness in reducing poverty tax-transfer system is far more pensioner-biased,
and this implies the need to define a meaningful as is also clear in Dang et al. (2006). Such ag-
poverty line. One hotly debated issue is whether gregate redistribution data confirm in another
to adopt a relative or ‘absolute’ measure. The rea- way the distinctiveness of welfare regimes since
son that the vast majority prefer the former is to intra-regime similarities in redistribution are
capture the meaning of being poor within the overall surprisingly strong—the only real excep-
society one lives in. An American family with a tion being Italy which looks more similar to the
$10,000 annual income might appear very rich Anglo-Saxon group.
 Esping-Andersen and Myles—The Welfare State and Redistribution • 55

Table 6.1. Expenditure Profiles in Three Welfare Regimes

Public Social Private as % Non-health Services Targeting: %
Spending Total Social as % of Total of Transfers to
( % GDP)1 Spending Public Spending Bottom Quintile2
Nordic 25 5 18 34
Anglo 19 19 4 43
Continental Europe 26 8 5 30
Source: calculations from Adema and Ladaique (2005: Table 6) and from Forster and d’Ercole (2005).
1
Data refer to net social spending.
2
Excludes retired households.

A major shortcoming of cross-national com- seriously adverse consequences for later outcomes
parisons of this type is that they provide only such as schooling, health and social integration
snapshots and fail, accordingly, to capture how (Duncan and Brooks-Gunn, 1997; Vleminckx
welfare states respond to changes in economic and Smeeding, 2001). A welfare state’s invest-
inequality. Market incomes have, in the majority ment in children’s well-being is a potentially very
of OECD countries, experienced a strong ine- powerful tool for promoting greater equality of
galitarian trend over the past decades (Atkinson, opportunities (Esping-Andersen, 2007b).
2003). In some countries, like the UK and the Virtually all studies conclude similarly that
US, the inegalitarian tide has been quite dra- poverty reduction, in particular among families
matic. This raises the question as to the welfare with children, is closely associated with levels
state’s ability to stem, or even roll-back, the tide of social expenditure. The Nordic countries are
via an added redistributive effort. Research, so typically the most redistributive while the Anglo
far, suggests that in most OECD countries the Saxon (especially the US) are the least (McFate
jump in market inequalities has not been paral- et.al., 1995; Jantti and Danziger, 2000; Bradbury
leled by a similar jump for disposable incomes and Jantti, 2001; Smeeding, 2005a and 2005b).
(Esping-Andersen, 2007a). ‘Stemming the tide’ Smeeding (2005a: Figure 1) shows a correlation
should provide a revealing indicator of a welfare of .6 between levels of cash social spending and
state’s commitment to equality. Kenworthy and poverty among non-elderly households. Corak
Pontusson’s (2005: Figure 7) data suggest that (2005) shows, similarly, a strong correlation
the UK and the US have done the least to stem between welfare state spending on families and
the tide, whereas Germany excels in the other poverty reduction among children.
direction. The Nordic countries have increased The country differences are in fact far more
redistribution in perfect tandem with rising accentuated than when we use aggregate redis-
inequality. tribution indicators like the Gini coefficient. At
Poverty reduction is arguably the single most one extreme we find, once again, the US where
relevant measure of welfare state redistribution government redistribution has only a very minor
and, unsurprisingly, it has become the favoured impact on child poverty and at the other extreme
approach in empirical research. Theoretically, it we find Sweden where child poverty almost dis-
provides a good test of the Rawlsian maximin appears. In Sweden transfers account for almost
principle of justice, namely that any redistribu- 70 percent of income in poor households, com-
tion should be to the greatest benefit of the worst pared to little more than 30 percent in the US;
off. It also speaks most directly to vertical, Robin in the US, the primary policy vehicle is social
Hood–redistribution. For two reasons, research assistance; in Sweden, targeted assistance plays a
has especially centered on child poverty. One peripheral role (Rainwater and Smeeding, 2003;
is that poverty in childhood is known to have Smeeding, 2005b: Table 8).
56 • Part II: Inequality in Comparative Perspective

Table 6.2. Poverty Reduction in Families with Children, Mid-1990s1

Post- Percent
Redistribution Reduction of
Market Poverty Poverty Poverty
Denmark 30 6 80
Finland 18 3 83
Norway 29 5 83
Sweden 39 4 90
Regime Mean 29 5 84
Australia 32 17 47
Canada 29 16 45
Ireland 28 15 46
UK 39 21 46
US 31 26 16
Regime Mean 32 19 40
Belgium 31 6 81
France 40 10 75
Germany 31 12 61
Italy 37 21 43
Netherlands 25 8 68
Spain 30 13 57
Regime Mean 32 12 64
Source: LIS-based estimates, from Bradbury and Jantti (2001: 83).
1
Poverty is less than 50% of median equivalent income.

Maitre et.al. (2005) adopt a welfare regime again, quite heterogeneous. Italy’s performance
approach similar to ours and find a distinctly lies closer to the Anglo-Saxon regime.
strong, respectively weak, redistributive in- Turning to elderly poverty, the most basic
cidence in the Nordic, respectively Southern problem is that welfare states typically compel
European, regimes when examining poverty re- the elderly to retire, thus forcing upon this pop-
duction across all households. Table 6.2 presents ulation a prevalence of zero or very low market
data on poverty reduction in child families. earnings. We must assume that this varies con-
A remarkable feature is that pre-redistribution siderably across countries for a number of rea-
poverty rates vary little between countries. In sons including normal age of retirement, rules
terms of market incomes there is, in fact, more regarding pension entitlements and continued
child poverty in Sweden than in the US. This earnings, and the prevalence of wives’ incomes.
reflects, of course, the artificiality of the ‘orig- In studies that focus on old age poverty, the bias
inal’ income distribution. The universality of that is created by women’s employment patterns
paid maternity leave in Sweden and the lack is especially acute since the large majority of poor
thereof in the US implies a large number of elderly households are composed of widows with
zero-earner mothers in Sweden compared to the meager pension savings. A second major prob-
US. The regime differences emerge when we lem lies in the public-private mix of retirement
shift attention to redistribution: a very homo- savings and, in particular, in how welfare states’
geneous and strong redistributive effect in the encourage private pension plans.
Nordic countries. The Anglo group is, with the Lefebvre (2007) presents a rare attempt to
US as an extreme case, homogeneously very little disentangle all such effects cross-nationally, and
redistributive-only half as much as the Nordic. finds that poverty is the norm when we examine
The Continental European countries are, once pre-transfer poverty, ranging from 90+ percent
 Esping-Andersen and Myles—The Welfare State and Redistribution • 57

in Austria, Belgium, and France to a low of ———. 2003 ‘Income inequality in OECD coun-
54 percent in Finland. More to the point, he tries: data and explanations’. CESifo Economic
decomposes the marginal effect of earnings, Studies, 49: 479–513.
property income, private pensions and public Bradbury, B., and M. Jäntti. 2001. Child Poverty
Across Twenty-five Countries. The Dynamics of
transfers on final inequality. Earnings contrib-
Child Poverty in Industrialised Countries, 62–69.
ute very importantly to inequality, in particular Bradbury, B., Jenkins, S., and Micklewright, J. 2001
because the highly skilled are more likely to con- The Dynamics of Child Poverty in Industrialized
tinue working. This suggests that compulsory Countries. Cambridge: Cambridge University
retirement reduces inequalities. Property income Press.
and private pensions are, likewise, a source of Corak, M. 2005 ‘Principles and practicalities for
inequality. Their impact is minor in most Euro- measuring child poverty in rich countries’. LIS
pean countries, but potentially large in the US. Working Paper, 406.
All this considered, it nonetheless remains that Dang, T. Immervoll, H., Mantovani, D., Orsini,
K. and Sutherland, H. 2006 ‘An age perspective
government transfers account for most of the
on economic well-being and social protection in
post-redistribution inequality among the elderly. nine OECD countries’. IZA Working Paper, 2173.
Lefebvre finally concludes, like Korpi and Palme Duncan, G. J., and J. Brooks-Dunn. 1997. Con-
(1998), that welfare states that stress targeting to sequences of Growing Up Poor. Russell Sage
the poor elderly are far less redistributive than Foundation.
are comprehensive (and therefore generous) pen- Duncan, G., Gustafsson, B., Hauser, R., Schmaus,
sion programs (Lefebvre, 2007: 10). G., Muffels, R., Nolan, B. and Ray, J. 1993 ‘Pov-
So far we only have simulations that exam- erty dynamics in eight countries’. Journal of Pop-
ine specific components of the welfare state, ulation Economics, 6: 295–334.
Esping-Andersen, G. 1990 The Three Worlds of Wel-
such as family benefits systems. The great and,
fare Capitalism. Cambridge: Polity Press.
as yet, unmet challenge would be to adapt this
———. 1999 Social Foundations of Postindustrial
methodology to address the equalizing effects Economies. Oxford: Oxford University Press.
of welfare states as such and, in particular, to ———. 2007a ‘More inequality and fewer oppor-
attack the counter-factual problem. Until we tunities? Structural determinants and human
have a more precise idea of how, and how much, agency in the dynamics of income distribution’.
market-based inequalities are patterned by the Pp. 216–51 in D. Held and A. Kaya, eds. Global
welfare state we are essentially unable to answer Inequality. Cambridge: Polity Press.
the question posed in this chapter. ———. 2007b ‘Childhood investments and skill
formation’. International Tax and Finance
NOTES Journal.
1. This section draws heavily on Esping-Andersen Forster, M. 2000 ‘Trends and driving forces in in-
(1990; 1999). come distribution and poverty in the OECD
area’. OECD Labor Market and Social Policy Oc-
casional Papers, 42.
REFERENCES
Forster, M. and d’Ercole, M. 2005 ‘Income distri-
Adema, W., and Ladaique, M. 2005. Net Social
bution and poverty in OECD Countries in the
Expenditure, 2005 Edition: More Comprehen-
second half of the 1990s’. OECD Social, Employ-
sive Measures of Social Support. No. 29. OECD
ment and Migration Working Paper, 22.
Publishing.
Fritzell, J. 2000 ‘Still different? Income distribution
Albertini, M., M. Kohli, and C. Vogel. 2007. “Inter-
in the Nordic countries in a European compari-
national Transfers of Time and Money in Euro-
son’. LIS Working Paper, 238.
pean Families.” Journal of European Social Policy
Jantti, M. and Danziger, S. 2000. ‘Income poverty in
117: 319–334.
advanced countries’ in A. Atkinson and F. Bour-
Atkinson, A.B. 1995 ‘The welfare state and eco-
guignon, eds. Handbook of Income Distribution.
nomic performance’. National Tax Journal, 48:
Amsterdam: North Holland.
171–98.
58 • Part II: Inequality in Comparative Perspective

Kenworthy, L. 1999 ‘Do social welfare policies re- system effectiveness in the 1980s’. Pp. 67–108
duce poverty?’ Social Forces, 77: 1119–39. in K. McFate, R. Lawson and W. J. Wilson, eds.
Kenworthy, L. and Pontusson, J. 2005 ‘Rising in- Poverty, Inequality and the Future of Social Policy.
equality and the politics of redistribution in afflu- New York: Russell Sage.
ent countries’. Perspectives on Politics, 3: 449–71. Mitchell, D. 1991 Income Transfers in Ten Welfare
Korpi, W., and J. Palme. 1998. “The Paradox of Re- States. Sydney: Avebury.
distribution and Strategies of Equality.” American Rainwater, L. and Smeeding, T. 2003 Poor Kids in a
Sociological Review 63: 661–687. Rich Country. New York: Russell Sage.
Lefebvre, M. 2007 ‘The redistributive effects of pen- Smeeding, T. 2005a ‘Government programs and so-
sion systems in Europe: a survey of the evidence’. cial outcomes: The United States in comparative
LIS Working Paper, 457. perspective’. LIS Working Paper, 426.
Maitre, B., Nolan, B. and Whelan, C. 2005 ‘Welfare ———. 2005b. ‘Poor people in rich nations: the
regimes and household income packaging in the United States in comparative perspective’. LIS
European Union’. Journal of European Social Pol- Working Paper, 419.
icy, 15:157–71. Vleminckx, K. and Smeeding, T. 2001 Child
McFate, K., Smeeding, T. and Rainwater, L. 1995 Well-being, Child Poverty, and Child Policy in
‘Markets and states: poverty trends and transfer Modern Nations. Bristol: Policy Press.
 Atkinson, Piketty, and Saez—Top Incomes in the Long Run of History • 59

Trends in Economic Inequality

7 • Anthony B. Atkinson, Thomas Piketty, and Emmanuel Saez

Top Incomes in the Long Run of History

There has been a marked revival of interest in the We focus on the data series produced in this
study of the distribution of top incomes using project on the grounds that they are fairly ho-
income tax data. mogenous across countries, annual, long-run,
In contrast to existing international data- and broken down by income source for most
bases, generally restricted to the post-1970 or countries. They cover twenty-two countries,
post-1980 period, the top income data cover a [and] cover periods that range from 15 years
much longer period, which is important because (China) and 30 years (Italy) to 120 years (Japan)
structural changes in income and wealth distri- and 132 years (Norway). Hence they offer a
butions often span several decades. In order to unique opportunity to better understand the
properly understand such changes, one needs dynamics of income and wealth distribution and
to be able to put them into broader historical the interplay between inequality and growth.
perspective. The new data provide estimates The complete database is available online at the
that cover much of the twentieth century and in Paris School of Economics at http://g-mond.
some cases go back to the nineteenth century—a parisschoolofeconomics.eu/topincomes/.
length of time familiar to economic historians We obtain three main empirical results. First,
but unusual for most economists. Moreover, the most countries experienced a sharp drop in top
tax data typically allow us to decompose income income shares in the first half of the twentieth
inequality into labor income and capital income century. In these countries, the fall in top income
components. Economic mechanisms can be very shares is often concentrated around key episodes
different for the distribution of labor income such as the World Wars or the Great Depression.
(demand and supply of skills, labor market insti- In some countries however, especially those that
tutions, etc.) and the distribution of capital in- stayed outside World War II, the fall is more
come (capital accumulation, credit constraints, gradual during the period. In all countries for
inheritance law and taxation, etc.), so that it is which income composition data are available, in
difficult to test these mechanisms using data on the first part of the century top percentile in-
total incomes. comes were overwhelmingly composed of capital

Anthony B. Atkinson, Thomas Piketty, and Emmanuel Saez. “Top Incomes in the Long Run of History,”
Journal of Economic Literature 49:1 (2011), pp. 3–5, 40–46, 48–50. Used by permission of the American
Economic Association. Emmanuel Saez. “Striking It Richer: The Evolution of Top Incomes in the United
States (Updated with 2009 and 2010 Estimates),” updated from Pathways 2008, http://elsa.berkeley.edu/~saez/
saez-UStopincomes-2010.pdf, pp. 1–10. Used by permission of the Center on Poverty and Inequality, Stanford
University.
60 • Part II: Inequality in Comparative Perspective

income (as opposed to labor income). Therefore,

the fall in the top percentile share is primarily a Striking It Richer: Findings from the
capital income phenomenon: top income shares United States*
fall because of a reduction in top wealth concen- We begin with the U.S. case. The recent dra-
tration. In contrast, upper income groups below matic rise in income inequality in the United
the top percentile such as the next 4 percent or States is well documented. But we know less
the second vingtile, which are comprised primar- about which groups are winners and which are
ily of labor income, fall much less than the top losers, or how this may have changed over time.
percentile during the first half of the twentieth Is most of the income growth being captured by
century. an extremely small income elite? Or is a broader
By 1949, the dispersion in top percentile upper middle class profiting? And are capital-
income shares across the countries studied had ists or salaried managers and professionals the
become small. In the second half of the twen- main winners? We explore these questions with a
tieth century, top percentile shares experienced uniquely long-term historical view that allows us
a U-shape pattern, with further declines during to place current developments in deeper context
the immediate postwar decades followed by in- than is typically the case.
creases in recent decades. However, the degree of Efforts at analyzing long-term trends are
the U-shape varies dramatically across countries. often hampered by a lack of good data. In the
In all of the Western English speaking countries United States, and most other countries, house-
(in Europe, North America, and Australia and hold income surveys virtually did not exist prior
New Zealand), and in China and India, there to 1960. The only data source consistently avail-
was a substantial increase in top income shares able on a long-run basis is tax data. The U.S.
in recent decades, with the United States leading government has published detailed statistics on
the way both in terms of timing and magnitude income reported for tax purposes since 1913,
of the increase. Southern European countries when the modern federal income tax started.
and Nordic countries in Europe also experience These statistics report the number of taxpay-
an increase in top percentile shares although ers and their total income and tax liability for
less in magnitude than in English speaking a large number of income brackets. Combining
countries. In contrast, Continental European these data with population census data and ag-
countries (France, Germany, Netherlands, Swit- gregate income sources, one can estimate the
zerland) and Japan experience a very flat U-shape share of total personal income accruing to var-
with either no or modest increases in top income ious upper-income groups, such as the top 10
shares in recent decades. percent or top 1 percent.
Third, as was the case for the decline in the We define income as the sum of all income
first half of the century, the increase in top in- components reported on tax returns (wages and
come shares in recent decades has been quite salaries, pensions received, profits from busi-
concentrated with most of the gains accruing to nesses, capital income such as dividends, inter-
the top percentile with much more modest gains est, or rents, and realized capital gains) before
(or even none at all) for the next 4 percent or the individual income taxes. We exclude government
second vingtile. However, in most countries, a transfers such as Social Security retirement ben-
significant portion of the gains are due to an in- efits or unemployment compensation benefits
crease in top labor incomes, and especially wages from our income definition. Non-taxable fringe
and salaries. As a result, the fraction of labor in- benefits such as employer provided health insur-
come in the top percentile is much higher today ance is also excluded from our income definition.
in most countries than earlier in the twentieth Therefore, our income measure is defined as cash
century. market income before individual income taxes.
 Atkinson, Piketty, and Saez—Top Incomes in the Long Run of History • 61

Figure 7.1. The Top Decile Income Share, 1917–2010 United States

50%

45%
Top 10% Income Share

40%

35%

30%
Including capital gains
Excluding capital gains
25%
1917
1922
1927
1932
1937
1942
1947
1952
1957
1962
1967
1972
1977
1982
1987
1992
1997
2002
2007
Income is defined as market income (and excludes government transfers). In 2010, top decile includes all families with
annual income above $108,000.

Figure 7.1 presents the income share of the surpasses 1928, the peak of stock market bubble
top decile from 1917 to 2010 in the United in the “roaring” 1920s. In 2010, the top decile
States. In 2010, the top decile includes all share is equal to 47.9 percent.
families with market income above $108,000. Figure 7.2 decomposes the top decile into
The overall pattern of the top decile share over the top percentile (families with income above
the century is U-shaped. The share of the top $352,000 in 2010) and the next 4 percent
decile is around 45 percent from the mid-1920s (families with income between $150,000 and
to 1940. It declines substantially to just above $352,000 in 2010), and the bottom half of
32.5 percent in four years during World War the top decile (families with income between
II and stays fairly stable around 33 percent $108,000 and $150,000 in 2010). Interestingly,
until the 1970s. Such an abrupt decline, con- most of the fluctuations of the top decile are due
centrated exactly during the war years, cannot to fluctuations within the top percentile. The
easily be reconciled with slow technological drop in the next two groups during World War
changes and suggests instead that the shock of II is far less dramatic, and they recover from the
the war played a key and lasting role in shap- WWII shock relatively quickly. Finally, their
ing income concentration in the United States. shares do not increase much during the recent
After decades of stability in the post-war period, decades. In contrast, the top percentile has gone
the top decile share has increased dramatically through enormous fluctuations along the course
over the last twenty-five years and has now re- of the twentieth century, from about 18 percent
gained its pre-war level. Indeed, the top decile before WWI, to a peak of almost 24 percent in
share in 2007 is equal to 49.7 percent, a level the late 1920s, to only about 9 percent during
higher than any other year since 1917 and even the 1960s-1970s, and back to almost 23.5
62 • Part II: Inequality in Comparative Perspective

Figure 7.2. Decomposing the Top Decile U.S. Income Share into 3 Groups, 1913–2010

Income is defined as market income including capital gains. Top 1% denotes the top percentile (families with annual
income above $352,000 in 2010). Top 5–1% denotes the next 4% (families with annual income between $150,000 and
$352,000 in 2010). Top 10–5% denotes the next 5% (bottom half of the top decile, families with annual income between
$108,000 and $150,000 in 2010).

percent by 2007. Those at the very top of the in- growth of real incomes per family over the pe-
come distribution therefore play a central role in riod 1993–2010.
the evolution of U.S. inequality over the course The 1993–2010 period encompasses, how-
of the twentieth century. ever, a dramatic shift in how the bottom 99 per-
The implications of these fluctuations at cent of the income distribution fared. Table 7.1
the very top can also be seen when we examine next distinguishes between five sub-periods: (1)
trends in real income growth per family between the 1993–2000 expansion of the Clinton ad-
the top 1 percent and the bottom 99 percent in ministrations, (2) the 2000–2002 recession, (3)
recent years, as illustrated in Table 7.1. From the 2002–2007 expansion of the Bush admin-
1993 to 2010, for example, average real incomes istrations, (4) the 2007–2009 Great Recession,
per family grew by only 13.8% over this 17 (5) and 2009–2010, the first year of recovery.
year period (implying an annual growth rate of During both expansions, the incomes of the top
.76%). However, if one excludes the top 1 per- 1 percent grew extremely quickly by 98.7% and
cent, average real incomes of the bottom 99% 61.8% respectively. However, while the bottom
grew only by 6.4% from 1993 to 2010 (imply- 99 percent of incomes grew at a solid pace of
ing an annual growth rate of .37%). Top 1 per- 20.3% from 1993 to 2000, these incomes grew
cent incomes grew by 58% from 1993 to 2010 only 6.8% percent from 2002 to 2007. As a re-
(implying a 2.7% annual growth rate). This sult, in the economic expansion of 2002–2007,
implies that top 1 percent incomes captured the top 1 percent captured two thirds of income
slightly more than half of the overall economic growth. Those results may help explain the
 Atkinson, Piketty, and Saez—Top Incomes in the Long Run of History • 63

Table 7.1. Real Income Growth by Groups, 1993–2010, United States

Top 1% Bottom 99% Fraction of Total Growth
Average Income Incomes Real Incomes Real (or Loss) Captured by
Real Growth Growth Growth Top 1%
(1) (2) (3) (4)
Full Period
13.8% 58.0% 6.4% 52%
1993–2010
Clinton Expansion
31.5% 98.7% 20.3% 45%
1993–2000
2001 Recession
-11.7% -30.8% -6.5% 57%
2000–2002
Bush Expansion
16.1% 61.8% 6.8% 65%
2002–2007
Great Recession
-17.4% -36.3% -11.6% 49%
2007–2009
Recovery
2.3% 11.6% 0.2% 93%
2009–2010
Notes: Computations based on family market income including realized capital gains (before individual taxes).
Incomes exclude government transfers (such as unemployment insurance and social security) and non-taxable fringe
benefits. Incomes are deflated using the Consumer Price Index. Column (4) reports the fraction of total real family
income growth (or loss) captured by the top 1%. For example, from 2002 to 2007, average real family incomes grew
by 16.1 % but 65% of that growth accrued to the top 1% while only 35% of that growth accrued to the bottom 99%
of US families. From 2009 to 2010, average real family incomes increased by 2.3% and the top 1% captured 93% of
those gains.
Source: Piketty and Saez (2003), series updated to 2010 in March 2012 using IRS tax statistics.

disconnect between the economic experiences of is the largest fall on record in any two year period
the public and the solid macroeconomic growth since the Great Depression of 1929–1933.
posted by the U.S. economy from 2002 to 2007. From 2009 to 2010, average real income per
Those results may also help explain why the dra- family grew by 2.3% (Table 7.1) but the gains
matic growth in top incomes during the Clinton were very uneven. Top 1% incomes grew by
administration did not generate much public 11.6% while bottom 99% incomes grew only
outcry while there has been a great attention to by 0.2%. Hence, the top 1% captured 93%
top incomes in the press and in the public debate of the income gains in the first year of recov-
since 2005. ery.1 Such an uneven recovery can possibly ex-
During both recessions, the top 1 percent in- plain the recent public demonstrations against
comes fell sharply, by 30.8% from 2000 to 2002, inequality.
and by 36.3% from 2007 to 2009. The primary The top percentile share declined during
driver of the fall in top incomes during those re- WWI, recovered during the 1920s boom, and
cessions is the stock market crash which reduces declined again during the great depression and
dramatically realized capital gains, and, especially WWII. This very specific timing, together with
in the 2000–2002 period, the value of executive the fact that very high incomes account for a
stock-options. However, bottom 99 percent in- disproportionate share of the total decline in
comes fell by 11.6% from 2007 to 2009 while inequality, strongly suggests that the shocks in-
they fell only by 6.5 percent from 2000 to 2002. curred by capital owners during 1914 to 1945
Therefore, the top 1 percent absorbed a larger (depression and wars) played a key role.2 Indeed,
fraction of losses in the 2000–2002 recession from 1913 and up to the 1970s, very top in-
(57%) than in the Great Recession (49%). The comes were mostly composed of capital income
11.6 percent fall in bottom 99 percent incomes (mostly dividend income) and to a smaller
64 • Part II: Inequality in Comparative Perspective

extent business income, the wage income share in the United States show a precipitous decline
being very modest. Therefore, the large decline in the first part of the century with only fairly
of top incomes observed during the 1914–1960 modest increases in recent decades. The evidence
period is predominantly a capital income suggests that top incomes earners today are not
phenomenon. “rentiers” deriving their incomes from past
Interestingly, the income composition pat- wealth but rather are “working rich,” highly paid
tern at the very top has changed considerably employees or new entrepreneurs who have not
over the century. The share of wage and salary yet accumulated fortunes comparable to those
income has increased sharply from the 1920s accumulated during the Gilded Age. Such a pat-
to the present, and especially since the 1970s. tern might not last for very long. The drastic cuts
Therefore, a significant fraction of the surge in of the federal tax on large estates could certainly
top incomes since 1970 is due to an explosion of accelerate the path toward the reconstitution of
top wages and salaries. Indeed, estimates based the great wealth concentration that existed in the
purely on wages and salaries show that the share U.S. economy before the Great Depression.
of total wages and salaries earned by the top 1 The labor market has been creating much
percent wage income earners has jumped from more inequality over the last thirty years, with
5.1 percent in 1970 to 12.4 percent in 2007.3 the very top earners capturing a large fraction
Evidence based on the wealth distribution is of macroeconomic productivity gains. A num-
consistent with those facts. Estimates of wealth ber of factors may help explain this increase in
concentration, measured by the share of total inequality, not only underlying technological
wealth accruing to top 1 percent wealth holders, changes but also the retreat of institutions de-
constructed by Wojciech Kopczuk and myself veloped during the New Deal and World War
from estate tax returns for the 1916–2000 period II—such as progressive tax policies, powerful

Figure 7.3. The Top 0.01 % Income Share, 1913–2010

6%
Including capital gains
5%
Excluding capital gains
Top 0.01% Income Share

4%

3%

2%

1%

0%
1913
1918
1923
1928
1933
1938
1943
1948
1953
1958
1963
1968
1973
1978
1983
1988
1993
1998
2003
2008

Income is defined as market income including (or excluding) capital gains. In 2010, top .01% includes the 15,617 top
families with annual income above $7,890,000.
 Atkinson, Piketty, and Saez—Top Incomes in the Long Run of History • 65

unions, corporate provision of health and re- geographical proximity but also on proximity of
tirement benefits, and changing social norms the historical evolution of top income shares. In
regarding pay inequality. We need to decide as a all cases, we have used series excluding realized
society whether this increase in income inequal- capital gains (as only a subset of countries in
ity is efficient and acceptable and, if not, what the present series including capital gains, and
mix of institutional and tax reforms should be in those cases, series excluding capital gains have
developed to counter it. also been produced). We have used the same
y-axis scale in all four figures to facilitate com-
parisons across figures. Western English speaking
The Comparative Findings countries in Figure 7.4 display a clear U-shape
We now turn to the comparative data on income over the century. Continental Central European
inequality. We depict the annual top 1 percent countries and Japan in Figure 7.5 display an
share of total gross income for twenty-two indi- L-shape over the century. Nordic and Southern
vidual countries grouped in Figures 7.4–7.7 as European countries display a pattern in between
follows: Figure 7.4—Western English speaking a U and a L shape in Figure 7.6, as the drop in
countries (United States, Canada, United King- the early part of the period is much more pro-
dom, Ireland, Australia, New Zealand); Figure nounced than the rebound in the late part of pe-
7.5—Continental Central European countries riod. Finally, developing countries in Figure 7.7
(France, Germany, Netherlands, Switzerland) also display a U/L shape pattern, although there
and Japan; Figure 7.6—Nordic European coun- is substantial heterogeneity in this group.
tries (Norway, Sweden, Finland) and Southern Let us summarize first the evidence in the
European countries (Portugal, Spain, Italy); and middle of the twentieth century. The first col-
Figure 7.7—Developing countries (China, India, umns in Table 7.2 show the position in 1949
Singapore, Indonesia, Argentina). As we shall see, (1950).4 We take this year as one for which
the grouping is made not only on cultural or we have estimates for all except four of the

Figure 7.4. Top 1 Percent Share: English Speaking Countries (U-shaped), 1910–2005
30

United States United Kingdom

25 Canada Australia
Ireland New Zealand
Top percentile share (in percent)

20

15

10

0
1910

1915

1920

1925

1930

1935

1940

1945

1950

1955

1960

1965

1970

1975

1980

1985

1990

1995

2000

2005

Source: Atkinson and Piketty (2007, 2010).

66

Figure 7.5. Top 1 Percent Share: Middle Europe and Japan (L-shaped), 1900–2005
30

France Germany

25 Netherlands Switzerland
Japan
Top percentile share (in percent)

20

15

10

0
1900

1905

1910

1915

1920

1925

1930

1935

1940

1945

1950

1955

1960

1965

1970

1975

1980

1985

1990

1995

2000

2005
Source: Atkinson and Picketty (2007, 2010).

Figure 7.6. Top 1 Percent Share: Nordic and Southern Europe (U/L-shaped), 1900–2006
30

Sweden Finland Norway

25
Spain Portugal Italy
Top percentile share (in percent)

20

15

10

0
1900

1905

1910

1915

1920

1925

1930

1935

1940

1945

1950

1955

1960

1965

1970

1975

1980

1985

1990

1995

2000

2005

Source: Atkinson and Picketty (2007, 2010).

 Atkinson, Piketty, and Saez—Top Incomes in the Long Run of History • 67

Figure 7.7. Top 1 Percent Share: Developing Countries, 1920–2005

30

China Indonesia
Argentina India
25
Singapore
Top percentile share (in percent)

20

15

10

0
1920

1925

1930

1935

1940

1945

1950

1955

1960

1965

1970

1975

1980

1985

1990

1995

2000

2005
Source: Atkinson and Picketty (2007, 2010).

twenty-two countries, and as one when most Argentina, and (colonial) Indonesia. The top 1
countries had begun to return to normality after percent is of course just one point on the distri-
the Second World War (for Germany and the bution. If we look at the top 0.1 percent, shown
Netherlands we take 1950). Moreover, it was be- in Table 7.2 for eighteen countries (Portugal re-
fore the 1950–51 commodity price boom that placing Finland), then we find that again twelve
affected top shares in Australia, New Zealand, lie within a (± 20 percent) range around 3.25
and Singapore. percent from 2.6 to 3.9 percent. Leaving out the
If we start with the top 1 percent—the group three outliers at each end, the top 0.1 percent
on which attention is commonly focused and had between twenty-six and thirty-nine times
which is depicted on Figures 7.4–7.7—then we the average income.
can see from Table 7.2 that the shares of total We also report in Table 7.2 the inverse Pareto-
gross income are strikingly similar when we take Lorenz coefficients ȕ associated to the upper
account of the possible margins of error. There tail of the observed distribution in the various
are eighteen countries for which we have esti- countries in 1949 and 2005. ȕ measures the
mates. If we take 10 percent as the central value average income of people above y, relative to y
(the median is in fact around 10.8), then twelve and provides a direct intuitive measure of the
of the eighteen lie within the range 8 to 12 per- fatness of the upper tail of the distribution.
cent (i.e., with an error margin of ± 20 percent). Coming back to 1949, we find that ten of the
In countries as diverse as India, Norway, France, twenty countries for which y coefficient values
New Zealand, and the United States, the top 1 are shown in Table 7.2 lie between 1.88 and
percent had on average between eight to twelve 2.00 in 1949. Countries as different as Spain,
times average income. Three countries were only Norway, the United States, and (colonial) Sin-
just below 8 percent: Japan, Finland, and Swe- gapore had Pareto coefficients that differed only
den. The countries above the range were Ireland, in the second decimal place. As of 1949, the
68 • Part II: Inequality in Comparative Perspective

Table 7.2. Comparative Top Income Shares

around around
 
Share of Share of G Share of Share of G
Top 1% Top 0.1% Coefficient Top 1% Top 0.1 % Coefficient
Indonesia 19.87 7.03 2.22
Argentina 19.34 7.87 2.56 16.75 7.02 2.65
Ireland 12.92 4.00 1.96 10.30 2.00
Netherlands 12.05 3.80 2.00 5.38 1.08 1.43
India 12.00 5.24 2.78 8.95 3.64 2.56
Germany 11.60 3.90 2.11 11.10 4.40 2.49
United Kingdom 11.47 3.45 1.92 14.25 5.19 2.28
Australia 11.26 3.31 1.88 8.79 2.68 1.94
United States 10.95 3.34 1.94 17.42 7.70 2.82
Canada 10.69 2.91 1.77 13.56 5.23 2.42
Singapore 10.38 3.24 1.98 13.28 4.29 2.04
New Zealand 9.98 2.42 1.63 8.76 2.51 1.84
Switzerland 9.88 3.23 2.06 7.76 2.67 2.16
France 9.01 2.61 1.86 8.73 2.48 1.83
Norway 8.88 2.74 1.96 11.82 5.59 3.08
Japan 7.89 1.82 1.57 9.20 2.40 1.71
Finland 7.71 1.63 7.08 2.65 2.34
Sweden 7.64 1.96 1.69 6.28 1.91 1.93
Spain 1.99 8.79 2.62 1.90
Portugal 3.57 1.94 9.13 2.26 1.65
Italy 9.03 2.55 1.82
China 5.87 1.20 1.45
Notes:
(1) 1939 for Indonesia, 1943 for Ireland, 1950 for Germany and the Netherlands, 1954 for Spain.
(2) 1995 for Switzerland, 1998 for Germany, 1999 for Netherlands, 1999–2000 for India, 2000 for Canada and
Ireland, 2002 for Australia, 2003 for Portugal, 2004 for Argentina, Italy, Norway and Sweden.
(3) ȕ coefficients are calculated using share of top 0.1 percent in top 1 percent (see table 13A.24 in Atkinson
and Piketty 2010), with the following exceptions: (i) ȕ coefficient for Finland in 1949 calculated using share of
top 1 percent in top 5 percent; (ii) ȕ coefficient for Spain in 1949 calculated using share of top 0.01 percent in
top 0.05 percent; (iii) ȕ coefficient for Portugal in 1949 calculated using share of top 0.01 percent in top 0.1
percent; (iv) ȕ coefficient for Ireland in 2000 calculated using share of top 0.5 percent in top 1 percent.
Source: Atkinson and Picketty (2007, 2010).

only countries with y coefficients above 2.5 were thirteen and fifty-six times the average income
Argentina and India. (in 1949 these figures had been twenty and
1949 is of interest not just for being fifty-two). In terms of the ȕ coefficients, only
mid-century but also because later years did not four of the twenty-two countries had values be-
exhibit the degree of similarity described above. tween 1.88 and 2.00. Of the countries present in
The right-hand part of Table 7.2 assembles es- 1949, five now have values of ȕ in excess of 2.5.
timates for 2005 (or a close year). The central
value for the share of the top 1 percent is not too
different from that in 1949: 9 percent. But we Before 1949
now find more dispersion. For the top 1 percent, Before examining the recent period in detail, we
nine out of twenty-one countries lie outside the look at the first half of the century (and back
range of ± 20 percent. Leaving out the two outli- into the nineteenth century). What happened
ers at each end, the top 0.1 percent had between before 1949 is relevant for several reasons. The
 Atkinson, Piketty, and Saez—Top Incomes in the Long Run of History • 69

behavior of the income distribution in today’s of unequal America is only suggestive and in-
rich countries may provide a guide as to what complete” (p. 192). Using large samples of Pa-
can be expected in today’s fast-growing econ- risian and national estate tax returns over the
omies. We can learn from nineteenth-century 1807–1994 period, Piketty, Gilles Postel-Vinay,
data, such as those for Norway or Japan, that and Jean-Laurent Rosenthal (2006) find that
cover the period of industrialization. Events in wealth concentration rose continuously during
today’s world economy may resemble those in the 1807–1914 period (with an acceleration of
the past. If we are concerned as to the distribu- the trend in the last three to four decades prior to
tional impact of recession, then there may be 1914) and that the downturn did not start until
lessons to be learned from the 1930s. the First World War. Due to the lack of similar
The first striking conclusion is that the top wealth series for other countries, it is difficult to
shares in 1949 were much lower than thirty years know whether this is a general pattern.
earlier (1919) in the great majority of countries.
Of the eighteen countries for which we can make
the comparison with 1919 (or in some cases with The Postwar Picture
the early 1920s), no fewer than thirteen showed Returning to more recent times, we can see that
a strong decline in top income shares. In only there was considerable diversity of experience
one case (Indonesia) was there an increase in over the period from 1949 to the beginning of
the top shares. In half of the countries, the fall the twenty-first century. If we ask in how many
caused the shares to be at least halved between cases the share of the top 1 percent rose or fell
1919 and 1949. For countries where one can by more than 1 percentage points between 1949
compare 1949 with 1913–14, the fall generally and 2005 (bearing in mind that two-thirds were
seems at least as large. in the range 8 to 12 percent in 1949), then we
What happened before 1914? In five cases, find the seventeen countries more or less evenly
we have data for a number of years before the divided: six had a fall of two points or more, five
First World War. Naturally the evidence has to had a rise of two points or more, and six had a
be treated with caution and has evident limita- smaller or no change. If we ask in how many
tions: for example, the German figures relate cases the inverted-Pareto-Lorenz ȕ coefficient
only to Prussia. But it is interesting that in the changed by more than 0.1, then this was true
two Nordic countries (Sweden and Norway) the of fifteen out of twenty countries in Table 7.2
top shares seems to have fallen somewhat at the with twelve showing a rise (a move to greater
very beginning of the twentieth century, a pe- concentration). Examination of the annual top
riod when they might have been in the upward 1 percent share data for individual countries de-
part of the Kuznets inverted-U. As is noted in picted on Figures 7.4–7.7 confirms that, during
Aaberge and Atkinson (2010) for Norway and the fifty-plus years since 1949, individual coun-
Roine and Waldenström (2008) for Sweden, tries followed different time paths.
at that time Norway and Sweden were largely Can we nonetheless draw any common con-
agrarian economies. In neither Japan nor the clusions? Is it for example the case that all were
United Kingdom is there evidence of a trend following a U-shape, and that the differences
in top shares. In order to explore the pre-1914 when comparing 2005 and 1949 arise simply
period further, data apart from the income tax because some countries are further advanced?
records needs to be applied. Using a variety of Is the United States leading the way, with other
sources, including wealth data, Lindert (2000) countries lagging? In Table 7.3, we summarize
concludes that, in the United States, “we know the time paths from 1949 to 2005 for the sixteen
that income inequality must have risen some- countries for which we have fairly complete data
time between 1774 and any of these three com- over this period for the share of the top 1 per-
peting peak-inequality dates: 1860, 1913 and cent and top 0.1 percent. In focusing on change,
1929. . . . Beyond this, the evidence on the rise we are not interested in small differences after
70 • Part II: Inequality in Comparative Perspective

Table 7.3. Summary of Changes in Shares of Top 1 Percent and 0.1 Percent Between 1949 and 2005
Country Share of top 1 percent Share of top 0.1 percent
No change; rose 1 point Fell 1 point between 1949 and early 1980s;
France
between 1998 and 2005 rose 0.4 point between 1998 and 2005
United Kingdom Fell 6; rose 7½ points Fell 2; rose 3 points
United States Fell 3; rose 10 points Fell 1; rose 6 points
Fell 3; rose 6 points (up to
Canada Fell 1; rose 3½ points (up to 2000)
2000)
Australia Fell 7; rose 4 points Fell 2; rose 1½ points
New Zealand Fell 3; rose 4 points Fell 1; rose 1½ points
Germany No sustained change No sustained change
Netherlands Fell 6½ points (up to 1999) Fell 3 points (up to 1999)
Switzerland No sustained change No sustained change
Fell 7½; rose 4½ points (up
India Fell 4; rose 2½ points (up to 1999)
to 1999)
No sustained change up to
1999; No sustained change up to 1999;
Japan
rose 1½ points between 1999 rose ¾ point between 1999 and 2005
and 2005
No sustained change from
1960 to 1998; No sustained change from 1960 to 1990s;
Singapore
rose 2 points between 1998 rose 2 points between 1990s and 2005
and 2005
Argentina Fell 12; rose 4 points Fell 5½; rose 3 points
Sweden Fell 3½; rose 2 points Fell 1¼; rose 1¼ points
Rose 2 points up to early
Finland 1960s; fell 6 points; rose 3½
points
Norway Fell 4½; rose 8 points Fell ¾; rose 4½ points
Notes:
(1) “No change” means change less than 2 percentage points for top 1 percent; less than 0.65 percentage point for top 0.1
percent.
(2) Data coverage incomplete for part of the period for Argentina.
Source: Atkinson and Picketty (2007, 2010).

the decimal points. The criterion applied in the fall/rise. The share of the top 1 percent in Fin-
case of the share of the top 1 percent is that used land rose from below 8 percent in 1949 (it had
above: a change of 2 percentage points or more. been lower before then) to around 10 percent in
For the share of the top 0.1 percent, we apply a the early 1960s. Of the remaining fifteen coun-
criterion of 0.65 percentage points (i.e., scaled tries, one can distinguish a group of six “flat”
by 3.25/10). In applying this, we consider only countries (France, Germany, Switzerland, the
sustained changes. This means that we do not Netherlands, Japan, Singapore) and a group of
recognize changes due to tax reforms that distort nine “U-shaped” countries (United Kingdom,
the figures as in the case of Norway (Aaberge and United States, Canada, Australia, New Zealand,
Atkinson 2010) or New Zealand (Atkinson and India, Argentina, Sweden, Norway).
Leigh 2008), those due to the commodity price The ten countries belonging to the second
boom of the early 1950s as for Australia, New group appear to fit, to varying degrees, the
Zealand, and Singapore, or other changes that U-shape hypothesis that top shares have first
are not maintained for several years. fallen and then risen over the postwar period. In
Applying this criterion, there is just one most countries, the initial fall was of limited size.
case—Finland—where there is a pattern of rise/ As may be seen from Table 7.3, the initial falls in
 Atkinson, Piketty, and Saez—Top Incomes in the Long Run of History • 71

Figure 7.8. Inverted-Pareto ȕ Coefficients: English-Speaking Countries, 1910–2005

4.0

United States United Kingdom

Canada Australia
3.5
New Zealand

3.0
Pareto-Lorenz coefficient

2.5

2.0

1.5

1.0
1910

1915

1920

1925

1930

1935

1940

1945

1950

1955

1960

1965

1970

1975

1980

1985

1990

1995

2000

2005
Source: Atkinson and Piketty (2007, 2010).

top shares were less marked in the United States, significantly larger concentration at the top than
Canada, and New Zealand than in the United other continental European countries. This is
Kingdom, Australia, and India. The share of the also apparent in the observed patterns of Pareto
top 1 percent was much the same in the United ȕ coefficients, which more generally depict the
States and United Kingdom in 1949 but, in the same contrast between L-shaped and U-shaped
United Kingdom, the share then halved over countries as top income shares (see Figure 7.8).
the next quarter century, whereas in the United What about countries for which we have only
States it fell by only a little over a quarter. a shorter time series? The time series for China
The frontier between the U-shaped countries is indeed short, but there too the top of the dis-
and the flat countries is somewhat arbitrary and tribution is heading for greater concentration.
should not be overstressed. In France, after an For instance, the top 1 percent income share in
initial reduction in concentration, the top 1 per- China have gradually risen from 2.6 percent in
cent income share has begun to rise since the late 1986 to 5.9 percent in 2003 (Figure 7.7). This
1990s (Figure 7.5). In Japan and Singapore, the is still a very low top 1 percent share by inter-
rebound in recent years is even more pronounced national and historical standards, but the trend
(Figures 7.5 and 7.7). The only three countries is strong (and the levels are probably underes-
with no sign of a rise in income concentration timated due to the fact that China’s estimates
during the most recent period, namely Switzer- are based on survey data and not tax data, see
land, Germany, and the Netherlands, are coun- Piketty and Nancy Qian 2009). China has a way
tries where our series stop in the late 1990s. to go, but the degree of concentration is heading
There exists some reasonable presumption that in the direction of the values in OECD coun-
when data become available for the 2000s, these tries. Regarding the other countries with limited
countries might also display an upward trend. time coverage (Spain, Portugal, and Italy), one
Finally, note that Switzerland and especially also observes a significant rise in income concen-
Germany have always been characterized by a tration during the most recent period.
72 • Part II: Inequality in Comparative Perspective

However, the conclusion that most of the gains from

Conclusion economic growth was captured by the top 1 percent
In this chapter, we have argued that the study of is not in doubt.
top incomes is important from the standpoint of 2. The negative effect of the wars on top incomes
can be explained in part by the large tax increases
overall inequality and of the design of public pol-
enacted to finance the wars. During both wars,
icy. The tax data, on which the studies reviewed the corporate income tax was drastically increased
here are based, are subject to serious limitations, and this reduced mechanically the distributions to
which we have examined at length. The data can stockholders.
however, in our judgment, be used for distribu- 3. Interestingly, this dramatic increase in top
tional analysis, and they are the only source cov- wage incomes has not been mitigated by an increase
ering such a long run of years. The data cover in mobility at the top of the wage distribution. As
much of the twentieth century, including the Wojciech Kopczuk, myself, and Jae Song have shown
Great Depression, the Golden Age, and the Roar- in a separate paper, the probability of staying in the
top 1 percent wage income group from one year to
ing Nineties. In some cases, the data reach back
the next has remained remarkably stable since the
before the First World War and into the nine- 1970s.
teenth century. The estimates presented here are 4. In the case of New Zealand, we have used the
designed to be broadly comparable and provide estimates of Atkinson and Leigh (2008: table 1) that
evidence for more than twenty countries, con- adjust for the change in the tax unit in 1953. For
taining more than half of the world’s population. Indonesia we have taken the 1939 estimate and for
It will be clear to the reader that much re- Ireland that for 1943.
mains to be done. Major countries, such as Brazil
and Russia, are still missing from the database; REFERENCES
Aaberge, Roll, and Anthony B. Atkinson. 2010.
and Latin America is represented only by Argen-
“Top Incomes in Norway.” In Top Incomes: A
tina. Only a start has so far been made on test-
Global Perspective, ed. Anthony B. Atkinson and
ing different explanations and on evaluating the Thomas Piketty, 448–81. Oxford and New York:
impact of policy. The results from income tax Oxford University Press.
data need to be combined with those from other Atkinson, Anthony B. 2010. “Top Incomes in a
sources of evidence, such as the data on inher- Rapidly Growing Economy: Singapore.” In Top
ited wealth (Piketty 2009), on long-run studies Incomes: A Global Perspective, ed. Anthony B. At-
of company data (for example, Carola Frydman kinson and Thomas Piketty, 220–52. Oxford and
and Raven E. Saks 2010 and Steven N. Kaplan New York: Oxford University Press.
and Joshua Rauh 2010) and of key sectors such Atkinson, Anthony B., and Andrew Leigh. 2008.
“Top Incomes in New Zealand 1921–2005: Un-
as the financial industry (Thomas Philippon and
derstanding the Effects of Marginal Tax Rates,
Ariell Reshef 2009). We hope however to have Migration Threat, and the Macroeconomy.” Re-
demonstrated the potential of the field and we view of Income and Wealth, 54(2): 149–65.
hope that the data will provide a rich source for Atkinson, Anthony B., and Thomas Piketty, eds.
future researchers. 2007. Top Incomes over the Twentieth Century:
A Contrast between Continental European and
NOTES English-Speaking Countries. Oxford and New
* Note: This section draws from the following York: Oxford University Press.
article: Emmanuel Saez. 2010. “Striking it Richer: ———, eds. 2010. Top Incomes: A Global Perspective.
The Evolution of Top Incomes in the United States.” Oxford and New York: Oxford University Press.
Updated from Pathways 2008. http://elsa.berkeley Atkinson, Anthony B., Thomas Piketty, and Em-
.edu/~saez/saez-ustopincomes.2010.pdf. manuel Saez. 2010. “Top Incomes in the Long
1. The exact percentage (93 percent) is sensi- Run of History.’’ In Top Incomes: A Global Per-
tive to measurement error, especially the growth in spective, ed. Anthony B. Atkinson and Thomas
the total number of families from 2009 to 2010, Piketty, 664–759. Oxford and New York: Oxford
estimated from the Current Population Survey. University Press.
 Goldin and Katz—The Race Between Education and Technology • 73

Frydman, Carola, and Raven E. Saks. 2010. “Ex- Piketty, Thomas, Gilles Postel-Vinay, and
ecutive Compensation: A New View from a Jean-Laurent Rosenthal. 2006. “Wealth Con-
Long-Term Perspective, 1936–2005.” Review of centration in a Developing Economy: Paris and
Financial Studies, 23(5): 2099–2138. France, 1807–1994.” American Economic Review,
Kaplan, Steven N., and Joshua Rauh. 2010. “Wall 96(1): 236–56.
Street and Main Street: What Contributes to the Piketty, Thomas, and Emmanuel Saez. 2003. “Income
Rise in the Highest Incomes?” Review of Finan- Inequality in the United States, 1913–1998.”
cial Studies, 23(3):1004–50. Quarterly Journal of Economics, 118(1): 1–39.
Kuznets, Simon. 1953. Shuras of Upper Income ———. 2006. “The Evolution of Top Incomes: A
Groups in Income and Savings. Cambridge, Mass.: Historical and International Perspective.” Ameri-
National Bureau of Economic Research. can Economic Review, 96(2): 200–205.
Lindert, Peter H. 2000. “Three Centuries of In- Piketty, Thomas, and Nancy Qian. 2009. “Income
equality in Britain and America.” In Handbook Inequality and Progressive Income Taxation in
of Income Distribution. Volume 1, ed. Anthony B. China and India.” American Economic Journal:
Atkinson and Francois Bourguignon, 167–216. Applied Economics, 1(2): 53–63.
Amsterdam; New York and Oxford: Elsevier Sci- Roine, Jesper, Jonas Vlachos, and Daniel Walden-
ence, North-Holland. ström. 2009. “The Long-Run Determinants of
Philippon, Thomas, and Ariell Reshef. 2009. “Wages Inequality: What Can We Learn from Top In-
and Human Capital in the U.S. Financial Indus- come Data?” Journal of Public Economics, 93(7–
try: 1909–2006.” National Bureau of Economic 8): 974–88.
Research Working Paper 14644. Roine, Jesper, and Daniel Waldenström. 2008. “The
Piketty, Thomas. 2009. “On the Long-Run Evo- Evolution of Top Incomes in an Egalitarian Soci-
lution of Inheritance: France 1820–2050.” ety: Sweden, 1903–2004.” Journal of Public Eco-
Unpublished. nomics, 92(1–2): 366–87.

8 • Claudia Goldin and Lawrence F. Katz

The Race Between Education and Technology

Economic inequality since 1980 increased The expanding gap also occurred within groups,
greatly. The earnings of college graduates rose at even within educational levels. Among college
a far greater clip than did the earnings of those graduates, for example, those with degrees from
who stopped at high school graduation. The institutions with higher standards for admissions
incomes of top managers and professionals in- earned relatively more over time. Those who
creased at a much faster rate than did those of went to more prestigious law schools did better
ordinary workers. relative to other law school graduates. The wid-
The increase in inequality was more ening occurred within virtually all groups in a
all-encompassing than a widening between dif- manner that is not easily explained by the usual
ferent education levels or occupational groups. observable factors such as years of schooling. At

Claudia Goldin and Lawrence F. Katz. Race Between Education and Technology (Cambridge, MA: The Belknap
Press of Harvard University Press, 2008), pp. 87, 90–92, 95–99, 125, 395–402, 443, 445, 446–449, 451–454,
457–459, 460, 462–463, 466–467. Copyright © 2008 by the President and Fellows of Harvard College.
Reprinted by permission of Harvard University Press.
74 • Part II: Inequality in Comparative Perspective

almost all educational and experience levels, for and omnipresent, cutting across various produc-
example, the earnings for those near the top of tion methods and services. In addition, the tech-
the distribution increased considerably relative nological innovation must have diffused during a
to those near the middle or close to the bottom.1 fairly brief period. Finally, it must have required
The pervasive and rapid increase in economic workers to think, adjust, and reconfigure the
inequality has led many to search for explana- workplace. Computerization would seem to be
tory factors that are themselves pervasive and the perfect culprit.
rapid. A key suspect is skill-biased technological As we demonstrate in this chapter, it is clear
change, particularly that involved in the use of that technological change—computerization
computers.2 Chief among other factors that have in particular—is part of the explanation for
been mentioned are increased international trade rising inequality in the past 25 years. But, al-
and outsourcing, the greater immigration of though computerization and other technologi-
low-wage workers, the decline in private-sector cal changes were culprits in fostering inequality,
unionization, the erosion of the real value of these “criminals” were not acting alone. The rea-
the federal minimum wage, and changes in so- soning is simple.
cial norms concerning the pay of executives and New technologies alter the relative demand
other top-end earners. Here we mainly discuss for different types of labor; however, the overall
the role of technological change. impact of a new technology on the wage struc-
The central idea concerning the role of tech- ture reflects not only these demand shifts but
nology in affecting inequality is that certain also the supply responses by individuals attend-
technologies are difficult for workers and con- ing various types of schools or obtaining skills on
sumers to master, at least initially. Individuals the job or in other ways. Just because a technol-
with more education and higher innate abilities ogy places increasing demands on the skill, edu-
will be more able to grasp new and complicated cation, and know-how of the workforce does not
tools. Younger individuals are often better able necessarily mean that economic inequality will
to master new-fangled equipment than are rise and, if it does, that the increase will be sus-
older individuals. Employers, in turn, will be tained over a long period. If the supply of skills
more willing to hire those with the education rises to accommodate the increase in demand for
and other observable characteristics that endow skill, then wage inequality need not change.
them with the capacity to learn and use the new In other words, the evolution of the wage
technologies. Existing employees who are slow structure reflects, at least in part, a race between
to grasp new tools will not be promoted and the growth in the demand for skills driven by
might see their earnings reduced. Those who are technological advances and the growth in the
quicker will be rewarded. supply of skills driven by demographic change,
The type of technological change that is nec- educational investment choices, and immigra-
essary to explain the pervasive and rapid increase tion.4 This framework suggests that the rise in
in economic inequality in the latter part of the educational wage differentials and wage inequal-
twentieth century and the early twenty-first cen- ity since 1980 resulted from an acceleration in
tury must meet various criteria. First, it must demand shifts from technological change, or a
have affected a large segment of the workforce, deceleration in the growth of the supply of skills,
both production line workers and those in the or some combination of the two.
office, both highly educated professionals and The “computers did it” account also lacks
ordinary staff. As such, the innovation would historical perspective regarding technological
probably have to be of the “general purpose tech- change and inequality. Other critical moments
nology” form.3 A general purpose technology is existed in U.S. history when general purpose
one that is not specific to a particular firm, in- technologies swept the factory, office, and home.
dustry, product, or service. Instead, it is pervasive Consider, for example, the advent of motive
 Goldin and Katz—The Race Between Education and Technology • 75

power in the form of water wheels and later continuous during most of the twentieth cen-
steam engines or, better yet, the electrification tury than has been previously suspected. Sim-
of the factory, home, and urban transportation. ilar amounts of “skill bias” can be measured
The notion that computerization provided the during much of the twentieth century. Thus, the
first or the most momentous instance in U.S. demand-side argument, by itself, cannot explain
economic history of a complex technology that both parts of the twentieth century inequality ex-
placed greater demands on the knowledge, abil- perience. One cannot fully resolve the divergent
ity, and flexibility of virtually all workers and inequality experiences of the two halves of the
consumers is gravely mistaken. twentieth century by appealing to a recent accel-
eration in the degree of skill-biased technological
change brought on by the computer revolution.
Lessons from History Computers, to be sure, have given us much that
In the early twentieth century, a wide range is novel, time-saving, informative, entertaining,
of industries, particularly the newer and more and convenient. Yet, in terms of the skill bias to
technologically dynamic ones, demanded technological change and the increase in the rela-
more-educated workers. The workers to whom tive demand for skill, the era of computerization
we refer, were not necessarily of the professional has brought little that is new. . . .
class and they were not all working in an office, The evolution of the college wage premium
a boardroom, or on the sales floor. That is, they and the educational composition of the U.S.
were not always white-collar workers. Rather, workforce from 1950 to 2005 are given in Table
they included ordinary production line workers 8.1. The relative supply of college workers rose
who were using more complicated and valuable throughout the more than half-century era con-
machinery. For these workers, having more edu- sidered. The fraction of all full-time workers
cation meant having some high school and pos- who were high school dropouts was almost 59
sibly a high school diploma. For the professional percent in 1950 and that for college graduates
positions, the more-educated individuals would was barely 8 percent. In 2005, high school drop-
have gone to college. Our point is that new and outs were 8 percent and college graduates were
more complex technologies have had a long his- almost 32 percent. Those with some college rose
tory of transforming the workplace, as well as ev- from 9 percent to 29 percent. At the same time,
eryday life, in ways that reward quick-thinking, however, the wage premium for college gradu-
flexible, often young, and educated individuals. ates relative to high school workers more than
The most important historical point we want doubled, from 36.7 percent in 1950 to 86.6
to make in this chapter concerns a unified ex- percent in 2005. But most of the increase in
planation for long-term trends in inequality in the college premium occurred since 1980, with
America. The inequality story of the twentieth the decline in the college wage premium in the
century contains two parts: an era of initially de- 1970s offsetting much of the earlier increases in
clining inequality and a more recent one of rising the 1950s and 1960s.
inequality. But can the two parts of the inequality The key implication of the two central facts
experience have a unified explanation in the con- of a rising college wage premium and rising rel-
text of a demand-side framework? If technologi- ative supply of college workers since 1980 can
cal change was skill-biased in the latter part of the be best understood with reference to Figure
twentieth century so that the more skilled and 8.1, which is a schematic representation of the
educated did relatively better, was the opposite market for skilled and unskilled labor.5 In this
true of the earlier part of the century so that the simplified depiction, the workforce consists of
less skilled and educated fared relatively better? two groups—the skilled or highly educated (Ls)
The answers lie in the fact that skill-biased and the unskilled or less educated (Lu). Relative
technological change was far more rapid and wages for the two groups (ws/wu) are determined
76 • Part II: Inequality in Comparative Perspective

Table 8.1. U.S. Educational Composition of Employment and the College/High School Wage Premium:
1950 to 2005
full-time equivalent employment shares () by education
High School High School Some College College/High School
Dropouts Graduates College Graduates Wage Premium
1950 Census 58.6 24.4 9.2 7.8 0.313
1960 Census 49.5 27.7 12.2 10.6 0.396
1970 Census 35.9 34.7 15.6 13.8 0.465
1980 Census 20.7 36.1 22.8 20.4 0.391
1980 CPS 19.1 38.0 22.0 20.9 0.356
1990 CPS 12.7 36.2 25.1 26.1 0.508
1990 Census 11.4 33.0 30.2 25.4 0.549
2000 CPS 9.2 32.4 28.7 29.7 0.579
2000 Census 8.7 20.6 32.0 29.7 0.607
2005 CPS 8.4 30.9 28.9 31.8 0.596
Sources: Data for 1950 to 1990 are from Autor, Katz, and Krueger (1998, table 1). Data for 2000 and 2005 are from
the 2000 and 2005 Merged Outgoing Rotation Groups (MORG) of the CPS and 2000 Census IPUMS using the same
approach as in Autor, Katz, and Krueger (1998).
Notes: The college/high school wage premium is expressed in logs. Full-time equivalent (FTE) employment shares are
calculated for samples that include all individuals 18 to 65 years old in paid employment during the survey reference week
for each census and CPS sample. FTE shares are defined as the share of total weekly hours supplied by each education
group. The tabulations are based on the 1940 to 2000 Census IPUMS; the 1980, 1990, 2000, and 2005 CPS MORG
samples. The log (college/high school) wage premium for each year is a weighted average of the estimated college (exactly
16 years of schooling or bachelor’s degree) and post-college (17+ years of schooling or a post-baccalaureate degree) wage
premium relative to high school workers (those with exactly 12 years of schooling or a high school diploma) for the year
given. The weights are the employment shares of college and post-college workers in 1980. Educational wage differentials
in each year are estimated using standard cross-section log hourly earnings regressions for wage and salary workers in each
sample with dummies for single year of schooling (or degree attainment) categories, a quartic in experience, three region
dummies, a part-time dummy, a female dummy, a nonwhite dummy, and interaction terms between the female dummy
and quartic in experience and the nonwhite dummy. The levels of the log college wage premium can be compared across
census samples and across CPS samples respectively. But the levels of the CPS and census wage differentials cannot be
directly compared with each other due to differences in the construction of hourly wages in the two surveys. For further
details see Autor, Katz, and Krueger (1998).

by the intersection of a downward-sloping rela- to S2000. It is also the case—and this is a major
tive demand curve and an upward-sloping rela- point we would like to make—that the relative
tive supply curve. The short-run relative supply demand function must also have shifted outward.
of more-skilled workers is assumed to be inelas- We have drawn such a shift in Figure 8.1 as from
tic, since it is predetermined by factors such as D1980 to D2000.6 But why did the relative demand
past educational investments, immigration, and curve shift outward?
fertility.
Although one cannot directly observe the
entire demand and supply functions, one can Evidence on Skill-Biased
observe the equilibrium relative wages and rel- Technological Change
ative skills employed, as given in Table 8.1. Fig- Just because the relative demand for more highly
ure 8.1 illustrates the recent changes using data skilled workers increased does not necessarily
from 1980 and 2000. The actual labor market mean that there was skill-biased technological
outcomes from Table 8.1 shifted from point A change. Another possibility, and one that has
(year 1980) to point B (year 2000). Thus, the received considerable attention, is that the man-
relative supply function of more-skilled workers ufacturing jobs taken by the less well educated in
shifted outward from 1980 to 2000, from S1980 the United States have gone overseas. The relative
 Goldin and Katz—The Race Between Education and Technology • 77

Figure 8.1. A Schematic Representation of the Relative Supply and Demand for Skill: 1980 and 2000

Point A is approximately the values for (ws/wu) and (Ls/Lu) from Table 8.1 (using the 1980 Census data and exponentiat-
ing the college/high school wage premium) and point B is the same for 2000 (using the 2000 Census data), where Ls/Lu is
the ratio of college graduates to those with less than a college degree as per the selection criteria given in Table 8.1. Thus,
point A is e0.391 = 1.48 for (ws/wu) and 0.26 for (Ls/Lu).

demand for skill would then rise even if there had New technologies and greater capital in-
been no technological change. But there is very tensity, as shown in many studies, are strongly
strong evidence for the technology explanation. and positively associated with higher relative
In the first place, relative employment of utilization of more-skilled workers in firms and
more-educated workers and of nonproduction industries.8 A clear positive relationship has
workers increased rapidly within industries and been found between the relative employment of
within establishments during the 1980s and more-skilled workers and measures of technol-
1990s in the United States. The increased em- ogy and capital, such as computer investments,
ployment occurred despite the fact that the rela- the growth of employee computer use, R&D
tive cost of hiring such workers greatly increased. expenditures, the utilization of scientists and en-
Even though international outsourcing has been gineers, and increased capital intensity.9
blamed for the decreased utilization of the less Case studies of the banking, auto repair,
educated, the facts in this case argue against and valve industries show that the introduc-
that explanation as being the primary factor. tion of new computer-based technologies is
Large within-industry shifts toward more skilled strongly associated with shifts in demand toward
workers occurred in sectors with little or no for- more-educated workers.10 Surveys of human re-
eign outsourcing activity, at least in the 1980s source managers reveal that large investments in
and up to the late 1990s. Between-industry information technology, particularly those that
product demand shifts cannot be the main decentralize decisionmaking and increase worker
culprit. The magnitude of employment shifts autonomy, increase the demand for more highly
to skill-intensive industries, as measured by educated workers.11 The evidence is consistent
between-industry demand shift indices, is sim- with the “computers did it” view of widening in-
ply too small.7 equality. But we have more direct confirmation.
78 • Part II: Inequality in Comparative Perspective

How have computers increased the relative intensity and technology investments through-
demand for educated and skilled workers? A mul- out the 1950s to the 1970s.14 . . .
titude of reasons exist and often differ by work-
place. Computerized offices have routinized many
white-collar tasks, and the simpler and more “It Isn’t Just Technology—Stupid”
repetitive tasks are more amenable to comput- Great technological advances in recent decades
erization than are the more complex and idiosyn- have increased the relative demand for skill;
cratic. On the shop floor, microprocessor-based but, surprising as it may seem, the early part
technologies have facilitated the automation of of the twentieth century also experienced great
many production processes. Where hundreds of advances that increased the relative demand for
production workers once stood, often a handful skill, possibly to an equal degree. Technological
remains together with a small team of workers changes, however, were not always skill biased.
operating the computer. Computers, the Inter- We can locate a turning-point in the late nine-
net, and electronic commerce have raised the re- teenth century when technological changes be-
turns to marketing and problem-solving skills.12 came, on net, skill biased.
The share of U.S. workers directly using comput- Technological changes are not, in themselves,
ers on the job increased from 25 percent in 1984 responsible for the increase in inequality in the
to 57 percent in 2003.13 Even though computers recent period, just as they are not responsible for
have been easy for younger and highly educated the decrease in inequality during the earlier part
workers to use, they have been daunting for many of the twentieth century. Thus the central point
others, at least initially. of this chapter, to paraphrase a mantra of the
The empirical and logical case for skill-biased 1992 presidential campaign, is that “it isn’t just
technological change, as a substantial source of technology—stupid.” Since “it isn’t just technol-
demand shifts favoring more-educated work- ogy,” the reason for rising inequality cannot be
ers since 1980, would appear very strong. But located solely on the demand side. The other im-
that does not necessarily mean that the driving portant part of the answer can be found on the
force behind rising wage inequality since 1980 supply side.
was an alteration in the type or a quickening in
the rate of technological change. The reason is NOTES
that capital-skill complementarity was present 1. See Autor, Katz, and Kearney (2005) and Le-
mieux (2006) on the recent patterns of the evolution
throughout the twentieth century as was rapid
of U.S. residual (within-group) wage inequality.
skill-biased technological change. These effects
2. Skill-biased technological change refers to any
of technological change occurred even during introduction of a new technology, change in produc-
periods of declining or stable educational wage tion methods, or change in the organization of work
differentials and narrowing economic inequality. that increases the demand for more-skilled labor
The evidence that skill-biased technological (e.g., college graduates) relative to less-skilled labor
change has been ongoing for at least the last half (e.g., non-college workers) at fixed relative wages.
century begins with a pioneering article by Zvi 3. See Bresnahan and Trajtenberg (1995) on gen-
Griliches (1969), who found a substantial degree eral purpose technologies and their contributions to
of capital-skill complementarity in U.S. manu- economic growth.
4. The percentage college wage premium (ad-
facturing during the 1950s. Other researchers
justed for demographics) can be derived by exponen-
following Griliches’ path documented strong tiating the log college/high school wage differential
within-sector skill upgrading, even in the face shown in the final column of Table 8.1, subtracting
of rising educational wage differentials during 1, and then multiplying by 100. The levels of the
the 1950s and 1960s, and a strong positive cor- log college wage premium from the Census and CPS
relation of industry skill demand with capital are not fully comparable due to differences in the
 Goldin and Katz—The Race Between Education and Technology • 79

construction of hourly wages in the two surveys. 12. Autor, Levy, and Murnane (2003) posit such
Thus, we add the 2000 to 2005 CPS change in the an organizational complementarity between com-
log college premium to the 2000 Census log college puters and workers who possess both greater cogni-
premium to get a 2005 log college premium that tive skills and greater people skills.
can be compared to the 1950 Census log college 13. Friedberg (2003) uses questions on computer
premium. use at work from a series of CPS Computer and In-
5. See Bound and Johnson (1995), Katz and ternet Use supplements to document the growth of
Autor (1999), and Katz and Murphy (1992) for U.S. employee computer usage.
more detailed expositions of this framework. 14. See, for example, Autor, Katz, and Krueger
6. Changes in institutional factors or norms of (1998).
wage setting that lead to deviations from competi-
tive labor market outcomes could have played a role, REFERENCES
although the basic logic of the framework would Autor, David H., Lawrence F. Katz, and Melissa S.
still hold, as would the implication that the relative Kearney. 2005. “Trends in U.S. Wage Inequality:
demand function shifted outward as long as firms Re-Assessing the Revisionists.” NBER Working
remain on their labor demand curves. It is possible Paper no. 11627 (September).
that institutional factors, such as unions, produced Autor, David, Lawrence F. Katz, and Alan B. Krueger.
employment levels off the demand curve and that 1998. “Computing Inequality: Have Computers
declines in union strength could have led to a re- Changed the Labor Market?,” Quarterly Journal of
duction in the relative wage and employment of less Economics 113 (November), pp. 1169–213.
highly educated workers even in the absence of de- Autor, David H., Frank Levy, and Richard J. Mur-
mand shifts against them. nane. 2003. “The Skill Content of Recent Tech-
7. Autor, Katz, and Krueger (1998) find that nological Change: An Empirical Investigation,”
growth in the employment and wage bill shares of Quarterly Journal of Economics 118 (November),
more-educated workers from 1960 to 1996 is domi- pp. 1279–333.
nated by within industry changes using data on U.S. Bartel, A., Ichniowski, C., and K. Shaw. 2007. “How
three-digit industries. Borjas, Freeman, and Katz does Information Technology Really Affect Pro-
(1997) illustrate that between industry labor demand ductivity? Plant-Level Comparisons of Product
shifts from international trade explain only a modest Innovation, Process Improvement, and Worker
portion of the rise in the demand for more-skilled Skills.” Quarterly Journal of Economics.
U.S. workers from 1980 to 1995. Borjas, George J., Richard B. Freeman, and Lawrence
8. Doms, Dunne, and Troske (1997) provide a F. Katz. 1997. “How Much Do Immigration and
detailed plant-level analysis on the correlates of the Trade Affect Labor Market Outcomes?” Brookings
adoption of new technologies in U.S. manufacturing Papers on Economic Activity, no. 1, pp. 1–90.
in the 1980s and 1990s. Bound, John, and George Johnson. 1995. “What
9. Autor, Katz, and Krueger (1998) document Are the Causes of Rising Wage Inequality in the
strong positive correlations of skill upgrading with United States?” Economic Policy Review 1.1.
computer investments, increases in capital intensity, Bresnahan, Timothy F., Erik Brynjolfsson, and Lorin
R&D investments, and increased employee com- M. Hitt. February 2002. “Information Technol-
puter usage for U.S. industries. ogy, Workplace Organization, and the Demand
10. On the banking sector, see Autor, Levy, and for Skilled Labor: Firm-Level Evidence,” Quar-
Murnane (2003) and Levy and Murnane (2004). terly Journal of Economics, Vol. 117, pp. 339–376.
Levy, Beamish, Murnane, and Autor (1999) examine Bresnahan, Timothy, and Manuel Trajtenberg.
auto repair and Bartel, Ichniowski, and Shaw (2007) 1995. “General Purpose Technologies ‘Engines
study the valve industry. of Growth?’” NBER Working Paper No. w4148.
11. Bresnahan, Brynjolfsson, and Hitt (2002) Doms, Mark, Timothy Dunne, and Kenneth R.
study these issues combining a detailed survey of Troske. 1997. “Workers, Wages, and Technol-
senior human resource managers on organizational ogy,” Quarterly Journal of Economics. 1997. CXII,
practices and labor force characteristics with detailed 253–290.
information on information technology investment Friedberg., L. 2003. The impact of technological
for U.S. companies in the mid-1990s. change on older workers: evidence from data on
80 • Part II: Inequality in Comparative Perspective

computer use. Industrial and Labor Relations Re- Katz, Lawrence F., and Kevin M. Murphy. 1992.
view, 56(3), 511–529. “Changes in Relative Wages, 1963–87: Supply
Griliches, Z. 1969. “Capital-skill Complementarity.” and Demand Factors,” Quarterly Journal of Eco-
The Review of Economics and Statistics, 51(4): nomics 107 (February), pp. 35–78.
465–468. Lemieux, Thomas. 2006. “Postsecondary Education
Katz, Lawrence F., and David H. Autor. 1999. and Increased Wage Inequality,” American Eco-
“Changes in the Wage Structure and Earnings nomic Review 96 (May), pp. 195–99.
Inequality,” in Handbook of Labor Economics, Vol. Levy, Frank, and Richard J. Murnane. 2004. The
3, O. Ashenfelter and D. Card, eds. (Amsterdam: New Division of Labor. New York: Russell Sage.
North-Holland).

9 • Bruce Western and Jake Rosenfeld

Unions, Norms, and the Rise in U.S.

Wage Inequality

The decline of organized labor in the United other institutions like the minimum wage (Card
States coincided with a large increase in wage and DiNardo 2002; DiNardo, Fortin, and Le-
inequality. From 1973 to 2007, union member- mieux 1996); and limited, accounting for only
ship in the private sector declined from 34 to 8 a small fraction of rising inequality and only
percent for men and from 16 to 6 percent for among men (Card, Lemieux, and Riddell 2004).
women. During this time, wage inequality in We revisit the effects of union decline on in-
the private sector increased by over 40 percent. equality by examining union effects on nonunion
Union decline forms part of an institutional wages, considering whether wage inequality is
account of rising inequality that is often con- lower among nonunion workers in highly union-
trasted with a market explanation. In the market ized regions and industries.
explanation, technological change, immigration, Union effects on nonunion workers can work
and foreign trade increased demand for highly in several ways. Nonunion employers may raise
skilled workers, raising the premium paid to col- wages to avert the threat of union organization
lege graduates (for reviews, see Autor, Katz, and (Leicht 1989). We argue that unions also con-
Kearney 2008; Gottschalk and Danziger 2005; tribute to a moral economy that institutionalizes
Lemieux 2008). norms for fair pay, even for nonunion workers.
Compared to market forces, union decline is In the early 1970s, when 1 in 3 male workers
often seen as a modest source of rising inequality were organized, unions were often prominent
(Autor et al. 2008). Scholars view unions’ effects voices for equity, not just for their members,
as indirect, mediating the influence of techno- but for all workers. Union decline marks an ero-
logical change (Acemoglu 2002); secondary to sion of the moral economy and its underlying

Bruce Western and Jake Rosenfeld. “Unions, Norms, and the Rise in U.S. Wage Inequality,” American Socio-
logical Review 76:4 (August 2011), pp. 513, 514, 517, 518, 520–522, 524, 528, 532, 533. Copyright © 2011
by the American Sociological Association. Reprinted by permission of SAGE Publications.
 Western and Rosenfeld—Unions, Norms, and the Rise in U.S. Wage Inequality • 81

distributional norms. Wage inequality in the found less inequality among nonunion workers
nonunion sector increased as a result. in highly unionized industries (cf. Belman and
Our analysis estimates union effects on wage Heywood 1990).
inequality by decomposing the growth in hourly
wage inequality for full-time workers in the pri-
vate sector. Analyses of the Current Population Unions and the Moral Economy
Survey (CPS) show that union decline explains The theory of union threat takes a rationalist
a fifth of the increase in inequality among men view of employers and a minimalist view of labor
and none of the increase among women if only market institutions. Employers minimize labor
union wages are considered. The effect of union costs and only raise wages when threatened with
decline grows when we account for the link even greater pay increases through unionization.
between unionization and nonunion wages. In Institutions are conceived minimally in the sense
this case, deunionization explains a fifth of the that unions are the key distortion in an other-
inequality increase for women and a third for wise competitive labor market.
men. We relax these assumptions, arguing that the
labor market is embedded in a moral economy
in which norms of equity reduce inequality in
Unions and Nonunion Wages pay. The moral economy consists of norms pre-
Union wages have been a main focus of research scribing fair distribution that are institutional-
on inequality, but organized labor also affects ized in the market’s formal rules and customs. In
nonunion workers. Economists often contrast a robust moral economy, violation of distribu-
the effects of spillover and threat. When unions tional norms inspires condemnation and charges
raise wages for their members, employers may of injustice. We often think of the moral econ-
cut union employment, forcing unemployed omy historically—determining, for example, fair
workers to find jobs in the nonunion sector. prices for bread and flour under the British Corn
Spillover of workers into the nonunion labor Laws (Thompson 1971) or the relative rank and
market causes wages to fall. The threat effect standing of English workers in the nineteenth
results from nonunion employers raising wages century (Polanyi [1944] 1957).1
to the union level to avert the threat of union- Unions are pillars of the moral economy in
ization. The two theories yield opposing pre- modern labor markets. Across countries and over
dictions: unions reduce nonunion wages with time, unions widely promoted norms of equity
spillover but increase nonunion wages with that claimed the fairness of a standard rate for
threat. Empirical studies tend to support the low-pay works and the injustice of unchecked
threat effect, showing that nonunion wages are earnings for managers and owners (Hyman and
higher in highly unionized industries, localities, Brough 1976; Webb and Webb 1911). Com-
and firms (Farber 2005; Leicht 1989; Neumark parative researchers emphasize the role of distri-
and Wachter 1995). butional norms governing European industrial
The theory of union threat has distributional relations (Elster 1989b; Swenson 1989). The
implications. If unions threaten to organize U.S. labor movement never exerted the broad in-
low-wage workers, employers may raise wages, fluence of the European unions, but U.S. unions
thereby equalizing the wage distribution. Testing often supported norms of equity that extended
this theory, Kahn and Curme (1987) estimated beyond their own membership. In our theory
the effect of industry unionization on the vari- of the moral economy, unions help material-
ance of nonunion wages for detailed industries ize labor market norms of equity (1) culturally,
and occupations in the 1979 CPS. Consistent through public speech about economic inequal-
with the equalizing effect of union threat, they ity, (2) politically, by influencing social policy,
82 • Part II: Inequality in Comparative Perspective

and (3) institutionally, through rules governing We measure between-group wage inequality
the labor market. by the variance of the conditional means,
Culturally, industrial unions often use a lan-
guage of social solidarity in public discourse and (1)
within firms. Politically, U.S. unions have been
frequent advocates for redistributive public pol- where wi is the sample weight for respondent i,
icy. Institutionally, U.S. industrial relations often normed to sum to 1, and is the grand mean of
extend union conditions to nonunion workers. log wages. We measure within-group inequality
Unions also helped establish pay norms in local by the residual variance,
labor markets. In some industries, union influ-
ence was amplified by law. W (2)

That each respondent has a variance, may be

Data and Methods counterintuitive, but it is simply estimated by
We link union decline to rising inequality by the squared residual from the regression on
decomposing the variance of log wages. The yi. Because unions are mostly associated with
decomposition uses a variance function regres- within-group inequality, we expect deunioniza-
sion in which the mean and the variance of an tion to be closely associated with an increase in
outcome depend on independent variables, pro- within-group variance, W. Summing within-
viding a model of between- and within-group and between-group components yields the total
inequality (Western and Bloome 2009). variance, a measure of overall inequality:
The decomposition is based on a regression
on the log hourly wage, yi, for respondent i V = B + W.
(i=1, . . . ,N) for a given year of the CPS. Our
key predictors are an indicator for union mem- We decompose the rise in inequality with two
bership, ui, and a continuous variable, ūi, that adjusted variances that fix in a baseline year ei-
records for each respondent the unionization ther the coefficients or the distribution of some
rate for the industry and region in which they predictors. First, focusing on individual union
work. Covariates, including schooling, age, race, membership, we calculate the level of inequal-
ethnicity, and region, are collected in the vector, ity assuming unionization had remained at its
xi. The model includes equations for the condi- 1973 level. We estimate this compositional ef-
tional mean of log wages, fect by reweighting the data to preserve the 1973
unionization rate across all years, from 1973 to
ǔi = xiĮ1 + uiĮ2 + njiĮ3 2007. We then use the adjusted weights in Equa-
tions 1 and 2 to calculate adjusted variances for
and the conditional variance, all years. Second, we add the effects of union
threat and equitable wage norms on nonunion
log ıi = xi ȕ1 + ui ȕ2 + nji ȕ3 workers by calculating wage inequality assum-
ing industry-region unionization and its coeffi-
We expect union membership and cients remain constant at the 1973 level. Fixing
industry-region unionization to have positive industry-region unionization coefficients, and
effects on average wages, but negative effects reweighting to hold union density constant,
on the variance of log wages. Economy-wide yields adjusted values of ǔi and that are plugged
changes in average wages and within-group in- into Equations 1 and 2 to obtain adjusted mea-
equality, perhaps due to general shifts in norms sures of between- and within-group inequality.
or the macroeconomy, are captured through the We compiled data from the annual May files
regression intercepts. of the CPS from 1973 to 1981 and the annual
 Western and Rosenfeld—Unions, Norms, and the Rise in U.S. Wage Inequality • 83

Merged Outgoing Rotation Group files of the Inequality grew slightly more for women than
CPS from 1983 to 2007 (Unicon Research for men, but increasing women’s wage in-
Corporation various years; U.S. Bureau of the equality is unrelated to union membership.
Census various years). We exclude 1982, when Within-group inequality increased by about
union questions were omitted from the survey, .047 points among women but the adjusted se-
and 1994 and three-quarters of 1995, when al- ries, with 1973 unionization held constant, in-
location flags for wages were missing. The anal- creases nearly as much, by .043 points. Similarly,
ysis includes men and women working full-time between-group inequality increased by .051
(i.e., 30 hours a week or more) in the private points and the adjusted between-group variance,
sector. with union membership fixed, increased by .054
The dependent variable is log hourly wages points. Consistent with other research, holding
adjusted for inflation to 2001 dollars. Several the unionization rate constant explains almost
adjustments improve the quality and continu- none of the rise in women’s wage inequality.
ity of the wage data. Nonresponse to wage and Union decline explains more of the rise in
income questions increases over time, and by wage inequality once we account for the link
2007, about a third of CPS workers did not between unions and nonunion wages. The link
report wages. The CPS imputes wages to non- between organized labor and nonunion work-
respondents, but regression coefficients for non- ers is captured by the effects of industry-region
matched criteria are attenuated and the residual unionization on the mean and variance of wages.
variance for wages is sensitive to the imputed The total effect of deunionization can be mea-
data (Hirsch 2004; Mouw and Kalleberg 2010). sured by fixing the 1973 unionization rate as
We thus omit imputed wages from the analysis. before, and also fixing industry-region unioniza-
tion rates and their coefficients at 1973 values
(see Figure 7). This adjustment indicates the
Results strong relationship between union decline and
To study the effects of declining union member- rising within-group inequality. Among men, ad-
ship on wage inequality, we fix the unionization justed within-group inequality does not increase
rate at 1973 levels. Similar to earlier research, between 1973 and 2007, suggesting effects of
we find the largest effect of declining union union threat and eroding norms of equity on
membership on men’s within-group inequality. the wage distribution. Deunionization’s effect
The observed within-group variance increases on union and nonunion wages is associated
by .046 points, but the adjusted variance (with with about a third of the rise in wage inequality.
the 1973 unionization rate) increases by only This accounting, which includes unions’ effects
.028 points. Over a third of the increase in on nonunion wages, is 50 to 100 percent larger
within-group inequality is associated with de- than the union effects on inequality reported in
clining union membership ([.046–.028]/.046 other decompositions (cf. Card 2001; DiNardo
= .40). Between-group inequality increases by et al. 1996).
.055 variance points and the adjusted series in- Among women, the association between
creases by almost as much, by .053, indicating union decline and wage inequality is smaller,
that union decline explains little of the rise in but greater than zero, once nonunion wages
men’s between-group inequality. Summing the are taken into account (see Figure 7, panel b).
between- and within-group effects, the decline Declining industry-region unionization ac-
in unionization from 34 to 8 percent explains counts for over half of the increase in women’s
about a fifth of the rise in inequality in hourly within-group inequality. Adding between and
wages among full-time, private sector men. within-group effects together, union decline is
Among women, the effect of declining union associated with about a fifth of the rise in wage
membership on wage inequality is smaller. inequality among women.
84

Figure 9.1. Observed and Adjusted Within- and Between-Group Variances of Log Hourly Wages, Full-
Time, Private Sector Men and Women, 1973 to 2007; Adjusted Variances Fix Union Membership at the
1973 Level
 Western and Rosenfeld—Unions, Norms, and the Rise in U.S. Wage Inequality • 85

DiNardo, James, Nicole M. Fortin, and Thomas

Discussion Lemieux. 1996. “Labor Market Institutions and
These results suggest unions are a normative the Distribution of Wages, 1973–1992.” Econo-
presence that help sustain the labor market as metrica 64:1001–1044.
Elster, Jon. 1989a. “Social Norms and Economic
a social institution, in which norms of equity
Theory.” The Journal of Economic Perspectives,
shape the allocation of wages outside the union 3(4): 99–117.
sector. Elster, Jon. 1989b. “Wage Bargaining and Social
The key comparison implied by our analysis Norms.” Acta Sociologica 32:113–36.
is fundamentally historical, from the early 1970s Farber, Henry S. 2005. “Nonunion Wage Rates and
to the 2000s. In the earlier period, unions of- the Threat of Unionization.” Industrial and Labor
fered an alternative to an unbridled market logic, Relations Review 58:335–52.
and this institutional alternative employed over Gottschalk, Peter and Sheldon Danziger. 2005. “In-
a third of all male private sector workers. The equality of Wage Rates, Earnings and Family In-
come in the United States, 1975–2002.” Review
social experience of organized labor bled into
of Income and Wealth 51:231–54.
nonunion sectors, contributing to greater equal- Hirsch, Barry T. 2004. “Reconsidering Union Wage
ity overall. As unions declined, not only did the Effects: Surveying New Evidence on an Old
logic of the market encroach on what had been Topic.” Journal of Labor Research 25:233–66.
the union sector, but the logic of the market Hyman, Richard and Ian Brough. 1976. Social Val-
deepened in the nonunion sector, too, contrib- ues and Industrial Relations. Oxford, UK: Basic
uting to the rise in wage inequality. Blackwell.
Kahn, Lawrence M. and Michael Curme. 1987.
NOTES “Unions and Nonunion Wage Dispersion.” Re-
1. Thompson (1971) coined the term “moral view of Economics and Statistics 69:600–607.
economy.” Elster (1989a) argued that distributional Leicht, Kevin T. 1989. “On the Estimation of Union
norms are departures from strict rationality, eliciting Threat Effects.” American Sociological Review
strong emotions in response to violation. 54:1035–1057.
Lemieux, Thomas. 2008. “The Changing Nature of
REFERENCES Wage Inequality.” Journal of Population Economics
Acemoglu, Daron. 2002. “Technical Change, In- 21:21–48.
equality, and the Labor Market.” Journal of Eco- Mouw, Ted and Arne Kalleberg. 2010. “Occupa-
nomic Literature 40:7–72. tions and the Structure of Wage Inequality in the
Autor, David H., Lawrence F. Katz, and Melissa S. United States, 1980s to 2000s.” American Socio-
Kearney. 2008. “Trends in U.S. Wage Inequality: logical Review 75:402–431.
Revising the Revisionists.” Review of Economics Neumark, David and Michael L. Wachter. 1995.
and Statistics 90:300–323. “Union Effects on Nonunion Wages: Evidence
Belman, Dale and John S. Heywood. 1990. “Union from Panel Data on Industries and Cities.” Indus-
Membership, Union Organization and the Dis- trial and Labor Relations Review 49:20–38.
persion of Wages.” Review of Economics and Sta- Polanyi, Karl. [1944] 1957. The Great Transforma-
tistics 72:148–53. tion: The Political and Economic Origins of Our
Card, David. 2001. “The Effect of Unions on Wage Time. Boston: Beacon Press.
Inequality in the U.S. Labor Market.” Industrial Swenson, Peter. 1989. Fair Shares: Unions, Pay and
and Labor Relations Review, 54: 296–315. Politics in Sweden and West Germany. Ithaca, NY:
Card, David and John E. DiNardo. 2002. Cornell University Press.
“Skill-Biased Technological Change and Rising Thompson, Edward Palmer. 1971. “The Moral
Wage Inequality: Some Problems and Puzzles.” Economy of the English Crowd in the 18th Cen-
Journal of Labor Economics 20:733–83. tury.” Past and Present 50:76–136.
Card, David, Thomas Lemieux, and W. Craig Rid- Unicon Research Corporation. Various years. CPS
dell. 2004. “Unions and Wage Inequality.” Jour- Utilities Merged Outgoing Rotation Group and
nal of Labor Research 25:519–62. May Files. Santa Monica, CA.
86 • Part II: Inequality in Comparative Perspective

U.S. Bureau of the Census. Various years. Current Webb, Sidney and Beatrice Webb. 1911. Industrial
Population Survey 1980 May File. ICPSR ed. Democracy. London: Longmans, Green.
Ann Arbor, MI: Inter-university Consortium Western, Bruce and Deirdre Bloome. 2009. “Vari-
for Political and Social Research [producer and ance Function Regressions for Studying Inequal-
distributor]. ity.” Sociological Methodology 39:293–326.

10 • Robert Frank

Why Is Income Inequality Growing?

The bursting of the American housing bubble in Solving a problem requires an accurate di-
2008 sent the economy into free fall, with output agnosis of its cause, and much of the received
and employment tumbling even more rapidly wisdom on the causes of rising inequality is spu-
than during the early stages of the Great De- rious. Perhaps the most commonly held view
pression in the 1930s. And as even free-market is that inequality has grown because corporate
enthusiasts like former Federal Reserve chairman miscreants have successfully suppressed competi-
Alan Greenspan agreed, irresponsible lending tion. But although vivid examples of such abuses
practices by financial industry executives helped persist, most markets are in fact more competi-
precipitate the crisis. tive now than they have ever been.
So it is little wonder that there was immediate In their influential book Winner-Take-All Pol-
and widespread public outrage over the billions itics (2010), the political scientists Jacob Hacker
of taxpayer dollars used to bail out large banks and Paul Pierson persuasively explain how a ris-
whose leaders had been receiving eight-figure ing tide of campaign contributions has led Con-
annual bonuses. Nor should it be any surprise gress to enact lower tax rates for top earners and
that with the economy still struggling and those less stringent regulations for the corporations
bonuses again surging, populist anger continues. they own and manage.1 It is a compelling story
By fall 2011 this anger had spawned the Oc- whose importance has been underscored by the
cupy Wall Street movement. For the first time US Supreme Court’s decision in Citizens United
in more than eighty years, rising income in- v. Federal Election Commission in 2010. But al-
equality became a subject of widespread public though Hacker and Pierson are correct that the
discussion. reductions in top tax rates won by campaign
The specific public policy proposal offered contributors have contributed to the growth of
most frequently in response to populist outrage inequality, those reductions are of only second-
has been a government-imposed cap on executive ary importance. Across a broad swath of mar-
pay. In weighing this proposal, officials would do kets, not just those in which regulation has been
well to recall the words of Marcus Aurelius in relaxed, the far more important dimension of the
Meditations: “How much more grievous are the story has been the change in the distribution of
consequences of anger than the causes of it.” pretax incomes.

Original article prepared for this fourth edition of Social Stratification: Class, Race, and Gender in Sociological
Perspective, edited by David B. Grusky. Copyright © 2014 Westview Press. Used with permission of the author.
 Frank—Why Is Income Inequality Growing? • 87

For the first three decades after World War II, time. Institutional investors and leveraged buy-
pretax incomes grew at roughly the same rate— out specialists are now a much more powerful
slightly less than 3 percent a year—for house- force in the market than they were three decades
holds at all rungs of the income ladder. Since ago. Because these investors know that share
the mid-1970s, however, the pattern has been prices fall when a CEO packs the board with
dramatically different. The inflation-adjusted cronies, they can profit by intervening when they
median hourly wage for American men is actu- see evidence of that happening.
ally lower now than it was in 1975. Real median In one celebrated instance, Kohlberg, Kra-
household incomes have grown by roughly 15 vis, Roberts won the bidding for control of
percent since then, primarily because of large in- RJR-Nabisco at a time when the corporation
creases in female labor force participation. Only was perceived to be severely underperforming
those in the top quintile, whose incomes have under CEO Ross Johnson. KKR’s first move was
roughly doubled since the mid-1970s, have es- to fire Johnson. The buyout firm managed to
caped the income slowdown. break even on this transaction, even though it
That last statement is somewhat misleading, paid share prices twice as high as they had been
however, because the income growth picture is under Johnson’s leadership. In short, the growing
much the same within the top quintile as for influence of large outside investors has made ex-
the population as a whole. That is, those at the ecutive misbehavior less widespread than it used
bottom of the top quintile have seen little real to be.
income growth, the lion’s share of which has Why, then, has CEO pay at the nation’s larg-
been concentrated among top earners in the est companies risen more than tenfold during
group. Real incomes among the top 5 percent, the past three decades? As Philip Cook and I
for example, were more than two and a half argued in our 1995 book The Winner-Take-All
times larger in 2007 than in 1979, whereas those Society, the simplest explanation rests on techno-
among the top 1 percent were almost four times logical changes that not only have increased the
larger. In 1976 only 8.9 percent of the nation’s value of the most talented business leaders, but
total pretax incomes went to the top 1 percent also have resulted in more open competition for
of earners, but by 2007 that group was receiving their services.2
23.5 percent of the total. Falling shipping costs, advances in produc-
The pattern is similar for virtually all groups. tion technology, and the information revolution
It is true for dentists, for example, and also for have combined to increase the scope of many
college graduates. In every group, pretax real in- markets in recent decades. Companies that were
comes have been largely stagnant for all but the local became regional, then national, and now
top earners. international. Scale matters. When a corporal
Some of the most spectacular growth in top errs, fewer soldiers suffer than when a general
incomes has occurred in the financial services missteps. The same holds true in the corporate
industry, where, as Hacker and Pierson explain, world. As enterprises have grown, the economic
relaxed regulation has been an important reason. consequences of decisions made by top manage-
But deregulation and tax favors are not the cen- ment have increased in tandem.
tral story for most of the economy. A simple numerical example illustrates the
Others argue that chief executive officer extent to which growing scale increases the
(CEO) pay has soared because executives pack leverage of executive talent. Measured by mar-
their boards with cronies who gratefully grant ket capitalization, Apple is currently the world’s
them outsized salaries. That obviously has hap- largest company. Its future earnings growth de-
pened in specific cases. But because this is hardly pends heavily on its ability to ramp up produc-
a new phenomenon, it fails as an explanation of tion quickly enough to serve the rapidly growing
why CEO pay has risen so dramatically over demand for its smartphones and tablets. The
88 • Part II: Inequality in Comparative Perspective

company’s current CEO, Tim Cook, is widely managerial talent were just what the company
regarded as one of the world’s most talented needed, and that subordinates could compensate
supply chain managers. Now suppose, conserva- for Gerstner’s gaps in technical knowledge. The
tively, that Cook is just 3 percent better than the company’s bet paid off spectacularly, of course,
next best candidate Apple could have tapped for and in the years since then an existing trend to-
the same task. If Apple’s fiscal 2012 earnings of ward hiring CEOs from outside has accelerated
approximately $40 billion were 3 percent higher in most industries.
because of Cook’s performance, he was worth Most companies still promote CEOs from
$1.2 billion more to the company’s shareholders within, but even in those cases the spot market
that year than the second-best CEO candidate for talent has completely transformed the climate
would have been. Cook’s 2012 pretax com- in which salary negotiations take place. Internal
pensation package has been reported as $376 candidates can now credibly threaten to go else-
million—no small sum, to be sure, but only a where if they are not remunerated in accordance
fraction of even a conservative estimate of his with the market’s estimate of their value.
value to the company. Critics point to examples of disastrously
But growth in executive leverage alone can- unsuccessful CEOs to support their claim that
not explain the explosive increase in executive executive pay is completely divorced from per-
salaries. The decisions made by Charles Erwin formance. Certainly corporate boards never
Wilson, who headed General Motors from 1941 know for sure how much a given executive will
to 1953, had as great an impact on that com- add to or subtract from the company’s bottom
pany’s annual bottom line as the corresponding line. They are forced to rely on crude estimates,
decisions by today’s average Fortune 500 CEO. and hiring decisions based on them sometimes
Yet even after adjusting for inflation, Wilson’s don’t pan out. But even crude estimates are in-
total career earnings at GM were just a fraction formative, and companies that did not gamble
of what today’s top CEOs earn every year. in accordance with them would eventually cede
This is so because a second factor necessary ground to rivals that did.
to explain explosive CEO pay growth—an open The simple fact is that in large organizations,
market for CEOs—did not exist in Wilson’s day. even marginally better executive performance
Until recently, most corporate boards shared an has an enormous impact on a company’s bottom
implicit belief that the only viable candidates line. It is because executive labor markets are be-
for top executive positions were those who had coming more, rather than less, competitive that
spent all or most of their careers with the com- executive compensation in large American corpo-
pany. There was usually a leading internal can- rations has risen more than tenfold since 1980.
didate to succeed a retiring CEO and seldom The winner-take-all perspective is by no
more than a few others who were even credible means confined to the executive suite. The econ-
choices. Under the circumstances, CEO pay was omist’s traditional explanation of why some peo-
a matter of negotiation between the board and ple earn more than others is that people are paid
the anointed successor. in rough proportion to their human capital: an
That nearly exclusive focus on insiders has amalgam of characteristics such as intelligence,
softened in recent decades, a change driven in training, experience, temperament, and the like.
no small part by one particularly visible out- The human capital model directs our attention
side hire. In 1993 IBM decided to hire Louis J. to the worker rather than the job. Yet a given
Gerstner away from RJR Nabisco. At the time level of human capital will realize its full value
outside observers were extremely skeptical that a only if placed in a position with adequate scope
former tobacco CEO would be able to turn the and opportunity. For example, whereas having
struggling computer giant around. But IBM’s a slightly more talented salesperson may mean
board thought that Gerstner’s motivational and little if the task is to sell children’s shoes, it will
 Frank—Why Is Income Inequality Growing? • 89

mean a great deal if the task is to sell securities to persuasive showing that existing freedoms im-
the world’s largest pension funds. pose undue costs on others.
An economist under the influence of the As it turns out, that is relatively easy to
human capital metaphor might ask: Why not demonstrate. As I have argued elsewhere, run-
save money by hiring two mediocre people to away income growth at the top has spawned
fill an important position, instead of paying the changes in spending patterns that have made
exorbitant salary required to attract the best? Al- it more difficult for middle-income families to
though that sort of substitution might succeed achieve basic life goals.3 Consider, for exam-
with physical capital, it does not necessarily ple, the middle-income family whose goal is
work with human capital. Two average surgeons to send its children to a school of at least av-
or CEOs or novelists or quarterbacks are often a erage quality. A good school is an inherently
poor substitute for a single gifted one. context-dependent concept. It compares favor-
Technology has greatly extended the power ably with other schools in the same area. In al-
and reach of the planet’s most gifted performers. most every jurisdiction, the better schools tend
The printing press let a relatively few gifted sto- to serve children who live in more expensive
rytellers displace any number of minstrels and neighborhoods. The upshot is that the median
village raconteurs. Now that we listen mostly to family must outbid 50 percent of other parents
recorded music, the world’s best soprano can be merely to send its children to a school of average
everywhere at once. The electronic newswire al- quality.
lowed a small number of syndicated columnists Achieving that goal has become much more
to displace a host of local journalists. And the expensive in recent decades. Between 1970 and
proliferation of personal computers has enabled 2007, for example, the median size of a newly
a handful of software developers to replace thou- constructed single-family home in the United
sands of tax accountants. States increased by more than 50 percent. But
The result is that for positions in which ad- if real median wages were essentially stagnant
ditional talent has great value to the employer during that period, why did the median house
or the marketplace, there is no reason to expect grow so much bigger?
that the market will compensate individuals in That question is difficult to answer without
proportion to their human capital. For such po- reference to an expenditure cascade launched
sitions—which confer the greatest leverage or by higher spending at the top. Top earners built
“amplification” of human talent—small incre- bigger houses simply because they had so much
ments of talent may have enormous value and more money. That changed the frame of refer-
may be greatly rewarded as a result of the normal ence that shapes demands among earners just
competitive market process. This simple insight below the top, who travel in similar social circles,
lies at the core of the winner-take-all perspective so they, too, built bigger houses. And so on, all
on growing inequality. the way down the income ladder.
Of course the mere fact that explosive pay Similar cascades have influenced how much
growth at the top may have been largely a con- families must spend to celebrate special occa-
sequence of market forces does not make it an sions. The average American wedding in 2009
improper object of public policy concern. But cost $27,000, almost three times as much, in real
neither does the mere fact that the highest sal- terms, as its counterpart in 1980.
aries make some people feel angry and resent- Capping executive pay would be an
ful mean that it would be a good idea for the ill-considered response to the expenditure cas-
government to cap executive pay. Any legislative cades that have made life more difficult for the
action to abridge the freedom to negotiate mu- middle class. It would undercut one of the most
tually satisfactory contracts between executives important functions of the labor market, which
and their employers must be predicated on a is to steer people to the jobs that will make the
90 • Part II: Inequality in Comparative Perspective

most productive use of their talents. Having a parties would grow less rapidly. And because it
slightly more talented CEO improves corporate is relative expenditure that beyond some point
earnings by a larger amount in a big company determines whether people achieve what they set
than in a small one. So if each CEO candidate out to do, top earners wouldn’t sacrifice anything
accepts the highest offer received, both total cor- important by spending less.
porate earnings and total CEO earnings will be Hacker and Pierson are right that the grow-
maximized. Executive pay caps would reduce ing influence of money on politics has been
those totals by eliminating the signals that steer pernicious. But although the campaign contri-
people to the jobs that make best use of their butions of top earners have bought them lower
talents. But why would anyone prefer a world tax rates and less stringent regulations on their
with more, say, personal injury lawyers and fewer companies, such political shenanigans have not
corporate executives? been the most important cause of rising income
Fortunately, we can address the most impor- inequality. The big story has been the explosive
tant costs of runaway income inequality without growth in pretax incomes of the top 1 percent,
compromising the labor market’s ability to al- and in most sectors of the economy that growth
locate talent efficiently. Beyond a certain point, has been largely a consequence of intensifying
what matters most in people’s decisions to take market forces.
particular jobs is relative pay, not absolute pay. Those forces have produced important social
If any one company were to reduce its CEO pay benefits, but have also generated enormous so-
offer unilaterally, it would risk losing its best can- cial costs. The good news is that simple changes
didates to rival firms. But if all companies cut in tax policy can curb those costs without sacri-
their salaries in half, most executives would end ficing the most important benefits.
up in the same jobs as before. Companies cannot
organize across-the-board pay cuts on their own, NOTES
but the government can accomplish the same 1. Jacob Hacker and Paul Pierson, Winner-Take-
All Politics: How Washington Made the Rich Richer—
goal through the tax structure.
and Turned Its Back on the Middle Class (New York:
Higher tax rates on top earners would attenu-
Simon & Schuster, 2010).
ate the expenditure cascades that have been mak- 2. Robert H. Frank and Philip J. Cook, The
ing life so much more costly for middle-income Winner-Take-All Society (New York: The Free Press,
families. With fewer after-tax dollars to spend, 1995).
the additions to top earners’ mansions and 3. Robert H. Frank, The Darwin Economy (Princ-
the expenses for their children’s coming-of age eton, NJ: Princeton University Press, 2011).
 Hacker and Pierson—Winner-Take-All Politics • 91

11 • Jacob S. Hacker and Paul Pierson

Winner-Take-All Politics
Public Policy, Political Organization, and the Precipitous Rise of
Top Incomes in the United States

The three salient trends [of our time]—that who have been left behind. Put another way, the
income has become hyper-concentrated at the distribution of educational gains over the last
top, that the increase in income hyperconcen- twenty-five years—who finishes college or gains
tration has been sustained, and that this hyper- advanced degrees—has been much broader than
concentration has produced few “trickle-down” the distribution of economic gains. Only a very
benefits for the vast majority of American small slice of the new educational elite has en-
households—raise difficult problems for stan- tered the new economic elite.
dard economic analyses of rising inequality that Another problem for the standard economic
emphasize autonomous market changes that account is that the United States looks distinct
have widened the gap among broad skill and from other nations, despite the fact that all these
educational groups. They also call into ques- nations have presumably been buffeted by sim-
tion some central features of recent works that ilar market and technological forces. American
move beyond this economic emphasis to bring inequality is the highest in the advanced indus-
in politics. trial world. Yet gaps in skills are not measurably
By far the dominant economic explanation larger in the United States than they are in other
for rising inequality emphasizes “skill-biased affluent nations. And while the return to school-
technological change”—a shift toward greater ing is higher in the United States, this explains
emphasis on specialized skills, knowledge, and only a trivial portion of American inequality rel-
education—that has fueled a growing divide ative to inequality in other nations.2
between the highly educated and the rest of American distinctiveness is particularly pro-
American workers.1 The evidence on the chang- nounced when it comes to the hyperconcentra-
ing income distribution shows, however, that tion of income at the top. Figure 11.1 shows the
American inequality is not mainly about the share of income, excluding capital gains, going
gap between the well educated and the rest, or to the top 1 percent in twelve rich nations.
indeed about educational gaps in general. It is The first bar shows the share in the mid-1970s
about the extraordinarily rapid pulling away of (1973–75); the second shows the share around
the very top. Those at the top are often highly the millennium (1998–2004). As the figure
educated, but so too are those just below them makes clear, the United States leads the pack

Jacob S. Hacker and Paul Pierson. “Winner-Take-All Politics: Public Policy, Political Organization, and the
Precipitous Rise of Top Incomes in the United States,” Politics & Society 38:2 (2010), pp. 159–161, 168–169,
182–204. Copyright © 2010. Reprinted by permission of Sage Publications.
92 • Part II: Inequality in Comparative Perspective

Figure 11.1. The Top 1 Percent’s Share of National Income, Mid-1970s versus Circa 2000

Source: Andrew Leigh. “How Closely Do Top Incomes Track Other Measures of Inequality!” Economic Journal 117, no.
524 (2007): 619–33, http://econrsss.anu.edu.au/~aleigh/pdf/ToplncomesPanel.xls.

with regard to both the level (16 percent) and percentage increase—despite starting from a
increase (virtually a doubling) of the top 1 per- higher base.)
cent’s share of income. Note that half of the na- Moreover, the trajectory of the two coun-
tions in Figure 11.1—France, Germany, Japan, tries that are most often compared to the United
the Netherlands, Sweden, and Switzerland—ex- States’s, the United Kingdom and Canada, can-
perienced little or no increase. Note, too, that not be viewed as wholly independent of the rise
the United States was similar to many of these of America’s winner-take-all economy. The rise
nations in terms of the share of income going in the compensation of the highest earners, espe-
to the top 1 percent in the 1970s—indeed, the cially corporate executives and financial managers,
shares in the United States and Sweden track drives much of the outsized gains at the top in the
very closely until around 1980. United States. Companies in English-speaking
It is true that the other English-speaking na- Canada and the United Kingdom compete for
tions in this group—Australia, Canada, Ireland, these workers, and thus have faced the most
New Zealand, and the United Kingdom—have pressure to match the massive salaries on offer in
followed a path more like the United States’s, the United States. There is substantial evidence
suggesting overlapping policy and political that much of the (considerably smaller) rise in
trends. Still, the United States stands out even executive compensation in Canada is driven by
among these nations, experiencing a doubling American developments, rather than reflecting
of the income share of the top 1 percent be- an independent example of the same phenom-
tween the mid-1970s and 2000, compared with enon.3 While the contagion effect of the United
around half that in percentage terms in the other States is difficult to quantify, some (and perhaps
five nations. (The difference is smaller if 1980 is much) of the increase in top incomes in other
used as the base year, rather than the mid-1970s, English-speaking nations may reflect competitive
but the United States still experienced a larger pressure to match the more dramatic rise in the
 Hacker and Pierson—Winner-Take-All Politics • 93

United States—a rise that we shall see has a great

deal to do with U.S. public policy. An Organizational-Policy
Some economists have accepted that Ameri- Perspective on Winner-Take-All
can inequality has been distinctly top-heavy, but Inequality
insist that the market still rules. The economist A convincing political account of American
(and former Bush administration official) Greg- inequality must explain the defining feature of
ory Mankiw has evocatively likened the Ameri- American inequality, namely, the stunning shift
can superrich to the winners of the golden ticket of income toward the very top. Equally im-
in Charlie and the Chocolate Factory.4 Most of portant, it must explain how public policy has
the educated receive only the chocolate bar; a contributed to this trend. This means not only
lucky few find a ticket to vast riches within the identifying public policies that can be linked to
bar’s wrapper. But Mankiw’s analogy is silent large increases in inequality; it also means pro-
on the question of how the tickets were placed in viding an account of the political processes that
the chocolate bar and why some of the educated have led to the generation of those policies.
get the ticket and others do not. It implies that These are not easy tasks. They require an un-
both the presence of the tickets and the selection derstanding of the connections, often subtle, be-
is market-driven, when in fact, as we shall sec, tween policy structures and economic outcomes.
Mankiw’s “golden tickets’’ were in substantial Moreover, they require close examination of the
part created by government, and their distri- political forces behind policy change. Depend-
bution has been deeply shaped by the political ing on the policy involved, the relevant decision
clout of their beneficiaries. makers may be legislators, regulators, or presi-
To be sure, market processes and techno- dents. New enactments may be required to shift
logical changes have played a significant part in policy. As we will see, however, policy change
shaping the distribution of rewards at the top. often occurs when groups with the ability to
Revolutionary changes in information technol- block change effectively resist the updating of
ogy have fostered more concentrated rewards in policy over an extended period of time in the
fields of endeavor—such as sports or entertain- face of strong contrary pressure and strong evi-
ment—where the ability to reach large audiences dence that policy is failing to achieve its initial
is the principal determinant of economic retum.5 goals—what we call policy “drift.”6 These com-
Computers, increased global capital flows, and plex connections between political action and
the development of new financial instruments social outcomes are not likely to be established
have made it possible for savvy investors to reap without sustained attention to the evolving con-
(or lose) huge fortunes almost instantaneously. tent of public policies.
Other examples of such technologically driven Existing political accounts—while a vast im-
winner-take-all inequality can be found. But provement over the standard economic diagno-
these accounts do not come close to explain- sis—are not particularly successful in identifying
ing the concentrated gains at the very top of plausible links among politics, policy, and rising
the American economic ladder, especially those inequality. McCarty, Poole, and Rosenthal focus
driven by rising executive pay and financial mar- on the way in which gridlock has prevented pol-
ket compensation. They do not explain why icy updating in a few areas (the minimum wage,
these trends have been much more pronounced public assistance policies, the estate tax), but for
in the United States than elsewhere. Nor do they the most part say little about how public policy
explain why market structures conducive to such has influenced rising inequality—and virtually
outcomes arose when they did, much less why, nothing about how it has fueled the meteoric rise
as we show, those structures were fostered by of top incomes. Bartels, for his part, fails to show
government. how partisan control of the executive by itself
94 • Part II: Inequality in Comparative Perspective

could plausibly be connected to the dramatic rise exactly has benefited? Moreover, what does the
in winner-take-all outcomes that has occurred pattern of policy change suggest about political
since the late 1970s. dynamics? When, and through what mecha-
At the root of these weaknesses, in our view, is nisms has policy changed?
a conception of politics that focuses overwhelm- Thanks to work by Thomas Piketty and Em-
ingly on the voter-politician relationship—a manuel Saez, we are now in a much better posi-
view we call “politics as electoral spectacle.” By tion to answer some of these questions.7 Piketty
contrast, we believe the rise of winner-take-all and Saez have generated stunning data on the
inequality can only be convincingly explained changing structure of federal taxation. The data
with a very different perspective—which we call allow Piketty and Saez to investigate tax inci-
“politics as organized combat.” dence—not just changes in the marginal rates in
the tax code but the actual tax levels that house-
holds pay once deductions and other maneuvers
Government’s Reshaping of the are taken into account.
American Economy Piketty and Saez’s results, presented in Figure
If the politics of electoral spectacle is about win- 11.2, are striking in several respects. First, they
ning elections, the politics of organized combat suggest that the role of taxes in rising inequality
is about transforming what government does. is much more pronounced if one concentrates on
Did the shifting balance of organized interests the very top income groups. The changes of the
lead to major changes in the governance of the tax rate for those at the ninetieth percentile, and
American political economy? The answer is yes, even the ninety-eighth percentile, have actually
and we document these changes in four crucial been quite modest over the past four decades. By
policy arenas: taxation, industrial relations, exec- contrast, there have been startlingly large changes
utive compensation, and financial markets. for those in the top 1 percent. This is mostly be-
We seek to make a plausible case for two cause of the declining role of the corporate in-
claims. First, there is substantial evidence come tax and the estate tax. Progressivity used to be
that policy developments of the past three de- very pronounced at the very top of the tax code; now
cades—through both enactments and drift— it is almost entirely absent. As Piketty and Saez
have made a central contribution to the surge summarize their findings, “The 1960 federal tax
of winner-take-all economic outcomes in the system was very progressive even within the top
United States. Second, there is also substantial percentile, with an average tax rate of around 35
evidence that organized interests were highly percent in the bottom half of the top percentile
motivated, mobilized, and involved in many of to over 70 percent in the top 0.01 percent.”8
these developments. Second, not only has the shift in policy
been highly concentrated on the very affluent,
the magnitude of the shift is quite large. Again,
Taxes Piketty and Saez,
Taxes represent perhaps the most visible way in
which policy makers influence the distribution [T]he pre-tax share of income for the top
of income. Furthermore, even casual observers 0.1 percent rose from 2.6 percent in 1970
are aware that policy since the Reagan adminis- to 9.3 percent in 2000. The rise in after-tax
tration has often involved significant tax cuts for income shares was from 1.2 percent in 1970
the well-to-do. Yet crucial questions remain un- to over 7.3 percent in 2000. In percentage
answered—questions we will address with regard point terms, the increase in pretax incomes
to all four of the policy areas we examine. How is slightly greater than the increase in post-
big has the policy shift been? Has it made a sig- tax incomes. But in terms of observing what
nificant contribution to rising inequality? Who those with very high incomes can afford to
 Hacker and Pierson—Winner-Take-All Politics • 95

Figure 11.2. Average Tax Rates for Top Income Groups, 1960–2004

Source: Thomas Piketty and Emmanual Saez, “How Progressive Is the U.S. Federal Tax System? A Historical and
International Perspective,” Journal of Economic Perspectives 21, no. 1 (2007): 3–24.

consume, the after-tax share of income for confidence of compensation boards that execu-
those in this income group multiplied by a tives will reap much of the astronomical salaries
factor of 6.1, while the pretax share of income that boards grant.)
multiplied by a factor of 3.5. The tax reduc- Equally striking is the enduring nature of
tions enacted in 2001 and 2003 have further the policy shift. Although the early years of the
weakened the redistributive power of the fed- Reagan administration figure prominently, the
eral income tax today.9 change began in the 1970s. The initial drop
came through large cuts in the capital gains tax
One can ask, counterfactually, what if the gap and other taxes on the well-to-do passed, after
between pre- and posttax income of the top 1 very intense business lobbying, by a Congress
percent had not declined since 1970? Accord- composed of large Democratic majorities in
ing to Piketty and Saez, the top 0.1 percent had both chambers and signed into law by a Demo-
about 7.3 percent of after-tax income in 2000. cratic president.
If the gap between their pre- and posttax income Moreover, it is difficult to trace these devel-
had remained what it was in 1970, they would opments to any straightforward shift in public
have had about 4.5 percent of after-tax income. opinion regarding taxation.10 Research on both
In other words, changes in tax incidence account the tax cuts of the early 1980s and those of the
for roughly one-third of the total gains in income past few years suggest that organized interests
share for the top 0.1 percent in the last four de- have played a prominent role, both in keeping
cades. (Moreover, this is just the direct effect of tax cuts on the agenda and shaping policy to
the decline in effective taxation at the top. Many focus the gains of tax-policy changes on those at
experts believe the rise in executive pay that we the very top of the income distribution.11 The
will discuss in a moment has been fostered in- rise of supply-side economics, with its empha-
directly by the decline, which has increased the sis on the negative effects of high marginal tax
96 • Part II: Inequality in Comparative Perspective

rates on the well-off, surely helped the progress The United States has always stood out on
of tax cuts. Yet the supply-side rationale actu- measures of union strength as a nation with a
ally gained credence only after the initial flurry weak organized labor movement. Nonethe-
of tax cuts in the 1970s, and it was largely ex- less, over the past three decades, the structure
hausted as a serious intellectual force by the of American industrial relations has changed
time the Bush administration and congressional profoundly. Union density—the share of the
Republicans slashed upper-income taxes in the workforce covered by collective bargaining—
early 2000s. has fallen precipitously, especially in the private
A number of specific developments within sector. From organizing roughly one-third of
the broad arc of tax policy making suggest an the workforce, union organization has fallen to
impressive focus on directing benefits not just under one-tenth of private sector employees. . . .
to the very well-to-do (say, the bottom half of [There are] at least two important ramifi-
the top 1 percent) but to the superrich. David cations of union power. First, unions have the
Cay Johnston provides copious evidence, for capacity to play an important role in corporate
example, that there has been a steep decline, governance.14 At least under certain institutional
encouraged by Congress, in IRS oversight of conditions, they have the resources and incen-
high-income returns, combined with a politically tives necessary to provide a check on the scale of
induced shift of resources to oversight of Earned executive compensation, and to push for com-
Income Tax Credit returns of the poor and lower pensation designs that align executive incentives
middle class.12 Studies of specific battles over the more closely with those of their firms. Indeed,
estate tax and the alternative minimum tax have even with their current weakness, American
suggested that policy makers repeatedly chose unions represent one of only two organized in-
courses of action that strongly advantaged the terests providing a potential check on managerial
very wealthy at the expense of the much larger autonomy—the other being “investor collec-
group of the merely well-to-do.13 In all these tives” like public employee pension systems and
accounts, the influence of organized interests— (more problematically) mutual funds.
particularly lobbyists representing business and Second, unions may play a significant role in
the wealthy, conservative antitax groups such political conflicts related to the distribution of
as Americans for Tax Reform, and free-market income. On the one hand, they may push policy
think tanks like the Cato Institute—loom much makers to address issues of mounting inequal-
larger than the sway of voters or pull of general ity. On the other, they may recognize, highlight,
public sentiment. and effectively resist policy changes that further
inequality. Consider just one example of how
the contemporary weakness of organized labor
Industrial Relations shows up in recent policy developments. Well
The evolution of industrial relations in the over a thousand registered lobbyists in Wash-
United States provides the second crucial chap- ington identify taxes as one of their areas of
ter in the tale of winner-take-all inequality’s rise. activity. Yet during recent fights over the estate
Research in comparative political economy has tax—a policy issue with large and obvious distri-
long maintained that the organization of rela- butional consequences—organized labor could
tionships between employers and workers is of supply only one union lobbyist to address the
fundamental significance for a wide range of issue.15 Amazingly, even this lobbyist was avail-
economic interactions. For good reason: there able only quarter-time to work on all tax issues.
are wide and highly consequential differences in In fact, the biggest organized opposition to es-
these arrangements within the universe of afflu- tate tax repeal came not from organized labor
ent democracies that have been repeatedly linked but from a group of billionaires led by William
to major distributional and market differences Gates Sr. (Bill Gates’s father). It is hard to imag-
across nations. ine a more telling illustration of the existing
 Hacker and Pierson—Winner-Take-All Politics • 97

Table 11.1. Union Share of Wage and Salary Workers in the United States and Canada
United States (%) Canada (%) Percentage point difference
1960 30.4 32.3 1.9
1970 26.4 33.6 7.2
1980 22.2 35.7 13.5
1990 15.3 34.5 19.2
2000 13.5 32.4 18.9
2005 12.5 32.0 19.5
Sources: David Card, Thomas Lemieux, and W. Craig Riddell, “Unionization and Wage Inequality:
A Comparative Study of the U.S., the U.K. and Canada” (Working Paper 9473, National Bureau of
Economic Research, Cambridge, MA, 2003); Sylvia Allegreto, Lawrence Mishel, and Jared Bernstein,
The Slate of Working America (Ithaca, NY: Cornell University Press, 2008); and Statistics Canada,
Labour Force Historical Review 2008 (Table Cd3T09an), Ottawa, Statistics Canada, 2009 (Cat. No.
71F0004XCB).

organizational inequalities in Washington on the composition of the National Labor Relations

economic issues. Board after his election. As Henry Farber and
Even if one accepts labor’s potential signifi- Bruce Western have cogently argued, however, a
cance for politics and policy making, however, heavy emphasis on these overt initiatives is diffi-
there is the issue of how to account for union cult to reconcile with a close examination of the
decline. Many would maintain that the decline timing and patterns of union decline.17
of unions in the United States is almost exclu- Our own account would emphasize alterna-
sively a market phenomenon. Unions are fad- tive policy mechanisms and focus on the con-
ing, it is often suggested, because of changes in sequences of government inaction rather than
the global economy that relentlessly send their action. During the recent transformation of
jobs overseas. Yet as in so many aspects of the the American political economy, the evolution
subject of inequality, a comparative view casts of industrial relations is perhaps the most con-
doubt on the idea that market imperatives are sequential instance of policy drift. The absence
the only story, and that extreme union decline of an updating of industrial relations policy has
in the private sector is inevitable. While unions had brutal effects on the long-term prospects of
have declined in significance in many Western organized labor. It is well understood that the
nations, their presence has fallen little or none American industrial relations system contained
in others.16 And one of those nations has the certain structural features that gravely threat-
virtue for comparative analysis of being other- ened unions after the 1960s.18 Well established
wise quite similar to the United States—namely, in certain manufacturing industries in particu-
Canada. Once more limited in reach than their lar states, unions were acutely vulnerable to the
American counterparts, Canadian unions movement of manufacturing jobs to states where
now enjoy much broader membership (about labor rights were more limited, as well as shifts in
one-third of the nonagricultural workforce) and employment to sectors that had not previously
have seen little decline—despite similar worker been organized.
attitudes toward unions in the two nations An updating of industrial relations policy
(Table 11.1). could have addressed some of these weaknesses.
The contrast with Canada suggests the possi- Careful comparison with Canada is revealing. As
bility that fall in union density has been in sig- Table 11.1 suggests, the two union movements
nificant part a political process. And indeed there diverged dramatically in their capacity to cope
is considerable evidence this is the case. Popular with a shifting economic environment. The Ca-
advocates of this view would stress events like nadian economist W. Craig Riddell has found
Ronald Reagan’s efforts to break the strike by air that little of the divergence can be explained
traffic controllers (PATCO), as well as changes in by structural differences in the two nations’
98 • Part II: Inequality in Comparative Perspective

Figure 11.3. The Evolution of Executive Compensation (Mean Pay of Top Executives)

Source: Carols Frydman and Raven Saks, “Historical Trends in Executive Compensation, 1936–2003” (Working
paper, Massachusetts Institute of Technology, Cambridge, MA, 2005).
Note: Sum of salaries, bonuses, long-term bonus payments, and the Black-Scholes value of stock option grants
for the three highest-paid officers in the largest fifty firms in 1940, 1960, and 1990.

economies, or even by differing worker propen- them. They then used their organizational re-
sities to join a union.19 Rather, the difference sources to exploit the resulting drift and launch a
is due to the much lower (and declining) like- vigorous assault on American labor, with effects
lihood in the United States that workers who felt not just in the economic sphere but also in
have an interest in joining a union will actually American politics.
belong to one. There is considerable evidence
that differences in labor law are a major part of
the explanation. Prominent institutional differ- Corporate Governance and
ences include Canadian practices that allow for Executive Compensation
card certification and first-contract arbitration, While changes in taxation and the reach of
ban permanent striker replacements, and impose unions are obviously germane to inequality, the
strong limits on employer speech.20 Meanwhile, case for discussion of corporate governance re-
aggressive antiunion activities by employers in quires some defense. No one disputes that rising
the United States have met little resistance from executive pay has played a central role in mount-
public authorities. ing American inequality. Piketty and Saez sug-
In short, the severe decline of organized labor gest that perhaps half of the pretax gains of the
in the United States was in part a political out- top 1 percent reflect the explosion of executive
come, driven by new antiunion enactments as compensation.21 Figure 11.3 shows the devel-
well the failure to update policy to reflect the opment of total compensation for the top three
increasing relative strength of employers in a executives at the nation’s largest firms since the
more global, service-oriented economy. There mid-1930s. As the figure shows, executive pay
were policy alternatives that would have reduced skyrockets after 1980, with the largest increases
the decline, and that had advocates within the coming in the 1990s.
United States. The opponents of such reforms, Here we enter a sphere that Americans
possessing formidable and growing organiza- have typically treated as outside of politics, but
tional resources, mobilized effectively to stop which comparative evidence suggest is strongly
 Hacker and Pierson—Winner-Take-All Politics • 99

influenced by patterns of political contestation.22 encourage good performance. Indeed, in much

Cross-nationally there is tremendous variation in the same way that our view of interest groups
corporate governance practices and in economic suggests opportunities to outflank disorganized
outcomes. Rising executive compensation is “outsiders” (exploiting structural advantages
much more evident in Anglo countries than in under conditions of asymmetric information),
other rich democracies, and most evident of all Bebchuk and Fried demonstrate the promi-
in the United States. Compensation structures nence in corporate arrangements of what they
remain much different in most of the advanced call “camouflage.” Patterns of executive compen-
industrial world than they are in the United sation seem designed to mitigate public outrage
States. For instance, where stock options are rather than limit excessive pay or link it more
used, they are often linked to long-term rather closely to value.
than short-term performance, as well as to firm Most accounts of American inequality, if
performance relative to industry norms. Thus, they touch on these issues at all, regard them as
for example, when a rising price of oil drives matters of markets, not politics. At a minimum,
up the share price of energy companies, CEOs however, policy makers (like police officers who
would receive extra compensation only if their studiously look the other way) have done little
firm’s performance exceeded industry averages. to constrain the dramatic shift that has taken
The rise in executive pay seems related to place. This is in sharp contrast to the experience
a broader shift in structures of corporate gov- abroad, where—even though executive pay is
ernance, ostensibly toward maximization of much lower—there have been substantial efforts
“shareholder value” but arguably toward what to monitor and impose limits on executive pay,
Peter Gourevitch and James Shinn call “manager- and where sources of countervailing power ap-
ism,” in which opportunities for well-positioned pear to be much stronger. Again, the compar-
elites to extract resources increase. The hypoth- ative evidence of American (or at least Anglo)
esis to consider is that the capacity of managers exceptionalism with regard to executive pay sug-
to engage in such extraction has increased.23 The gests that there is nothing about the structure
issue, as the financier John Bogle has recently of modern capitalism that makes such extraor-
put it, is whether the United States moved to- dinary increases in executive salaries inevitable
ward an “ownership society” in which managers or even likely.
serve owners or an “agency society” in which In fact, one can see strong links between
managers serve themselves. American policy making and the explosion of
A school of thought (prominent in the field executive compensation. The most direct, and
of law and economics) sees the United States’s consequential, concerns the development of
and other systems of diffused ownership as rep- stock options. This constitutes another highly
resenting the best protections for stockholders.24 significant example of policy drift. During the
That view relies heavily on a principal-agent 1990s, stock options became the central vehicle
analysis that sees boards of directors as protect- for enhancing executive compensation—indeed,
ing shareholders through “arm’s-length negotia- roughly 50 percent of executive compensation
tions” with executives. Yet in many cases, boards came through stock options by 2001. Although
are not playing the role outlined in this theory. ostensibly a vehicle for linking pay to perfor-
Lucian Bebchuk and Jesse Fried provide many mance, these options were almost always struc-
findings more consistent with a “board-capture” tured in ways that lowered the visibility of high
view, in which boards are so beholden to man- payouts (by removing them from financial ac-
agers that they offer little countervailing au- counts at the time they were granted). More-
thority.25 Perhaps most telling, the design of over, options were granted without creating
CEO compensation often varies markedly from strong connections between payout and man-
what one would expect if it were intended to agerial effectiveness, even though instruments
100 • Part II: Inequality in Comparative Perspective

for establishing such links were well known and lower than it was thirty years ago.) Even staid
widely used abroad.26 The value of options sim- corporate giants got into the act. In 1980, GE
ply rose along with stock prices, even if stock earned 92 percent of its profit from manufac-
price gains were fleeting, or a firm’s performance turing. In 2007, more than half of GE’s profits
badly trailed that of other companies in the same came from its financial businesses. The home
sector. In the extreme but widespread practice address of the winner-take-all economy has been
of “backdating,” option values were reset retro- neither Hollywood nor Silicon Valley, but Wall
actively to provide big gains for executives—a Street.
practice akin to repositioning the target after The rise of finance is virtually synonymous
the fact to make sure the archer’s shot hits the with the rise of winner-take-all, since in no
bull’s-eye. major sector of the economy are gains so highly
The structure of American corporate gover- (and increasingly) concentrated at the top. In
nance—and its associated, distinctive patterns part, this is just a chapter of the broader rise of
of executive compensation—is a prime contrib- executive pay. But the other part is the runaway
utor to American winner-take-all inequality. To rewards that have flowed into the pockets of
a considerable if largely unrecognized extent this the rich out of America’s widening range of ex-
is a political outcome. The policy alternative is otic new financial institutions—from boutique
not just hypothetical; other countries, including hedge funds to massive financial conglomerates
ones with market systems relatively similar to crossing once-inviolable regulatory boundaries.
that of the United States’, have moved to facili- These rewards have involved the development of
tate organized countervailing powers.27 Yet in the complex new financial products that, for most
United States, huge differences remain in the rel- Americans, offered limited benefits—and some-
ative capacity for organized action of managers times real economic risks—but which held out
and shareholders. And there is substantial evi- the prospect of big returns from every financial
dence that political authorities remain far more transaction and spectacular incomes for those
attentive to the interests of the former than the within the industry.
latter. At the very top, the gains were mind-boggling.
In 2002 it took $30 million to make it to the
top twenty-five hedge fund incomes; in 2004,
Financial Deregulation $100 million; in 2005, $130 million. That year,
With the pillars of high finance now battered, five hedge fund managers made $500 million
it is easy to forget how dramatic the rise of the or more. These were just the biggest of the big
American financial sector has been. Wages and winners, however: in the two years before they
salaries in U.S. financial services roughly dou- began reporting losses that dwarfed the profits
bled their share in the economy over the past of prior years and brought many of their stock-
three decades, expanding from 5 percent to holders to ruin, the venerable firms of Goldman
nearly 10 percent of all wages and salaries be- Sachs, Merrill Lynch, Morgan Stanley, Lehman
tween 1975 and 2007. The percentage of the Brothers, and Bear Stearns paid their employees
economy composed of financial industry activ- bonuses of $75 billion. Wages in the financial
ities exploded—from less than 2 percent just sector took off in the 1980s. But what were con-
after World War II to more than 8 percent. sidered princely sums at that time simply set the
Between 1980 and 2007, financial service com- floor for what was to follow. The pace of the rise
panies expanded their proportion of company actually accelerated in the 1990s, and again after
profits from around 13 percent to more than the millennium. Current revelations about Wall
27 percent. (In the early 2000s, the share nearly Street excesses make abundantly clear that the
reached one-third of all corporate profits—de- central chapter in the chronicle of winner-take-
spite the fact that employment in the sector was all inequality is a tale of American finance.
 Hacker and Pierson—Winner-Take-All Politics • 101

As with executive compensation, attempts real incomes of investment bankers and share
to discuss the political roots of this economic traders.”28
transformation were long dismissed out of Nonetheless, the gradual shredding of the
hand. Until a few years ago, high finance was post–New Deal rulebook for financial markets
depicted as the purest of markets. When ana- did not simply result from the impersonal forces
lysts referenced the preferences of “Wall Street,” of “financial innovation.” Titans of finance are
it was taken as almost a synonym for economic wont to airbrush the role of government out of
rationality itself, rather than as a set of specific their tales of individual economic success. San-
economic interests. Yet financial markets, like ford Weill, the former chairman of Citigroup,
others, are not prepolitical. Our financial system put it this way: “People can look at the last 25
has always rested on an extensive set of govern- years and say this is an incredibly unique period
ment interventions. Public policy establishes the of time. We didn’t rely on somebody else to
legal environment for financial transactions, in- build what we built.”29 But Weill and the other
cluding such crucial issues as what constitutes financial chieftains who prospered so greatly
insider dealing or an unacceptable conflict of in- during this “unique” period relied a great deal
terest, how much monitoring and transparency on supportive politicians in government. Weill,
there will be in major financial transactions, and for example, helped lead the industry assault
what levels of leverage and risk are acceptable on the Glass-Steagall Act, stripping away key
given the potentially massive externalities asso- conflict-of-interest and transparency rules. These
ciated with large-scale speculation. In response “reforms” legalized powerful financial conglom-
to market failures on all these dimensions in erates, such as the one produced by Weill’s
the run-up to the Great Depression, extensive merger of Travelers Insurance and Citibank to
new federal regulations were designed during form Citigroup (an entity that would become
the New Deal to ensure investor confidence a ward of the American taxpayer only a decade
and align private ambitions more closely with later). . . .
broad economic goals such as financial stability. Policy—both what government has done and
Regulations sought to limit conflicts of interest, what, as a result of drift, it has failed to do—has
encourage transparency, and discourage reckless played an absolutely central role in the rise of
risk taking that placed the desire for private gain winner-take-all economic outcomes. It is not the
in conflict with the integrity and security of the only thing that has mattered, but it has mattered
financial system. a lot. Moreover, in the main areas where the role
Over the last three decades, these relatively of government appears most significant, we see
quiet and stable financial markets gave way to a consistent pattern: active, persistent, and con-
ones that were both far more dynamic and—for sequential action on the part of organized in-
good or ill—had far more pervasive effects on the terests that stood to gain from a transformation
rest of the economy. Some of this dramatic shift of government’s role in the American economy.
was clearly driven by changes in the nature of A winner-take-all politics accompanied, and
economic activity and the possibilities for finan- helped produce a winner-take-all economy.
cial intermediation. Technological innovation
made possible the development of new financial
instruments and facilitated spectacular experi- Conclusion
ments with securitization. Moreover, as Robert Explaining the remarkable rise of a winner-take-
Gordon and Ian Dew-Becker note, technology all economy requires a true political economy—
helped “shift Wall Street from million-share that is, a perspective that sees modern capitalism
trading days in the 1980s to billion-share trad- and modern electoral democracies as deeply
ing days since the late 1990s,” which “must also interconnected. On the one side, government
have contributed to the multi-fold increase in profoundly influences the economy through an
102 • Part II: Inequality in Comparative Perspective

extensive range of policies that shape and reshape 3. Perhaps most telling, there is little sign of the
markets. On the other side, economic actors— same meteoric rise at the top among French-speaking
especially when capable of sustained collective portions of Canada, where executives do not appear
action on behalf of shared material interests— to be competing in the same common labor market
that has allowed American pay levels at the top to
have a massive and ongoing impact on how po-
diffuse to the rest of Canada. Emmanuel Saez and
litical authority is exercised. Michael R. Veall. “The Evolution of High Incomes
Recent economic accounts have missed the in Northern America: Lessons from Canadian Evi-
first side of this relationship. Conceptualizing dence,” American Economic Review 95, no. 3 (2005):
government’s role in an excessively narrow way, 831–49.
they have attributed highly concentrated gains 4. Gregory N. Mankiw. “The Wealth Trajectory:
to impersonal technological forces. While this Rewards for the Few,” New York Times, April 20,
interpretation has some basis, neither the Amer- 2008.
ican experience nor comparative evidence sug- 5. Phillip J. Cook and Robert H. Frank, The
Winner-Take-All Society (New York: Free Press,
gests it can bear the weight that economists have
1995).
placed on it. 6. Jacob S. Hacker, “Privatizing Risk without Pri-
Perhaps surprisingly, the limits of these ac- vatizing, the Welfare State: The Hidden Politics of
counts flow from a similar source. Too many U.S. Social Policy Retrenchment,” American Political
[social scientists] have treated the American Science Review 98. no. 2 (2004): 243–60.
political economy as an atomized space, and fo- 7. Thomas Piketty and Emmanual Saez, “How
cused their analysis on individual actors, from Progressive Is the U.S. Federal Tax System? A His-
voters and politicians to workers and consum- torical and International Perspective.” Journal of Eco-
ers. But the American political economy is an nomic Perspectives 21, no. 1 (2007): 3–24.
8. Ibid., 12.
organized space, with extensive government pol-
9. Ibid., 15.
icies shaping markets, and increasingly power-
10. Hacker and Pierson, Off Center, chap. 3.
ful groups who favor winner-take-all outcomes 11. For tax cuts of the early 1980s, see Edsall,
playing a critical role in politics. Finding allies New Politics of Inequality, 226–33; Ferguson and
in both political parties, organized groups with Rogers, Right Turn, 119–24; and David Stockman,
a long view have successfully pushed new ini- The Triumph of Politics: Why the Reagan Revolution
tiatives onto the American political agenda and Failed (New York: Harper & Row, 1987). For tax
exploited the opportunities created by Amer- cuts in the past few years, see Hacker and Pierson,
ican political institutions to transform U.S. Off Center, chap. 3.
12. David Cay Johnston, Perfectly Legal (New
public policy—through new enactments and
York: Portfolio Books, 2003).
pervasive policy drift. In the process, they have
13. Michael J. Graetz and Ian Shapiro, Death
fundamentally reshaped the relative economic by a Thousand Cuts: The Fight over Taxing Inherited
standing and absolute well-being of millions Wealth (Princeton, NJ: Princeton University Press,
of ordinary Americans. Politics and governance 2005). Hacker and Pierson, Off Center.
have been central to the rise of winner-take-all 14. Peter A. Gourevitch and James Shinn, Polit-
inequality. ical Power and Corporate Control: The New Global
Politics of Corporate Governance (Princeton, NJ: Uni-
NOTES versity of Princeton Press, 2005).
1. Eli Berman, John Bound, and Stephen 15. Graetz and Shapiro, Death by a Thousand
Machin, “Implications of Skill-Biased Technological Cuts.
Change,” Quarterly Journal of Economics 113, no. 4 16. According to the ICTWSS Database (In-
(1998): 1245–79; and Claudia Goldin and Lawrence stitutional Characteristics of Trade Unions, Wage
F. Katz, The Race between Education and Technology Setting, State Intervention and Social Pacts), of the
(Cambridge, MA: Harvard University Press, 2008). nineteen rich democracies for which data are avail-
2. Richard Freeman, America Works: Critical able from the early 1970s through the mid-2000s,
Thoughts on the Exceptional U.S. Labor Market (New union density increased in four (Finland, Belgium,
York: Russell Sage, 2007). Denmark, and Sweden) and declined by less than
 Hacker and Pierson—Winner-Take-All Politics • 103

15 percent in four others (Norway, Canada, Italy, 22. Mark Roe, Political Determinants of Corpo-
and Luxembourg). In seven nations (Switzerland, rate Governance: Political Context, Corporate Impact
the United Kingdom, Denmark, Ireland, the Neth- (Oxford, UK: Oxford University Press, 2003); and
erlands, Japan, and Austria), the decline was be- Gourevitch and Shinn, Political Power and, Corporate
tween one-third and one-half. In four more, it was Control.
greater than 50 percent: Australia (55 percent), New 23. As noted earlier, there is a connection be-
Zealand (59 percent), France (63 percent), and the tween the previous discussion of tax policy and the
United States (57 percent). Only France had lower current discussion of executive compensation. The
union density at the end of the period, though in sharp fall in true tax rates on very high incomes
France, unlike in the United States, the effects of may have stimulated the rise in executive pay since
statutory bargaining extend to nearly the whole the recipients capture so much more of any rise in
workforce. Jelle Visser, University of Amsterdam, compensation. Carola Frydman and Raven Sax esti-
http://www.uva-aias.net/208. mate that “had tax rates been at their year 2000 level
17. Henry Farber and Bruce Western, “Ronald for the entire sample period, the level of executive
Reagan and the Politics of Declining Union Orga- compensation would have been 35 percent higher
nization,” British Journal of Industrial Relations 40 in the 1950s and 1960s.” “Executive Compensation:
(2002): 385–402. A New View from a Long-Term Perspective, 1936–
18. Nelson Lichtenstein, State of the Union 2005” (Working Paper W14145, National Bureau of
(Princeton, NJ: Princeton University Press, 2002); Economic Research, Cambridge, MA, June 2008).
and Paul Osterman, Securing Prosperity—The Amer- http://ssm. eom/abstract=1152686.
ican Labor Market: How It Has Changed and What 24. Kevin J. Murphy, “Politics, Economics, and
To Do about It (Princeton, NJ: Princeton University Executive Compensation,” University of Cincinnati
Press, 1999). Law Review 63 (1995): 714–46.
19. W. Craig Riddell, “Unionization in Canada 25. Lucian Bebchuk and Jesse Fried, Pay
and the United States: A Tale of Two Countries,” in without Performance: The Unfulfilled Promise of
Small Differences that Matter: Labor Markets and In- Executive-Compensation (Cambridge, MA: Harvard
come Maintenance in Canada and the United States, University Press, 2004).
ed. David Card and Richard Freeman (Chicago: 26. Arthur Levitt, Take on the Street: How to Fight
University of Chicago Press and National Bureau of for Your Financial Future (New York: Vintage, 2002).
Economic Research, 1993), 109–48. 27. Gourevitch and Shinn have an interesting
20. John Goddard, “Do Labor Laws Matter? The argument about why the British system (similar in
Density Decline and Convergence Thesis Revisited,” important respects to the American one) seems to be
Industrial Relations 42. no. 3 (2003): 458–92. less vulnerable to regulatory capture by managers.
21. Piketty and Saez, “How Progressive Is the They stress the majoritarian structure of political in-
U.S. Federal Tax System?” Also see Ian Dew-Becker stitutions, which puts greater checks on the political
and Robert J. Gordon, “Where Did the Productivity influence of highly concentrated interests.
Growth Go? Inflation Dynamics and the Distribu- 28. Gordon and Dew-Becker, “Controversies
tion of Income,” Brookings Papers in Economic Activ- about the Rise of American Inequalities,” 25.
ity 2 (2005): 67–150; and Robert J. Gordon and Ian 29. Louis Uchitelle and Amanda Cox, “The
Dew-Becker, “Controversies about the Rise of Amer- Richest of the Rich, Proud of a New Gilded Age,’’
ican Inequality: A Survey” (Working Paper 13982, New York Times, July 15, 2007, A1.
National Bureau of Economic Research, Cambridge,
MA, May 2008).
104 • Part II: Inequality in Comparative Perspective

12 • Thomas A. DiPrete, Gregory M. Eirich, and Matthew Pittinsky

Leapfrogs and the Surge in Executive Pay

Introduction of “superstar” labor markets in a manner that

Income inequality has been rising in the U.S. is consistent with a “shareholder value” model
since the late 1970s, with most of the increase of corporate governance (Jensen and Murphy,
since the early 1990s being at the top of the in- 1990; Himmelberg and Hubbard, 2000; Mur-
come distribution. Some attribute the increase phy and Zábojník, 2004; Gabaix and Landier,
primarily to technical factors that have increased 2008). Others counter that CEO compensation
demand for education, but others argue that this rose so rapidly because CEOs gained too much
explanation is implausible: technical change is power over the setting of their own pay. Defend-
global in character, and yet earnings inequal- ers of the shareholder value theory cite positive
ity rose much more rapidly in the U.S. than firm-level correlations between pay and perfor-
in other industrialized countries. This anomaly mance. Proponents of the managerial power the-
has led scholars to argue that the explanation ory call attention to numerous characteristics of
for inequality trends must involve institutional compensation practices that appear inconsistent
arrangements that determine the setting of pay with efficient market practices (Core et al., 1999;
(Ebbinghaus and Kittel, 2005; Kenworthy, Bebchuk et al., 2002; Ang and Nagel, 2006;
2007), the terms of employment (DiPrete et al., Armstrong et al., 2008). Defenders respond that
2006; Maurin and Postel-Vinay, 2005), and, the large increase in compensation during the
more abstractly, the level of “rent” that can be 1990s occurred even as corporate governance in
extracted from specific classes and occupations general appeared to be tightening, not loosening.
(Sørensen, 2000; Weeden, 2002; Morgan and Critics find plausible evidence of a correlation
Tang, 2007).1 between the quality of governance in the firm
These debates about labor market trends in and the compensation of its executives net of
general mirror debates about the pay of Amer- performance and other market-relevant factors.
ican corporate executives, which has increased Defenders point to other evidence that questions
rapidly even as earnings in general have risen the impact of firm-level governance on the rising
at the top of the American distribution. Some compensation trend.
scholars argue that corporations have bid up Neither of these positions adequately cap-
executive compensation according to the rules tures the implications of the interaction between

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Thomas A. DiPrete, Gregory M. Elrich, and Matthew Pittinsky, “Compensation
Benchmarking, Leapfrogs, and the Surge in Executive Pay,” American Journal of Sociology 115:6 (May 2010),
pp. 1671–1712, published by the University of Chicago Press. The article printed here was originally prepared
by Thomas A. DiPrete, Gregory M. Eirich, and Matthew Pittinksky for this fourth edition of Social Stratifi-
cation: Class, Race, and Gender in Sociological Perspective, edited by David B. Grusky. Copyright © 2014 by
Westview Press.
 DiPrete, Eirich, and Pittinsky—Leapfrogs and the Surge in Executive Pay • 105

firm and environment, which provides a struc- the documentation that corporate governance
tural basis for rent extraction. We favor a third responds to institutional as well as market im-
explanation that emphasizes the systemic impli- peratives, with a principal institutional force
cations of managerial power. According to this being the historically changing set of cultural
third explanation, the complexity and dynamism conceptions about how markets should work
of compensation comparison groups produce a (Fligstein, 2002). Bebchuk et al. (2002) (see also
mechanism by which governance failures in Bebchuk and Fried, 2004), meanwhile, question
individual firms propagate through corporate from a legal perspective the effectiveness of cor-
networks via the normal workings of the bench- porate governance in their critique of “optimal
marking system to raise executive salaries. We contracting theory,” via their core assertion that
argue, in short, that governance is a characteristic “there is no contract that would perfectly align
of networks as much as of firms. Poor governance the interests of managers and shareholders” (p.
at the firm level can produce excessive compen- 751).
sation for the CEO of that firm. And when firms Much of the empirical evidence put forward
that are “well-governed” use poorly governed to support either the managerial power or the
firms with overpaid CEOs to benchmark the pay for performance theory searches for statis-
compensation of their own CEO, then the conse- tical patterns between firm-level, industry-level,
quences of poor governance spread from “poorly and executive-level characteristics on the one
governed” to “well-governed” firms. This diffu- hand and compensation on the other. Economic
sion process raises the compensation of CEOs sociology, in contrast, favors explanations that
throughout the distribution. In other words, rent focus either on the interaction between local
extraction takes place even when CEOs are paid governance and the broader governance envi-
their “market wage” as established by competitive ronment, or on the impact of network ties on
benchmarks. This argument supports the recent economic outcomes. We see three important
attention in the stratification literature to rent questions from the economic sociology perspec-
extraction as a manifestation of occupational tive that need to be addressed in order to estab-
power rather than an outcome of the bargaining lish the differences in these approaches. First,
power of individual workers. what part of the market does the firm’s compen-
sation committee look at when it sets pay, i.e.,
what is the network structure and the content of
Inequality, Rent Extraction, and the the network ties between executives and the en-
Governance of Markets vironment that are implied by pay comparisons?
A key theoretical issue in the debate over ex- Second, how controlling are these comparisons
ecutive compensation concerns the distinction on pay determination? Third, how does pay set-
between the questions: “where does governance ting in response (or not) to these comparisons
come from?” and “how well does governance propagate through the comparison networks
work?” From the perspective of the theory of that are set up by the compensation process?
shareholder value, these questions amount to the Answers to all three of these questions involve
same thing, i.e., the answer to “where does gov- the system of compensation benchmarking.
ernance come from?” is “the market,” and the an- The benchmarking process is well described
swer to “how well does governance work” is that in contemporary compensation manuals pro-
“market discipline produces optimal shareholder duced by compensation consultants (e.g., Reda
value.” Economic sociology sees these questions et al., 2008). Some parts of this “standard model”
as fundamentally separate. As argued in recent have been incorporated into regulatory practices
reviews of this literature (Davis, 2005; Fligstein monitored by regulatory agencies, and company
and Choo, 2005; Fligstein and Dauter, 2007), adherence at least to the formal characteristics
much of the sociological literature concerns of this process has been well documented.2 To
106 • Part II: Inequality in Comparative Perspective

accomplish this process, compensation con- distribution of CEO compensation over time.
sultants are called in and a peer group for the We do not argue that leapfrogging is the only
company’s executives is identified. New com- mechanism by which upward ratcheting of pay
pensation is determined relative to the compen- could occur. Indeed, we expect that managerial
sation of the executives in the peer group via power can and would be expressed through other
negotiations with the CEO. A number of stud- mechanisms, including the control of informa-
ies of executive compensation have documented tion about the value of equity and other incen-
the widespread use of peer groups in the process tive components of executive compensation
of setting executive compensation (Porac et al., (Sørensen 2000; Hall and Murphy 2003) and
1999; Bizjak et al., 2007; Faulkender and Yang, the mechanism involving below-median pay ex-
2007), but the interpretation of this practice amined by Ang and Nagel.3 We do argue, how-
varies widely. Holmstrom and Kaplan recently ever, that leapfrogging is a powerful mechanism
wrote that “Prices, including wages, are ulti- for ratcheting pay because it allows individual
mately set by supply and demand, and bench- expressions of managerial power at some firms to
marking is nothing more than looking at market change benchmarks and thereby affect the pay of
prices” (Holmstrom and Kaplan, 2003, p. 19). CEOs throughout the corporate world.
They went on to argue that “the main problem
with executive pay levels is not the overall level
but rather the extreme skew in the awards. To Data and Methods
deal with this problem, we need more effective We analyzed information obtained from the
benchmarking, not less of it.” The environmen- Standard & Poor’s ExecuComp database, which
tal perspective, however, questions this logic as provides information about executive compensa-
fundamentally flawed to the extent that the fail- tion and firm performance for firms in the S&P
ure of governance at one firm can then influence 500, S&P mid-cap 400, and S&P small-cap 600
the competition at other firms through the se- indices at some point since 1992. These data,
lection of the overcompensated CEO as a peer. which are described in detail on the S&P Web
Our study examines how a relatively small site (http://umi.compustat.com), contain infor-
fraction of above normative jumps to the right mation about total yearly compensation for the
tail of the benchmark distribution, which we five highest-paid executives in the firm each year,
describe as “leapfrogging,” could affect the evo- including salary, bonus, stock option grants, and
lution of the entire distribution of executive grants of stock. Basic demographic information
compensation over time. For scholars such as is also available including age and gender and
Holmstrom and Kaplan (2003), leapfrogging date of hire and of separation if applicable. The
(at least if it is “excessive”) becomes a problem starting and ending dates of tenure as CEO are
of “excessive skew” and an illustration of gover- available for CEOs, and mobility among corpo-
nance failure in a small number of firms within rations can be tracked because the names of each
a system of governance that works well because it executive are contained in the data. The Execu-
identifies the appropriate “market wage” for the Comp database contain several important com-
great majority of CEOs. This view is inadequate, pany characteristics including industry, sales,
because the benchmarking system contains feed- assets, number of employees, revenues, return on
back loops that transform governance problems assets, return on equity, net income, stock price,
in some firms into a market reality for others. and yearly return to shareholders (including div-
In this section, we discuss a set of mechanisms idends). We limited the sample to current CEOs
that potentially can link benchmarking to secu- for the analyses reported in this paper.
lar trends in corporate executive compensation, Figure 12.1 shows the tremendous rate of
and then we explore how relatively low rates of increase in mean total compensation for CEOs
leapfrogging can influence the evolution of the in the ExecuComp data. Adjusted for inflation,
 DiPrete, Eirich, and Pittinsky—Leapfrogs and the Surge in Executive Pay • 107

Figure 12.1. Mean CEO Compensation, by Year

the median salary/bonus increased from 1993 of our results to the particular form of the op-
through 2005 by 40%, while the mean increased erationalization. Because of evidence from re-
by 58%. Adjusted for inflation, the median total cent data that actual peer group construction is
compensation went from $1.6 million to $3.2 biased toward more highly compensated peers
million, a 106% increase, while the mean in- (Faulkender and Yang, 2007; Pittinsky and
creased by 116%. In 2006, the typical salary/ DiPrete, 2012), we also operationalize the idea
bonus package dropped considerably in the Ex- that corporation compensation committees
ecuComp data, but the total compensation rose often choose an “aspiration” instead of a compa-
to a mean of $7.1 million, a 175% increase over rable peer group for the process of setting execu-
its value in 1993. In contrast, mean annual com- tive compensation.
pensation in domestic industries in the U.S. as a The first step in our procedure is to simulate
whole rose in this period by only 15%.4 the setting of CEO compensation by bench-
To estimate the impact of leapfrogging on the marking CEOs against their normative (im-
overall rate of growth of executive compensation, puted) peer groups in order to recover a trend
we analyzed compensation trends in the Execu- that approximates what is actually found in the
Comp data by means of counterfactual simula- data. We presume that last year’s compensation
tions. In the historical data, the peer groups that for peer CEOs is observed by the compensation
were actually used by compensation commit- committee, and that they update this lagged
tees are not known (an SEC requirement that compensation distribution by a certain amount
peer groups be named was not instituted until to bring it to present-day levels. We then pre-
2006). What we have instead are the principles sume that compensation committees paid the
that compensation consultants and companies CEO in each year according to his actual rank
say they used in choosing peer groups. We used in the compensation distribution of his peers.
this information to impute peer groups. Because As an important initial step, therefore, we com-
there is more than one way to construct a rea- pared the focal CEO’s compensation in year t
sonable operationalization, we used multiple with the updated distribution of compensation
approaches in order to establish the sensitivity in the peer group, and we thereby identified the
108 • Part II: Inequality in Comparative Perspective

Table 12.1. Definition of Peer Groups increases at the 75th percentile of their escalated
Peer Group Description
peer group distribution unless they were above
the 70th percentile in the previous year, or they
Two-digit SIC codes define
industry groups. Each industry were a new or newly mobile CEO, or if their es-
1. Industry
Priority
group is divided into two peer timated probability of being at the 80th percen-
groups (top and bottom half ) tile or higher in their peer group was in the top
based on market capitalization.
30% of all CEOs based on a logistic regression
One-digit SIC codes define of percentile location in the peer distribution on
industry groups. Each industry
2. Size Priority group is divided into 10 peer
the firm’s market value, change in market value,
groups (deciles) based on market and average change in market value in their peer
capitalization. group.
Industries are divided into In any given year only about 5% of CEOs
four large groups (Agriculture, were capped by our least strict criterion. How-
Mining, and Construction; ever, the identities of capped CEOs differ from
Manufacturing and
Transportation; Wholesale and year to year. The cumulative effect of capping oc-
Retail Trade; Finance, Services, curs because capped CEO compensation affects
3. Performance
and Administration). Each the distribution of compensation in the bench-
group is divided into quintiles
mark distributions. It is this environmental effect
based on market capitalization,
and each market-cap quintile is that causes the suppression of leapfrogging to
divided into quintiles based on affect compensation of other CEOs in our sim-
return on assets. ulations. Because we use actual ranking within
Peer groups are defined as in Size normative peer groups as a key input variable, it
Priority. Each year, executives follows that the capping of any particular CEO’s
are benchmarked against the
4. Aspiration compensation at the 75th percentile of the es-
companies in the same industry
who are one decile larger in calated benchmark distribution would have no
terms of market capitalization. direct effect on his future compensation in our
simulation. Compensation in our model is not a
actual rank of every CEO in his normative peer function of base-year compensation, but only of
group. The ranks of CEOs within normative the compensation of peers and one’s actual rank
peer groups in each year are empirical facts that in that benchmark distribution as computed
are derived from the actual data. each year from the actual compensation data.
We then use these ranks as key input data The cumulative effects come from the spreading
in order to simulate pay increases each year and impact of capping (alternatively, leapfrogging)
thereby approximate the actual rate of increase on benchmark distributions.
of the mean and median of the compensation
distribution over time. Having produced a base-
line simulation that recovers the approximate Results
pattern of compensation change over time via Figures 12.2 and 12.3 show the results for mean
an explicit benchmarking mechanism, we then total compensation and median total compensa-
performed a series of counterfactual simulations tion for the four representations of peer groups.
to estimate the effect of capping compensation The dotted line shows the actual change in
for CEOs who moved to the upper tail of their total compensation at the mean over time. The
benchmark distributions. We did this in three smooth line without year markers (labelled “as
ways, each of which was progressively less restric- simulated” in the legend) shows the movement
tive, first for total compensation and then for of the simulated compensation distribution,
the sum of salary and bonus. In the least restric- which, by construction, approximates the move-
tive simulation, we capped CEO compensation ment of the mean of the actual distribution. The
 109

Figure 12.2. Growth of Simulated Mean Total Compensation

Figure 12.3. Growth of Simulated Median Total Compensation

110 • Part II: Inequality in Comparative Perspective

other three solid lines (each with data markers) capped CEOs the very same compensation that
are the consequences of constraining compen- they did in fact receive but keep this compen-
sation by controlling leapfrogging as defined sation below the 80th percentile of a peer dis-
above. Because the ungoverned simulated dis- tribution if sufficiently highly paid CEOs were
tribution cannot exactly match the true distri- chosen as the peers. In our simulation, we used
bution, one should compare the constrained CEOs of companies that were one decile larger
simulations with the unconstrained simulation in market cap as the way to operationalize aspi-
in order to assess the impact of capping leap- ration peer groups. With this choice, we see that
frog compensation at the 75th percentile on the the capping of compensation jumps continues to
overall growth in executive compensation. The have a noticeable effect on the rate of growth of
specific quantitative features of these different compensation. Had we chosen aspiration groups
growth curves depend upon the assumptions that were two deciles ahead of the focal com-
underlying each simulation and would vary if pany, the effect of capping would no doubt have
we shifted these assumptions. We focus, there- diminished still further. Of course, a CEO who
fore, on the qualitative comparisons, which are is placed in the middle of a distribution of CEOs
stable across the various simulations that we have who are “peers” in name only obscures the jumps
conducted. but does not eliminate them.
Even when we use the least strict tenure and The differing results obtained from the use
performance-based capping, we see a consider- of normative and aspiration peer groups arises
ably slower rate of growth of mean total com- most obviously because leapfrogging is necessar-
pensation when we restrict leapfroggers to the ily less common when aspiration peer groups are
75th percentile of their benchmark distributions. used and the benchmark distribution is thereby
In each of the four panels, it can be seen that shifted upward. This fact is of great importance
the rate of increase of mean total compensation and highlights the dual mechanisms that can be
would still have been very high during the thir- used at the level of specific firms to pay greater
teen year period, going from about $3 million in compensation to an executive. Either one can
real terms in 1993 to nearly $5 million dollars by locate an executive higher up in the normative
2005 for industry- and size-priority peer groups. benchmark distribution, or one can raise the
At the same time, it is apparent in each of the mean of that distribution by composing it of
charts that leapfrogging had a considerable effect higher-paid executives. The use of aspiration peer
on change in mean total compensation; up to groups masks leapfrogging, but amounts to the
half of the growth is removed when leapfrogging same thing for an executive who receives outsize
is controlled. By comparing the size-priority as- increases relative to the executives that constitute
pirations peer groups in the bottom right panel his natural comparison group. Either way, these
with the size-priority peer groups in the top right pay increases then have system-wide implications
panel in Figure 12.2, we see that the effect of as they diffuse to other executives through the
capping (and hence the effect of leapfrogging it- workings of the benchmarking system.
self ) is diminished when aspiration peer groups
are used for benchmarking. The reason for this
is obvious; the higher the compensation of the Discussion
benchmark distribution, the lower will be the The literature on executive compensation rec-
ranking of the focal CEO in that distribution. ognizes that the framework for corporate gov-
The size of his compensation increase depends ernance as it applies to executive compensation
both on where he is placed in the peer distribu- derives from tax and regulatory authorities and
tion and on how elevated the peer distribution is therefore national in scope even as the details
is relative to his own compensation. Clearly, and effectiveness of corporate governance vary
it would be possible to have given almost all by firm. This governance process requires that
 DiPrete, Eirich, and Pittinsky—Leapfrogs and the Surge in Executive Pay • 111

compensation committees pay attention to the of the institutional dynamics that affect both
behavior of other firms when making their de- higher-earning and lower-earning occupations
cisions. In form, the executive compensation (see also Zhou, 1993; Weeden, 2002).
literature is similar to much of the economic so-
ciology literature, which generally pays attention NOTES
to broad environmental context (e.g., resource 1. The standard definition of economic rent is
the difference between the returns to a given factor
dependence theory [Aldrich, 1979], or Fligstein’s
of production and its opportunity cost, i.e., its re-
[2002] “political-cultural” approach), and to the
turns in the best alternative. Those who win access
ways that heterogeneous actors pay attention to to high-paying positions that are in limited supply
the behavior of other actors in their environment can be paid much better in these positions than they
(DiMaggio and Powell, 1983). Coleman (1986) could get elsewhere, which implies a high “rent”
argued that models such as these reflect the component of their earnings.
omnipresent “macro-to-micro” processes that 2. Crystal (1991) found extensive evidence for
cause individual behavior to be constrained by the use of benchmarking in the 1980s. More recently,
the social environment. He further argues that Bizjak et al. (2007) found that 96 of 100 S&P 500
the obverse “micro-to-macro” process, which he firms who were sampled in 1997 asserted that they
used peer groups in their proxies. Ang and Nagel
termed “the means by which purposive actions
(2006) found the same percentage in their S&P 500
of individuals combine to produce a social out- sample. Whether compensation committees actually
come” (p. 1321) is also always present in social consistently behave in ways that compensation con-
action, but is generally underdeveloped in social sultants recommend is, of course, an open question,
science theories. and one whose answer might be different under the
In the case of executive compensation, the current regulatory environment than it was in the
purposive action that forms the micro-to-macro 1990s.
link is rent-seeking on the part of individual 3. Shifts in accounting and tax policies as well
executives. This rent-seeking, when combined as the growing dominance of the shareholder value
paradigm may have contributed to the increased
with individual firm-level problems in corpo-
weight of equity compensation in the typical exec-
rate governance or even when combined with
utive compensation package. Patton (1951) argued
random shocks, affects the environment as well that the high marginal tax rates of the late 1940s
as the outcomes for that individual. Corporate caused companies to limit the ratio of equity to cash
governance affects rent-seeking, but rent-seeking compensation so that their executives would be able
also affects governance in such a way that even to use their cash in order to pay the taxes on their
good governance leads to higher rent extraction. equity compensation. More recently, IRS and ac-
A full elaboration of this feedback loop between counting regulations concerning the implications of
purposive action and the environment remains equity compensation for corporate profitability cre-
ated incentives for corporations to use equity-based
underdeveloped in many other problems of
pay for their top executives.
long-standing interest in economic sociology
4. Table 622, U.S. Statistical Abstract, 2008 (for
(e.g., the study of merger waves), just as it year 2005), and Table 672, U.S. Statistical Abstract,
does in the study of rent-extraction in general 1995 (for year 1993).
and of executive compensation in particular.
Other scholars (e.g., Morgan and Tang, 2007; REFERENCES
DiPrete, 2007; Levy and Temin, 2007) have ar- Aldrich, Howard. 1979. Organizations and Environ-
gued that institutions matter for the evolution ments. Stanford, Calif.: Stanford University Press.
of the compensation structure at the lower end Ang, James S., and Gregory L. Nagel. 2006. “The
of the earnings distribution. In the context of Influence of Peers on CEO Pay Since 1970.” Un-
published manuscript.
the current paper, we would conclude that a
Armstrong, Christopher S., Christopher D. Ittner,
comprehensive explanation for inequality trends and David F. Larcker. 2008. “Economic Char-
in industrial countries must involve a study acteristics, Corporate Governance, and the
112 • Part II: Inequality in Comparative Perspective

Influence of Compensation Consultants on Ex- Fligstein, Neil. 2002. The Architecture of Markets:
ecutive Pay Levels.” Unpublished manuscript. An Economic Sociology of Twenty-first Century
Bebchuk, LA, JM Fried, and DI Walker. 2002. Capitalist Societies. Princeton, N.J., Woodstock:
“Managerial Power and Rent Extraction in the Princeton University Press.
Design of Executive Compensation.” U. Chi. L. Fligstein, Neil, and Jennifer Choo. 2005. “Law and
Rev. 69:751. Corporate Governance.” Annual Review of Law
Bebchuk, Lucian A., and Jesse M. Fried. 2004. Pay and Social Science 1:61–84.
Without Performance: The Unfulfilled Promise of Fligstein, Neil, and Luke Dauter. 2007. “The So-
Executive Compensation. Cambridge, Mass.; Lon- ciology of Markets.” Annual Review of Sociology
don: Harvard University Press. 33:105–128.
Bizjak, John M., Michael L. Lemmon, and Lalitha Gabaix, Xaviar, and Augustin Landier. 2008. “Why
Naveen. 2007. “Has the Use of Peer Groups Has CEO Pay Increased So Much?” Quarterly
Contributed to Higher Levels of Executive Com- Journal of Economics 123:49–100.
pensation?” Unpublished manuscript. Hall, Brian J., and Kevin J. Murphy. 2003. “The
Coleman, James S. 1986. “Social Theory, Social Re- Trouble with Stock Options.” Journal of Eco-
search, and a Theory of Action.” American Jour- nomic Perspectives 17:49–70.
nal of Sociology 91:1309–35. Himmelberg, Charles P., and R. Glenn Hubbard.
Core, John E., Robert W. Holthausen, and David 2000. “Incentive Pay and the Market for CEOs:
F. Larcker. 1999. “Corporate Governance, Chief An Analysis of Pay-For-Performance Sensitivity.”
Executive Officer Compensation, and Firm Working Paper, Columbia University Business
Performance.” Journal of Financial Economics School.
51:371–406. Holmstrom, Bengt, and Steven N. Kaplan. 2003.
Crystal, Graef S. 1991. In Search of Excess: The Over- “The State of U.S. Corporate Governance:
compensation of American Executives. New York: What’s Right and What’s Wrong?” Journal of Ap-
W. W. Norton & Company. plied Corporate Finance 15:8–20.
Davis, Gerald F. 2005. “New Directions in Corpo- Jensen, M.C., and Kevin J. Murphy. 1990. “Perfor-
rate Governance.” Annual Review of Sociology mance Pay and Top-Management Incentives.”
31:143–162. Journal of Political Economy 98:225.
DiMaggio, Paul J., and Walter W. Powell. 1983. Kenworthy, Lane. 2007. “Inequality and Sociology.”
“The Iron Cage Revisited: Institutional Isomor- American Behavioral Scientist 50:584–602.
phism and Collective Rationality.” American So- Levy, Frank, and Peter Temin. 2007. “Inequality and
ciological Review 48:147–160. Institutions in 20th Century America.” Working
DiPrete, Thomas A. 2007. “What has Sociology to Paper 13106. National Bureau of Economic
Contribute to the Study of Inequality Trends? A Research.
Historical and Comparative Perspective.” Ameri- Maurin, Eric, and Fabien Postel-Vinay. 2005. “The
can Behavioral Scientist 50:603–618. European Job Security Gap.” Work and Occupa-
DiPrete, Thomas A., Dominique Goux, Eric Mau- tions 32:229–252.
rin, and Amelie Quesnel-Vallee. 2006. “Work Morgan, Stephen L., and Zun Tang. 2007. “Social
and Pay in Flexible and Regulated Labor Mar- class and workers’ rent, 1983–2001.” Research in
kets: A Generalized Perspective on Institutional Social Stratification and Mobility 25:273–293.
Evolution and Inequality Trends in the U.S. and Murphy, Kevin J., and Ján Zábojník. 2004. “CEO
Europe.” Research in Social Stratification and Mo- Pay and Appointments: A Market-Based Expla-
bility 24:311–332. nation for Recent Trends.” American Economic
Ebbinghaus, Bernard, and Bernard Kittel. 2005. Review 94:192–196.
“European Rigidity Versus American Flexibility? Patton, A. 1951. “Current practices in executive com-
The Institutional Adaptability of Collective Bar- pensation.” Harvard Business Review 29:56–64.
gaining.” Work and Occupations 32:163. Pittinsky, Matthew, and Thomas A. DiPrete. 2012.
Faulkender, Michael W., and Jun Yang. 2007. “In- “Peer Group Ties and Executive Compensation
side the Black Box: The Role and Composition Networks.” Working paper, SSRN.
of Compensation Peer Groups.” Unpublished Porac, Joseph F., James B. Wade, and Timothy G. Pol-
manuscript. lock. 1999. “Industry Categories and the Politics
 Liu and Grusky—The Winners of the Third Industrial Revolution • 113

of the Comparable Firm in CEO Compensation.” Weeden, Kim A. 2002. “Why Do Some Occupa-
Administrative Science Quarterly 44:112–144. tions Pay More than Others? Social Closure and
Reda, James F., Stewart Reifler, and Laura G. Earnings Inequality in the United States.” Amer-
Thatcher. 2008. The Compensation Committee ican Journal of Sociology 108:55–101.
Handbook, Third Edition. New York: John Wiley Zhou, Xueguang. 1993. “Occupational Power, State
and Sons. Capacities, and the Diffusion of Licensing in the
Sørensen, Aage B. 2000. “Toward a Sounder Basis American States: 1980–1950.” American Socio-
for Class Analysis.” American Journal of Sociology logical Review 58:536–52.
105:1523–1558.

13 • Yujia Liu and David B. Grusky

The Winners of the Third Industrial Revolution

The United States and other late industrial coun- sociological literature on postindustrialism,
tries remain in the throes of a “third industrial which, like the SBTC account, treats the rising
revolution” that appears to be reshaping the pay- demand for skilled labor as a central feature of
off for skilled labor in fundamental ways. The contemporary economies (Bell 1999, xvii).
two previous industrial revolutions, defined by This account is so frequently related that
the introduction of steam power in the early one forgets how little it contributes to under-
nineteenth century and the mechanization of standing how the third industrial revolution has
production in the mid-nineteenth century, were changed the demand and payoff for skill. What
characterized by bursts of both skill-biased and types of skills are increasingly in demand and
skill-degrading technologies. The latest revo- increasingly well paid in the third industrial rev-
lution, by contrast, is widely understood as a olution? Is Krueger (1993) right in suggesting
mainly skill-biased one. The “canonical model” that it is a computer revolution in which the pay-
(Acemoglu and Autor 2010) of this third revo- off for computer use and knowledge is taking
lution posits that accelerating technical change off and driving the rise in inequality (Cappelli
(e.g., computerization) has increased the de- and Carter 2000; Black and Lynch 2001; Di-
mand for skilled workers, raised the payoff for Nardo and Pischke 1997)? Or are we instead
skill, and thereby ramped up earnings inequality in the midst of a far broader technocratic revo-
(see also Goldin and Katz 2008; Autor, Katz, lution that entails a growing payoff for scientific
and Kearney 2008). This model is expressed not and other forms of technical knowledge (e.g.,
only in the economics literature on skill-biased Bell 1999; Castells 2000; Powell and Snellman
technological change (SBTC), but also in 2004; Giddens 2008; Esping-Andersen 2009)?

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Yujia Liu and David B. Grusky, “Inequality in the Third Industrial Revolution,”
American Journal of Sociology 118:6 (May 2013), published by the University of Chicago Press. The article
printed here was originally prepared by Yujia Liu and David B. Grusky for this fourth edition of Social Strat-
ification: Class, Race, and Gender in Sociological Perspective, edited by David B. Grusky. Copyright © 2014 by
Westview Press.
114 • Part II: Inequality in Comparative Perspective

Or is creativity rather than knowledge driving viewed in such simple terms, it is typically oper-
the third industrial revolution? Is Florida (2002), ationalized with a measure of schooling, either a
for example, correct in suggesting that there is a continuous measure (e.g., years of schooling) or
burgeoning creative economy and that the rising a categorical one (e.g., high school dropout, high
payoff for creative skills is the core labor mar- school degree, college degree). Worse yet, skill is
ket trend of our time (see also Giddens 2008)? sometimes simply equated with the residual in a
Or are all these theories wrongly defaulting to conventional earnings model, the presumption
a “knowledge theory of value” (Bell 1999, xvii) being that education-based measures of skill are
and overlooking the rising demand for direct- incomplete and that any residual variability in
ing, controlling, and supervising labor? In other earnings can be attributed to unmeasured skill
words, are we witnessing a managerial revolution (see DiPrete 2007; Lemieux 2006; Autor, Katz,
in which the globalization of markets, the com- and Kearney 2008; Card and DiNardo 2000;
plexity of the modern firm, and the “low road” Acemoglu 2002; Katz and Autor 1999; Juhn,
strategy of intensive supervision have created an Murphy, and Pierce 1993). If rising demands
accelerating demand for oversight (see, e.g., Gor- for skill are in fact so central to the third indus-
don 1996)? trial revolution, one might imagine we could be
For each of these theories about the third in- bothered to measure it directly.
dustrial revolution, a supporting literature has of It would be wrong, however, to blame the
course emerged, but such balkanized accounts poor measurement of skill entirely on this will-
are unsatisfying because the competing narra- ingness to default to a unidimensional view of
tives cannot be convincingly assessed in isolation it. Even within the SBTC literature, Neal (1995,
from one another. In all cases, they are at least 670) presciently argued for “directly captur-
partly about the type of labor that is increasingly ing important skill specificities” (see also Card
in demand and increasingly rewarded, and a and Lemieux 1996), whereas Acemoglu (2002,
convincing test accordingly requires us to esti- 13) noted that “models based on a single skill
mate the net returns for each type of skill in the index” cannot well explain recent trends in in-
context of a full model of earnings. The purpose equality, and Kambourov and Manovskii (2009)
of this chapter is to begin that task of developing argued on behalf of the “occupational speci-
and estimating just such a model for the US case. ficity of human capital” (63). It follows that,
How is it that we still know so little about the even among SBTC adherents, there is ample
types of skill underlying the third industrial rev- awareness that the various types of skill are not
olution? The main reason for this state of affairs perfect substitutes for one another and that it
is the relatively poor measurement of skill within is accordingly important to distinguish between
the economics and sociological literatures. The them. Likewise, sociologists who study postin-
entire SBTC apparatus is predicated on the dustrialism and skill (e.g., Handel 2003; Fel-
assumptions that skill can be usefully treated stead et al. 2007) have long known that skill is
as a unidimensional concept, that one can ac- a multidimensional concept, yet their interest in
cordingly divide the labor force into skilled and evaluating hypotheses about skill upgrading has
unskilled workers, and that the third industrial induced them to then revert to a unidimensional
revolution has increased the demand for the for- formulation (e.g., Jonsson 1998; Gallie 1991).
mer category and diminished the demand for the There has been little interest, moreover, in build-
latter (see Handel 2003 for a relevant review). ing credible sociological models of the changing
Although most renditions of the SBTC account payout for this panoply of skills (see Carbonaro
appreciate that skill might be measured in con- 2005, 2006, 2007 for important exceptions; see
tinuous rather than dichotomous form, it is still also Felstead et al. 2007).
understood unidimensionally, with workers sim- The key challenge facing the field is to trans-
ply having more or less of it. Given that skill is late this awareness of skill unsubstitutability
 Liu and Grusky—The Winners of the Third Industrial Revolution • 115

into a full-blown, multidimensional model of computerization of the workplace, that is then

earnings. Insofar as specific types of skills have presumed to raise the demand for skilled labor
been directly measured, the tendency has been to to a level in excess of the available supply. The
estimate earnings models that include just a few conventional SBTC account does not identify
types, thereby leaving the analysis wide open to particular types of skill for which demand is in-
charges of omitted variable bias. Most famously, creasing. In at least some variants, it presumes
Krueger (1993) reported a 15 percent “computer that technological change will uniformly in-
wage premium” in a model that included a mea- crease demand for all types of skill, with the rise
sure of computer skills, a result that DiNardo of computer-based skills serving in this context
and Pischke (1997) later criticized because the as but one illustration of the broader upgrading
putative computer effect could be recast as a pen- process. The accounts reviewed here instead
cil effect by replacing the measure of computer focus on exogenous shifts that may increase de-
usage with one of pencil usage (see Cappelli and mand in skill-specific ways.
Carter 2000; Black and Lynch 2001; Dickerson Will such increases in demand yield a
and Green 2004; Green et al. 2007). long-term increase in the payout for a skill? The
There are of course a few two-factor mod- answer depends on how the supply of the requi-
els of skill. The most famous one rests on the site labor responds. In a competitive economy,
distinction between cognitive and noncogni- workers should respond to price signals by in-
tive skills and shows that the latter contribute vesting in the training needed for high-return
substantially to wages (Heckman, Stixrud, and jobs, thereby increasing the supply of labor for
Urzua 2006; Heckman and Rubinstein 2001; those jobs and ultimately driving down their
Dunifon and Duncan 1998; Moss and Tilly wages. The temporary “quasi-rents” generated by
1996). The two-factor extension is an important a transitory shortage of skilled labor will in this
step forward and has been very influential, yet it way be eliminated over time through the usual
could usefully be extended by assessing whether equilibrating flows (see, e.g., Sørensen 2008).
the so-called cognitive factor is itself a compos- We should therefore observe wage premiums in
ite of many correlated dimensions (e.g., creative the long term only insofar as they rest on (1)
skills, analytic skills, managerial skills). This is abilities that are in fixed supply and cannot be
all to make the seemingly obvious point that one acquired through education or training or (2)
cannot well estimate the effects of particular skill access to training, credentials, or certificates that
types without teasing out the possibly correlated is likewise restricted or rationed. The implica-
effects of other types. Although the need for a tion is that the quasi-rents accruing to those with
truly multidimensional model of skill thus seems high-demand skills will translate into permanent
obvious, such a model is nonetheless strangely rents only when closure of some type prevents
missing from the field (but see Felstead et al. labor from meeting demand.
2007 for an early and important effort). Is this demand-led mechanism the only pos-
sible source of price increases? In principle, the
tastes for particular types of work activities can
Why Does the Payoff for also shift over time, with the result that more or
Skill Change? fewer workers will invest in the skills required for
We begin by considering the main mechanisms such activities. It is often suggested, for example,
by which the price of skills might change. The that women have historically had a special “taste”
most obvious prerequisite for any increase in the for caregiving and are therefore willing to accept
payout for a particular type of skill is that de- relatively low earnings in exchange for the on-the-
mand for that skill exceeds the supply of labor job rewards of tending children or the sick. But
that can provide it. For example, the SBTC ac- what happens when such a taste for caregiving
count features a technical change, such as the becomes less widespread as enlightened parents
116 • Part II: Inequality in Comparative Perspective

Table 13.1. Means and Standard Deviations of Individual-Level Variables

Variable Mean SD
Wage:
Natural log of hourly wage 2.74 .6
Race:
Black .09 …
Hispanic .09 …
Gender:
Female .48 …
Sector:
Manufacturing industry .17 …
Public sector .18 …
Spatial:
Metropolitan .76 …
South .3 …
Demographic:
Not married .4 …
Work experience 19.26 12.39
Part-time worker .17 …
Education:
College degree or higher .26 …
High school degree .61 …

Note: Means and standard deviations are averaged across the survey years. “White” is the omitted category for race and
ethnicity. “High school dropout” is the omitted category for education. Work experience is calculated as age minus years
of education minus 5. Data are from the 1979–2010 Current Population Surveys outgoing rotation group.

and peers encourage women to develop other between the ages of sixteen and sixty-five, and
types of tastes? This is a classic example of a (3) have a valid occupation. We do not exclude
supply-driven source of rising payout. As egal- part-time workers and instead include a control
itarian views spread employers have no choice for part-time status. For the analyses presented
but to raise wages for caregiving, because there here, missing wages are allocated (with the al-
are no longer enough women with a taste for location depending on the major occupation
caregiving that compensates adequately for low group), because we prefer to reduce the amount
caregiving wages. We may accordingly refer to of missing data. We have, however, also repli-
both demand-driven and supply-driven changes cated our analyses after excising the 1994 and
in the prices accruing to different types of skill. 1995 surveys (because those surveys do not iden-
tify cases with allocated wages) and eliminating
all cases in other surveys with missing wage data.
Measuring Skill and Its The analyses based on the full complement of
Determinants survey years (and allocated wages) contain on av-
The foundation for any credible estimate of the erage 84,383 men and 79,308 women per year
payout for skill is a strong individual-level model and yield a grand total of 5,238,144 cases.
of earnings. We estimate our earnings model Throughout the analyses presented here,
with microdata from the monthly outgoing earnings are measured as hourly wages converted
rotation group supplement of the 1979–2010 into constant 2010 dollars using the Consumer
Current Population Surveys (CPS-ORG). In all Price Indexes (see McCully, Moyer, and Stew-
analyses, we include nonmilitary wage and salary art 2007). When wages have been truncated by
workers who (1) are currently working, (2) are the Bureau of Labor Statistics (BLS) to maintain
 Liu and Grusky—The Winners of the Third Industrial Revolution • 117

confidentiality (i.e., “topcoding”), we have pro- on each skill variable range from 0 (low skill)
ceeded by multiplying such wages by 1.4, which to 6 (high skill) and are reported by O*NET as
is the conventional approach (e.g., Card and occupational means (averaged across all raters).
DiNardo 2000). We then model logged hourly We have used the O*NET databases to build
wages as a function of race, gender, sector, re- an eight-factor representation of workplace skill
gion, marital status, work experience, employ- (see Felstead et al. 2007 for a related scheme). As
ment status, and education. These variables, shown in Table 13.2, each of the eight workplace
which are presented in Table 13.1, are opera- factors captures a form of cognitive, creative,
tionalized in standard ways, as reported in the technical, or social skill, and each is measured as
note to that table. This table further reports the a composite based on two to fifteen indicators.
means and standard deviations (where appropri- The resulting loadings, which are also presented
ate) for each of the variables. in Table 13.2, are based on a confirmatory fac-
The centerpiece of our effort is a compre- tor analysis of the O*NET skills for the 494 oc-
hensive measurement of workplace skill using cupations in the SOC scheme. These estimates
occupation-level data from the Occupational come from a model that constrains the loadings
Information Network (O*NET). The O*NET for the incumbent and analyst skills to be equal
data, which include both analyst and incumbent (which ensures that the factors are constructed
ratings of occupational skills, are a new resource identically over time and are thereby nominally
designed to build on and extend the Dictionary comparable).
of Occupational Titles (DOT). The “analyst rat- The foregoing analyses are complicated by
ings” of occupations are just the long-standing changes in the SOC occupational scheme in
DOT expert ratings of occupational skills, but the 1983, 1992, and 2003 CPS-ORG surveys.
O*NET analysts have recoded them into the The O*NET ratings are only available for 2000
highly detailed occupations defined by the 2000 SOC codes, yet the CPS-ORG data are coded
Standard Occupational Classification System into simpler precursor schemes for all years prior
(SOC). These ratings pertain to occupational to 2003. How can we assign skill ratings to oc-
skills when the DOT was last updated, which cupations appearing in these precursor schemes?
for most occupations was in 1977, but for a This problem is easily addressed because the BLS
small subset was in 1991. The O*NET archive has double-coded the occupations of a subset of
also provides a new “incumbent database” with respondents (see Weeden and Grusky 2005).
more recent evidence on occupational skills. With the double-coded results in hand, we can
These ratings were developed by identifying identify the subset of 2000 SOC occupations
organizations that employ workers in targeted into which the incumbents of precursor occu-
occupations, randomly sampling those workers, pations would have been coded, allowing us to
and then distributing a detailed survey of the oc- calculate the rating for a precursor occupation as
cupational skills required in their jobs. Because a simple weighted average of the ratings for the
this survey was designed to be consistent with constituent 2000 SOC occupations. We rely on
the DOT analyst survey, the two ratings are this approach throughout the analyses.
nominally comparable, with the analyst ratings
pertaining to occupational skills in 1977 (and in
some cases 1991) and the incumbent ratings per- Trends in the Price of Skills
taining to occupational skills in the 2000s (see We proceed by estimating a hierarchical linear
Peterson et al. 1999). Although the DOT rat- model. This model allows us to first eliminate
ings were completed by analysts and the O*NET any individual-level compositional differences
ratings by incumbents, we have elsewhere devel- across occupations and then estimate the effects
oped methods to render them comparable (see of skill on the resulting purged occupational
Liu and Grusky 2013 for details). The ratings means. We can represent this model as follows:
118 • Part II: Inequality in Comparative Perspective

Table 13.2. Indicators and Loadings for Workplace Skill Factors

Skill Factor Correlation Skill Factor Correlation

Cognitive: Creative:
Verbal: Creative:
Oral comprehension .93 Originality .95
Written comprehension .93 Thinking creatively .79
Oral expression .93 Innovation .81
Written expression .95 Technical:
Reading comprehension .91 Computer:
Active listening .88 Programming .76
Writing .92 Computers and electronics .82
Speaking .89 Interacting with computers .93
Quantitative: Science and engineering
Mathematical reasoning .94 Operations analysis .69
Number facility .86 Technology design .96
Mathematics .84 Engineering and technology .83
Analytic: Design .86
Fluency of ideas .79 Mechanical .85
Problem sensitivity .87 Physics, Chemistry, Biology .53
Deductive reasoning .95 Social:
Inductive reasoning .96 Managerial:
Critical thinking .91 Management of financial resources .83
Active learning .87 Management of material resources .87
Learning strategies .89 Management of personnel resources .85
Problem identification .87 Admin and Management .94
Information gathering .83 Developing and building teams .77
Information organization .85 Directing and motivating subordinates .83
Synthesis and reorganization .96 Care work:
Idea generation .91 Service orientation .95
Idea evaluation .93 Assisting and caring for others .82
Implementation planning .80
Solution appraisal .83

Note: Data are from the Occupational Information Network.

P
Xpij and Zsj are the pth and sth variables in the
InWageij = β0 j + ∑ β p X pij + rij
p =1 individual-level and occupation-level equations.
S (1) The individual-level model is intended to re-
β0 j = γ 00 + ∑ γ s Z sj + μ0 j
s =1
move compositional differences across occu-
pations and thus allow the true effects of skills
on wages to be revealed. These compositional
where i refers to individuals, j refers to oc- differences, if left uncontrolled, could bias our
cupations, p indexes the variables in the estimates of returns on skill.
individual-level equation (and P refers to the The estimates for workplace skill are pre-
number of such variables), s indexes the skill sented in Figure 13.1. The first and most
variables in the occupation-level equation (and important conclusion is that the payoff for
S refers to the number of such variables), and workplace skill is not increasing in any simple,
 Liu and Grusky—The Winners of the Third Industrial Revolution • 119

Figure 13.1. Wage Payoff for Workplace Skills

1.10 1.10

1.08 1.08

1.06 1.06

1.04 1.04

1.02 1.02

1.00 1.00
9

0
7

1
19

19

19

19

20

20

20

19

19

19

19

20

20

20
1.10 1.10

1.08 1.08

1.06 1.06

1.04 1.04

1.02 1.02

1.00 1.00
79

85

90

95

00

05

10

79

85

90

95

00

05

10
19

19

19

19

20

20

20

19

19

19

19

20

20

20
Estimates are from the model of equation (1). S & E = science and engineering.

across-the-board fashion. Of our eight skill Although skill is often treated in undifferen-
measures, the payout is increasing for only four tiated ways by SBTC and postindustrial theo-
(analytical, computer, managerial, nurturing), rists, some variants of these accounts imply an
is decreasing for three (quantitative, verbal, cre- especially prominent takeoff in the payout for
ative), and is stable for one (S&E). This result cognitive skills (e.g., Bell 1999) or computer
is inconsistent with postindustrial or SBTC skills (e.g., Krueger 1993). The results in Figures
accounts that treat skill in an undifferentiated 13.1a and 13.1b reveal that, contrary to these
fashion and imply that the payout for all skills more delimited accounts, the payout for verbal
should be rising. If one wishes to explain the rise and quantitative skills is actually decreasing. At
of inequality through skill bias, it surely does not the same time, the payout for computer skill is
help that some skills have a diminishing payoff increasing (see Figure 13.1c), but it is not a very
(i.e., are converging to zero) and are therefore steep increase and can hardly figure prominently
introducing ever less heterogeneity into the wage in explaining the takeoff in inequality. The up-
distribution. The steady increase in the demand shot is that neither a generic nor a delimited ver-
for most skills is consistent with a postindustrial sion of SBTC receives much support in Figure
or SBTC account (see Liu and Grusky 2013 13.1 (cf. Dickerson and Green 2004; Green et
for analyses of changing demand). However, al. 2007).
because such changes do not reliably translate There is only one type of cognitive skill for
into a rising payoff, any account of the takeoff which the payout is indeed taking off. As Figure
in inequality has to be more complicated than 13.1a reveals, the rising returns on analytical skill
suggested in the standard commentary. are striking, with the slope well in excess of that
120 • Part II: Inequality in Comparative Perspective

for any other workplace skill. We are not aware hence drive down wages for that labor (despite
of an SBTC or postindustrialist account that sin- the rising demand for it).
gles out analytical skill in a way consistent with The SBTC and postindustrial accounts also
this result. In fact, Bell (1999) places rather more have not attended much to the increasing de-
emphasis on the rise of purely technical knowl- mand for social skills, an omission that arises
edge, as do many SBTC analysts (e.g., Krueger because this development is again driven by
1993). institutional forces (e.g., the growing marketi-
Why, then, are analytical skills the unex- zation of carework), not the conventionally
pected winner over the last three decades? The privileged technical ones (cf. Garicano and
demand for these skills may be increasing be- Rossi-Hansberg 2004). We have shown else-
cause (1) they require intuitive problem solving where (Liu and Grusky 2013) that the demand
that cannot easily be replaced with computer for social work, both in its managerial and nur-
programming or software (unlike quantitative turing forms, has taken off in the third industrial
and verbal skills), and (2) the accelerating “cre- revolution. In Figure 13.1d, we show that the
ative destruction” of modern capitalism places payoff for managerial and nurturing skills is also
a growing premium on innovation, problem increasing, albeit from very different bases. The
solving, and rapid response to changing mar- penalty for nurturing skill was once quite sub-
ket conditions (e.g., Castells 2000). On the stantial, but over the course of thirty years that
supply side, elite university credentials are now penalty diminished by a full 53.2 percent. The
required for almost all analytical positions, yet premium on managerial skill started off, by con-
such credentials are rationed rather than sold trast, from a small positive base and rose by 75.9
in a profit-maximizing fashion. The resulting percent from 1979 to 2010. These results sug-
supply-demand imbalance could bring about ris- gest that increases in demand have not triggered
ing returns of the sort revealed in Figure 13.1a. corresponding increases in supply. In Liu and
The case of creative skill provides an instruc- Grusky (2013), we have speculated that a sup-
tive contrast. Although analytical and creative ply bottleneck for managers arises from a grow-
skills are both in increasingly high demand (see ing reliance on credential-based recruitment,
Liu and Grusky 2013), the payout for creative whereas a supply bottleneck for carework arises
skills, unlike that for analytical skills, is rapidly because women are now less likely to develop
deteriorating. As Figure 13.1b shows, there is a and pursue “tastes” for nurturing (Grusky and
long-standing wage penalty for creative skills, Weeden 2011). As England (2010) has noted,
which speaks to the widespread taste for creative gender-typed tastes have evolved in a very asym-
work and an associated willingness to forego earn- metrical fashion, with women adopting char-
ings for the privilege of holding a creative job. acteristically male tastes without men in turn
The more striking result shown in Fig- adopting characteristically female tastes. This
ure 13.1b, however, is that this wage penalty pattern of change is probably responsible in part
is growing ever larger, even as the demand for for a burgeoning supply bottleneck for nurturing
creative work increases and as some commen- skill and consequent higher prices for that skill.
tators (e.g., Florida 2002) celebrate the rise of a
creative economy. What is driving this decline?
It has long been argued that the postmaterial- Conclusions
ist turn (e.g., Inglehart 2008) is creating a bur- The foregoing results imply that the third in-
geoning class of would-be dancers, journalists, dustrial revolution is not a computer revolu-
poets, sculptors, writers, artists, and all manner tion, not a technocratic revolution, and not a
of creative types. The simple hypothesis here is creative revolution. What type of revolution is
that late industrialism creates a reserve army of it? The most striking result here is that rewards
workers who are captivated by creative labor and are increasingly going to those who engage in
 Liu and Grusky—The Winners of the Third Industrial Revolution • 121

critical thinking, problem solving, and deduc- It is not simply that institutional change
tive reasoning. The return to managerial skill often drives the demand for skill. Even more
also increased rapidly and by 2010 was higher important, our institutions also affect how labor
than that of any other skill except analytical. It responds to changes in demand, in particular
follows that there are two roads to getting ahead whether labor can take advantage of rising prices
in the third industrial era, the first being prob- for particular skills by acquiring them. If the
lem solving, innovation, and critical thinking labor market does not allow workers to freely
(i.e., the “intellectual road”) and the second flow to such opportunities, then the price of
being direct authority and supervision (i.e., the some skills can remain high for protracted pe-
“command road”). The payoff for technical skill riods of time. The price of analytical labor, for
(e.g., computer use), which is typically featured example, remains unusually high because ana-
in discussions of the third industrial revolution, lytical positions are typically only open to work-
has been oversold and simply does not compare ers with elite educational credentials, and such
to the payoff for analysis and command. credentials are rationed out by major private
We suspect that scholars have become fixated universities rather than being freely sold on the
on computers because of the presumption that market (Grusky and Weeden 2011). The supply
technical innovation is the main force behind bottleneck for nurturing skill is partly attribut-
the changing payoff for skill. The evidence from able, by contrast, to an emerging commitment
our models suggests that, contrary to this view, among enlightened parents, teachers, and peers
the real driving force is skill-biased institutional to draw out and cultivate a broader range of tal-
change rather than skill-biased technological ents among girls and young women (see England
change. The role of institutions in generating 2010).
skill-biased change is twofold. First, the modified The ruthlessly technological lens thus misses
demand for skill often originates in institutional the rise of a new, fast-moving form of capital-
rather than technical change, an obvious exam- ism in which market conditions change rap-
ple of this process being the effects of rising fe- idly, stock markets become difficult to fathom,
male labor force participation on the demand for product cycles become shorter, and the payoff
child care, nursing, and other nurturing skills. As for inspired and insightful analysis is accordingly
women increasingly enter the formal economy, ratcheted up. It does not hurt, moreover, that the
many types of formerly domestic work are grad- winners of this analytical contest have access to
ually marketized, and the demand for nurturing increasingly global markets and are therefore in-
skills accordingly increases. Likewise, the rising creasingly bigger winners. The financial analyst
demand for managerial work partly reflects our who devises a successful new financial product
institutional commitment to low-road supervi- stands to reap substantial returns from a market
sory forms of organization, whereas the growing that is now more nearly global. With a few im-
demand for analytical labor reflects the acceler- portant exceptions (Frank and Cook 1996; Saez
ating creative destruction of modern capitalism 2008), the literature on inequality has ignored
and the associated premium on innovation, the role of globalization in opening up new op-
problem solving, and rapid response to changing portunities for the privileged, preferring instead
market conditions. These institutional develop- to focus on how it harms the less privileged by
ments would not appear to be simple reactions to weakening unions and exposing them to com-
technological change. When the world is none- petition from foreign workers (see Western and
theless viewed through the blinders of technolog- Rosenfeld 2011). The increasing return on ana-
ical change, the result is theories about the rising lytical labor suggests that globalization may be
payoff for technical skill that ignore how broader a many-headed hydra that increases inequality
institutional developments have also altered the not just by undermining returns at the bottom
demand for analytical and social skills. of the class structure, but also by affording new
122 • Part II: Inequality in Comparative Perspective

opportunities to collect marginal product and DiNardo, John E., and Jorn-Steffen Pischke. 1997.
rent at the top. “The Returns to Computer Use Revisited: Have
Pencils Changed the Wage Structure Too?” Quar-
REFERENCES terly Journal of Economics 112 (1): 291–303.
Acemoglu, Daron. 2002. “Technical Change, In- DiPrete, Thomas A. 2007. “What Has Sociology to
equality, and the Labor Market.” Journal of Eco- Contribute to the Study of Inequality Trends? A
nomic Literature 60: 7–72. Historical and Comparative Perspective.” Ameri-
Acemoglu, Daron, and David Autor. 2010. “Skills, can Behavioral Scientist 50 (5): 603–618.
Tasks and Technologies: Implications for Em- Dunifon, Rachel, and Greg J. Duncan. 1998.
ployment and Earnings.” In Handbook of Labor “Long-Run Effects of Motivation on
Economics, edited by Orley Ashenfelter and David Labor-Market Success.” Social Psychology Quar-
E. Card, 4(B): 1043–1171. Amsterdam: Elsevier. terly 61 (1): 33–48.
Autor, David H., Lawrence F. Katz, and Melissa S. England, Paula. 2010. “Gender Revolution: Uneven
Kearney. 2008. “Trends in U.S. Wage Inequality: and Stalled.” Gender and Society 24 (2): 149–166.
Revising the Revisionists.” Review of Economics Esping-Andersen, Gøsta. 2009. Changing Classes:
and Statistics 90 (2): 300–323. Stratification and Mobility in Post-Industrial So-
Bell, Daniel. 1999. The Coming of Post-Industrial So- cieties. Newbury Park, CA: Sage.
ciety: A Venture in Social Forecasting. New York: Felstead, Alan, Duncan Gallie, Francis Green, and
Basic Books. Ying Zhou. 2007. “Skills at Work, 1986 to
Black, Sandra E., and Lisa M. Lynch. 2001. “How to 2006.” Working paper, ESRC Centre on Skills,
Compete: The Impact of Workplace Practices and Knowledge and Organisational Performance,
Information Technology on Productivity.” Review University of Oxford.
of Economics and Statistics 83 (3): 434–445. Florida, Richard. 2002. The Rise of the Creative Class:
Cappelli, Peter, and William Carter. 2000. “Com- And How It’s Transforming Work, Leisure, Commu-
puters, Work Organization, and Wage Out- nity, and Everyday Life. New York: Basic Books.
comes.” NBER Working Paper. Frank, Robert H., and Philip J. Cook. 1996. The
Carbonaro, William. 2005. “Explaining Variable Winner-Take-All Society. New York: Penguin.
Returns to Cognitive Skill across Occupations.” Gallie, Duncan. 1991. “Patterns of Skill Change:
Social Science Research 34: 165–188. Upskilling, Deskilling or the Polarisation of
———. 2006. “Cross-National Differences in the Skills.” Work, Employment, and Society 5 (3):
Skill-Earnings Relationship: The Role of Skill 319–351.
Demands and Labor Market Institutions.” Social Garicano, L., and E. Rossi-Hansberg. 2004. “In-
Forces 84: 1819–1842. equality and the Organization of Knowledge
———. 2007. “The Effects of Education and Cog- Economy.” American Economic Review 94:
nitive Skill on Earnings: How Much Do Oc- 197–202.
cupations and Jobs Matter?” Research in Social Giddens, Anthony. 2008. “Social Justice and Social
Stratification and Mobility 25: 57–71. Divisions.” In Social Stratification: Class, Race,
Card, David, and John DiNardo. 2000. “Skill-Biased and Gender in Sociological Perspective, edited by
Technological Change and Rising Wage Inequal- David B. Grusky, in collaboration with Manwai
ity: Some Problems and Puzzles.” Journal of Labor C. Ku and Szonja Szelényi, 1016–1022. Boulder,
Economics 20 (4): 733–783. CO: Westview Press.
Card, David, and Thomas Lemieux. 1996. “Wage Goldin, Claudia, and Lawrence F. Katz. 2008. The
Dispersion, Returns to Skill, and Black-White Race Between Education and Technology. Cam-
Wage Differentials.” Journal of Econometrics 74: bridge, MA: Harvard University Press.
319–361. Gordon, David M. 1996. Fat and Mean: The Corpo-
Castells, Manuel. 2000. The Rise of the Network So- rate Squeeze of Working Americans and the Myth
ciety. Malden, MA: Blackwell. of Managerial “Downsizing”. New York: The Free
Dickerson, Andy, and Francis Green. 2004. “The Press.
Growth and Valuation of Computing and Other Green, Francis, Alan Felstead, Duncan Gallie, and
Generic Skills.” Oxford Economic Papers 56 (3): Ying Zhou. 2007. “Computers and Pay.” Na-
371–406. tional Institute Economic Review 201 (63): 63–75.
 Liu and Grusky—The Winners of the Third Industrial Revolution • 123

Grusky, David B., and Kim A. Weeden. 2011. “Is Rising Demand for Skill?” American Economic
Market Failure Behind the Takeoff in Inequal- Review 96 (3): 461–498.
ity?” In The Inequality Reader, edited by David Lui, Yujia, and David B. Grusky. 2013. “Inequality
Grusky and Szonja Szelenyi, 90–97. Boulder, in the Third Industrial Revolution.” American
CO: Westview Press. Journal of Sociology 118 (6): 1330–1374.
Handel, Michael J. 2003. Implications of Informa- McCully, Clinton P., Brian C. Moyer, and Kenneth
tion Technology for Employment, Skills, and Wages: J. Stewart. 2007. “Comparing the Consumer
A Review of Recent Research. Arlington, VA: SRI Price Index and Personal Consumption Expen-
International. ditures Price Index.” Working paper, U.S. Bureau
Heckman, James J., and Yona Rubinstein. 2001. of Economic Analysis.
“The Importance of Noncognitive Skills: Les- Moss, Philip, and Chris Tilly. 1996. “‘Soft Skills
sons from the GED Testing Program.” American and Race: An Investigation of Black Men’s Em-
Economic Review 91 (2): 145–149. ployment Problems.” Work and Occupations 23:
Heckman, James J., Jora Stixrud, and Sergio Urzua. 252–276.
2006. “The Effects of Cognitive and Noncog- Neal, Derek. 1995. “Industry-Specific Human Cap-
nitive Abilities on Labor Market Outcomes and ital: Evidence from Displaced Workers.” Journal
Social Behavior.” Journal of Labor Economics 24 of Labor Economics 13 (4): 653–677.
(3): 411–482. Peterson, Norman G., Michael Mumford, Walter
Inglehart, Ronald F. 2008. “Changing Values among C. Borman, Richard P. Jeanneret, and Edwin A.
Western Publics from 1970 to 2006.” West Euro- Fleishman, eds. 1999. An Occupational Informa-
pean Politics 31 (1–2): 130–146. tion System for the 21st Century: The Development
Jonsson, Jan O. 1998. “Class and the Changing Na- of O*NET. Washington, DC: American Psycho-
ture of Work: Testing Hypotheses of Deskilling logical Association.
and Convergence among Swedish Employees.” Powell, Walter W., and Kaisa Snellman. 2004. “The
Work, Employment and Society 12: 603–633. Knowledge Economy.” Annual Review of Sociol-
Juhn, Chinhui, Kevin M. Murphy, and Brooks ogy 30: 199–220.
Pierce. 1993. “Wage Inequality and the Rise in Saez, Emmanuel. 2008. “Striking It Richer: The
Returns to Skill.” Journal of Political Economy 101 Evolution of Top Incomes in the United States.”
(3): 410–442. Pathways Magazine (Winter 2008): 6–7.
Kambourov, Gueorgui, and Lourii Manovskii. 2009. Sørensen, Aage. 2008. “Foundations of a Rent-Based
“Occupational Specificity of Human Capital.” Class Analysis.” In Social Stratification: Class,
Review of Economic Studies 76 (2): 731–759. Race, and Gender in Sociological Perspective, ed-
Katz, Lawrence F., and David H. Autor. 1999. ited by David B. Grusky, in collaboration with
“Changes in the Wage Structure and Earnings In- Manwai C. Ku and Szonja Szelényi, 219–234.
equality.” In Handbook of Labor Economics, edited Boulder, CO: Westview Press.
by O. Ashenfelter and D. Card, 3A:1463–1555. Weeden, Kim A., and David B. Grusky. 2005. “The
New York: Elsevier. Case for A New Class Map.” American Journal of
Krueger, Alan B. 1993. “How Computers Have Sociology 111 (1): 141–212.
Changed the Wage Structure: Evidence from Western, Bruce, and Jake Rosenfeld. 2011. “Unions,
Microdata, 1984–1989.” The Quarterly Journal Norms, and the Rise in U.S. Wage Inequality.”
of Economics 108 (1): 33–60. American Sociological Review 76 (4): 512–537.
Lemieux, Thomas. 2006. “Increasing Residual Wage
Inequality: Composition Effects, Noisy Data, or
 PART III

The Structure of Inequality

Marxian Theories of Class

14 Alienation and Social Classes 127
Classes in Capitalism and Pre-Capitalism 131
Ideology and Class 141
15 Class and Class Conflict in Industrial Society 143
16 A General Framework for the Analysis of Class Structure 149
17 Class Conflict in the Capitalist World Economy 162

Weberian Theories of Class

18 Class, Status, Party 165
Status Groups and Classes 175
Open and Closed Relationships 179
19 The Class Structure of the Advanced Societies 183
20 Marxism and Class Theory: A Bourgeois Critique 193
21 Is There a Status Order in Contemporary British Society? 202

Durkheimian Theories of Class

22 The Division of Labor in Society 217
23 The Changing Form of Inequality 222

Class Gradationalism
24 Occupational Prestige in Comparative Perspective 233
25 Occupational Grading and Occupational Prestige 237
26 Prestige or Socioeconomic Scales in the Study of Occupational
Achievement? 244
27 Socioeconomic Indexes for Occupations: A Review,
Update, and Critique 246
126 • Part III: The Structure of Inequality

The New Gradationalism?

28 Foundations of a Rent-Based Class Analysis 251
29 From Income Inequality to Economic Inequality 260
 Marx—Alienation and Social Classes • 127

Marxian Theories of Class

14 • Karl Marx

Alienation and Social Classes

We shall begin from a contemporary economic deprived of the most essential things not only of
fact. The worker becomes poorer the more life but also of work. Labour itself becomes an
wealth he produces and the more his produc- object which he can acquire only by the great-
tion increases in power and extent. The worker est effort and with unpredictable interruptions.
becomes an ever cheaper commodity the more So much does the appropriation of the object
goods he creates. The devaluation of the human appear as alienation that the more objects the
world increases in direct relation with the in- worker produces the fewer he can possess and
crease in value of the world of things. Labour the more he falls under the domination of his
does not only create goods; it also produces itself product, of capital.
and the worker as a commodity, and indeed in the All these consequences follow from the fact
same proportion as it produces goods. that the worker is related to the product of his
This fact simply implies that the object labour as to an alien object. For it is clear on
produced by labour, its product, now stands this presupposition that the more the worker
opposed to it as an alien being, as a power inde- expends himself in work the more powerful
pendent of the producer. The product of labour becomes the world of objects which he creates
is labour which has been embodied in an object in face of himself, the poorer he becomes in his
and turned into a physical thing; this product is inner life, and the less he belongs to himself. It
an objectification of labour. The performance of is just the same as in religion. The more of him-
work is at the same time its objectification. The self man attributes to God the less he has left in
performance of work appears in the sphere of himself. The worker puts his life into the object,
political economy as a vitiation of the worker, and his life then belongs no longer to himself but
objectification as a loss and as servitude to the ob- to the object. The greater his activity, therefore,
ject, and appropriation as alienation. the less he possesses. What is embodied in the
So much does the performance of work ap- product of his labour is no longer his own. The
pear as vitiation that the worker is vitiated to the greater this product is, therefore, the more he is
point of starvation. So much does objectification diminished. The alienation of the worker in his
appear as loss of the object that the worker is product means not only that his labour becomes

Karl Marx. “Alienation and Social Classes.” “The Economic and Philosophical Manuscripts,” in Karl Marx:
Early Writings, edited and translated by T. B. Bottomore, pp. 121–131. Published by C. A. Watts (London)
and McGraw-Hill (New York). Copyright © T. B. Bottomore 1963, 1964. Reprinted by permission of The
McGraw-Hill Companies and the Heirs of T. B. Bottomore. The Holy Family: A Critique of Critical Criti-
cism,” in The Marx-Engels Reader by Karl Marx and Friedrich Engels, edited by Robert C.Tucker, pp. 133–135.
Copyright © 1978, 1972 by W. W. Norton & Company, Inc. Used by permission of—W. W. Norton &
Company, Inc.
128 • Part III: The Structure of Inequality

an object, assumes an external existence, but that the worker is not his own spontaneous activ-
it exists independently, outside himself, and alien ity. It is another’s activity and a loss of his own
to him, and that it stands opposed to him as an spontaneity.
autonomous power. The life which he has given We arrive at the result that man (the worker)
to the object sets itself against him as an alien feels himself to be freely active only in his animal
and hostile force. . . . functions—eating, drinking and procreating,
So far we have considered the alienation of the or at most also in his dwelling and in personal
worker only from one aspect; namely, his relation- adornment—while in his human functions he
ship with the products of his labour. However, alien- is reduced to an animal. The animal becomes
ation appears not merely in the result but also in human and the human becomes animal.
the process of production, within productive activity Eating, drinking and procreating are of
itself. How could the worker stand in an alien re- course also genuine human functions. But ab-
lationship to the product of his activity if he did stractly considered, apart from the environment
not alienate himself in the act of production itself? of human activities, and turned into final and
The product is indeed only the résumé of activity, sole ends, they are animal functions.
of production. Consequently, if the product of We have now considered the act of alienation
labour is alienation, production itself must be ac- of practical human activity, labour, from two as-
tive alienation—the alienation of activity and the pects: (1) the relationship of the worker to the
activity of alienation. The alienation of the object product of labour as an alien object which domi-
of labour merely summarizes the alienation in the nates him. This relationship is at the same time
work activity itself. the relationship to the sensuous external world,
What constitutes the alienation of labour? to natural objects, as an alien and hostile world;
First, that the work is external to the worker, (2) the relationship of labour to the act of pro-
that it is not part of his nature; and that, con- duction within labour. This is the relationship of
sequently, he does not fulfil himself in his work the worker to his own activity as something alien
but denies himself, has a feeling of misery rather and not belonging to him, activity as suffering
than well-being, does not develop freely his (passivity), strength as powerlessness, creation as
mental and physical energies but is physically emasculation, the personal physical and mental
exhausted and mentally debased. The worker, energy of the worker, his personal life (for what is
therefore, feels himself at home only during his life but activity?), as an activity which is directed
leisure time, whereas at work he feels homeless. against himself, independent of him and not be-
His work is not voluntary but imposed, forced longing to him. This is self-alienation as against
labour. It is not the satisfaction of a need, but the above-mentioned alienation of the thing.
only a means for satisfying other needs. Its alien We have now to infer a third characteris-
character is clearly shown by the fact that as soon tic of alienated labour from the two we have
as there is no physical or other compulsion it is considered.
avoided like the plague. External labour, labour Man is a species-being not only in the sense
in which man alienates himself, is a labour of that he makes the community (his own as well as
self-sacrifice, of mortification. Finally, the exter- those of other things) his object both practically
nal character of work for the worker is shown by and theoretically, but also (and this is simply an-
the fact that it is not his own work but work for other expression for the same thing) in the sense
someone else, that in work he does not belong to that he treats himself as the present, living spe-
himself but to another person. cies, as a universal and consequently free being.
Just as in religion the spontaneous activity of Species-life, for man as for animals, has its
human fantasy, of the human brain and heart, physical basis in the fact that man (like animals)
reacts independently as an alien activity of gods lives from inorganic nature, and since man is
or devils upon the individual, so the activity of more universal than an animal so the range of
 Marx—Alienation and Social Classes • 129

inorganic nature from which he lives is more a conscious life activity. It is not a determina-
universal. Plants, animals, minerals, air, light, tion with which he is completely identified.
etc. constitute, from the theoretical aspect, a part Conscious life activity distinguishes man from
of human consciousness as objects of natural sci- the life activity of animals. Only for this rea-
ence and art; they are man’s spiritual inorganic son is he a species-being. Or rather, he is only a
nature, his intellectual means of life, which he self-conscious being, i.e. his own life is an object
must first prepare for enjoyment and perpetu- for him, because he is a species-being. Only for
ation. So also, from the practical aspect, they this reason is his activity free activity. Alienated
form a part of human life and activity. In prac- labour reverses the relationship, in that man be-
tice man lives only from these natural products, cause he is a self-conscious being makes his life
whether in the form of food, heating, clothing, activity, his being, only a means for his existence.
housing, etc. The universality of man appears The practical construction of an objective
in practice in the universality which makes the world, the manipulation of inorganic nature,
whole of nature into his inorganic body: (1) as is the confirmation of man as a conscious
a direct means of life; and equally (2) as the ma- species-being, i.e. a being who treats the species
terial object and instrument of his life activity. as his own being or himself as a species-being.
Nature is the inorganic body of man; that is to Of course, animals also produce. They construct
say nature, excluding the human body itself. To nests, dwellings, as in the case of bees, beavers,
say that man lives from nature means that na- ants, etc. But they only produce what is strictly
ture is his body with which he must remain in a necessary for themselves or their young. They
continuous interchange in order not to die. The produce only in a single direction, while man
statement that the physical and mental life of produces universally. They produce only under
man, and nature, are interdependent means sim- the compulsion of direct physical needs, while
ply that nature is interdependent with itself, for man produces when he is free from physical
man is a part of nature. need and only truly produces in freedom from
Since alienated labour: (1) alienates nature such need. Animals produce only themselves,
from man; and (2) alienates man from himself, while man reproduces the whole of nature. The
from his own active function, his life activity; products of animal production belong directly
so it alienates him from the species. It makes to their physical bodies, while man is free in
species-life into a means of individual life. In the face of his product. Animals construct only in
first place it alienates species-life and individual accordance with the standards and needs of the
life, and secondly, it turns the latter, as an ab- species to which they belong, while man knows
straction, into the purpose of the former, also in how to produce in accordance with the standards
its abstract and alienated form. of every species and knows how to apply the
For labour, life activity, productive life, now appropriate standard to the object. Thus man
appear to man only as means for the satisfaction constructs also in accordance with the laws of
of a need, the need to maintain his physical ex- beauty.
istence. Productive life is, however, species-life. It is just in his work upon the objective world
It is life creating life. In the type of life activ- that man really proves himself as a species-being.
ity resides the whole character of a species, its This production is his active species-life. By
species-character; and free, conscious activity is means of it nature appears as his work and his
the species-character of human beings. Life itself reality. The object of labour is, therefore, the ob-
appears only as a means of life. jectification of man’s species-life; for he no longer
The animal is one with its life activity. It does reproduces himself merely intellectually, as in
not distinguish the activity from itself. It is its consciousness, but actively and in a real sense,
activity. But man makes his life activity itself and he sees his own reflection in a world which
an object of his will and consciousness. He has he has constructed. While, therefore, alienated
130 • Part III: The Structure of Inequality

labour takes away the object of production from whom does it belong? If my own activity does
man, it also takes away his species-life, his real ob- not belong to me but is an alien, forced activ-
jectivity as a species-being, and changes his ad- ity, to whom does it belong? To a being other
vantage over animals into a disadvantage in so far than myself. And who is this being? The gods?
as his inorganic body, nature, is taken from him. It is apparent in the earliest stages of advanced
Just as alienated labour transforms free and production, e.g. temple building, etc. in Egypt,
self-directed activity into a means, so it trans- India, Mexico, and in the service rendered to
forms the species-life of man into a means of gods, that the product belonged to the gods. But
physical existence. the gods alone were never the lords of labour.
Consciousness, which man has from his And no more was nature. What a contradiction
species, is transformed through alienation it would be if the more man subjugates nature
so that species-life becomes only a means by his labour, and the more the marvels of the
for him. (3) Thus alienated labour turns the gods are rendered superfluous by the marvels of
species-life of man, and also nature as his men- industry, the more he should abstain from his joy
tal species-property, into an alien being and into in producing and his enjoyment of the product
a means for his individual existence. It alienates for love of these powers.
from man his own body, external nature, his The alien being to whom labour and the
mental life and his human life. (4) A direct product of labour belong, to whose service la-
consequence of the alienation of man from the bour is devoted, and to whose enjoyment the
product of his labour, from his life activity and product of labour goes, can only be man him-
from his species-life, is that man is alienated from self. If the product of labour does not belong to
other men. When man confronts himself he also the worker, but confronts him as an alien power,
confronts other men. What is true of man’s rela- this can only be because it belongs to a man other
tionship to his work, to the product of his work than the worker. If his activity is a torment to him
and to himself, is also true of his relationship to it must be a source of enjoyment and pleasure to
other men, to their labour and to the objects of another. Not the gods, nor nature, but only man
their labour. himself can be this alien power over men.
In general, the statement that man is alien- Consider the earlier statement that the rela-
ated from his species-life means that each man is tion of man to himself is first realized, objectified,
alienated from others, and that each of the others through his relation to other men. If he is related
is likewise alienated from human life. to the product of his labour, his objectified la-
Human alienation, and above all the relation bour, as to an alien, hostile, powerful and inde-
of man to himself, is first realized and expressed pendent object, he is related in such a way that
in the relationship between each man and other another alien, hostile, powerful and independent
men. Thus in the relationship of alienated labour man is the lord of this object. If he is related to
every man regards other men according to the his own activity as to unfree activity, then he is
standards and relationships in which he finds related to it as activity in the service, and under
himself placed as a worker. the domination, coercion and yoke, of another
We began with an economic fact, the alien- man. . . .
ation of the worker and his production. We have Thus, through alienated labour the worker
expressed this fact in conceptual terms as alien- creates the relation of another man, who does
ated labour, and in analysing the concept we have not work and is outside the work process, to
merely analysed an economic fact. this labour. The relation of the worker to work
Let us now examine further how this concept also produces the relation of the capitalist (or
of alienated labour must express and reveal it- whatever one likes to call the lord of labour) to
self in reality. If the product of labour is alien work. Private property is, therefore, the product,
to me and confronts me as an alien power, to the necessary result, of alienated labour, of the
 Marx—Classes in Capitalism and Pre-Capitalism • 131

external relation of the worker to nature and to The possessing class and the proletarian class rep-
himself. resent one and the same human self-alienation.
Private property is thus derived from the anal- But the former feels satisfied and affirmed in
ysis of the concept of alienated labour; that is, this self-alienation, experiences the alienation as
alienated man, alienated labour, alienated life, a sign of its own power, and possesses in it the ap-
and estranged man. pearance of a human existence. The latter, how-
We have, of course, derived the concept of ever, feels destroyed in this alienation, seeing in
alienated labour (alienated life) from political it its own impotence and the reality of an in-
economy, from an analysis of the movement of human existence. To use Hegel’s expression, this
private property. But the analysis of this concept class is, within depravity, an indignation against
shows that although private property appears to this depravity, an indignation necessarily aroused
be the basis and cause of alienated labour, it is in this class by the contradiction between its
rather a consequence of the latter, just as the gods human nature and its life-situation, which is a
are fundamentally not the cause but the product blatant, outright and all-embracing denial of
of confusions of human reason. At a later stage, that very nature.
however, there is a reciprocal influence. Within the antagonism as a whole, therefore,
Only in the final stage of the development private property represents the conservative side
of private property is its secret revealed, namely, and the proletariat the destructive side. From the
that it is on one hand the product of alienated former comes action aimed at preserving the
labour, and on the other hand the means by antagonism; from the latter, action aimed at its
which labour is alienated, the realization of this destruction.
alienation. The Holy Family:
The Economic and A Critique of Critical Criticism,
Philosophical Manuscripts, pp. 133–134
pp. 121–131

Karl Marx

Classes in Capitalism and Pre-Capitalism

The history of all hitherto existing society1 is the constant opposition to one another, carried on
history of class struggles. an uninterrupted, now hidden, now open fight,
Freeman and slave, patrician and plebeian, a fight that each time ended, either in a revolu-
lord and serf, guild-master2 and journeyman, tionary re-constitution of society at large, or in
in a word, oppressor and oppressed, stood in the common ruin of the contending classes.

Karl Marx. “The Communist Manifesto,” in Selected Works, Vol. I (Moscow: Progress Publishers, 1964), pp.
108–119. Reprinted by permission. The Poverty of Philosophy (New York: International Publishers, 1963), pp.
172–175. Reprinted by permission of International Publishers. “The Eighteenth Brumaire of Louis Bona-
parte,” in Selected Works, Vol. I (Moscow: Progress Publishers, 1963), pp. 478–479. Reprinted by permission.
Capital, Vol. III (Moscow: Progress Publishers, 1967), pp. 885–886. Reprinted by permission.
132 • Part III: The Structure of Inequality

In the earlier epochs of history, we find al- of manufacture was taken by the giant, Modern
most everywhere a complicated arrangement of Industry, the place of the industrial middle class,
society into various orders, a manifold gradation by industrial millionaires, the leaders of whole
of social rank. In ancient Rome we have patri- industrial armies, the modern bourgeois.
cians, knights, plebeians, slaves; in the Middle Modern industry has established the
Ages, feudal lords, vassals, guild-masters, jour- world-market, for which the discovery of Amer-
neymen, apprentices, serfs; in almost all of these ica paved the way. This market has given an
classes, again, subordinate gradations. immense development to commerce, to naviga-
The modern bourgeois society that has tion, to communication by land. This develop-
sprouted from the ruins of feudal society has not ment has, in its turn, reacted on the extension
done away with class antagonisms. It has but es- of industry; and in proportion as industry,
tablished new classes, new conditions of oppres- commerce, navigation, railways extended, in
sion, new forms of struggle in place of the old the same proportion the bourgeoisie developed,
ones. increased its capital, and pushed into the back-
Our epoch, the epoch of the bourgeoisie, ground every class handed down from the Mid-
possesses, however, this distinctive feature: it dle Ages.
has simplified the class antagonisms. Society as We see, therefore, how the modern bourgeoi-
a whole is more and more splitting up into two sie is itself the product of a long course of devel-
great hostile camps, into two great classes directly opment, of a series of revolutions in the modes
facing each other: Bourgeoisie and Proletariat. of production and of exchange.
From the serfs of the Middle Ages sprang the Each step in the development of the bour-
chartered burghers of the earliest towns. From geoisie was accompanied by a corresponding po-
these burgesses the first elements of the bour- litical advance of that class. An oppressed class
geoisie were developed. under the sway of the feudal nobility, an armed
The discovery of America, the rounding of and self-governing association in the mediaeval
the Cape, opened up fresh ground for the rising commune3; here independent urban republic (as
bourgeoisie. The East-Indian and Chinese mar- in Italy and Germany), there taxable “third es-
kets, the colonisation of America, trade with the tate” of the monarchy (as in France), afterwards,
colonies, the increase in the means of exchange in the period of manufacture proper, serving
and in commodities generally, gave to com- either the semi-feudal or the absolute monar-
merce, to navigation, to industry, an impulse chy as a counterpoise against the nobility, and,
never before known, and thereby, to the revolu- in fact, cornerstone of the great monarchies in
tionary element in the tottering feudal society, a general, the bourgeoisie has at last, since the
rapid development. establishment of Modern Industry and of the
The feudal system of industry, under which world-market, conquered for itself, in the mod-
industrial production was monopolised by closed ern representative State, exclusive political sway.
guilds, now no longer sufficed for the growing The executive of the modern State is but a com-
wants of the new markets. The manufacturing mittee for managing the common affairs of the
system took its place. The guild-masters were whole bourgeoisie.
pushed on one side by the manufacturing mid- The bourgeoisie, historically, has played a
dle class; division of labour between the different most revolutionary part. The bourgeoisie, wher-
corporate guilds vanished in the face of division ever it has got the upper hand, has put an end
of labour in each single workshop. to all feudal, patriarchal, idyllic relations. It has
Meantime the markets kept ever growing, the pitilessly torn asunder the motley feudal ties
demand ever rising. Even manufacture no lon- that bound man to his “natural superiors,” and
ger sufficed. Thereupon, steam and machinery has left remaining no other nexus between man
revolutionised industrial production. The place and man than naked self-interest, than callous
 Marx—Classes in Capitalism and Pre-Capitalism • 133

“cash payment.” It has drowned the most heav- sober senses, his real conditions of life, and his
enly ecstasies of religious fervour, of chivalrous relations with his kind.
enthusiasm, of hilistine sentimentalism, in the The need of a constantly expanding market
icy water of egotistical calculation. It has re- for its products chases the bourgeoisie over the
solved personal worth into exchange value, and whole surface of the globe. It must nestle every-
in place of the numberless indefeasible chartered where, settle everywhere, establish connexions
freedoms, has set up that single, unconscionable everywhere.
freedom—Free Trade. In one word, for exploita- The bourgeoisie has through its exploita-
tion, veiled by religious and political illusions, it tion of the world-market given a cosmopolitan
has substituted naked, shameless, direct, brutal character to production and consumption in
exploitation. every country. To the great chagrin of Reaction-
The bourgeoisie has stripped of its halo every ists, it has drawn from under the feet of indus-
occupation hitherto honoured and looked up to try the national ground on which it stood. All
with reverent awe. It has converted the physi- old-established national industries have been
cian, the lawyer, the priest, the poet, the man of destroyed or are daily being destroyed. They are
science, into its paid wage-labourers. dislodged by new industries, whose introduction
The bourgeoisie has torn away from the fam- becomes a life and death question for all civilised
ily its sentimental veil, and has reduced the fam- nations, by industries that no longer work up
ily relation to a mere money relation. indigenous raw material, but raw material drawn
The bourgeoisie has disclosed how it came from the remotest zones; industries whose prod-
to pass that the brutal display of vigour in the ucts are consumed, not only at home, but in
Middle Ages, which Reactionists so much ad- every quarter of the globe. In place of the old
mire, found its fitting complement in the most wants, satisfied by the productions of the coun-
slothful indolence. It has been the first to show try, we find new wants, requiring for their satis-
what man’s activity can bring about. It has ac- faction the products of distant lands and climes.
complished wonders far surpassing Egyptian In place of the old local and national seclusion
pyramids, Roman aqueducts, and Gothic ca- and self-sufficiency, we have intercourse in every
thedrals; it has conducted expeditions that put direction, universal inter-dependence of nations.
in the shade all former Exoduses of nations and And as in material, so also in intellectual pro-
crusades. duction. The intellectual creations of individual
The bourgeoisie cannot exist without con- nations become common property. National
stantly revolutionising the instruments of pro- one-sidedness and narrow-mindedness become
duction, and thereby the relations of production, more and more impossible, and from the numer-
and with them the whole relations of society. ous national and local literatures, there arises a
Conservation of the old modes of production in world literature.
unaltered form, was, on the contrary, the first The bourgeoisie, by the rapid improvement
condition of existence for all earlier industrial of all instruments of production, by the im-
classes. Constant revolutionising of production, mensely facilitated means of communication,
uninterrupted disturbance of all social con- draws all, even the most barbarian, nations into
ditions, everlasting uncertainty and agitation civilisation. The cheap prices of its commodities
distinguish the bourgeois epoch from all earlier are the heavy artillery with which it batters down
ones. All fixed, fast-frozen relations, with their all Chinese walls, with which it forces the bar-
train of ancient and venerable prejudices and barians’ intensely obstinate hatred of foreigners
opinions, are swept away, all new-formed ones to capitulate. It compels all nations, on pain of
become antiquated before they can ossify. All extinction, to adopt the bourgeois mode of pro-
that is solid melts into air, all that is holy is pro- duction; it compels them to introduce what it
faned, and man is at last compelled to face with calls civilisation into their midst, i.e., to become
134 • Part III: The Structure of Inequality

bourgeois themselves. In one word, it creates a productive forces; they became so many fetters.
world after its own image. They had to be burst asunder; they were burst
The bourgeoisie has subjected the country to asunder.
the rule of the towns. It has created enormous Into their place stepped free competition, ac-
cities, has greatly increased the urban population companied by a social and political constitution
as compared with the rural, and has thus res- adapted to it, and by the economical and politi-
cued a considerable part of the population from cal sway of the bourgeois class.
the idiocy of rural life. Just as it has made the A similar movement is going on before our
country dependent on the towns, so it has made own eyes. Modern bourgeois society with its
barbarian and semibarbarian countries depen- relations of production, of exchange and of
dent on the civilised ones, nations of peasants property, a society that has conjured up such
on nations of bourgeois, the East on the West. gigantic means of production and of exchange,
The bourgeoisie keeps more and more doing is like the sorcerer, who is no longer able to con-
away with the scattered state of the population, trol the powers of the nether world whom he
of the means of production, and of property. It has called up by his spells. For many a decade
has agglomerated population, centralised means past the history of industry and commerce is but
of production, and has concentrated property the history of the revolt of modern productive
in a few hands. The necessary consequence of forces against modern conditions of production,
this was political centralisation. Independent, against the property relations that are the condi-
or but loosely connected provinces, with sepa- tions for the existence of the bourgeoisie and of
rate interests, laws, governments and systems its rule. It is enough to mention the commercial
of taxation, became lumped together into one crises that by their periodical return put on its
nation, with one government, one code of laws, trial, each time more threateningly, the existence
one national class-interest, one frontier and one of the entire bourgeois society. In these crises a
customs-tariff. great part not only of the existing products, but
The bourgeoisie, during its rule of scarce also of the previously created productive forces,
one hundred years, has created more massive are periodically destroyed. In these crises there
and more colossal productive forces than have breaks out an epidemic that, in all earlier epochs,
all preceding generations together. Subjection would have seemed an absurdity—the epidemic
of Nature’s forces to man, machinery, applica- of over-production. Society suddenly finds itself
tion of chemistry to industry and agriculture, put back into a state of momentary barbarism;
steam-navigation, railways, electric telegraphs, it appears as if a famine, a universal war of dev-
clearing of whole continents for cultivation, ca- astation had cut off the supply of every means of
nalisation of rivers, whole populations conjured subsistence; industry and commerce seem to be
out of the ground—what earlier century had destroyed; and why? Because there is too much
even a presentiment that such productive forces civilisation, too much means of subsistence, too
slumbered in the lap of social labour? much industry, too much commerce. The pro-
We see then: the means of production and ductive forces at the disposal of society no longer
of exchange, on whose foundation the bour- tend to further the development of the condi-
geoisie built itself up, were generated in feudal tions of bourgeois property; on the contrary,
society. At a certain stage in the development they have become too powerful for these con-
of these means of production and of exchange, ditions, by which they are fettered, and so soon
the conditions under which feudal society pro- as they overcome these fetters, they bring dis-
duced and exchanged, the feudal organisation of order into the whole of bourgeois society, en-
agriculture and manufacturing industry, in one danger the existence of bourgeois property. The
word, the feudal relations of property became no conditions of bourgeois society are too narrow
longer compatible with the already developed to comprise the wealth created by them. And
 Marx—Classes in Capitalism and Pre-Capitalism • 135

how does the bourgeoisie get over these crises? Modern industry has converted the little
On the one hand by enforced destruction of a workshop of the patriarchal master into the
mass of productive forces; on the other, by the great factory of the industrial capitalist. Masses
conquest of new markets, and by the more thor- of labourers, crowded into the factory, are or-
ough exploitation of the old ones. That is to say, ganised like soldiers. As privates of the industrial
by paving the way for more extensive and more army they are placed under the command of a
destructive crises, and by diminishing the means perfect hierarchy of officers and sergeants. Not
whereby crises are prevented. only are they slaves of the bourgeois class, and
The weapons with which the bourgeoisie of the bourgeois State; they are daily and hourly
felled feudalism to the ground are now turned enslaved by the machine, by the overlooker, and,
against the bourgeoisie itself. above all, by the individual bourgeois manufac-
But not only has the bourgeoisie forged the turer himself. The more openly this despotism
weapons that bring death to itself; it has also proclaims gain to be its end and aim, the more
called into existence the men who are to wield petty, the more hateful and the more embittering
those weapons—the modern working class—the it is.
proletarians. The less the skill and exertion of strength
In proportion as the bourgeoisie, i.e., capital, implied in manual labour, in other words, the
is developed, in the same proportion is the pro- more modern industry becomes developed, the
letariat, the modern working class, developed—a more is the labour of men superseded by that of
class of labourers, who live only so long as they women. Differences of age and sex have no lon-
find work, and who find work only so long as ger any distinctive social validity for the working
their labour increases capital. These labourers, class. All are instruments of labour, more or less
who must sell themselves piecemeal, are a com- expensive to use, according to their age and sex.
modity, like every other article of commerce, and No sooner is the exploitation of the labourer
are consequently exposed to all the vicissitudes of by the manufacturer, so far, at an end, and he
competition, to all the fluctuations of the market. receives his wages in cash, than he is set upon by
Owing to the extensive use of machinery and the other portions of the bourgeoisie, the land-
to division of labour, the work of the proletari- lord, the shopkeeper, the pawnbroker, etc.
ans has lost all individual character, and, conse- The lower strata of the middle class—the
quently, all charm for the workman. He becomes small tradespeople, shopkeepers, and retired
an appendage of the machine, and it is only tradesmen generally, the handicraftsmen and
the most simple, most monotonous, and most peasants—all these sink gradually into the pro-
easily acquired knack, that is required of him. letariat, partly because their diminutive capital
Hence, the cost of production of a workman is does not suffice for the scale on which Modern
restricted, almost entirely, to the means of subsis- Industry is carried on, and is swamped in the
tence that he requires for his maintenance, and competition with the large capitalists, partly
for the propagation of his race. But the price of a because their specialised skill is rendered worth-
commodity, and therefore also of labour, is equal less by new methods of production. Thus the
to its cost of production. In proportion, there- proletariat is recruited from all classes of the
fore, as the repulsiveness of the work increases, population.
the wage decreases. Nay more, in proportion as The proletariat goes through various stages
the use of machinery and division of labour in- of development. With its birth begins its strug-
creases, in the same proportion the burden of gle with the bourgeoisie. At first the contest is
toil also increases, whether by prolongation of carried on by individual labourers, then by the
the working hours, by increase of the work ex- workpeople of a factory, then by the operatives
acted in a given time or by increased speed of the of one trade, in one locality, against the individ-
machinery, etc. ual bourgeois who directly exploits them. They
136 • Part III: The Structure of Inequality

direct their attacks not against the bourgeois beforehand for these occasional revolts. Here
conditions of production, but against the instru- and there the contest breaks out into riots.
ments of production themselves: they destroy Now and then the workers are victorious,
imported wares that compete with their labour, but only for a time. The real fruit of their bat-
they smash to pieces machinery, they set factories tles lies, not in the immediate result, but in the
ablaze, they seek to restore by force the vanished ever-expanding union of the workers. This union
status of the workman of the Middle Ages. is helped on by the improved means of commu-
At this stage the labourers still form an in- nication that are created by modern industry and
coherent mass scattered over the whole country, that place the workers of different localities in
and broken up by their mutual competition. If contact with one another. It was just this con-
anywhere they unite to form more compact bod- tact that was needed to centralise the numerous
ies, this is not yet the consequence of their own local struggles, all of the same character, into one
active union, but of the union of the bourgeoisie, national struggle between classes. But every class
which class, in order to attain its own political struggle is a political struggle. And that union,
ends, is compelled to set the whole proletariat in to attain which the burghers of the Middle Ages,
motion, and is moreover yet, for a time, able to with their miserable highways, required centu-
do so. At this stage, therefore, the proletarians do ries, the modern proletarians, thanks to railways,
not fight their enemies, but the enemies of their achieve in a few years.
enemies, the remnants of absolute monarchy, This organisation of the proletarians into a
the landowners, the non-industrial bourgeois, class, and consequently into a political party, is
the petty bourgeoisie. Thus the whole historical continually being upset again by the competi-
movement is concentrated in the hands of the tion between the workers themselves. But it ever
bourgeoisie; every victory so obtained is a victory rises up again, stronger, firmer, mightier. It com-
for the bourgeoisie. pels legislative recognition of particular interests
But with the development of industry the of the workers, by taking advantage of the di-
proletariat not only increases in number; it visions among the bourgeoisie itself. Thus the
becomes concentrated in greater masses, its ten-hours’ bill in England was carried.
strength grows, and it feels that strength more. Altogether collisions between the classes of
The various interests and conditions of life the old society further, in many ways, the course
within the ranks of the proletariat are more and of development of the proletariat. The bourgeoi-
more equalised, in proportion as machinery sie finds itself involved in a constant battle. At
obliterates all distinctions of labour, and nearly first with the aristocracy; later on, with those
everywhere reduces wages to the same low level. portions of the bourgeoisie itself, whose inter-
The growing competition among the bourgeois, ests have become antagonistic to the progress
and the resulting commercial crises, make the of industry; at all times, with the bourgeoisie of
wages of the workers ever more fluctuating. The foreign countries. In all these battles it sees itself
unceasing improvement of machinery, ever more compelled to appeal to the proletariat, to ask for
rapidly developing, makes their livelihood more its help, and thus, to drag it into the political
and more precarious; the collisions between in- arena. The bourgeoisie itself, therefore, supplies
dividual workmen and individual bourgeois take the proletariat with its own elements of political
more and more the character of collisions be- and general education, in other words, it fur-
tween two classes. Thereupon the workers begin nishes the proletariat with weapons for fighting
to form combinations (Trades’ Unions) against the bourgeoisie.
the bourgeois; they club together in order to Further, as we have already seen, entire sec-
keep up the rate of wages; they found perma- tions of the ruling classes are, by the advance
nent associations in order to make provision of industry, precipitated into the proletariat,
 Marx—Classes in Capitalism and Pre-Capitalism • 137

or are at least threatened in their conditions of to his wife and children has no longer anything
existence. These also supply the proletariat with in common with the bourgeois family-relations;
fresh elements of enlightenment and progress. modern, industrial labour, modern subjection
Finally, in times when the class struggle nears to capital, the same in England as in France,
the decisive hour, the process of dissolution in America as in Germany, has stripped him of
going on within the ruling class, in fact within every trace of national character. Law, morality,
the whole range of old society, assumes such a religion, are to him so many bourgeois preju-
violent, glaring character, that a small section of dices, behind which lurk in ambush just as many
the ruling class cuts itself adrift, and joins the bourgeois interests.
revolutionary class, the class that holds the fu- All the preceding classes that got the upper
ture in its hands. Just as, therefore, at an earlier hand, sought to fortify their already acquired
period, a section of the nobility went over to the status by subjecting society at large to their con-
bourgeoisie, so now a portion of the bourgeoisie ditions of appropriation. The proletarians can-
goes over to the proletariat, and in particular, a not become masters of the productive forces of
portion of the bourgeois ideologists, who have society, except by abolishing their own previous
raised themselves to the level of comprehending mode of appropriation, and thereby also every
theoretically the historical movement as a whole. other previous mode of appropriation. They
Of all the classes that stand face to face with have nothing of their own to secure and to for-
the bourgeoisie today, the proletariat alone is a tify; their mission is to destroy all previous secu-
really revolutionary class. The other classes decay rities for, and insurances of, individual property.
and finally disappear in the face of Modern In- All previous historical movements were
dustry; the proletariat is its special and essential movements of minorities, or in the interests of
product. minorities. The proletarian movement is the
The lower middle class, the small manufac- self-conscious, independent movement of the
turer, the shopkeeper, the artisan, the peasant, all immense majority, in the interests of the im-
these fight against the bourgeoisie, to save from mense majority. The proletariat, the lowest stra-
extinction their existence as fractions of the mid- tum of our present society, cannot stir, cannot
dle class. They are therefore not revolutionary, raise itself up, without the whole superincum-
but conservative. Nay more, they are reaction- bent strata of official society being sprung into
ary, for they try to roll back the wheel of history. the air.
If by chance they are revolutionary, they are so Though not in substance, yet in form, the
only in view of their impending transfer into the struggle of the proletariat with the bourgeoisie
proletariat; they thus defend not their present, is at first a national struggle. The proletariat of
but their future interests, they desert their own each country must, of course, first of all settle
standpoint to place themselves at that of the matters with its own bourgeoisie.
proletariat. In depicting the most general phases of
The “dangerous class,” the social scum, that the development of the proletariat, we traced
passively rotting mass thrown off by the lowest the more or less veiled civil war, raging within
layers of old society, may, here and there, be existing society, up to the point where that war
swept into the movement by a proletarian rev- breaks out into open revolution, and where the
olution; its conditions of life, however, prepare violent overthrow of the bourgeoisie lays the
it far more for the part of a bribed tool of reac- foundation for the sway of the proletariat.
tionary intrigue. Hitherto, every form of society has been
In the conditions of the proletariat, those of based, as we have already seen, on the antago-
old society at large are already virtually swamped. nism of oppressing and oppressed classes. But in
The proletarian is without property; his relation order to oppress a class, certain conditions must
138 • Part III: The Structure of Inequality

conquered from their feudal lords and masters local

be assured to it under which it can, at least, con-
self-government and political rights as the “Third
tinue its slavish existence. The serf, in the period Estate.” Generally speaking, for the economical de-
of serfdom, raised himself to membership in the velopment of the bourgeoisie, England is here taken
commune, just as the petty bourgeois, under the as the typical country; for its political development,
yoke of feudal absolutism, managed to develop France. [Note by Engels to the English edition of 1888.]
into a bourgeois. The modern labourer, on the This was the name given their urban commu-
contrary, instead of rising with the progress of nities by the townsmen of Italy and France, after
industry, sinks deeper and deeper below the they had purchased or wrested their initial rights of
self-government from their feudal lords. [Note by En-
conditions of existence of his own class. He be-
gels to the German edition of 1890.]
comes a pauper, and pauperism develops more
rapidly than population and wealth. And here The Communist Manifesto,
pp. 108–119
it becomes evident, that the bourgeoisie is unfit
any longer to be the ruling class in society, and to
impose its conditions of existence upon society The first attempts of workers to associate among
as an overriding law. It is unfit to rule because themselves always take place in the form of
it is incompetent to assure an existence to its combinations.
slave within his slavery, because it cannot help Large-scale industry concentrates in one
letting him sink into such a state, that it has to place a crowd of people unknown to one an-
feed him, instead of being fed by him. Society other. Competition divides their interests. But
can no longer live under this bourgeoisie, in the maintenance of wages, this common inter-
other words, its existence is no longer compati- est which they have against their boss, unites
ble with society. them in a common thought of resistance—com-
The essential condition for the existence, and bination. Thus combination always has a dou-
for the sway of the bourgeois class, is the forma- ble aim, that of stopping competition among
tion and augmentation of capital; the condition the workers, so that they can carry on general
for capital is wage-labour. Wage-labour rests ex- competition with the capitalist. If the first aim
clusively on competition between the labourers. of resistance was merely the maintenance of
The advance of industry, whose involuntary pro- wages, combinations, at first isolated, constitute
moter is the bourgeoisie, replaces the isolation of themselves into groups as the capitalists in their
the labourers, due to competition, by their rev- turn unite for the purpose of repression, and in
olutionary combination, due to association. The face of always united capital, the maintenance of
development of Modern Industry, therefore, cuts the association becomes more necessary to them
from under its feet the very foundation on which than that of wages. This is so true that English
the bourgeoisie produces and appropriates prod- economists are amazed to see the workers sacri-
ucts. What the bourgeoisie, therefore, produces, fice a good part of their wages in favour of asso-
above all, is its own grave-diggers. Its fall and the ciations, which, in the eyes of these economists,
victory of the proletariat are equally inevitable. are established solely in favour of wages. In this
struggle—a veritable civil war—all the elements
NOTES necessary for a coming battle unite and develop.
1. That is, all written history. In 1847, the Once it has reached this point, association takes
pre-history of society, the social organisation existing
on a political character.
previous to recorded history, was all but unknown.
Economic conditions had first transformed
[Note by Engels to the English edition of 1888.]
2. Guild-master, that is, a full member of a guild, the mass of the people of the country into work-
a master within, not a head of a guild. [Note by Engels ers. The combination of capital has created for
to the English edition of 1888.] this mass a common situation, common inter-
3. “Commune” was the name taken, in ests. This mass is thus already a class as against
France, by the nascent towns even before they had capital, but not yet for itself. In the struggle, of
 Marx—Classes in Capitalism and Pre-Capitalism • 139

which we have noted only a few phases, this mass The working class, in the course of its de-
becomes united, and constitutes itself as a class velopment, will substitute for the old civil so-
for itself. The interests it defends become class ciety an association which will exclude classes
interests. But the struggle of class against class is and their antagonism, and there will be no more
a political struggle. political power properly so-called, since political
In the bourgeoisie we have two phases to power is precisely the official expression of antag-
distinguish: that in which it constituted itself onism in civil society.
as a class under the regime of feudalism and Meanwhile the antagonism between the pro-
absolute monarchy, and that in which, already letariat and the bourgeoisie is a struggle of class
constituted as a class, it overthrew feudalism against class, a struggle which carried to its high-
and monarchy to make society into a bourgeois est expression is a total revolution. Indeed, is it
society. The first of these phases was the lon- at all surprising that a society founded on the
ger and necessitated the greater efforts. This opposition of classes should culminate in brutal
too began by partial combinations against the contradiction, the shock of body against body, as
feudal lords. its final dénouement?
Much research has been carried out to trace Do not say that social movement excludes
the different historical phases that the bourgeoi- political movement. There is never a political
sie has passed through, from the commune up to movement which is not at the same time social.
its constitution as a class. It is only in an order of things in which there
But when it is a question of making a precise are no more classes and class antagonisms that so-
study of strikes, combinations and other forms cial evolutions will cease to be political revolutions.
in which the proletarians carry out before our Till then, on the eve of every general reshuffling
eyes their organization as a class, some are seized of society, the last word of social science will al-
with real fear and others display a transcendental ways be:
disdain. “Le combat ou la mort; la lutte sanguinaire ou
An oppressed class is the vital condition le néant. C’est ainsi que la question est invincible-
for every society founded on the antagonism ment posée.” 2
of classes. The emancipation of the oppressed
class thus implies necessarily the creation of a NOTES
new society. For the oppressed class to be able 1. Estates here in the historical sense of the estates
of feudalism, estates with definite and limited priv-
to emancipate itself it is necessary that the pro-
ileges. The revolution of the bourgeoisie abolished
ductive powers already acquired and the existing
the estates and their privileges. Bourgeois society
social relations should no longer be capable of knows only classes. It was, therefore, absolutely in
existing side by side. Of all the instruments of contradiction with history to describe the proletariat
production, the greatest productive power is the as the “fourth estate.” [Note by F. Engels to the Ger-
revolutionary class itself. The organization of man edition, 1885.]
revolutionary elements as a class supposes the 2. “Combat or death; bloody struggle or extinc-
existence of all the productive forces which could tion. It is thus that the question is inexorably put.”
be engendered in the bosom of the old society. George Sand, Jean Ziska.
Does this mean that after the fall of the old The Poverty of Philosophy, pp. 172–175
society there will be a new class domination cul-
minating in a new political power? No. The small-holding peasants form a vast mass, the
The condition for the emancipation of the members of which live in similar conditions but
working class is the abolition of every class, just without entering into manifold relations with
as the condition for the liberation of the third one another. Their mode of production iso-
estate, of the bourgeois order, was the abolition lates them from one another instead of bringing
of all estates1 and all orders. them into mutual intercourse. The isolation is
140 • Part III: The Structure of Inequality

increased by France’s bad means of communi- of income are wages, profit and ground-rent,
cation and by the poverty of the peasants. Their in other words, wage-labourers, capitalists and
field of production, the small holding, admits of landowners, constitute then three big classes of
no division of labour in its cultivation, no ap- modern society based upon the capitalist mode
plication of science and, therefore, no diversity of production.
of development, no variety of talent, no wealth In England, modern society is indisputably
of social relationships. Each individual peasant most highly and classically developed in eco-
family is almost self-sufficient; it itself directly nomic structure. Nevertheless, even here the
produces the major part of its consumption and stratification of classes does not appear in its
thus acquires its means of life more through ex- pure form. Middle and intermediate strata even
change with nature than in intercourse with so- here obliterate lines of demarcation everywhere
ciety. A small holding, a peasant and his family; (although incomparably less in rural districts
alongside them another small holding, another than in the cities). However, this is immaterial
peasant and another family. A few score of these for our analysis. We have seen that the continual
make up a village, and a few score of villages tendency and law of development of the capi-
make up a Department. In this way, the great talist mode of production is more and more to
mass of the French nation is formed by simple divorce the means of production from labour,
addition of homologous magnitudes, much as and more and more to concentrate the scattered
potatoes in a sack form a sack of potatoes. In so means of production into large groups, thereby
far as millions of families live under economic transforming labour into wage-labour and the
conditions of existence that separate their mode means of production into capital. And to this
of life, their interests and their culture from those tendency, on the other hand, corresponds the
of the other classes, and put them in hostile op- independent separation of landed property from
position to the latter, they form a class. In so far capital and labour, or the transformation of all
as there is merely a local interconnection among landed property into the form of landed prop-
these small-holding peasants, and the identity of erty corresponding to the capitalist mode of
their interests begets no community, no national production.
bond and no political organisation among them, The first question to be answered is this:
they do not form a class. They are consequently What constitutes a class?—and the reply to this
incapable of enforcing their class interests in follows naturally from the reply to another ques-
their own name, whether through a parliament tion, namely: What makes wage-labourers, cap-
or through a convention. They cannot represent italists and landlords constitute the three great
themselves, they must be represented. Their social classes?
representative must at the same time appear as At first glance—the identity of revenues
their master, as an authority over them, as an un- and sources of revenue. There are three great
limited governmental power that protects them social groups whose members, the individu-
against the other classes and sends them rain and als forming them, live on wages, profit and
sunshine from above. The political influence of ground-rent respectively, on the realisation
the small-holding peasants, therefore, finds its of their labour-power, their capital, and their
final expression in the executive power subordi- landed property.
nating society to itself. However, from this standpoint, physicians
The Eighteenth Brumaire of Louis Bonaparte, and officials, e.g., would also constitute two
pp. 478–479 classes, for they belong to two distinct social
groups, the members of each of these groups
The owners merely of labour-power, owners of receiving their revenue from one and the same
capital, and landowners, whose respective sources source. The same would also be true of the
 Marx—Ideology and Class • 141

infinite fragmentation of interest and rank into owners of forests, mine owners and owners of
which the division of social labour splits labour- fisheries.
ers as well as capitalists and landlords—the lat- Capital, Vol. III, pp. 885–886
ter, e.g., into owners of vineyards, farm owners,

Karl Marx

Ideology and Class

The ideas of the ruling class are in every epoch to be the dominant idea and is expressed as an
the ruling ideas, i.e. the class which is the ruling “eternal law.”
material force of society, is at the same time its The division of labour manifests itself in the
ruling intellectual force. The class which has the ruling class as the division of mental and ma-
means of material production at its disposal, has terial labour, so that inside this class one part
control at the same time over the means of men- appears as the thinkers of the class (its active,
tal production, so that thereby, generally speak- conceptive ideologists, who make the perfecting
ing, the ideas of those who lack the means of of the illusion of the class about itself their chief
mental production are subject to it. The ruling source of livelihood), while the others’ attitude
ideas are nothing more than the ideal expres- to these ideas and illusions is more passive and
sion of the dominant material relationships, the receptive, because they are in reality the active
dominant material relationships grasped as ideas; members of this class and have less time to make
hence of the relationships which make the one up illusions and ideas about themselves. Within
class the ruling one, therefore, the ideas of its this class this cleavage can even develop into a
dominance. The individuals composing the rul- certain opposition and hostility between the two
ing class possess among other things conscious- parts, which, however, in the case of a practical
ness, and therefore think. Insofar, therefore, as collision, in which the class itself is endangered,
they rule as a class and determine the extent automatically comes to nothing, in which case
and compass of an epoch, it is self-evident that there also vanishes the semblance that the rul-
they do this in its whole range, hence among ing ideas were not the ideas of the ruling class
other things rule also as thinkers, as producers and had a power distinct from the power of this
of ideas, and regulate the production and distri- class. The existence of revolutionary ideas in a
bution of the ideas of their age: thus their ideas particular period presupposes the existence of a
are the ruling ideas of the epoch. For instance, revolutionary class.
in an age and in a country where royal power, If now in considering the course of history
aristocracy, and bourgeoisie are contending for we detach the ideas of the ruling class from the
mastery and where, therefore, mastery is shared, ruling class itself and attribute to them an in-
the doctrine of the separation of powers proves dependent existence, if we confine ourselves to

Karl Marx. “Ideology and Class,” in The German Ideology (New York: International Publishers, 1970), pp.
64–66, edited by C. J. Arthur. Reprinted by permission of International Publishers.
142 • Part III: The Structure of Inequality

saying that these or those ideas were dominant at therefore, benefits also many individuals of the
a given time, without bothering ourselves about other classes which are not winning a dominant
the conditions of production and the producers position, but only insofar as it now puts these
of these ideas, if we thus ignore the individuals individuals in a position to raise themselves into
and world conditions which are the source of the ruling class. When the French bourgeoisie
the ideas, we can say, for instance, that during overthrew the power of the aristocracy, it thereby
the time that the aristocracy was dominant, the made it possible for many proletarians to raise
concepts honour, loyalty, etc. were dominant, themselves above the proletariat, but only inso-
during the dominance of the bourgeoisie the far as they become bourgeois. Every new class,
concepts freedom, equality, etc. The ruling class therefore, achieves its hegemony only on a
itself on the whole imagines this to be so. This broader basis than that of the class ruling previ-
conception of history, which is common to all ously, whereas the opposition of the non-ruling
historians, particularly since the eighteenth cen- class against the new ruling class later develops
tury, will necessarily come up against the phe- all the more sharply and profoundly. Both these
nomenon that increasingly abstract ideas hold things determine the fact that the struggle to be
sway, i.e. ideas which increasingly take on the waged against this new ruling class, in its turn,
form of universality. For each new class which aims at a more decided and radical negation of
puts itself in the place of one ruling before it, the previous conditions of society than could all
is compelled, merely in order to carry through previous classes which sought to rule.
its aim, to represent its interest as the common This whole semblance, that the rule of a cer-
interest of all the members of society, that is, tain class is only the rule of certain ideas, comes
expressed in ideal form: it has to give its ideas to a natural end, of course, as soon as class rule
the form of universality, and represent them as in general ceases to be the form in which society
the only rational, universally valid ones. The is organised, that is to say, as soon as it is no lon-
class making a revolution appears from the very ger necessary to represent a particular interest as
start, if only because it is opposed to a class, not general or the “general interest” as ruling.
as a class but as the representative of the whole
of society; it appears as the whole mass of soci- NOTES
ety confronting the one ruling class.1 It can do 1. Universality corresponds to (1) the class ver-
this because, to start with, its interest really is sus the estate, (2) the competition, worldwide inter-
course, etc., (3) the great numerical strength of the
more connected with the common interest of
ruling class, (4) the illusion of the common interests
all other non-ruling classes, because under the (in the beginning this illusion is true), (5) the de-
pressure of hitherto existing conditions its inter- lusion of the ideologists and the division of labour.
est has not yet been able to develop as the par- [Marginal note by Marx.]
ticular interest of a particular class. Its victory,
 Dahrendorf—Class and Class Conlict in Industrial Society • 143

15 • Ralf Dahrendorf

Class and Class Conlict in Industrial Society

One of the main questions which the pres- recognition and expanded into all branches of
ent investigation is supposed to answer is: Do economic activity. Today, more than two-thirds
classes and class conflicts belong to that group of all companies in advanced industrial societies
of phenomena by which only the capitalist type are joint-stock companies, and their property
of industrial society is characterized, or is their exceeds four-fifths of the total property in eco-
existence a consequence of industrial production nomic enterprises. The enterprise owned and
itself, and are they therefore a lasting feature of run by an individual, or even a family, has long
industrial societies? This question will accom- ceased to be the dominant pattern of economic
pany us throughout the following analysis of organization. . . .
changes in the structure of industrial societies According to the radical view, joint-stock
since Marx. companies involve a complete break with ear-
lier capitalist traditions. By separating what has
come to be called ownership and control, they
Ownership and Control, or the give rise to a new group of managers who are ut-
Decomposition of Capital terly different from their capitalist predecessors.
Marx was right in seeking the root of social Thus for Marx, the joint-stock company involves
change in capitalist society in the sphere of a complete alienation of capital “from the real
industrial production, but the direction these producers, and its opposition as alien property to
changes took turned out to be directly contrary all individuals really participating in production,
to Marx’s expectations. With respect to capi- from the manager down to the last day-laborer”
tal, he had, in his later years, at least a vision (1953, Vol. III, p. 478). In other words, by sep-
of what was going to happen, as his brief and arating ownership and control, the joint-stock
somewhat puzzled analysis of joint-stock com- company reduces the distance between manager
panies shows. Joint-stock companies were le- and worker while at the same time removing the
gally recognized in Germany, England, France, owners altogether from the sphere of production
and the United States in the second half of the and thereby isolating their function as exploiters
nineteenth century. Laws often indicate the con- of others. It is merely a step from this kind of
clusion of social developments, and indeed early analysis to the thesis that, as Renner has it, the
forms of joint-stock companies can be traced “capitalists without function” yield to the “func-
back at least to the commercial companies and tionaries without capital,” and that this new rul-
trade societies of the seventeenth century. But it ing group of industry bears little resemblance to
was in the nineteenth and early twentieth centu- the old “full capitalists” (1953, pp. 182, 198).
ries that this type of enterprise first gained wide Burnham, Geiger, Sering, and others followed

Ralf Dahrendorf. Class and Class Conflict in Industrial Society, pp. 41–43, 47–48, 50–51, 64–67, 165–167,
170–173. Copyright © 1959 by the Board of Trustees of the Leland Stanford Junior University. Renewed
1987. All rights reserved. Used with the permission of Stanford University Press, www.sup.org.
144 • Part III: The Structure of Inequality

Marx (and Renner) in this radical interpretation the decomposition of capital involves a change
of the social effects of joint-stock companies. in the patterns of conflict. One might question
The conservative view, on the other hand, whether this new conflict, in which labor is no
holds that the consequences of the apparent longer opposed to a homogeneous capitalist
separation of ownership and control have been class, can still be described as a class conflict at
vastly overrated. It is argued that in fact owners all. In any case, it is different from the division
and controllers, i.e., stockholders and managers, of the whole society into two great and homoge-
are a fairly homogeneous group. There are often neous hostile camps with which Marx was con-
direct connections between them, and where cerned. While I would follow the radical view
this is not the case, their outlook is sufficiently of the separation of ownership and control in
similar to justify insisting on the old assumption industry to this point, there is one thing to be
of a homogeneous class of capitalists opposed to said in favor of the conservative view. Changes in
an equally homogeneous class of laborers. This the composition of conflict groups, of the issues,
view is not often heard in the West nowadays, and of patterns of conflict do not imply the ab-
although traces of it are evident in the work of olition of conflict or even of the specific conflict
C. Wright Mills (1954, 1956). It may be added between management and labor in industry. De-
that this conservative view is clearly contrary to spite the effects of the decomposition of capital
Marx’s own analysis. . . . on class structure, we have no reason to believe
There is little reason to follow Marx and de- that antagonisms and clashes of interest have
scribe the condition of separation of ownership now been banned from industrial enterprises.
and control as a transitional form of historical
development. It is no more transitional than any
other stage of history, and it has already proven Skill and Stratification, or the
quite a vital pattern of social and economic Decomposition of Labor
structure. But I think that we can follow Marx While Marx had at least a premonition of things
in his radical interpretation of this phenome- to come with respect to capital, he remained un-
non. The separation of ownership and control aware of developments affecting the unity and
has replaced one group by two whose positions, homogeneity of labor. Yet in this respect, too,
roles, and outlooks are far from identical. In the sphere of production which loomed so large
taking this view, one does of course agree with in Marx’s analyses became the starting point of
Marx against himself. For it follows from this changes that clearly refute his predictions. The
that the homogeneous capitalist class predicted working class of today, far from being a homo-
by Marx has in fact not developed. Capital—and geneous group of equally unskilled and impover-
thereby capitalism—has dissolved and given way ished people, is in fact a stratum differentiated by
in the economic sphere, to a plurality of partly numerous subtle and not-so-subtle distinctions.
agreed, partly competing, and partly simply dif- Here, too, history has dissolved one position, or
ferent groups. The effect of this development on role, and has substituted for it a plurality of roles
class conflict is threefold: first, the replacement that are endowed with diverging and often con-
of capitalists by managers involves a change in flicting expectations. . . .
the composition of the groups participating in Analysis of industrial conditions suggests
conflict; second, and as a consequence of this quite clearly that within the labor force of ad-
change in recruitment and composition, there vanced industry we have to distinguish at least
is a change in the nature of the issues that cause three skill groups: a growing stratum of highly
conflicts, for the interests of the functionaries skilled workmen who increasingly merge with
without capital differ from those of full-blown both engineers and white-collar employees, a
capitalists, and so therefore do the interests of relatively stable stratum of semiskilled workers
labor vis-à-vis their new opponents; and third, with a high degree of diffuse as well as specific
 Dahrendorf—Class and Class Conlict in Industrial Society • 145

industrial experience, and a dwindling stratum impoverished slave of industry who is indistin-
of totally unskilled laborers who are character- guishable from his peers in terms of his work, his
istically either newcomers to industry (begin- skill, his wage, and his prestige, has left the scene.
ners, former agricultural laborers, immigrants) What is more, it appears that by now he has been
or semi-unemployables. It appears, furthermore, followed by his less depraved, but equally alien-
that these three groups differ not only in their ated successor, the worker. In modern industry,
level of skill, but also in other attributes and “the worker” has become precisely the kind of
determinants of social status. The semiskilled abstraction which Marx quite justly resented so
almost invariably earn a higher wage than the much. In his place, we find a plurality of status
unskilled, whereas the skilled are often salaried and skill groups whose interests often diverge.
and thereby participate in white-collar status. Demands of the skilled for security may injure
The hierarchy of skill corresponds exactly to the semiskilled; wage claims of the semiskilled
the hierarchy of responsibility and delegated au- may raise objections by the skilled; and any in-
thority within the working class. From numer- terest on the part of the unskilled is bound to set
ous studies it would seem beyond doubt that it their more highly skilled fellow workmen worry-
also correlates with the hierarchy of prestige, at ing about differentials.
the top of which we find the skilled man whose Again, as in the case of capital, it does not
prolonged training, salary, and security convey follow from the decomposition of labor that
special status, and at the bottom of which stands there is no bond left that unites most workers—
the unskilled man who is, according to a recent at least for specific goals; nor does it follow that
German investigation into workers’ opinions, industrial conflict has lost its edge. But here,
merely “working” without having an “occupa- too, a change of the issues and, above all, of
tion” proper (see Kluth, 1955, p. 67). Here as the patterns of conflict is indicated. As with the
elsewhere Marx was evidently mistaken. “Every- capitalist class, it has become doubtful whether
where, the working class differentiates itself more speaking of the working class still makes much
and more, on the one hand into occupational sense. Probably Marx would have agreed that
groups, on the other hand into three large cate- class “is a force that unites into groups people
gories with different, if not contradictory, inter- who differ from one another, by overriding the
ests: the skilled craftsmen, the unskilled laborers, differences between them” (Marshall, 1950, p.
and the semiskilled specialist workers” (Philip, 114), but he certainly did not expect the dif-
1955, p. 2). ferences to be so great, and the uniting force so
In trying to assess the consequences of this precarious as it has turned out to be in the case
development, it is well to remember that, for both of capital and of labor. . . .
Marx, the increasing uniformity of the working
class was an indispensable condition of that in-
tensification of the class struggle which was to The Institutionalization of Class
lead, eventually, to its climax in a revolution. Conflict
The underlying argument of what for Marx be- A historian might argue that all the tendencies
came a prediction appears quite plausible. For of change here described as changes in the struc-
there to be a revolution, the conflicts within a ture of industrial societies since Marx had in fact
society have to become extremely intense. For begun before and in some cases long before Marx
conflicts to be intense, one would indeed expect died in 1883. . . . There is, however, one line of
its participants to be highly unified and homo- social development in industrial societies which
geneous groups. But neither capital nor labor has both originated and spread since about the
have developed along these lines. Capital has dis- time of Marx’s death, and which is directly rele-
solved into at least two, in many ways distinct, vant to our problem. Geiger, who has described
elements, and so has labor. The proletarian, the this change as the “institutionalization of class
146 • Part III: The Structure of Inequality

conflict,” says: “The tension between capital and Nobody can, of course, ever be sure that a
labor is recognized as a principle of the structure given pattern of conflict regulation will always
of the labor market and has become a legal in- prove successful. There are still strikes, and for
stitution of society. . . . The methods, weapons, all we know they will continue to occur. But it
and techniques of the class struggle are recog- has proved possible for industrial society to get
nized—and are thereby brought under control. along with the clashes of interest arising from
The struggle evolves according to certain rules of its industrial and political structure—and it has
the game. Thereby the class struggle has lost its proved possible for interest groups to get along
worst sting, it is converted into a legitimate ten- with industrial society. Instead of a battlefield,
sion between power factors which balance each the scene of group conflict has become a kind
other. Capital and labor struggle with each other, of market in which relatively autonomous forces
conclude compromises, negotiate solutions, and contend according to certain rules of the game,
thereby determine wage levels, hours of work, by virtue of which nobody is a permanent win-
and other conditions of work” (1949, p. 184). ner or loser. This course of development must
Marx displayed a certain sociological naïveté naturally be bitter for the orthodox and the dog-
when he expressed his belief that capitalist so- matic, but theirs is the kind of bitterness which
ciety would be entirely unable to cope with the makes liberal minds rejoice. . . .
class conflict generated by its structure. In fact,
every society is capable of coping with whatever
new phenomena arise in it, if only by the sim- Power and Authority
ple yet effective inertia which can be described, a One of the central theses of this study consists in
little pretentiously, as the process of institution- the assumption that the differential distribution
alization. In the case of class conflict, institution- of authority invariably becomes the determining
alization assumed a number of successive and factor of systematic social conflicts of a type that
complementary forms. It began with the painful is germane to class conflicts in the traditional
process of recognition of the contending parties (Marxian) sense of this term. The structural ori-
as legitimate interest groups. Within industry, gin of such group conflicts must be sought in the
a “secondary system of industrial citizenship” arrangement of social roles endowed with expec-
(Marshall, 1950, p. 68) enabled both workers tations of domination or subjection. Wherever
and entrepreneurs to associate and defend their there are such roles, group conflicts of the type
interests collectively. Outside industry, the pri- in question are to be expected. Differentiation of
mary system of political citizenship had the same groups engaged in such conflicts follows the lines
effect. And while, in the stage of organization, of differentiation of roles that are relevant from
conflict may develop a greater visible intensity, the point of view of the exercise of authority.
organization has at least two side effects which Identification of variously equipped authority
operate in the opposite direction. Organization roles is the first task of conflict analysis;1 concep-
presupposes the legitimacy of conflict groups, tually and empirically all further steps of analysis
and it thereby removes the permanent and in- follow from the investigation of distributions of
calculable threat of guerrilla warfare. At the same power and authority.
time, it makes systematic regulations of conflicts “Unfortunately, the concept of power is not
possible. Organization is institutionalization, a settled one in the social sciences, either in po-
and whereas its manifest function is usually an litical science or in sociology” (Parsons, 1957,
increasingly articulate and outspoken defense p. 139). Max Weber (1947), Pareto (1955),
of interests, it invariably has the latent function Mosca (1950), later Russell (1938), Bendix
also of inaugurating routines of conflict which (1952), Lasswell (1936), and others have ex-
contribute to reducing the violence of clashes of plored some of the dimensions of this category;
interest. . . . they have not, however, reached such a degree
 Dahrendorf—Class and Class Conlict in Industrial Society • 147

of consensus as would enable us to employ the element. (3) Such expectations attach to rela-
categories of power and authority without at tively permanent social positions rather than
least brief conceptual preliminaries. So far as the to the character of individuals; they are in this
terms “power” and “authority” and their distinc- sense legitimate. (4) By virtue of this fact, they
tion are concerned, I shall follow in this study always involve specification of the persons sub-
the useful and well-considered definitions of ject to control and of the spheres within which
Max Weber. For Weber, power is the “probabil- control is permissible. Authority, as distinct from
ity that one actor within a social relationship will power, is never a relation of generalized control
be in a position to carry out his own will despite over others. (5) Authority being a legitimate re-
resistance, regardless of the basis on which this lation, noncompliance with authoritative com-
probability rests”; whereas authority (Herrschaft) mands can be sanctioned; it is indeed one of the
is the “probability that a command with a given functions of the legal system (and of course of
specific content will be obeyed by a given group quasi-legal customs and norms) to support the
of persons” (1947, p. 28). The important differ- effective exercise of legitimate authority.
ence between power and authority consists in the Alongside the term “authority,” we shall em-
fact that whereas power is essentially tied to the ploy in this study the terms “domination” and
personality of individuals, authority is always as- “subjection.” These will be used synonymously
sociated with social positions or roles. The dem- with the rather clumsy expressions “endowed
agogue has power over the masses to whom he with authority” or “participating in the exercise
speaks or whose actions he controls; but the con- of authority” (domination), and “deprived of
trol of the officer over his men, the manager over authority” or “excluded from the exercise of au-
his workers, the civil servant over his clientele thority” (subjection).
is authority, because it exists as an expectation It seems desirable for purposes of conflict
independent of the specific person occupying the analysis to specify the relevant unit of social or-
position of officer, manager, civil servant. It is ganization in analogy to the concept of social
only another way of putting this difference if we system in the analysis of integration. To speak of
say—as does Max Weber—that while power is specification here is perhaps misleading. “Social
merely a factual relation, authority is a legitimate system” is a very general concept applicable to
relation of domination and subjection. In this all types of organization; and we shall want to
sense, authority can be described as legitimate employ an equally general concept which differs
power. from that of social system by emphasizing a dif-
In the present study we are concerned exclu- ferent aspect of the same organizations. It seems
sively with relations of authority, for these alone to me that Max Weber’s category “imperatively
are part of social structure and therefore per- coordinated association” (Herrschaftsverband)
mit the systematic derivation of group conflicts serves this purpose despite its clumsiness. . . .
from the organization of total societies and asso- Empirically it is not always easy to identify
ciations within them. The significance of such the border line between domination and sub-
group conflicts rests with the fact that they are jection. Authority has not remained unaffected
not the product of structurally fortuitous rela- by the modern process of division of labor. But
tions of power but come forth wherever author- even here, groups or aggregates can be identi-
ity is exercised—and that means in all societies fied which do not participate in the exercise of
under all historical conditions. (1) Authority re- authority other than by complying with given
lations are always relations of super- and subor- commands or prohibitions. Contrary to all cri-
dination. (2) Where there are authority relations, teria of social stratification, authority does not
the superordinate element is socially expected to permit the construction of a scale. So-called hier-
control, by orders and commands, warnings and archies of authority (as displayed, for example, in
prohibitions, the behavior of the subordinate organization charts) are in fact hierarchies of the
148 • Part III: The Structure of Inequality

“plus-side” of authority, i.e., of the differentia- picture of a plurality of competing dominant

tion of domination; but there is, in every associa- (and, conversely, subjected) aggregates. This,
tion, also a “minus-side” consisting of those who again, is a problem for the analysis of specific
are subjected to authority rather than participate historical societies and must not be confounded
in its exercise. with the clearer lines of differentiation within
In two respects this analysis has to be spec- any one association. Within the latter, the distri-
ified, if not supplemented. First, for the indi- bution of authority always sums up to zero, i.e.,
vidual incumbent of roles, domination in one there always is a division involving domination
association does not necessarily involve dom- and subjection.
ination in all others to which he belongs, and I need hardly emphasize that from the point
subjection, conversely, in one association does of view of “settling” the concepts of power and
not mean subjection in all. The dichotomy of authority, the preceding discussion has raised
positions of authority holds for specific associa- more problems than it has solved. I believe, how-
tions only. In a democratic state, there are both ever, that for the purposes of this study, and of
mere voters and incumbents of positions of au- a sociological theory of conflict, little needs to
thority such as cabinet ministers, representatives, be added to what has been stated here. In order
and higher civil servants. But this does not mean somewhat to substantiate this perhaps rather
that the “mere voter” cannot be incumbent of a bold assertion, it seems useful to recapitulate
position of authority in a different context, say, briefly the heuristic purpose and logical status of
in an industrial enterprise; conversely, a cabinet the considerations of this section.
minister may be, in his church, a mere member, I have introduced, as a structural determinant
i.e., subject to the authority of others. Although of conflict groups, the category of authority as
empirically a certain correlation of the authority exercised in imperatively coordinated associa-
positions of individuals in different associations tions. While agreeing with Marx that source and
seems likely, it is by no means general and is level of income—even socioeconomic status—
in any case a matter of specific empirical con- cannot usefully be conceived as determinants
ditions. It is at least possible, if not probable, of conflict groups, I have added to this list of
that if individuals in a given society are ranked erroneous approaches Marx’s own in terms of
according to the sum total of their authority po- property in the means of production. Authority
sitions in all associations, the resulting pattern is both a more general and a more significant
will not be a dichotomy but rather like scales social relation. The former has been shown in
of stratification according to income or prestige. our critique of Marx; the latter will have to be
For this reason it is necessary to emphasize that demonstrated elsewhere (see Dahrendorf 1959).
in the sociological analysis of group conflict the The concept of authority is used, in this context,
unit of analysis is always a specific association in a specific sense. It is differentiated from power
and the dichotomy of positions within it. by what may roughly be referred to as the ele-
As with respect to the set of roles associated ment of legitimacy; and it has to be understood
with an individual, total societies, also, do not throughout in the restricted sense of authority
usually present an unambiguously dichotomic as distributed and exercised in imperatively co-
authority structure. There are a large number ordinated associations. While its “disruptive”
of imperatively coordinated associations in any or conflict-generating consequences are not the
given society. Within every one of them we can only aspect of authority, they are the one rele-
distinguish the aggregates of those who domi- vant in terms of the coercion model of society.
nate and those who are subjected. But since Within the frame of reference of this model, (1)
domination in industry does not necessarily in- the distribution of authority in associations is
volve domination in the state, or a church, or the ultimate “cause” of the formation of conflict
other associations, total societies can present the groups, and (2) being dichotomous, it is, in any
 Wright—A General Framework for the Analysis of Class Structure • 149

Harold Lasswell. 1936. Politics—Who Gets What,

given association, the cause of the formation of
When, and How? New York.
two, and only two, conflict groups. T.H. Marshall. 1950. Citizenship and Social Class.
Cambridge.
NOTES Karl Marx. 1953. Das Kapital. Vols. I, III. New ed.
1. To facilitate communication, I shall employ Berlin.
in this study a number of abbreviations. These must C.W. Mills. 1954. The New Men of Power. New York.
not however be misunderstood. Thus, “conflict anal- ———. 1956. The Power Elite. New York.
ysis” in this context stands for “analysis of group con- Gaetano Mosca. 1950. Die herrschende Klasse. Bern.
flicts of the class type, class being understood in the Vilfredo Pareto. 1955. Allgemeine Soziologie. Trans-
traditional sense.” At no point do I want to imply a lated and edited by C. Brinkmann. Tübingen.
claim for a generalized theory of social conflict. Talcott Parsons. 1957. “The Distribution of Power
in American Society,” World Politics X, No. 1,
REFERENCES October.
Reinhard Bendix. 1952. “Bureaucracy and the Prob- André Philip. 1955. La Démocratie Industrielle. Paris.
lem of Power.” Reader in Bureaucracy. Edited by Karl Renner. 1953. Wandlungen der Modernen Ge-
R.K. Merton, A.P. Gray, B. Hockey, and H.C. sellschaft: Zwei Abhandlungen Uber die Probleme
Selvin. Glencoe. der Nachkriegszeit. Vienna.
Ralf Dahrendorf. 1959. Class and Class Conflict in Bertrand Russell. 1938. Power: A New Social Anal-
Industrial Society. Stanford. ysis. London.
Theodor Geiger. 1949. Die Klassengesellschaft im Paul Sering. 1947. Jenseits des Kapitalismus.
Schmelztiegel. Cologne and Hagen. Nürnberg.
Heinz B. Kluth. 1955. “Arbeiterjugend—Begriff Max Weber. 1947. Wirtschaft und Gesellschaft (Grun-
und Wirklichkeit,” in Helmut Schelsky, ed., Ar- driss der Sozialökonomik, section III). 4th ed.
beiterjugend—gestern und heute. Heidelberg. Tübingen.

16 • Erik Olin Wright

A General Framework for the Analysis of

Class Structure

The Point of Departure: Neo-Marxist abstract concept of class relations. Yet, at least at
Analyses of Class Structure first glance, the concrete class structures of con-
At the heart of the recent resurgence of Marxist temporary advanced capitalist societies look any-
theorizing on the problem of class has been what thing but polarized. This empirical evidence of a
might be termed the “embarrassment” of the large middle class has provided critics of Marx-
middle class. For all of their disagreements, all ism with one of their principal arguments against
Marxists share a basic commitment to a polarized Marxist class theory. In response, a variety of

Eric Olin Wright. “A General Framework for the Analysis of Class Structure,” Politics and Society 13:4 (1984),
pp. 383–397, 399–402, 417–422. Copyright © 1984 by Sage Publications, Inc. Reprinted by permission of
Sage Publications, Inc.
150 • Part III: The Structure of Inequality

solutions to the problem of the middle class have structure: it tends to shift the analysis of class
been proposed in the recent Marxist debates. relations from exploitation to domination; and
Without going into any detail, it is possible it implicitly regards socialism—a society within
to identify four broadly different strategies that which the working class is the “ruling class”—as
Marxists have adopted to deal with the concep- the only possible alternative to capitalism.
tual problem of nonpolarized class positions
within a logic of polarized class relations.1 First,
the class structure of advanced capitalist societies Domination Versus Exploitation
really is polarized; the “middle class” is strictly Throughout the development of the concept
an ideological illusion. This position deals with of contradictory class locations I have insisted
the problem of the middle class by denying the that this was a reformulation of a distinctively
problem itself. Second, the middle class should Marxist class concept. As part of the rhetoric of
be viewed as a segment of some other class, typi- such an enterprise, I affirmed the relationship
cally a “new petty bourgeoisie” or “new working between class and exploitation. Nevertheless, in
class.”2 In this strategy the basic class map of cap- practice the concept of contradictory locations
italism remains intact, but significant internal within class relations rested almost exclusively on
differentiations within classes are added to the relations of domination rather than exploitation.
analysis of class structure. Third, the middle class Reference to exploitation functioned more as a
is really a new class in its own right, completely background concept to the discussion of classes
distinct from either the bourgeoisie, the proletar- than as a constitutive element of the analysis of
iat, or the petty bourgeoisie. Sometimes this class class structures. Managers, for example, were
is given a specific name, such as the Professional basically defined as a contradictory location
Managerial Class,3 sometimes it is simply called because they were simultaneously dominators
“the New Class.”4 By adding entirely new classes and dominated. Domination relations were
to the class structure, this approach more radi- also decisive in defining the class character of
cally alters the class map of capitalism than the “semiautonomous employees”—locations that,
class-segment strategy. Fourth, the positions ag- I argued, were simultaneously petty bourgeois
gregated under the popular rubric “middle class” and proletarian by virtue of their self-direction
are not really in a class at all. Rather they should within the labor process—since “autonomy”
be viewed as locations that are simultaneously defines a condition with respect to domination.
in more than one class, positions that I have This same tendency of substituting domination
characterized as “contradictory locations within for exploitation at the core of the concept of class
class relations.”5 Managers, for example, should is found in most other neo-Marxist conceptual-
be viewed as simultaneously in the working class izations of class structure.
(in so far as they are wage laborers dominated For some people, of course, marginalizing the
by capitalists) and in the capitalist class (in so concept of exploitation is a virtue, not a sin. My
far as they control the operation of production own view, however, is that this is a serious weak-
and the labor of workers). This strategy departs ness. The marginalization of exploitation both
most from the traditional Marxist vision of class undermines claims that classes have “objective”
structure since the very meaning of a “location” interests and erodes the centrality Marxists have
is altered: there is no longer a one-to-one corre- accorded class in social theory.
spondence between structural locations filled by The concept of domination does not in and
individuals and classes. of itself imply any specific interest of actors. Par-
I no longer feel that this fourth solution ents dominate small children, but this does not
is satisfactory. Specifically, it suffers from two im- imply that they have intrinsically opposed inter-
portant problems that it shares with most other ests to their children. What would make those
neo-Marxist conceptualizations of class interests antagonistic is if the relation of parents
 Wright—A General Framework for the Analysis of Class Structure • 151

to children were exploitative as well. Exploita- not unambiguously refuted, this historical vi-
tion, unlike domination, intrinsically implies a sion. As I have argued elsewhere, it is necessary
set of opposing material interests. If we wish to to at least entertain the possibility of postcapi-
retain some sense in which the interests of in- talist class structures.6 The difficulty is that with
dividuals as members of classes are not simply very few exceptions, the conceptual frameworks
whatever interests those individuals subjectively adopted by Marxists for analyzing capitalist class
hold, then the shift to a domination-centered relations do not contain adequate criteria for un-
concept renders this more difficult. derstanding postcapitalist classes.7 In particular,
Domination-centered concepts of class also all of the class categories in my analysis of con-
tend to slide into what can be termed “the mul- tradictory locations within class relations were
tiple oppressions” approach to understanding either situated firmly within capitalist relations
society. Societies, in this view, are characterized (bourgeoisie, managers, workers) or in contra-
by a plurality of oppressions each rooted in a dictory locations involving basically precapitalist
different form of domination—sexual, racial, relations (semiautonomous employees, the petty
national, economic—none of which has any bourgeoisie, small employers). There were no el-
explanatory priority over any other. Class, then, ements within this analysis of class relations in
becomes just one of many oppressions, with capitalist society that could point the direction
no particular centrality for social and historical for the analysis of postcapitalist classes. The re-
analysis. How important class is in a given soci- sult is a tendency for discussions of postcapitalist
ety becomes an historically contingent question. class structures—the class structures of “actually
Again, this displacement of class from the existing socialism”—to have a very ad hoc char-
center stage may be viewed as an achievement acter to them.
rather than a problem. It may be that class Given these conceptual problems—the shift
should not occupy a privileged place in social from exploitation to domination and the lack of
theory. But if one believes, as Marxists tra- a conceptual basis for analyzing postcapitalist
ditionally have believed, that only by giving classes—there are really two theoretical alterna-
class this central place is it possible to develop tives that could be pursued. One possibility is to
a scientific theory of the trajectory of historical celebrate the shift to a domination-centered con-
development, and in particular, a theory of the cept and use this new class concept as the basis
real historical alternatives to capitalism, then the for analyzing both capitalist and postcapitalist
domination-centered concept of class risks erod- society. This would lead class analysis firmly in
ing the theoretical justification for Marxian class the direction of Dahrendorf ’s analysis of classes
analysis itself. as positions within authority relations.8 A second
alternative is to attempt to restore exploitation as
the center of class analysis in such a way that it
Classes in Postcapitalist Societies can both accommodate the empirical complexi-
Classical Marxism was absolutely unequivocal ties of the middle class within capitalism and the
about the historical prognosis for capitalism: so- historical reality of postcapitalist class structures.
cialism—and ultimately communism—was the It is this second course of action that I will pur-
future of capitalist societies. The bearer of that sue in the rest of this paper.
necessary future was the working class. The po- The basis for this reconstruction of an
larized class structure within capitalism between exploitation-centered concept of class comes
the bourgeoisie and the proletariat thus paral- from the recent work of John Roemer.9 While
leled the polarized historical alternatives between Roemer himself has not been particularly con-
capitalism and socialism. cerned with problems of empirical investigation
The actual historical experience of the twen- or the elaboration of concrete maps of class
tieth century has called into question, although structures, nevertheless his work does provide
152 • Part III: The Structure of Inequality

a rich foundation for such endeavors. As I will labor” from one category of actors to another
attempt to show, with suitable modification and in the course of various exchange relations; the
extension, his strategy of analysis can provide a second involves adopting a kind of game-theory
rigorous basis for resolving the problems in the approach to specifying different forms of ex-
concept of contradictory class locations. ploitation. Let us briefly examine each of these
in turn.

Roemer’s Account of Class and

Exploitation The Labor-Transfer Approach
The Concept of Exploitation The analysis of labor transfers is an extension of
We observe inequalities in the distribution of in- the traditional Marxist view of exploitation, al-
comes, the real consumption packages available though Roemer self-consciously does not rely on
to individuals, families, groups. The concept of the labor theory of value in order to explore such
exploitation is a particular way of analyzing such labor transfers. The main target of his analysis is
inequalities. To describe an inequality as reflect- the view, commonly held by Marxists, that a nec-
ing exploitation is to make the claim that there essary condition for the exploitation of labor in a
exists a particular kind of causal relationship be- market economy is the institution of wage labor.
tween the incomes of different actors. More con- Roemer demonstrates two basic propositions.
cretely, we will say that the rich exploit the poor First, Roemer demonstrates that exploitation can
when two things can be established: that the wel- occur in an economy in which all producers own
fare of the rich causally depends on the depriva- their own means of production and in which
tions of the poor—the rich are rich because the there is no market in labor power and no credit
poor are poor; and that the welfare of the rich market (that is, no borrowing). The only things
depends upon the effort of the poor—the rich, that are traded are products. In such an economy
through one mechanism or another, appropriate if different producers own different amounts of
part of the fruits of labor of the poor. The first of productive assets such that different producers
these criteria by itself defines economic oppression, have to work different numbers of hours to pro-
but not exploitation. Unemployed workers, in duce the exchange-equivalent of their own sub-
these terms, are economically oppressed but not sistence, then free trade among these producers
exploited. Exploitation implies both economic will lead to exploitation of the asset poor by the
oppression and appropriation of at least part of asset rich. What Roemer shows in this simple
the social surplus by the oppressor. economy is not simply that some producers
The traditional Marxist concept of exploita- work less than others for the same subsistence,
tion is clearly a special case of this general but that the workers who work less are able to
concept. In Marxian exploitation one class ap- do so because the less-endowed producers have to
propriates the surplus labor performed by an- work more. The critical proof in this example is
other class through various mechanisms. The that if the asset-poor person simply stopped pro-
income of the exploiting class comes from the ducing—died—and the asset-rich person took
labor performed by the exploited class. There is over the asset-poor’s assets, then the asset-rich
thus a straightforward causal linkage between the producer would have to work longer hours than
poverty and effort of the exploited and the afflu- before to maintain the same subsistence. There
ence of the exploiter. The latter benefits at the is thus not merely an inequality among the pro-
expense of the former. ducers in this economy, but exploitation as well.
Roemer has attempted to elaborate this view Second, Roemer demonstrates that there is
of exploitation using two strategies. The first of complete symmetry in the structure of exploita-
these involves studying through a series of for- tion in a system in which capital hires wage la-
mal mathematical models the flows of “surplus borers and in a system in which workers rent
 Wright—A General Framework for the Analysis of Class Structure • 153

capital (that is, systems with credit and labor idea is to compare different systems of exploita-
markets). For this analysis, he compares the class tion by treating the organization of production
structures and patterns of exploitation on the as a “game” and asking if a coalition of players
two imaginary islands, “labor-market island” would be better off if they withdrew from the
and “credit-market island.” On both islands game under certain specified procedures. Dif-
some people own no means of production and ferent types of exploitation are defined by the
other people own varying amounts of the means withdrawal rules that would make certain agents
of production. The distribution of these assets better off.
is identical on the two islands. And on both is- More formally, Roemer argues that a coali-
lands people have the same motivations: they tion of actors S can be said to be exploited, and
all seek to minimize the amount of labor-time another coalition S´ (the complement of S) can
they must expend to achieve a common level of be said to be exploiting, if “there is no alterna-
subsistence. The two islands differ in only one tive, which we may conceive of as hypothetically
respect: on the labor-market island people are feasible, in which S would be better off than in
allowed to sell their labor power, whereas on its present situation, [and if ] under this alterna-
the credit-market island people are prohibited tive, the complement to S . . . would be worse off
from selling their labor power but are allowed to than at present.”10 The counterfactual in these
borrow, at some interest rate, the means of pro- two conditions is meant to convey the sense in
duction. Roemer shows that on each island there which the welfare of S´ is causally dependent
is a strict correspondence between class location upon the deprivation of S.
(derived from ownership of differing amounts Roemer uses this strategy to define three
of means of production, including no means of kinds of exploitation: feudal exploitation, cap-
production) and exploitation status (having one’s italist exploitation, and what he refers to as so-
surplus labor appropriated by someone else). cialist exploitation. Let’s begin with capitalist
This is what he terms the “Class-Exploitation exploitation. Workers own no physical assets
Correspondence Principle.” He also shows that (means of production) and sell their labor power
the two class structures are completely isomor- to capitalists for a wage. Are workers exploited
phic: every individual on one island would be in under capitalism? The answer to this question, in
exactly the same exploitation status on the other the game theoretic formulation, requires posing
island. an alternative game to the game of capitalism
The upshot of these two propositions (and within which the two conditions specified above
others that Roemer explores) is the claim that hold. What is the alternative? It is a game within
market-based exploitation is strictly a conse- which each worker receives his/her per capita
quence of inequalities in the distribution of the share of society’s total productive assets. What Ro-
means of production. However, while this may emer demonstrates is that if the coalition of all
typically play itself out through a labor market, wage-earners were to leave the game of capital-
this is only one concrete institutional form for ism with their per capita share of society’s assets,
such exploitation: it is not the necessary condi- then they would be better off than staying in
tion for the exploitation to occur. capitalism, and capitalists would be worse off.
The “withdrawal rule” in this case—leaving the
game with per capita shares of physical assets—
The Game-Theory Approach then becomes the formal “test” of whether or
While the labor-transfer analyses of exploitation not a particular social system involves capitalistic
were primarily designed to reveal the underly- exploitation.
ing logic of exploitation in market exchanges, In contrast, the withdrawal rule to specify
the game-theory approach is used by Roemer to feudal exploitation is leaving the game with one’s
compare different systems of exploitation. The personal assets (rather than one’s per capita share
154 • Part III: The Structure of Inequality

of total social assets). This is equivalent to the These conclusions have led Roemer to chal-
feudal serf being freed from all obligations based lenge directly the tendency of Marxists (like my-
on personal bondage. Peasants would be better self ) to define class relations primarily in terms
off under such circumstances; feudal lords would of domination relations within production. Of
be worse off.11 course, exploiting classes dominate exploited
The concept of the socialist exploitation is classes in the sense of preventing the exploited
the least systematically worked out in Roemer’s classes from taking the exploiting class’s produc-
analysis. The withdrawal rule in this case is leav- tive assets. But domination within production,
ing the game with one’s per capita share of in- Roemer insists, is not a central part of defining
alienable assets (skills). A coalition will be said to class relations as such.
be socialistically exploited if it would improve In previous work I have criticized Roemer’s
its position by leaving with its per capita skills position on this issue.14 I argued that class re-
while its complement would be worse off under lations intrinsically involved domination at the
such circumstances. This implies that people point of production, not simply in the repressive
with high levels of skills in the game receive high protection of the property relations as such. I
income not simply because they have high skills, now think that Roemer is correct on this point.
but because of the differentials in skill levels That capitalists boss workers around within pro-
across actors. The highly skilled would become duction is unquestionably an important feature
worse off if the unskilled obtained skills; they of most historic forms of capitalist production
thus have an interest in maintaining skill differ- and may play an important role in explaining
entials, and this is what underpins the claim that the forms of class organization and class conflict
their income reflects exploitation.12 If a skilled within production. However, the basis of the
person’s income reflected no more than the capital-labor relation should be identified with
amount of time and resources it takes to obtain relations of effective control (that is, real eco-
the skill, then there would be no skill-based ex- nomic ownership) over productive assets as such.
ploitation. The higher incomes would simply be One of the reasons why I resisted Roemer’s
reimbursement for real costs incurred. The ar- conceptualization of classes in terms of property
gument behind skill exploitation is that people relations is that it seemed to blur the difference
with scarce skills receive incomes above the costs between Marxist definitions of class and Webe-
of producing those skills, a “rent” component to rian definitions. Weberian definitions, as I con-
their income; it is this element that constitutes strued them, were “market based” definitions of
exploitation. class, whereas Marxist definitions were “produc-
tion based.” The reputed advantage of the latter
was that production was more “fundamental”
Class and Exploitation than exchange, and therefore production-based
The central message of both of Roemer’s strat- class concepts had more explanatory power than
egies for analyzing exploitation is that the ma- market-based concepts.
terial basis of exploitation is inequalities in What now seems clear to me is that definitions
distributions of productive assets, or what is of classes in terms of property relations should
usually referred to as property relations. On the not be identified with strictly market-based defi-
one hand, inequalities of assets are sufficient to nitions. Property-relations accounts of classes
account for transfers of labor surplus; on the do not define classes by income shares, by the
other hand, different forms of asset inequality results of market transactions, but by the pro-
specify different systems of exploitation. Classes ductive assets that classes control, which lead
are then defined as positions within the social them to adopt certain strategies within exchange
relations of production derived from these rela- relations and which thereby determine the out-
tions of exploitation.13 comes of those market transactions.
 Wright—A General Framework for the Analysis of Class Structure • 155

symmetrical: feudal exploitation is based on in-

Toward a General Framework of
equalities generated by ownership of labor-power
Class Analysis
assets; capitalist exploitation on inequalities gen-
Extending Roemer’s Analysis erated by ownership of alienable assets; socialist
The heart of Roemer’s analysis is the linkage exploitation on inequalities generated by own-
between the distribution of productive assets of ership of inalienable assets. And corresponding
various sorts and exploitation. Different mech- to each of these exploitation-generating inequal-
anisms of exploitation are defined by different ities of assets, there is a specific class relation:
kinds of assets, and different class systems are lords and serfs in feudalism, bourgeoisie and
defined by which of these assets is most impor- proletariat in capitalism, experts and workers in
tant for shaping the patterns of exploitation in socialism.
the society. But how, it might be asked, should “actually
In Roemer’s own explicit formulation, only existing socialist societies” be theorized within
two kinds of assets are formally considered: these categories? The anticapitalist revolution
physical assets (alienable assets in his terminol- in Russia resulted in the virtual elimination of
ogy) and skill assets (inalienable assets). The dis- private property in the means of production:
tinction between exploitation in feudalism and individuals cannot own means of production,
exploitation in capitalism revolves around the they cannot inherit them or dispose of them on
nature of the withdrawal rules with respect to a market, and so on. And yet it seems unsatis-
physical assets (withdrawing with one’s personal factory to characterize such societies simply in
assets to define feudal exploitation versus with- terms of skill-based exploitation. Experts do not
drawing with one’s per capita share of assets to appear to be the “ruling class” in those societies,
define capitalist exploitation). The feudal case, and the dynamic of the societies does not seem
however, can be characterized in a somewhat dif- to revolve around skill inequalities as such.
ferent way. Labor power is a productive asset.15 In Roemer recognized this problem and intro-
capitalist societies everyone owns one unit of this duced what he termed “status exploitation” to
asset, namely themselves. In feudalism, on the deal with it. The exploitation exercised by bu-
other hand, ownership rights over labor power reaucrats is the prototypical example. “If these
are unequally distributed: feudal lords have more positions,” Roemer writes, “required special
than one unit, serfs have less than one unit. To skills, then one might be justified in calling the
be sure, it is not typical of feudalism for serfs differential remuneration to these positions an
to own no labor power—they are generally not aspect of socialist [skill-based] exploitation. . . .
slaves divested of all ownership rights in their [However] there is some extra remuneration to
own labor power—but they do not have com- holders of those positions which accrues solely
plete effective control over their own persons as by virtue of the position and not by virtue of the
productive actors, and this is what it means to skill necessary to carry out the tasks associated
“own” one’s own labor power assets. The with- with it. These special payments to positions give
drawal rule that defines feudal exploitation can rise to status exploitation.”16
then be specified as leaving the feudal game with Roemer’s concept of status exploitation is un-
one’s per capita share of society’s assets in labor satisfactory for two principal reasons. First, it is
power, namely one unit. Feudal exploitation is outside of the logic of the rest of his analysis of
thus exploitation (transfers of labor) that results exploitation. In each of the other cases, exploita-
from inequalities in the distribution of assets in tion is rooted in relations to the forces of pro-
labor power. duction. Each of the other forms of exploitation
Reformulating feudal exploitation in this is “materialist” not only because the concept is
manner makes the game-theory specification meant to explain material distribution, but also
of different exploitations in Roemer’s analysis because it is based on the relation to the material
156 • Part III: The Structure of Inequality

conditions of production. “Status” exploitation organized at the center. The control over orga-
has no necessary relationship to production at nization assets is no longer simply the task of
all. Second, it is hard to rigorously distinguish firm-level managers but extends into the central
status exploitation from feudal exploitation. The organs of planning within the state. Exploitation
“lord” receives remuneration strictly because of in such societies is thus based on bureaucratic
an incumbency in a position, not because of power: the control over organization assets de-
skills or ownership of capital. Yet, it hardly seems fines the material basis for class relations and
reasonable to consider the logic of exploitation exploitation.
and class in the contemporary Soviet Union and This notion of organization assets bears a
in fourteenth-century feudal Europe as being es- close relation to the problem of authority and
sentially the same. hierarchy. The asset is organization. The activity
The problems with the concept of status of using that asset is coordinated decision mak-
exploitation can be solved by analyzing ex- ing over a complex technical division of labor.
ploitation based on a fourth element in the in- When that asset is distributed unequally, so
ventory of productive assets, an asset that can some positions have effective control over much
be referred to as “organization.” As both Adam more of the asset than others, then the social re-
Smith and Marx noted, the technical division lation with respect to that asset takes the form of
of labor among producers is itself a source of hierarchical authority. Authority, however, is not
productivity. The way the production process is the asset as such; organization is the asset and is
organized is a productive resource independent controlled through a hierarchy of authority.
of the expenditure of labor power, the use of The claim that effective control over organi-
means of production, or the skills of the pro- zation assets is a basis of exploitation is equiva-
ducer. Of course there is an interrelationship lent to saying that nonmanagers would be better
between organization and these other assets, just off and managers/bureaucrats worse off if non-
as there is an interdependence between means managers were to withdraw with their per capita
of production and skills. But organization—the share of organization assets (or equivalently, if
conditions of coordinated cooperation among organizational control were democratized); and
producers in a complex division of labor—is a that by virtue of effectively controlling organiza-
productive resource in its own right. tion assets managers/bureaucrats control part or
How is this asset distributed in different all of the socially produced surplus.17
kinds of societies? In contemporary capitalism,
organization assets are generally controlled by
managers and capitalists: managers control the A Typology of Class Structures,
organization assets within specific firms under Assets, and Exploitation
constraints imposed by the ownership of the If we add organization assets to the list in Ro-
capital assets by capitalists. Entrepreneurial cap- emer’s analysis, we generate the more complex
italists directly control both kinds of assets (and typology presented in Table 16.1. Let us briefly
probably skill assets as well); pure rentier capital- look at each row of this table and examine its
ists (“coupon clippers”) only own capital assets. logic. Feudalism is a class system based on un-
Because of the anarchy of the capitalist market, equal distribution of ownership rights in labor
no set of actors controls the technical division of power. What “personal bondage” means is that
labor across firms. feudal lords have partial effective economic con-
In state bureaucratic socialism, organiza- trol over vassals. The empirical manifestation of
tion assets assume a much greater importance. this unequal distribution of ownership rights
Controlling the technical division of labor—the over labor power in classical feudalism is the co-
coordination of productive activities within and ercive extraction of labor dues from serfs. When
across labor processes—becomes a societal task corvée labor is commuted to rents in kind and
 Wright—A General Framework for the Analysis of Class Structure • 157

Table 16.1. Assets, Exploitation, and Classes

eventually money rents, the feudal character of production, but not having sufficient assets to
the exploitation relation is reflected in legal pro- actually produce commodities without the assis-
hibitions on the movement of peasants off the tance of merchant capitalists. Such workers were
land. The “flight” of a peasant to the city is, in still being capitalistically exploited even though
effect, a form of theft: the peasant is stealing part there was no formal labor market with wages. In
of the labor power owned by the lord. Feudal all capitalist exploitation, the mediating mecha-
lords may also have more means of production nism is market exchanges. Unlike in feudalism,
than serfs, more organizational assets, and more surplus is not directly appropriated from workers
productive skills (although this is unlikely), and in the form of coerced labor. Rather, it is appro-
thus they may be exploiters with respect to these priated through market exchanges: workers are
assets as well. What defines the society as “feu- paid a wage that covers the costs of production
dal,” however, is the primacy of the distinctively of their labor power; capitalists receive an in-
feudal mechanisms of exploitation. Accordingly, come from the sale of the commodities produced
feudal class relations will be the primary struc- by workers. The difference in these quantities
tural basis of class struggle. constitutes the exploitative surplus appropriated
The bourgeois revolutions radically redistrib- by capitalists.
uted productive assets in people: everyone, at Anticapitalist revolutions attempt to elimi-
least in principle, owns one unit. This is what nate the distinctively capitalist form of exploita-
is meant by “bourgeois freedoms,” and in this tion, exploitation based on private ownership of
sense capitalism can be regarded as an histori- the means of production. The nationalization of
cally progressive force. But capitalism raises the the principal means of production is, in effect, a
second type of exploitation, exploitation based radical equalization of ownership of capital: ev-
on property relations in means of production, to eryone owns one citizen-share. Such revolutions,
an unprecedented level. however, do not eliminate, and indeed may con-
The typical institutional form of capitalist siderably strengthen and deepen, inequalities
class relations is capitalists having full ownership of effective control over organization assets.
rights in the means of production and workers Whereas in capitalism the control over organi-
none. Other possibilities, however, have existed zation assets does not extend beyond the firm,
historically. Cottage industries in early capitalism in state bureaucratic socialism the coordinated
involved workers owning some of their means of integration of the division of labor extends to
158 • Part III: The Structure of Inequality

the whole society through institutions of cen- the skill may be restricted through various mech-
tral state planning. The mechanism by which anisms, creating an artificial scarcity of trained
this generates exploitative transfers of surplus people; third, a certification system may be es-
involves the centrally planned bureaucratic ap- tablished that prohibits uncertified people from
propriation and distribution of the surplus along being employed to use the skill even if they have
hierarchical principles. The corresponding class it. In all of these cases, the exploitation comes
relation is therefore between managers/bureau- from the skilled/certified individual receiving an
crats—people who control organization assets— income that is above the costs of production of
and nonmanagers. the skills by virtue of the scarcity of the availabil-
The historical task of revolutionary transfor- ity of the skill.
mation of state bureaucratic socialism revolves In this conceptualization of socialism, a so-
around the equalization of effective economic cialist society is essentially a kind of democratic
control over organization assets, or, equivalently, technocracy. Experts control their own skills and
the democratization of bureaucratic apparatuses knowledge within production, and by virtue of
of production. This does not imply total direct such control are able to appropriate some of the
democracy, where all decisions of any conse- surplus out of production. However, because of
quence are directly made in democratic assem- the democratization of organization assets, ac-
blies. There will still inevitably be delegated tual planning decisions will not be made under
responsibilities, and there certainly can be rep- the direct control of experts but will be made
resentative forms of democratic control. But it through some kind of democratic procedure
does mean that the basic parameters of planning (this is in effect what democratization of organi-
and coordinating social production are made zation assets means: equalizing control over the
through democratic mechanisms and that in- planning and coordinating of social production).
cumbency within delegated positions of respon- This means that the actual class power of a so-
sibility does not give incumbents any personal cialist technocratic exploiting class will be much
claims on the social surplus. Such equalization, weaker than the class power of exploiting classes
however, would not necessarily affect exploita- in other class systems. Their ownership rights ex-
tion based on skills/credentials. Such exploita- tend to only a limited part of the social surplus.
tion would remain a central feature of socialism. This much more limited basis of domination
“Skill” in this context is not a trivial concept. implied by skill-based exploitation is consistent
The mere possession of enhanced laboring ca- with the spirit, if not the letter, of Marx’s claim
pabilities acquired through training is not suffi- that socialism is the “lower stage” of “commu-
cient to generate relations of exploitation, since nism,” since classes are already in a partial state
the income of such trained labor may simply of dissolution in a society with only skill-based
reflect the costs of acquiring the training. In exploitation. Communism itself, then, would be
such cases there is neither a transfer of surplus, understood as a society within which skill-based
nor would the untrained be better off under the exploitation itself had “withered away,” that is, in
game-theory specification of exploitation. For a which ownership rights in skills had been equal-
skill to be the basis of exploitation, therefore, it ized. This does not mean, it must be stressed,
has to be in some sense scarce relative to its de- that all individuals would actually possess the
mand, and there must be a mechanism through same skills in communism, any more than elim-
which individual owners of scarce skills are able inating property rights in means of production
to translate that scarcity into higher incomes. implies that all individuals would actively use the
There are basically three ways that skills can same amount of physical capital. What is equal-
become scarce: first, they may require special ized is effective control over skills as a productive
talents that are naturally scarce in a population; resource and claims to differential incomes re-
second, access to the training needed to develop sulting from differential use of skills.18 . . .
 Wright—A General Framework for the Analysis of Class Structure • 159

Table 16.2. Basic Typology of Exploitation and Class

wage-earners (for example, professionals)

The Middle Classes and Contradictory in capitalism are a good example: they are
Locations capitalistically exploited because they lack
The framework in Table 16.1 enables us to pose assets in capital, and yet they are skill ex-
the problem of middle classes in a new way. Two ploiters. Such positions are what are typi-
different kinds of nonpolarized class locations cally referred to as the “new middle class” of
can be defined in the logic of this framework: a given system.
1. There are class locations that are neither ex-
ploiters nor exploited, that is, people who Table 16.2 presents a schematic typology of
have precisely the per capita level of the rele- such complex class locations for capitalism. The
vant asset. A petty bourgeois, self-employed typology is divided into two segments: one for
producer with average capital stock, for owners of the means of production and one for
example, would be neither exploiter nor nonowners. Within the wage-earner section of
exploited within capitalist relations. These the typology, locations are distinguished by the
kinds of positions are what can be called two subordinate relations of exploitation char-
the “traditional” or “old” middle class of a acteristic of capitalist society—organization as-
particular kind of class system. sets and skill/credential assets. It is thus possible
2. Since concrete societies are rarely, if ever, within this framework to distinguish a whole
characterized by a single mode of produc- terrain of class locations in capitalist society that
tion, the actual class structures of given are distinct from the polarized classes of the cap-
societies will be characterized by complex italist mode of production: expert managers, non-
patterns of intersecting exploitation rela- managerial experts, nonexpert managers, and so
tions. There will therefore tend to be some on.19
positions that are exploiting along one di- What is the relationship between this het-
mension of exploitation relations and are erogeneous exploitation definition of the mid-
exploited along another. Highly skilled dle class and my previous conceptualization of
 160 • Part III: The Structure of Inequality

such positions as contradictory locations within One of the upshots of this reconceptualiza-
class relations? There is still a sense in which such tion of the middle class is that it is no longer
positions could be characterized as “contradic- axiomatic that the proletariat is the unique, or
tory locations,” for they will typically hold con- perhaps even the central, rival to the capitalist
tradictory interests with respect to the primary class for class power in capitalist society. That
forms of class struggle in capitalist society, the classical Marxist assumption depended upon
struggle between labor and capital. On the one the thesis that there were no other classes within
hand, they are like workers, in being excluded capitalism that could be viewed as the “bearers”
from ownership of the means of production. of an historical alternative to capitalism. Social-
On the other hand, they have interests opposed ism (as the transition to communism) was the
to workers because of their effective control of only possible future for capitalism. What Table
organization and skill assets. Within the strug- 16.3 suggests is that there are other class forces
gles of capitalism, therefore, these new middle within capitalism that potentially pose an alter-
classes do constitute contradictory locations, or native to capitalism. This does not imply that
more precisely, contradictory locations within there is any inevitability to the sequence feudal-
exploitation relations. ism–capitalism–state bureaucratic socialism–so-
This conceptualization of the middle classes cialism–communism; state bureaucrats are not
also suggests that historically the principal forms inevitably destined to be the future ruling class of
of contradictory locations will vary depending present-day capitalisms. But it does suggest that
upon the particular combinations of exploitation the process of class formation and class struggle
relations in a given society. These principal con- is considerably more complex and indeterminate
tradictory locations are presented in Table 16.3. than the traditional Marxist story has allowed.
In feudalism, the critical contradictory location
is constituted by the bourgeoisie, the rising class NOTES
of the successor mode of production. Within 1. For a more detailed review of these alterna-
tives, see E.O. Wright, “Varieties of Marxist Con-
capitalism, the central contradictory location
cepts of Class Structure,” Politics and Society, vol. 9,
within exploitation relations is constituted by
no. 3 (1980).
managers and state bureaucrats. They embody 2. The leading proponent of the concept of the
a principle of class organization that is quite dis- “new petty bourgeoisie” is N. Poulantzas, Classes in
tinct from capitalism and that potentially poses Contemporary Capitalism (London: Verso, 1975).
an alternative to capitalist relations. This is par- For the new-working-class concept, see S. Mallet, La
ticularly true for state managers who, unlike Nouvelle Classe Ouvrière (Paris: Seuil, 1963).
corporate managers, are less likely to have their 3. B. Ehrenreich and J. Ehrenreich, “The Profes-
careers tightly integrated with the interests of the sional and Managerial Class,” Radical America, vol.
capitalist class. Finally, in state bureaucratic so- 11, no. 2 (1977).
4. A. Gouldner, The Future of Intellectuals and
cialism, the “intelligentsia” broadly defined con-
the Rise of the New Class (New York: Seabury Press,
stitutes the pivotal contradictory location. 1979); and G. Konrad and I. Szelényi, Intellectuals
on the Road to Class Power (New York: Harcourt,
Table 16.3. Basic Classes and Contradictory Locations Brace, Jovanovich, 1979).
in Successive Modes of Production 5. E.O. Wright, “Class Boundaries in Advanced
Capitalist Societies,” New Left Review, no. 98 (1976);
and Class, Crisis and the State (London: New Left
Books, 1978). See also G. Carchedi, The Economic
Identification of Social Classes (London: Routledge
and Kegan Paul, 1977).
6. E.O. Wright, “Capitalism’s Futures,” Socialist
Review, no. 68 (1983).
 Wright—A General Framework for the Analysis of Class Structure • 161

7. A partial exception to this can be found in their imputed contribution to economic productiv-
arguments for the existence of a “new class” of in- ity” (Future of Intellectuals, p. 21). While Gouldner
tellectuals and/or bureaucrats in capitalist and post- does not characterize this income allocation process
capitalist society. See: A. Gouldner, The Future of in terms of exploitation, Roemer’s exploitation con-
Intellectuals; and I. Szelényi and W. Martin, New cept would fit comfortably within Gouldner’s gen-
Class Theory and Beyond (unpublished book manu- eral approach.
script, Department of Sociology, University of Wis- 14. E.O. Wright, “The Status of the Political in
consin, 1985). the Concept of Class Structure,” Politics and Society,
8. R. Dahrendorf, Class and Class Conflict in In- vol. 11, no. 3 (1982).
dustrial Society (Palo Alto: Stanford University Press, 15. See G.A. Cohen, Karl Marx’s Theory of His-
1959). tory: A Defense (Princeton: Princeton University
9. Roemer is a Marxist economist engaged in a Press, 1978), pp. 40–41, for a discussion of why
long-term project of elaborating what he calls the labor power should be considered part of the forces
“microfoundations” of Marxist theory. His most of production (that is, a productive asset).
important work is entitled A General Theory of Ex- 16. Roemer, A General Theory, p. 243.
ploitation and Class (Cambridge: Harvard University 17. This “control of the surplus,” it must be
Press, 1982). noted, is not the equivalent of the actual personal
10. Roemer, A General Theory, pp. 194–95. consumption income of managers and bureaucrats,
11. But note: workers in capitalism are not feu- any more than capitalist profits or feudal rents are
dalistically exploited; they would be worse off, not the equivalent of the personally consumed income
better off, if they withdrew from the game of cap- of capitalists and feudal lords. It is historically vari-
italism with only their personal assets. As Roemer able both within and between types of societies
argues, the claim by neoclassical theorists that wage what fraction of the surplus effectively controlled
earners in capitalism are not exploited is generally by exploiting classes is used for personal consump-
equivalent to the claim that they are not feudalis- tion and what portion is used for other purposes
tically exploited, that is, that they are not subjected (feudal military expenditures, capitalist accu-
to surplus extraction based on relations of personal mulation, organization growth). The claim that
bondage. See Roemer, A General Theory, p. 206. managers-bureaucrats would be “worse off ” under
12. The asset-exploitation nexus thus depends conditions of a redistribution of organization assets
upon the capacity of asset-holders to deprive others refers to the amount of income they effectively con-
of that asset. The social basis of exploitation, under- trol, which is therefore potentially available for per-
stood in this way, is quite similar to Frank Parkin’s sonal appropriation, not simply the amount they
characterization of Weber’s concept of social closure personally consume.
as “the process by which social collectivities seek to 18. It may be utopian to imagine a society with-
maximize rewards by restricting access to resources out skill-based exploitation, or even a society with-
and opportunities to a limited circle of eligibles.” F. out organization-asset exploitation, particularly if we
Parkin, Marxism and Class Theory: A Bourgeois Cri- reject the claim that a future society will ever exist
tique (New York: Columbia University Press, 1979). in a state of absolute abundance. In the absence of
While Parkin’s central concern is with the kinds of absolute abundance, all societies will face dilemmas
attributes that serve as the basis for closure—race, and trade-offs around the problem of distribution
religion, language—Roemer’s is with the nature of of consumption, and such dilemmas may pose in-
the resources (productive assets) over which closure tractable incentive problems in the absence of ex-
is organized. ploitation. For a careful exposition of the problem of
13. Roemer’s conceptualization of the relation- utopian fantasies in Marxist theory, see A. Nove, The
ship between class and exploitation is similar in cer- Economics of Feasible Socialism (Hemel Hempstead:
tain aspects to Alvin Gouldner’s, although Roemer is George Allen and Unwin, 1983).
unaware of Gouldner’s work. Gouldner defines the 19. The labor-force data in this table come from
“New Class” as a cultural bourgeoisie defined by its the comparative project on class structure and class
control over “cultural capital,” where “capital” is de- consciousness, University of Wisconsin. Details of
fined as “any produced object used to make saleable the coding of categories and the operationalization
utilities, thus providing its possessor with incomes, of variables can be found in E.O. Wright, Classes
or claims to incomes defined as legitimate because of (London: Verso, 1985), appendix 2.
162 • Part III: The Structure of Inequality

17 • Immanuel Wallerstein

Class Conlict in the Capitalist World Economy

What is capitalism as a mode of production? we would not have capitalism. This is in fact
This is not an easy question, and for that reason approximately what had happened in various
is not in fact a widely discussed one. It seems pre-capitalist systems.
to me that there are several elements that com- Capitalism involves in addition structures
bine to constitute the ‘model’. Capitalism is the and institutions which reward primarily that
only mode of production in which the maxi- subsegment of the owners and controllers who
mization of surplus creation is rewarded per se. use the surplus value only in part for their own
In every historical system, there has been some consumption, and in another (usually larger)
production for use, and some production for ex- part for further investment. The structure of the
change, but only in capitalism are all producers market ensures that those who do not accumu-
rewarded primarily in terms of the exchange late capital (but merely consume surplus value)
value they produce and penalized to the extent lose out economically over time to those who do
they neglect it. The ‘rewards’ and ‘penalties’ are accumulate capital.
mediated through a structure called the ‘market’. We may thereupon designate as the bourgeoi-
It is a structure but not an institution. It is a sie those who receive a part of the surplus value
structure molded by many institutions (political, they do not themselves create and use some of
economic, social, even cultural), and it is the it to accumulate capital. What defines the bour-
principal arena of economic struggle. geois is not a particular profession and not even
Not only is surplus maximized for its own the legal status of proprietor (although this was
sake, but those who use the surplus to accumu- historically important) but the fact that the bour-
late more capital to produce still more surplus geois obtains, either as an individual or a mem-
are further rewarded. Thus the pressure is for ber of some collectivity, a part of the surplus that
constant expansion, although the individualistic he did not create and is in the position to invest
premise of the system simultaneously renders (again either individually or as part of a collectiv-
constant expansion impossible. ity) some of this surplus in capital goods.
How does the search for profit operate? It op- There is a very large gamut of organizational
erates by creating legal protections for individual arrangements which can permit this, of which
firms (which can range in size from individuals the classic model of the ‘free entrepreneur’ is
to quite large organizations, including parastatal only one. Which organizational arrangements
agencies) to appropriate the surplus value created prevail at particular moments of time in partic-
by the labor of the primary producers. Were all ular states (for these arrangements are depen-
or most of this surplus value however consumed dent on the legal framework) is a function of
by the few who owned or controlled the ‘firms’, the state of development of the world-economy

Immanuel Wallerstein. “Class Conflict in the Capitalist World-Economy,” in The Capitalist World-Economy
(Cambridge: Cambridge University Press, 1979), pp. 285–286, 291–293. Copyright © 1979 by Immanuel
Wallerstein. Reprinted by permission of the author.
 Wallerstein—Class Conlict in the Capitalist World Economy • 163

as a whole (and the role of a particular state account for recurring initial thrusts to specializa-
in that world-economy) on the one hand, and tion (natural or socio-historical advantages in the
the consequent forms of class struggle in the production of one or another commodity), but
world-economy (and within the particular state) it is the state system which encrusts, enforces,
on the other. Hence, like all other social con- and exaggerates the patterns, and it has regularly
structs, the ‘bourgeoisie’ is not a static phenome- required the use of state machinery to revise the
non. It is the designation of a class in the process pattern of the worldwide division of labor.
of perpetual re-creation and hence of constant Furthermore, the ability of states to interfere
change of form and composition. . . . with flows becomes differentiated. That is, core
The fundamental role of the state as an in- states become stronger than peripheral states,
stitution in the capitalist world-economy is to and use this differential power to maintain a
augment the advantage of some against others differential degree of interstate freedom of flow.
in the market—that is, to reduce the ‘freedom’ of Specifically, core states have historically arranged
the market. Everyone is in favor of this, as long that worldwide and over time, money and goods
as one is the beneficiary of the ‘distortion’, and have flowed more ‘freely’ than labor. The reason
everyone opposed to the extent that one loses. It for doing this is that core states have thereby re-
is all a matter of whose ox is being gored. ceived the advantages of ‘unequal exchange’.
The modes of augmenting advantage are In effect, unequal exchange is simply a part
many. The state can transfer income by taking of the worldwide process of the appropriation
it from some and giving it to others. The state of surplus. We analyze falsely if we try to take
can restrict access to the market (of commodi- literally the model of one proletarian relating to
ties or of labor) which favor those who thereby one bourgeois. In fact, the surplus value that the
share in the oligopoly or oligopsony. The state producer creates passes through a series of per-
can restrain persons from organizing to change sons and firms. It is therefore the case that many
the actions of the state. And, of course, the state bourgeois share the surplus value of one prole-
can act not only within its jurisdiction but be- tarian. The exact share of different groups in the
yond it. This may be licit (the rules concerning chain (property owner, merchants, intermediate
transit over boundaries) or illicit (interference in consumers) is subject to much historical change
the internal affairs of another state). Warfare is of and is itself a principal analytical variable in the
course one of the mechanisms used. functioning of the capitalist world-economy.
What is crucial to perceive is that the state is This chain of the transfer of surplus value fre-
a special kind of organization. Its ‘sovereignty’, quently (often? almost always?) traverses national
a notion of the modern world, is the claim to boundaries and, when it does, state operations
the monopolization (regulation) of the legit- intervene to tilt the sharing among bourgeois to-
imate use of force within its boundaries, and wards those bourgeois located in core states. This
it is in a relatively strong position to interfere is unequal exchange, a mechanism in the overall
effectively with the flow of factors of produc- process of the appropriation of surplus value.
tion. Obviously also it is possible for particular One of the socio-geographic consequences
social groups to alter advantage by altering state of this system is the uneven distribution of the
boundaries; hence both movements for secession bourgeoisie and proletariat in different states,
(or autonomy) and movements for annexation core states containing a higher percentage na-
(or federation). tionally of bourgeois than peripheral states.
It is this realistic ability of states to interfere In addition, there are systematic differences
with the flow of factors of production that pro- in kinds of bourgeois and proletarians located
vides the political underpinnings of the structural in the two zones. For example, the percentage
division of labor in the capitalist world-economy of wage-earning proletarians is systematically
as a whole. Normal market considerations may higher in core states.
164 • Part III: The Structure of Inequality

Since states are the primary arena of political via the intervening variable of their impact on
conflict in a capitalist world-economy, and since worldwide class consciousness of the proletariat.
the functioning of the world-economy is such Core and periphery then are simply phrases
that national class composition varies widely, it is to locate one crucial part of the system of surplus
easy to perceive why the politics of states differ- appropriation by the bourgeoisie. To oversim-
entially located in relation to the world-economy plify, capitalism is a system in which the surplus
should be so dissimilar. It is also then easy to value of the proletarian is appropriated by the
perceive that using the political machinery of bourgeois. When this proletarian is located in
a given state to change the social composition a different country from this bourgeois, one of
and world-economic function of national pro- the mechanisms that has affected the process of
duction does not per se change the capitalist appropriation is the manipulation of controlling
world-system as such. flows over state boundaries. This results in pat-
Obviously, however, these various national terns of ‘uneven development’ which are sum-
thrusts to a change in structural position (which marized in the concepts of core, semiperiphery,
we misleadingly often call ‘development’) do in and periphery. This is an intellectual tool to help
fact affect, indeed over the long run do in fact analyze the multiple forms of class conflict in the
transform, the world-system. But they do so capitalist world-economy.
 Weber—Class, Status, Party • 165

Weberian Theories of Class

18 • Max Weber

Class, Status, Party

Economically Determined Power and typical big speculator, deliberately relinquishes

the Social Order social honor. Quite generally, ‘mere economic’
Law exists when there is a probability that an power, and especially ‘naked’ money power, is by
order will be upheld by a specific staff of men no means a recognized basis of social honor. Nor
who will use physical or psychical compulsion is power the only basis of social honor. Indeed,
with the intention of obtaining conformity with social honor, or prestige, may even be the basis of
the order, or of inflicting sanctions for infringe- political or economic power, and very frequently
ment of it.1 The structure of every legal order has been. Power, as well as honor, may be guar-
directly influences the distribution of power, anteed by the legal order, but, at least normally,
economic or otherwise, within its respective it is not their primary source. The legal order
community. This is true of all legal orders and is rather an additional factor that enhances the
not only that of the state. In general, we un- chance to hold power or honor; but it cannot
derstand by ‘power’ the chance of a man or of always secure them.
a number of men to realize their own will in a The way in which social honor is distributed
communal action even against the resistance of in a community between typical groups partici-
others who are participating in the action. pating in this distribution we may call the ‘social
‘Economically conditioned’ power is not, of order.’ The social order and the economic order
course, identical with ‘power’ as such. On the are, of course, similarly related to the ‘legal order.’
contrary, the emergence of economic power may However, the social and the economic order
be the consequence of power existing on other are not identical. The economic order is for us
grounds. Man does not strive for power only merely the way in which economic goods and
in order to enrich himself economically. Power, services are distributed and used. The social order
including economic power, may be valued ‘for is of course conditioned by the economic order
its own sake.’ Very frequently the striving for to a high degree, and in its turn reacts upon it.
power is also conditioned by the social ‘honor’ Now: ‘classes,’ ‘status groups,’ and ‘parties’ are
it entails. Not all power, however, entails social phenomena of the distribution of power within
honor: The typical American Boss, as well as the a community.

Max Weber. “Class, Status, Party,” in Economy and Society: An Outline of Interpretive Sociology, Vol. 1, edited
by Guenther Roth and Claus Wittich, translated by Ephraim Rischoff, Hans Gerth, A. M. Henderson, et al.
Copyright © 1978 by the Regents of the University of California. Published by the University of California
Press.
166 • Part III: The Structure of Inequality

or indirectly in returns on capital. All this holds

Determination of Class-Situation by true within the area in which pure market con-
Market-Situation ditions prevail. ‘Property’ and ‘lack of property’
In our terminology, ‘classes’ are not commu- are, therefore, the basic categories of all class sit-
nities; they merely represent possible, and fre- uations. It does not matter whether these two
quent, bases for communal action. We may categories become effective in price wars or in
speak of a ‘class’ when (1) a number of people competitive struggles.
have in common a specific causal component of Within these categories, however, class situa-
their life chances, in so far as (2) this component tions are further differentiated: on the one hand,
is represented exclusively by economic interests according to the kind of property that is usable
in the possession of goods and opportunities for for returns; and, on the other hand, according
income, and (3) is represented under the condi- to the kind of services that can be offered in
tions of the commodity or labor markets. [These the market. Ownership of domestic buildings;
points refer to ‘class situation,’ which we may productive establishments; warehouses; stores;
express more briefly as the typical chance for a agriculturally usable land, large and small hold-
supply of goods, external living conditions, and ings—quantitative differences with possibly
personal life experiences, in so far as this chance qualitative consequences; ownership of mines;
is determined by the amount and kind of power, cattle; men (slaves); disposition over mobile
or lack of such, to dispose of goods or skills for instruments of production, or capital goods of
the sake of income in a given economic order. all sorts, especially money or objects that can be
The term ‘class’ refers to any group of people exchanged for money easily and at any time; dis-
that is found in the same class situation.] position over products of one’s own labor or of
It is the most elemental economic fact that others’ labor differing according to their various
the way in which the disposition over material distances from consumability; disposition over
property is distributed among a plurality of transferable monopolies of any kind—all these
people, meeting competitively in the market for distinctions differentiate the class situations of
the purpose of exchange, in itself creates specific the propertied just as does the ‘meaning’ which
life chances. According to the law of marginal they can and do give to the utilization of prop-
utility this mode of distribution excludes the erty, especially to property which has money
non-owners from competing for highly valued equivalence. Accordingly, the propertied, for in-
goods; it favors the owners and, in fact, gives to stance, may belong to the class of rentiers or to
them a monopoly to acquire such goods. Other the class of entrepreneurs.
things being equal, this mode of distribution mo- Those who have no property but who offer
nopolizes the opportunities for profitable deals services are differentiated just as much accord-
for all those who, provided with goods, do not ing to their kinds of services as according to the
necessarily have to exchange them. It increases, way in which they make use of these services,
at least generally, their power in price wars with in a continuous or discontinuous relation to a
those who, being propertyless, have nothing to recipient. But always this is the generic conno-
offer but their services in native form or goods in tation of the concept of class: that the kind of
a form constituted through their own labor, and chance in the market is the decisive moment
who above all are compelled to get rid of these which presents a common condition for the in-
products in order barely to subsist. This mode of dividual’s fate. ‘Class situation’ is, in this sense,
distribution gives to the propertied a monopoly ultimately ‘market situation.’ The effect of naked
on the possibility of transferring property from possession per se, which among cattle breeders
the sphere of use as a ‘fortune,’ to the sphere of gives the nonowning slave or serf into the power
‘capital goods’; that is, it gives them the entrepre- of the cattle owner, is only a forerunner of real
neurial function and all chances to share directly ‘class’ formation. However, in the cattle loan
 Weber—Class, Status, Party • 167

and in the naked severity of the law of debts in Societal action, on the other hand, is oriented to
such communities, for the first time mere ‘pos- a rationally motivated adjustment of interests.]
session’ as such emerges as decisive for the fate The rise of societal or even of communal action
of the individual. This is very much in contrast from a common class situation is by no means a
to the agricultural communities based on labor. universal phenomenon.
The creditor-debtor relation becomes the basis The class situation may be restricted in its
of ‘class situations’ only in those cities where a effects to the generation of essentially similar
‘credit market,’ however primitive, with rates reactions, that is to say, within our terminol-
of interest increasing according to the extent of ogy, of ‘mass actions.’ However, it may not have
dearth and a factual monopolization of credits, even this result. Furthermore, often merely an
is developed by a plutocracy. Therewith ‘class amorphous communal action emerges. For ex-
struggles’ begin. ample, the ‘murmuring’ of the workers known
Those men whose fate is not determined by in ancient oriental ethics: the moral disapproval
the chance of using goods or services for them- of the work-master’s conduct, which in its prac-
selves on the market, e.g. slaves, are not, how- tical significance was probably equivalent to an
ever, a ‘class’ in the technical sense of the term. increasingly typical phenomenon of precisely the
They are, rather, a ‘status group.’ latest industrial development, namely, the ‘slow
down’ (the deliberate limiting of work effort) of
laborers by virtue of tacit agreement. The degree
Communal Action Flowing from in which ‘communal action’ and possibly ‘socie-
Class Interest tal action,’ emerges from the ‘mass actions’ of the
According to our terminology, the factor that members of a class is linked to general cultural
creates ‘class’ is unambiguously economic inter- conditions, especially to those of an intellectual
est, and indeed, only those interests involved in sort. It is also linked to the extent of the contrasts
the existence of the ‘market.’ Nevertheless, the that have already evolved, and is especially linked
concept of ‘class-interest’ is an ambiguous one: to the transparency of the connections between
even as an empirical concept it is ambiguous as the causes and the consequences of the ‘class sit-
soon as one understands by it something other uation.’ For however different life chances may
than the factual direction of interests following be, this fact in itself, according to all experience,
with a certain probability from the class situation by no means gives birth to ‘class action’ (commu-
for a certain ‘average’ of those people subjected to nal action by the members of a class). The fact of
the class situation. The class situation and other being conditioned and the results of the class sit-
circumstances remaining the same, the direction uation must be distinctly recognizable. For only
in which the individual worker, for instance, is then the contrast of life chances can be felt not
likely to pursue his interests may vary widely, as an absolutely given fact to be accepted, but as
according to whether he is constitutionally qual- a resultant from either (1) the given distribution
ified for the task at hand to a high, to an average, of property, or (2) the structure of the concrete
or to a low degree. In the same way, the direc- economic order. It is only then that people may
tion of interests may vary according to whether react against the class structure not only through
or not a communal action of a larger or smaller acts of an intermittent and irrational protest, but
portion of those commonly affected by the ‘class in the form of rational association. There have
situation,’ or even an association among them, been ‘class situations’ of the first category (1), of
e.g. a ‘trade union,’ has grown out of the class a specifically naked and transparent sort, in the
situation from which the individual may or may urban centers of Antiquity and during the Mid-
not expect promising results. [Communal ac- dle Ages; especially then, when great fortunes
tion refers to that action which is oriented to the were accumulated by factually monopolized
feeling of the actors that they belong together. trading in industrial products of these localities
168 • Part III: The Structure of Inequality

or in foodstuffs. Furthermore, under certain cir- kind of ‘legal order.’ Each kind of class situation,
cumstances, in the rural economy of the most and above all when it rests upon the power of
diverse periods, when agriculture was increas- property per se, will become most clearly effica-
ingly exploited in a profit-making manner. The cious when all other determinants of reciprocal
most important historical example of the second relations are, as far as possible, eliminated in their
category (2) is the class situation of the modern significance. It is in this way that the utilization
‘proletariat.’ of the power of property in the market obtains its
most sovereign importance.
Now ‘status groups’ hinder the strict carry-
Types of ‘Class Struggle’ ing through of the sheer market principle. In the
Thus every class may be the carrier of any one of present context they are of interest to us only
the possibly innumerable forms of ‘class action,’ from this one point of view. Before we briefly
but this is not necessarily so: In any case, a class consider them, note that not much of a general
does not in itself constitute a community. To nature can be said about the more specific kinds
treat ‘class’ conceptually as having the same value of antagonism between ‘classes’ (in our mean-
as ‘community’ leads to distortion. That men in ing of the term). The great shift, which has been
the same class situation regularly react in mass going on continuously in the past, and up to our
actions to such tangible situations as economic times, may be summarized, although at the cost
ones in the direction of those interests that are of some precision: the struggle in which class
most adequate to their average number is an im- situations are effective has progressively shifted
portant and after all simple fact for the under- from consumption credit toward, first, compet-
standing of historical events. Above all, this fact itive struggles in the commodity market and,
must not lead to that kind of pseudo-scientific then, toward price wars on the labor market.
operation with the concepts of ‘class’ and ‘class The ‘class struggles’ of antiquity—to the extent
interests’ so frequently found these days, and that they were genuine class struggles and not
which has found its most classic expression in the struggles between status groups—were initially
statement of a talented author, that the individual carried on by indebted peasants, and perhaps
may be in error concerning his interests but that also by artisans threatened by debt bondage
the ‘class’ is ‘infallible’ about its interests. Yet, if and struggling against urban creditors. For debt
classes as such are not communities, nevertheless bondage is the normal result of the differenti-
class situations emerge only on the basis of com- ation of wealth in commercial cities, especially
munalization. The communal action that brings in seaport cities. A similar situation has existed
forth class situations, however, is not basically among cattle breeders. Debt relationships as such
action between members of the identical class; it produced class action up to the time of Cataline.
is an action between members of different classes. Along with this, and with an increase in provi-
Communal actions that directly determine the sion of grain for the city by transporting it from
class situation of the worker and the entrepreneur the outside, the struggle over the means of sus-
are: the labor market, the commodities market, tenance emerged. It centered in the first place
and the capitalistic enterprise. But, in its turn, the around the provision of bread and the determi-
existence of a capitalistic enterprise presupposes nation of the price of bread. It lasted through-
that a very specific communal action exists and out antiquity and the entire Middle Ages. The
that it is specifically structured to protect the pos- propertyless as such flocked together against
session of goods per se, and especially the power those who actually and supposedly were inter-
of individuals to dispose, in principle freely, over ested in the dearth of bread. This fight spread
the means of production. The existence of a cap- until it involved all those commodities essential
italistic enterprise is preconditioned by a specific to the way of life and to handicraft production.
 Weber—Class, Status, Party • 169

There were only incipient discussions of wage economically determined ‘class situation’ we
disputes in antiquity and in the Middle Ages. wish to designate as ‘status situation’ every typ-
But they have been slowly increasing up into ical component of the life fate of men that is
modern times. In the earlier periods they were determined by a specific, positive or negative,
completely secondary to slave rebellions as well social estimation of honor. This honor may be
as to fights in the commodity market. connected with any quality shared by a plurality,
The propertyless of antiquity and of the and, of course, it can be knit to a class situation:
Middle Ages protested against monopolies, class distinctions are linked in the most varied
pre-emption, forestalling, and the withholding ways with status distinctions. Property as such
of goods from the market in order to raise prices. is not always recognized as a status qualification,
Today the central issue is the determination of but in the long run it is, and with extraordinary
the price of labor. regularity. In the subsistence economy of the
This transition is represented by the fight for organized neighborhood, very often the richest
access to the market and for the determination man is simply the chieftain. However, this often
of the price of products. Such fights went on be- means only an honorific preference. For exam-
tween merchants and workers in the putting-out ple, in the so-called pure modern ‘democracy,’
system of domestic handicraft during the transi- that is, one devoid of any expressly ordered status
tion to modern times. Since it is quite a general privileges for individuals, it may be that only the
phenomenon we must mention here that the families coming under approximately the same
class antagonisms that are conditioned through tax class dance with one another. This example
the market situation are usually most bitter be- is reported of certain smaller Swiss cities. But
tween those who actually and directly partici- status honor need not necessarily be linked with
pate as opponents in price wars. It is not the a ‘class situation.’ On the contrary, it normally
rentier, the share-holder, and the banker who stands in sharp opposition to the pretensions of
suffer the ill will of the worker, but almost ex- sheer property.
clusively the manufacturer and the business ex- Both propertied and propertyless people can
ecutives who are the direct opponents of workers belong to the same status group, and frequently
in price wars. This is so in spite of the fact that they do with very tangible consequences. This
it is precisely the cash boxes of the rentier, the ‘equality’ of social esteem may, however, in the
share-holder, and the banker into which the long run become quite precarious. The ‘equality’
more or less ‘unearned’ gains flow, rather than of status among the American ‘gentlemen,’ for
into the pockets of the manufacturers or of the instance, is expressed by the fact that outside the
business executives. This simple state of affairs subordination determined by the different func-
has very frequently been decisive for the role the tions of ‘business,’ it would be considered strictly
class situation has played in the formation of repugnant—wherever the old tradition still pre-
political parties. For example, it has made pos- vails—if even the richest ‘chief,’ while playing bil-
sible the varieties of patriarchal socialism and liards or cards in his club in the evening, would
the frequent attempts—formerly, at least—of not treat his ‘clerk’ as in every sense fully his equal
threatened status groups to form alliances with in birthright. It would be repugnant if the Ameri-
the proletariat against the ‘bourgeoisie.’ can ‘chief ’ would bestow upon his ‘clerk’ the con-
descending ‘benevolence’ marking a distinction
of ‘position,’ which the German chief can never
Status Honor dissever from his attitude. This is one of the most
In contrast to classes, status groups are nor- important reasons why in America the German
mally communities. They are, however, often of ‘clubby-ness’ has never been able to attain the at-
an amorphous kind. In contrast to the purely traction that the American clubs have.
170 • Part III: The Structure of Inequality

of status is essentially a question of stratification

Guarantees of Status Stratification resting upon usurpation. Such usurpation is the
In content, status honor is normally expressed normal origin of almost all status honor. But
by the fact that above all else a specific style of the road from this purely conventional situation
life can be expected from all those who wish to to legal privilege, positive or negative, is easily
belong to the circle. Linked with this expectation traveled as soon as a certain stratification of the
are restrictions on ‘social’ intercourse (that is, in- social order has in fact been ‘lived in’ and has
tercourse which is not subservient to economic achieved stability by virtue of a stable distribu-
or any other of business’s ‘functional’ purposes). tion of economic power.
These restrictions may confine normal marriages
to within the status circle and may lead to com-
plete endogamous closure. As soon as there is not ‘Ethnic’ Segregation and ‘Caste’
a mere individual and socially irrelevant imita- Where the consequences have been realized to
tion of another style of life, but an agreed-upon their full extent, the status group evolves into a
communal action of this closing character, the closed ‘caste.’ Status distinctions are then guar-
‘status’ development is under way. anteed not merely by conventions and laws,
In its characteristic form, stratification by but also by rituals. This occurs in such a way
‘status groups’ on the basis of conventional styles that every physical contact with a member of
of life evolves at the present time in the United any caste that is considered to be ‘lower’ by the
States out of the traditional democracy. For ex- members of a ‘higher’ caste is considered as mak-
ample, only the resident of a certain street (‘the ing for a ritualistic impurity and to be a stigma
street’) is considered as belonging to ‘society,’ is which must be expiated by a religious act. In-
qualified for social intercourse, and is visited and dividual castes develop quite distinct cults and
invited. Above all, this differentiation evolves in gods.
such a way as to make for strict submission to In general, however, the status structure
the fashion that is dominant at a given time in reaches such extreme consequences only where
society. This submission to fashion also exists there are underlying differences which are held
among men in America to a degree unknown to be ‘ethnic.’ The ‘caste’ is, indeed, the normal
in Germany. Such submission is considered to form in which ethnic communities usually live
be an indication of the fact that a given man side by side in a ‘societalized’ manner. These
pretends to qualify as a gentleman. This submis- ethnic communities believe in blood relation-
sion decides, at least prima facie, that he will be ship and exclude exogamous marriage and social
treated as such. And this recognition becomes intercourse. Such a caste situation is part of the
just as important for his employment chances in phenomenon of ‘pariah’ peoples and is found all
‘swank’ establishments, and above all, for social over the world. These people form communities,
intercourse and marriage with ‘esteemed’ fami- acquire specific occupational traditions of hand-
lies, as the qualification for dueling among Ger- icrafts or of other arts, and cultivate a belief in
mans in the Kaiser’s day. As for the rest: certain their ethnic community. They live in a ‘diaspora’
families resident for a long time, and, of course, strictly segregated from all personal intercourse,
correspondingly wealthy, e.g. ‘F. F. V., i.e. First except that of an unavoidable sort, and their
Families of Virginia,’ or the actual or alleged de- situation is legally precarious. Yet, by virtue of
scendants of the ‘Indian Princess’ Pocahontas, of their economic indispensability, they are toler-
the Pilgrim fathers, or of the Knickerbockers, the ated, indeed, frequently privileged, and they live
members of almost inaccessible sects and all sorts in interspersed political communities. The Jews
of circles setting themselves apart by means of are the most impressive historical example.
any other characteristics and badges . . . all these A ‘status’ segregation grown into a ‘caste’
elements usurp ‘status’ honor. The development differs in its structure from a mere ‘ethnic’
 Weber—Class, Status, Party • 171

segregation: the caste structure transforms the out the hidden honor of the pariah people. This
horizontal and unconnected coexistences of eth- simple state of affairs, and not the ‘resentment’
nically segregated groups into a vertical social which is so strongly emphasized in Nietzsche’s
system of super- and subordination. Correctly much admired construction in the Genealogy of
formulated: a comprehensive societalization in- Morals, is the source of the religiosity cultivated
tegrates the ethnically divided communities into by pariah status groups. In passing, we may note
specific political and communal action. In their that resentment may be accurately applied only
consequences they differ precisely in this way: to a limited extent; for one of Nietzsche’s main
ethnic coexistences condition a mutual repulsion examples, Buddhism, it is not at all applicable.
and disdain but allow each ethnic community to Incidentally, the development of status
consider its own honor as the highest one; the groups from ethnic segregations is by no means
caste structure brings about a social subordina- the normal phenomenon. On the contrary, since
tion and an acknowledgment of ‘more honor’ in objective ‘racial differences’ are by no means
favor of the privileged caste and status groups. basic to every subjective sentiment of an ethnic
This is due to the fact that in the caste struc- community, the ultimately racial foundation
ture ethnic distinctions as such have become of status structure is rightly and absolutely a
‘functional’ distinctions within the political so- question of the concrete individual case. Very
cietalization (warriors, priests, artisans that are frequently a status group is instrumental in the
politically important for war and for building, production of a thoroughbred anthropological
and so on). But even pariah people who are most type. Certainly a status group is to a high degree
despised are usually apt to continue cultivating effective in producing extreme types, for they
in some manner that which is equally peculiar to select personally qualified individuals (e.g. the
ethnic and to status communities: the belief in Knighthood selects those who are fit for warfare,
their own specific ‘honor.’ This is the case with physically and psychically). But selection is far
the Jews. from being the only, or the predominant, way in
Only with the negatively privileged status which status groups are formed: Political mem-
groups does the ‘sense of dignity’ take a specific bership or class situation has at all times been at
deviation. A sense of dignity is the precipitation least as frequently decisive. And today the class
in individuals of social honor and of conven- situation is by far the predominant factor, for of
tional demands which a positively privileged course the possibility of a style of life expected
status group raises for the deportment of its for members of a status group is usually condi-
members. The sense of dignity that characterizes tioned economically.
positively privileged status groups is naturally
related to their ‘being’ which does not transcend
itself, that is, it is to their ‘beauty and excellence.’ Status Privileges
Their kingdom is ‘of this world.’ They live for For all practical purposes, stratification by sta-
the present and by exploiting their great past. tus goes hand in hand with a monopolization
The sense of dignity of the negatively privileged of ideal and material goods or opportunities, in
strata naturally refers to a future lying beyond a manner we have come to know as typical. Be-
the present, whether it is of this life or of an- sides the specific status honor, which always rests
other. In other words, it must be nurtured by the upon distance and exclusiveness, we find all sorts
belief in a providential ‘mission’ and by a belief of material monopolies. Such honorific prefer-
in a specific honor before God. The ‘chosen peo- ences may consist of the privilege of wearing spe-
ple’s’ dignity is nurtured by a belief either that cial costumes, of eating special dishes taboo to
in the beyond ‘the last will be the first,’ or that others, of carrying arms—which is most obvious
in this life a Messiah will appear to bring forth in its consequences—the right to pursue certain
into the light of the world which has cast them non-professional dilettante artistic practices, e.g.
172 • Part III: The Structure of Inequality

to play certain musical instruments. Of course, his salon-like ‘studio’ and those forms of musical
material monopolies provide the most effective practice that are acceptable to the status group.
motives for the exclusiveness of a status group;
although, in themselves, they are rarely suffi-
cient, almost always they come into play to some Economic Conditions and Effects of
extent. Within a status circle there is the ques- Status Stratification
tion of intermarriage: the interest of the families The frequent disqualification of the gainfully
in the monopolization of potential bridegrooms employed as such is a direct result of the prin-
is at least of equal importance and is parallel to ciple of status stratification peculiar to the social
the interest in the monopolization of daughters. order, and of course, of this principle’s opposi-
The daughters of the circle must be provided for. tion to a distribution of power which is regulated
With an increased inclosure of the status group, exclusively through the market. These two fac-
the conventional preferential opportunities for tors operate along with various individual ones,
special employment grow into a legal monopoly which will be touched upon below.
of special offices for the members. Certain goods We have seen above that the market and its
become objects for monopolization by status processes ‘knows no personal distinctions’: ‘func-
groups. In the typical fashion these include ‘en- tional’ interests dominate it. It knows nothing
tailed estates’ and frequently also the possessions of ‘honor.’ The status order means precisely the
of serfs or bondsmen and, finally, special trades. reverse, viz.: stratification in terms of ‘honor’ and
This monopolization occurs positively when the of styles of life peculiar to status groups as such. If
status group is exclusively entitled to own and to mere economic acquisition and naked economic
manage them; and negatively when, in order to power still bearing the stigma of its extra-status
maintain its specific way of life, the status group origin could bestow upon anyone who has won
must not own and manage them. it the same honor as those who are interested in
The decisive role of a ‘style of life’ in status status by virtue of style of life claim for them-
‘honor’ means that status groups are the specific selves, the status order would be threatened at
bearers of all ‘conventions.’ In whatever way it its very root. This is the more so as, given equal-
may be manifest, all ‘stylization’ of life either ity of status honor, property per se represents an
originates in status groups or is at least conserved addition even if it is not overtly acknowledged
by them. Even if the principles of status con- to be such. Yet if such economic acquisition and
ventions differ greatly, they reveal certain typical power gave the agent any honor at all, his wealth
traits, especially among those strata which are would result in his attaining more honor than
most privileged. Quite generally, among privi- those who successfully claim honor by virtue of
leged status groups there is a status disqualifi- style of life. Therefore all groups having interests
cation that operates against the performance of in the status order react with special sharpness
common physical labor. This disqualification is precisely against the pretensions of purely eco-
now ‘setting in’ in America against the old tra- nomic acquisition. In most cases they react the
dition of esteem for labor. Very frequently every more vigorously the more they feel themselves
rational economic pursuit, and especially ‘entre- threatened. Calderon’s respectful treatment of
preneurial activity,’ is looked upon as a disqual- the peasant, for instance, as opposed to Shake-
ification of status. Artistic and literary activity speare’s simultaneous and ostensible disdain
is also considered as degrading work as soon as of the canaille illustrates the different way in
it is exploited for income, or at least when it is which a firmly structured status order reacts as
connected with hard physical exertion. An ex- compared with a status order that has become
ample is the sculptor working like a mason in economically precarious. This is an example of
his dusty smock as over against the painter in a state of affairs that recurs everywhere. Precisely
 Weber—Class, Status, Party • 173

because of the rigorous reactions against the kind of overt participation in economic acquisi-
claims of property per se, the ‘parvenu’ is never tion as absolutely stigmatizing.
accepted, personally and without reservation, by With some over-simplification, one might
the privileged status groups, no matter how com- thus say that ‘classes’ are stratified according to
pletely his style of life has been adjusted to theirs. their relations to the production and acquisition
They will only accept his descendants who have of goods; whereas ‘status groups’ are stratified
been educated in the conventions of their status according to the principles of their consumption
group and who have never besmirched its honor of goods as represented by special ‘styles of life.’
by their own economic labor. An ‘occupational group’ is also a status group.
As to the general effect of the status order, For normally, it successfully claims social honor
only one consequence can be stated, but it is a only by virtue of the special style of life which
very important one: the hindrance of the free may be determined by it. The differences be-
development of the market occurs first for those tween classes and status groups frequently over-
goods which status groups directly withheld lap. It is precisely those status communities most
from free exchange by monopolization. This strictly segregated in terms of honor (viz. the
monopolization may be effected either legally Indian castes) who today show, although within
or conventionally. For example, in many Hel- very rigid limits, a relatively high degree of indif-
lenic cities during the epoch of status groups, ference to pecuniary income. However, the Brah-
and also originally in Rome, the inherited estate mins seek such income in many different ways.
(as is shown by the old formula for indication As to the general economic conditions mak-
against spendthrifts) was monopolized just as ing for the predominance of stratification by
were the estates of knights, peasants, priests, and ‘status,’ only very little can be said. When the
especially the clientele of the craft and merchant bases of the acquisition and distribution of
guilds. The market is restricted, and the power goods are relatively stable, stratification by status
of naked property per se, which gives its stamp to is favored. Every technological repercussion and
‘class formation,’ is pushed into the background. economic transformation threatens stratification
The results of this process can be most varied. by status and pushes the class situation into the
Of course, they do not necessarily weaken the foreground. Epochs and countries in which the
contrasts in the economic situation. Frequently naked class situation is of predominant signifi-
they strengthen these contrasts, and in any case, cance are regularly the periods of technical and
where stratification by status permeates a com- economic transformations. And every slowing
munity as strongly as was the case in all politi- down of the shifting of economic stratifications
cal communities of antiquity and of the Middle leads, in due course, to the growth of status
Ages, one can never speak of a genuinely free structures and makes for a resuscitation of the
market competition as we understand it today. important role of social honor.
There are wider effects than this direct exclusion
of special goods from the market. From the con-
trariety between the status order and the purely Parties
economic order mentioned above, it follows that Whereas the genuine place of ‘classes’ is within
in most instances the notion of honor peculiar to the economic order, the place of ‘status groups’ is
status absolutely abhors that which is essential within the social order, that is, within the sphere
to the market: higgling. Honor abhors higgling of the distribution of ‘honor.’ From within these
among peers and occasionally it taboos higgling spheres, classes and status groups influence one
for the members of a status group in general. another and they influence the legal order and
Therefore, everywhere some status groups, and are in turn influenced by it. But ‘parties’ live in a
usually the most influential, consider almost any house of ‘power.’
174 • Part III: The Structure of Inequality

Their action is oriented toward the acquisi- specially modern forms of domination. We shall
tion of social ‘power,’ that is to say, toward in- also designate as parties the ancient and medie-
fluencing a communal action no matter what val ‘parties,’ despite the fact that their structure
its content may be. In principle, parties may differs basically from the structure of modern
exist in a social ‘club’ as well as in a ‘state.’ As parties. By virtue of these structural differences
over against the actions of classes and status of domination it is impossible to say anything
groups, for which this is not necessarily the case, about the structure of parties without discussing
the communal actions of ‘parties’ always mean the structural forms of social domination per se.
a societalization. For party actions are always Parties, which are always structures struggling
directed toward a goal which is striven for in for domination, are very frequently organized in
planned manner. This goal may be a ‘cause’ (the a very strict ‘authoritarian’ fashion. . . .
party may aim at realizing a program for ideal or Concerning ‘classes,’ ‘status groups,’ and ‘par-
material purposes), or the goal may be ‘personal’ ties,’ it must be said in general that they necessar-
(sinecures, power, and from these, honor for the ily presuppose a comprehensive societalization,
leader and the followers of the party). Usually and especially a political framework of commu-
the party action aims at all these simultaneously. nal action, within which they operate. This does
Parties are, therefore, only possible within com- not mean that parties would be confined by the
munities that are societalized, that is, which have frontiers of any individual political community.
some rational order and a staff of persons avail- On the contrary, at all times it has been the order
able who are ready to enforce it. For parties aim of the day that the societalization (even when
precisely at influencing this staff, and if possible, it aims at the use of military force in common)
to recruit it from party followers. reaches beyond the frontiers of politics. This has
In any individual case, parties may represent been the case in the solidarity of interests among
interests determined through ‘class situation’ or the Oligarchs and among the democrats in Hel-
‘status situation,’ and they may recruit their fol- las, among the Guelfs and among Ghibellines in
lowing respectively from one or the other. But the Middle Ages, and within the Calvinist party
they need be neither purely ‘class’ nor purely during the period of religious struggles. It has
‘status’ parties. In most cases they are partly class been the case up to the solidarity of the land-
parties and partly status parties, but sometimes lords (international congress of agrarian land-
they are neither. They may represent ephemeral lords), and has continued among princes (holy
or enduring structures. Their means of attaining alliance, Karlsbad decrees), socialist workers,
power may be quite varied, ranging from naked conservatives (the longing of Prussian conserva-
violence of any sort to canvassing for votes with tives for Russian intervention in 1850). But their
coarse or subtle means: money, social influence, aim is not necessarily the establishment of new
the force of speech, suggestion, clumsy hoax, and international political, i.e. territorial, dominion.
so on to the rougher or more artful tactics of In the main they aim to influence the existing
obstruction in parliamentary bodies. dominion.2
The sociological structure of parties differs in
a basic way according to the kind of communal NOTES
action which they struggle to influence. Par- 1. Wirtschaft und Gesellschaft, part III, chap. 4,
pp. 631–40. The first sentence in paragraph one
ties also differ according to whether or not the
and the several definitions in this chapter which are
community is stratified by status or by classes.
in brackets do not appear in the original text. They
Above all else, they vary according to the struc- have been taken from other contexts of Wirtschaft
ture of domination within the community. For und Gesellschaft.
their leaders normally deal with the conquest of 2. The posthumously published text breaks off
a community. They are, in the general concept here. We omit an incomplete sketch of types of ‘war-
which is maintained here, not only products of rior estates.’
 Weber—Status Groups and Classes • 175

Max Weber

Status Groups and Classes

The Concepts of Class and Class necessarily happen. The concepts of class and
Situation class situation as such designate only the fact
The term ‘class situation’1 will be applied to of identity or similarity in the typical situation
the typical probability that a given state of (a) in which a given individual and many others
provision with goods, (b) external conditions of find their interests defined. In principle control
life, and (c) subjective satisfaction or frustration over different combinations of consumer goods,
will be possessed by an individual or a group. means of production, investments, capital funds
These probabilities define class situation in so or marketable abilities constitute class situations
far as they are dependent on the kind and extent which are different with each variation and
of control or lack of it which the individual has combination. Only persons who are completely
over goods or services and existing possibilities of unskilled, without property and dependent on
their exploitation for the attainment of income employment without regular occupation, are in a
or receipts within a given economic order. strictly identical class situation. Transitions from
A ‘class’ is any group of persons occupying one class situation to another vary greatly in flu-
the same class situation. The following types of idity and in the ease with which an individual
classes may be distinguished: (a) A class is a ‘prop- can enter the class. Hence the unity of ‘social’
erty class’ when class situation for its members is classes is highly relative and variable.
primarily determined by the differentiation of
property holdings; (b) a class is an ‘acquisition
class’ when the class situation of its members is The Significance of Property Classes
primarily determined by their opportunity for The primary significance of a positively privi-
the exploitation of services on the market; (c) leged property class lies in the following facts:
the ‘social class’ structure is composed of the (i) Its members may be able to monopolize the
plurality of class situations between which an purchase of high-priced consumer goods. (ii)
interchange of individuals on a personal basis or They may control the opportunities of pursu-
in the course of generations is readily possible ing a systematic monopoly policy in the sale of
and typically observable. On the basis of any economic goods. (iii) They may monopolize
of the three types of class situation, associative opportunities for the accumulation of property
relationships between those sharing the same through unconsumed surpluses. (iv) They may
class interests, namely, corporate class organi- monopolize opportunities to accumulate capi-
zations may develop. This need not, however, tal by saving, hence, the possibility of investing

Max Weber. “Status Groups and Classes,” in The Theory of Social and Economic Organization, edited by Talcott
Parsons, translated by A. M. Henderson and Talcott Parsons, pp. 424–429. Copyright © 1947, copyright ©
renewed 1975 by Talcott Parsons. Reprinted with the permission of the Free Press of Simon and Schuster, Inc.
176 • Part III: The Structure of Inequality

property in loans and the related possibility of may, on the contrary, be concerned in the first
control over executive positions in business. (v) instance only with a redistribution of wealth.
They may monopolize the privileges of socially These may be called ‘property revolutions.’
advantageous kinds of education so far as these A classic example of the lack of class antag-
involve expenditures. onism has been the relation of the ‘poor white
Positively privileged property classes typically trash,’ originally those not owning slaves, to the
live from property income. This may be derived planters in the Southern States of the United
from property rights in human beings, as with States. The ‘poor whites’ have often been much
slaveowners, in land, in mining property, in fixed more hostile to the Negro than the planters who
equipment such as plant and apparatus, in ships, have frequently had a large element of patriar-
and as creditors in loan relationships. Loans may chal sentiment. The conflict of outcast against
consist of domestic animals, grain, or money. Fi- the property classes, of creditors and debtors,
nally they may live on income from securities. and of landowners and outcasts are best illus-
Class interests which are negatively privileged trated in the history of Antiquity.
with respect to property belong typically to one
of the following types: (a) They are themselves
objects of ownership, that is they are unfree. (b) The Significance of Acquisition and
They are ‘outcasts,’ that is ‘proletarians’ in the Social Classes
sense meant in Antiquity. (c) They are debtor The primary significance of a positively priv-
classes and, (d) the ‘poor.’ ileged acquisition class is to be found in two
In between stand the ‘middle’ classes. This directions. On the one hand it is generally possi-
term includes groups who have all sorts of prop- ble to go far toward attaining a monopoly of the
erty, or of marketable abilities through training, management of productive enterprises in favour
who are in a position to draw their support from of the members of the class and their business
these sources. Some of them may be ‘acquisition’ interests. On the other hand, such a class tends
classes. Entrepreneurs are in this category by vir- to insure the security of its economic position by
tue of essentially positive privileges; proletarians, exercising influence on the economic policy of
by virtue of negative privileges. But many types political bodies and other groups.
such as peasants, craftsmen, and officials do not The members of positively privileged acquisi-
fall in this category. tion classes are typically entrepreneurs. The fol-
The differentiation of classes on the basis lowing are the most important types: merchants,
of property alone is not ‘dynamic,’ that is, it shipowners, industrial and agricultural entrepre-
does not necessarily result in class struggles or neurs, bankers and financiers. Under certain
class revolutions. It is not uncommon for very circumstances two other types are also members
strongly privileged property classes, such as of such classes, namely, members of the ‘liberal’
slaveowners, to exist side by side with such far professions with a privileged position by virtue
less privileged groups as peasants or even out- of their abilities or training, and workers with
casts without any class struggle. There may even special skills commanding a monopolistic posi-
be ties of solidarity between privileged property tion, regardless of how far they are hereditary or
classes and unfree elements. However, such con- the result of training.
flicts as that between land owners and outcast Acquisition classes in a negatively privileged
elements or between creditors and debtors, the situation are workers of the various principal
latter often being a question of urban patricians types. They may be roughly classified as skilled,
as opposed to either rural peasants or urban semi-skilled and unskilled.
craftsmen, may lead to revolutionary conflict. In this connexion as well as the above, inde-
Even this, however, need not necessarily aim pendent peasants and craftsmen are to be treated
at radical changes in economic organization. It as belonging to the ‘middle classes.’ This category
 Weber—Status Groups and Classes • 177

often includes in addition officials, whether they Organized activity of class groups is favoured
are in public or private employment, the liberal by the following circumstances: (a) The possi-
professions, and workers with exceptional mo- bility of concentrating on opponents where the
nopolistic assets or positions. immediate conflict of interests is vital. Thus
Examples of ‘social classes’ are: (a) The ‘work- workers organize against management and not
ing’ class as a whole. It approaches this type the against security holders who are the ones who
more completely mechanized the productive really draw income without working. Simi-
process becomes. (b) The petty bourgeoisie.2 larly peasants are not apt to organize against
(c) The ‘intelligentsia’ without independent landlords. (b) The existence of a class situation
property and the persons whose social position which is typically similar for large masses of peo-
is primarily dependent on technical training ple. (c) The technical possibility of being easily
such as engineers, commercial and other offi- brought together. This is particularly true where
cials, and civil servants. These groups may differ large numbers work together in a small area, as
greatly among themselves, in particular accord- in the modern factory. (d) Leadership directed
ing to costs of training. (d) The classes occupy- to readily understandable goals. Such goals are
ing a privileged position through property and very generally imposed or at least are interpreted
education. by persons, such as intelligentsia, who do not
The unfinished concluding section of Karl belong to the class in question.
Marx’s Kapital was evidently intended to deal
with the problem of the class unity of the pro-
letariat, which he held existed in spite of the Status and Status Group
high degree of qualitative differentiation. A de- The term of ‘status’3 will be applied to a typically
cisive factor is the increase in the importance of effective claim to positive or negative privilege
semi-skilled workers who have been trained in with respect to social prestige so far as it rests on
a relatively short time directly on the machines one or more of the following bases: (a) mode of
themselves, at the expense of the older type of living, (b) a formal process of education which
‘skilled’ labour and also of unskilled. However, may consist in empirical or rational training
even this type of skill may often have a mo- and the acquisition of the corresponding modes
nopolistic aspect. Weavers are said to attain the of life, or (c) on the prestige of birth, or of an
highest level of productivity only after five years’ occupation.
experience. The primary practical manifestations of sta-
At an earlier period every worker could be tus with respect to social stratification are conu-
said to have been primarily interested in be- bium, commensality, and often monopolistic
coming an independent small bourgeois, but appropriation of privileged economic opportu-
the possibility of realizing this goal is becom- nities and also prohibition of certain modes of
ing progressively smaller. From one generation acquisition. Finally, there are conventions or tra-
to another the most readily available path to ditions of other types attached to a status.
advancement both for skilled and semi-skilled Status may be based on class situation di-
workers is into the class of technically trained rectly or related to it in complex ways. It is not,
individuals. In the most highly privileged classes, however, determined by this alone. Property and
at least over the period of more than one genera- managerial positions are not as such sufficient to
tion, it is coming more and more to be true that lend their holder a certain status, though they
money is overwhelmingly decisive. Through the may well lead to its acquisition. Similarly, pov-
banks and corporate enterprises members of the erty is not as such a disqualification for high sta-
lower middle class and the salaried groups have tus though again it may influence it.
certain opportunities to rise into the privileged Conversely, status may partly or even wholly
class. determine class situation, without, however,
178 • Part III: The Structure of Inequality

being identical with it. The class situation of an whereas status groups develop and subsist most
officer, a civil servant, and a student as deter- readily where economic organization is of a mo-
mined by their income may be widely different nopolistic and liturgical character and where the
while their status remains the same, because they economic needs of corporate groups are met on
adhere to the same mode of life in all relevant a feudal or patrimonial basis. The type of class
respects as a result of their common education. which is most closely related to a status group
A ‘status group’ is a plurality of individuals is the ‘social’ class, while the ‘acquisition’ class is
who, within a larger group, enjoy a particular the farthest removed. Property classes often con-
kind and level of prestige by virtue of their po- stitute the nucleus of a status group.
sition and possibly also claim certain special Every society where status groups play a
monopolies. prominent part is controlled to a large extent
The following are the most important sources by conventional rules of conduct. It thus cre-
of the development of distinct status groups: (a) ates economically irrational conditions of con-
The most important is by the development of a sumption and hinders the development of free
peculiar style of life including, particularly, the markets by monopolistic appropriation and by
type of occupation pursued. (b) The second basis restricting free disposal of the individual’s own
is hereditary charisma arising from the successful economic ability. This will have to be discussed
claim to a position of prestige by virtue of birth. further elsewhere.
(c) The third is the appropriation of political or
hierocratic authority as a monopoly by socially NOTES
distinct groups. 1. Although Parsons chooses to translate Klasse
as ‘class status’ in this context, to do so is potentially
The development of hereditary status groups
confusing because Weber so carefully distinguishes
is usually a form of the hereditary appropriation
between the concepts of class and status. I have
of privileges by an organized group or by indi- therefore followed the lead of Roth and Wittich
vidual qualified persons. Every well-established (Economy and Society, 1968) and opted for the term
case of appropriation of opportunities and abil- ‘class situation’ throughout this essay.—Ed.
ities, especially of exercising imperative powers, 2. I have again followed Roth and Wittich (Econ-
has a tendency to lead to the development of dis- omy and Society, 1968) in translating the German
tinct status groups. Conversely, the development term Kleinbürgertum as ‘petty bourgeoisie,’ whereas
of status groups has a tendency in turn to lead Parsons opted for the more ambiguous term ‘lower
to the monopolistic appropriation of govern- middle’ class.—Ed.
3. For the purposes of consistency with the other
ing powers and of the corresponding economic
selections, I have translated the term ständische Lage
advantages. as ‘status’ (see Roth and Wittich, Economy and Soci-
Acquisition classes are favoured by an eco- ety, 1968), whereas Parsons opted for the terms ‘so-
nomic system oriented to market situations, cial status,’ ‘stratifactory status,’ and the like.—Ed.
 Weber—Open and Closed Relationships • 179

Max Weber

Open and Closed Relationships

Social Relationships There are various ways in which it is possible

A social relationship, regardless of whether it is for a closed social relationship to guarantee its
communal or associative in character, will be monopolized advantages to the parties. (a) Such
spoken of as “open” to outsiders if and insofar as advantages may be left free to competitive strug-
its system of order does not deny participation gle within the group, (b) they may be regulated
to anyone who wishes to join and is actually in or rationed in amount and kind, or (c) they may
a position to do so. A relationship will, on the be appropriated by individuals or sub-groups
other hand, be called “closed” against outsiders on a permanent basis and become more or less
so far as, according to its subjective meaning and inalienable. The last is a case of closure within,
its binding rules, participation of certain persons as well as against outsiders. Appropriated ad-
is excluded, limited, or subjected to conditions. vantages will be called “rights.” As determined
Whether a relationship is open or closed may by the relevant order, appropriation may be
be determined traditionally, affectually, or ra- (1) for the benefit of the members of particular
tionally in terms of values or of expediency. It communal or associative groups (for instance,
is especially likely to be closed, for rational rea- household groups), or (2) for the benefit of in-
sons, in the following type of situation: a social dividuals. In the latter case, the individual may
relationship may provide the parties to it with enjoy his rights on a purely personal basis or in
opportunities for the satisfaction of spiritual such a way that in case of his death one or more
or material interests, whether absolutely or in- other persons related to the holder of the right
strumentally, or whether it is achieved through by birth (kinship), or by some other social rela-
co-operative action or by a compromise of inter- tionship, may inherit the rights in question. Or
ests. If the participants expect that the admission the rights may pass to one or more individuals
of others will lead to an improvement of their specifically designated by the holder. These are
situation, an improvement in degree, in kind, in cases of hereditary appropriation. Finally, (3)
the security or the value of the satisfaction, their it may be that the holder is more or less fully
interest will be in keeping the relationship open. empowered to alienate his rights by voluntary
If, on the other hand, their expectations are of agreement, either to other specific persons or to
improving their position by monopolistic tactics, anyone he chooses. This is alienable appropria-
their interest is in a closed relationship. tion. A party to a closed social relationship will

Max Weber. “Open and Closed Relationships,” in Economy and Society: An Outline of Interpretive Sociology,
Vol. 1, edited by Guenther Roth and Claus Wittich, translated by Ephraim Rischoff, Hans Gerth, A. M. Hen-
derson, et al., pp. 43–46, 341–342, 344. Copyright © 1978 by the Regents of the University of California.
Published by the University of California Press.
180 • Part III: The Structure of Inequality

be called a “member”; in case his participation their monopolistic position. The same phenom-
is regulated in such a way as to guarantee him enon is not uncommon in monastic orders and
appropriated advantages, a privileged member religious sects which have passed from a stage of
(Rechtsgenosse). Appropriated rights which religious proselytizing to one of restriction in the
are enjoyed by individuals through inheritance interest of the maintenance of an ethical stan-
or by hereditary groups, whether communal or dard or for the protection of material interests.
associative, will be called the “property” of the There is a similar close relationship between the
individual or of groups in question; and, insofar extension of market relationships in the interest
as they are alienable, “free” property. of increased turnover on the one hand, their mo-
The apparently gratuitous tediousness in- nopolistic restriction on the other. The promo-
volved in the elaborate definition of the above tion of linguistic uniformity is today a natural
concepts is an example of the fact that we often result of the interests of publishers and writers, as
neglect to think out clearly what seems to be ob- opposed to the earlier, not uncommon, tendency
vious, because it is intuitively familiar. for status groups to maintain linguistic peculiari-
ties or even for secret languages to emerge.
1. a. Examples of communal relationships, 2. Both the extent and the methods of reg-
which tend to be closed on a traditional ulation and exclusion in relation to outsiders
basis, are those in which membership is de- may vary widely, so that the transition from a
termined by family relationship. state of openness to one of regulation and clo-
b. Personal emotional relationships are usually sure is gradual. Various conditions of partic-
affectually closed. Examples are erotic rela- ipation may be laid down; qualifying tests, a
tionships and, very commonly, relations of period of probation, requirement of possession
personal loyalty. of a share which can be purchased under certain
c. Closure on the basis of value-rational com- conditions, election of new members by ballot,
mitment to values is usual in groups sharing membership or eligibility by birth or by virtue
a common system of explicit religious belief. of achievements open to anyone. Finally, in case
d. Typical cases of rational closure on grounds of closure and the appropriation of rights within
of expediency are economic associations of the group, participation may be dependent on
a monopolistic or a plutocratic character. the acquisition of an appropriated right. There is
A few examples may be taken at random. a wide variety of different degrees of closure and
Whether a group of people engaged in con- of conditions of participation. Thus regulation
versation is open or closed depends on its con- and closure are relative concepts. There are all
tent. General conversation is apt to be open, as manner of gradual shadings as between an exclu-
contrasted with intimate conversation or the sive club, a theatrical audience the members of
imparting of official information. Market rela- which have purchased tickets, and a party rally
tionships are in most, or at least in many, cases to which the largest possible number has been
essentially open. In the case of many relation- urged to come; similarly, from a church service
ships, both communal and associative, there is open to the general public through the rituals
a tendency to shift from a phase of expansion of a limited sect to the mysteries of a secret cult.
to one of exclusiveness. Examples are the guilds 3. Similarly, closure within the group may
and the democratic city-states of Antiquity and also assume the most varied forms. Thus a caste,
the Middle Ages. At times these groups sought a guild, or a group of stock exchange brokers,
to increase their membership in the interest of which is closed to outsiders, may allow to its
improving the security of their position of power members a perfectly free competition for all the
by adequate numbers. At other times they re- advantages which the group as a whole monop-
stricted their membership to protect the value of olizes for itself. Or it may assign every member
 Weber—Open and Closed Relationships • 181

strictly to the enjoyment of certain advantages,

such as claims over customers or particular busi- Economic Relationships
ness opportunities, for life or even on a hereditary One frequent economic determinant [of closure]
basis. This is particularly characteristic of India. is the competition for a livelihood—offices, cli-
Similarly, a closed group of settlers (Markgenos- ents and other remunerative opportunities.
senschaft) may allow its members free use of the When the number of competitors increases in re-
resources of its area or may restrict them rigidly lation to the profit span, the participants become
to a plot assigned to each individual household. interested in curbing competition. Usually one
A closed group of colonists may allow free use of group of competitors takes some externally iden-
the land or sanction and guarantee permanent tifiable characteristic of another group of (actual
appropriation of separate holdings. In such cases or potential) competitors—race, language, reli-
all conceivable transitional and intermediate gion, local or social origin, descent, residence,
forms can be found. Historically, the closure of etc.—as a pretext for attempting their exclusion.
eligibility to fiefs, benefices, and offices within It does not matter which characteristic is chosen
the group, and the appropriation on the part of in the individual case: whatever suggests itself
those enjoying them, have occurred in the most most easily is seized upon. Such group action
varied forms. Similarly, the establishment of may provoke a corresponding reaction on the
rights to and possession of particular jobs on the part of those against whom it is directed.
part of workers may develop all the way from the In spite of their continued competition
“closed shop” to a right to a particular job. The against one another, the jointly acting com-
first step in this development may be to prohibit petitors now form an “interest group” toward
the dismissal of a worker without the consent of outsiders; there is a growing tendency to set up
the workers’ representatives. The development of some kind of association with rational regula-
the “works councils” [in Germany after 1918] tions; if the monopolistic interests persist, the
might be a first step in this direction, though it time comes when the competitors, or another
need not be. . . . group whom they can influence (for example, a
4. The principal motives for closure of a political community), establish a legal order that
relationship are: (a) The maintenance of qual- limits competition through formal monopolies;
ity, which is often combined with the interest from then on, certain persons are available as
in prestige and the consequent opportunities “organs” to protect the monopolistic practices,
to enjoy honor, and even profit; examples are if need be, with force. In such a case, the in-
communities of ascetics, monastic orders, es- terest group has developed into a “legally priv-
pecially, for instance, the Indian mendicant or- ileged group” (Rechtsgemeinschaft) and the
ders, religious sects like the Puritans, organized participants have become “privileged members”
groups of warriors, of ministeriales and other (Rechtsgenossen). Such closure, as we want
functionaries, organized citizen bodies as in the to call it, is an ever-recurring process; it is the
Greek states, craft guilds; (b) the contraction of source of property in land as well as of all guild
advantages in relation to consumption needs and other group monopolies.
(Nahrungsspielraum); examples are monopolies The tendency toward the monopolization
of consumption, the most developed form of of specific, usually economic opportunities is
which is a self-subsistent village community; (c) always the driving force in such cases as: “coop-
the growing scarcity of opportunities for acqui- erative organization,” which always means closed
sition (Erwerbsspielraum). This is found in trade monopolistic groups, for example, of fishermen
monopolies such as guilds, the ancient monop- taking their name from a certain fishing area;
olies of fishing rights, and so on. Usually motive the establishment of an association of engineer-
(a) is combined with (b) or (c). . . . ing graduates, which seeks to secure a legal, or
182 • Part III: The Structure of Inequality

at least factual, monopoly over certain positions; (2) have proven their qualification, and (3)
the exclusion of outsiders from sharing in the sometimes have passed through further waiting
fields and commons of a village; “patriotic” asso- periods and met additional requirements. This
ciations of shop clerks; the ministeriales, knights, development follows a typical pattern in groups
university graduates and craftsmen of a given re- ranging from the juvenile student fraternities,
gion or locality; ex-soldiers entitled to civil ser- through knightly associations and craft-guilds,
vice positions—all these groups first engage in to the qualifications required of the modern of-
some joint action (Gemeinschaftshandeln) and ficials and employees. It is true that the interest
later perhaps an explicit association. This mo- in guaranteeing an efficient performance may
nopolization is directed against competitors who everywhere have some importance; the partici-
share some positive or negative characteristics; its pants may desire it for idealistic or materialis-
purpose is always the closure of social and eco- tic reasons in spite of their possibly continuing
nomic opportunities to outsiders. Its extent may competition with one another: local craftsmen
vary widely, especially so far as the group mem- may desire it for the sake of their business rep-
ber shares in the apportionment of monopolistic utation, ministeriales and knights of a given
advantages. . . . association for the sake of their professional
This monopolistic tendency takes on specific reputation and also their own military security,
forms when groups are formed by persons with and ascetic groups for fear that the gods and de-
shared qualities acquired through upbringing, mons may turn their wrath against all members
apprenticeship and training. These character- because of faulty manipulations. (For example,
istics may be economic qualifications of some in almost all primitive tribes, persons who sang
kind, the holding of the same or of similar of- falsely during a ritual dance were originally slain
fices, a knightly or ascetic way of life, etc. If in in expiation of such an offense.) But normally
such a case an association results from social this concern for efficient performance recedes
action, it tends toward the guild. Full mem- behind the interest in limiting the supply of can-
bers make a vocation out of monopolizing the didates for the benefices and honors of a given
disposition of spiritual, intellectual, social and occupation. The novitiates, waiting periods,
economic goods, duties and positions. Only masterpieces and other demands, particularly
those are admitted to the unrestricted practice the expensive entertainment of group members,
of the vocation who (1) have completed a no- are more often economic than professional tests
vitiate in order to acquire the proper training, of qualification.
 Giddens—The Class Structure of the Advanced Societies • 183

19 • Anthony Giddens

The Class Structure of the Advanced Societies

The Weberian Critique two may vary independently. There is yet a third
For the most significant developments in the dimension, so the argument continues, which
theory of classes since Marx, we have to look Weber recognised as an independently variable
to those forms of social thought whose authors, factor in ‘stratification’, but which Marx treated
while being directly influenced by Marx’s ideas, as directly contingent upon class interests. This
have attempted at the same time to criticise or is the factor of ‘power’.2
to reformulate them. This tendency has been Evaluation of the validity of this interpreta-
strongest, for a combination of historical and tion is difficult because there is no doubt that
intellectual reasons, in German sociology, where Weber himself accepted it—or certain elements
a series of attempts have been made to provide a of it. What is often portrayed in the secondary
fruitful critique of Marx—beginning with Max literature as a critique of ‘Marx’s conception of
Weber, and continuing through such authors as class’ actually takes a stilted and impoverished
Geiger, Renner and Dahrendorf.1 Weber’s cri- form of crude Marxism as its main target of at-
tique of Marx here has been of particular im- tack. But this sort of determinist Marxism was
portance. But, especially in the English-speaking already current in Germany in Weber’s lifetime,
world, the real import of Weber’s analysis has and since Weber himself set out to question this
frequently been misrepresented. The customary determinism, the true lines of similarity and dif-
procedure has been to contrast Weber’s discus- ference between his and Marx’s analysis of classes
sion of ‘Class, status and party’, a fragment of are difficult to disentangle. . . . 3
Economy and Society, with the conception of class In the two versions of ‘Class, status and party’
supposedly taken by Marx, to the detriment of which have been embodied in Economy and So-
the latter. Marx, so it is argued, treated ‘class’ as ciety,4 Weber provides what is missing in Marx:
a purely economic phenomenon and, moreover, an explicit discussion of the concept of class.
regarded class conflicts as in some way the ‘in- There are two principal respects in which this
evitable’ outcome of clashes of material interest. analysis differs from Marx’s ‘abstract model’ of
He failed to realise, according to this argument, classes. One is that which is familiar from most
that the divisions of economic interest which secondary accounts—the differentiation of ‘class’
create classes do not necessarily correspond to from ‘status’ and ‘party’. The second, however, as
sentiments of communal identity which consti- will be argued below, is equally important: this is
tute differential ‘status’. Thus, status, which de- that, although Weber employs for some purposes
pends upon subjective evaluation, is a separate a dichotomous model which in certain general
‘dimension of stratification’ from class, and the respects resembles that of Marx, his viewpoint

Anthony Giddens. The Class Structure of the Advanced Societies, pp. 41–44, 47–49, 102–112. Copyright ©
1973 by Anthony Giddens. Reprinted by permission of HarperCollins Publishers and the author.
184 • Part III: The Structure of Inequality

strongly emphasises a pluralistic conception of capitalism—educational qualifications take on

classes. Thus Weber’s distinction between ‘own- a particular significance in this respect; but the
ership classes’ (Besitzklassen) and ‘acquisition monopolisation of trade skills by manual work-
classes’ (Erwerbsklassen) is based upon a fusion ers is also important.
of two criteria: ‘on the one hand . . . the kind of Weber insists that a clear-cut distinction must
property that is usable for returns; and, on the be made between class ‘in itself ’ and class ‘for
other hand . . . the kind of services that can be itself ’: ‘class’, in his terminology, always refers
offered on the market’, thus producing a com- to market interests, which exist independently
plex typology. The sorts of property which may of whether men are aware of them. Class is
be used to obtain market returns, although di- thus an ‘objective’ characteristic influencing the
viding generally into two types—creating own- life-chances of men. But only under certain con-
ership (rentier) and acquisition (entrepreneurial) ditions do those sharing a common class situa-
classes—are highly variable, and may produce tion become conscious of, and act upon, their
many differential interests within dominant mutual economic interests. In making this em-
classes: phasis, Weber undoubtedly intends to separate
his position from that adopted by many Marx-
Ownership of dwellings; workshops; ware- ists, involving what he calls a ‘pseudo-scientific
houses; stores; agriculturally usable land operation’ whereby the link between class and
in large or small-holdings—a quantitative class consciousness is treated as direct and imme-
difference with possibly qualitative conse- diate.6 Such a consideration evidently also un-
quences; ownership of mines; cattle; men derlies the emphasis which Weber places upon
(slaves); disposition over mobile instruments ‘status groups’ (Stände) as contrasted to classes.
of production, or capital goods of all sorts, The contrast between class and status group,
especially money or objects that can easily be however, is not, as often seems to be assumed,
exchanged for money; disposition over prod- merely, nor perhaps even primarily, a distinc-
ucts of one’s own labour or of others’ labour tion between subjective and objective aspects
differing according to their various distances of differentiation. While class is founded upon
from consumability; disposition over trans- differentials of economic interest in market re-
ferable monopolies of any kind—all these lationships, Weber nowhere denies that, under
distinctions differentiate the class situations certain given circumstances, a class may be a
of the propertied . . . 5 subjectively aware ‘community’. The importance
of status groups—which are normally ‘commu-
But the class situations of the propertyless are nities’ in this sense—derives from the fact that
also differentiated, in relation both to the types they are built upon criteria of grouping other
and the degree of ‘monopolisation’ of ‘market- than those stemming from market situation.
able skills’ which they possess. Consequently, The contrast between classes and status groups
there are various types of ‘middle class’ which is sometimes portrayed by Weber as one between
stand between the ‘positively privileged’ classes the objective and the subjective; but it is also one
(the propertied) and the ‘negatively privileged’ between production and consumption. Whereas
classes (those who possess neither property class expresses relationships involved in produc-
nor marketable skills). While these groupings tion, status groups express those involved in con-
are all nominally propertyless, those who pos- sumption, in the form of specific ‘styles of life’.
sess skills which have a definite ‘market value’ Status affiliations may cut across the relation-
are certainly in a different class situation from ships generated in the market, since membership
those who have nothing to offer but their (un- of a status group usually carries with it various
skilled) labour. In acquisition classes—i.e., those sorts of monopolistic privileges. Nonetheless,
associated particularly with the rise of modern classes and status groups tend in many cases to
 Giddens—The Class Structure of the Advanced Societies • 185

be closely linked, through property: possession class are not necessarily conscious of the fact).
of property is both a major determinant of class The notion of social class is important because
situation and also provides the basis for follow- it introduces a unifying theme into the diver-
ing a definite ‘style of life’. The point of Weber’s sity of cross-cutting class relationships which
analysis is not that class and status constitute may stem from Weber’s identification of ‘class
two ‘dimensions of stratification’, but that classes situation’ with ‘market position’. If the latter is
and status communities represent two possible, applied strictly, it is possible to distinguish an
and competing, modes of group formation in almost endless multiplicity of class situations.
relation to the distribution of power in society. But a ‘social class’ exists only when these class
Power is not, for Weber, a ‘third dimension’ in situations cluster together in such a way as to
some sense comparable to the first two. He is create a common nexus of social interchange
quite explicit about saying that classes, status between individuals. In capitalism, Weber dis-
groups and parties are all ‘phenomena of the tinguishes four main social class groupings: the
distribution of power’.7 The theorem inform- manual working class; the petty bourgeoisie;
ing Weber’s position here is his insistence that propertyless white-collar workers: ‘technicians,
power is not to be assimilated to economic dom- various kinds of white-collar employees, civil
ination—again, of course, a standpoint taken in servants—possibly with considerable social dif-
deliberate contrast to that of Marx. The party, ferences depending on the cost of their train-
oriented towards the acquisition or maintenance ing’; and those ‘privileged through property and
of political leadership, represents, like the class education’.8 Of these social class groupings, the
and the status group, a major focus of social or- most significant are the working class, the prop-
ganisation relevant to the distribution of power ertyless ‘middle class’ and the propertied ‘upper
in a society. It is, however, only characteristic of class’. Weber agrees with Marx that the category
the modern rational state. . . . of small property-owners (Kleinbürgertum) tends
In his conceptual discussion of class, besides to become progressively more restricted with the
distinguishing the purely economic Besitzklassen increasing maturity of capitalism. The result of
and Erwerbsklassen, Weber also refers to what this process, however, is not normally that they
he calls ‘social classes’. A social class, in Weber’s ‘sink into the proletariat’, but that they become
sense, is formed of a cluster of class situations absorbed into the expanding category of skilled
which are linked together by virtue of the fact manual or non-manual salaried workers.
that they involve common mobility chances, To emphasise, therefore, that Weber’s ‘ab-
either within the career of individuals or across stract model’ of classes is a pluralistic one is not
the generations. Thus while a worker may fairly to hold that he failed to recognise unifying ties
readily move from an unskilled to a semi-skilled between the numerous combinations of class in-
manual occupation, and the son of an unskilled terests made possible by his conception of ‘class
worker may become a semi-skilled or perhaps situation’. But there is no doubt that his view-
a skilled worker, the chances of either intra- or point drastically amends important elements
inter-generational mobility into non-manual of Marx’s picture of the typical trend of devel-
occupations are much less. While the concep- opment of the capitalist class structure. Even
tion of the ‘social class’ remains relatively unde- Weber’s simplified (‘social class’) model of cap-
veloped in Weber’s writings, it is of particular italism diverges significantly from the Marxian
interest in relation to his model of capitalist de- conception, in treating the propertyless ‘middle
velopment. As Weber himself points out, the no- class’ as the category which tends to expand
tion of ‘social class’ comes much closer to that of most with the advance of capitalism. Moreover,
‘status group’ than does the conception of purely the social classes do not necessarily constitute
economic class (although, as with economic class ‘communities’, and they may be fragmented by
situation, individuals who are in the same social interest divisions deriving from differentials in
186 • Part III: The Structure of Inequality

market position; and finally, as Weber shows in sanction. It would be departing too much from
his historical writings, the relationship between usual terminology to refer to capital and to the
class structure and the political sphere is a con- labour-power of the worker both as ‘property’;
tingent one. . . . and, anyway, the point is rather that ‘property’
(capital) is a particular case of capacity to de-
termine the bargaining outcome, rather than
Rethinking the Theory of Class vice versa. So I shall continue to speak below of
The deficiency in Weber’s reinterpretation of ‘property’ (in the means of production) in a con-
Marx’s view is that it is not sufficiently radi- ventional sense, and shall use the term ‘market
cal. While Weber recognises the unsatisfactory capacity’ in an inclusive manner to refer to all
character of the Marxian standpoint, particu- forms of relevant attributes which individuals may
larly as regards the undifferentiated category of bring to the bargaining encounter.
the ‘propertyless’, he does not pursue the im- It is an elementary fact that where ownership
plications of his own conception far enough. of property is concentrated in the hands of a mi-
Dahrendorf has suggested that we may stand nority and in a society in which the mass of the
the Marxian concept of property on its head in population is employed in industrial production,
terms of its relation to authority;9 the implica- the vast majority consequently offer their labour
tions of the Weberian analysis, however, are that for sale on the market. Because of his general em-
the conception of property may be ‘inverted’ or phasis upon ‘productive labour’, and because of
generalised in a different way, which does not his expectation that it is in the nature of modern
sacrifice the economic foundation of the concept technology to reduce productive operations to a
of class. ‘Property’ refers, not to any character- homogeneous skill-level, Marx failed to recog-
istics of physical objects as such, but to rights nise the potential significance of differentiations
which are associated with them, which in turn of market capacity which do not derive directly
confer certain capacities upon the ‘owner’. . . . In from the factor of property ownership. Such dif-
the market, of course, the significance of capital ferentiations, it seems clear, are contingent upon
as private property is that it confers certain very the scarcity value of what the individual ‘owns’
definite capacities upon its possessor as compared and is able to offer on the market. As Weber indi-
to those who are ‘propertyless’—those who do cates, possession of recognised ‘skills’—including
not own their means of production. But we can educational qualifications—is the major factor
readily perceive that, even in the Marxian view, influencing market capacity. Differentiations in
the notion of ‘propertylessness’ is something of market capacity may be used, as various recent
a misnomer. For if ‘property’ is conceived of as authors have indicated, to secure economic re-
a set of capacities of action with reference to turns other than income as such. These include,
the operations of the market, it is evident that principally, security of employment, prospects of
the wage-labourer does possess such capaci- career advancement, and a range of ‘fringe ben-
ties. The ‘property’ of the wage-labourer is the efits’, such as pension rights, etc.10 In the same
labour-power which he brings for sale in enter- way as the capacities which individuals bring to
ing into the contractual relation. While this fun- the bargaining process may be regarded as a form
damentally disadvantages him in the competitive of ‘property’ which they exchange on the mar-
bargaining situation in relation to the owner of ket, so these material returns may be regarded as
capital, this is not simply a one-way power rela- forms of ‘good’ which are obtained through the
tionship: the ‘property’ which the wage-labourer sale of labour-power.
possesses is needed by the employer, and he In the market structure of competitive capi-
must pay at least some minimal attention to the talism, all those who participate in the exchange
demands of the worker—providing a basis for process are in a certain sense in (interest) con-
the collective withdrawal of labour as a possible flict with one another for access to scarce returns.
 Giddens—The Class Structure of the Advanced Societies • 187

Conflict of interest may be created by the ex- capitalism. While it would by no means be true
istence of many sorts of differential market ca- to hold that Marx ignored this completely, it can
pacities. Moreover, the possible relationships be said that he gave only little attention to the
between property and ‘propertyless’ forms of modes in which classes, founded in a set of eco-
market capacity are various. Speculative invest- nomic relationships, take on or ‘express’ them-
ment in property may, for example, be one of selves in definite social forms.
the specific market advantages used by those in Nor has the matter been adequately dealt
certain occupations (thus directors are often able with in the writings of later authors. In fact,
to use ‘inside knowledge’ to profit from property one of the leading dilemmas in the theory of
deals). Marx himself, of course, recognised the class—which figures prominently, for example,
existence of persistent conflicts of interest within in Aron’s discussion—is that of identifying the
property-owning groupings: notably, between fi- ‘reality’ of class.11 Not only has there been some
nancial and industrial sectors of the large bour- considerable controversy over whether class is
geoisie, and between large and petty bourgeoisie. a ‘real’ or ‘nominal’ category, but many have
The difficulty of identifying ‘class’ with com- argued that, since it is difficult or impossible
mon market capacity has already been alluded to draw the ‘boundaries’ between classes with
to with reference to Weber. While Weber’s con- any degree of clarity, we should abandon the
cept of ‘market situation’ successfully moves notion of class as a useful sociological concept
away from some of the rigidities of the Marxian altogether.12 Only Dahrendorf seems to have at-
scheme, it tends to imply the recognition of a tempted to give attention to the problem within
cumbersome plurality of classes. There would the framework of an overall theory of class, and
appear to be as many ‘classes’, and as many ‘class since his identification of class with authority
conflicts’, as there are differing market positions. divisions is unacceptable [for reasons outlined
The problem here, however, is not the recog- elsewhere],13 his analysis does not help greatly.
nition of the diversity of the relationships and The major problems in the theory of class, I
conflicts created by the capitalist market as such, shall suggest, do not so much concern the na-
but that of making the theoretical transition from ture and application of the class concept itself,
such relationships and conflicts to the identification as what, for want of a better word, I shall call
of classes as structured forms. The unsatisfactory the structuration of class relationships.14 Most
and ill-defined character of the connections attempts to revise class theory since Marx have
between ‘class position’, the typology of Besitz- sought to accomplish such a revision primarily
klassen and Erwerbsklassen, and ‘social classes’ in by refining, modifying, or substituting an alto-
Weber’s work has already been mentioned. But gether different notion for the Marxian concept
the problem is by no means confined to Weber’s of class. While it is useful to follow and develop
theoretical scheme. Marx was certainly conscious certain of Weber’s insights in this respect, the
of the problematic character of the links between most important blank spots in the theory of
class as a latent set of characteristics generated class concern the processes whereby ‘economic
by the capitalist system and class as an historical, classes’ become ‘social classes’, and whereby in
dynamic entity, an ‘historical actor’. But his con- turn the latter are related to other social forms.
trast between class ‘in itself ’ and class ‘for itself ’ As Marx was anxious to stress in criticising the
is primarily one distinguishing between class re- premises of political economy, all economic rela-
lationships as a cluster of economic connections tionships, and any sort of ‘economy’, presuppose
on the one hand and class consciousness on the a set of social ties between producers. In arguing
other. This emphasis was very much dictated by for the necessity of conceptualising the structur-
the nature of Marx’s interests, lying as they did ation of class relationships, I do not in any way
above all in understanding and promoting the wish to question the legitimacy of this insight,
rise of a revolutionary class consciousness within but rather to focus upon the modes in which
188 • Part III: The Structure of Inequality

‘economic’ relationships become translated into of some significance, and bears directly upon the
‘non-economic’ social structures. problem of class ‘boundaries’. For the divisions
One source of terminological ambiguity and between strata, for analytical purposes, may be
conceptual confusion in the usage of the term drawn very precisely, since they may be set upon
‘class’ is that it has often been employed to refer a measurement scale—as, for example, with ‘in-
both to an economic category and to a specifiable come strata’. The divisions between classes are
cluster of social groupings. Thus Weber uses the never of this sort; nor, moreover, do they lend
term in both of these ways, although he seeks themselves to easy visualisation, in terms of any
terminologically to indicate the difference be- ordinal scale of ‘higher’ and ‘lower’, as strata
tween ‘class’ (as a series of ‘class positions’) and do—although this sort of imagery cannot be
‘social class’. But in order to insist that the study escaped altogether. Finally we must distinguish
of class and class conflict must concern itself clearly between class and elite. Elite theory, as
with the interdependence of economy and soci- formulated by Pareto and Mosca, developed in
ety, it is not necessary to identify the term ‘class’ part as a conscious and deliberate repudiation of
with the divisions and interests generated by the class analysis.16 In place of the concept of class
market as such. Consequently, in the remainder relationships, the elite theorists substituted the
of this [essay], I shall use the term in the sense of opposition of ‘elite’ and ‘mass’; and in place of
Weber’s ‘social class’—appropriately explicated. the Marxian juxtaposition of class society and
While there may be an indefinite multiplicity classlessness they substituted the idea of the cy-
of cross-cutting interests created by differential clical replacement of elites in perpetuo. . . .
market capacities, there are only, in any given
society, a limited number of classes.
It will be useful at this juncture to state what The Structuration of Class
class is not. First, a class is not a specific ‘en- Relationships
tity’—that is to say, a bounded social form in It is useful, initially, to distinguish the mediate
the way in which a business firm or a university from the proximate structuration of class re-
is—and a class has no publicly sanctioned iden- lationships. By the former term, I refer to the
tity. It is extremely important to stress this, since factors which intervene between the existence of
established linguistic usage often encourages us certain given market capacities and the forma-
to apply active verbs to the term ‘class’; but the tion of classes as identifiable social groupings,
sense in which a class ‘acts’ in a certain way, or that is to say which operate as ‘overall’ connect-
‘perceives’ elements in its environment on a par ing links between the market on the one hand
with an individual actor, is highly elliptical, and and structured systems of class relationships
this sort of verbal usage is to be avoided wher- on the other. In using the latter phrase, I refer
ever possible. Similarly, it is perhaps misleading to ‘localised’ factors which condition or shape
to speak of ‘membership’ of a class, since this class formation. The mediate structuration of
might be held to imply participation in a defi- class relationships is governed above all by the
nite ‘group’. This form of expression, however, distribution of mobility chances which pertain
is difficult to avoid altogether, and I shall not within a given society. Mobility has sometimes
attempt to do so in what follows. Secondly, class been treated as if it were in large part separable
has to be distinguished from ‘stratum’, and class from the determination of class structure. Ac-
theory from the study of ‘stratification’ as such. cording to Schumpeter’s famous example, classes
The latter, comprising what Ossowski terms a may be conceived of as like conveyances, which
gradation scheme, involves a criterion or set of may be constantly carrying different ‘passen-
criteria in terms of which individuals may be gers’ without in any way changing their shape.
ranked descriptively along a scale.15 The distinc- But, compelling though the analogy is at first
tion between class and stratum is again a matter sight, it does not stand up to closer examination,
 Giddens—The Class Structure of the Advanced Societies • 189

especially within the framework I am suggesting sense. In capitalism, the division of labour in the
here.17 In general, the greater the degree of ‘clo- enterprise is in principle governed by the pro-
sure’ of mobility chances—both intergeneration- motion of productive efficiency in relation to the
ally and within the career of the individual—the maximisation of profit; but while responding to
more this facilitates the formation of identifiable the same exigencies as the capitalist market in
classes. For the effect of closure in terms of in- general, the influence of the division of labour
tergenerational movement is to provide for the must be analytically separated as a distinctive
reproduction of common life experience over the source of structuration (and, as will be discussed
generations; and this homogenisation of expe- later, as a significant influence upon class con-
rience is reinforced to the degree to which the sciousness). The division of labour, it is clear,
individual’s movement within the labour mar- may be a basis of the fragmentation as well as the
ket is confined to occupations which generate consolidation of class relationships. It furthers
a similar range of material outcomes. In general the formation of classes to the degree to which
we may state that the structuration of classes is it creates homogeneous groupings which cluster
facilitated to the degree to which mobility closure along the same lines as those which are fostered
exists in relation to any specified form of market by mediate structuration. Within the modern
capacity. There are three sorts of market capacity industrial order, the most significant influence
which can be said to be normally of importance upon proximate structuration in the division of
in this respect: ownership of property in the labour is undoubtedly that of technique. The ef-
means of production; possession of educational fect of industrial technique (more recently, how-
or technical qualifications; and possession of ever, modified by the introduction of cybernetic
manual labour-power. In so far as it is the case systems of control) is to create a decisive separa-
that these tend to be tied to closed patterns of tion between the conditions of labour of manual
inter- and intragenerational mobility, this yields and non-manual workers. ‘Machine-minding’,
the foundation of a basic three-class system in cap- in one form or another, regardless of whether
italist society: an ‘upper’, ‘middle’, and ‘lower’ it involves a high level of manual skill, tends
or ‘working’ class. But as has been indicated to create a working environment quite distinct
previously, it is an intrinsic characteristic of the from that of the administrative employee, and
development of the capitalist market that there one which normally enforces a high degree of
exist no legally sanctioned or formally prescribed physical separation between the two groupings.18
limitations upon mobility, and hence it must be This effect of the division of labour thus over-
emphasised that there is certainly never anything laps closely with the influence of the mediate
even approaching complete closure. In order to structuration of class relationships through the
account for the emergence of structured classes, differential apportionment of mobility chances;
we must look in addition at the proximate but it is, in turn, potentially heavily reinforced
sources of structuration. by the typical authority system in the enterprise.
There are three, related, sources of proximate In so far as administrative workers participate in
structuration of class relationships: the division the framing, or merely in the enforcement, of
of labour within the productive enterprise; the authoritative commands, they tend to be sepa-
authority relationships within the enterprise; rated from manual workers, who are subject to
and the influence of what I shall call ‘distributive those commands. But the influence of differen-
groupings’. I have already suggested that Marx tial authority is also basic as a reinforcing agent
tended to use the notion of ‘division of labour’ of the structuration of class relationships at the
very broadly, to refer both to market relation- ‘upper’ levels. Ownership of property, in other
ships and to the allocation of occupational tasks words, confers certain fundamental capacities of
within the productive organisation. Here I shall command, maximised within the ‘entrepreneur-
use the term only in this second, more specific, ial’ enterprise in its classical form. To the extent
190 • Part III: The Structure of Inequality

to which this serves to underlie a division at ‘the most significant distributive groupings in this
top’, in the control of the organisation (some- respect are those formed through the tendency
thing which is manifestly influenced, but not at towards community or neighbourhood segrega-
all destroyed, if certain of the suppositions ad- tion. Such a tendency is not normally based only
vanced by the advocates of the theory of the sep- upon differentials in income, but also upon such
aration of ‘ownership and control’ are correct) it factors as access to housing mortgages, etc. The
supports the differentiation of the ‘upper’ from creation of distinctive ‘working-class neighbour-
the ‘middle’ class. hoods’ and ‘middle-class neighbourhoods’, for
The third source of the proximate structur- example, is naturally promoted if those in man-
ation of class relationships is that originating ual labour are by and large denied mortgages
in the sphere of consumption rather than pro- for house buying, while those in non-manual
duction. Now according to the traditional in- occupations experience little difficulty in ob-
terpretations of class structure, including those taining such loans. Where industry is located
of Marx and Weber, ‘class’ is a phenomenon of outside of the major urban areas, homogeneous
production: relationships established in con- ‘working-class communities’ frequently develop
sumption are therefore quite distinct from, and through the dependence of workers upon hous-
secondary to, those formed in the context of ing provided by the company.
productive activity. There is no reason to deviate In summary, to the extent to which the various
from this general emphasis. But without drop- bases of mediate and proximate class structura-
ping the conception that classes are founded ulti- tion overlap, classes will exist as distinguishable
mately in the economic structure of the capitalist formations. I wish to say that the combination of
market, it is still possible to regard consumption the sources of mediate and proximate structuration
patterns as a major influence upon class struc- distinguished here, creating a threefold class struc-
turation. Weber’s notions of ‘status’ and ‘status ture, is generic to capitalist society. But the mode
group’, as I previously pointed out, confuse two in which these elements are merged to form a
separable elements: the formation of groupings specific class system, in any given society, differs
in consumption, on the one hand, and the for- significantly according to variations in economic
mation of types of social differentiation based and political development. It should be evident
upon some sort of non-economic value provid- that structuration is never an all-or-nothing mat-
ing a scale of ‘honour’ or ‘prestige’ on the other. ter. The problem of the existence of distinct class
While the two may often coincide, they do not ‘boundaries’, therefore, is not one which can be
necessarily do so, and it seems worthwhile to settled in abstracto: one of the specific aims of
distinguish them terminologically. Thus I shall class analysis in relation to empirical societies
call ‘distributive groupings’ those relationships in- must necessarily be that of determining how
volving common patterns of the consumption strongly, in any given case, the ‘class principle’
of economic goods, regardless of whether the has become established as a mode of structura-
individuals involved make any type of conscious tion. Moreover, the operation of the ‘class prin-
evaluation of their honour or prestige relative ciple’ may also involve the creation of forms of
to others; ‘status’ refers to the existence of such structuration within the major class divisions.
evaluations, and a ‘status group’ is, then, any set One case in point is that which Marx called the
of social relationships which derives its coher- ‘petty bourgeoisie’. In terms of the preceding
ence from their application.19 analysis, it is quite easy to see why ownership of
In terms of class structuration, distributive small property in the means of production might
groupings are important in so far as they inter- come to be differentiated both from the upper
relate with the other sets of factors distinguished class and from the (‘new’) middle class. If it is the
above in such a way as to reinforce the typical sep- case that the chances of mobility, either inter- or
arations between forms of market capacity. The intragenerationally, from small to large property
 Giddens—The Class Structure of the Advanced Societies • 191

ownership are slight, this is likely to isolate the ‘Class awareness’, as I use the term here, does
small property-owner from membership of the not involve a recognition that these attitudes
upper class as such. But the fact that he enjoys and beliefs signify a particular class affiliation,
directive control of an enterprise, however mi- or the recognition that there exist other classes,
nute, acts to distinguish him from those who are characterised by different attitudes, beliefs, and
part of a hierarchy of authority in a larger or- styles of life; ‘class consciousness’, by contrast, as
ganisation. On the other hand, the income and I shall use the notion, does imply both of these.
other economic returns of the petty bourgeois The difference between class awareness and class
are likely to be similar to the white-collar worker, consciousness is a fundamental one, because
and hence they may belong to similar distribu- class awareness may take the form of a denial
tive groupings. A second potentially important of the existence or reality of classes.22 Thus the
influence upon class formation is to be traced to class awareness of the middle class, in so far as
the factor of skill differential within the general it involves beliefs which place a premium upon
category of manual labour. The manual worker individual responsibility and achievement, is of
who has undergone apprenticeship, or a compa- this order.
rable period of training, possesses a market ca- Within ethnically and culturally homoge-
pacity which sets him apart from the unskilled or neous societies, the degree of class structura-
semi-skilled worker. This case will be discussed tion will be determined by the interrelationship
in more detail [elsewhere];20 it is enough merely between the sources of structuration identified
to indicate at this point that there are certain previously. But many, if not the majority, of
factors promoting structuration on the basis of capitalist societies are not homogeneous in these
this differentiation in market capacity (e.g., that respects. Traditionally, in class theory, racial or
the chances of intergenerational mobility from religious divisions have been regarded as just so
skilled manual to white-collar occupations are many ‘obstacles’ to the formation of classes as
considerably higher than they are from unskilled coherent unities. This may often be so, where
and semi-skilled manual occupations). these foster types of structuration which devi-
So far I have spoken of structuration in ate from that established by the ‘class principle’
a purely formal way, as though class could be (as typically was the case in the battles fought
defined in terms of relationships which have by the rearguard of feudalism against the forces
no ‘content’. But this obviously will not do: if promoting the emergence of capitalism). The
classes become social realities, this must be man- idea that ethnic or cultural divisions serve to
ifest in the formation of common patterns of be- dilute or hinder the formation of classes is also
haviour and attitude. Since Weber’s discussion of very explicitly built into Weber’s separation of
classes and status groups, the notion of ‘style of (economic) ‘class’ and ‘status group’. But this,
life’ has normally come to be identified as solely in part at least, gains its cogency from the con-
pertaining to the mode whereby a status group trast between estate, as a legally constituted cat-
expresses its claim to distinctiveness. However, egory, and class, as an economic category. While
in so far as there is marked convergence of the it may be agreed, however, that the bases of the
sources of structuration mentioned above, classes formation of classes and status groups (in the
will also tend to manifest common styles of life. sense in which I have employed these concepts)
An initial distinction can be drawn here are different, nonetheless the tendency to class
between ‘class awareness’ and ‘class conscious- structuration may receive a considerable impe-
ness’.21 We may say that, in so far as class is a tus where class coincides with the criteria of status
structurated phenomenon, there will tend to group membership—in other words, where struc-
exist a common awareness and acceptance of turation deriving from economic organisation
similar attitudes and beliefs, linked to a common ‘overlaps’ with, or, in Dahrendorf ’s terms, is ‘su-
style of life, among the members of the class. perimposed’ upon, that deriving from evaluative
192 • Part III: The Structure of Inequality

categorisations based upon ethnic or cultural dif- 15. Stanislaw Ossowski, Class Structure in the So-
ferences.23 Where this is so, status group mem- cial Consciousness (London 1963).
bership itself becomes a form of market capacity. 16. Vilfredo Pareto, The Mind and Society (New
Such a situation frequently offers the strongest York 1935); Gaetano Mosca, The Ruling Class (New
York 1939).
possible source of class structuration, whereby
17. We may, however, agree with Schumpeter
there develop clear-cut differences in attitudes, that ‘The family, not the physical person, is the true
beliefs and style of life between the classes. unit of class and class theory’ (Joseph Schumpeter,
Where ethnic differences serve as a ‘disqualify- Imperialism, Social Classes, Cleveland 1961). This is
ing’ market capacity, such that those in the cate- actually completely consistent with the idea that mo-
gory in question are heavily concentrated among bility is fundamental to class formation.
the lowest-paid occupations, or are chronically 18. Lockwood, The Blackcoated Worker, op. cit.
unemployed or semi-employed, we may speak 19. It might be pointed out that it would eas-
of the existence of an underclass.24 ily be possible to break down the notion of status
group further: according, for example, to whether
the status evaluations in question are made primarily
NOTES
by others outside the group, and rejected by those
1. Theodor Geiger, Die Klassengesellschaft im
inside it, etc.
Schmeltztiegel (Cologne 1949); Karl Renner, Wand-
20. Giddens, The Class Structure of the Advanced
lungen der modernen Gesellschaft (Vienna 1953); Ralf
Societies.
Dahrendorf, Class and Class Conflict in Industrial So-
21. This is not, of course, the same as Lukács’
ciety (Stanford 1959).
‘class-conditioned unconsciousness’; but I believe
2. For a cogent representation of this view, see
that Lukács is correct in distinguishing qualita-
W. G. Runciman, ‘Class, status and power’, in J. A.
tively different ‘levels’ of class consciousness. Georg
Jackson, Social Stratification (Cambridge 1968).
Lukács, History and Class Consciousness (London
3. See my Capitalism and Modern Social Theory
1971), pp. 52ff.
(Cambridge 1971), pp. 185ff. and passim.
22. cf. Nicos Poulantzas, Pouvoir politique et
4. Economy and Society, vol. 2 (New York 1968),
classes sociales de l’état capitaliste (Paris 1970). It is
pp. 926–40, and vol. 1, pp. 302–7.
misleading, however, to speak of classes sans con-
5. Ibid., vol. 2, p. 928.
science, as Crozier does. See Michel Crozier, ‘Classes
6. Ibid., p. 930.
sans conscience ou préfiguration de la société sans
7. Ibid., p. 927.
classes’, Archives européenes de sociologie 1, 1960; also
8. Ibid., p. 305.
‘L’ambiguité de la conscience de classe chez les em-
9. Dahrendorf, Class and Class Conflict.
ployés et les petits fonctionnaires’, Cahiers interna-
10. See, for example, David Lockwood, The
tionaux de sociologie 28, 1955.
Blackcoated Worker (London 1958), pp. 202–4;
23. Or, to use another terminology, where there
Frank Parkin, Class Inequality and Political Order
is ‘overdetermination’ (Louis Althusser, For Marx,
(London 1971).
London 1969, pp. 89–128).
11. Raymond Aron, La lutte des classes (Paris
24. Marx’s Lumpenproletariat, according to this
1964).
usage, is only an underclass when the individuals
12. See Robert A. Nisbet, ‘The decline and fall
in question tend to derive from distinctive ethnic
of social class’, Pacific Sociological Review 2, 1959.
backgrounds. Leggett has referred to the underclass
13. Anthony Giddens, The Class Structure of the
as the ‘marginal working class’, defining this as ‘a
Advanced Societies (New York 1973), ch. 4.
sub-community of workers who belong to a subor-
14. What I call class structuration, Gurvitch calls
dinate ethnic or racial group which is usually prole-
negatively ‘résistance à la pénétration par la société
tarianised and highly segregated’ (John C. Leggett,
globale’. Georges Gurvitch, Le concept de classes socia-
Class, Race, and Labor, New York 1968, p. 14).
les de Marx à nos jours (Paris 1954), p. 116 and passim.
 Parkin—Marxism and Class Theory • 193

20 • Frank Parkin

Marxism and Class Theory

A Bourgeois Critique

The ‘Boundary Problem’ in Sociology contrast between the empirical model of class
The persistent attractions of Marxist class theory and the general conception of capitalist society.
have almost certainly been boosted by the less The strongest case that could be made out
than inspiring alternative offered by academic for identifying the line between manual and
sociology. In so far as there is any sort of tac- non-manual labour as the focal point of class
itly agreed upon model of class among western conflict would be one that treated capitalist so-
social theorists it takes the form of the familiar ciety as the industrial firm writ large. It is only
distinction between manual and non-manual la- within the framework of ‘factory despotism’ that
bour. No other criterion for identifying the class the blue-collar/white-collar divide closely corre-
boundary seems to enjoy such widespread accep- sponds to the line of social confrontation over
tance among those who conduct investigations the distribution of spoils and the prerogatives of
into family structure, political attitudes, social command. And this is particularly the case in
imagery, life-styles, educational attainment, and those industrial settings where even the lowest
similar enquiries that keep the wheels of empir- grades of white-collar staff are cast in the role
ical sociology endlessly turning. Paradoxically, of managerial subalterns physically and emo-
however, although the manual/non-manual tionally removed from the shop-floor workers.
model is felt to be highly serviceable for research Within the microcosm of capitalism represented
purposes, it is not commonly represented as a by the typical industrial firm, the sociological
model of class cleavage and conflict. That is to model of class has something to recommend it
say, the two main social categories distinguished as an alternative to one constructed around the
by sociology for purposes of class analysis are not rights of property.
invested with antagonistic properties compara- The drawback is, however, that social rela-
ble to those accorded to proletariat and bour- tions within the capitalist firm are a less accurate
geoisie in Marxist theory. This would be less guide to class relations within capitalist society
cause for comment if proponents of the manual/ than they might once have been. The reason for
non-manual model normally construed the so- this is that the post-war expansion of the pub-
cial order as a harmonious and integrated whole; lic sector has given rise to an ever-increasing
but to construe it instead in terms of conflict, assortment of non-manual groups in local gov-
dichotomy, and cleavage, as most of these writers ernment and welfare services that cannot in any
now appear to do, seems to reveal an awkward real sense be thought of as the tail-end of a broad

Frank Parkin. Marxism and Class Theory: A Bourgeois Critique, pp. 11–13, 23–25, 27–28, 44–50, 53–58,
62–64, 67–73, 112–113. Copyright © 1979 by Frank Parkin. Used by permission of Columbia University
Press and Taylor & Francis Books UK.
194 • Part III: The Structure of Inequality

managerial stratum aligned against a manual What this suggests is that manual and
workforce. Frequently, in fact there is no man- non-manual groups can usefully be thought of as
ual workforce to confront in the occupational entities socially differentiated from each other in
settings within which these white-collar groups terms of life-chances and opportunities, but not
are employed. And even where teachers, social as groups standing in a relationship of exploiter
workers, nurses, local government clerks, lower and exploited, of dominance and subordination,
civil servants, and the like do form part of an in the manner presumably required of a genuine
organization that includes janitors, orderlies, conflict model. Expressed differently, the current
cleaners, and other workers by hand, they do sociological model does not fulfil even the min-
not usually stand in the same quasi-managerial imal Weberian claim that the relations between
relationship to them as does the staff employee classes are to be understood as ‘aspects of the
to the industrial worker in the capitalist firm. distribution of power’. Instead of a theoretical
The usual rationale for treating intermediate framework organized around the central ideas
and lower white-collar groups as a constituent of mutual antagonism and the incompatibility
element of a dominant class is that these groups of interests we find one organized around the
traditionally have identified themselves with the recorded facts of mere social differentiation. . . .
interests of capital and management rather than
with the interests of organized labour. But for
various reasons this identification is easier to ac- The ‘Boundary Problem’ in Marxism
complish in the sphere of private industry and The variety of [Marxist] interpretations on offer
commerce than in the public sector. In the latter, make it more than usually difficult to speak of
as already pointed out, not only is there usually ‘the’ Marxist theory of class. In some respects the
no subordinate manual group physically present range of differences within this camp has tended
to inspire a sense of white-collar status elevation, to blur the simple contrast between Marxist
but also the charms of management are likely to and bourgeois theories; and this is particularly
seem less alluring when the chain of command so given the tendency for Marxists to adopt fa-
stretches ever upwards and out of sight into miliar sociological categories under substitute
the amorphous and unlovely body of the state. names. The most striking example of this is the
Moreover, public sector employees do not have tacit acknowledgment of the role of authority in
the same opportunities as those in the commer- the determination of bourgeois status. This arises
cial sector for transferring their special skills and from the need to find some theoretical principle
services to different and competing employers; by which the managerial stratum, in particular,
all improvements in pay and conditions must can be assigned to the same class as the owners
be negotiated with a monopoly employer, and of capital. Although allusions may occasionally
one who is under close budgetary scrutiny. All be made to the fact that managers are sometimes
this makes for a relationship of some tension be- shareholders in the companies that employ
tween white-collar employees and the state qua them, it is clear that this is a contingent feature
employer, a condition more akin to that found of managerial status and could not be regarded
between manual labour and management than as theoretically decisive. Managers with and
between white-collar employees and manage- without private company shares do not appear
ment in the private sector. Thus, the validity of to be different political and ideological animals.
the manual/non-manual model as a representa- The exercise of discipline over the workforce,
tion of class conflict relies more heavily upon a on the other hand, is a necessary feature of the
view of the commercial employee as the proto- managerial role, not a contingent one; and as
typical case of the white-collar worker than re- such it recommends itself as a major criterion of
ally is justified, given the enormous growth of bourgeois class membership. Indeed, for some
public-sector employment. Marxists managerial authority has in certain
 Parkin—Marxism and Class Theory • 195

respects superseded property ownership as the mental labour were present not only in societies
defining attribute of a capitalist class. According where the manager was ‘capital personified’, but
to Carchedi, ‘the manager, rather than the capi- also in societies where he was the party person-
talist rentier, is the central figure, he, rather than ified? Marxists would then be faced with the
the capitalist rentier, is the non-labourer, the unwelcome choice of either having to expand
non-producer, the exploiter. He, rather than the the definition of capitalism to embrace socialist
capitalist rentier, is capital personified’ (Carchedi society, or of disowning the cherished concepts
1975:48). of private property and surplus extraction upon
Interestingly, by proclaiming that the supervi- which their class theory is grounded. The obvi-
sion and control of subordinates is the new hall- ous reluctance to engage in the comparative anal-
mark of bourgeois status, Marxist theorists have ysis of class under the two ostensibly different
come surprisingly close to endorsing Dahren- modes of production is therefore understandable
dorf ’s view of the determinate role of authority enough. As for the credibility of Marxist class
in establishing the class boundary (Dahrendorf theory, it would seem that the advent of socialist
1959). Their strict avoidance of this term in fa- society is about the worst thing that could have
vour of some synonym or circumlocution (‘men- happened to it.
tal labour’, ‘global function of capital’, ‘labour of A further difficulty encountered by this
superintendence’) is perhaps a tacit admission of theory is the attempt to arrive at some general
this embarrassing affinity with Dahrendorf ’s po- principles by which to demarcate the estab-
sition. Although none of these writers would ac- lished professions from routine white-collar
cept Dahrendorf ’s proposition that authority is a employees, a distinction required by the evi-
general phenomenon that encompasses property, dent self-identification of the former with the
it is nevertheless the case that their treatment of general interests of the bourgeoisie. In place of
authority relations, however phrased, takes up any general principles, however, resort is had
far more of their analysis than the discussion of to an eclectic assortment of descriptive indices
property relations. demonstrating that ‘higher’ white-collar groups
To make property the centrepiece of class are in various ways simply better off than ‘lower’
analysis would bring with it the duty of explain- white-collar groups. Braverman, for example,
ing precisely why the apparatus of managerial lists advantages such as higher pay, security of
authority and control was thought to grow out employment, and the privileged market position
of the institution of private ownership. Presum- of the professions (Braverman 1974: Chapter
ably it has come to the attention of western 18). In similar vein, Westergaard and Resler
Marxists that societies that have done away with suggest drawing a line of class demarcation be-
property in its private forms nevertheless have neath professional and managerial groups on
their own interesting little ways of seeing to the the grounds that ‘they are not dependent on the
‘superintendence of labour’. The view that class markets in which they sell their labour in any-
and authority relations under capitalism are a thing like the way that other earners are’ (West-
unique product of private ownership must rest ergaard and Resler 1975:92). Their incomes ‘are
on a belief that these things are ordered in a determined by market rules and mechanisms
very different way under the socialist mode of over which, in effect, they themselves have con-
production. The fact that this mode of produc- siderable influence in their own corners of the
tion figures not at all in any of the class analyses market’ (Westergaard and Resler 1975:346).
referred to suggests that Marxists are none too The one notable thing about this kind of
happy about drawing the very comparisons that analysis is that despite its avowedly Marxist prov-
are so essential to their case. After all, supposing enance it is indistinguishable from the approach
it was discovered that factory despotism, the co- of modern bourgeois social theory. It is, after
ercive uses of knowledge, and the privileges of all, Weber rather than Marx who provides the
196 • Part III: The Structure of Inequality

intellectual framework for understanding class of the distribution of power, which for Weber is
in terms of market opportunities, life-chances, practically synonymous with stratification. As a
and symbolic rewards. The focus upon income result, the usefulness of the concept for the study
differences and other market factors is difficult of class and similar forms of structured inequal-
to reconcile with the standard Marxist objection ity becomes conditional on the acceptance of
to bourgeois sociology that it mistakenly oper- certain refinements and enlargements upon the
ates on the level of distribution instead of on original usage.
the level of productive relations. It might also An initial step in this direction is to extend
be said that it is from Weber rather than Marx the notion of closure to encompass other forms
that the postulated link between class position of collective social action designed to maximize
and bureaucratic authority most clearly derives. claims to rewards and opportunities. Closure
The fact that these normally alien concepts of strategies would thus include not only those of
authority relations, life-chances, and market an exclusionary kind, but also those adopted by
rewards have now been comfortably absorbed the excluded themselves as a direct response to
by contemporary Marxist theory is a handsome, their status as outsiders. It is in any case hardly
if unacknowledged, tribute to the virtues of possible to consider the effectiveness of exclusion
bourgeois sociology. Inside every neo-Marxist practices without due reference to the counter-
there seems to be a Weberian struggling to get vailing actions of socially defined ineligibles. As
out. . . . Weber acknowledges: ‘Such group action may
provoke a corresponding reaction on the part of
those against whom it is directed’ (Weber [eds
Social Closure Roth and Wittich] 1968:342). In other words,
By social closure Weber means the process by collective efforts to resist a pattern of dominance
which social collectivities seek to maximize governed by exclusion principles can properly
rewards by restricting access to resources and be regarded as the other half of the social clo-
opportunities to a limited circle of eligibles. sure equation. This usage is in fact employed by
This entails the singling out of certain social Weber in his discussion of ‘community closure’
or physical attributes as the justificatory basis which, as Neuwirth has shown, bears directly
of exclusion. Weber suggests that virtually any upon those forms of collective action mounted
group attribute—race, language, social origin, by the excluded—that is, ‘negatively privileged
religion—may be seized upon provided it can status groups’ (Neuwirth 1969).
be used for ‘the monopolization of specific, usu- The distinguishing feature of exclusionary
ally economic opportunities’ (Weber [eds Roth closure is the attempt by one group to secure for
and Wittich] 1968:342). This monopolization itself a privileged position at the expense of some
is directed against competitors who share some other group through a process of subordination.
positive or negative characteristic; its purpose is That is to say, it is a form of collective social ac-
always the ‘closure of social and economic op- tion which, intentionally or otherwise, gives rise
portunities to outsiders’ (Weber [eds Roth and to a social category of ineligibles or outsiders.
Wittich] 1968:342). The nature of these exclu- Expressed metaphorically, exclusionary closure
sionary practices, and the completeness of social represents the use of power in a ‘downward’ di-
closure, determine the general character of the rection because it necessarily entails the creation
distributive system. of a group, class, or stratum of legally defined in-
Surprisingly, Weber’s elaboration of the clo- feriors. Countervailing action by the ‘negatively
sure theme is not linked in any immediate way privileged’, on the other hand, represents the
with his other main contributions to stratifica- use of power in an upward direction in the sense
tion theory, despite the fact that processes of ex- that collective attempts by the excluded to win a
clusion can properly be conceived of as an aspect greater share of resources always threaten to bite
 Parkin—Marxism and Class Theory • 197

into the privileges of legally defined superiors. It In modern capitalist society the two main
is in other words a form of action having usur- exclusionary devices by which the bourgeoi-
pation as its goal. Exclusion and usurpation may sie constructs and maintains itself as a class
therefore be regarded as the two main generic are, first, those surrounding the institutions of
types of social closure, the latter always being a property; and, second, academic or professional
consequence of, and collective response to, the qualifications and credentials. Each represents a
former (see Parkin 1974). set of legal arrangements for restricting access to
Strategies of exclusion are the predominant rewards and privileges: property ownership is a
mode of closure in all stratified systems. Where form of closure designed to prevent general ac-
the excluded in their turn also succeed in closing cess to the means of production and its fruits;
off access to remaining rewards and opportu- credentialism is a form of closure designed to
nities, so multiplying the number of substrata, control and monitor entry to key positions in
the stratification order approaches the furthest the division of labour. The two sets of bene-
point of contrast to the Marxist model of class ficiaries of these state-enforced exclusionary
polarization. The traditional caste system and practices may thus be thought of as the core
the stratification of ethnic communities in the components of the dominant class under mod-
United States provide the clearest illustrations ern capitalism. . . .
of this closure pattern, though similar processes The case for restoring the notion of property
are easily detectable in societies in which class into the centre of class analysis is that it is the
formation is paramount. Strategies of usurpation most important single form of social closure
vary in scale from those designed to bring about common to industrial societies. That is to say,
marginal redistribution to those aimed at total rights of ownership can be understood not as a
expropriation. But whatever their intended scale special case of authority so much as a specific
they nearly always contain a potential challenge form of exclusion. As Durkheim expresses it, ‘the
to the prevailing system of allocation and to the right of property is the right of a given individual
authorized version of distributive justice. to exclude other individual and collective enti-
All this indicates the ease with which the lan- ties from the usage of a given thing’ (Durkheim
guage of closure can be translated into the lan- 1957:142). Property is defined negatively by ‘the
guage of power. Modes of closure can be thought exclusion it involves rather than the prerogatives
of as different means of mobilizing power for the it confers’ (Durkheim 1957:142). Durkheim’s
purpose of engaging in distributive struggle. To reference to individual rights of exclusion clearly
conceive of power as a built-in attribute of clo- indicates that once again he has possessions in
sure is at the very least to dispense with those mind, and that, characteristically, he sees no im-
fruitless searches for its ‘location’ inspired by portant distinction between objects of personal
Weber’s more familiar but completely unhelp- ownership, and the control of resources resulting
ful definition in terms of the ubiquitous struggle in the exercise of power.
between contending wills. Moreover, to speak of It is clearly necessary to distinguish property
power in the light of closure principles is quite as possessions from property as capital, since
consistent with the analysis of class relations. only the latter is germane to the analysis of class
Thus, to anticipate the discussion, the familiar systems. Property as capital is, to paraphrase
distinction between bourgeoisie and proletariat, Macpherson, that which ‘confers the right to
in its classic as well as in its modern guise, may deny men access to the means of life and labour’
be conceived of as an expression of conflict be- (Macpherson 1973). This exclusionary right can
tween classes defined not specifically in relation obviously be vested in a variety of institutional
to their place in the productive process but in forms, including the capitalist firm, a national-
relation to their prevalent modes of closure, ex- ized industry, or a Soviet enterprise. All these are
clusion and usurpation, respectively. . . . examples of property that confers legal powers
198 • Part III: The Structure of Inequality

upon a limited few to grant or deny general ‘The development of the diploma from
access to the means of production and the dis- universities, and business and engineering
tribution of its fruits. Although personal posses- colleges, and the universal clamour for the
sions and capital both entail rights of exclusion, creation of educational certificates in all fields
it is only the exclusionary rights embedded in make for the formation of a privileged stra-
the latter that have important consequences for tum in bureaus and offices. Such certificates
the life-chances and social condition of the ex- support their holders’ claims for intermar-
cluded. To speak of property in the context of riages with notable families . . . , claims to
class analysis is, then, to speak of capital only, be admitted into the circles that adhere to
and not possessions. “codes of honour,” claims for a “respectable”
Once property is conceptualized as a form remuneration rather than remuneration for
of exclusionary social closure there is no need work well done, claims for assured advance-
to become entangled in semantic debates over ment and old-age insurance, and, above all,
whether or not workers in socialist states are claims to monopolize social and economically
‘really’ exploited. The relevant question is not advantageous positions. When we hear from
whether surplus extraction occurs, but whether all sides the demand for an introduction of
the state confers rights upon a limited circle of regular curricula and special examinations,
eligibles to deny access to the ‘means of life and the reason behind it is, of course, not a sud-
labour’ to the rest of the community. If such denly awakened “thirst for education” but
exclusionary powers are legally guaranteed and the desire for restricting the supply of these
enforced, an exploitative relationship prevails positions and their monopolization by the
as a matter of definition. It is not of overriding owners of educational certificates. Today the
importance to know whether these exclusion- “examination” is the universal means of this
ary powers are exercised by the formal owners monopolization, and therefore examinations
of property or by their appointed agents, since irresistibly advance’ (Weber [eds Gerth and
the social consequences of exclusion are not de- Mills] 1946:241–42).
monstrably different in the two cases. Carchedi
and other neo-Marxists may therefore be quite The use of credentials for closure purposes, in
correct in suggesting that ‘the manager is capi- the manner elaborated by Weber, has accompa-
tal personified’; but all that needs to be added nied the attempt by an ever-increasing number
is first, that this dictum holds good not only for of white collar occupations to attain the status
monopoly capitalism, but for all, including so- of professions. Professionalization itself may
cialism, systems in which access to property and be understood as a strategy designed, amongst
its benefices is in the legal gift of a select few; other things, to limit and control the supply of
and, second, that it squares far more comfortably entrants to an occupation in order to safeguard
with the assumptions of bourgeois, or at least or enhance its market value. Much of the litera-
Weberian, sociology than with classical Marxist ture on the professions has tended to stress their
theory. differences from workaday occupations, usually
Of equal importance to the exclusionary accepting the professions’ own evaluation of
rights of property is that set of closure practices their singularity in creating rigorous codes of
sometimes referred to as ‘credentialism’—that is, technical competence and ethical standards. It
the inflated use of educational certificates as a is perfectly possible to accept that the monop-
means of monitoring entry to key positions in olization of skills and services does enable the
the division of labour. Well before the onset of professions to exercise close control over the
mass higher education, Weber had pointed to moral and technical standards of their members,
the growing use of credentials as a means of ef- whilst also endorsing Weber’s judgment that
fecting exclusionary closure. ‘normally this concern for efficient performance
 Parkin—Marxism and Class Theory • 199

recedes behind the interest in limiting the supply of the apprenticeship system or certain forms
of candidates for the benefices and honours of a of the closed shop. Some unskilled occupations
given occupation’ (Weber [eds Roth and Wit- such as dock work and market-portering have
tich] 1968:344). . . . also sought to restrict entry to the kinsmen of
The supreme advantage of occupational clo- those already employed, though this does not
sure based upon credentials is that all those in normally guarantee control over the actual vol-
possession of a given qualification are deemed ume of labour supply. The crucial difference
competent to provide the relevant skills and between these attempts at occupational exclu-
services for the rest of their professional lives. sion by manual trades and those adopted by the
There is no question of retesting abilities at a professions is that the latter generally seek to
later stage in the professional career. The profes- establish a legal monopoly over the provision of
sional bodies’ careful insistence that members of services through licensure by the state. Whereas
the lay public are not competent to sit in judge- the learned professions have been remarkably
ment on professional standards effectively means successful in winning for themselves the status
that a final certificate is a meal ticket for life. In of what Weber calls ‘legally privileged groups’, it
the sporting and entertainment professions, by has been far less common for the manual trades
contrast, the skills and abilities of the perform- to secure the blessing of the state for their ex-
ers are kept under continuous open review by clusionary tactics. Indeed, the resort to ‘restric-
the public; those who consume the services are tive practices’ on the part of organized labour
themselves the ultimate arbiters of an individu- is commonly condemned as a breach of indus-
al’s competence and hence his market value, as trial morality that should be curbed rather than
expressed via their aggregate purchasing power. sanctified by law. Presumably the fact that gov-
There can be no resort to the umbrella pro- ernments have usually been reluctant to legislate
tection of a professional licence when sporting formally against such practices is not unrelated
prowess and the ability to entertain are felt to be to the awkwardness that might arise in drawing
in decline in the eyes of those who pass collective legal distinctions between these practices and the
judgement. exclusionary devices of the professions, includ-
Against this exacting yardstick, then, creden- ing the profession of law itself.
tialism stands out as a doubly effective device A further point of difference between profes-
for protecting the learned professions from the sional closure and restrictive practices by trade
hazards of the marketplace. Not merely does it unions is that the main purpose behind the lat-
serve the convenient purpose of monitoring and ter activity has been the attempt to redress in
restricting the supply of labour, but also effec- some small part the disadvantages accruing to
tively masks all but the most extreme variations labour in its uneven contest with capital. Closure
in the level of ability of professional members, by skilled workers has been a strategy embarked
thereby shielding the least competent from ru- upon in the course of struggle against a superior
inous economic punishment. The small irony is and highly organized opponent, and not pri-
that credentialist strategies aimed at neutraliz- marily with the conscious intent of reducing the
ing the competitive effects of the market confer material opportunities of other members of the
most benefit upon that class that is most prone labour force. Credentialism, on the other hand,
to trumpet the virtues of a free market economy cannot be seen as a response to exploitation by
and the sins of collectivism. powerful employers; the learned or free profes-
The use of systematic restrictions upon oc- sions were never directly subordinate to an em-
cupational entry has not of course been wholly ploying class during the period when they were
confined to the white-collar professions. Certain effecting social closure. Their conflict, concealed
skilled manual trades have adopted similar tech- beneath the rhetoric of professional ethics was, if
niques designed to regulate supply, as in the case anything, with the lay public. It was the struggle
200 • Part III: The Structure of Inequality

to establish a monopoly of certain forms of class will strive to put their own advantages to
knowledge and practice and to win legal protec- the service of their children, while asserting at
tion from lay interference. The aim was to ensure the same time that bourgeois forms of closure are
that the professional-client relationship was one not exactly tailor-made for self-recruiting pur-
in which the organized few confronted the disor- poses. In fact exclusionary institutions formed
ganized many. Under modern conditions, where under capitalism do not seem to be designed
many professionals are indirectly in the service first and foremost to solve the problem of class
of the state and occasionally in conflict with the reproduction through the family line. The kin-
government of the day over pay and conditions, ship link can only be preserved as a result of
a somewhat better case could perhaps be made adaptation by the bourgeois family to the de-
for likening the position of professions to that mands of institutions designed to serve a differ-
of craft unions, in so far as both could be said to ent purpose; it does not come about as a natural
employ closure for purposes of bargaining with a consequence of the closure rules themselves. In
more powerful agency. But however acrimonious systems based on aristocratic, caste, or racial
relations may become between professional bod- exclusion, families of the dominant group can
ies and the state, it is worth noting that the state expect to pass on their privileged status to their
rarely if ever threatens to take sanctions against own descendants as a direct result of the closure
professions in the way that would most seriously rules in operation, however socially lethargic
damage their interests—namely, by rescinding those families might be. The bourgeois family,
their legal monopoly. by contrast, cannot rest comfortably on the as-
On all these grounds it is necessary to regard sumption of automatic class succession; it must
credentialism as a form of exclusionary social make definite social exertions of its own or face
closure comparable in its importance for class the very real prospect of generational decline. In
formation to the institution of property. Both other words, although the typical bourgeois fam-
entail the use of exclusionary rules that confer ily will certainly be better equipped than most
benefits and privileges on the few through deny- to cope with the closure system on its children’s
ing access to the many, rules that are enshrined behalf, it must still approach the task more in the
in law and upheld by the coercive authority of manner of a challenge with serious risks attached
the state. It follows from this that the dominant than as a foregone conclusion. Even when it is
class under modern capitalism can be thought successful it must face the prospect of sharing
of as comprising those who possess or control bourgeois status with uncomfortably large num-
productive capital and those who possess a bers of parvenus. What kind of system is this
legal monopoly of professional services. These to provoke such anxieties in the breasts of those
groups represent the core body of the dominant supposedly in command?
or exploiting class by virtue of their exclusion- The answer must be that it is a system de-
ary powers which necessarily have the effect of signed to promote a class formation biased more
creating a reciprocal class of social inferiors and in the direction of sponsorship and careful se-
subordinates. . . . lection of successors than of hereditary trans-
mission. Although both aims might be held
desirable, the first takes ideological precedence
Class Reproduction over the second, so that succession along kin-
There is a definite tension between the commit- ship lines must be accomplished in conformity
ment to closure by way of property and creden- with the application of criteria that are osten-
tials on the part of one generation and the desire sibly indifferent to the claims of blood. There
to pass on benefits to subsequent generations is nothing especially bizarre about an arrange-
of kith and kin. It is not in the least necessary ment whereby a dominant class relinquishes its
to deny that most members of the exclusionary children’s patrimony in order to ensure that the
 Parkin—Marxism and Class Theory • 201

calibre of its replacements is of the highest possi- benefit, nor of irresolvable and fatal contradic-
ble order. It would only appear strange to those tion. Rather, the relationship is understood as
unable to conceive that the attachment to doc- one of mutual antagonism and permanent ten-
trine could ever take precedence over the claims sion; that is, a condition of unrelieved distribu-
of kinship. . . . tive struggle that is not necessarily impossible to
‘contain’. Class conflict may be without cease,
but it is not inevitably fought to a conclusion.
Conclusion The competing notions of harmony, contradic-
By way of concluding this part of the discussion, tion, and tension could thus be thought of as the
it might be appropriate to offer some general re- three broad possible ways of conceptualizing the
marks on the explanatory status of the closure relation between classes, and on which all class
model. This model, like any other, recommends models are grounded.
the use of a particular sociological vocabulary
and an attendant battery of concepts that con- REFERENCES
tain barely disguised moral assumptions about Braverman, H. (1974) Labor and Monopoly Capital.
New York: Monthly Review Press.
the nature of class society. It is not strictly speak-
Carchedi, G. (1975) On the Economic Identifica-
ing a ‘theory’ of class but a way of conceptual-
tion of the New Middle Class. Economy and So-
izing it that differs from that proposed by other ciety 4 (1).
variants of bourgeois sociology or by Marxism. Dahrendorf, R. (1959) Class and Class Conflict in
Most of what we conventionally call theories of Industrial Society. London: Routledge.
class are in fact conceptual methods of this kind. Davis, K. and Moore, W. E. (1945) Some Principles
They are, for the most part, take-it-or-leave-it of Stratification. American Sociological Review X
moral classifications, not sets of propositions that (2).
stand or fall under the impact of evidence. What Durkheim, E. (1957) Professional Ethics and Civic
conceivable social facts could destroy either the Morals. London: Routledge.
Macpherson, C. B. (1973) A Political Theory of
Marxist conception of class as an exploitative re-
Property. In Democratic Theory: Essays in Re-
lationship, or the liberal conception of class as an trieval. London: Oxford University Press.
exchange relationship? Since conceptual models Mills, C. W. (1956) White Collar. New York: Oxford
are ways of presenting social reality, it follows University Press.
that the preference for one presentation over an- Neuwirth, G. (1969) A Weberian Outline of a The-
other entails a personal judgement of some kind ory of Community: Its Application to the ‘Dark
about the moral standing of class society. Ghetto’. British Journal of Sociology 20 (2).
On this score, the closure model is almost Orwell, G. (1949) Nineteen Eighty-Four. London:
bound to appear defective by liberal and Marxist Secker and Warburg.
Parkin, F. (ed.) (1974) The Social Analysis of Class
theorists alike. Liberal theory endorses a contrac-
Structure. London: Tavistock.
tual view of class, in which the notion of mutual
Parsons, T. (1951) The Social System. London:
interest and harmony is the essential ingredient. Routledge.
Marxism, on the other hand, assumes not merely Weber, M. (1946) From Max Weber. Gerth, H. H.
the absence of harmony and common class in- and Mills, C. W. (eds) New York: Oxford Uni-
terests, but, more importantly, the presence of versity Press.
irresolvable antagonisms that drive the system ———. (1968) Economy and Society. Roth, G. and
to ultimate breakdown. The neo-Weberian po- Wittich, C. (eds) New York: Bedminster Press.
sition advanced here is that the relation between Westergaard, J. and Resler, H. (1975) Class in a Cap-
italist Society. London: Heinemann.
classes is neither one of harmony and mutual
202 • Part III: The Structure of Inequality

21 • Tak Wing Chan and John H. Goldthorpe

Is There a Status Order in Contemporary

British Society?
Evidence from the Occupational Structure of Friendship

Introduction (cf. Dahrendorf, 1959: 74–77; Giddens, 1973:

The main question we consider in this paper is 106).
that of whether present-day British society has At least up to the middle decades of the last
a recognisable status order. We treat this ques- century, it was in fact commonplace for sociol-
tion, however, in the context of a larger problem ogists to distinguish between status and class
concerning the stratification of modern societies much on the lines indicated above (e.g. see Mills,
in general: namely, that of whether it is still em- 1951; Marshall, 1953/1963; Lockwood, 1958).
pirically defensible and conceptually valuable to Subsequently, though, this approach has been far
distinguish, on broadly Weberian lines (Weber, less often followed, as a result, it would seem,
1922/1968: 302–307, 926–939), between a sta- of a number of only rather loosely connected
tus order and a class structure. developments.
By a status order we understand a set of hi- On the one hand, doubts have arisen over
erarchical relations that express perceived and whether in the more advanced societies of the
typically accepted social superiority, equality or present day well-defined status orders still exist.
inferiority of a quite generalised kind, attaching Historians, drawing in part on earlier sociologi-
not to qualities of particular individuals but cal work, have produced evidence of a long-term
rather to social positions that they hold or to decline in deference. Individuals treated by oth-
certain of their ascribed attributes (e.g. ‘birth’ ers as social inferiors would appear to have be-
or ethnicity). A class structure, in contrast, we come less ready to accept such derogation or at
would see as being grounded specifically, and all events to acknowledge it through words or ac-
quite objectively, in the social relations of eco- tions—such as the use of honorifics, curtseying,
nomic life—i.e. in the social relations of labour cap-touching etc. The overt expression of social
markets and production units. While typically superiority has also, perhaps, become less accept-
generating differential, and often extreme, ad- able (cf. Runciman, 1997: 153–158 esp). At the
vantage and disadvantage, a class structure does same time, ‘local status systems’ that in an ear-
not necessarily take on the consistently hier- lier period were often documented in fascinating
archical form that is inherent to a status order detail in community studies (e.g. Warner and

Tak Wing Chan and John H. Goldthorpe, “Is There a Status Order in Contemporary British Society? Evidence
from the Occupational Structure of Friendship,” European Sociological Review 20, no. 5 (2004), pp. 383–391,
393–397. Used by permission of Oxford University Press.
 Chan and Goldthorpe—Is There a Status Order in Contemporary British Society? • 203

Lunt, 1941; Warner et al., 1949; Plowman et al., stratification that can be empirically as well
1962) have been seen as increasingly threatened as conceptually differentiated from class
by the mobility and anonymity of modern ‘mass’ structure?
society (Goldthorpe, 1978). (ii) In so far as this is so, what is the relative im-
On the other hand, though, sociolo- portance of these two forms of stratification
gists would seem to have become attracted to as determinants of individuals’ experience
‘one-dimensional’ understandings of social strat- and action in different areas of their social
ification, which discount or override the status/ lives—or, in other words, of the pattern of
class distinction as much out of methodological their life-chances and life-choices?
or theoretical predilections as of any respon-
siveness to actual social change. Thus, in North In the present chapter, our primary concern
America the popularity of the concept of ‘socio- is with the first of these questions which we take
economic status’ has clearly reflected the advan- up in the case of present-day British society.
tages, as demonstrated by Duncan and others, We aim to show that the question can in fact
of being able to treat stratification through a be answered positively, i.e. a status order can be
single continuous measure, such as the Duncan identified and one that, we suggest, maps onto
SEI (Duncan, 1961), in work using correlation the class structure in an intelligible, though far
and OLS regression techniques (e.g. Duncan from straightforward, way. In subsequent papers,
and Hodge, 1963; Blau and Duncan, 1967). we will then address the second question on the
In Europe, in contrast, a more important influ- basis of an ongoing research programme.
ence has probably been the theoretical efforts of
Bourdieu (1984) to ‘re-think’, and indeed over-
come, Weber’s ‘opposition’ of status and class: Methodological Approach
that is, by treating status as the symbolic aspect In our attempt to trace a status order in
of class structure, which is itself seen as not re- present-day Britain we follow the approach pi-
ducible to ‘economic’ relations alone. For Bour- oneered in the USA by Laumann (1966, 1973;
dieu and his followers attention thus focuses on Laumann and Guttman 1966) in seeking to
the lifestyle or underlying habitus that is specific move beyond ‘small town’ studies of social strat-
and distinctive to different classes (or ‘class frac- ification such as those of Warner and his asso-
tions’) rather than, as in Weber’s case, on ‘the ciates previously cited. Laumann accepted that
most varied ways’ in which class and status, and differential association could be taken as a key
thus lifestyle, may be contingently related to indicator of status. He recognised, however, that
each other. in the urban ‘mass’ society of the later twentieth
However, while the developments in question century individuals’ associational networks were
cannot be lightly disregarded, they do not, in our not restricted by the boundaries of local commu-
view, amount in themselves to a demonstration nities but could extend over a wide geographical
that, in the study of contemporary societies, the area. Thus, in studying differential association,
distinction between status and class as quali- ethnographic work, reliant on participant obser-
tatively different forms of stratification is now vation, would need to give way to survey-based
either empirically outmoded or conceptually re- research of a more spatially extensive kind.
dundant. In particular, the two following issues More specifically, Laumann proceeded by
remain to be determined: collecting information from samples of urban
(i) Can status orders still be identified in mod- populations on respondents’ own occupations
ern societies, even if of a less overt, less and on the occupations of other individuals
sharply demarcated and less localised kind within respondents’ more immediate social net-
than previously existed, and as a form of works. For different occupational groupings of
204 • Part III: The Structure of Inequality

respondents, the occupational distributions of unlikely—which, from our point of view, is the
reported associates could then be established crucial consideration.
and, in turn, the extent to which these distribu- In this regard, it is relevant to refer to the
tions differed from one grouping to another. Fi- work of Prandy and his associates, who have
nally, by using the (at the time) novel technique also sought to apply Laumann’s approach to the
of multidimensional scaling, the dimensionality British case. In their initial work in construct-
of these differences could be investigated, and ing a stratification scale, the ‘Cambridge Scale’,
the further questions addressed of whether a (Stewart et al., 1973, 1980), this group also rely
dimension emerged that could be plausibly on occupationally linked data on friendship, but
interpreted as one of status and, if so, of what define friendship in a deliberately loose way so
ordering occupational groupings took along as to allow even relatively transient associations
this dimension. In his earlier work, it should be to be included. However, for the purposes of a
added, Laumann sought information on a rela- revision and updating of the scale (Prandy and
tively wide range of respondents’ associates, in- Lambert, 2003), they then abandon analysis of
cluding kin, neighbours and friends; but in later data of this kind in favor of data on the (current)
work he made what for us is a significant change occupations of married couples, mainly because
in concentrating on friends alone. they can then draw on samples of census data
Laumann’s approach depends on two basic and gain the advantage of working with large
assumptions, both of which we believe are de- numbers of cases.
fensible. The first is that in modern societies oc- Whether the revised ‘CAMSIS’ scale that is
cupation is one of the most salient characteristics based on such marriage data is essentially the
to which status attaches. That this is indeed the same as the original Cambridge scale based on
case would, at all events, appear to be a matter of friendship data, as Prandy and Lambert claim, is,
some consensus among sociologists of otherwise in our view, debatable; and they indeed concede
often divergent views (e.g. Blau and Duncan, that ‘[c]ertainly, the strength of the relationship
1967; Parkin, 1971; Coxon and Jones, 1978; between marriage partners is less, statistically
Stewart et al., 1980; Bourdieu, 1984; Grusky speaking, than it is between friends’ (Prandy
and Sørensen, 1998). and Lambert, 2003: 401). But what must be
The second assumption is that recurrent as- recognised here is that a fundamental theoretical
sociation is a good indicator of a state of social difference exists between our position and that
equality between individuals and is, moreover, a of Prandy and his associates. In an early state-
better indicator of such equality, the ‘freer’ the ment, they assert that ‘the Weberian distinction
choice of associates and the closer or more in- of classes . . . from status groups . . . is neither
timate the association—which would appear to useful nor necessary’ (Stewart et al., 1980: 28).
be the reason for Laumann’s eventual decision More recently, Bottero and Prandy (2003: 180)
to work with data on friends only. Again, this maintain that ‘social interaction distance is taken
assumption would seem a plausible one, and, for as a stratification order in its own right’ and that
Britain at least, there is evidence to show that ‘research has tended to eliminate the distinction
differential association is in fact more marked between class and status, or the economic and
in the case of friends who are regarded as ‘close’ the cultural, which was once seen as central an-
than in the case of those who are simply fre- alytically to conventional stratification theory’.
quent leisure time companions (Goldthorpe, The implication then is that their scale reflects
1987: ch.7). While association with the latter ‘stratification arrangements’ in some quite gen-
may result primarily from shared interests and eral and undifferentiated sense. In contrast, we
activities (cultural, sporting, hobby, political would believe that the Weberian distinction be-
etc.), relations with the former are more likely tween class and status is conceptually clear and
to be based on ‘pure’ sociability, thus making so- potentially highly important; that it cannot be
cial inequality within the relationship especially rejected by fiat; and that whether or not a status
 Chan and Goldthorpe—Is There a Status Order in Contemporary British Society? • 205

Table 21.1. Occupational Categories Used in the Analysis and Their Constituent Minor Occupational
Groups

Code Descriptive title OPCS MOGs %

GMA General managers and administrators 10, 13, 15 2.5
PDM Plant, depot and site managers 11, 14, 16 2.7
SM Specialist managers 12 2.7
MPS Managers and proprietors in services 17 4.4
OMO Managers and officials, not elsewhere classified 19 2.0
SET Scientists, engineers and technologists 20, 21 1.9
HP Higher professionals 22, 24, 25, 26, 27, 29 3.3
TPE Teachers and other professionals in education 23 4.5
API Associate professionals in industry 30, 31, 32, 33, 39 3.9
APH Associate professionals in health and welfare 34, 37 4.8
APB Associate professionals in business 35, 36, 38 2.6
AOA Administrative officers and assistants 40 2.1
NCC Numerical clerks and cashiers 41 3.7
FRC Filing and record clerks 42 1.9
OCW Other clerical workers 43 3.5
SDC Store and dispatch clerks 44, 49 2.1
SEC Secretaries and receptionists 45, 46 3.3
SMC Skilled and related manual workers in
construction and maintenance 50, 52 3.5
SMM Skilled and related manual workers in metal trade 51, 53, 54 3.5
SMO Skilled and related manual workers not elsewhere
classified 55, 56, 57, 58, 59 3.9
PSP Protective service personnel 60, 61 1.9
CW Catering workers 62 2.3
PSW Personal service workers 63, 66, 67, 69 2.2
HW Health workers 64 2.6
CCW Childcare workers 65 2.6
BSR Buyers and sales representatives 70, 71 1.6
SW Sales workers 72, 73, 79 6.3
PMO Plant and machine operatives 80, 81, 82, 83, 84, 85, 86, 89 6.2
TO Transport operatives 87, 88 3.3
GL General labourers 90, 91, 92, 93, 99 2.2
RWS Routine workers in services 94, 95 6.1

order is still identifiable in contemporary British, there to be a common status order for men
or any other, society must be treated as an em- and women together rather than two separate,
pirical question. gender-specific orders. By including women, we
can then of course investigate how far this expec-
tation holds good.
Data and Analytical Techniques In wave 10 of the BHPS, respondents were
The data we use come from wave 10 (year 2000) asked to think of three people they considered
of the British Household Panel Study (BHPS). to be their closest friends. Information about
We restrict our analysis to respondents aged these friends, such as their age, sex, employment
20–64 but, unlike Laumann and most others status, and their relationship to respondents
who have subsequently taken his approach, we was recorded. For the first-mentioned friend,
include women, categorised on the basis of their respondents were also asked to report his or
own occupations. The idea of a status order her occupational title. Data on the current (or
from which we begin could be described as ‘gen- last) occupation of these ‘first’ close friends, to-
der neutral’. That is to say, one would expect gether with similar data on the current (or last)
206 • Part III: The Structure of Inequality

occupation of respondents, form the basis of our

empirical analyses.
These data were coded to the three-digit unit
groups of the UK standard occupational classifi-
cation (OPCS, UK, 1990) and could thence be
allocated to the 77 two-digit minor occupational
groups (MOGs). Although the BHPS affords where xAm and xBm are the coordinates of points
us a relatively large sample size (n = 9160), the A and B in the mth dimension, and Į and ȕ are
MOGs were still a more detailed classification parameters.
than we could sensibly employ. A 77 × 77 con-
tingency table would, on average, have had less
than two observations per cell. Some collapsing Results
of the MOGs was therefore necessary. Opera- In Figure 21.1 we report stress-values from our
tionally, we combined MOGs according to their MDSCAL analyses. It can be seen that, if men
functions, while taking into account their rela- and women are taken together in the analy-
tive size and the work milieu of their constituent sis, the three-dimensional solution achieves a
occupations at the same time. In the end, we stress-value of 0.075, indicating a rather good
worked with the 31 categories of Table 21.1. As fit. However, if men and women are analyzed
can be seen from the last column of the table, separately, the fit is generally worse, requiring, in
each of these categories accounts for between 1.6 the case of men, a four-dimensional solution to
and 6.3 percent of our respondents. obtain a stress value that falls below 0.10. Since
Our basic data arrays are therefore 31 × 31 it is difficult to visualise a space of four, or more,
contingency tables, for men and women sepa- dimensions, we would incline here not to go be-
rately and together, in which respondent’s oc- yond the three-dimensional solution.
cupational group is related to the occupational So what, then, are the main features of our fa-
group of respondent’s ‘first’ close friend. We voured MDSCAL solution? The panels of Figure
subject these contingency tables to multidimen- 21.2 show the positions of our 31 occupational
sional scaling (MDSCAL), the technique used categories as projected onto the three planes in-
by Laumann and most of those following him. volved—with men and women included in the
Our MDSCAL exercise proceeds as follows. analysis together.
We first generate ‘outflow’ rates from our con-
Figure 21.1. Stress Values of Multidimensional
tingency tables, i.e. the percentage distributions Scaling Applied to 2000 BHPS Best Friend Data,
of friends across our occupational categories for Using Data for All, and for Male and Female
each category of respondent. We then compute Respondents Separately
the index of dissimilarity for each pair of rows
of outflow rates. This gives us a measure of the
between-category dissimilarity, į. We then use
the half-matrix of ds as input to the simplest
form of MDSCAL analysis.
That is, we seek to represent our 31 occupa-
tional categories as points in a Euclidean space,
such that the distance between category A and
category B in this space, dAB, best approximates
the observed dissimilarity between the two cate-
gories įAB. Formally, this idea can be represented
as follows:
 Chan and Goldthorpe—Is There a Status Order in Contemporary British Society? • 207

The first dimension, shown horizontally in Figure 21.2. The Three Dimensional MDSCAL Solu-
the first and second panels of Figure 21.2 is, tion Projected onto Three Basic Planes
at least prima facie, that which captures status.
Thus, at the left extreme of the dimension are
located six categories comprising manual occu-
pations, while moving in from the right extreme
come the categories of Higher professionals,
Associate professionals in business, Specialist
managers, Teachers and other professionals in
education, General managers and administra-
tors, Associate professionals in industry, and
Scientists, engineers and technologists. In be-
tween, the categories can be seen as ordered ac-
cording to what might be described as a manual/
non-manual continuum that we discuss further
below.
In this respect, our findings are in fact much
in line with those of Laumann and others.
That is to say, these earlier investigators have
also found a first dimension in their MDSCAL
analyses that is likewise interpretable as reflect-
ing status. However, it has been a recurrent
finding in previous work that while more than
one dimension is required in order to obtain
a well-fitting MDSCAL solution, the further
dimensions introduced have not been open to
interpretation, or only in a rather speculative
way (e.g. see Laumann, 1966: 102–104, 1973:
79–80; Pappi, 1973; Stewart et al., 1980: 41–44;
Prandy, 1998).
In fact, it turns out that this second dimen-
sion can be interpreted rather convincingly—
and in a way that follows directly from our
inclusion of women in the analysis. It can be
taken as a dimension that expresses the degree to
which our 31 occupational categories are charac-
terised by sex segregation. Thus, looking at the
first panel of Figure 21.2, where the dimension
is shown vertically, one finds at the top occupa-
tional categories in which men predominate—
Plant, depot and site managers, Skilled and
related manual workers in metal trades, Skilled
and related manual workers in construction
and maintenance, and Scientists, engineers and
technologists; while at the bottom come catego-
ries in which women predominate—Childcare
 208 • Part III: The Structure of Inequality

workers, Health workers, Sales workers, Routine status, dimension for men and women are plot-
workers in services, and Secretaries and recep- ted against each other. As can be seen, there is
tionists. We can check this interpretation of our a quite high correlation (r = 0.75). Moreover,
second dimension by relating the scores of our the two obvious ‘outlier’ categories, which are
categories on it to their sex composition (i.e. per- identified, have one feature in common that
centage female) within our sample. We obtain a suggest that their significance should not be
correlation of r =0.92. over-estimated: they fall towards the extremes of
If, then, we do capture here the effects on the second dimension (see Figure 21.2), i.e. they
the occupational structure of friendship of dif- are markedly sex-segregated. Only 2.3 percent
ferences in work environments according to the of the Secretaries and receptionists are men, and
degree to which they constitute male or female only 3.1 percent of the Skilled and related man-
‘worlds’—the effects of, say, the engineering ual workers in construction and maintenance are
shop, building site or maintenance department women. We would then believe that these dis-
as against those of the care home, supermarket crepant scores are likely to come about mainly by
floor or word-processing room—an important chance, because respondents of the ‘minority’ sex
implication follows. Because our second di- are too few in number to allow reliable results.
mension in this way represents the gendering From this point of view, the outlier categories
of opportunities for friendship formation, our are, at least to some extent, artefacts of inescap-
first dimension, being thus ‘purified’ of this in- able limitations of our study in terms of sample
fluence, should, if it does indeed reflect actual size; and in turn, we would regard Figure 21.3 as
friendship choice as influenced by status, apply giving quite strong support to the idea that the
to men and women alike. For, as we earlier ob- first dimension of our three-dimensional MD-
served, the conception of a status order with SCAL solution does reflect a status order that
which we operate entails gender-neutrality. We is, as it should be, common to men and women.
can easily check whether or not this expectation Finally here we should comment briefly on
is met by turning to our MDSCAL analyses for the third dimension of our MDSCAL solution.
men and women separately. It should be recognised that the dimensions of
The relevant results are shown in Figure 21.3 MDSCAL solutions do not have to be inter-
in which category scores on the first, putatively preted in terms of some single factor: they may
Figure 21.3. First Dimension MDSCAL Scores simply ‘mop up’ a number of different effects.
Estimated Separately for Men and for Women Plotted With this point in mind, it may be noted that on
Against Each Other the third dimension—as a detailed examination
of the plots of Figure 21.2 will confirm—sev-
eral categories come into close proximity whose
members will tend to occupy similar occupa-
tional situses (Morris and Murphy, 1959): for ex-
ample, Plant, depot and site managers, our three
categories of Skilled and related manual workers
and Plant and machine operatives; or Managers
and proprietors in services and Sales workers,
Catering workers and Personal service workers;
or again, Associate professionals in health and
welfare and Health workers and Childcare work-
ers. In so far, then, as the third dimension, like
the second, even if in a less straightforwardly
interpretable way, does capture features of the
occupational structure of friendship that derive
 Chan and Goldthorpe—Is There a Status Order in Contemporary British Society? • 209

primarily from the different opportunities for that cover occupations, falling mostly within the
friendship that are offered by particular work services sector, that tend to some significant de-
environments, rather than from actual friend- gree to have both non-manual and manual com-
ship choices expressing status considerations, ponents—the former usually involving some
we are further encouraged to believe that in our kind of ‘people processing’. And, finally, cate-
first dimension it is status per se that is primarily gories 26–31 comprise occupations that require
reflected. the performance of predominantly manual tasks,
in effect working with things rather than with
either symbols or people.
The Status Dimension in More Detail The second feature of the ranking that
We now turn to a more detailed examination of we would pick out relates specifically to its
what we would take to be the status dimension non-manual range—i.e. categories 1–18. It can
of our MDSCAL analyses, although, as will be be seen that within this range there is a tendency
seen, with a continuing concern for the validity for managerial categories to rank lower than do
of this interpretation. This concern is impor- professional categories. It is true that the two
tant since the ordering of categories along any highest ranking categories, Higher professionals
MDSCAL dimension is in itself indicative only and Associate professionals in business, are fol-
of their relative closeness or separation accord- lowed in third place by a managerial category,
ing to some metric. Thus, it could be suggested Specialist managers. But individuals falling in
that our first dimension reflects no more than this latter category are more likely than other
homophily—the tendency for people to make managers to have professional qualifications
friends with others like themselves. To sustain and to be operating to some extent in a profes-
the stronger claim that it captures a hierarchical sional role. Following Teachers and other pro-
ordering by status, further evidence is called for, fessionals in other professionals in education,
external to that of the MDSCAL analysis. General managers and administrators then rank
To begin with, we show in Table 21.2 the fifth, but other managers and officials rank only
rank-ordering of our 31 categories on the puta- ninth, below in fact the highest ranking cleri-
tive status dimension and, to make our discus- cal category, and the two remaining managerial
sion somewhat more concrete, we also identify categories, Managers and proprietors in services
‘representative’ occupations: that is, occupations and Plant, depot and site managers are found to-
that account for relatively large numbers of in- gether at the very end of the non-manual range.
dividuals within each category and at the same These two features of the ranking we would
time give some idea of its range. There are two regard as having particular significance in that
features of the rank-ordering to which we would they lend support to the idea that it does indeed
draw attention. express status rather than homophily alone. This
The first, on which we have in fact already is so because both can be taken as providing
commented, is that the categories can be seen as rather clear ‘echoes’ of the relatively explicit and
ordered overall according to the degree of ‘man- well-defined status order that prevailed in British
uality’ of the work involved in their constitu- society from, say, the later nineteenth through
ent occupations. More specifically, occupations to the mid-twentieth century. Sociologists and
in categories 1–7 in the ranking are essentially historians would appear largely to concur (for
non-manual in character, and categories 8–18, useful reviews see Runciman, 1997: 153–163,
only slightly less so. Or, somewhat more specif- 212–229; McKibbin, 2000: chs I–IV) that the
ically, one might say that these occupations re- non-manual/manual distinction marked a major
quire working predominantly with symbols and/ boundary within this order, and one that was
or people rather than directly with inanimate strongly upheld even in fact as the distinction
material entities. Categories 19–25 are then ones became less consequential in terms of economic
210 • Part III: The Structure of Inequality

Table 21.2. The 31 Occupational Categories Ranked by Status Score and Representative Occupations
Within Each Category
g y
Code Representative Occupations
1 HP Chartered accountants, clergy, medical practitioners, solicitors
2 APB Journalists, investment analysts, insurance brokers, designers
3 SM Company treasurers, financial managers, computer systems managers, personnel
managers
4 TPE College lecturers, education officers and inspectors, school teachers
5 GMA Bank and building society managers, general managers in industry, national and local
government officers
6 API Computer analysts and programmers, quantity surveyors, vocational and industrial
trainers
7 SET Civil and structural engineers, clinical biochemists, industrial chemists, planning
engineers, software engineers
8 FRC Conveyancing clerks, computer clerks, library assistants
9 OMO Security managers, cleaning managers
10 AOA Clerical officers in national and local government
11 NCC Accounts assistants, bank clerks
12 APH Community workers, nurses, occupational therapists, youth workers
13 SEC Personal assistants, receptionists, secretaries, word processor operators
14 OCW General assistants, commercial and clerical assistants
15 BSR Buyers and purchasing officers, technical sales representatives, wholesale representatives
16 CCW Educational assistants, nursery nurses
17 MPS Catering managers, hoteliers, publicans, shopkeepers and managers
18 PDM Clerks of works, farm managers, maintenance managers, transport managers, works
managers
19 SW Cash desk and check-out operators, sales and shop assistants, window dressers
20 HW Ambulance staff, dental nurses, nursing auxiliaries
21 PSW Caretakers and housekeepers, hairdressers and beauticians, travel attendants, undertakers
22 PSP Fire service and police officers, security guards
23 RWS Car park attendants, cleaners, counter-hands, couriers and messengers, hotel porters,
postal workers
24 CW Bar staff, chefs, cooks, waiters and waitresses
25 SDC Despatch and production control clerks, storekeepers
26 SMO Gardeners and groundsmen, printers, textile workers, woodworkers
27 TO Bus and coach drivers, lorry and van drivers, taxi drivers
28 SMC Bricklayers, electricians, painters and decorators, plasterers, roofers, telephone repairmen
29 SMM Fitters, setters, setter-operators, sheet metal workers, turners, welders
30 PMO Assemblers, canners, fillers and packers, food processors, moulders and extruders, routine
inspectors and testers
31 GL Agricultural workers, factory labourers, goods porters, refuse collectors

conditions; and, further, that professional em- occupations that have concurrently expanded
ployment was generally regarded as being so- are ones to which the distinction does not all
cially superior to managerial employment, and that easily apply. As we have observed, categories
especially to managerial employment in industry 19–25 are largely made up of occupations in ser-
or ‘trade’. vices that involve both non-manual and manual
At the same time, though, it should also be work, and official statistics would indicate (cf.
recognised that, today, the non-manual/man- Table 21.1 and also Jackson, 2002) that these
ual distinction is in itself less clear-cut than it occupations now account for as much as a quar-
was previously, chiefly as a result of the growth ter of the total employed population. Even if,
of the services sector of the economy. Many then—as we would believe to be the case—the
 Chan and Goldthorpe—Is There a Status Order in Contemporary British Society? • 211

degree of ‘manuality’ of work remains an im-

portant influence on social status, it would seem Status and Class
likely that the resulting lines of division will now Having now presented evidence to suggest that
be rather more blurred than they were half a cen- through the analysis of the occupational struc-
tury ago. ture of friendship a status order in contemporary
Finally, change in this regard may be in part British society can still be identified, at least in
related to one further feature of the ranking of its broad lines, we come finally to the issue that is
Table 21.2 that calls for some comment: that is, salient in the Weberian perspective that we have
the relatively low positions of the skilled manual adopted: that of the relationship between status
worker categories. To some degree, this result and class. As we have already indicated, we take
may be artefactual in that the official occupa- the view that the distinction between class and
tional groupings from which these categories are status is, conceptually, a well-defined and coher-
constructed are not, despite their labels, drawn ent one and that its applicability and value in the
up according to skill in any very strict way context of present-day British society, as in any
(hence the ‘skilled and related . . . ’ formulation). other, has to be a matter for investigation, not
However, we would doubt if this is a factor of assertion. In this regard, therefore, a question of
major importance, and would emphasise, rather, immediate interest that we seek here to address
the real changes that have occurred over recent is that of how the status order we hope to have
decades in the nature and occupational distribu- identified in present-day Britain maps onto the
tion of skills, in consequence of developments class structure.
in economic structure, technology and organi- The Goldthorpe class schema, which treats
sation (cf. Gallie et al., 1998: ch. 2), and that class positions as being defined by social rela-
indeed underlie the difficulties that arise in using tions in economic life or, more specifically, by
earlier nomenclature. It is usually supposed (e.g. employment relations (Goldthorpe, 1997, 2000:
Roberts, 1971: ch. 1) that, under the ‘old’ status ch. 10) has been shown to possess an acceptable
order, skilled, ‘time-served’ craftsmen formed an level of criterion validity and also of construct
‘aristocracy of labour’ and ranked clearly above validity in being strongly predictive of individ-
semi- and unskilled manual workers and also uals’ economic security, stability and prospects
above the typical service works of the day—do- (Evans and Mills, 1998; Goldthorpe and Mc-
mestics, shop hands etc. But the fact that we do Knight, 2004). The nine-class version of this
not reproduce this pattern has to be understood, schema is displayed in Table 21.3. Figure 21.4
we would suggest, in relation to the declining then gives a first indication of how status is dis-
demand for, or dilution of, many traditional tributed within and between classes. It is evident
craft skills at the same time as new kinds of skill that there is a status gradient across classes, as
with different sectoral and occupational link- might be expected. In terms of the median or
ages have emerged—including technical skills interquartile range of the status of their mem-
as, say, in connection with computerisation, bers, the non-manual classes (I, II and IIIa) rank
but also communication and ‘social’ skills more clearly above the manual classes (V, VI and VII);
generally. and, within the non-manual classes, the median
In sum, while the present-day status order status of members of Class I is above that of
that is suggested by our empirical analyses can members of Class II who in turn rank above the
claim to show continuities with that of an earlier members of Class IIIa.
period that, we believe, can scarcely be dismissed However, it also appears from Figure 21.4
as coincidental, this order would at the same that the spread of status within classes is often
time appear to be in several respects less sharply quite considerable and that there is a good deal
demarcated than previously, as well as being less of overlap in status between classes, both in
openly recognised and acknowledged. the case of the non-manual and manual classes
 212 • Part III: The Structure of Inequality

Table 21.3. The Goldthorpe Class Schema (Nine-Class Version)

Class Description
I Professional, administrative and managerial employees, higher grade; large employers
II Professional, administrative and managerial employees, lower grade; technicians, higher grade
IIIa Routine non-manual employees, higher grade
IIIb Routine non-manual employees, lower grade
IVac Small employers (other than professionals) including farmers
IVb Self-employed workers (other than professionals)
V Technicians, lower grade; supervisors of manual workers
VI Skilled manual workers
VII Non-skilled (i.e., semi- and unskilled) manual workers

considered separately and across the non-manual/ for these possibilities, the lack of congruence be-
manual divide. These features of Figure 21.4, we tween status and class has still to be regarded as
should recognise, may well be in some degree far from negligible, and likewise the differences
artefactual—most obviously as the result simply in the degree of such congruence from one re-
of any measurement error in regard to both sta- gion of Figure 21.4 to another.
tus and class. Further, as we have acknowledged, To provide further information on the inter-
the categories on which our status ordering is relation between status and class, and especially
based are by no means as refined as we would on status stratification within classes, we show
ideally wish. Thus, when these categories are in Figure 21.5 the composition of each class in
related to classes, the category members repre- terms of our occupational categories as ordered
sented in one class may in fact have higher or by status.
lower status than those represented in another. To begin with the salariat, it could be said
In other words, classes may pick up variation in that in Class I status stratification is quite lim-
status within categories in a systematic way. But, ited: 77 percent of those in this class are in fact
even with all reasonable allowance being made covered by the seven highest categories in the
status order. Moreover, so far as the remainder
Figure 21.4. Distribution of Social Status Within and are concerned, the point made above concerning
Between Classes artefactual effects could well apply. For example,
we know that the Plant, depot and site managers
who are the main discrepant—i.e. relatively low
status—category within Class I will be employed
in large establishments and individuals in this
subset of the category may then have higher sta-
tus than their counterparts employed in small
establishments and allocated to Class II. When
we turn to Class II itself, however, stratification
by status is far more extensive and would seem
less likely to be of an artefactual kind. From Fig-
ure 21.5 three broad status levels can in fact be
identified. First comes a grouping of profession-
als and specialist managers, secondly, one of ad-
ministrative officials and associate professionals
in health, and thirdly, one in which managers in
Note: The boxes are drawn with width proportional to the
square-root of n of the classes. Because of the small n of classes IVc industry and services predominate, i.e. those oc-
and VIIb, they are collapsed with classes IVa and VIIa respectively. cupational groupings that, as we have seen tend
 Chan and Goldthorpe—Is There a Status Order in Contemporary British Society? • 213

Figure 21.5. Distribution of Respondents by Social Class, and by Occupational Group within Social
Class

Note: The occupational groups are numbered according to their status ranking.

to have low status relative to both their income occupational categories that are in fact neigh-
and educational qualifications. bours in our status order.
With the classes of routine non-manual em- Turning next to the two classes of ‘indepen-
ployees, Classes IIIa and IIIb, we again find only dents’, IVac and IVb, status stratification is in
rather limited status stratification. In IIIa, two these cases quite marked, as might be expected
status levels might perhaps be distinguished— given the occupational range that is covered. In
the higher comprising routine non-manual both cases alike, Figure 21.5 points to two main
employees working in predominantly adminis- groupings, the higher comprising those running
trative contexts, and the lower, such employees largely service or small industrial enterprises and
working in sales and services. But in IIIb two the lower involved in enterprises entailing various
thirds of those in the class fall within just three kinds of mainly artisanal, that is, manual work.
214 • Part III: The Structure of Inequality

Finally, with the ‘blue-collar’ classes, V, VI small scale of enterprises and establishments and
and VII, the finding of variable degrees of sta- perhaps also because, as we have previously re-
tus stratification persists. Such stratification is marked, in work in this sector the manual/
least apparent in Class VI, that of skilled manual non-manual division is often rather blurred.
workers, with 85 percent of those in the class However, for our present purposes, the out-
being accounted for by five neighbouring cate- come of main importance here is that the map-
gories in the status order. In contrast, in Classes ping of status onto class does not appear to be so
V and VII status stratification appears far more close as to make further research into their in-
extensive. The main line of division that is in- terrelation and its consequences unduly difficult
dicated in Class V is that between technicians and, at the same time, rather pointless. There are
and supervisors of service workers, on the one at least certain ‘regions’ in which significant dis-
hand, and supervisors of manual workers outside junctions between class position and status level
of the services sector, on the other; and in Class would appear to occur.
VII, a somewhat similar division is apparent be-
tween unskilled manual workers in services and
those employed in manufacturing, construction, Conclusions
transport etc. In this paper we have been concerned, as a first
To revert, then, to the question of how closely step in a larger research programme, with inves-
the status order of contemporary British society tigating how far in contemporary British society
maps onto the class structure, what our inves- it is still possible to identify a status order, de-
tigations thus far would lead us to say is that, spite an evident decline in deference and in the
if we view the matter in terms of status strat- readiness of individuals to openly assert their so-
ification within classes, the mapping is much cial superiority over others. Using the approach
closer in some cases than it is in others. More pioneered by Laumann, and focusing on the oc-
specifically, status homogeneity appears rela- cupational structure of close friendship—within
tively high in Class I, the higher division of the which social equality can be supposed—we have
salariat, in Classes IIIa and IIIb, those of routine presented empirical analyses to show that there
non-manual employees, and in Class VI, that is one dimension of this structure that can be
of skilled manual workers. But in the remain- plausibly interpreted as reflecting a hierarchy
ing classes status stratification would appear far of status and one that is, as it should be, essen-
from negligible. tially ‘gender-neutral’. We attach particular sig-
Why such a pattern should exist calls for nificance to the fact that this hierarchy displays
further inquiry, although one pointer to emerge clear continuities with that depicted for the later
from the foregoing, relating to the services sec- nineteenth and earlier twentieth centuries in his-
tor, may be noted. In this sector, and especially torical and earlier sociological research, although
in sales and personal services, ‘white-collar’, that there are also indications that the present-day hi-
is, managerial and other non-manual, workers erarchy is less sharply demarcated. This would
would appear to have relatively low status, as appear to be largely the result of the growth
is seen within Class II (and also perhaps Class of occupations, especially within the services
IIIa), while independents, and blue-collar super- sector of the economy, to which the manual/
visory and manual workers have relatively high non-manual distinction does not easily apply.
status, as is seen within Classes IVac, IVb, V and Also, as regards status and class, we have ar-
VII. The suggestion then is that one source of gued that it is important to treat these as two
status stratification within classes may lie in a distinct concepts and then to consider as an em-
tendency for the occupational status hierarchy pirical question in what way, in any particular
within the services sector to be more compressed society at any particular time, the status order
than elsewhere—perhaps because of the typically and the class structure relate to each other. So
 Chan and Goldthorpe—Is There a Status Order in Contemporary British Society? • 215

far as present-day Britain is concerned—and as- Coxon, A. and Jones, C. (1978). The Images of Occu-
suming of course that a status order does indeed pational Prestige. London, UK: Macmillan.
exist broadly on the lines we have claimed—we Dahrendorf, R. (1959). Class and Class Conflict in
find that while some classes show a rather high Industrial Societies. London, UK: Routledge &
Kegan Paul.
degree of status homogeneity, in others the ex-
Duncan, O. D. (1961). A socioeconomic index for
tent of status stratification is quite extensive. all occupations. In Reiss, A. (Ed.), Occupations
And in this connection also we have raised the and Social Status. New York, NY: Free Press.
possibility that the growth of the services sector Duncan, O. D. and Hodge, R. W. (1963). Educa-
may play a significant role. tion and occupational mobility. American Journal
Following our programme, we are presently of Sociology, 68, 629–644.
addressing the second question that we initially Evans, G. and Mills, C. (1998). Identifying
posed: that of the relative importance of status class structure: A latent class analysis of the
and class as determinants of individuals’ expe- criterion-related and construct validity of the
Goldthorpe class schema. European Sociological
rience and action in different domains of their
Review, 14, 87–106.
social lives. Thus, we are examining the rela- Gallie, D., White, M., Cheng, Y. and Tomlinson,
tionship between the positions individuals hold M. (Eds.) (1998). Restructuring the Employment
within the status order, as we would envisage it, Relationship. Oxford, UK: Clarendon Press.
and their cultural tastes and preferences and, in Giddens, A. (1973). The Class Structure of the Ad-
particular, in areas where ‘high’ and ‘low’ tastes vanced Societies. London, UK: Hutchinson.
and preferences are widely recognised as mark- Goldthorpe, J. H. (1978). The current inflation: to-
ing out distinctive lifestyles. To the extent that in wards a sociological account. In Hirsch, F. and
this regard we can show that the effects of status Goldthorpe, J. H. (Eds), The Political Economy
of Inflation. London, UK: Martin Robertson,
clearly outweigh those of class—and our prelim-
186–214.
inary results indicate that they in fact do—then
———. (1987). Social Mobility and Class Structure
the validity of the status order that we have pro- in Modern Britain, second edn. Oxford, UK:
posed will be further confirmed. And moreover, Clarendon Press.
to the extent that we can in other domains—as, ———. (1997). The ‘Goldthorpe’ class schema. In
for example, those of experience in economic Rose, D. and O’Reilly, K. (Eds), Constructing
life and in turn the perception of economic and Classes. London, UK: Office of National Statistics.
political interests—show that the reverse is the ———. (2000). On Sociology: Numbers, Narratives,
case—i.e. that class is a more potent force than and the Integration of Research and Theory. Ox-
ford, UK: Oxford University Press.
status—then the value of a Weberian perspective
Goldthorpe, J. H. and McKnight, A. (2004). The
will be underlined and the way opened up to
economic basis of social class. CASEpaper 80, Lon-
an understanding of the form of stratification of don School of Economics.
modern societies more consistent with the com- Grusky, D. B. and Sørensen, J. B. (1998). Can class
plexity that, we would believe, it does indeed analysis be salvaged? American Journal of Sociol-
display. ogy, 103, 1187–1234.
Hirsch, F. and Goldthorpe, J. H. (Eds.) (1978). The
REFERENCES Political Economy of Inflation. London, UK: Mar-
Blau, P. M. and Duncan, O. D. (1967). The Ameri- tin Robertson, 186–214.
can Occupational Structure. New York, NY: John Jackson, M. (2002). Explaining Class Mobility: Mer-
Wiley and Sons. itocracy, Education and Employers. DPhil thesis,
Bottero, W. and Prandy, K. (2003). Social interac- University of Oxford, UK.
tion distance and stratification. British Journal of Laumann, E. O. (1966). Prestige and Association
Sociology, 54, 177–197. in an Urban Community. Indianapolis, IN:
Bourdieu, P. (1984). Distinction: A Social Critique of Bobbs-Merrill.
the Judgement of Taste. London, UK: Routledge ———. (1973). Bonds of Pluralism. New York, NY:
& Kegan Paul. Wiley.
216 • Part III: The Structure of Inequality

Laumann, E. O. and Guttman, L. (1966). The rel- Plowman, D., Minchinton, W. and Stacey, M.
ative associational contiguity of occupations in (1962). Local social status in England and Wales.
an urban setting. American Sociological Review, Sociological Review, 10, 161–202.
31, 169–178. Prandy, K. (1998). Deconstructing classes. Work,
Lockwood, D. (1958). The Blackcoated Worker: A Employment and Society, 12, 743–753.
Study in Class Consciousness. London, UK: Allen Prandy, K. and Lambert, P. S. (2003). Marriage, so-
& Unwin. cial distance and the social space: An alternative
Marshall, T. (1953/1963). Sociology at the Crossroads derivation and validation of the Cambridge scale.
and Other Essays. London, UK: Heinemann. Sociology, 37, 397–411.
McKibbin, R. (2000). Classes and Cultures: England Roberts, R. (1971). The Classic Slum. Harmonds-
1918–1951. Oxford, UK: Oxford University Press. worth, UK: Penguin.
Mills, C. W. (1951). White Collar: The American Runciman, W. G. (1997). Applied Social Theory, Vol.
Middle Classes. New York, NY: Oxford Univer- 3 of a Treatise on Social Theory. Cambridge, UK:
sity Press. Cambridge University Press.
Morris, R. and Murphy, R. (1959). The situs dimen- Stewart, A., Prandy, K. and Blackburn, R. M.
sion in occupational structure. American Sociolog- (1973). Measuring the class structure. Nature,
ical Review, 24, 231–239. 245 (5426), 415–417.
OPCS, UK (1990). Standard Occupational Classifi- ———. (1980). Social Structure and Occupations.
cation, Vol. 1. London: HMSO. London, UK: Macmillan.
Pappi, F. (1973). Sozialstruktur und soziale Schich- Warner, W. and Lunt, P. (1941). The Social Life of a
tung in einer Kleinstadt mit heterogener Bev- Modern Community. New Haven, CT: Yale Uni-
olkerung. In Friedrichs, J., Mayer, K. and versity Press.
Schluchter, W. (Eds). Soziologische Theorie und Warner, W., Meeker, M. and Eells, K. (1949). Social
Empirie. Opladen: Westdeutschen Verlag. Class in America. Chicago, IL: Science Research
Parkin, F. (1971). Class Inequality and Political Order. Associates.
London, UK: MacGibbon and Kee. Weber, M. (1922/1968). Economy and Society.
Berkeley and Los Angeles, CA: University of
California Press.
 Durkheim—The Division of Labor in Society • 217

Durkheimian Theories of Class

22 • Emile Durkheim

The Division of Labor in Society

In The Division of Labor in Society, we empha- that opinion shows itself to be about the way in
sise the state of legal and moral anomie in which which such vague obligations are fulfilled. Those
economic life exists at the present time. In fact, actions most blameworthy are so often excused
in this particular sphere of activity, professional by success that the boundary between the per-
ethics only exist in a very rudimentary state. missible and the prohibited, between what is just
There are professional ethics for the lawyer and and what is unjust, is no longer fixed in any way,
magistrate, the soldier and professor, the doctor but seems capable of being shifted by individuals
and priest, etc. Yet if we attempted to express in an almost arbitrary fashion. So vague a mo-
in somewhat more precise terms contemporary rality, one so inconsistent, cannot constitute any
ideas of what should be the relationship between kind of discipline. The upshot is that this entire
employer and white-collar worker, between the sphere of collective life is for the most part re-
industrial worker and the factory boss, between moved from the moderating action of any rules.
industrialists in competition with one another It is to this state of anomie that must be at-
or between industrialists and the public, how tributed the continually recurring conflicts and
imprecise would be the statements that we disorders of every kind of which the economic
could formulate! Some vague generalities about world affords so sorry a spectacle. For, since
the loyalty and commitment that employees of nothing restrains the forces present from react-
every kind owe to those who employ them, or ing together, or prescribes limits for them that
about the moderation that employers should they are obliged to respect, they tend to grow
manifest in exercising their economic superior- beyond all bounds, each clashing with the other,
ity, a certain condemnation of any competition each warding off and weakening the other. . . .
that is too blatantly unfair, or of any too glaring Political society as a whole, or the state,
exploitation of the consumer: this is almost the clearly cannot draw up the system of rules that
sum total of what the ethical consciousness of is now lacking. Economic life, because it is very
these professions comprises. Moreover, most of special and is daily becoming increasingly spe-
these precepts lack any juridical character. They cialised, lies outside their authority and sphere
are backed only by public opinion and not by of action. Activity within a profession can only
the law—and it is well known how indulgent be effectively regulated through a group close

Emile Durkheim. The Division of Labor in Society, with an introduction by Lewis A. Coser, and translated by
W. D. Halls, pp. xxxi, xxxii, xxxv, xxxix–xl, xlii–xlv, l–lv, lviii–lix. Introduction copyright © 1984 by Lewis A.
Coser. Translation copyright © 1984 by The Higher & Further Education Division, Macmillan Publishers Ltd.
Reprinted with the permission of The Free Press, a Division of Simon and Schuster Adult Publishing Group
and Palgrave Macmillan. All rights reserved.
218 • Part III: The Structure of Inequality

enough to that profession to be thoroughly cog- memories that survive of the corporations during
nisant of how it functions, capable of perceiving the ancien régime only confirm this impression.
all its needs and following every fluctuation in We incline to visualise them in the future as they
them. The sole group that meets these condi- were towards the end of their former existence,
tions is that constituted by all those working in intent above all on maintaining or increasing
the same industry, assembled together and or- their privileges and monopolies. We fail to see
ganised in a single body. This is what is termed a how such narrow vocational concerns might
corporation, or professional group. have any beneficial effect upon the morality of
Yet in the economic field the professional the corporation or its members.
group no more exists than does a professional However, we should refrain from extending
ethic. Since the last century when, not without to the entire corporative system what may have
reason, the ancient corporations were dissolved, been true of certain corporations during a very
hardly more than fragmentary and incomplete short period in their development. Far from the
attempts have been made to reconstitute them system having been, because of its very constitu-
on a different basis. Doubtless, individuals who tion, infected by a kind of moral sickness, during
are busy in the same trade are in contact with the greater part of its existence it played above all
one another by the very fact that their activities a moral role. This is especially evident with the
are similar. Competition with one another en- Roman corporation. ‘Among the Romans,’ de-
genders mutual relationships. But these are in no clares Walzing, ‘the corporations of artisans were
way regular; depending upon chance meetings, far from having so pronounced a professional
they are very often entirely of an individual na- character as in the Middle Ages. We come across
ture. One industrialist finds himself in contact no regulations concerning methods, no obliga-
with another, but the body of industrialists in tory apprenticeship, and no monopoly. Nor was
some particular speciality do not meet to act in their purpose to accumulate the capital necessary
concert. Exceptionally, we do see all members to exploit an industry.’1 Doubtless their associat-
of the same profession come together at a con- ing together gave them more power to safeguard
ference to deal with some problem of common the common interest, when the need arose. But
interest. But such conferences last only a short this was only one of the useful by-products that
while: they do not survive the particular circum- the institution engendered. It was not the justi-
stances that gave rise to them. Consequently the fication for its existence, nor its main function.
collective life for which they provided an oppor- Above all else, the corporation was a collegiate
tunity dies more or less entirely with them. . . . religious body. Each one possessed its own par-
Can we legitimately believe that corporative ticular god, who, when the means were available,
organisation is called upon to play in contempo- was worshipped in a special temple. Just as every
rary societies a more considerable part? If we deem family had its Lar familiaris and every city its
it indispensable it is not because of the services it Genius publicus, so every collegiate body had its
might render the economy, but on account of the protecting divinity, the Genius collegii. Naturally
moral influence it could exercise. What we partic- this professional form of worship was not with-
ularly see in the professional grouping is a moral out its festivities, and sacrifices and banquets
force capable of curbing individual egoism, nur- were celebrated in common together. Moreover,
turing among workers a more envigorated feeling all kinds of circumstances would serve as the oc-
of their common solidarity, and preventing the casion for festive gatherings; distribution of food
law of the strongest from being applied too bru- and money was often made at the expense of the
tally in industrial and commercial relationships. community. . . .
Yet such a grouping is deemed unfit for such a The facts cited adequately demonstrate that
role. Because it springs from temporal interests, it a professional grouping is not at all incapable of
can seemingly only serve utilitarian ends, and the exerting a moral effect. The very important place
 Durkheim—The Division of Labor in Society • 219

that religion held in its life highlights very partic- with others. But, as we have often had occasion
ularly the true nature of its functions, for in such to show,2 blood kinship has in no way the ex-
times every religious community constituted a traordinary effectiveness attributed to it. The
moral environment, just as every kind of moral proof of this is that in a large number of socie-
discipline necessarily tended to take on a reli- ties relations not linked by the blood tie are very
gious form. Moreover, this characteristic of cor- numerous in a family. Thus so-called artificial
porative organisation is due to the effect of very kinship is entered into very readily and has all
general causes which we can see at work in dif- the effects of natural kinship. Conversely, very
ferent circumstances. Within a political society, frequently those closely knit by ties of blood are
as soon as a certain number of individuals find morally and legally strangers to one another. For
they hold in common ideas, interests, sentiments example, this is true of blood kin in the Roman
and occupations which the rest of the popula- family. Thus the family does not derive its whole
tion does not share in, it is inevitable that, under strength from unity of descent. Quite simply, it
the influence of these similarities, they should is a group of individuals who have drawn close
be attracted to one another. They will seek one to one another within the body politic through a
another out, enter into relationships and associ- very specially close community of ideas, feelings
ate together. Thus a restricted group is gradually and interests. Blood kinship was able to make
formed within society as a whole, with its own such a concentration of individuals easier, for it
special features. Once such a group is formed, a naturally tends to have the effect of bringing dif-
moral life evolves within it which naturally bears ferent consciousnesses together. Yet many other
the distinguishing mark of the special conditions factors have also intervened: physical proximity,
in which it has developed. It is impossible for solidarity of interest, the need to unite to fight a
men to live together and be in regular contact common danger, or simply to unite, have been
with one another without their acquiring some causes of a different kind which have made peo-
feeling for the group which they constitute ple come together.
through having united together, without their Such causes are not peculiar to the family
becoming attached to it, concerning themselves but are to be found, although in different forms,
with its interests and taking it into account in within the corporation. Thus if the former group
their behaviour. And this attachment to some- has played so important a role in the moral his-
thing that transcends the individual, this subor- tory of humanity, why should not also the latter
dination of the particular to the general interest, be capable of so doing? Undoubtedly one differ-
is the very well-spring of all moral activity. Let ence will always exist between them, inasmuch
this sentiment only crystallise and grow more de- as family members share in common their entire
terminate, let it be translated into well-defined existence, whereas the members of a corporation
formulas by being applied to the most common share only their professional concerns. The fam-
circumstances of life, and we see gradually being ily is a kind of complete society whose influence
constituted a corpus of moral rules. extends to economic activity as well as to that
Domestic morality did not arise any differ- of religion, politics, and science, etc. Everything
ently. Because of the prestige that the family of any importance that we do, even outside the
retains in our eyes, if it appears to us to have home, has repercussions upon it and sparks off
been and continue to be a school of altruism an appropriate reaction. In one sense the cor-
and abnegation, the highest seat of morality, poration’s sphere of influence is more limited.
it is through the very special characteristics it Yet we must not forget the ever more important
is privileged to possess, ones that could not be place that our profession assumes in our lives as
found at any level elsewhere. We like to believe work becomes increasingly segmented. . . .
that in blood kinship there exists an extraordi- What past experience demonstrates above
narily powerful reason for moral identification all is that the organisational framework of the
220 • Part III: The Structure of Inequality

professional group should always be related to with local or regional needs, the regulations for
that of economic life. It is because this condition a profession. Thus economic activity could be
was not fulfilled that the system of corporations regulated and demarcated without losing any of
disappeared. Thus, since the market, from being its diversity.
municipal as it once was, has become national Moreover, we must reject the belief that the
and international, the corporation should as- corporation’s sole role should consist in laying
sume the same dimensions. Instead of being re- down and applying rules. It is undoubtedly true
stricted exclusively to the artisans of one town, that wherever a group is formed, a moral disci-
it must grow so as to include all the members of pline is also formed. But the institution of that
one profession scattered over the whole country,3 discipline is only one of the numerous ways in
for in whatever region they may be, whether they which any collective activity manifests itself. A
live in town or countryside, they are all linked to group is not only a moral authority regulating
one another and share a common life. Since this the life of its members, but also a source of life
common life is in certain respects independent sui generis. From it there arises a warmth that
of any territorial boundaries, a suitable organ- quickens or gives fresh life to each individual,
ism must be created to give expression to this which makes him disposed to empathise, caus-
life and to regulate its functions. Because of the ing selfishness to melt away. Thus in the past the
dimensions that it assumes, such an organism family has been responsible for legislating a code
should necessarily be closely in contact and di- of law and morality whose severity has often been
rectly linked with the central organism of the life carried to an extreme of harshness. But it has also
of the collectivity. Events important enough to been the environment where, for the first time,
affect a whole category of industrial enterprises men have learnt to appreciate the outpouring of
within a country necessarily have wide repercus- feeling. We have likewise seen how the corpora-
sions of which the state cannot fail to be aware. tion, both in Rome and during the Middle Ages,
This impels it to intervene. Thus for good reason created these same needs and sought to satisfy
the royal power tended instinctively not to leave them. The corporations of the future will be as-
large-scale industry outside its ambit as soon as it signed even greater and more complex functions,
appeared. It could not fail to take an interest in a because of their increased scope. Around their
form of activity which by its very nature is always purely professional functions will be grouped
liable to affect society as a whole. Yet such regu- others which at present are exercised by the com-
latory action, although necessary, should not de- munes and private associations. Among these are
generate into utter subordination, as happened functions of mutual assistance which, in order to
in the seventeenth and eighteenth centuries. The be entirely fulfilled, assume between helpers and
two organisms, although in contact with each helped feelings of solidarity as well as a certain
other, should remain distinct and autonomous; homogeneity of intellect and morals, such as that
each has functions that it alone can perform. If readily engendered by the exercise of the same
it falls to political assemblies to lay down the profession. Many educational activities (tech-
general principles for industrial legislation, they nical education, adult education, etc.) should
are not capable of diversifying them according also, it seems, find in the corporation their nat-
to the various types of industry. It is this diversi- ural habitat. The same is also true for a certain
fication that is the corporation’s proper task.4 A type of artistic activity. It would seem in accor-
unitary organisation over a whole country also dance with the nature of things that such a noble
in no way precludes the formation of secondary form of diversion and recreation should develop
organisations which include similar workers in alongside the more serious aspects of life, act-
the same region or locality. Their role could be ing as a balancing and restorative influence. In
to spell out even more specifically, in accordance fact we now already see trade unions acting at
 Durkheim—The Division of Labor in Society • 221

the same time as friendly societies, and others that we each belong to a commune or a dépar-
are setting up communal centres where courses tement, but the ties binding us to them become
are organised, and concerts and dramatic perfor- daily more loose and tenuous. These geograph-
mances held. Hence the activity of a corporation ical divisions are in the main artificial, and no
can take on the most varied forms. longer arouse deep emotions within us. The pro-
We may even reasonably suppose that the vincial spirit has vanished beyond recall. ‘Parish
corporation will be called upon to become the pump’ patriotism has become an anachronism
foundation, or one of the essential foundations, that cannot be restored at will. Strictly local or
of our political organisation. We have seen that, département matters hardly affect or enthrall us
although it first began outside the social system, either any longer, save in so far as they go hand
it tended to become more and more closely in- in hand with matters relating to our profession.
volved in it as economic life developed. We have Our activity extends much beyond these groups,
therefore every reason to anticipate that, if prog- which are too narrow for it; moreover, much of
ress continues on the same lines, the corporation what happens within them leaves us indifferent.
is destined to assume an ever more central and Thus what might be described as the sponta-
preponderant place in society. It was once the neous collapse of the old social structure has
elementary division of communal organisation. occurred. But this internal organisation cannot
Now that the commune, from being the auton- disappear without something taking its place. A
omous unit that it once was, has been absorbed society made up of an extremely large mass of
into the state just as the municipal market was unorganised individuals, which an overgrown
absorbed into the national market, may we not state attempts to limit and restrain, constitutes
legitimately think that the corporation should a veritable sociological monstrosity. For collec-
also undergo a corresponding transformation tive activity is always too complex to be capable
and become the elementary division of the of finding expression in the one single organ
state, the basic political unit? Society, instead of the state. Moreover, the state is too remote
of remaining what it is today—a conglomerate from individuals, its connections with them too
of land masses juxtaposed together—would be- superficial and irregular, to be able to penetrate
come a vast system of national corporations. The the depths of their consciousness and socialise
demand is raised in various quarters for electoral them from within. This is why, when the state
colleges to be constituted by professions and not constitutes the sole environment in which men
by territorial constituencies. Certainly in this can fit themselves for the business of living in
way political assemblies would more accurately common, they inevitably ‘contract out,’ detach-
reflect the diversity of social interests and their ing themselves from one another, and thus so-
interconnections. They would more exactly epit- ciety disintegrates to a corresponding extent. A
omise social life as a whole. Yet if we state that nation cannot be maintained unless, between the
the country, in order to become conscious of it- state and individuals, a whole range of second-
self, should be grouped by professions, is not this ary groups are interposed. These must be close
to acknowledge that the organised profession or enough to the individual to attract him strongly
the corporation should become the essential to their activities and, in so doing, to absorb him
organ of public life? into the mainstream of social life. We have just
In this way a serious gap in the structure of demonstrated how professional groupings are
European societies, and in our own in particular, fitted to perform this role, and how indeed ev-
would be filled. We shall see how, as history un- erything marks them out for it. Hence we can
folds, an organisation based on territorial group- comprehend how important it is, particularly in
ings (village, town, district or province, etc.) the economic sphere, that they should emerge
becomes progressively weaker. There is no doubt from that inchoate and disorganised state in
222 • Part III: The Structure of Inequality

which they have lain for a century, since profes- both sides to meet on the governing councils of the
sions of this kind today absorb the greater part corporation it is no less indispensable for them to
of the energies of society.5 constitute distinct and independent groups at the
lower level of corporative organisation, because too
NOTES often their interests vie with one another and are op-
1. Waltzing, Etude historique sur les corporations posing. To feel that they exist freely, they must be
profession chez les Romains, vol. I, p. 194. aware of their separate existence. The two bodies so
2. Cf. especially Année sociologique, vol. I, pp. constituted can then appoint their representatives to
313 ff. the common assemblies.
3. We need not discuss the international organi- 5. Moreover, we do not mean that territorial
sation which, because of the international character constituencies are destined to disappear completely,
of the market, would necessarily develop at a level but only that they will fade into the background.
above that of the national organisation. For at pres- Old institutions never vanish in the face of new
ent the latter alone can constitute a legal entity. In ones to such an extent that they leave no trace of
the present state of European law the former can themselves. They persist not only by the mere fact
only result from arrangements freely concluded be- of survival, but also because there persists some trace
tween national corporations. of the needs to which they corresponded. Material
4. This specialisation could not occur without proximity will always constitute a link between men.
the help of elected assemblies charged with repre- Consequently the political and social organisation
senting the corporation. In the present state of in- based on territory will certainly subsist. But it will
dustry, these assemblies, as well as those tribunals no longer enjoy its present predominance, precisely
entrusted with the task of applying the regulations because that link is losing some of its force. What is
of an occupation, should clearly include represen- more, we have shown above that, even at the base of
tatives of employees and employers, as is already the the corporation will still be found geographical di-
case with the industrial arbitration tribunals. The visions. Moreover, between the various corporations
proportion of each should correspond to the respec- from a same locality or region there will necessarily
tive importance attributed by public opinion to these be special relationships of solidarity which will, from
two factors of production. But if it is necessary for time to time, demand an appropriate organisation.

23 • Kim A. Weeden and David B. Grusky

The Changing Form of Inequality

The spectacular increase in income inequality income inequality has been accompanied by the
in the United States has drawn scholars to ex- rise of a gradational inequality regime in which
plore changes in the amount of inequality rather one’s life chances, lifestyles, and attitudes are
than possible changes in its form. We simply do increasingly defined by a simple income gradi-
not know, for example, whether the growth in ent. We likewise do not know whether we are

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Kim A. Weeden and David B. Grusky, “The Three Worlds of Inequality,” American
Journal of Sociology 117:6 (2012), pp. 1723–1785, published by the University of Chicago Press. The article
printed here was originally prepared by Kim A. Weeden and David B. Grusky for this fourth edition of Social
Stratification: Class, Race, and Gender in Sociological Perspective, edited by David B. Grusky. Copyright © 2014
by Westview Press.
 Weeden and Grusky—The Changing Form of Inequality • 223

instead moving toward a categorical inequality 2001; Weeden and Grusky 2005); and (3) a gra-
regime in which life conditions are increasingly dational world in which classes, either big classes
defined by membership in big social classes (e.g., or microclasses, take an increasingly hierarchical
the professional-managerial class) or in small oc- form.
cupations or “microclasses” (e.g., doctor, lawyer, This chapter examines whether any of these
carpenter). three worlds is ascendant and, obversely, whether
There nonetheless is some evidence on the any is in decline. Although there is no shortage
changing form of inequality. With the advent of of narratives of change in inequality scholarship,
“death of class” arguments in the 1990s, a small they typically do not explicitly address the mech-
and self-contained literature on the effects of so- anisms that might underlie changes in the form
cial class has developed, but it unfortunately fo- of inequality. We thus next lay out some possible
cuses on a restricted set of outcomes, mainly life mechanisms.
chances, political attitudes, and political behav-
ior. Worse yet, virtually no research assesses the
underlying “gradationalism” of inequality, even Big-Class Trend
though it seems plausible that it is on the rise. It The most common accounts of change at
is striking that once-standard gradational models, the big-class level predict a shift away from a
such as socioeconomic scales, have largely disap- big-class world. The exemplar here is the post-
peared from the day-to-day practice of inequality modern narrative, but the featured causes within
research, almost entirely without fanfare (but see this narrative also figure prominently in other
Hauser and Warren 1997). With class models scholarly traditions (see, e.g., Pakulski and Wa-
under attack and gradational models out of favor, ters 2008; Hechter 2004; Clark and Lipset 2001;
the emerging convention within American sociol- Hall 2001; Inglehart 1997; Bradley 1996). In
ogy is to resort to individual or family measures of this family of arguments, the weakening of big
income, earnings, or education, all of which are classes occurs through two types of change: a
now colloquially referred to as “class measures” decline in the class-based ideologies of old in-
(see Lareau and Conley 2008). The underlying, termediary organizations and an increase in
and largely unacknowledged, assumption of this “class-orthogonal” ideologies propagated by new
individualistic approach is that inequality is no intermediary organizations.
longer rooted in the division of labor. The first subnarrative assumes that much
We are left with many competing models ideological work is needed to convert “imag-
of the inequality space and either vocal claims ined” and geographically disparate big classes
or tacit assumptions about their relevance, but into culturally coherent communities. This
little systematic empirical evidence on actual work was once a main task of unions and
trends. In this chapter we begin to fill this gap working-class political parties, but it is no longer
by assessing trends in the strength of big class, as reliably performed by these intermediary orga-
microclass, and gradational forms of inequality nizations as they die off or define their strategies
in the United States over the past forty years. more narrowly. In the absence of organizations
We couch our analysis as a contest among three that explicitly train members into a class-based
ideal-typical inequality regimes: (1) a big-class worldview, big classes are little more than the
world in which inequalities in life conditions imagined fabrications of scholars, no longer the
are organized around such aggregate categories institutionalized communities of the past.
as professional, manager, service worker, craft The second subnarrative emphasizes the
worker, and laborer;1 (2) a microclass world in emergence of new types of intermediary or-
which inequalities are organized around the ganizations that fracture classes into diverse
local, organic, and deeply institutionalized cat- communities. These so-called new social move-
egories that emerge in occupations and become ments organize around issues that do not di-
microclasses (see, e.g., Grusky and Sørensen rectly map onto class (e.g., gender, climate, race,
224 • Part III: The Structure of Inequality

environment, abortion) and accordingly frac- boundaries (e.g., Casey 1996; Kim and Saka-
ture classes rather than reinforce class-specific moto 2008). Ostensibly, this development has
cultures. Similarly, highly differentiated media been hastened by deunionization and rising mo-
provide big-class members with a range of life- bility, which prevent workers from negotiating
style and attitudinal options that are not directly for fixed and stable job definitions (e.g., Kam-
linked to class, again having the effect of weak- bourev and Manovskii 2009; Mouw and Kalle-
ening big-class culture. berg 2010). As occupational boundaries become
Do any scholars describe a shift toward a more amorphous and affiliations more ephem-
big-class world? The Marxian story of class po- eral, occupational identities become weaker and
larization is unabashedly positive about the rise occupations no longer structure life conditions.
of a big-class regime, but contemporary scholars This variant of postoccupationalism is argu-
only rarely push that narrative. The Occupy Wall ably the dominant account of trends in micro-
Street movement has triggered efforts to char- class inequality, but other changes in the social
acterize top earners as a class (i.e., the “1 per- organization of work may have had counteract-
cent”), but this characterization relies more on ing effects. In particular, contemporary firms
income than on position in the division of labor. may now slot core employees into postoccupa-
Although growing income inequality is relevant, tional jobs, but rely on deeply occupationalized
then, to big-class formation, it probably works labor outside the firm (e.g., accounting, legal
to bring about an increasingly gradational form work, advertising, custodial services) for sup-
of big-class inequality, not an increasingly cate- port services. Similarly, the growing reliance on
gorical form (as discussed in more detail below). employment intermediaries (e.g., temp-work
For the most part, the class-analytic faithful organizations, headhunters) and computer tech-
have adopted a defensive posture, with the most nology to filter applications may have strength-
prominent responses emphasizing, at best, the ened occupational labor markets, at least insofar
continuing relevance of conventional big classes. as they rely on occupational categories to win-
In some cases, class analysts have implicitly con- now pools of workers.
ceded that they overreached in the past, none- The diffusion of an occupational logic is also
theless arguing for the continuing role of class evident in the proliferation of occupational clo-
analysis in understanding a narrower range of sure through licenses, credentials, and certifica-
outcomes. For example, Goldthorpe and Chan tions (Weeden 2002). In the United States, the
(2007) have argued that big classes continue percentage of the labor force that works in li-
to govern life chances and political behavior, censed occupations increased from 10 percent in
whereas detailed occupational affiliations matter the 1970s to 20 percent in 2004, an increase so
most in understanding consumption practices fast that licensees now outnumber the declining
and other cultural outcomes. pool of union members (see Kleiner and Krueger
2008). This increase reflects not only the growth
of long-closed professions (e.g., lawyers), but
Microclass Trend also the emergence of licensure in formerly
The mechanisms driving trends at the micro- “open” occupations. Moreover, even unlicensed
class level are even less well specified. The only occupations may increasingly require credentials
available theory comes in the form of “postoccu- that, like licenses, can support closure (Weeden
pational” accounts that describe a gradual break- 2002). Finally, occupational associations and
down of occupational distinctions. According to occupation-specific unions have fared far better
such narratives, the contemporary workplace of late than have classic industrial unions, again
increasingly relies on teamwork, cross-training, making the trend away from a microclass world
and multi-activity jobs, all of which promote less certain than the postoccupational narrative
labor force flexibility and weaken occupational would have us believe.
 Weeden and Grusky—The Changing Form of Inequality • 225

Gradational Trend Measuring Trend

The main force behind a possible increase in These competing accounts can be examined in
gradationalism is the well-known growth in in- four topical domains: (1) life chances (e.g., in-
come inequality over the last thirty-five years. As come, education, home ownership); (2) lifestyles
income gaps between classes widen, differences (e.g., consumption practices, institutional par-
in the consumption opportunities available to ticipation); (3) sentiments (e.g., political pref-
class incumbents also increase, which may in erences, social attitudes); and (4) demographic
turn generate more distinctive class cultures and composition (e.g., race, ethnicity). The variables
interests. It follows that professionals, even those pertaining to life chances, institutional participa-
who themselves are not well paid, will increas- tion, and demographic composition are drawn
ingly display the tolerance, cultural cosmopol- primarily from the 1972–2002 March Current
itanism, and largesse of an economically secure Population Survey (CPS; BLS 2012), and the
group, whereas unskilled laborers will increas- variables pertaining to social and political atti-
ingly display the intolerance, parochialism, and tudes are drawn from the 1972–2010 General
economic protectionism of a class that is under Social Survey (GSS; Davis, Smith, and Marsden
threat and anxious. Moreover, because income 2012).
gaps between microclasses are growing at a faster Within each domain, we chose variables that
rate than gaps between big classes (Weeden et offer consistent coverage across survey years, sim-
al. 2007), any increase in gradationalism should ilar wording of items over time, and at least two
be especially pronounced at the microclass level. thousand valid cases in each time period. These
The income inequality narrative offers a plau- conditions yield forty variables for our analysis of
sible account of increasing gradationalism, but a trends in big-class and microclass inequality and
countervailing narrative about the rise of non- thirty-eight variables for our analysis of trends in
economic and “postmaterial” values makes the gradationalism (see Weeden and Grusky 2012).2
net trend unclear. As Pakulski and Waters (2008) The annual or biannual samples of the GSS and
argue, the heyday of economic determinism in CPS surveys are pooled into three time periods:
politics was some fifty years ago, a period during 1972–1982, 1983–1992, and 1993–2010 (GSS)
which unions and political parties formulated or 1993–2002 (CPS).3
class narratives that pitted the interests of mon- We define the big-class and microclass
ied classes against those of manual and working schemes by coding 1970 detailed SOC codes
classes. This simple formula broke down, how- into 126 microclasses and then in turn nest-
ever, as party platforms and messages decou- ing those microclasses into 12 big classes. The
pled from particular class income levels. The big-class scheme includes self-employed profes-
Republican Party, for example, has fashioned a sionals, employed professionals, employed man-
message that brings managers and the working agers, self-employed managers, sales workers,
class together by valorizing free markets and clerical workers, craft workers, operatives, service
delegitimating redistribution, favoring military workers, laborers, farmers, and farm laborers.
spending over welfare spending, and celebrat- The microclass scheme, which is described else-
ing (Christian) “family values.” The Democratic where in more detail (Weeden and Grusky 2005,
Party, for its part, has striven to create a message 2012), is designed to capture the main institu-
that appeals to professionals as well as residual tionalized boundaries in the division of labor.4
sectors of the working class (e.g., Hout and The occupational earnings scale, which is
Moodie 2008). The unappreciated implication constructed from weekly earnings data in the
of the postmodernist narrative is that big classes 1973–1982 CPS surveys, measures the percent-
are breaking down mainly because their grada- age of workers in each occupation who exceed a
tionalist foundation is disintegrating. threshold earnings level (see Hauser and Warren
226 • Part III: The Structure of Inequality

Figure 23.1. Total and Gradational Class-Outcome Association at Baseline (1972–1981)

Big Class Micro-Class
share of total association share of total association
share of total association share of total association
that is gradational that is gradational

Life chances

Lifestyles: consumption

Lifestyles: institutional
participation

Sentiments: political attitudes

Sentiments: social attitudes

Demographic structuration

All outcomes

-75 % p -50 p-25 0 25 50 75 %

p

1997). The big-class variant of the scale averages gender, and self-employment status (where rel-
the percentage of above-threshold earners in evant). We then apply simple log-linear models
all occupations comprising a big class, and the to estimate (1) the total microclass and big-class
microclass variant takes each occupation’s devia- association, and (2) the extent of microclass and
tion from its big-class value. We use this scale to big-class gradationalism.5
assess the extent to which classes are organized The first result of note is that the majority
in simple gradational form (see Weeden and of class-based inequality in life conditions was
Grusky 2012 for details). found at the microclass level (see Figure 23.1).
Averaged across outcomes, 58 percent of the
total class association is located between the
Which World Prevailed in the 1970s? microclasses comprising big classes, whereas 42
We begin by describing the baseline form of in- percent is located between big classes. Only in
equality that prevailed in the 1970s. For this anal- the life chances domain, which includes items
ysis, we use the 1972–1981 data to cross-classify measuring education, employment, and home
each of the outcome variables by microclass, ownership, do we find more big-class association
 Weeden and Grusky—The Changing Form of Inequality • 227

Figure 23.2. Preferred Models of Trends in Class Inequality for 40 Outcomes

Model 1: constant association, gender effect

Model 2: time shift effects, gender effect

Model 3: big-class and micro-class time shift

effects, gender effect

Model 4: gender- and level-specific time shift

effects

0 10 20 30
Number of outcomes

(60 percent) than microclass association (40 to differ by gender. This model constrains the
percent). class-by-outcome association to take on the same
As Figure 23.1 also shows, big-class asso- pattern and remain at the same overall strength
ciation in the 1970s was largely gradational, in each time period.
whereas microclass association was largely poly- The second model, “global trend,” adds shift
morphous or “nongradational.” Averaged across effects that allow the association between class
all outcomes, approximately three-quarters (73 and outcome to expand or contract in each time
percent) of the big-class association is captured period. These effects are applied to the global
by an income gradient. This result, which has class-outcome association and do not allow for
not been well appreciated in prior research, is differing trends at the big-class and microclass
inconsistent with big-class theorizing that down- levels. The contrast in fit statistics with the prior
plays the hierarchical dimension of class (e.g., model speaks to the change in the class-outcome
Erikson and Goldthorpe 2002; Bourdieu 1984). association.
By contrast, our measure of occupational earn- The third model, “class-specific trend,” as-
ings accounts on average for just 6 percent of the sesses whether changes in big-class and mi-
total microclass association, with the maximum croclass association diverge. It fits two sets of
value of 16 percent appearing in the life chances multiplicative shift effects, one that captures
domain. trends in the big-class association (after purging
any confounding effect of changes in the occu-
pational composition of big classes), and another
Trends in Class Inequalities that captures trends in the microclass association
We next fit four log-multiplicative models to (net of big-class effects). The contrast between
the class-by-outcome arrays (Goodman 1991; this model and the global trend model reveals
Xie 1992; see Weeden and Grusky 2012 for whether big-class and microclass trends are mov-
equations and identifying constraints). The first ing in lockstep. As a robustness check, we also
model, “no trend,” allows the class and outcome fit a fourth model that allows for gender-specific
marginal effects to vary over time and by gender, big-class and microclass trends.
saturates the two-way interaction of class and Figure 23.2 provides a simple count of the
outcome, and fits a multiplicative shift effect that number of outcomes for which each of the four
permits the strength of this two-way interaction models is preferred. According to conventional
228 • Part III: The Structure of Inequality

Figure 23.3. Trends in Big-Class and Microclass Inequality by Domain

1.00
Life chances Micro class
0.98

0.98 Big class

Lifestyles: consumption
0.74

1.01
Lifestyles: institutional participation
0.86

1.29
Sentiments: political attitudes
0.78

0.97
Sentiments: social attitudes
0.86

0.88
Demographic structuration
0.88

0.99
All outcomes
0.85

0.6 1.0 1.6

=ln(-0.5) =ln(0.5)

Values are the slopes of multiplicative time effects graphed on a logarithmic scale.

significance tests, the “no trend” model is pre- conditions is weakening on average by approxi-
ferred for only two of the forty outcomes and mately 15 percent per time period. This result
the “global trend” model for only one, whereas belies the long-dominant view that class systems
one of the two level-specific trend models is pre- are stable and timeless. It should be noted, more-
ferred for the remaining thirty-seven outcomes. over, that this trend is off a 1970s baseline in
We conclude that (1) the association between which big classes captured only a minority of the
class and outcome is rarely stable, and (2) the available association in the division of labor (see
big-class and microclass trends tend to differ in Figure 23.1). The simple conclusion is that the
strength or direction. big-class model, long the sociologist’s preferred
We next calculate, for each item, measuring tool, was not only a relatively weak
size-weighted slope coefficients on trend at the signal of inequality in the 1970s, but is becom-
big-class and microclass levels. If the “no trend” ing ever weaker.
model is preferred, the slope coefficient is fixed This shift away from a big-class world has
at 1. If the “no trend” model is rejected, the slope not been accompanied by a corresponding de-
coefficients are calculated from the global trend cline in microclass structure. The all-outcome
parameters or the level-specific trend parameters, slope coefficients (see Figure 23.3) indicate a
depending on which model is preferred.6 mere 1 percent decline (per period) in micro-
The means of these coefficients are graphed class structure. That said, most domains show
on a logarithmic scale in Figure 23.3. We find no evidence of a strengthening of microclass
that the association between big class and all life organization, a result that is inconsistent with
 Weeden and Grusky—The Changing Form of Inequality • 229

a strong-form occupationalization narrative. the logged shift effects. The main conclusion is
The results instead suggest either trendless fluc- again straightforward: we find no evidence of an
tuation or no substantial trend in microclass increase in microclass gradationalism across all
inequality. outcomes. In the few domains where the increase
This general pattern of big-class decline and is substantial (e.g., consumption patterns), it is
microclass stability holds across most domains, off an exceedingly low 1970s baseline (Figure
but there are notable variations. The political 23.1).
domain, for example, is seemingly the poster The big-class story is completely different.
child for a microclass narrative. In this domain The gradational association between big-class
the big-class effects weakened by 22 percent per earnings and life conditions, which was substan-
period, whereas microclass effects strengthened tial in the 1970s (see Figure 23.1), declined on
by 29 percent per period (Figure 23.3). This average by 12 percent per period (Figure 23.4).
result should be interpreted cautiously, because The trend is exaggerated in some domains (e.g.,
the trend in the average strength of the big-class the political domain) and absent in others (e.g.,
and political attitudes association may conceal the life chances domain), but the common con-
the realignment of the professional class with the clusion is that gradationalism is declining.
political Left (see Hout and Moodie 2008). We note, finally, that this decline in big-class
If the divergence between big-class and mi- gradationalism was less dramatic than the decline
croclass trends is exaggerated in the political in nongradational big-class effects. As shown in
domain, it is muted in the life chances domain, Figures 23.3 and 23.4, the mean slope coefficient
where the data show little evidence of change in pertaining to big-class gradationalism is slightly
the strength of class-based inequalities. This re- larger than the mean slope coefficient pertaining
sult is consistent with a narrative that emphasizes to total big-class inequality, which implies that
the weakening of class cultures (e.g., attitudes, the downward trend in gradationalism is flatter
lifestyles), but not a weakening in the structural than the downward trend in nongradational as-
backbone of class analysis (Goldthorpe and sociation. It follows that the remaining big-class
McKnight 2006). inequality is increasingly gradational in form.
This increase in the gradational share is the only
sense in which we are moving toward a grada-
Trends in Gradationalism tional world.
We can assess trends in gradationalism by apply-
ing variants of the four log-multiplicative shift
effect models. Instead of saturating the under- Conclusion
lying association between class and outcome, We find that the amount of big-class organiza-
we now fit an RC model (Goodman 1991) that tion is declining in the United States. Moreover,
constrains the class effects to take on a grada- at the big-class level, the gradational princi-
tional pattern defined by our income scales. We ple was already dominant in the 1970s and, if
find that a model that assumes no change in gra- anything, has become more dominant in the
dationalism is preferred for only four outcomes, decades since. By contrast, microclass inequal-
whereas models that allow for disparate trends in ity remains as strong as ever in absolute terms,
big-class and microclass gradationalism are pre- constitutes a growing share of total class-based
ferred for the remaining thirty-four outcomes. inequalities, and cannot be reduced to simple
In Figure 23.4 the trend in gradational gradational terms.
inequality is presented by domain, using These results have practical implications for
a model that allows for class-specific (but measuring inequality. It is unquestioned prac-
gender-invariant) trends and slopes from a tice, especially in European sociology, to use
weighted least squares model regressing time on big-class schemes to represent “background
230 • Part III: The Structure of Inequality

Figure 23.4. Trends in Big-Class and Microclass Gradationalism by Domain

0.89
Life chances
1.03

1.09
Lifestyles: consumption
0.88

0.93
Micro class
Lifestyles: institutional participation
0.84
Big class

0.79
Sentiments: political attitudes
0.75

1.11
Sentiments: social attitudes
0.87

0.85
Demographic structuration
0.88

1.00
All outcomes
0.88

0.6 1.0 1.6

Values are the slopes of multiplicative time effects graphed on a logarithmic scale.

effects,” analyze social mobility, and model the We are left with two possible measurement
various effects of inequality. We show here that approaches. The first, which we applied here, is
defaulting to big classes is especially problem- to represent the inequality space with a full com-
atic when the goal is to measure trends. If we plement of big-class and microclass effects. The
continue to use big-class measurement, social second is to forego the division of labor alto-
background effects may appear to be declining, gether and simply represent the inequality space
the mobility regime may appear to be opening with measures of individual income, education,
up, or individual choice in lifestyles may appear or wealth. The latter approach is fast becoming
to be expanding. But such appearances could the default, but without systematic evidence on
be deceiving because the big-class measurement its behalf or even acknowledgment of the radical
tool is conveying progressively less information shift in assumptions about the form of the in-
about the inequality space. At the same time, equality space that it implies.
unidimensional occupational scales fare no bet- The results presented here also speak to
ter, particularly at the microclass level, where a prevailing theories of change in the inequality
substantial and increasing share of positional in- system. Although big-class differences in life
equalities is located. chances (e.g., education, income) have remained
 Weeden and Grusky—The Changing Form of Inequality • 231

largely intact, big-class differences in cultures outcome, and (2) the three-way association among
(e.g., attitudes, lifestyles) have not. This pattern microclass, gender, and outcome. This prevents our
is consistent with the postmodern claim that in- log-likelihood ratios from being polluted by associ-
termediary organizations no longer convert big ation that is not predicted by either the big-class or
the microclass approach.
classes into culturally coherent communities.
6. For ease of presentation, we calculated slopes
These narratives fall short, however, insofar as using the gender-invariant coefficients, but the re-
they suggest that all class forms are dissipating. sults are much the same when gender-specific trends
We show instead that microclass inequalities have are allowed.
remained largely unchanged. This result is not
entirely surprising given that the intermediary REFERENCES
organizations that generate occupation-specific Bourdieu, Pierre. 1984. Distinction: A Social Critique
cultures, interests, and life chances remain com- of the Judgement of Taste. Translated by Richard
paratively strong. The unions organized around Nice. New York: Cambridge University Press.
Bradley, Harriet. 1996. Fractured Identities: Chang-
occupations (e.g., teachers, airline pilots) have
ing Patterns of Inequality. Oxford: Polity Press.
fared better than large industrial unions; occu-
Bureau of Labor Statistics (BLS). 2012. March De-
pational licenses, certifications, and associations mographic Files of the Current Population Sur-
have proliferated; and new forms of hiring tend veys. Machine-readable data file. Washington,
to rely on occupationalized labor markets. Be- DC: Bureau of the Census.
cause occupations remain deeply institutional- Casey, Catherine. 1996. “Corporate Transforma-
ized, they can develop cultures that affect what tions: Designer Culture, Designer Employees
we believe, how we vote, and the lifestyles we and ‘Post-Occupational’ Solidarity.” Organiza-
lead. The result is a shift toward a labor mar- tion 3 (3): 317–339.
Clark, Terry, and Seymour Martin Lipset, eds. 2001.
ket that is characterized by the “moral poly-
The Breakdown in Class Politics. A Debate on
morphism” to which Durkheim (1958, 7) so
Post-Industrial Stratification. Washington, DC:
famously referred. Woodrow Wilson Center Press; Baltimore, MD:
Johns Hopkins University Press.
NOTES Davis, James A., Tom W. Smith, and Peter H. Mars-
1. There are of course many varieties of big-class den. 2012. General Social Surveys 1972–2010.
models. For our purposes, the distinctive features of Chicago: National Opinion Research Center.
all big-class formulations are that the classes are (a) Durkheim, Emile. 1958 [1893]. The Division of La-
few in number, (b) large in size, and (c) “analytically” bour in Society. Oxford: Oxford University Press.
defined rather than deeply institutionalized. Erikson, Robert, and John H. Goldthorpe. 1992.
2. The class measure pertains to the respondent’s The Constant Flux: A Study of Class Mobility in
class position for all outcomes except educational at- Industrial Societies. Oxford: Clarendon Press.
tainment (for which we use the father’s class position). ———. 2002. “Intergenerational Inequality: A So-
3. The GSS and CPS samples are restricted to ciological Perspective.” Journal of Economic Per-
adult respondents in the civilian labor force who spectives 16(3): 31–44.
are between ages twenty-five and sixty-four. The Featherman, David L., and Robert M. Hauser. 1978.
CPS samples are further restricted to households in Opportunity and Change. New York: Academic.
months one through four of the sampling rotation Fischer, Claude S., and Michael Hout. 2006. Cen-
to prevent duplicate observations. Both samples are tury of Difference: How America Changed in the
weighted by the appropriate survey weights. Last Hundred Years. New York: Russell Sage
4. This nesting is imperfect because the big-class Foundation.
map incorporates self-employment status, whereas Goldthorpe, John H., and Tak Wing Chan. 2007.
the microclass map does not. We address this prob- “Social Status and Newspaper Readership.” Amer-
lem by constraining microclass effects within the ican Journal of Sociology 112(4): 1095–1134.
two professional and two managerial categories to Goldthorpe, John H., and Abigail McKnight. 2006.
be identical while allowing for an employment status “The Economic Basis of Social Class.” In Social
effect at the big-class level. Mobility and Inequality: Frontiers of Research from
5. We purge the data of (1) the three-way asso- Sociology and Economics, edited by Stephen L.
ciation among self-employment status, gender, and
232 • Part III: The Structure of Inequality

Morgan, David B. Grusky, and Gary Fields, 109– Lareau, Annette, and Dalton Conley, eds. 2008. So-
136. Stanford, CA: Stanford University Press. cial Class: How Does it Work? New York: Russell
Goodman, Leo. 1991. “Models, Measures, and Sage Foundation.
Graphical Displays in the Analysis of Cross- Manza, Jeff, and Clem Brooks. 1999. Social Cleav-
Classified Data (With Discussion).” Journal of ages and Political Change: Voter Alignment and
the American Statistical Association 86 (416): U.S. Party Coalitions. New York: Oxford Univer-
1085–1138. sity Press.
Grusky, David B., and Jesper B. Sørensen. 2001. Massey, Douglas S. 2007. Categorically Unequal: The
“Can Class Analysis be Salvaged?” American Jour- American Stratification System. New York: Russell
nal of Sociology 103 (5): 1187–1234. Sage Foundation.
Hall, Stuart. 2001. “The Meaning of New Times.” In Mayo, Anthony J., Nitin Nohria, and Laura G. Sin-
Social Stratification: Class, Race, and Gender in So- gleton. 2007. Paths to Power: How Insiders and
ciological Perspective, 2nd ed., edited by David B. Outsiders Shaped American Business Leadership.
Grusky, 859–865. Boulder, CO: Westview Press. Boston: Harvard Business Review.
Hauser, Robert M., and John Robert Warren. 1997. Mouw, Ted, and Arne L. Kalleberg. 2010. “Occupa-
“Socioeconomic Indexes of Occupational Status: tions and the Structure of Wage Inequality in the
A Review, Update, and Critique.” Sociological United States, 1980s-2000s.” American Sociologi-
Methodology 27(1): 177–298. cal Review 75(3): 402–431.
Hechter, Michael. 2004. “From Class to Culture.” Pakulski, Jan, and Malcolm Waters. 2008. “The
American Journal of Sociology 110 (2): 400–445. Death of Class.” In Social Stratification: Class,
Hout, Michael, and Benjamin Moodie. 2008. “The Race, and Gender in Sociological Perspective, 3rd
Realignment of U.S. Presidential Voting, 1948– ed., edited by David B. Grusky, Manwai C. Ku,
2004.” In Social Stratification: Class, Race, and and Szonja Szelenyi, 1022–1030. Boulder, CO:
Gender in Sociological Perspective, 3rd ed., edited Westview Press.
by David B. Grusky, Manwai C. Ku, and Szonja Weeden, Kim A. 2002. “Why Do Some Occupa-
Szelényi, 945–953. Boulder, CO: Westview Press. tions Pay More than Others? Social Closure and
Inglehart, Ronald. 1997. Modernization and Post- Earnings Inequality in the United States.” Amer-
modernization. Princeton, NJ: Princeton Univer- ican Journal of Sociology 108 (1): 55–101.
sity Press. ———. 2008. “Occupational Closure and Earnings
Jonsson, Jan O., David B. Grusky, Matthew Di Inequality.” In Social Stratification: Class, Race,
Carlo, Reinhard Pollak, and Mary C. Brinton. and Gender in Sociological Perspective, 3rd ed.,
2009. “Microclass Mobility: Social Reproduction edited by David B. Grusky, 176–186. Boulder,
in Four Countries.” American Journal of Sociology CO: Westview Press.
114 (6): 977–1036. Weeden, Kim A., and David B. Grusky. 2005. “The
Kambourev, Gueorgui, and Iourii Manovskii. 2009. Case for a New Class Map.” American Journal of
“Occupational Mobility and Wage Inequality.” Re- Sociology 111 (1): 141–212.
view of Economic Studies 76(2):731–759. ———. 2012. “The Three Worlds of Inequal-
Kim, ChangHwan, and Arthur Sakamoto. 2008. ity.” American Journal of Sociology 117 (6):
“The Rise of Intra-Occupational Wage Inequal- 1723–1785.
ity in the U.S., 1983–2002. American Sociological Weeden, Kim A., Young-Mi Kim, Matthew Di
Review 73: 129–157. Carlo, and David B. Grusky. 2007. “Social Class
Kingston, Paul. 2000. The Classless Society. Stanford, and Earnings Inequality.” American Behavioral
CA: Stanford University Press. Scientist 50 (5): 702–736.
Kleiner, Morris M., and Alan B. Krueger. 2008. Wilensky, Harold L., and Jack Ladinsky. 1967.
“The Prevalence and Effects of Occupational “From Religious Community to Occupational
Licensing.” Working paper 14308. Cambridge, Group: Structural Assimilation among Profes-
MA: NBER. sors, Lawyers and Engineers.” American Sociolog-
Kohn, Melvin. 2008. “Job Complexity and Adult ical Review 32 (4): 541–561.
Personality.” In Social Stratification: Class, Race, Wright, Erik Olin. 1997. Class Counts. Cambridge:
and Gender in Comparative Perspective, 3rd ed., Cambridge University Press.
edited by David B. Grusky, Manwai C. Ku, and Xie, Yu. 1992. “The Log-Multiplicative Layer Effect
Szonja Szelényi, 532–540. Boulder, CO: West- Model for Comparing Mobility Tables.” Ameri-
view Press. can Sociological Review 57(3): 380–395.
 Trieman—Occupational Prestige in Comparative Perspective • 233

Classic Gradationalism

24 • Donald J. Treiman

Occupational Prestige in Comparative Perspective

In the three decades since World War II there importance since it allows us to make use of
have been some eighty-five studies of occupa- data drawn from rather poor samples of the
tional prestige conducted in more than sixty population—for example, students, members of
countries throughout the world, ranging from voluntary organizations, representatives of spe-
highly industrialized places such as the U.S. to cial subcultures—without fear that if we had a
traditional societies such as India, Thailand, Ni- different sample we would get different answers.
geria, and New Guinea. Although these studies The third noteworthy feature is that although
vary somewhat in their specific details, they all the distribution of the labor force in various
utilize the same basic procedure: a sample of occupations varies substantially from place to
the population is asked to rate or rank a set of place, the same sorts of occupations tend to exist
occupational titles with respect to their prestige everywhere. Even if there are not many airplane
or social standing. These ratings are then aggre- pilots or professors in a given country, there tend
gated into mean scores (or other measures of to be at least a few of them and the population
central tendency) and the scores are treated as at large knows what these jobs are. In general,
indicators of the relative prestige of the evaluated the organization of work into specific jobs is
occupations. amazingly uniform across societies. Pretty much
A remarkable feature of these studies is that everywhere there are distinctions between weav-
they yield the same results regardless of the exact ers and tailors, and between carpenters, painters,
wording of the questionnaire. It does not matter and plumbers. And the uniformity in occupa-
whether respondents are asked about the “pres- tions across societies is reflected in the unifor-
tige” or “social standing” or “respect” accorded mity of occupational titles appearing in prestige
certain occupations, or whether they are asked studies. As a result, matching occupational titles
to rate occupations on a scale or to rank them across countries is a less onerous task than might
in any other way. The results are the same. A be expected.
second striking feature is that the educated and These three features, uniform results regard-
uneducated, the rich and poor, the urban and less of measurement procedures, minimal sub-
rural, the old and young, all on the average have group variations, and similarity of occupational
the same perceptions of the prestige hierarchy. titles, make possible a systematic comparison of
There is no systematic subgroup variation in the occupational prestige hierarchies among coun-
relative ratings of jobs. This is of considerable tries. The basic procedure is to match titles across

Donald J. Treiman. “A Standard Occupational Prestige Scale for Use with Historical Data,” from The Journal
for Interdisciplinary History VII (1976), pp. 285–290. Copyright © 1976 by the Massachusetts Institute of
Technology and The Journal for Interdisciplinary History, Inc. Reprinted with permission of the editors of The
Journal for Interdisciplinary History and the MIT Press, Cambridge, Massachusetts.
234 • Part III: The Structure of Inequality

countries, e.g., “physician” in the U.S. with To derive generic prestige scores for each
“doctor” in Australia, “medecin” in Mauritania, of these occupations, I converted all the data
“medico” in Argentina, “laege” in Denmark, and to a standard metric and then simply averaged
then to compute a product moment correlation scores across all countries in which a given title
between the prestige scores for all matching titles appeared. Scores for higher levels of aggregation
for each pair of countries. The correlation coef- (unit groups, minor groups, and major groups)
ficients thus generated can be taken as measures were derived by various averaging procedures.3
of the similarity of prestige evaluations between How good is the prestige scale created by this
each pair of countries. The fundamental conclu- procedure? The answer is—very good. Evidence
sion from such computations is that there is sub- for contemporary societies is extremely convinc-
stantial uniformity in occupational evaluations ing. The average correlation of the new Standard
throughout the world: the average intercorrela- Scale with the reported prestige hierarchies of
tion between pairs of countries is .81.1 As such fifty-five countries is .91 and only seven of the
numbers go, it is extremely high and fully jus- correlations are less than .87. Thus the Standard
tifies treatment of the prestige hierarchy in any Scale is, on the average, the best available predic-
given country as reflecting, in large part, a single tor of the prestige of occupations in any contem-
worldwide occupational prestige hierarchy. On porary society. . . . 4
the basis of this result, and in view of the need
for a standard occupational scaling procedure, it
seemed desirable to attempt to construct a stan- A Theory of the Determinants of
dard occupational prestige scale which could be Prestige
applied to any country. Analysis of the universally shared occupational
In order to match occupational titles across prestige hierarchy suggests that high prestige is
countries, it was necessary to devise a compre- allocated to those occupations which require a
hensive occupational classification scheme. To high degree of skill or which entail authority
do this, I took advantage of an already existing over other individuals or control over capital.
scheme: the International Standard Classification Moreover, the nature of occupational special-
of Occupations, Revised Edition (ISCO).2 This ization is such that specific occupations are rel-
classification is a “nested” scheme which clusters atively invariant in these characteristics across
occupations into nine major groups, eighty-three time and space. As a result, the prestige of spe-
minor groups, 284 unit groups, and 1,506 spe- cific occupations is relatively invariant as well. In
cific occupations. It was developed by the Inter- fact, it is so uniform that a single occupational
national Labour Office as a guide for national prestige scale will capture the basic features of
census offices to encourage the comparability of the occupational hierarchy of any society.
occupational statistics. Many foreign census bu- Specialization of functions into distinct oc-
reaus do, in fact, utilize the ISCO scheme and cupational roles necessarily results in inequalities
do publish occupational statistics according to among occupations with respect to skill, author-
its guidelines. ity, and control over capital. Some occupations,
However, since the 1,506 ISCO occupational by their very definitions, require specific skills.
categories did not correspond very well to the For example, literacy is required of clerks because
occupational titles for which I had prestige rat- one cannot be a clerk if one is illiterate. Similarly,
ings, I followed the ISCO scheme (with minor some occupations require control over capital
variations) down to the unit group level and or authority over other individuals as inherent
then, within this level, made distinctions among definitions of their functions. For instance, a
specific occupations when they appeared war- managerial job is one which involves “planning,
ranted by my prestige data. The resulting classi- direction, control, and co-ordination”;5 other-
fication contains 509 distinct occupational titles. wise, it is not a managerial job but something
 Trieman—Occupational Prestige in Comparative Perspective • 235

else. Examples of these inherent inequalities can is, their power and privilege.6 Since occupa-
be located throughout the prestige hierarchy. tions are differentiated with respect to power,
Skill, authority, and economic control are they will in turn be differentiated with respect
singled out as the basic resources which differ- to privilege and prestige. Thus, if this theory of
entiate occupations because these are the fun- prestige determinants is correct, these attributes
damental aspects of power—they provide the of occupations will be highly correlated across
crucial means to the achievement of desired societies.7 In particular, skill level, authority, eco-
goals. But the more powerful an occupation, nomic power, wealth, income, and prestige will
the more important it is that it be performed be highly intercorrelated with one another and
well, since the consequences of competent or will be highly correlated across countries.
incompetent performances are more telling for In my work on occupational prestige in con-
such occupations. For example, if a garbage col- temporary societies, I amassed data on the edu-
lector does his job poorly, little is lost; but if a cation levels of occupations for fifteen nations (as
surgeon is incompetent, a life can be lost. Or, a surrogate measure for “skill”) and on income
similarly, if a chain store manager makes a poor levels of occupations for eleven countries; no
business decision it may cost a firm a few hun- comparable measures of authority or control of
dred or at the most a few thousand dollars; but capital were available. These data indicate a strik-
a poor decision on the part of a major executive ing uniformity in occupational hierarchies. Like
can run into millions. Consequently, the more occupational prestige evaluations, occupational
powerful an occupation, the greater the incen- variations in education and in income proved to
tive to attract competent personnel to it. And be highly similar from society to society. When
since the basic mechanism for inducing people measures of the average level of schooling of in-
to perform tasks is to reward them, it follows cumbents of each occupation were computed for
that the most powerful positions will also be the U.S., Argentina, Canada, West Germany,
the most highly rewarded. Ghana, Great Britain, India, Israel, Japan, the
Other factors do enter into the determina- Netherlands, Norway, Taiwan, the U.S.S.R., Yu-
tion of rewards, so that the relationship between goslavia, and Zambia and these measures were
power and privilege is not perfect. Some func- intercorrelated, the average intercorrelation was
tions are in greater demand than others, depend- .76, which is almost as high as the average pres-
ing upon the needs of society at any particular tige intercorrelation reported above (.81). And
time. For example, in a hunting and gathering when measures of the average income of incum-
society, hunting is in greater demand than farm- bents of each occupation were computed for the
ing and thus hunting is more highly rewarded. U.S., Canada, Ceylon, Costa Rica, India, New
And in a commercial economy, law is of great Zealand, Pakistan, Surinam, Sweden, Taiwan,
importance and therefore highly rewarded. But and Yugoslavia and these measures were inter-
these differences are relatively minor compared correlated, the average intercorrelation was .65,
to the differences in occupational requisites and which is still a substantial correlation. Moreover,
perquisites which are inherent in the definitions education, income, and prestige levels of occupa-
of jobs and therefore stable across time and tions were highly correlated within each country:
space. the average correlations were, respectively, .77
Not only is the relative power and privilege between education and income, .72 between ed-
of occupations essentially similar across societies, ucation and prestige, and .69 between income
but so is the prestige accorded them, for prestige and prestige. The average correlations with the
is granted in recognition of power and privilege. Standard Scale were .79 for education and .70
Prestige is the metric of “moral worth,” and the for income.8
moral worth of positions reflects their control In short, the available data indicate that in the
over socially valued resources and rewards, that contemporary world occupational hierarchies are
236 • Part III: The Structure of Inequality

substantially invariant from place to place, even Measurement in the Comparative Study of Occupa-
among countries varying widely in level of in- tional Mobility,” Social Science Research, IV (1975),
dustrialization. This finding lends considerable 183–230.
empirical support to the theoretical argument 5. I.L.O., International Standard Classification,
95.
outlined above.
6. Cf. Edward Shils, “Deference,” in John A.
Jackson (ed.), Social Stratification (Cambridge,
NOTES
1968), 104–132.
This is a revised version of a paper presented
7. Some readers will recognize the similarity
at the Conference on International Comparisons
between this theory and that of Kingsley Davis
of Social Mobility in Past Societies held in 1972.
and Wilbert E. Moore, “Some Principles of Strat-
Preparation of the paper was supported by a grant
ification,” American Sociological Review, X (1945),
from the National Science Foundation to Columbia
242–249. The principal difference between the two
University (NSF #28050). I am grateful to the fol-
lies in the claim by Davis and Moore that prestige is
lowing for making unpublished material available or
granted by society as an inducement to competent
for giving leads to the work of others: Peter Decker,
people to fill important jobs. My claim is that occu-
Sigmund Diamond, Clyde Griffen, James Henretta,
pational income may be seen as such an inducement
David Herlihy, Theodore Hershberg, Richard Hop-
but that prestige must be viewed as a measure of
kins, Michael Katz, William Sewell, Jr., James Smith,
moral worth, that is, of the extent to which an oc-
and Stephan Thernstrom. Thanks are also extended
cupation embodies that which is valued by members
to Vincent Covello, Theodore Riccardi, Jr., Rose M.
of society. Since power and privilege are universally
Cascio, Michael Freeman, John Hammond, Jr., Her-
valued and since hierarchies of power and privilege
bert Klein, and Jane Ferrar. The comments of the
are relatively invariant, prestige will also be relatively
participants at the MSSB Conference, and especially
invariant.
those of Griffen, were extremely helpful in preparing
8. The education data typically derive from
the revision.
the population censuses of each of the countries
1. This average correlation was computed over all
in question. The income data also typically derive
pairs of countries with at least 10 occupational titles
from population censuses, but in some cases they
rated in common.
are from enterprise censuses. Ordinarily annual in-
2. International Labour Office (Geneva, 1969).
come was utilized but in some instances weekly or
3. These are described more fully in Donald J.
monthly wage rates were available rather than an-
Treiman, Occupational Prestige in Comparative Per-
nual income. In practice, alternative measures of the
spective (New York, 1976), ch. 8.
relative income of occupational groups tend to be
4. I have shown that the Standard Scale does a
highly correlated, despite differences among occupa-
uniformly better job of predicting occupational pres-
tions in part-time or seasonal employment rates. See
tige hierarchies in individual countries than do oc-
Treiman, Occupational Prestige, ch. 5, Tables 5.1 and
cupational status scales developed specifically for use
5.2. Data on both income and education levels were
in occupational mobility studies in those countries.
available for only five countries: the U.S., Canada,
See Donald J. Treiman, “Problems of Concept and
India, Taiwan, and Yugoslavia.
 Goldthorpe and Hope—Occupational Grading and Occupational Prestige • 237

25 • John H. Goldthorpe and Keith Hope

Occupational Grading and Occupational Prestige

Introduction power which are of a symbolic, rather than of

Over the last forty years or so, there has accu- an economic or political nature. That is to say,
mulated in the literature of sociology and social such advantage and power imply the ability of
psychology a relatively large number (probably an actor to exploit—in the pursuit of his goals—
several score) of studies in which respondents meanings and values, rather than superior mate-
have been required to grade a selection of occu- rial resources or positions of authority or of force
pations in some hierarchical fashion. It has be- majeure.
come customary to refer to such studies as being From this conception it follows that a hierar-
ones of ‘occupational prestige’. Indeed, when the chy of prestige is constituted by intersubjective
matter of occupational prestige is now consid- communication among actors, and must there-
ered, it is almost invariably in terms of studies fore be characterized in attitudinal and relational
of the kind in question. Furthermore, the data terms. It cannot be characterized—other than
provided by these enquiries have come to play misleadingly—as a distribution in which units
an important part both in theoretical discussion have differing amounts of some particular sub-
and in the conduct of empirical investigation in stance or quality. As a provisional statement, a
the general problem-area of social stratification prestige hierarchy might be one in which actors
and mobility. Yet, oddly enough, ‘occupational (i) defer to their superiors—that is, acknowl-
prestige’ studies have rarely been subjected to edge by speech or other action their own
critical examination other than from a technical social inferiority—and seek, or at least ap-
point of view. . . . preciate, association with superiors;
(ii) accept their equals as partners, associates etc.
in intimate social interaction—entertain-
The Meaning of Prestige ment, friendship, courtship, marriage, etc.;
An appropriate starting point for a more radi- (iii) derogate1 their inferiors, if only by accept-
cal appraisal of ‘occupational prestige’ studies ing their deference and avoiding association
than seems hitherto to have been made is with with them other than where their own su-
the concept of ‘prestige’ itself. In a sociological periority is confirmed.
context, we would suggest, prestige can be most
usefully understood as referring to a particular The attributes of roles or collectivities which
form of social advantage and power, associated differentiate actors in respect of their prestige
with the incumbency of a role or membership are various. What they have in common is
of a collectivity: specifically, to advantage and some symbolic significance—some generally

John H. Goldthorpe and Keith Hope. “Occupational Grading and Occupational Prestige,” in The Analysis of
Social Mobility: Methods and Approaches, edited by Keith Hope, pp. 23–24, 26–27, 30–37, 78–79. Copyright
© 1972 by Oxford University Press. Used by permission of Oxford University Press.
238 • Part III: The Structure of Inequality

recognised meaning—which, in conjunction that is likely to be interpreted as indicative of so-

with prevailing values, constitutes a claim to cial superiority or inferiority, with corresponding
social superiority or, conversely, some stigma of interactional consequences.
inferiority. For example, having the role of doc- We may, therefore, now go on to ask such
tor and working in a hospital or clinic implies questions as: (i) whether such a conception of
having knowledge of, control over and close in- occupational prestige has been that generally
volvement with matters which are generally re- held by the authors of conventional occupa-
garded as ones of ultimate concern—matters of tional prestige studies; (ii) whether the results of
life and death. Belonging to an aristocratic fam- such studies provide valid indicators of prestige
ily and owning a landed estate signifies descent in the sense in question; (iii) whether the uses to
from illustrious forebears and participation in which results have been put have been appropri-
an historically-rooted, distinctive and exclusive ate ones. . . .
way of life. Working as a clerk in a bank evokes
such generally valued characteristics as honesty,
trustworthiness, discretion and dependability, The Interpretation of Occupational
and again in relation to ‘important’—in this Prestige Ratings
case, financial—matters. In all of these cases, It has been regularly remarked that in occupa-
then, ‘deference-entitlements’ (Shils, 1968) tional prestige ratings, as conventionally carried
exist, and are likely to be honoured at least by out, both cognitive and evaluative processes are
some actors in some contexts. In contrast, being, involved. However, precisely what are supposed
say, a gypsy scrap-metal dealer or a West Indian to be the objects of these processes has rarely
refuse-collector is likely to mean relatively fre- been made clear. For example, if it really were
quent exposure to derogation, on account both occupational prestige in the sense we would fa-
of the symbolic significance of the ethnic mem- vour which was being assessed, then what would
berships in question and of the implied occu- have to be cognized (or, rather, recognized) and
pational contact with what is spoiled, discarded evaluated would be the symbolic significance of
and dirty.2 In other words, prestige positions do certain features of an occupation with regard
not derive directly from the attributes of a role to the chances of those engaged in the occu-
or collectivity ‘objectively’ considered, but rather pation meeting with deference, acceptance or
from the way in which certain of these attributes derogation in their relations with others. If, for
are perceived and evaluated in some culturally instance, the occupational ‘stimulus’ given were
determined fashion. . . . that of ‘coal miner’, a possible response might be
on the lines of
‘dirty, degrading work’
Occupational Prestige ‘rough, uncultivated men’
Assuming that a conception of prestige consis- ‘likely to be looked down on by most groups’
tent with classical analyses is adopted, then the
reference of ‘occupational prestige’ follows from or, alternatively perhaps
it directly: it is to the chances of deference, ac- ‘difficult, dangerous work’
ceptance and derogation associated with the ‘able, courageous men’
incumbency of occupational roles and member- ‘likely to be respected by many groups’.
ship in occupational collectivities. Such prestige
will be related to the ‘objective’ attributes of But is this in fact the kind of thing that usu-
occupations—their rewards, requisite qualifica- ally happens? There is little reason to believe so,
tions, work-tasks, work environments etc.—but at least if we are guided by respondents’ own
only indirectly: only, that is, in so far as these accounts of what chiefly influenced their rat-
attributes carry symbolic significance of a kind ings.3 Rather, we would suggest, the operation
 Goldthorpe and Hope—Occupational Grading and Occupational Prestige • 239

that most respondents have tended to perform way. Knowledge about the more general char-
(perhaps in accordance with the principle of acteristics of other than rather esoteric occupa-
least effort) is a far more obvious and simple tions is relatively ‘open’; and, again in general
one: namely, that of rating the occupations on terms, the kinds of thing thought of as ‘good’
the basis of what they know, or think they know, in a job are unlikely to give rise to systematic
about a number of objective characteristics, eval- differences in ratings, especially since there is, in
uated in terms of what they contribute to the any case, a clear tendency for such advantages to
general ‘goodness’ of a job. In other words— go together. To take a particular example—from
and consistently with their own accounts—re- the NORC data—it is not surprising, given an
spondents in occupational prestige studies have interpretation of the kind we have proposed,
not typically been acting within a distinctively that individuals should quite often disagree
‘prestige’ frame of reference at all. The sensitiv- about the ratings of ‘building contractor’ vis-à-
ity to symbolic indications of social superiority vis ‘welfare-worker’—nor that, at the same time,
and inferiority which this would imply has not in the case of age, sex, regional, occupational or
usually been evoked by the task of grading set other categories, the former job should invari-
them. Rather, this task has led them to assess oc- ably have the higher mean rating. (See Reiss
cupations only in some far less specific fashion, 1961, pp. 55–6, 225–8.)6
according to a composite judgment on an assort- On the other hand, if we were to suppose that
ment of their attributes which might be thought ‘occupational prestige’ scores did give a valid in-
of as more or less desirable.4 dication of a structure of prestige relations, then
Such an interpretation of what ‘occupational the degree of consensus that is shown among dif-
prestige’ ratings are actually about would seem, ferent social groups would indeed be remarkable,
moreover, to fit far better with what is known at least in those societies where other research
of the pattern of variation in such ratings than has indicated some notable diversity in value sys-
would the idea that they relate to prestige stricto tems and in particular between members of dif-
sensu. The basic feature of this pattern is that ferent social strata. For in this case it would not
while some considerable amount of disagree- be a matter of evaluative consensus simply on
ment in rating occurs as between individuals, what attributes make a job ‘good’, but rather on
differences between the mean ratings of age, certain symbolic criteria of generalized superior-
sex, regional, occupational and other collectiv- ity and inferiority, with all their attitudinal and
ities are never very great. If one assumes that in behavioural implications. As Shils has observed,
making their judgments, respondents more or the conditions necessary for an entirely, or even
less consciously (i) consider a number of differ- a largely, ‘integrated’ prestige order to exist are
ent occupational attributes which they take as in fact demanding ones. It would seem, there-
determining how ‘good’ a job is; (ii) attach some fore, the safest assumption to make that, within
subjective ‘weight’ to each of these; (iii) for each modern industrial societies, such conditions will
occupation presented apply their rating ‘formula’ prevail only locally, transiently or imperfectly,
to what they know about the occupation, and and thus that social relations expressive of a
thus (iv) come to some overall assessment of it— prestige order will occur only in an intermittent
then one might well anticipate some appreciable or discontinuous fashion. On the basis of avail-
degree of variation in ratings at the individual able empirical data, one might suggest that while
level. Individuals are likely to differ in their fa- derogation is still quite widely manifest—as, for
miliarity with particular jobs and in their priori- example, in the form of differential association
ties as regards what makes a job ‘good’. However, or status-group exclusivity—the claim to superi-
one would not expect—other than in somewhat ority thus made by one group is not necessarily,
special and limited cases5—that such differences or even usually, acknowledged by those regarded
would be socially structured in any very striking as inferior; that is to say, the latter are often not
240 • Part III: The Structure of Inequality

inclined to display deference.7 This refusal may regarded because of their functional importance
be revealed passively—by disregard for the claim and also because they are those which require the
to superiority, in that no particular ‘respect’ is most training and ability and those to which the
shown, and little concern to reduce social dis- highest rewards accrue. Thus, ‘any major prestige
tance from the ‘superior’ group; or, perhaps, inversion would produce a great deal of inconsis-
some direct challenge to the claim may be made tency in the stratification system’ (p. 310).
where real interests are felt to be threatened by In this way, therefore, it is clearly indicated
it—as, say, by ‘exclusivity’ in housing areas, use how occupational prestige data may further be
of amenities, etc. . . . employed in support of a general theory of so-
cial stratification of a structural-functional type.
Such an application has in fact been made quite
The Uses of ‘Occupational Prestige’ explicitly in the work of Barber (1957). Fol-
Ratings lowing a Parsonian approach, Barber takes the
One notable use of the data in question results results of the Inkeles-Rossi study as the main
from the fact that over the last two decades oc- empirical foundation for the view that the fac-
cupational prestige studies have been carried out tual order of stratification in modern societies
in a steadily increasing number of countries at tends in the main to be consistent with the
different levels of economic development. The dominant normative order. Inequality in social
opportunity has therefore arisen of making rewards and relationships, it is held, is structured
cross-national comparisons which, it has been in accordance with functional ‘needs’, and this
supposed, can throw light on the relationship be- arrangement is then seen as receiving general
tween value systems and social structural charac- moral support: ‘functionally important roles are
teristics and are thus relevant to the thesis of the congruent with or partly determine a system of
‘convergent’ development of societies as industri- values’ (p. 6).
alism advances. For example, Inkeles and Rossi Clearly, for occupational prestige data to be
(1956), comparing occupational prestige ratings used in the ways in question, it is necessary to
in studies from six industrial societies, showed assume that such data reflect prevailing values
that a high degree of similarity prevailed. On and norms of a particular kind: ones pertaining
this basis, they concluded that common struc- to the ‘goodness’—in the sense of the ‘fairness’
tural features of these societies were of greater or ‘justice’—of the existing distribution of social
influence on the evaluation of occupations than power and advantage. However, in view of our
were differences in cultural traditions. Subse- previous discussion, it is difficult to regard such
quently, however, occupational prestige ratings an assumption as a valid one or indeed to under-
from several countries as yet little industrialized stand why it ever should have been made. Even
have also been shown to be broadly in line with if it were to be supposed that data on publicly
the hierarchy found in economically advanced recognized occupational hierarchies do indicate
societies—in so far, that is, as comparisons can a prestige order in something approximating
be made. This result has then led to the modified the classical conception, it still then would not
argument (Hodge, Treiman and Rossi, 1966) follow that they can provide evidence that the
that what is chiefly reflected in prestige ratings is objective reality of stratification is morally legit-
the set of structural features shared by national imated. For while prestige relations do depend
societies of any degree of complexity—‘special- upon a certain range of shared understandings,
ized institutions to carry out political, religious, consensus on principles of distributive justice is
and economic functions, and to provide for the not necessarily involved.8 Moreover, as we have
health, education and welfare of the popula- argued, by far the most plausible interpretation
tion . . . ’. Occupations at the top of these in- is that occupational prestige ratings reflect pre-
stitutional structures, it is suggested, are highly vailing ideas at a much lower level of abstraction:
 Goldthorpe and Hope—Occupational Grading and Occupational Prestige • 241

that is, ideas of what is ‘good’ in the sense simply occupation encountering deference, acceptance
of what is generally found desirable in an occu- or derogation in their social lives. In this case,
pation. And if this is the case, then the consensus therefore, mobility between different occupa-
that exists is obviously of no very great moral tional levels, other than of a marginal kind, may
or legitimatory significance at all. Apart from be interpreted as involving the probability of
quite unsurprising agreement on such matters subcultural and relational discontinuity. While
as, for example, that high pay is preferable to such a perspective does not necessarily mean that
low pay, more security to less, qualifications to society is seen as divided up into more or less
lack of qualifications, etc., the consensus that is discrete strata, it does imply that social mobil-
implied is of a cognitive and perceptual kind, ity, as measured, is not just a matter of individ-
not an evaluative one. The fact that, on average, uals gaining more qualifications, more income,
all groups and strata agree that certain occupa- more interesting work etc., but further of their
tions should be rated higher than others tells one experiencing changes in their life-styles and pat-
nothing at all about whether the occupational terns of association. The difficulty is, however,
hierarchy that is thus represented is regarded as as already remarked, that the validity of occu-
that which ought to exist. And in so far as the pational prestige ratings construed in this way
publicly recognized hierarchy corresponds to has never been established, and that there are in-
that proposed by structural-functional theorists, deed strong grounds for doubting their validity.
this would seem to indicate no more than that In other words, we are simply not in a position
broadly similar sets of rating criteria are being to infer, with any acceptable degree of precision
applied: i.e. occupational rewards and occupa- and certitude, what are the typical consequences
tional requirements.9 of mobility, as measured via occupational pres-
Thus, as regards the utilization of occupa- tige ratings, for the actual social experience of
tional prestige data in the advancement of strat- those deemed to be mobile.
ification theory, our view must be that this has (ii) Prestige ratings may be taken as indica-
been fundamentally misguided. What, now, tive of the status of occupations in the generic
of their application in research? Primarily, of sense—that is, as being in effect comparable
course, occupational prestige ratings have been with composite measures of ‘socio-economic’
used in studies of social mobility, in which they status, derived from data on income, educa-
have constituted the hierarchy—scalar or cate- tion, housing, possessions etc. Justification for
gorical—in the context of which mobility has this position is twofold: first, to [quote] the
been assessed. Assumptions about what pres- observation of Reiss (1961), respondents in
tige ratings rate are thus necessarily involved in prestige-rating studies appear ‘to emphasize
the interpretation of mobility patterns, and the the relevance of indicators sociologists use to
crucial issues that arise are once more ones of measure socio-economic status . . . ’; secondly,
‘validity’. as shown by Duncan (1961), it is possible, at
Concerning the question: What, in mobility least in the American case, to predict prestige
studies, may occupational prestige ratings be ratings fairly accurately from census data on oc-
taken to indicate?—three main positions can be cupational income and education. If then, ‘oc-
distinguished. These can be usefully considered cupational prestige’ is understood in the way in
in turn, together with their implications and question, some reasonable basis may be claimed
problems. for interpreting occupationally-measured mo-
bility in terms of movement between grades of
(i) Ratings may be taken—as, for example, by occupation differentiated chiefly by their levels
Svalastoga (1959)—as indicative of the position of rewards and requirements. At the same time,
of an occupation within a prestige order; that is, though, it must be emphasized that in this case
as indicative of the chances of those holding that no good grounds exist for any interpretation in
242 • Part III: The Structure of Inequality

terms of prestige stricto sensu, and, of course, Artis (1951) and Reiss (1961), that their validity
no basis at all for any consideration of how far lies in the degree of consensus which emerges,
mobility may be incongruent from one form of despite the use of quite various criteria of evalua-
stratification to another. Precisely because of the tion. The argument that this consensus points to
inevitably ‘synthetic’ nature (Ossowski, 1963) of ‘the existence of an underlying and agreed upon
socio-economic status, as indicated by prestige structure of occupational prestige’ is difficult to
scores, the analysis of mobility must be strictly sustain once it is recognized just what consen-
unidimensional. These limitations would lead sus on a prestige order entails. But the idea of a
one to suggest, therefore, that if it is accepted broadly agreed upon ordering of occupations in
that occupational prestige ratings are not valid terms of ‘goodness’ does, on the evidence in ques-
indicators of a prestige order but are being used tion, receive some clear—and not very surpris-
simply to stand proxy for socio-economic status, ing—support. Furthermore, if prestige ratings are
then it would be preferable, where possible, to taken as indicative of an occupational hierarchy
seek to measure the latter more directly—and of this kind, then the fact that they represent syn-
without any concern to combine components thetic judgments and cannot be ‘disaggregated’ is
so that a good ‘fit’ with prestige scores may be no longer a problem in the analysis of mobility
obtained. To discard the notion of prestige alto- patterns. For if mobility is being interpreted as
gether would, in this case, mean losing nothing being simply between grades of occupation of
but the possibility of terminological confusion; differing desirability in some overall sense, a
and developing separate indices of occupational unidimensional approach would appear the ap-
income, education etc., as well as some compos- propriate one. However, it must be added that
ite measure, would permit the analysis of mo- what would then be a dubious and potentially
bility in a multi-dimensional manner. In short, dangerous step would be to shift from such an
there seems no good argument for basing mobil- interpretation of specifically occupational mo-
ity research on occupational prestige ratings, in- bility to one in which conclusions were drawn
terpreted as socio-economic status scores, other regarding the stability of status groups, income
than where a lack of data on the socio-economic classes, or social strata in any sense whatsoever;
attributes of occupations makes this procedure that is, conclusions regarding social mobility as
an unavoidable pis aller. generally understood. In effect, of course, a shift
(iii) Prestige ratings may be taken as indicat- of this nature has been made in most large-scale
ing popular evaluations of the relative ‘goodness’ mobility studies carried out in the recent past.
of occupations in terms of the entire range of But while it might reasonably be held that such
prevailing criteria. In this case, related mobility a manoeuvre is unlikely to be very misleading
data are open to interpretation as showing, basi- so far as the ‘gross’ patterns of social mobility
cally, the chances of individuals entering more or are concerned, the difficulty is (apart from the
less desirable grades of occupation, given certain limitation of unidimensionality) that we have no
grades of origin. While an interpretation of the way of knowing at just what point and in what
data on these lines has rarely, if ever, been pur- ways it might turn out to be quite deceptive. Yet
sued consistently throughout a mobility study, again, the problem of validity recurs.
it is that which, on grounds of validity, could
best be defended. First, as we have already ar- The general—and rather pessimistic—conclu-
gued, grading occupations according to notions sion to which one is led is, therefore, the fol-
of their general ‘goodness’ is what respondents in lowing: that to the extent that the meaning of
occupational prestige studies appear, in the main, occupational prestige ratings is correctly con-
to be doing. Secondly, it is in regard to this un- strued, the less useful they appear to be as a basis
derstanding of prestige scores that it would seem for mobility studies which pursue the ‘classical’
most relevant to claim, following Duncan and sociological interests of mobility research.
 Goldthorpe and Hope—Occupational Grading and Occupational Prestige • 243

NOTES paragraph—by Gusfield and Schwartz (1963) in a

1. We use ‘derogate’ in this context following paper that has been curiously neglected by subse-
Shils (1968). Were it not that its usual connotations quent American writers on occupational grading.
go beyond its strict meaning, ‘disparage’—literally 9. It is a well-known problem of the structural-
‘to make unequal’—might be a preferable term. functional theory of stratification that other usable
2. On ‘stigma symbols’ as the obverse of ‘prestige criteria of the functional importance of occupational
symbols’, see Goffman (1963). roles are hard to find: employing the two criteria in
3. See Reiss (1961). question does, of course, introduce a serious degree
4. As regards the NORC [National Opinion of circularity into the argument.
Research Center] study, it is worth recalling what
is usually forgotten: that this enquiry, at least in the REFERENCES
view of those who devised it, was in fact specifically Barber, B. (1957). Social stratification. Harcourt,
aimed at finding out what people thought were the Brace & Co., New York.
best jobs, in the sense of the most desirable. Where Duncan, O.D. (1961). A socioeconomic index for
‘prestige’ and ‘standing’ are referred to in the initial all occupations. In A.J. Reiss, Occupations and
report on the study, they are obviously equated with social status. Free Press of Glencoe, New York.
desirability. See N.O.R.C. (1947). Duncan, O.D. and Artis, J.W. (1951). Social strati-
5. E.g. where respondents are rating occupations fication in a Pennsylvania rural community. Penn-
within their own status or situs areas, c.f. Gerstl and sylvania State College: Agricultural Experiment
Cohen (1964). Station Bulletin 543.
6. Our interpretation of the meaning of ‘occu- Gerstl, J. and Cohen, L.K. (1964). Dissensus, situs
pational prestige’ ratings is also consistent with the and egocentrism in occupational ranking. Brit. J.
fact that certain variations in the task set to respon- Sociol., 15, 254–61.
dents appear to make little difference to the results Goffman, E. (1963). Stigma. Prentice Hall, Engle-
achieved: e.g. whether respondents are asked to rate wood Cliffs.
occupations according to their ‘social prestige’, ‘social Goldthorpe, J.H., Lockwood, D., Bechhofer, F. and
standing’, ‘social status’, ‘general desirability’ etc.: or Platt, J. (1969). The affluent worker in the class
whether they are asked for their own opinions or structure. Cambridge University Press.
what they believe are generally prevailing opinions. Gusfield, J.R. and Schwartz, M. (1963). The mean-
It seems reasonable to suppose that if respondents ing of occupational prestige: reconsideration of
are required to grade occupations according to any the NORC scale. Amer. Sociol. Rev., 28, 265–71.
one criterion which, while rather imprecise, implies Hodge, R.W., Treiman, D.J. and Rossi, P. (1966).
a ‘better-worse’ dimension, they will produce results A comparative study of occupational prestige. In
of the kind in question; and further, that the level of Class, status and power (2nd edn.). (ed. R. Bendix
consensus in this respect is such that the distinction and S.M. Lipset). Free Press of Glencoe, New York.
between personal and general opinion is of little con- Inkeles, A. and Rossi, P. (1956). National compari-
sequence—provided that there is no suggestion of a sons of occupational prestige. Amer. J. Sociol., 61,
normative judgment being required; that is, one in 329–39.
terms of which jobs ought to be the best. N.O.R.C. (1947). Jobs and occupations: a popular
7. Cf. for example, Goldthorpe, Lockwood, evaluation. Opinion News, 9th September. Re-
Bechhofer and Platt (1969), chapters 4 and 5. printed in Class, status and power (1st edn.). (ed.
8. In fact, one might suggest the hypothesis that R. Bendix and S.M. Lipset). Free Press of Glen-
societies of the kind in which an integrated and sta- coe, New York.
ble prestige order is to be found will tend to be ones Ossowski, S. (1963). Class structure in the social con-
in which the factual order of stratification is not sciousness. Routledge, London.
commonly appraised in terms of distributive justice, Reiss, A.J. (1961). Occupations and social status. Free
or indeed envisaged as capable of being in any way Press of Glencoe, New York.
substantially different from what it is. Shils, E.A. (1968). Deference. In Social stratification
The distinction between the recognition of (ed. J.A. Jackson). Cambridge University Press.
prestige and the attribution of justice is foreshad- Svalastoga, K. (1959). Prestige, class and mobility.
owed—as are several other points in the above Gyldendal, Copenhagen.
244 • Part III: The Structure of Inequality

26 • David L. Featherman and Robert M. Hauser

Prestige or Socioeconomic Scales in the Study of

Occupational Achievement?

At least in the United States and Australia, the indicators of the dimensions of occupations
processes of allocation to educational and oc- pertinent to occupational mobility in industrial
cupational statuses from social origins (i.e., the societies and of the status attainment processes
process of stratification or of status attainment) operating therein than are socioeconomic scores.
seem largely socioeconomic in character (Feath- We reason from evidence for the United States
erman, Jones, and Hauser, 1975). Put another (Reiss, 1961; Siegel, 1971) and Great Britain
way, inter- and intragenerational movements of (Goldthorpe and Hope, 1974) that occupa-
men among categories of their own and their tional prestige scores represent a congeries of sa-
parents’ educations and occupations more closely lient dimensions or occupational characteristics.
follow the dimensions of social space defined For example, the British ratings of the “social
by the “socioeconomic” distances among occu- standing” of occupations are a linear combi-
pation groups than by the “prestige” distances nation (to the extent of 97% of their variance)
among occupations. Evidence for this interpreta- of four oblique dimensions: standard of living,
tion is drawn from parallel results for the United power and influence over other people, level of
States and Australia in which estimates for the qualifications, and value to society (Goldthorpe
structural equations of “status attainment” mod- and Hope, 1974: 14). Any two pairs of raters
els with occupations scaled in units of Duncan’s produce rankings which are modestly correlated
(1961) socioeconomic index (SEI) yield higher at best (r = .4), consistent with the notion that
coefficients of multiple determination (R2) than unique variance in prestige gradings is quite
do estimates based on occupations scaled in units high. Conversely, the mean ranks for the same
of NORC prestige (Siegel, 1971) or of Treiman’s occupations over socially and demographically
(1977) international prestige index. In addition, defined groups correlate in the range of 0.8 and
the canonical structure of generational and ca- 0.9. This common variance appears to be so-
reer occupational mobility in both societies more cioeconomic; that is, over three-quarters of the
nearly approximates a socioeconomic “space,” as linear variance in prestige scores is a reflection
the canonical weights for occupation categories of the educational and economic properties of
correlated higher with mean SEI scores for these the ranked occupations. Thus, while raters in
occupations than with mean Siegel or Treiman the United States and Britain used many and id-
scores. iosyncratic features of occupations in assessing
In interpreting these data we suggest that their relative social standing, apparently they all
prestige scores for occupations are less valid were aware of and utilized the socioeconomic

David L. Featherman and Robert M. Hauser. “Prestige or Socioeconomic Scales in the Study of Occupational
Achievement?” Sociological Methods and Research 4, issue 4 (May 1976), pp. 403–405. Copyright © 1976 by
Sage Publications. Reprinted by permission of Sage Publications.
 Featherman and Hauser—Prestige or Socioeconomic Scales? • 245

“desirability” of titles, to some extent, in reach- the social process one is studying. In instances
ing their decisions. of occupational mobility and related processes
The salience of the socioeconomic properties of status allocation, socioeconomic dimensions
of occupations across persons, groups, and per- and socioeconomic scores for occupations are
haps societies may follow from the rather similar the more central, and therefore are preferable
social organization of occupations in function- over prestige scores.
ally similar economic systems (e.g., industrial
capitalism). But more to the point of the rela- NOTES
tive centrality of “prestige” or socioeconomic di- This research was supported by NSF Grant
#44336, NICHHD Grant #HD-05876 and institu-
mensions to the process of status attainment, we
tional support from the College of Agricultural and
speculate that commonalities in prestige grades
Life Sciences, University of Wisconsin-Madison.
and in the responsiveness of these rankings to Any opinions, findings, conclusions, or recom-
socioeconomic attributes of occupations may re- mendations are those of the authors and do not
flect popular awareness of (what further compar- necessarily reflect the views of the National Science
ative research may show to be) similar processes Foundation.
of status allocation across societies. In at least
the cases of Australia and the United States, the REFERENCES
socioeconomic model, patterned after the work Blau, P. M. and O. D. Duncan (1967) The American
of Blau and Duncan (1967), yields estimates Occupational Structure. New York: Wiley.
Duncan, O.D. (1961) “A socioeconomic index for
of effect parameters which are substantially the
all occupations,” pp. 139–161 in A. Reiss, Occu-
same. Moreover, log-linear adjustments of mo-
pations and Social Status. New York: Free Press.
bility matrices for the effects of differential oc- Featherman, D. L., F. L. Jones, and R. M. Hauser
cupation structures (to wit, as provided in the (1975) “Assumptions of social mobility research
table margins) uncovers largely similar interac- in the U.S.: the case of occupational status.” So-
tions within the tables (to wit, constant patterns cial Science Research 4 (December): 329–360.
of inflow and outflow both between and within Goldthorpe, J. and K. Hope (1974) The Social Grad-
generations for both societies). ing of Occupations: A New Approach and Scale.
Our provisional conclusion is that prestige Oxford: Oxford Univ. Press.
———. (1972) “Occupational grading and occu-
scores are “error-prone” estimates of the socio-
pational prestige,” pp. 19–79 in K. Hope (ed.)
economic attributes of occupations. Whatever it
The Analysis of Social Mobility: Methods and Ap-
is that prestige scores scale—and this does not proaches. Oxford: Clarendon.
appear to be prestige in the classical sense of def- Reiss, A. J. (1961) Occupations and Social Status.
erence/derogation (see Goldthorpe and Hope, New York: Free Press.
1972)—it is substantively different from socio- Siegel, P. M. (1971) “Prestige in the American occu-
economic status. Yet one is best advised to use a pational structure.” Unpublished doctoral disser-
scale for occupations which most accurately cap- tation, University of Chicago.
tures the features of occupations having force for Treiman, D. J. (1977) Occupational Prestige in Com-
parative Perspective. New York: Academic Press.
246 • Part III: The Structure of Inequality

27 • Robert M. Hauser and John Robert Warren

Socioeconomic Indexes for Occupations

A Review, Update, and Critique

There are several reasons to focus more attention in the measurement of income or wealth. Job de-
on the collection, scaling, and analysis of occu- scriptions—contemporary or retrospective, from
pational data than has recently been the case. job-holders or from their family members—are
First, job-holding is the most important social imperfect, but the reliability and validity of
role held by most adults outside their family or carefully collected occupational data are high
household. When we meet someone new, often enough to support sustained analysis (Hauser,
our first question is, “What do you do?” and Sewell, and Warren 1994). Thus, even if we are
that is a very good question. Job-holding defines limited to retrospective questions, we can confi-
how we spend much of our time, and it pro- dently trace occupational trajectories across the
vides strong clues about the activities and cir- adult years. The same cannot be said of earnings
cumstances in which that time is spent. Second, trajectories, let alone other components of per-
job-holding tells us about the technical and so- sonal or household income or wealth.
cial skills that we bring to the labor market, and
for most people job-holding delimits current and
future economic prospects. Thus, even for per- Conceptual Issues
sons who are not attached to the labor market, It is important to distinguish between jobs and
past jobs or the jobs held by other members of occupations. A job is a specific and sometimes
the same family or household provide informa- unique bundle of activities carried out by a per-
tion about economic and social standing. Third, son in the expectation of economic remunera-
as market labor has become nearly universal tion. An occupation is an abstract category used
among adult women as well as men, it is in- to group and classify similar jobs. Such abstrac-
creasingly possible to characterize individuals in tions are often heterogeneous and idiosyncratic
terms of their own current or past jobs. Fourth, in construction, but they usually involve deter-
once we have a good job description, it is possi- minations of similarity in typical activities, in the
ble to map jobs into many classifications, scales, sites where work is performed, in the form of job
and measures. Fifth, measurement of jobs and tenure, in the skill requirements of the job, or in
occupations does not entail the same problems the product or service that results from the job.
of refusal, recall, reliability, and stability as occur There are multiple systems for the classification

The ideas, issues and theories considered in this brief piece are examined in greater detail in the following
publication: Robert M. Hauser and John Robert Warren, “Socioeconomic Indexes of Occupational Status: A
Review, Update, and Critique,” in Sociological Methodology (vol. 27), 1997, pp. 177–298. Used by permission
of the American Sociological Association. The article printed here was originally prepared by Robert M. Hauser
and John Robert Warren for Social Stratification: Class, Race, and Gender in Sociological Perspective, Second
Edition, pp. 281–286, edited by David B. Grusky. Copyright © 2001 by Westview Press.
 Hauser and Warren—Socioeconomic Indexes for Occupations • 247

of jobs and complex interdependencies between and on other job characteristics, as well as on
occupational and industrial classifications. Most occupational standing (see Jencks, Perman, and
social scientific uses of occupational data are Rainwater 1988).
based either on (a) the classification systems of In working with measures of occupational
the U.S. Bureau of the Census, which are revised social standing, we emphasize the social and
each decade at the time of the census, or (b) the economic grading of the occupational structure,
Dictionary of Occupational Titles, which is pro- rather than a priori constructions of distinct so-
duced by the Employment and Training Admin- cial classes. People are linked to jobs, not only
istration of the U.S. Department of Labor. through job-holding, but also through their rela-
Some measures of social class reflect job or tionships with others who hold or have held jobs.
personal characteristics, whereas others de- Jobs can be mapped into standard occupational
pend on occupation. For example, consider classifications, and the categories of those clas-
two widely used conceptions of “social class.” sifications may be linked to occupational char-
Wright’s (1985, 88) class typology combines acteristics. By working back through this series
concepts of ownership, authority, and expertise. of linkages, we can describe people in terms of
It requires information about a person’s educa- occupational characteristics. Such characteristics
tional attainment as well as ownership, authority, will be valid as descriptions of jobs only to the
supervision, and occupational classification. On degree that occupations are homogeneous and
the other hand, Erikson and Goldthorpe’s (1992, the intervening maps and linkages are sound. In
38–39) “class schema” is ultimately a grouping of our view, the remarkable thing about this way
occupational categories based upon Goldthorpe of measuring social and economic characteristics
and Hope’s (1974) study of occupational pres- is not that it is error prone, which would seem
tige in Great Britain (Goldthorpe 1980). Each obvious, but that it has such high reliability and
author sees his scheme as a theoretically refined validity. That it does so is a social fact, which
basis for identifying the membership of real and rests both on skill and care in classification and
discrete social classes. In Wright’s neo-Marxian coding but also on strong uniformities in social
classification, the aim is to identify modes of structure.
labor exploitation in the relations of produc-
tion. In Erikson and Goldthorpe’s neo-Weberian
classification, the class categories are designed to Occupational Prestige and
identify distinct combinations of occupational Socioeconomic Indexes of
function and employment status. Occupational Status
In our opinion, differences between the two What are the relevant status characteristics of oc-
class schemes and between them and our occu- cupations? Many discussions of occupations in
pational status measures lie more in the proxim- the stratification system begin with the concept
ity of constituent variables to jobs and persons of occupational prestige, the general level of so-
than in other theoretical or conceptual distinc- cial standing enjoyed by the incumbents of an
tions that have been debated by their authors. occupation. In the United States there have been
Other things being equal, we should expect a three major national surveys of occupational
classification based partly upon personal and job prestige, the most recent of which was carried
characteristics to be more direct and powerful in out in conjunction with the 1989 General So-
its influence than a classification based on occu- cial Survey (GSS) of the National Opinion Re-
pational characteristics would be alone. Rather search Center (NORC; Nakao and Treas 1994).
than relying on a predetermined combination of The main problem with all of these occupational
occupation and other social or economic char- prestige ratings is that they lack criterion valid-
acteristics, we suggest that investigators should ity. Prestige is not as highly correlated with other
use data on individual education and income, variables as are other measures of occupational
248 • Part III: The Structure of Inequality

social standing, specifically, measures of the so- prestige criterion and of the educational level
cioeconomic status of occupations, as indicated and wage rate of each occupation (see Hauser
by the average educational attainment and in- and Warren 1997, 203–17).
come of occupational incumbents. We then constructed socioeconomic indexes
Duncan (1961) created the first socioeco- for the total work force and, separately, for
nomic index (SEI) of occupational status. For men and for women. Our purpose in creating
forty-five census occupation lines, he ran the gender-specific indexes was to compare the be-
linear regression of the percentage of “good” havior of occupational characteristics between
or “excellent” ratings on measures of both oc- men and women, especially in relation to occu-
cupational education and occupational income pational prestige. We do not recommend routine
(see Duncan 1961 for details). This regression use of the gender-specific indexes in research.
yielded roughly equal weights for the two regres- Although the indexes for all workers, men, and
sors, a result that motivated some sociologists to women have roughly the same range and are in
characterize socioeconomic scales as a hybrid of the same metric, their statistical properties differ.
“social status” (as indexed by occupational edu- Findings based on the total, male, and female in-
cation) and economic status (see Hodge 1981). dexes are not strictly comparable (Warren, Sher-
The Duncan SEI has been updated or elabo- idan, and Hauser 1998), and, where researchers
rated in several ways, and researchers should be choose to use a composite socioeconomic index,
cautious in using the updates because of their we recommend the index based on the character-
potential lack of comparability. Most recently, as istics of all workers.
part of their work with prestige scores obtained Chastened and instructed by the example of
in the 1989 GSS, Nakao and Treas (1994) cre- Fox’s (1991) and Friendly’s (1991) reanalyses
ated socioeconomic scores for 1980-basis census of Duncan’s data, we also paid a good deal of
occupational lines by regressing their prestige attention to issues of fit and functional form.
ratings on the characteristics of male and female In the final set of regression analyses, we used
occupational incumbents in the 1980 census. several types of residual plots to identify influen-
The obvious next step is to create another set tial outliers. Based on these findings, we deleted
of socioeconomic scores, using the 1989 prestige several occupations from the regression analyses
scores as a criterion, but based upon characteris- used to estimate weights for the socioeconomic
tics of the work force in the 1990 census. scores. Several of the largest and most influen-
In an earlier paper (see Hauser and Warren tial exceptions to typical relationships among
1997), we made a special extract of occupational occupational education, wage rates, and prestige
education and earnings from the 1990 census 5 occur in common and visible jobs: business own-
percent public use sample. Throughout our anal- ers, farmers, clergy, secretaries, teachers, waiters
yses, we used the same definition of occupational and waitresses, janitors, and truck drivers. This
education as Nakao and Treas (1994), namely, finding reminds us that occupational prestige is
the percentage of people in an occupation who by no means the same as occupational socioeco-
had completed one or more years of college. nomic status, and we should respect both the
After experimenting with alternative treatments theoretical and empirical distinctions between
of earnings and income, we constructed the new them.
socioeconomic indexes using occupational wage Using estimates from our preferred mod-
rates, whereas Duncan (1961) used the percent- els, we computed total-based, male-based, and
age of occupational incumbents who had re- female-based SEI scores for all occupations. The
ported incomes of $3500 or more. Our prestige combined three sets of scores were transformed
criterion was the percentage of prestige ratings to range between 0 and 100. Our 1990-basis
above a fixed threshold. However, for statistical and 1980-basis total (TSEI), male (MSEI), and
reasons, we used a logistic transformation of the female (FSEI) scores for all occupation lines,
 Hauser and Warren—Socioeconomic Indexes for Occupations • 249

Figure 27.1. Illustrative Model of Intergenerational Stratification in Occupational Socioeconomic

Status

Father's Child's
Occ. Ed. Occ. Ed.
a a
e1
Father's Child's
SEI SEI

b b
Father's Child's
Occ. Occ.
Wage Wage
e2

the socioeconomic components of those scores, the status of a man or woman’s first occupation,
and the 1989 Nakao-Treas prestige scores and and the status of his or her current or last occu-
ratings are available elsewhere (Hauser and War- pation. In this rudimentary model, we specified
ren 1997; http://www.ssc.wisc.edu/cde/cdewp that the status of first occupation depends on
/1996papers.php). that of father’s occupation, whereas the status of
current or last occupation depends upon father’s
occupation and first occupation. To be sure, this
Structural Models of the is scarcely a complete model of the stratification
Socioeconomic Index process, but it is sufficient to generate new esti-
Subsequent analyses—presented in detail in mates of the weights of the socioeconomic index
Hauser and Warren (1997)—reveal three po- components. At each of the three stages of the
tential weaknesses in composite socioeconomic model—father’s occupation, first occupation,
indexes of occupational standing. First, gender and current or last occupation—we assume that
differences appear both in the relationships be- an SEI composite is completely determined by
tween occupational socioeconomic standing measures of occupational education and occu-
and prestige and in the socioeconomic charac- pational wages (Hauser and Warren 1997, 236).
teristics of occupational incumbents. Second, Figure 27.1 illustrates part of this model with
occupational wage rates appear to be far less measures of father’s and child’s occupational ed-
highly correlated, both within and across gen- ucation and occupational wage. The effects of
erations, than occupational education. Third, father on child identify the weights of occu-
the latter finding led us to wonder whether the pational education, a, and occupational wage,
use of prestige-validated socioeconomic indexes b, in the SEI. To our surprise, we found that
may overestimate the importance of the eco- the weight of the wage rate is negligible, that
nomic standing of occupations in the stratifi- is, b = 0. Thus, in the GSS data the process of
cation process. Thus, we developed structural occupational stratification is best described by
equation models in which the construction of relationships among occupation-based measures
socioeconomic indexes was embedded in the of educational attainment, not the combination
stratification process. These models were esti- of occupation-based measures of educational
mated using data for men and women in the attainment and wage rates. Our findings from
1994 GSS. the socioeconomic model point to occupational
For example, one of our models considered differentiation by education as a central feature
relationships among father’s occupational status, of the stratification process.
250 • Part III: The Structure of Inequality

Erikson, Robert, and John H. Goldthorpe. 1992.

Discussion The Constant Flux: A Study of Class Mobility in
We are thus led to question the value of tradi- Industrial Societies. Oxford: Clarendon Press.
tional socioeconomic indexes of occupational Fox, John. 1991. Regression Diagnostics, edited by
Michael S. Lewis-Beck. Quantitative Applica-
standing, including those that we constructed.
tions in the Social Sciences, Vol. 07-079. New-
If the 1994 GSS data are a reliable guide, we bury Park, CA: Sage.
would do better—in studies of the stratification Friendly, Michael. 1991. SAS System for Statistical
process—to index occupations by their edu- Graphics. SAS Series in Statistical Applications.
cational level alone than by any of the usual, Cary, NC: SAS Institute.
weighted combinations of educational level and Goldthorpe, John H. 1980. Social Mobility and Class
earnings. However, given the modest sensitivity Structure in Modern Britain. Oxford: Clarendon
of occupational status correlations to differences Press.
in model specification, we would not suggest Goldthorpe, John, and Keith Hope. 1974. The So-
cial Grading of Occupations: A New Approach and
any wholesale effort to reevaluate previous find-
Scale. Oxford: Clarendon Press.
ings about levels, trends, and differentials in oc- Hauser, Robert M., William H. Sewell, and John
cupational stratification. It would be sufficient, Robert Warren. 1994. “Education, Occupation,
we think, to suggest that previously estimated and Earnings in the Long Run: Men and Women
levels of correlation are slightly too low. Finally, from Adolescence to Midlife.” Paper presented at
we would caution that our findings about the the 1994 Meeting of the American Sociological
relative importance of occupational education Association. University of Wisconsin–Madison.
and occupational wage rates are specific to Hauser, Robert M., and John Robert Warren. 1997.
models of the stratification process. Just as the “Socioeconomic Indexes of Occupational Status:
A Review, Update, and Critique.” Pages 177–
relative weights of occupational education and
298 in Sociological Methodology, 1997, edited by
wage rates differ between prestige and socioeco-
Adrian Raftery. Cambridge: Blackwell Publishers.
nomic outcomes, so they may also differ across Hodge, Robert W. 1981. “The Measurement of
other outcomes (e.g., health, well-being, social Occupational Status.” Social Science Research
participation, or political choice). If there is 10:396–415.
any general conclusion to be drawn from our Jencks, Christopher, Lauri Perman, and Lee Rainwa-
analysis, it is that we ought to move toward a ter. 1988. “What Is a Good Job? A New Measure
more specific and disaggregated appraisal of the of Labor-Market Success.” American Journal of
effects of occupational characteristics on social, Sociology 93:1322–57.
Nakao, Keiko, and Judith Treas. 1994. “Updat-
psychological, economic, political, and health
ing Occupational Prestige and Socioeconomic
outcomes. Although composite measures of oc-
Scores: How the New Measures Measure Up.”
cupational status may have heuristic uses, the Pages 1–72 in Sociological Methodology, 1994,
global concept of occupational status is scien- edited by Peter Marsden. Washington, D.C.:
tifically obsolete. American Sociological Association.
Warren, John Robert, Jennifer T. Sheridan, and Rob-
REFERENCES ert M. Hauser. 1998. “Choosing a Measure of
Duncan, Otis Dudley. 1961. “A Socioeconomic Occupational Standing: How Useful are Com-
Index for All Occupations.” Pages 109–138 in posite Measures in Analyses of Gender Inequality
Occupations and Social Status, edited by Albert J. in Occupational Attainment?” Sociological Meth-
Reiss, Jr. New York: Free Press. ods and Research 27:3–76.
Wright, Erik Olin. 1985. Classes. London: Verso.
 Sørensen—Foundations of a Rent-Based Analysis • 251

The New Gradationalism?

28 • Aage B. Sørensen

Foundations of a Rent-Based
Class Analysis

There is an enormous literature on the concept economic theory and therefore avoids the prob-
of class that consists mostly of debates about lems of the labor theory of value.
which properties should be included in the
concept.
The theoretically most ambitious concept, Class and Wealth
the concept of class as exploitation originating Marx thought that classes were based on rights
with Marx, proposes a mechanism of how antag- to the payments on wealth, and Weber thought
onistic interests emerge and therefore how class property to be very important for the emergence
conflict is generated. However, the theory rests of economic classes. Exploitation is a question of
on a labor theory of value that has been aban- economic advantage obtained at the expense of
doned by economic theory. The various attempts someone else. The right to the returns on wealth
to resurrect the concept by invoking authority is indeed essential for the distribution of these
are unsatisfactory because it is not clear that au- returns, as I will show.
thority is a source of exploitation and antagonis- Rights to returns may reflect legal ownership.
tic interests. However, rights to the advantage provided by
There is another solution. This is to main- assets or resources need not be legal rights to
tain Marx’s insistence on property rights as the be effective. Following Barzel (1997), economic
source of exploitation, but to not see all wealth property rights are properly seen as reflecting an
as a source of exploitation. I propose instead individual’s ability to consume a good or asset
to restrict exploitation to inequality generated directly or consume through exchange, that is, to
by ownership or possession of rent-producing control the use of a good or an asset. Such eco-
assets. Rent-producing assets or resources cre- nomic rights may be enforceable by law and are
ate inequalities where the advantage to the then stronger, but they need not be supported by
owner is obtained at the expense of nonown- the state to be effective. Property rights are not
ers. These nonowners would be better off if the absolute and constant, and they can be changed
rent-producing asset was redistributed or elimi- through individual or collective action to protect
nated. A concept of class as exploitation based on and enhance the rights. Such action incurs trans-
the concept of rent is consistent with modern action costs that are the costs of transfer, capture,

Aage B. Sørensen. “Foundations of a Rent-Based Class Analysis,” in Approaches to Class Analysis, edited by
Erik Olin Wright (2005), pp. 119, 127–151, 193–206. Copyright © 2005 by Cambridge University Press.
Reprinted with the permission of Cambridge University Press.
252 • Part III: The Structure of Inequality

and protection of rights. When transaction costs but could also be social or psychological. Fur-
are positive, rights are not perfectly delineated, ther, let cij be the right of actor i to asset j or the
and the transfer or protection of rights will be control i exercises over j. Then the total wealth
impeded or made impossible. Positive transac- possessed by i will be:
tion costs may appear for a variety of reasons. wi = ∑ cijwj
Barzel (1997) emphasizes that some of the at- where wi is the wealth of actor i.
tributes of assets may be costly to measure and Individuals maximize their wealth by max-
not fully known to actual or potential owners. imizing the return on their assets, employing
These attributes are subject to capture by others them in the production of goods and services.
who then obtain rights to the benefits from these This usually means that they will need the use of
attributes. Transfer of rights allowing an actor to other assets controlled by other actors. Therefore,
realize the full value of his assets may be costly they need to transact with these other actors. A
because mobility is costly or prevented by force. farmer needs land to maximize the return on the
Collective action needed to rearrange property efforts and skills he may devote to farming; a
rights that create a monopoly is costly. worker needs an employer, machines, and raw
For example, in the modern corporation, materials, to realize the value of her main asset,
stockholders do not own all of the assets of the that is, her labor power. These assets will often be
organization but share it with other parties inside controlled or owned by other actors—the land-
and outside the organization that have rights to lord owns the land, the capitalist the machines.
gains from various attributes of the assets. Man- These assets can be bought by the actor needing
agers obtain some gains because stockholders them, or they can be rented. Rental here means
cannot fully control their use of assets because of transfer of use rights to the asset. The laborer
lack of information. Other employees may ob- can rent out her labor power to the capitalist in
tain advantages, for example, by retaining con- return for a wage, or the laborer can rent the
trol over their effort. That ownership is divided capitalist in return for a profit to him. Such rent-
does not mean that the concept of property as als are especially important with durable assets
the basis for exploitation should be abandoned, or resources, and even when assets are bought
as Dahrendorf (1959) proposes. For example, and owned, good accounting practice suggests
the absence of individual legal property rights calculating the payment of a rental to the owner.
to productive assets in socialist society does not The total wealth controlled by actors defines
mean that individuals do not gain from con- their class situation with respect to class as life
trolling the use of an asset in such a society. Only conditions. The assets controlled will determine
their property rights are more restricted in these their incomes and the variability of their in-
societies, and it may be difficult to identify those comes. Workers will obtain wages as a result of
who obtain the gains (Barzel 1997). their effort and skills, and their particular em-
The broader concept of property rights pro- ployment opportunities will be important for
posed by Barzel implies that individuals—even the variation in their earnings. Assets will be rel-
slaves—usually have some property rights to as- evant for the respect and prestige received from
sets under some circumstances (Barzel 1997, p. the community when knowledge of these assets
105). This means that all individuals will have permits a collective evaluation of the standing of
some wealth, even if it consists only of their abil- actors. The assets controlled will shape oppor-
ity to execute a task that can be exchanged for a tunities for transactions with other actors and
wage. therefore preferences or economic interests in
A simple formalization may be helpful. De- the meaning of interests suggested by Weber. By
note with vj the value of resource or asset j, where shaping welfare and well-being, as well as eco-
value is given by the returns to j over the lifetime nomic opportunities and the investments that
of the asset. These returns are usually monetary, maximize these opportunities, the total wealth
 Sørensen—Foundations of a Rent-Based Analysis • 253

and its composition create the behavioral dispo- prices for the rentals of assets needed to maxi-
sitions that are accountable for the inoculation mize the returns on actor j’s resources are costs
and socialization mechanisms associated with to j for those assets he does not own and returns
class as life conditions, which I will amplify fur- to j for those assets he does own. This means
ther below. his wealth is crucially dependent on the prices of
When individuals need to transact with other assets relevant for him. These prices depend in
actors to get access to assets they need to real- the usual manner on supply and demand in the
ize a return on their wealth, the actors may be market. If the supply of a certain asset, for which
able to control the supply of the needed asset. actor j pays a rental, increases, the price will fall
Costs of mobility or other costs may prevent ac- and actor j’s wealth increases since he has lower
cess to alternative suppliers, the supply may be costs. Demand in a similar manner will influence
inherently limited by nature or the supplier of the value of assets. This is the normal story.
the needed access may have created a monopoly. Suppose now that actor k controls the sup-
This may allow actors controlling the needed ply of something j needs to employ her or his
asset to charge for use of the asset that is greater assets. Actor k may own land actor j needs access
than the payment needed to cover the costs of to in order for j to obtain a return on his labor
the asset. For example, the owner of a mine in and farming skills. They will negotiate a price
an isolated location may gain an advantage from for using the land and this price is a cost to j
lower wages because workers are not able to find reducing the benefit he receives from his labor
alternative employment. Workers are thus pre- and skills. When negotiating the price, j and k
vented from realizing the returns on their labor will compare what other farmers pay for land.
that they could have obtained elsewhere, and the In the long run, competition will ensure that a
mine owner has lower costs of production and price emerges that will ensure j a sufficient reve-
therefore greater gains from production. The ad- nue to keep him alive and able to work the land
vantage thus gained from effectively being able and compensate the landlord for whatever costs
to control the supply of assets is an economic he covers, for example, for fencing. Of course,
rent. j may try to buy the land instead of renting it,
Rents may also reflect lack of full information. but for the eventual outcome this does not mat-
Executives of organizations may obtain benefits ter: the rental to the landowner is replaced by a
far in excess of what is needed to secure their em- rental of capital to finance the purchase, that is,
ployment because they are able to control cash of interest.
flows that stockholders are not able to monitor. For the competitive equilibrium to occur it
Or the supply of the asset will be limited because is important that the supply of land can vary in
its availability depends on the presence of other response to prices. It is also required that renters
specific assets. In general, rents are advantages of land are mobile and thus able to take advan-
that prevent other actors from realizing the full tage of the rental offers provided by landowners.
return on their assets. Rents are crucial for the If these conditions are not met, if land is in fixed
emergence of exploitation classes because those and limited supply or if renters of land—farmers
who benefit from rents have an interest in pro- or peasants—are prevented from being mobile
tecting their rights to the rent-producing assets, by force or law, the owners of land can charge
while those who are prevented from realizing a rental price that is larger than the hypotheti-
the full return on their assets have an interest cal competitive price that just covers their costs
in eliminating the rents. Rents thus may create related to the ownership of land. The difference
antagonistic interests and conflict. between the actual rental price and the competi-
To see how rents emerge, it is useful to con- tive price is what is called an economic rent.
sider the transactions involved in maximizing Rents are payments to assets that exceed the
returns on productive assets more closely. The competitive price or the price sufficient to cover
254 • Part III: The Structure of Inequality

costs and therefore exceeding what is sufficient by the wage because cost of monitoring prevents
to bring about the employment of the asset. the employer from adjusting wages to effort. In
Thus, a rent on asset i can be defined as: general, the salient property is that a component
of the payment obtained from the asset, or its
ri = via = vic
return, is in excess of what is needed to bring out
the supply of the asset.
where via is the actual value of i and vic is the The association of land with rent is not only
value that would have emerged under competi- an accident of history, it also reflects that tenancy
tion and equal to the costs of making the asset arrangements usually are long-term and rents
available. These values are given by the stream of therefore are long-term. In a competitive econ-
income generated by the asset over time. omy, rents may emerge in industrial production
The existence of rents depends on the ability as a result of an innovation or an import restric-
of the owner of the asset to control the supply. I tion. However, when others discover that there is
have already alluded to the classic example: the an excess profit or rent available from owning a
tenancy contract associated with feudalism. Part particular resource these others increase the sup-
of the benefit from the land goes to payment for ply of the resource if they can. This reduces the
the labor of the peasant, and another part of the excess profit and eventually makes it disappear.
benefit goes to payment for capital expenditures Marshall (1949 [1920]) calls such temporary
on the land by the landlord. The rent benefit rents quasi rents. These temporary rents are the
obtained from a tenancy arrangement is the typical rents in capitalist production and will be-
remainder, that is, the payment not needed to come important in our discussion below.
employ the peasant and keep the land fertile. It is For our analysis it is extremely important
an advantage going to the landlord because of his that rents are advantages to the owner of as-
rights to the returns on the asset that he controls. sets that are not needed to bring about the
But the rent benefit forces a disadvantage on the use of these assets. If the competitive payment
peasant, since he does not realize the full value of is enough to make the landlord willing to let
his labor and skill. the farmer use his land, then any excess is in a
The association of rents with land is not re- basic sense unnecessary. It is an advantage cost-
quired. Rent will emerge on all productive assets ing nothing. The farmer has a clear interest in
that are in fixed supply and that actors need to reducing, and if possible eliminating, the rent.
maximize their wealth; or rent may be present as The landowner has an equally clear and oppo-
a result of transaction costs involved in getting site interest in preserving the advantage provided
access to needed assets. Alfred Marshall (1949 by the rent. Rents therefore create antagonistic
[1920]) devoted much attention to the con- interests. Certain rents are especially important
cept of rent and generalized its applicability to for social structure and social change. These are
benefits received from any productive resource enduring rents that, resulting from enduring
or asset. He showed that rents also may appear property rights to rent producing assets, cause
as payments for the use of capital and labor in significant advantages and disadvantages. They
restricted supply; as payments for the use of are at the basis for class formation, as owners of
unique combinations of capitals and labors, such such assets will protect their property rights to
as those created by certain technologies; and as these assets, and nonowners will seek to elimi-
payments for unusual and rare individual abili- nate these rights.
ties that cannot be developed by training alone In summary, the individual’s total wealth, as
(musical talents, artistic creativity, athletic abil- defined by her control of assets, will determine
ity, etc.). Rents may be created in employment her life conditions and thus her class location in
relationships when workers control their effort terms of class as life conditions. It will be argued
in an attempt to increase the advantage obtained below that the consequences of these conditions
 Sørensen—Foundations of a Rent-Based Analysis • 255

are not only dependent on the total wealth, but Further, the variation in the returns on the
also on the overtime variability in the returns on wealth is important, particularly for the social-
that wealth (which define the variation in the ization patterns that emerge in different classes.
value of the wealth). Part of the total wealth An older literature found strong differences be-
may generate benefits obtained at the expense tween social classes in what was called “the abil-
of someone else, who would be better off with ity to defer gratification” (see, e.g., Schneider
a different distribution of control or property and Lysgaard 1953). This literature was largely
rights to the various attributes of the assets. This dismissed in the seventies because it was seen as
rent-generating part defines class as exploitation. reflecting an attempt to “blame the victim” (see
Below, I further develop these ideas, treating Ryan 1971). More recently, psychologists and
briefly first the idea of class as life conditions as economists have suggested a different formula-
total personal wealth and then treating in more tion of the same phenomena (see Ainslie 1992).
detail the exploitation class concepts based on People discount future rewards, often at very
rights to returns from rent-producing wealth. high rates. In particular, there are strong differ-
ences among social classes or different socioeco-
nomic levels in time orientation, with persons at
Wealth and Class As Life Conditions low socioeconomic levels having a much shorter
As noted above, there is an abundance of research time horizon than others. Those with high dis-
that shows that class as life conditions indeed is a count rates invest less in their health and edu-
powerful determinant of all kinds of outcomes. cation, and in the health and education of their
There is much less understanding of how these children.
outcomes come about. We have, of course, a rich These differences among classes in time ori-
literature on socialization that demonstrates that entation or deferred gratification patterns reflect
class is associated with important socialization the level of uncertainty in living conditions or
differences, we know about important value the variability of returns. Such uncertainty is not
differences among different classes, and we also the fault of the “victim,” but is a rational reac-
know about a host of lifestyle differences asso- tion to the expected high uncertainty of returns.
ciated with different classes. However, this only Banks also charge a higher interest rate with
moves the question one level back. What is it uncertain investments, and banks presumably
about the living conditions of different classes are acting rationally. The impact of uncertainty
that accounts for these differences? on people’s investments in themselves, and their
I propose that the answer is lifetime wealth children, should be greater, the lower the overall
and the expected variation in returns on that level of resources. Fewer resources give less of a
wealth for incumbents of different classes. There buffer.
is abundant evidence that social class accounts A person’s total wealth has two main compo-
for more outcomes the more homogeneous nents. One part is personal, human, and physi-
class categories are with respect to a variety of cal wealth that is acquired mostly outside of the
resources, or their wealth. It is important to labor market in and from families and schools,
consider not the cross-sectional distribution of but some is acquired from on-the-job training.
income, but the long-term wealth profile that The other component is wealth acquired from
determines what economists call permanent in- employment relations.
come and consumption patterns. A person who The personal part of wealth that exists, inde-
obtains a higher education will orient her life- pendent of the actor’s employment relationships,
style not to the level of income in her youth, has several components. The amount of human
but to the long-term expected living conditions capital obtained through investments in training
corresponding to the wealth associated with her and health is particularly important. There may
human capital. also be skills and abilities that command rents.
256 • Part III: The Structure of Inequality

Finally, the amount of wealth obviously de- and others with high levels of human capital and
pends on the endowments of physical capital enduring employment relations to secure the
provided by the family of origin and augmented success of their offspring.
by the person through entrepreneurship and in-
vestment independently of his involvement with
the labor market. This variation in endowments Wealth, Rents, and Exploitation
creates different incentives for investments in The issue for the formulation of a theory of ex-
human capital and the like, and these differences ploitation is to define a process by which some
explain the much emphasized nonvertical nature holder of an economic property right obtains
of some class schemes, for example, the life con- an advantage at the expense of persons with-
ditions of farmers. This component of personal out these rights. As shown above, wealth trans-
wealth is obviously of major importance for a fers made possible by the acquisition of rights
full analysis of the class structure. to assets generating economic rents satisfy this
Individuals also obtain wealth from their requirement.
employment relationships. They may have ac- Rents satisfy the requirements of the struc-
cess to on-the-job training opportunities that tural theory of inequality. Rents are created by so-
increase their human capital. A component of cial relationships of ownership of rent-producing
the human capital acquired on the job may be assets (with the obvious exceptions of rents on
specific to the job and the firm and give bar- natural abilities, to be treated later). Advantaged
gaining power to the worker. Because of specific exploitation classes are positions in the social
human capital or collective action, the worker structure that allow individuals to gain control
may gain above-market wages, increasing the or economic property rights over assets or attri-
value of his labor assets, and thus obtain a rent. butes of assets that generate rents; disadvantaged
The employment relationship will in these cir- exploitation classes are defined by the absence
cumstances be closed in contrast to the open of these rights. Changing the property relations
employment relationship characteristic of the that generate rents will change the distribution
competitive market. The resulting increase in of wealth and hence the class structure.
the expected duration of employment relation- The holder of a rent-producing asset has an
ships is crucially important for the variability interest in securing the continued flow of bene-
of returns and therefore for the consequences fits, and those denied the benefits, a clear interest
of differences in wealth, because the shorter the in obtaining the benefit by acquiring it, or by de-
employment relationships, the more variable stroying the social organizations that create the
will be the returns on wealth. The duration of rents. When actors act on their interest, they cre-
the employment relation also is important for ate social organization and processes to protect
the amount of wealth obtained in the relation- or destroy rent benefits. These arrangements,
ship. We should therefore expect persons in sta- well described by neo-Weberians, are processes
ble employment relationships to invest more in of closure and usurpation (Parkin 1979) and the
themselves and in their children. Professionals processes of moving from awareness of interests,
having large amounts of human capital and sta- through development of consciousness, to acting
ble employment relationships should invest the in pursuit of these interests (Giddens 1973). For
most. Therefore, what Erikson and Goldthorpe the scenario to unfold, not only membership,
(1992) call the “service class” should be espe- but also interests must be enduring.
cially successful securing their children’s future. The distinction between temporary rents and
Though no specific test of the idea proposed enduring rents is very important for the analy-
here exists, an abundance of research on social sis of class-formation processes. Class formation
mobility and inequality of educational oppor- not only depends on stability of membership
tunity demonstrates the ability of professionals in structural locations providing antagonistic
 Sørensen—Foundations of a Rent-Based Analysis • 257

interests, as pointed out by Goldthorpe and job; and (3) rents based on natural abilities and
Giddens. Class formation also depends on the talents, for example the height and ball-catching
rate of change in advantages or disadvantages ability to make a professional basketball team.
provided by rents. This immediately suggests I will consider some of the main properties of
that structural conflict or class conflict should monopoly rents below.
be more prevalent under feudalism than under
capitalism, for rents are more permanent under
feudalism. No revolution has occurred in an ad- Monopoly Rents
vanced capitalist society. “Artificial” or social constraints on production
The importance of the distribution over time, create monopoly rents. The monopoly may have
of the advantages provided by rents is often ig- emerged “naturally” because of increasing re-
nored. A cross-sectional inequality does not nec- turns to scale creating prohibitive costs of enter-
essarily imply a longer-term advantage provided ing production for others, as in the production
by an enduring rent. For example, according to of automobiles. Often monopolies are created by
human capital theory, the higher incomes of the governments as licenses or patents. Finally, social
higher educated compensate for higher training associations, such as trade unions or industry as-
costs and do not create a permanent advantage sociations, who agree to regulate the production
over the lifetime of the person. Thus, skills ac- of something, create monopolies. In all these
quired according to the mechanism proposed cases the supply of a product will not be sensitive
by human capital theory do not create rents and to price, and rents will appear and persist unless
therefore not classes. This is generally ignored in the monopoly is broken.
the so-called new class theory that sees classes For the sake of clarity, assume that the mo-
emerge on the basis of skills and education nopolist is working with production conditions
(Gouldner 1979; Konrad and Szélenyi 1979). that generate constant returns to scale so that
Education can, of course, create rents, but a average costs equal marginal costs. Nothing es-
measure of educational attainment as used, for sential in the present argument depends on this
example, in Wright’s class scheme cannot sep- assumption. Under perfect competition, output
arate the rent from the human capital compo- and price would be qc and pc and the price will
nent. The role of education in class analysis will correspond to the cost of the product. The mo-
be discussed further below. nopolist is able to charge a price pm above the
The types of social organization and processes price pc that would prevail with perfect compe-
that emerge around rent-producing assets differ tition. This will cause an increase in revenue per
according to which type of asset is being consid- unit produced that is an increase in the income
ered. Feudalism can be described as an elaborate of the producer over and above the amount
organization for the distribution of rent benefits needed to bring forth the production, that is, a
based in land and mercantilism as an extension of rent. In addition to the creation of the rent and
the arrangements to cover industrial production. the corresponding increase in inequality, there
In modern industrial society, there are three main will be a reduction in the wealth of society, for
types of rents to be considered, already identified less is produced at the price pm. This is the “dead-
by Marshall (1949 [1920]): (1) monopoly rents, weight loss” caused by monopoly rents and rep-
based on monopolization of the supply of an resents a welfare cost to society, which is a social
asset, for example when a cable company gains waste of resources.
a monopoly from a local government on distrib- The increase in revenue to the monopolist
uting TV signals; (2) composite rents, formed is, of course, an advantage others might want.
by unique combinations of productive rents or If others therefore successfully enter the mar-
asset specificity, for example when a worker has ket, the resulting competition might eventually
acquired skills only employable in a particular erase the monopoly rent, lowering the price to
258 • Part III: The Structure of Inequality

pc and increasing the quantity to qc. When this A variety of monopoly rents emerge in the
happens, the temporary advantage to the initial labor market. Employment rents emerge when
producer is a quasi rent. employment and jobs are closed to outsiders by
This scenario, of course, assumes that oth- the collective action of unions, by government
ers can enter production. If there are prohibitive approved certification of professions, and by
entry costs created by production technologies, other occupational licenses. Unions create rents
governments or trade associations, competition when they close shops or ration employment
will be about obtaining the monopoly. Such through apprenticeship systems. Unions may
competition is the typical case of rent seeking, that also significantly alter the distribution of rents
is, zero-sum competition over rent-producing when they obtain egalitarian wage systems that
assets. The efforts and other costs involved in will increase the wages to the least productive
trying to acquire the rent-producing property with the lowest market wages (for evidence see
or resource of course reduce the benefit of the Freeman and Medoff 1984). Professional associ-
monopoly. Indeed, those who wish to acquire ations create rents when they obtain certification
the monopoly should be willing to pay the limiting employment to the properly certified or
equivalent of benefits to obtain it, so that the when they gain control over the recruitment to
rent benefit completely disappears. The costs of the profession through control over educational
rent seeking do not increase the production of institutions; medical schools are a good exam-
society and therefore represent wasted resources ple of this. In general, educational credentials,
(Tullock 1980). This waste is in addition to the used as rationing devices for employment or for
waste represented by the deadweight loss. access to employment-specific education for pro-
The nature of the rent seeking depends on fessions, create monopoly rents to those holding
whether the monopoly can be traded in the mar- the credentials.
ket or not. If it can be traded, the sale may create Employment rents create rent seeking as
a large transfer of wealth to those who obtained zero-sum competition for positional goods
the monopoly first, and subsequent owners will (Hirsch 1976) in what I have called “vacancy
not realize a rent. For example, it is often ar- competition” (Sørensen 1983). Employment
gued that rents received by farmers, as agricul- rents are not only monopoly rents. Positions may
tural subsidies, produce higher land values and also be closed without the assistance of outside
therefore higher interest payments, eliminating agencies like unions or professional associations.
the initial advantage. Once established, the rent In internal labor markets, closed positions can be
creating monopoly is difficult to eliminate, even created without collective agreements because of
when the monopoly is fully capitalized and the the existence of composite rents created by asset
rent has disappeared. Clearly the new owners specificity, for example specific skills.
are strongly interested in receiving the rents
they have paid for, even though the advantage
to them has disappeared. An example, described Conclusion
by Tullock (1980), is a taxi medallion system, A sound basis for class concepts should be based
similar to the one in New York City. The me- on property rights to assets and resources that
dallions are sold, producing large gains to the generate economic benefits. Property rights
initial owners, but only normal rates of returns should be conceived of broadly. They are eco-
to subsequent owners. Their existence creates a nomic property rights defined as the ability to
welfare loss to consumers. This loss can be re- receive the return on an asset, directly or indi-
duced only by removing the restriction on taxi rectly through exchange (Barzel 1997). Some of
driving, something that is almost impossible to these rights may be supported by the state, and
do without forcing losses on the present owners they are then legal rights, but people also ob-
of medallions. tain advantages from rights that are not legally
 Sørensen—Foundations of a Rent-Based Analysis • 259

enforceable. Property rights define an actor’s The rent-based concept of class as exploita-
wealth and I suggest that the class as life condi- tion provides an explanation for the recent
tions reflects a person’s total wealth. Part of this increase in earnings inequality and for the prac-
wealth may be in assets that generate returns or tice of downsizing to destroy rents in the labor
payments that are rents. Rent distribution cre- market.
ates exploitation classes that may engage in col- The main class action will be by actors to
lective action. seek rents, to protect rent privileges, and to
Class as life condition is a very useful con- destroy rents in structural locations, such as
cept for analyses of how patterns of attitudes, internal labor markets. The argument here im-
behaviors, and socialization vary by location in plies that it is to the advantage of the capitalist
social structure. A prominent example is the class class to produce a labor market conforming to
concept proposed by Goldthorpe (1987; see also the assumption of neoclassical economics, and
Erikson and Goldthorpe 1992). A major objec- capitalism in the last decades has been success-
tive for constructing class schemes that account ful in eliminating rents to labor. Eliminating
for different living conditions is to identify rents in the labor market creates more efficient
homogeneous groupings with respect to total labor markets—that is, labor markets with less
wealth, type of wealth, and the variability of structure and more fluidity. A rents-free labor
wealth over time. Such groupings will differ in market will be one where simple class schemes
the amount and type of investments they make are increasingly less applicable. The destruction
in themselves and their children. We therefore of rents also creates more inequality within the
obtain class schemes that include nonvertical labor market and produces more wealth that ac-
dimensions reflecting the type of wealth pos- crues to some of those owning means of produc-
sessed and its variability over time, as generated, tion—for example, capitalists—whether they are
for example, by the stability of employment old families, new entrepreneurs, pension funds
relationships. or graduate students with mutual funds. The
The present proposal overcomes the evi- resulting society conforms to Marx’s predictions
dent problem associated with Weberian and about the nature of advanced capitalism: “The
neo-Weberian class analysis where there is no bourgeoisie, whenever it has got the upper hand,
proposal for why anyone should be upset about has put an end to all feudal, patriarchal, idyllic
their position in society and engage in class for- relations. . . . It has resolved personal worth into
mation. Enduring rents identify antagonistic in- exchange value, and in place of the numberless
terests. Those who do not own a rent-producing indefeasible chartered freedoms, has set up a
asset suffer a disadvantage as the result of the single, unconscionable freedom—Free Trade”
rent. It is in their interest to eliminate the rent, (Marx 1959a [1848], p. 323).
and in the interest of the rent receiver to protect Thus, the main prediction about the devel-
the advantage. The proposal I present here pro- opment of capitalism from rent-based class the-
vides new insights. The concept of quasi rents ory is that rents will disappear from structural
suggests that monopoly rents often are transitory locations in the labor market. This will result
and the associated interests therefore not endur- in a structureless society, without the nooks
ing. Thus, not only will stability of membership and crannies of social structure we have come
in structural locations and closure be important to expect because feudalism is slow to disappear.
for class formation, but variations over time in The result is the transfer of wealth to those who
rent advantages are important for predicting class have rights to rent-producing assets, even though
formation. Rents provide a new interpretation of these assets usually are quasi rents, for the wealth
credentialism as a device to preserve and transmit created by quasi rents is not destroyed when the
advantages from one generation to the next with rent is destroyed. As a result, we see increasing
uncertainty about the ability of offspring. wealth inequality (Wolff 1995).
260 • Part III: The Structure of Inequality

If Marx’s grand scenario for advanced cap- Hirsch, Fred, 1976 The Social Limits to Growth,
italism is interpreted as having to do with the Cambridge, MA, Harvard University Press.
distribution of rents, it is sustained. Rent seek- Konrad, György, and Ivan Szélenyi, 1979 The In-
ing creates the dynamics of capitalism, and the tellectuals on the Road to Class Power, New York,
Harcourt Brace Jovanovich.
destruction of rents in the labor market creates
Marshall, Alfred, 1949 [1920] Principles of Econom-
a structurally more homogenous working class, ics, 8th edn., London, Macmillan.
that is, a working class without structural sup- Marx, Karl, 1959a [1848] “The Communist Mani-
ports for its welfare. festo,” pp. 315–55 in Capital[illegible] The Com-
munist Manifesto, and Other Writings by Karl
REFERENCES Marx, edited by Ma[illegible] Eastman, New
Ainslie, George, 1992 Piconomics, New York, Cam- York, The Modern Library.
bridge University Press. Parkin, Frank, 1979 Marxism and Class Theory: A
Barzel, Yoram, 1997 Economic Analysis of Property Bourgeois Critique, New York, Columbia Univer-
Rights, 2nd edn., New York, Cambridge Univer- sity Press.
sity Press. Ryan, William, 1971 Blaming the Victim, New York,
Collins, Randall, 1979 The Credential Society: An Orbach and Chambers.
Historical Sociology of Education and Stratifica- Schneider, Louis, and Sverre Lysgaard, 1953 “The
tion, New York, Academic Press. Deferred Gratification Pattern,” American Socio-
Dahrendorf, R. 1959. Class and Conflict in Industrial So- logical Review 18:2, pp. 142–9.
ciety. Redwood City, CA, Stanford University Press. Sørensen, Aage B., 1983 “Processes of Allocation to
Erikson, Robert, and John H. Goldthorpe, 1992 The Open and Closed Positions in Social Structure,”
Constant Flux: A Study of Class Mobility in Indus- Zeitschrift für Soziologie 12, pp. 203–24.
trial Societies, Oxford, Oxford University Press. Tullock, Gordon, 1980 “The Transitional Gains
Freeman, Richard, and James L. Medoff, 1984 What Trap,” pp. 211–21 in Toward a Theory of the Rent
Do Unions Do?, New York, Basic Books. Seeking Society, edited by James S. Buchanan,
Giddens, Anthony, 1973 The Class Structure of the Robert D. Tollison, and Gordon Tullock, College
Advanced Societies, London, Hutchinson. Station, TX, Texas A&M University Press.
Goldthorpe, John H., 1987 Social Mobility and Class Wolff, Edward N., 1995 Top Heavy: A Study of In-
Structure in Modern Britain, 2nd edn., Oxford, creasing Inequality of Wealth in America, Wash-
Clarendon Press. ington, DC, Brookings Institution.
Gouldner, Alvin, 1979 The Future of Intellectuals and Wright, Erik Olin, 1979 Class Structure and Income
the Rise of the New Class, New York, Seabury Press. Determination, New York, Academic Press.

29 • Amartya K. Sen

From Income Inequality to Economic Inequality

Introduction ago, when I went to give a lecture at another

I begin by recounting a true story—a rather campus, I chose “Economic Inequality” as the
trivial and innocuous story, as it happens, but title of my talk. On arrival, I found the cam-
one with something of a lesson. Some years pus covered with posters announcing that I

Amartya K. Sen. “From Income Inequality to Economic Inequality,” Southern Economic Journal 64 (1997), pp.
383–401. Reprinted with the permission of Southern Economic Association.
 Sen—From Income Inequality to Economic Inequality • 261

was speaking on “Income Inequality.” When I United States as well as other countries, to il-
grumbled about it slightly, I encountered gentle, lustrate the force of this distinction. The more
but genuine, amazement that I wanted to fuss difficult issue concerns the problems involved
about such “an insignificant difference.” Indeed, in having an appropriately broad notion of eco-
the identification of economic inequality with nomic inequality that is both theoretically ade-
income inequality is fairly standard, and the two quate and empirically usable. This question, too,
are often seen as effectively synonymous in the I shall briefly try to address.
economic literature. If you tell someone that you
are working on economic inequality, it is quite
commonly assumed that you are studying in- The Need for Going Beyond Income
come distribution. Inequality
This implicit identification can be found in A convenient point of departure is A. B. Atkin-
the philosophical literature as well. For example, son’s (1970) pioneering move in the measure-
in his interesting and important paper “Equal- ment of inequality. He assessed inequality of
ity as a Moral Ideal,” Harry Frankfurt (1987), incomes by bringing in an overall social objective
the distinguished philosopher, provides a closely function and measured inequality of an income
reasoned critique of what he calls economic egal- distribution through the social loss (in terms of
itarianism, defining it as “the doctrine that there equivalent income) from that distribution in
should be no inequalities in the distribution of comparison with a corresponding equal distri-
money” (p. 21). bution. However, he took the individuals to be
The distinction, however, is important. Many symmetrical and also did not explicitly consider
of the criticisms of economic egalitarianism as a what the individuals respectively get out of their
value or a goal apply much more readily to the incomes and other circumstances.
narrow concept of income inequality than it does There is a case for going beyond this structure
to the broader notions of economic inequality. and for examining the nature of individual ad-
For example, giving a larger share of income to a vantages themselves as the constituent elements
person with more needs, say due to a disability, of social welfare (or, more generally, of social ob-
can be seen as militating against the principle of jectives). In this context, we have to take note of
equalizing incomes, but it does not go against the heterogeneities of the individuals and of their
the broader precepts of economic equality since respective nonincome circumstances.
the greater need for economic resources due The important point to note is that the val-
to the disability must be taken into account in uation of income is entirely as a means to other
judging the requirements of economic equality. ends and also that it is one means among others.
The subject of this paper is precisely the dif- A more inclusive list of means has been used by
ference between economic inequality and in- John Rawls in his theory of justice through his
come inequality. It will be argued that we ought concentration on primary goods, which include
to pay much more attention than we conven- rights, liberties and opportunities, income and
tionally do to economic inequality in an appro- wealth, and the social bases of self-respect (Rawls
priately broad sense, taking note of the fact that 1971, pp. 60–65). Income is, of course, a cru-
income inequality, on which economic analysis cially important means, but its importance lies
of inequality so often concentrates, gives a very in the fact that it helps the person to do things
inadequate and biased view of inequalities, even that she values doing and to achieve states of
of those inequalities that can be powerfully in- being that she has reasons to desire. The worth
fluenced by economic policy. There is a serious of incomes cannot stand separated from these
gulf here, and the distinction, I would argue, is deeper concerns, and a society that respects in-
of considerable importance for economic prac- dividual well-being and freedom must take note
tice as well as for economic theory. I shall also of these concerns in making interpersonal com-
present some empirical examples, involving the parisons as well as social evaluations.
262 • Part III: The Structure of Inequality

The relationship between income (and in public without shame” may require higher
other resources) on the one hand and individ- standards of clothing and other visible consump-
ual achievements and freedoms on the other is tion in a richer society than in a poorer one (as
not constant. Different types of contingencies Adam Smith [1776] had noted more than two
lead to systematic variations in the conversion centuries ago). The same parametric variability
of incomes into the distinct functionings we can may apply to the personal resources needed for
achieve (i.e., the various things we can do or be), the fulfillment of self-respect. This is primarily
and that affects the lifestyles we can enjoy. There an intersocietal variation rather than an interin-
are at least five important sources of parametric dividual variation within a given society, but the
variation. two issues are frequently interlinked.
(1) Personal heterogeneities: People have dis- (5) Distribution within the family: Incomes
parate physical characteristics connected with earned by one or more members of a family are
disability, illness, age, or gender, making their shared by all, nonearners as well as earners. The
needs diverse. For example, an ill person may family is, thus, the basic unit for consideration
need more income to fight her illness than a per- of incomes from the point of view of their use.
son without such an illness would need. While The well-being or freedom of individuals in a
the compensation needed for disadvantages will family will depend on how the family income
vary, some disadvantages may not be correctable is used in furtherance of the interests and objec-
even with more expenditure on treatment or tives of different members of the family. Thus,
care. intrafamily distribution of incomes is quite a
(2) Environmental diversities: Variations in crucial parametric variable in linking individual
environmental conditions, such as climatic cir- achievements and opportunities with the overall
cumstances (temperature ranges, rainfall, flood- level of family income. Distributional rules fol-
ing, and so on), can influence what a person gets lowed within the family (e.g., related to gender
out of a given level of income. or age or perceived needs) can make a major dif-
(3) Variations in social climate: The con- ference to the attainments and predicaments of
version of personal incomes and resources into individual members.
functionings is influenced also by social condi-
tions, including public health care and epidemi-
ology, public educational arrangements, and the Illustrations of Contrasts
prevalence or absence of crime and violence in I have presented elsewhere empirical examples
the particular location. Aside from public facili- of different types that illustrate the variability of
ties, the nature of community relationships can the relation between incomes and achievements
be very important, as the recent literature on so- (Sen 1981, 1985a, 1995, 1998). I shall take the
cial capital has tended to emphasize.1 liberty of dwelling on a few such illustrations to
(4) Differences in relational perspectives: The indicate what kind of contrasts may be involved.
commodity requirements of established patterns Figure 29.1 presents the gross national prod-
of behavior may vary between communities, uct (GNP) per head and life expectancy at birth
depending on conventions and customs. For of six countries (China, Sri Lanka, Namibia,
example, being relatively poor in a rich commu- Brazil, South Africa, and Gabon) and one size-
nity can prevent a person from achieving some able state (Kerala) within a country (India).2
elementary functionings (such as taking part in The income-poor people of Kerala or China
the life of the community) even though her in- or Sri Lanka enjoy enormously higher levels of
come, in absolute terms, may be much higher life expectancy than do the much richer popula-
than the level of income at which members of tions of Brazil, South Africa, and Namibia, not
poorer communities can function with great ease to mention Gabon. Since life expectancy varia-
and success. For example, to be able to “appear tions relate to a variety of economic influences,
 Sen—From Income Inequality to Economic Inequality • 263

Figure 29.1. GNP per Capita (US$) and Life Expectancy at Birth, 1994
8000

7000

LIFE EXPECTANCY AT BIRTH (YEARS) 1994

6000
GNP PER CAPITA (US $) 1994

5000

4000

3000

2000

1000

0
Kerala China Sri Namibia Brazil South Gabon
Lanka Africa

Sources: Country data, 1994, World Bank (1996); Kerala data, life expectancy, 1989–1993, sample registration system report-
ed in Government of India (1997b); domestic product per capita, 1992–1993, Government of India (1997a).

including epidemiological policies, health care, compare the survival prospects of African Amer-
educational facilities, and so on, the reach of eco- icans vis-à-vis the immensely poorer Chinese or
nomic opportunities is much broader than that Indians in Kerala. American blacks do much bet-
of income alone. I have had the occasion to dis- ter in terms of survival at low age groups (par-
cuss elsewhere how public policies in particular ticularly in terms of infant mortality), but the
have been quite crucial in influencing the qual- picture changes over the years.
ity of life and longevity of different populations It turns out that, in fact, the Chinese and the
(see Sen 1981; Drèze and Sen 1989). In terms Keralites decisively outlive American black men
of inequality analysis, even the direction of the in terms of surviving to older age groups. Even
inequality points oppositely when we compare American black women end up having similar
Kerala, China, and Sri Lanka on one side with survival patterns for high ages as the Chinese
Brazil, South Africa, Namibia, and Gabon on and decidedly lower survival rates than the In-
the other. dians in Kerala. So it is not only the case that
Figures 29.2 and 29.3 make a related, but American blacks suffer from relative deprivation
differently focused, comparison, bringing in the in the income space (vis-à-vis American whites),
U.S. itself. Even though the income per capita they are absolutely more deprived than the much
of African Americans is considerably lower than poorer Indians in Kerala and the Chinese (in the
that of the American white population, Afri- case of men) in terms of living to ripe, old ages.
can Americans are of course a great many times In explaining these differences between living
richer in income terms than the people of China standards judged by income per head and that
or Kerala (even after correcting for cost-of-living judged by the ability to survive to higher ages, a
differences). In this context, it is interesting to number of causal issues are relevant (including
264

Figure 29.2. Variations in Male Survival Rates by Region

Sources: U.S., 1991–1993, U.S. Department of Health and Human Services (1996); Kerala, 1991,
Government of India (1991); China, 1992, WHO (1994).

Figure 29.3. Variations in Female Survival Rates by Region

Sources: U.S., 1991–1993, U.S. Department of Health and Human Services (1996); Kerala, 1991, Govern-
ment of India (1991); China, 1992, WHO (1994).
 Sen—From Income Inequality to Economic Inequality • 265

medical insurance, public health care, elemen- closely together. For example, the GDP figures
tary education, law and order, etc.) that are not would put Haryana and Punjab very much
unrelated to economic policies and programs.3 higher than Tamil Nadu and Kerala, but in terms
Figure 29.4 compares, for different states of aspects of quality of life, exactly the opposite
within India, the values of gross domestic prod- is the case.
uct (GDP) per capita, literacy (female and male), As these illustrations exemplify (Figures
life expectancy at birth (female and male), and 29.1–4) and as can be confirmed by other sta-
total fertility rate. The last has eventual impor- tistics (see, e.g., Sen 1985a, 1995, 1998, and
tance for population growth, but its inclusion literature cited there), there are substantial dif-
here is mainly for its immediate role, at high ferences between the income-based view and the
levels, as a major restraint on the freedom and nonincome indicators of quality of life. Inequal-
wellbeing that young women can enjoy when ity comparisons will yield very different results
battered by continual bearing and rearing of depending on whether we concentrate only on
children. Since the last is viewed as a negative incomes or also on the impact of other economic
influence on the quality of life, it is measured and social influences on the quality of life.
in the opposite (downward) direction from the A further issue, which I shall not take up
zero line. in this paper (but that I have addressed else-
It is readily seen (as can also be confirmed where, particularly in Sen [1997]), concerns
by standard measures of statistical relations) that the severely negative impact of unemployment,
the relative values of GDP per capita figures are especially persistent unemployment, on the
much at variance with the nonincome indices lives that people can live. This is an especially
of aspects of quality of life (female literacy, male important issue for the assessment of quality
literacy, female life expectancy, male life expec- of life and inequality in contemporary Europe.
tancy, and low fertility rate), which all move very Even though unemployment benefits and social

Figure 29.4. Selected Indicators for Indian States

Sources: Drèze and Sen (1995, Table A.3).

266 • Part III: The Structure of Inequality

security may reduce the impact of the extraordi- just such a focus has been proposed and ele-
nary levels of high unemployment on European gantly explored already by James Meade (1976)
income inequality in particular, the persistence in his exploration of “the just economy.” So the
of unemployment leads to many other kinds of utility-based evaluation of inequality should be
deprivation (see Sen 1997 and literature cited examined first, before stepping on to the less
therein) that are not reflected at all in the income tried, and perhaps more hazardous, field of func-
statistics. An over-concentration on income in- tionings (or the freedom to function).
equality alone has permitted greater social and The possibility of interpersonal comparisons
political tolerance of unemployment in Europe of utilities was, of course, famously challenged
(and even some economic smugness vis-à-vis the by Lionel Robbins (1938) and others in the
achievement of low unemployment levels in the high days of simple positivist criticism of util-
U.S.) that cannot be justified if a broader view of itarian welfare economics. Certainly, the claim
economic inequality is taken. to a high scientific status of utility comparisons
is compromised by many practical difficulties in
relating observations to firm and indisputable
Interpersonal Utility Comparisons conclusions regarding interpersonal rankings
and Inequality of utilities and utility differences. On the other
The illustrations just presented of contrast be- hand, comparisons of pleasures and happiness
tween income and achievement deal with par- are made in our day-to-day reflections and dis-
ticular classes of indicators of quality of life course, and there is considerable discipline in the
(longevity, survival, literacy, fertility, employ- making of such comparisons. Indeed, as Donald
ment status). Illustrations can also be provided Davidson (1986) has pointed out, the nature of
to exemplify variability in the relation between our understanding and communication regard-
income and other substantive achievements such ing intrapersonal comparisons of states of happi-
as being healthy, being well-nourished, taking ness and desires are not radically different from
part in the life of the community, and so on. the corresponding interpersonal exercises. Also,
The acceptance of variability between income interpersonal comparison of utilities need not
and achievement is not, however, an adequate take an all-or-nothing form, and it is possible to
ground for a definitive rejection of income in- have “partial interpersonal comparability” with a
equality as the center of our attention in inequal- rigorous analytical structure (see Sen 1970a, b).
ity assessment—without considering whether The difficult issue in basing inequality analy-
an alternative approach would be workable and sis on interpersonal comparisons is not so much
satisfactory. Practical economics, no less than the impossibility of making such comparisons
politics, is the art of the possible, and that issue but the possibility of being misled by such
remains, even when the need for going beyond comparisons (particularly about important dif-
income inequality is well accepted. Can we really ferences in the substantive deals that people get
get an alternative, practically usable approach and the real predicaments from which they suf-
based on the broader concentration on function- fer). Our ability to take pleasure in very adverse
ings rather than incomes? circumstances tends to adapt to the hardship of
Before taking on this issue fully, I would like circumstances so that the badly placed under-
to examine a related question, proposing a differ- dogs do not typically spend their lives weeping
ent alternative to the focus on incomes. Are we over what they have missed. People learn to
not likely, it may be sensibly asked, to be served make the most of small opportunities and to cut
better by opting for a more familiar notion, like desires to size, that is, to levels that are realistic
utility, in shifting away from income inequality? under the circumstances. Thus, in the scale of
Why not the inequality of utilities as the central pleasures and desire fulfillment, the deprivation
focus of attention for inequality analysis? Indeed, of the persistent underdog finds rather muffled
 Sen—From Income Inequality to Economic Inequality • 267

and muted expression. Deprived people, varying beyond opulence, would be a complete non se-
from subjugated housewives in sexist societies to quitur. If instead of assuming that each person
the hopelessly poor in strongly stratified econo- gets the same utility as others do from the same
mies, come to terms with their deprivation, and commodity bundle, it were assumed that one
the psychological indicators of pleasure or desire gets exactly one tenth of the utility that another
fulfillment may fail to reflect the extent of real gets from each respective bundle, then that too
deprivation that these people suffer. would be perfectly consistent with all the behav-
There is, I believe, force in this criticism of ioral observations (including the shared demand
relying on interpersonal comparison of pleasures function). Congruent demand functions tell us
and desire fulfillment for making judgments nothing about the congruence of utility func-
about inequality or injustice. However, this tions, and this follows generally from the fact
critique does not touch at all the more modern that the observations on which demand func-
definition of utility as a numerical representation tions are based do not lend themselves to any
of individual choice behavior. In this interpreta- presumption about interpersonal comparisons
tion, to say that a person gets more utility from of well-being (only of commodity holdings and
x than from y is not essentially different from opulence).
saying that, given the straightforward choice be- This must not be seen as just a fussy diffi-
tween x and y, the person would choose x. The culty of theoretical interest; it can make a very
malleability or adaptation of pleasure-taking big difference in practice as well. Even if a per-
ability need not compromise the perspective of son who is disabled or ill or depressed happens
utility as real-valued representation of preference. to have the same demand function as another
However, if we see utility only as a numerical who is not disadvantaged in this way, it would be
representation of each person’s choice behavior, quite absurd to assume that she is having exactly
there is, then, no basis here for interpersonal the same utility or well-being from a given com-
comparisons of utility since each person’s choice modity bundle as the other can get from it. To
behavior is a distinct and separate entity. My attribute the same utility function to each and
choices may well reveal that I prefer a banana to treat that as the basis of interpersonal com-
to an apple, but no choice of mine would, in parison for the analysis of inequality or injustice
any obvious sense, reveal whether I prefer to be would be both epistemologically unsound and
someone else. Interpersonal comparisons deal ethically unfair.
with objects of comparison that are not objects Utility cannot, therefore, serve as a satis-
of actual choice. factory basis for interpersonal comparison for
This point is often missed when it is pre- inequality analysis, and this holds no matter
sumed that similarity of choice behavior over whether we interpret utility as pleasure or as de-
commodity space must reveal a congruence of sire fulfillment or as a numerical representation
utilities. It is often presumed that, when two of choice behavior. Indeed, as I have tried to dis-
persons are observed to have the same demand cuss, attempts to make that use arbitrarily can be
function, then they must be seen as having the pernicious for judgments of equity and justice.
same level of interpersonally comparable util-
ity for any given commodity bundle. Indeed,
much of real-income comparison proceeds on Quality of Life, Functionings, and
the basis of identifying individual advantages Capabilities
with the commodity basket enjoyed, evaluated The choice of nonutility variables in terms of
by a shared preference relation, and that proce- which inequality can be judged has been a mat-
dure is not illegitimate for making situational ter of some interest in recent years, and such
comparisons of different persons’ opulence. But concepts as the quality of life or freedom of liv-
to interpret them as utility comparison, going ing and other such notions have been invoked.4
268 • Part III: The Structure of Inequality

It is, however, important to emphasize that fo- in the relationship between resources and func-
cusing on the quality of life rather than on in- tionings. Primary goods themselves are mainly
come or wealth or on psychological satisfaction various types of general resources, and the use of
is not new in economics. Indeed, the origin of these resources to generate the capability to do
the subject of economics was strongly motivated things is subject to distinct types of variations
by the need to study the assessment of and causal (as has been already discussed), including per-
influences on the conditions of living. The moti- sonal heterogeneities, environmental diversities,
vation is stated explicitly, with reasoned justifica- variations in social climate, and differences in
tion, by Aristotle, but it is also strongly reflected relational perspective. We can have complete
in the early writings on national accounts and equality of the chosen index of primary goods,
economic prosperity by William Petty, Greg- and yet some people may be immensely more
ory King, Francois Quesnay, Antoine Lavoisier, deprived than others because of age, disabilities,
Joseph Louis Lagrange, and others. While the proneness to illness, epidemiological conditions,
national accounts devised by these pioneers es- and so on.
tablished the foundations of the modern concept I have tried to argue for some time now (Sen
of income, the focus of their attention was never 1980, 1985a, b, 1992) that, for many purposes,
confined to this one concept. They were also the appropriate space is neither that of utilities
very aware of the basic issue that the importance (as claimed by welfarists) nor that of primary
of income is instrumental and circumstantially goods (as demanded by Rawls). If the object is to
contingent rather than intrinsic and categorical. concentrate on the individual’s real opportunity
In traditional welfare economics, there has to pursue her objectives, then account would
been interest both in individual utilities and in have to be taken not only of the primary goods
individual incomes. When individuals are taken the person holds but also of the relevant personal
to be symmetrical, the two are closely linked. characteristics that govern the conversion of pri-
John Rawls has pointed to the important issue mary goods into the person’s ability to promote
that income is not the only versatile means that her ends. For example, a person who is disabled
facilitates a person’s pursuit of his or her respec- may have a larger basket of primary goods and
tive objectives. He has focused instead, as was yet have less chance to lead a normal life (or to
stated earlier, on the broader category of primary pursue her objectives) than an able-bodied per-
goods, which are general-purpose means that son with a smaller basket of primary goods. Sim-
help anyone to promote his or her ends (includ- ilarly, an older person or a person more prone to
ing “rights, liberties and opportunities, income illness can be more disadvantaged in a generally
and wealth, and the social bases of self-respect”). accepted sense even with a larger bundle of pri-
The concentration on primary goods in the mary goods.
Rawlsian framework relates to his accounting of The concept of functionings, which has dis-
individual advantage in terms of the opportu- tinctly Aristotelian roots, reflects the various
nities they enjoy to pursue their respective ob- things a person may value doing or being. The
jectives. Rawls’s Difference Principle, which is valued functionings may vary from such ele-
part of his theory of justice as fairness, assesses mentary ones as being adequately nourished and
efficiency as well as equity in terms of the re- being free from avoidable disease to very com-
spective holdings of primary goods, represented plex activities or personal states, such as being
by an index. able to take part in the life of the community
The broadening of the narrow concentration and having self-respect.
on incomes alone involved in this move is sig- The extent of each functioning enjoyed by a
nificant, but this widening of the informational person may be represented by a real number and,
focus from incomes to primary goods is not when this is the case, a person’s actual achieve-
adequate to deal with all the relevant variations ment is given by a functioning vector in an
 Sen—From Income Inequality to Economic Inequality • 269

n-dimensional space of n functionings (presum- I shall not go into the details of the approach
ing finiteness of distinct functionings). When here (which I have tried to present elsewhere
numerical representation of each functioning [Sen 1985a, 1992, 1993]). But it is useful to ask
is not possible, the analysis has to be done in whether the focus on capability is likely to be
terms of the more general framework of seeing very different from that on functionings. The ca-
the functioning achievements as a functioning pability approach can be used either with a focus
n-tuple and the capability set as a set of such on what options a person has—given by the
n-tuples in the appropriate space (this will, then, whole capability set—or on the actual function-
not be a vector space). The set of alternative ing combination she chooses—given by the cho-
functioning vectors available to her for choice is sen functioning vector. In the former procedure,
called her capability set. While the combination what may be called the options application, the
of functionings (strictly, n-tuples) a person un- focus can be on the entire K, whereas in the lat-
dertakes reflects her achievements, the capability ter, the choice application, the concentration is
set represents the freedom to achieve: the alter- more narrowly on x. The options application is
native functioning combinations from which directly concerned with the freedom to choose
this person can choose. over various alternatives, whereas the choice ap-
Figure 29.5 illustrates a functioning space plication is involved with the alternative that is
(two dimensional), with the capability set of a actually chosen. Both the versions of the capa-
person being given by region K, and from this bility approach have been used in the literature,
capability set K, the person chooses one func- and sometimes they have been combined.
tioning vector x (though this need not necessarily By a well-established tradition in economics,
be unique). It may be useful to think of choice in the real value of a set of options lies in the best
this space in terms of an indifference map of val- use that can be made of them and, given maxi-
ued living defined over the functioning vectors, mizing behavior and the absence of uncertainty,
and x can then be seen as belonging to the high- the use that is actually made. The use value of
est reachable indifference curve (as indicated).5 the opportunity, then, lies derivatively on the
The focus of this capability approach could be value of one element of it (to wit, the best op-
either on the realized functionings (what a per- tion or the actually chosen option). In this case,
son is actually able to do) or on the set of alter- the focusing on chosen functioning vector coin-
natives she has (her real opportunities). cides with concentration on the capability set.
With this type of elementary evaluation, the two
uses of the capability approach share not
Figure 29.5. Functioning Vectors and the Preference Map only the identification of a relevant space
(that of functionings) but also the focal
variable in that space (the chosen func-
tioning vector).
However, the options application can
be used in other ways as well since the
value of a set need not invariably be iden-
tified with the value of the best, or the
chosen, element of it. It is possible to at-
tach importance to having opportunities
that are not taken up. This is a natural di-
rection to go if the process through which
outcomes are generated is of significance
of its own. Indeed, choosing itself can be
seen as a valuable functioning and having
270 • Part III: The Structure of Inequality

an x when there is no alternative may be sensibly of commodity bundles (or their relative placing
distinguished from choosing x when substantial on a shared system of indifference map in the
alternatives exist.6 The importance of this type of commodity space) may tell us rather little about
consideration lies more in drawing attention to interpersonal comparisons of well-being.
broader concerns than in offering a quick reso- In evaluative traditions involving fuller spec-
lution of interpersonal comparison of freedoms ification, considerable heterogeneity is explicitly
(and thus of overall individual advantages that admitted. For example, in Rawlsian analysis,
take note of the significance of freedom). primary goods are taken to be constitutively di-
verse (including “rights, liberties and opportuni-
ties, income and wealth, and the social bases of
Weights, Valuations, and self-respect”), and Rawls (1971) deals with them
Explicitness through an overall index of primary goods hold-
I turn now to a crucial methodological issue ings. While a similar exercise of judging over a
that has received much attention in recent dis- space with heterogeneity is involved both in the
cussions involving the capability approach and Rawlsian approach and in the use of function-
related proposals. The heterogeneity of func- ings, the former is informationally poorer, for
tionings involves the need to weigh them against reasons discussed already, because of the para-
one another. This would apply to all approaches metric variation of resources and primary goods
geared to functionings, whether the concentra- vis-à-vis the opportunity of achieving high qual-
tion is on realized functioning vectors x (as with ity of living.
the choice application) or on the capability sets The problem of valuation is not, however,
K (as with the options application).7 one of an all-or-nothing kind. Some judgments,
Is this weighting requirement a special dif- with incomplete reach, follow immediately from
ficulty associated with the capability approach? the specification of a focal space. When some
This cannot be the case since heterogeneity of functionings are selected as significant, such a
factors that influence individual advantage is a focal space is specified, and the relation of dom-
pervasive feature of actual evaluation. While we inance itself leads to a partial ordering over the
can decide to close our eyes to this issue by sim- alternative states of affairs. If person i has more
ply assuming that there is something homoge- of a significant functioning than person j and
neous (e.g., income or utility) in terms of which at least as much of all such functionings, then
everyone’s overall advantage can be judged and i clearly has a higher valued functioning vector
interpersonally compared (and that variations than j has. This partial ordering can be extended
of needs, personal circumstances, etc., can be, by further specifying the possible weights. A
correspondingly, assumed away), this does not unique set of weights will, of course, be suffi-
resolve the problem—it only evades it. cient to generate a complete order, but it is typ-
Comparisons of real income involve reduc- ically not necessary. Given a range of weights
tion of bundles of different commodities into on which there is agreement (i.e., when it is
points on a real line and, in judging compara- agreed that the weights are to be chosen from
tive individual advantages, there is the further a specified range, even without any agreement
problem of interpersonal comparisons taking as to the exact point on that range), there will
note of variations of individual conditions and be a partial ordering based on the intersection
circumstances. As was discussed earlier, even of rankings (see Figure 29.6). This partial order-
when each person’s preference is taken to be the ing will get systematically extended as the range
ultimate arbitrator of well-being for that person is made more and more narrow. Somewhere in
and even when, to take a very special case, ev- the process of narrowing the range, possibly
eryone has the same demand function or prefer- well before the weights are unique, the partial
ence map, the comparison of market valuations ordering will become complete. But even with
 Sen—From Income Inequality to Economic Inequality • 271

Figure 29.6. Functioning Vectors and Dominance in nature but that are not captured by the
simple statistics of incomes and commodity
holdings. The case for such broadening of
informational focus also entails the need to
pay evaluative attention to heterogeneous
magnitudes and calls for the derivation of
partial orderings based on explicit or im-
plicit public acceptance. The normative
force of this acceptance rests substantially
on the quality and reach of public discus-
sions on matters of central social concern.
The subject of this essay, though nominally
about inequality, is ultimately as much
about the nature and importance of public
discussion on social evaluation.
In matters of public judgment, there
is no real escape from the evaluative need
for public discussion. The work of pub-
lic valuation cannot be replaced by some
an incomplete ordering, many decision prob- “super clever” assumption. Some assumptions
lems can be adequately resolved, and even those that give the appearance of working very well
that are not fully resolved can be substantially operate through hiding the choice of values and
simplified (through the rejection of dominated weights in some constructed opaqueness. For
alternatives). example, the assumption, often implicitly made,
It is thus crucial to ask, in any evaluative ex- that two persons with the same demand function
ercise of this kind, how the weights are to be must have the same relation between commod-
selected. This judgmental exercise can be re- ity bundles and well-being (no matter whether
solved only through reasoned evaluation. For one is ill and the other not, one disabled and
a given person who is making his or her own the other not, etc.) is basically a way of evading
judgments, the selection of weights will require the consideration of significant influences on
reflection rather than interpersonal agreement or well-being. That evasion becomes transparent,
a consensus. However, in arriving at an agreed as I have tried to illustrate, when we supplement
range for social evaluation (e.g., in social stud- income and commodity data by information of
ies of poverty), there has to be some kind of a other types (including matters of life and death).
reasoned consensus on weights or at least on a The exercise need not, however, be as ex-
range of weights. This is a social choice exercise acting as it may first appear and as it certainly
and requires public discussion and a democratic would be if we were not to settle for anything
understanding and acceptance. It is not a special less than getting complete orderings of interper-
problem that is associated only with the use of sonal advantages and inequalities. Our values
the functioning space. about inequality aversion are not typically of the
fine-tuning variety, getting the level of inequal-
ity “just right,” taking note of all its pros and
A Concluding Remark cons. Rather, the engagement is mainly about
The argument for shifting our attention from in- the avoidance of substantial inequalities and se-
come inequality to economic inequality relates rious injustice.
to the presence of causal influences on individ- As material for public discussion and for
ual well-being and freedom that are economic informed consensus or acceptance, the need is
272 • Part III: The Structure of Inequality

not so much for a complete ordering of inter- Sen (1991), Foster (1993), Arrow (1995), Herrero
personal advantages and of levels of inequality (1995), Puppe (1995), among others.
(which would be inevitably based on some crude 7. In the latter case, there is the further task of
assumptions and evasions) but for usable partial comparing sets rather than points in this space, and
it involves the additional issue that the importance
orderings that capture the big inequalities in a
of freedom can stretch well beyond the value of the
clear way, taking note of the various significant particular element that is chosen (except in the spe-
concerns that go well beyond the commodity cial case of elementary evaluation).
space. The focus has to be on the reach and rel-
evance of partial orderings that can be cogently REFERENCES
derived and used. Insistence on completeness Arneson, R. 1989. Equality and equality of opportu-
can be an enemy of informed and democratic nity for welfare. Philosophical Studies 56:77–93.
decision making. ———. 1990. Liberalism, distributive subjectivism,
and equal opportunity for welfare. Philosophy and
NOTES Public Affairs 19:158–94.
1. See particularly Coleman (1986) and Putnam, Arrow, K. J. 1993. Social choice and individual values.
Leonardi, and Nanetti (1993). 2nd edition. New York: Wiley.
2. While Kerala is merely a state rather than a ———. 1995. A note on freedom and flexibility. In
country, nevertheless, with its population of 29 mil- Choice, welfare and development: A festschrift for
lion, it is larger than the majority of countries in the Amartya K. Sen, edited by K. Basu, P. Pattanaik,
world. and K. Suzumura. Oxford, UK: Clarendon Press.
3. On this, see Sen (1993) and also the medical Atkinson, A. B. 1970. On the measurement of in-
literature cited therein. See also the discussion that equality. Journal of Economic Theory 2:244–63.
American black men from the Harlem district of rich Balestrino, A. 1994. Poverty and functionings: Issues
New York fall (in terms of survival) not only behind in measurement and public action. Giornale degli
the Chinese or the Indians in Kerala but also behind Economisti e Annali di Economia 53:389–406.
the famished population of Bangladesh. Balestrino, A., and E. Chiappero Martinetti. 1994.
4. See the literature considered in Nussbaum and Poverty, differentiated needs, and information.
Sen (1993). Mimeographed, University of Pisa and Univer-
5. The use of such an indifference map in ex- sity of Pavia.
plaining valuation of functionings may be of con- Casini, L., and I. Bernetti. 1997. Environment, sus-
siderable pedagogic value, especially in moving tainability, and Sen’s theory. Notizei di Politeia
from the familiarity of the commodity space to the 12:55–78.
unaccustomed functioning space. It is, neverthe- Cerioli, A., and S. Zani. 1990. A fuzzy approach
less, important to recognize that the nature of the to the measurement of poverty. In Income and
indifference map in the functioning space may not wealth distribution, inequality and poverty, ed-
altogether mirror what we standardly presume in ited by C. Dagum and M. Zenga. New York:
the case of commodity space. In particular, there Springer.
may be considerable areas of incompleteness as well Chiappero Martinetti, E. 1994. A new approach to
as fuzziness (see Sen 1985a). The recent literature evaluation of well-being and poverty by fuzzy
on fuzzy set theory can be helpful in analyzing the set theory. Giornale degli Economisti e Annali di
valuation of functioning vectors and capability sets Economia 53:367–88.
(see particularly Chiappero Martinetti 1994, 1997; ———. 1997. Standard of living evaluation based
Delbono 1989; Cerioli and Zani 1990; Balestrino on Sen’s approach: Some methodological sugges-
1994; Balestrino and Chiappero Martinetti 1994; tions. Notizie di Politeia, 12:37–53.
Ok 1995; Casini and Bernetti 1997; among other Coleman, J. S. 1986. Individual interests and collec-
contributions). tive action: Selected essays. Cambridge, UK: Cam-
6. See Sen (1985a, b). There remains the more bridge University Press.
difficult issue of determining how this process Davidson, Donald. 1986. Judging interpersonal in-
consideration should be incorporated. For various terests. In Foundations of social choice theory, ed-
alternative proposals and also axiomatized formu- ited by J. Elster and A. Hylland. Cambridge, UK:
las, see Suppes (1987), Pattanaik and Xu (1990), Cambridge University Press, pp. 195–211.
 Sen—From Income Inequality to Economic Inequality • 273

Delbono, F. 1989. Poverta come incapacita: Premesse ———. 1970b. Interpersonal aggregation and par-
teoriche, identificazione e misurazione. Rivista tial comparability. Econometrica 38:393–409.
Internazionale di Scienze Sociali 97. ———. 1980. Equality of what? In Tanner lectures
Drèze, J., and A. K. Sen. 1989. Hunger and public on human values, edited by S. McMurrin. Cam-
action. Oxford, UK: Clarendon Press. bridge, UK: Cambridge University Press.
———. 1995. India: Economic development and so- ———. 1981. Public action and the quality of life in
cial opportunity. Oxford, UK: Clarendon Press. developing countries. Oxford Bulletin of Econom-
Foster, J. E. 1993. Notes on effective freedom. ics and Statistics 43:287–319.
Unpublished paper, presented at the Stanford ———. 1985a. Commodities and capabilities. Am-
Workshop on Economic Theories of Inequality, sterdam: North-Holland.
sponsored by the MacArthur Foundation. ———. 1985b. Well-being, agency and freedom:
Frankfurt, H. 1987. Equality as a moral ideal. Ethics The Dewey lectures 1984. Journal of Philosophy
98:21–43. 82:169–221.
Government of India. 1991. Sample registration sys- ———. 1991. Welfare, preference and freedom.
tem: Fertility and mortality indicators. New Delhi. Journal of Econometrics 50:15–29.
———. 1997a. Economic survey 1996–97. New ———. 1992. Inequality reexamined. Oxford: Ox-
Delhi: Ministry of Finance. ford University Press and Cambridge, MA: Har-
———. 1997b. Women in India: A statistical profile. vard University Press.
New Delhi: Department of Education. ———. 1993. The economics of life and death. Sci-
Herrero, Carmen. 1995. Capabilities and utilities. entific American 268(May): 40–7.
Unpublished paper, University of Alicante and ———. 1995. Demography and welfare economics.
IVIE, Spain. Empirica 22:1–21.
Meade, J. E. 1976. The just economy. London: Allen ———. 1997. Inequality, unemployment and
and Unwin. contemporary Europe. Paper presented at the
Nussbaum, M., and A. K. Sen, eds. 1993. The qual- Calouste Gulbenkian Conference on A Joint
ity of life. Oxford, UK: Clarendon Press. Europe, May 1997. International Labour Review.
Ok, E. 1995. Fuzzy measurement of income in- ———. 1998. Mortality as an indicator of economic
equality: A class of fuzzy inequality measures. success or failure. (Text of the first Innocenti
Social Choice and Welfare 12:111–36. Lecture, UNICEF, Florence, 1995.) Economic
Pattanaik, P., and Y. Xu. 1990. On ranking opportu- Journal.
nity sets in terms of freedom of choice. Recherches Smith, A. 1776. An inquiry into the nature and causes
Economiques de Louvain 56:383–90. of the wealth of nations, edited by R. H. Campbell
Puppe, Clemens. 1995. Freedom of choice and and A. S. Skinner. Oxford: Clarendon Press.
rational decisions. Social Choice and Welfare Suppes, P. 1987. Maximizing freedom of decision:
12:137–53. An axiomatic analysis. In Arrow and the founda-
Putnam, R., R. Leonardi, and R. Y. Nanetti. 1993. tions of economic policy, edited by G. R. Feiwel.
Making democracy work: Civic traditions in mod- London: Macmillan.
ern Italy. Princeton, NJ: Princeton University U.S. Department of Health and Human Services.
Press. 1996. Health, United States 1995. Hyattsville,
Rawls, J. 1971. A theory of justice. Cambridge, MA: MD: U.S. Department of Health and Human
Harvard University Press. Services.
Robbins, L. 1938. Interpersonal comparisons of util- WHO. 1994. World health statistics annual 1994.
ity: A comment. Economic Journal 48:635–41. Geneva: World Health Organization.
Sen, A. K. 1970a. Collective choice and social welfare. World Bank. 1996. World development report. New
San Francisco: Holden Day. Republished Am- York: Oxford University Press.
sterdam: North-Holland, 1979.
 PART IV

The Ruling Class, Elites,

and the Upper Class

Classic Statements
30 The Ruling Class 276
31 The Power Elite 282
32 Elites and Power 292

Contemporary Statements
33 Who Rules America? 297
34 The Future of Intellectuals and the Rise of the New Class 302
35 Bobos in Paradise: The New Upper Class and How They Got There 310
36 Post-Communist Managerialism 316
37 China’s Evolving Oligarchy 322
276 • Part IV: The Ruling Class, Elites, and the Upper Class

Classic Statements

30 • Gaetano Mosca

The Ruling Class

1. Among the constant facts and tendencies that relationships of superiority or subordination, or
are to be found in all political organisms, one is in which all men would share equally in the di-
so obvious that it is apparent to the most casual rection of political affairs. If we reason otherwise
eye. In all societies—from societies that are very in theory, that is due partly to inveterate habits
meagerly developed and have barely attained that we follow in our thinking and partly to the
the dawnings of civilization, down to the most exaggerated importance that we attach to two
advanced and powerful societies—two classes of political facts that loom far larger in appearance
people appear—a class that rules and a class that than they are in reality.
is ruled. The first class, always the less numer- The first of these facts—and one has only to
ous, performs all political functions, monopo- open one’s eyes to see it—is that in every po-
lizes power and enjoys the advantages that power litical organism there is one individual who is
brings, whereas the second, the more numerous chief among the leaders of the ruling class as a
class, is directed and controlled by the first, in a whole and stands, as we say, at the helm of the
manner that is now more or less legal, now more state. That person is not always the person who
or less arbitrary and violent, and supplies the holds supreme power according to law. At times,
first, in appearance at least, with material means alongside of the hereditary king or emperor
of subsistence and with the instrumentalities there is a prime minister or a major-domo who
that are essential to the vitality of the political wields an actual power that is greater than the
organism. sovereign’s. At other times, in place of the elected
In practical life we all recognize the existence president the influential politician who has pro-
of this ruling class (or political class, as we have cured the president’s election will govern. Under
elsewhere chosen to define it).1 We all know that, special circumstances there may be, instead of
in our own country, whichever it may be, the a single person, two or three who discharge the
management of public affairs is in the hands of functions of supreme control.
a minority of influential persons, to which man- The second fact, too, is readily discernible.
agement, willingly or unwillingly, the majority Whatever the type of political organization,
defer. We know that the same thing goes on in pressures arising from the discontent of the
neighboring countries, and in fact we should be masses who are governed, from the passions by
put to it to conceive of a real world otherwise which they are swayed, exert a certain amount
organized—a world in which all men would of influence on the policies of the ruling, the po-
be directly subject to a single person without litical, class.

Gaetano Mosca. “The Ruling Class,” in The Ruling Class, edited by Arthur Livingston, translated by Hannah
D. Kahn (New York; McGraw Hill, 1939), pp. 50–54, 56–62, 65–66. Copyright © by Elisabeth Abbott.
 Mosca—The Ruling Class • 277

But the man who is at the head of the state differences between the political organizations
would certainly not be able to govern without of the United States and France, though both
the support of a numerous class to enforce re- countries are republics.
spect for his orders and to have them carried out; As we have already suggested, ingrained
and granting that he can make one individual, habits of thinking have long stood, as they still
or indeed many individuals, in the ruling class stand, in the way of scientific progress in this
feel the weight of his power, he certainly cannot matter. The classification mentioned above,
be at odds with the class as a whole or do away which divides governments into absolute mon-
with it. Even if that were possible, he would at archies, limited monarchies and republics, was
once be forced to create another class, without devised by Montesquieu and was intended to
the support of which action on his part would replace the classical categories of Aristotle, who
be completely paralyzed. On the other hand, divided governments into monarchies, aristocra-
granting that the discontent of the masses might cies and democracies. What Aristotle called a de-
succeed in deposing a ruling class, inevitably, mocracy was simply an aristocracy of fairly broad
as we shall later show, there would have to be membership. Aristotle himself was in a position
another organized minority within the masses to observe that in every Greek state, whether
themselves to discharge the functions of a ruling aristocratic or democratic, there was always one
class. Otherwise all organization, and the whole person or more who had a preponderant influ-
social structure, would be destroyed. ence. Between the day of Polybius and the day
2. From the point of view of scientific re- of Montesquieu, many writers perfected Aris-
search the real superiority of the concept of the totle’s classification by introducing into it the
ruling, or political, class lies in the fact that the concept of “mixed” governments. Later on the
varying structure of ruling classes has a prepon- modern democratic theory, which had its source
derant importance in determining the political in Rousseau, took its stand upon the concept
type, and also the level of civilization, of the dif- that the majority of the citizens in any state can
ferent peoples. According to a manner of classi- participate, and in fact ought to participate, in its
fying forms of government that is still in vogue, political life, and the doctrine of popular sover-
Turkey and Russia were both, up to a few years eignty still holds sway over many minds in spite
ago, absolute monarchies, England and Italy of the fact that modern scholarship is making it
were constitutional, or limited, monarchies, and increasingly clear that democratic, monarchical
France and the United States were classed as re- and aristocratic principles function side by side
publics. The classification was based on the fact in every political organism. We shall not stop
that, in the first two countries mentioned, head- to refute this democratic theory here, since that
ship in the state was hereditary and the chief was is the task of this work as a whole. Besides, it
nominally omnipotent; in the second two, his would be hard to destroy in a few pages a whole
office is hereditary but his powers and prerog- system of ideas that has become firmly rooted in
atives are limited; in the last two, he is elected. the human mind. As Las Casas aptly wrote in his
That classification is obviously superficial. life of Christopher Columbus, it is often much
Absolutisms though they were, there was little in harder to unlearn than to learn.
common between the manners in which Russia 3. We think it may be desirable, neverthe-
and Turkey were managed politically, the levels less, to reply at this point to an objection which
of civilization in the two countries and the orga- might very readily be made to our point of view.
nization of their ruling classes being vastly dif- If it is easy to understand that a single individ-
ferent. On the same basis, the regime in Italy, a ual cannot command a group without finding
monarchy, is much more similar to the regime in within the group a minority to support him, it
France, a republic, than it is to the regime in En- is rather difficult to grant, as a constant and nat-
gland, also a monarchy; and there are important ural fact, that minorities rule majorities, rather
278 • Part IV: The Ruling Class, Elites, and the Upper Class

than majorities minorities. But that is one of the As a rule the dominance of a warrior class
points—so numerous in all the other sciences— over a peaceful multitude is attributed to a su-
where the first impression one has of things is perposition of races, to the conquest of a rela-
contrary to what they are in reality. In reality the tively unwarlike group by an aggressive one.
dominion of an organized minority, obeying a Sometimes that is actually the case—we have ex-
single impulse, over the unorganized majority amples in India after the Aryan invasions, in the
is inevitable. The power of any minority is ir- Roman Empire after the Germanic invasions and
resistible as against each single individual in the in Mexico after the Aztec conquest. But more
majority, who stands alone before the totality of often, under certain social conditions, we note
the organized minority. At the same time, the the rise of a warlike ruling class in places where
minority is organized for the very reason that it there is absolutely no trace of a foreign conquest.
is a minority. A hundred men acting uniformly As long as a horde lives exclusively by the chase,
in concert, with a common understanding, will all individuals can easily become warriors. There
triumph over a thousand men who are not in will of course be leaders who will rule over the
accord and can therefore be dealt with one by tribe, but we will not find a warrior class rising
one. Meanwhile it will be easier for the former to exploit, and at the same time to protect, an-
to act in concert and have a mutual understand- other class that is devoted to peaceful pursuits.
ing simply because they are a hundred and not As the tribe emerges from the hunting stage and
a thousand. It follows that the larger the politi- enters the agricultural and pastoral stage, then,
cal community, the smaller will the proportion along with an enormous increase in population
of the governing minority to the governed ma- and a greater stability in the means of exerting
jority be, and the more difficult will it be for social influence, a more or less clean-cut division
the majority to organize for reaction against the into two classes will take place, one class being
minority. devoted exclusively to agriculture, the other class
However, in addition to the great advantage to war. In this event, it is inevitable that the war-
accruing to them from the fact of being orga- rior class should little by little acquire such as-
nized, ruling minorities are usually so consti- cendancy over the other as to be able to oppress
tuted that the individuals who make them up it with impunity. . . .
are distinguished from the mass of the governed 5. Everywhere—in Russia and Poland, in
by qualities that give them a certain material, in- India and medieval Europe—the ruling warrior
tellectual or even moral superiority; or else they classes acquire almost exclusive ownership of the
are the heirs of individuals who possessed such land. Land, as we have seen, is the chief source
qualities. In other words, members of a ruling of production and wealth in countries that are
minority regularly have some attribute, real or not very far advanced in civilization. But as civ-
apparent, which is highly esteemed and very in- ilization progresses, revenue from land increases
fluential in the society in which they live. proportionately. With the growth of population
4. In primitive societies that are still in the there is, at least in certain periods, an increase in
early stages of organization, military valor is the rent, in the Ricardian sense of the term, largely
quality that most readily opens access to the rul- because great centers of consumption arise—
ing, or political, class. In societies of advanced such at all times have been the great capitals and
civilization, war is the exceptional condition. It other large cities, ancient and modern. Eventu-
may be regarded as virtually normal in societies ally, if other circumstances permit, a very impor-
that are in the initial stages of their development; tant social transformation occurs. Wealth rather
and the individuals who show the greatest ability than military valor comes to be the characteristic
in war easily gain supremacy over their fellows— feature of the dominant class: the people who
the bravest become chiefs. The fact is constant, rule are the rich rather than the brave.
but the forms it may assume, in one set of cir- The condition that in the main is required
cumstances or another, vary considerably. for this transformation is that social organization
 Mosca—The Ruling Class • 279

shall have concentrated and become perfected to pressure upon the politicians who control pub-
such an extent that the protection offered by pub- lic administration. It does not prevent elections
lic authority is considerably more effective than from being carried on to the music of clinking
the protection offered by private force. In other dollars. It does not prevent whole legislatures
words, private property must be so well protected and considerable numbers of national congress-
by the practical and real efficacy of the laws as to men from feeling the influence of powerful cor-
render the power of the proprietor himself su- porations and great financiers.2 . . .
perfluous. This comes about through a series of 6. In societies in which religious beliefs are
gradual alterations in the social structure whereby strong and ministers of the faith form a special
a type of political organization, which we shall class a priestly aristocracy almost always arises
call the “feudal state,” is transformed into an es- and gains possession of a more or less impor-
sentially different type, which we shall term the tant share of the wealth and the political power.
“bureaucratic state.” We are to discuss these types Conspicuous examples of that situation would
at some length hereafter, but we may say at once be ancient Egypt (during certain periods), Brah-
that the evolution here referred to is as a rule man India and medieval Europe. Oftentimes
greatly facilitated by progress in pacific manners the priests not only perform religious functions.
and customs and by certain moral habits which They possess legal and scientific knowledge and
societies contract as civilization advances. constitute the class of highest intellectual cul-
Once this transformation has taken place, ture. Consciously or unconsciously, priestly hi-
wealth produces political power just as political erarchies often show a tendency to monopolize
power has been producing wealth. In a society learning and hamper the dissemination of the
already somewhat mature—where, therefore, methods and procedures that make the acqui-
individual power is curbed by the collective sition of knowledge possible and easy. To that
power—if the powerful are as a rule the rich, tendency may have been due, in part at least,
to be rich is to become powerful. And, in truth, the painfully slow diffusion of the demotic al-
when fighting with the mailed fist is prohibited phabet in ancient Egypt, though that alphabet
whereas fighting with pounds and pence is sanc- was infinitely more simple than the hieroglyphic
tioned, the better posts are inevitably won by script. The Druids in Gaul were acquainted with
those who are better supplied with pounds and the Greek alphabet but would not permit their
pence. rich store of sacred literature to be written down,
There are, to be sure, states of a very high requiring their pupils to commit it to memory
level of civilization which in theory are organized at the cost of untold effort. To the same outlook
on the basis of moral principles of such a char- may be attributed the stubborn and frequent use
acter that they seem to preclude this overbearing of dead languages that we find in ancient Chal-
assertiveness on the part of wealth. But this is a dea, in India, and in medieval Europe. Some-
case—and there are many such—where theoret- times, as was the case in India, lower classes have
ical principles can have no more than a limited been explicitly forbidden to acquire knowledge
application in real life. In the United States all of sacred books.
powers flow directly or indirectly from popular Specialized knowledge and really scientific
elections, and suffrage is equal for all men and culture, purged of any sacred or religious aura,
women in all the states of the Union. What is become important political forces only in a
more, democracy prevails not only in institu- highly advanced stage of civilization, and only
tions but to a certain extent also in morals. The then do they give access to membership in the
rich ordinarily feel a certain aversion to enter- ruling class to those who possess them. But in
ing public life, and the poor a certain aversion this case too, it is not so much learning in itself
to choosing the rich for elective office. But that that has political value as the practical applica-
does not prevent a rich man from being more tions that may be made of learning to the profit
influential than a poor man, since he can use of the public or the state. Sometimes all that is
280 • Part IV: The Ruling Class, Elites, and the Upper Class

required is mere possession of the mechanical who have long held possession of political func-
processes that are indispensable to the acquisi- tions, knowledge of it has never served as an or-
tion of a higher culture. This may be due to the dinary criterion for admitting to public offices
fact that on such a basis it is easier to ascertain persons who were barred from them by social
and measure the skill which a candidate has been station. The degree of mastery of the art of gov-
able to acquire—it is easier to “mark” or grade erning that a person possesses is, moreover, apart
him. So in certain periods in ancient Egypt the from exceptional cases, a very difficult thing to
profession of scribe was a road to public office determine if the person has given no practical
and power, perhaps because to have learned the demonstration that he possesses it.
hieroglyphic script was proof of long and pa- 7. In some countries we find hereditary
tient study. In modern China, again, learning castes. In such cases the governing class is explic-
the numberless characters in Chinese script has itly restricted to a given number of families, and
formed the basis of the mandarin’s education.3 In birth is the one criterion that determines entry
present-day Europe and America the class that into the class or exclusion from it. Examples are
applies the findings of modern science to war, exceedingly common. There is practically no
public administration, public works and pub- country of long-standing civilization that has
lic sanitation holds a fairly important position, not had a hereditary aristocracy at one period or
both socially and politically, and in our western another in its history. We find hereditary nobili-
world, as in ancient Rome, an altogether priv- ties during certain periods in China and ancient
ileged position is held by lawyers. They know Egypt, in India, in Greece before the wars with
the complicated legislation that arises in all peo- the Medes, in ancient Rome, among the Slavs,
ples of long-standing civilization, and they be- among the Latins and Germans of the Middle
come especially powerful if their knowledge of Ages, in Mexico at the time of the Discovery and
law is coupled with the type of eloquence that in Japan down to a few years ago.
chances to have a strong appeal to the taste of In this connection two preliminary observa-
their contemporaries. tions are in point. In the first place, all ruling
There are examples in abundance where we classes tend to become hereditary in fact if not
see that longstanding practice in directing the in law. All political forces seem to possess a qual-
military and civil organization of a community ity that in physics used to be called the force of
creates and develops in the higher reaches of the inertia. They have a tendency, that is, to remain
ruling class a real art of governing which is some- at the point and in the state in which they find
thing better than crude empiricism and better themselves. Wealth and military valor are easily
than anything that mere individual experience maintained in certain families by moral tradition
could suggest. In such circumstances aristocra- and by heredity. Qualification for important of-
cies of functionaries arise, such as the Roman fice—the habit of, and to an extent the capac-
senate, the Venetian nobility and to a certain ity for, dealing with affairs of consequence—is
extent the English aristocracy. Those bodies much more readily acquired when one has had
all stirred John Stuart Mill to admiration and a certain familiarity with them from childhood.
certainly they all three developed governments Even when academic degrees, scientific training,
that were distinguished for carefully considered special aptitudes as tested by examinations and
policies and for great steadfastness and sagacity competitions, open the way to public office,
in carrying them out. This art of governing is there is no eliminating that special advantage
not political science, though it has, at one time in favor of certain individuals which the French
or another, anticipated applications of a number call the advantage of positions déjà prises. In ac-
of the postulates of political science. However, tual fact, though examinations and competitions
even if the art of governing has now and again may theoretically be open to all, the majority
enjoyed prestige with certain classes of persons never have the resources for meeting the expense
 Mosca—The Ruling Class • 281

of long preparation, and many others are with- Now history very definitely shows the special
out the connections and kinships that set an in- abilities as well as the special defects—both very
dividual promptly on the right road, enabling marked—which have been displayed by aristoc-
him to avoid the gropings and blunders that are racies that have either remained absolutely closed
inevitable when one enters an unfamiliar envi- or have made entry into their circles difficult.
ronment without any guidance or support. The ancient Roman patriciate and the English
The democratic principle of election by and German nobilities of modern times give a
broadbased suffrage would seem at first glance ready idea of the type we refer to. Yet in dealing
to be in conflict with the tendency toward stabil- with this fact, and with the theories that tend to
ity which, according to our theory, ruling classes exaggerate its significance, we can always raise
show. But it must be noted that candidates who the same objection—that the individuals who
are successful in democratic elections are almost belong to the aristocracies in question owe their
always the ones who possess the political forces special qualities not so much to the blood that
above enumerated, which are very often hered- flows in their veins as to their very particular up-
itary. In the English, French and Italian parlia- bringing, which has brought out certain intellec-
ments we frequently see the sons, grandsons, tual and moral tendencies in them in preference
brothers, nephews and sons-in-law of members to others. . . .
and deputies, ex-members and ex-deputies. 8. Finally, if we were to keep to the idea of
In the second place, when we see a hereditary those who maintain the exclusive influence of
caste established in a country and monopoliz- the hereditary principle in the formation of rul-
ing political power, we may be sure that such a ing classes, we should be carried to a conclusion
status de jure was preceded by a similar status somewhat like the one to which we were carried
de facto. Before proclaiming their exclusive and by the evolutionary principle: The political his-
hereditary right to power the families or castes in tory of mankind ought to be much simpler than
question must have held the scepter of command it is. If the ruling class really belonged to a differ-
in a firm grasp, completely monopolizing all the ent race, or if the qualities that fit it for dominion
political forces of that country at that period. were transmitted primarily by organic heredity, it
Otherwise such a claim on their part would only is difficult to see how, once the class was formed,
have aroused the bitterest protests and provoked it could decline and lose its power. The pecu-
the bitterest struggles. liar qualities of a race are exceedingly tenacious.
Hereditary aristocracies often come to vaunt Keeping to the evolutionary theory, acquired
supernatural origins, or at least origins different capacities in the parents are inborn in their chil-
from, and superior to, those of the governed dren and, as generation succeeds generation, are
classes. Such claims are explained by a highly sig- progressively accentuated. The descendants of
nificant social fact, namely that every governing rulers, therefore, ought to become better and
class tends to justify its actual exercise of power better fitted to rule, and the other classes ought
by resting it on some universal moral principle. to see their chances of challenging or supplanting
This same sort of claim has come forward in our them become more and more remote. Now the
time in scientific trappings. A number of writers, most commonplace experience suffices to assure
developing and amplifying Darwin’s theories, one that things do not go in that way at all.
contend that upper classes represent a higher What we see is that as soon as there is a shift
level in social evolution and are therefore supe- in the balance of political forces—when, that is,
rior to lower classes by organic structure. Gum- a need is felt that capacities different from the
plowicz goes to the point of maintaining that the old should assert themselves in the management
divisions of populations into trade groups and of the state, when the old capacities, therefore,
professional classes in modern civilized countries lose some of their importance or changes in their
are based on ethnological heterogeneousness.4 distribution occur—then the manner in which
282 • Part IV: The Ruling Class, Elites, and the Upper Class

the ruling class is constituted changes also. If its nobles ceased to command the galleys and no
a new source of wealth develops in a society, if longer passed the greater part of their lives in
the practical importance of knowledge grows, if sailing the seas and in trading and fighting.
an old religion declines or a new one is born, In inorganic nature we have the example
if a new current of ideas spreads, then, simul- of our air, in which a tendency to immobility
taneously, far-reaching dislocations occur in the produced by the force of inertia is continuously
ruling class. One might say, indeed, that the in conflict with a tendency to shift about as the
whole history of civilized mankind comes down result of inequalities in the distribution of heat.
to a conflict between the tendency of dominant The two tendencies, prevailing by turn in vari-
elements to monopolize political power and ous regions on our planet, produce now calm,
transmit possession of it by inheritance, and the now wind and storm. In much the same way in
tendency toward a dislocation of old forces and human societies there prevails now the tendency
an insurgence of new forces; and this conflict that produces closed, stationary, crystallized rul-
produces an unending ferment of endosmosis ing classes, now the tendency that results in a
and exosmosis between the upper classes and more or less rapid renovation of ruling classes.
certain portions of the lower. Ruling classes de-
cline inevitably when they cease to find scope NOTES
for the capacities through which they rose to 1. Mosca, Teorica dei governi e governo parlamen-
tare, chap. I.
power, when they can no longer render the so-
2. Jannet, Le istituzioni politiche e sociali degli
cial services which they once rendered, or when
Stati Uniti d’America, part II, chap. X.
their talents and the services they render lose in 3. This was true up to a few years ago, the ex-
importance in the social environment in which amination of a mandarin covering only literary and
they live. So the Roman aristocracy declined historical studies—as the Chinese understood such
when it was no longer the exclusive source of studies, of course.
higher officers for the army, of administrators for 4. Der Rassenkampf. This notion transpires from
the commonwealth, of governors for the prov- Gumplowicz’s whole volume. It is explicitly formu-
inces. So the Venetian aristocracy declined when lated in book II, chap. XXXIII.

31 • C. Wright Mills

The Power Elite

The powers of ordinary men are circumscribed projects not their own, but from every side, such
by the everyday worlds in which they live, yet changes now press upon the men and women of
even in these rounds of job, family, and neigh- the mass society, who accordingly feel that they
borhood they often seem driven by forces they are without purpose in an epoch in which they
can neither understand nor govern. ‘Great are without power.
changes’ are beyond their control, but affect their But not all men are in this sense ordinary.
conduct and outlook none the less. The very As the means of information and of power are
framework of modern society confines them to centralized, some men come to occupy positions

C. Wright Mills. The Power Elite, pp. 3–18, 20–23, 365–367. Copyright © 1956 by C. Wright Mills. Renewed
1984 by Yaraslava Mills. New edition © 2000. Reprinted by permission of Oxford University Press, USA.
 Mills—The Power Elite • 283

in American society from which they can look such celebrities are not at the head of any dom-
down upon, so to speak, and by their decisions inating hierarchy, they do often have the power
mightily affect, the everyday worlds of ordinary to distract the attention of the public or afford
men and women. They are not made by their sensations to the masses, or, more directly, to
jobs; they set up and break down jobs for thou- gain the ear of those who do occupy positions
sands of others; they are not confined by simple of direct power. More or less unattached, as
family responsibilities; they can escape. They critics of morality and technicians of power, as
may live in many hotels and houses, but they spokesmen of God and creators of mass sensibil-
are bound by no one community. They need ity, such celebrities and consultants are part of
not merely ‘meet the demands of the day and the immediate scene in which the drama of the
hour’; in some part, they create these demands, elite is enacted. But that drama itself is centered
and cause others to meet them. Whether or not in the command posts of the major institutional
they profess their power, their technical and po- hierarchies.
litical experience of it far transcends that of the
underlying population. What Jacob Burckhardt
said of ‘great men,’ most Americans might well 1
say of their elite: ‘They are all that we are not.’1 The truth about the nature and the power of the
The power elite is composed of men whose elite is not some secret which men of affairs know
positions enable them to transcend the ordinary but will not tell. Such men hold quite various
environments of ordinary men and women; theories about their own roles in the sequence
they are in positions to make decisions having of event and decision. Often they are uncertain
major consequences. Whether they do or do not about their roles, and even more often they allow
make such decisions is less important than the their fears and their hopes to affect their assess-
fact that they do occupy such pivotal positions: ment of their own power. No matter how great
their failure to act, their failure to make deci- their actual power, they tend to be less acutely
sions, is itself an act that is often of greater con- aware of it than of the resistances of others to
sequence than the decisions they do make. For its use. Moreover, most American men of affairs
they are in command of the major hierarchies have learned well the rhetoric of public relations,
and organizations of modern society. They rule in some cases even to the point of using it when
the big corporations. They run the machinery they are alone, and thus coming to believe it.
of the state and claim its prerogatives. They di- The personal awareness of the actors is only one
rect the military establishment. They occupy the of the several sources one must examine in order
strategic command posts of the social structure, to understand the higher circles. Yet many who
in which are now centered the effective means believe that there is no elite, or at any rate none
of the power and the wealth and the celebrity of any consequence, rest their argument upon
which they enjoy. what men of affairs believe about themselves, or
The power elite are not solitary rulers. at least assert in public.
Advisers and consultants, spokesmen and There is, however, another view: those who
opinion-makers are often the captains of their feel, even if vaguely, that a compact and power-
higher thought and decision. Immediately below ful elite of great importance does now prevail in
the elite are the professional politicians of the America often base that feeling upon the histor-
middle levels of power, in the Congress and in ical trend of our time. They have felt, for exam-
the pressure groups, as well as among the new ple, the domination of the military event, and
and old upper classes of town and city and re- from this they infer that generals and admirals,
gion. Mingling with them in curious ways are as well as other men of decision influenced by
those professional celebrities who live by being them, must be enormously powerful. They hear
continually displayed but are never, so long as that the Congress has again abdicated to a hand-
they remain celebrities, displayed enough. If ful of men decisions clearly related to the issue
284 • Part IV: The Ruling Class, Elites, and the Upper Class

of war or peace. They know that the bomb was developments of decisive and immediate conse-
dropped over Japan in the name of the United quence now occur. . . .
States of America, although they were at no time Within each of the big three, the typical insti-
consulted about the matter. They feel that they tutional unit has become enlarged, has become
live in a time of big decisions; they know that administrative, and, in the power of its decisions,
they are not making any. Accordingly, as they has become centralized. Behind these develop-
consider the present as history, they infer that at ments there is a fabulous technology, for as insti-
its center, making decisions or failing to make tutions, they have incorporated this technology
them, there must be an elite of power. and guide it, even as it shapes and paces their
On the one hand, those who share this feel- developments.
ing about big historical events assume that there The economy—once a great scatter of small
is an elite and that its power is great. On the productive units in autonomous balance—has
other hand, those who listen carefully to the become dominated by two or three hundred
reports of men apparently involved in the great giant corporations, administratively and politi-
decisions often do not believe that there is an cally interrelated, which together hold the keys
elite whose powers are of decisive consequence. to economic decisions.
Both views must be taken into account, but The political order, once a decentralized set
neither is adequate. The way to understand the of several dozen states with a weak spinal cord,
power of the American elite lies neither solely has become a centralized, executive establish-
in recognizing the historic scale of events nor ment which has taken up into itself many powers
in accepting the personal awareness reported by previously scattered, and now enters into each
men of apparent decision. Behind such men and and every cranny of the social structure.
behind the events of history, linking the two, The military order, once a slim establishment
are the major institutions of modern society. in a context of distrust fed by state militia, has
These hierarchies of state and corporation and become the largest and most expensive feature of
army constitute the means of power; as such they government, and, although well versed in smil-
are now of a consequence not before equaled in ing public relations, now has all the grim and
human history—and at their summits, there clumsy efficiency of a sprawling bureaucratic
are now those command posts of modern so- domain.
ciety which offer us the sociological key to an In each of these institutional areas, the means
understanding of the role of the higher circles of power at the disposal of decision-makers have
in America. increased enormously; their central executive
Within American society, major national powers have been enhanced; within each of them
power now resides in the economic, the politi- modern administrative routines have been elabo-
cal, and the military domains. Other institutions rated and tightened up.
seem off to the side of modern history, and, on As each of these domains becomes enlarged
occasion, duly subordinated to these. No family and centralized, the consequences of its ac-
is as directly powerful in national affairs as any tivities become greater, and its traffic with the
major corporation; no church is as directly pow- others increases. The decisions of a handful of
erful in the external biographies of young men in corporations bear upon military and political as
America today as the military establishment; no well as upon economic developments around the
college is as powerful in the shaping of momen- world. The decisions of the military establish-
tous events as the National Security Council. ment rest upon and grievously affect political
Religious, educational, and family institutions life as well as the very level of economic activ-
are not autonomous centers of national power; ity. The decisions made within the political do-
on the contrary, these decentralized areas are main determine economic activities and military
increasingly shaped by the big three, in which programs. There is no longer, on the one hand,
 Mills—The Power Elite • 285

an economy, and, on the other hand, a polit- and economic, military, and political structures
ical order containing a military establishment are interlocked.
unimportant to politics and to money-making. At the pinnacle of each of the three enlarged
There is a political economy linked, in a thou- and centralized domains, there have arisen those
sand ways, with military institutions and deci- higher circles which make up the economic, the
sions. On each side of the world-split running political, and the military elites. At the top of
through central Europe and around the Asiatic the economy, among the corporate rich, there
rimlands, there is an ever-increasing interlocking are the chief executives; at the top of the political
of economic, military, and political structures.2 order, the members of the political directorate;
If there is government intervention in the corpo- at the top of the military establishment, the elite
rate economy, so is there corporate intervention of soldier-statesmen clustered in and around
in the governmental process. In the structural the Joint Chiefs of Staff and the upper echelon.
sense, this triangle of power is the source of the As each of these domains has coincided with
interlocking directorate that is most important the others, as decisions tend to become total
for the historical structure of the present. in their consequence, the leading men in each
The fact of the interlocking is clearly revealed of the three domains of power—the warlords,
at each of the points of crisis of modern capitalist the corporation chieftains, the political director-
society—slump, war, and boom. In each, men ate—tend to come together, to form the power
of decision are led to an awareness of the inter- elite of America.
dependence of the major institutional orders.
In the nineteenth century, when the scale of all
institutions was smaller, their liberal integration 2
was achieved in the automatic economy, by an The higher circles in and around these command
autonomous play of market forces, and in the posts are often thought of in terms of what their
automatic political domain, by the bargain and members possess: they have a greater share than
the vote. It was then assumed that out of the other people of the things and experiences that
imbalance and friction that followed the limited are most highly valued. From this point of view,
decisions then possible a new equilibrium would the elite are simply those who have the most of
in due course emerge. That can no longer be as- what there is to have, which is generally held to
sumed, and it is not assumed by the men at the include money, power, and prestige—as well as
top of each of the three dominant hierarchies. all the ways of life to which these lead.3 But the
For given the scope of their consequences, de- elite are not simply those who have the most, for
cisions—and indecisions—in any one of these they could not ‘have the most’ were it not for
ramify into the others, and hence top decisions their positions in the great institutions. For such
tend either to become co-ordinated or to lead institutions are the necessary bases of power, of
to a commanding indecision. It has not always wealth, and of prestige, and at the same time,
been like this. When numerous small entrepre- the chief means of exercising power, of acquir-
neurs made up the economy, for example, many ing and retaining wealth, and of cashing in the
of them could fail and the consequences still higher claims for prestige.
remain local; political and military authorities By the powerful we mean, of course, those
did not intervene. But now, given political ex- who are able to realize their will, even if oth-
pectations and military commitments, can they ers resist it. No one, accordingly, can be truly
afford to allow key units of the private corpo- powerful unless he has access to the command
rate economy to break down in slump? Increas- of major institutions, for it is over these insti-
ingly, they do intervene in economic affairs, and tutional means of power that the truly powerful
as they do so, the controlling decisions in each are, in the first instance, powerful. Higher politi-
order are inspected by agents of the other two, cians and key officials of government command
286 • Part IV: The Ruling Class, Elites, and the Upper Class

such institutional power; so do admirals and Now, we do not want by our definition to
generals, and so do the major owners and execu- prejudge whether the elite of the command posts
tives of the larger corporations. Not all power, it are conscious members of such a socially recog-
is true, is anchored in and exercised by means of nized class, or whether considerable proportions
such institutions, but only within and through of the elite derive from such a clear and distinct
them can power be more or less continuous and class. These are matters to be investigated. Yet
important. . . . in order to be able to recognize what we intend
If we took the one hundred most powerful to investigate, we must note something that all
men in America, the one hundred wealthiest, biographies and memoirs of the wealthy and the
and the one hundred most celebrated away from powerful and the eminent make clear: no mat-
the institutional positions they now occupy, ter what else they may be, the people of these
away from their resources of men and women higher circles are involved in a set of overlap-
and money, away from the media of mass com- ping ‘crowds’ and intricately connected ‘cliques.’
munication that are now focused upon them— There is a kind of mutual attraction among
then they would be powerless and poor and those who ‘sit on the same terrace’—although
uncelebrated. For power is not of a man. Wealth this often becomes clear to them, as well as to
does not center in the person of the wealthy. Ce- others, only at the point at which they feel the
lebrity is not inherent in any personality. To be need to draw the line; only when, in their com-
celebrated, to be wealthy, to have power requires mon defense, they come to understand what
access to major institutions, for the institutional they have in common, and so close their ranks
positions men occupy determine in large part against outsiders.
their chances to have and to hold these valued The idea of such ruling stratum implies that
experiences. most of its members have similar social origins,
that throughout their lives they maintain a net-
work of informal connections, and that to some
3 degree there is an interchangeability of position
The people of the higher circles may also be between the various hierarchies of money and
conceived as members of a top social stratum, power and celebrity. We must, of course, note
as a set of groups whose members know one an- at once that if such an elite stratum does exist,
other, see one another socially and at business, its social visibility and its form, for very solid
and so, in making decisions, take one another historical reasons, are quite different from those
into account. The elite, according to this con- of the noble cousinhoods that once ruled various
ception, feel themselves to be, and are felt by European nations.
others to be, the inner circle of ‘the upper social That American society has never passed
classes.’4 They form a more or less compact so- through a feudal epoch is of decisive importance
cial and psychological entity; they have become to the nature of the American elite, as well as
self-conscious members of a social class. People to American society as a historic whole. For it
are either accepted into this class or they are not, means that no nobility or aristocracy, estab-
and there is a qualitative split, rather than merely lished before the capitalist era, has stood in tense
a numerical scale, separating them from those opposition to the higher bourgeoisie. It means
who are not elite. They are more or less aware of that this bourgeoisie has monopolized not only
themselves as a social class and they behave to- wealth but prestige and power as well. It means
ward one another differently from the way they that no set of noble families has commanded the
do toward members of other classes. They accept top positions and monopolized the values that
one another, understand one another, marry one are generally held in high esteem; and certainly
another, tend to work and to think if not to- that no set has done so explicitly by inherited
gether at least alike. right. It means that no high church dignitaries
 Mills—The Power Elite • 287

or court nobilities, no entrenched landlords with No fixed ruling class, anchored in agrarian
honorific accouterments, no monopolists of high life and coming to flower in military glory, could
army posts have opposed the enriched bourgeoi- contain in America the historic thrust of com-
sie and in the name of birth and prerogative suc- merce and industry, or subordinate to itself the
cessfully resisted its self-making. capitalist elite—as capitalists were subordinated,
But this does not mean that there are no for example, in Germany and Japan. Nor could
upper strata in the United States. That they such a ruling class anywhere in the world contain
emerged from a ‘middle class’ that had no recog- that of the United States when industrialized vi-
nized aristocratic superiors does not mean they olence came to decide history. Witness the fate
remained middle class when enormous increases of Germany and Japan in the two world wars
in wealth made their own superiority possible. of the twentieth century; and indeed the fate
Their origins and their newness may have made of Britain herself and her model ruling class, as
the upper strata less visible in America than else- New York became the inevitable economic, and
where. But in America today there are in fact Washington the inevitable political capital of the
tiers and ranges of wealth and power of which western capitalist world.
people in the middle and lower ranks know very
little and may not even dream. There are fami-
lies who, in their well-being, are quite insulated 4
from the economic jolts and lurches felt by the The elite who occupy the command posts may
merely prosperous and those farther down the be seen as the possessors of power and wealth
scale. There are also men of power who in quite and celebrity; they may be seen as members of
small groups make decisions of enormous conse- the upper stratum of a capitalistic society. They
quence for the underlying population. may also be defined in terms of psychological
The American elite entered modern his- and moral criteria, as certain kinds of selected
tory as a virtually unopposed bourgeoisie. No individuals. So defined, the elite, quite simply,
national bourgeoisie, before or since, has had are people of superior character and energy.
such opportunities and advantages. Having no The humanist, for example, may conceive of
military neighbors, they easily occupied an iso- the ‘elite’ not as a social level or category, but
lated continent stocked with natural resources as a scatter of those individuals who attempt to
and immensely inviting to a willing labor force. transcend themselves, and accordingly, are more
A framework of power and an ideology for its noble, more efficient, made out of better stuff. It
justification were already at hand. Against mer- does not matter whether they are poor or rich,
cantilist restriction, they inherited the principle whether they hold high position or low, whether
of laissez-faire; against Southern planters, they they are acclaimed or despised; they are elite be-
imposed the principle of industrialism. The cause of the kind of individuals they are. The rest
Revolutionary War put an end to colonial pre- of the population is mass, which, according to
tensions to nobility, as loyalists fled the country this conception, sluggishly relaxes into uncom-
and many estates were broken up. The Jackso- fortable mediocrity.5
nian upheaval with its status revolution put an This is the sort of socially unlocated concep-
end to pretensions to monopoly of descent by tion which some American writers with conser-
the old New England families. The Civil War vative yearnings have recently sought to develop.
broke the power, and so in due course the pres- But most moral and psychological conceptions
tige, of the antebellum South’s claimants for of the elite are much less sophisticated, concern-
the higher esteem. The tempo of the whole ing themselves not with individuals but with the
capitalist development made it impossible for stratum as a whole. Such ideas, in fact, always
an inherited nobility to develop and endure in arise in a society in which some people possess
America. more than do others of what there is to possess.
288 • Part IV: The Ruling Class, Elites, and the Upper Class

People with advantages are loath to believe that candidates for the elite, and we shall do so in
they just happen to be people with advantages. terms of the major institutions making up the
They come readily to define themselves as total society of America; within and between
inherently worthy of what they possess; they each of these institutions, we shall trace the in-
come to believe themselves ‘naturally’ elite; and, terrelations of wealth and power and prestige.
in fact, to imagine their possessions and their But our main concern is with the power of those
privileges as natural extensions of their own elite who now occupy the command posts, and with
selves. In this sense, the idea of the elite as com- the role which they are enacting in the history
posed of men and women having a finer moral of our epoch.
character is an ideology of the elite as a privileged Such an elite may be conceived as omnipo-
ruling stratum, and this is true whether the ide- tent, and its powers thought of as a great hid-
ology is elite-made or made up for it by others. den design. Thus, in vulgar Marxism, events
In eras of equalitarian rhetoric, the more in- and trends are explained by reference to ‘the will
telligent or the more articulate among the lower of the bourgeoisie’; in Nazism, by reference to
and middle classes, as well as guilty members ‘the conspiracy of the Jews’; by the petty right in
of the upper, may come to entertain ideas of a America today, by reference to ‘the hidden force’
counter-elite. In western society, as a matter of of Communist spies. According to such notions
fact, there is a long tradition and varied images of the omnipotent elite as historical cause, the
of the poor, the exploited, and the oppressed elite is never an entirely visible agency. It is, in
as the truly virtuous, the wise, and the blessed. fact, a secular substitute for the will of God,
Stemming from Christian tradition, this moral being realized in a sort of providential design,
idea of a counter-elite composed of essentially except that usually non-elite men are thought
higher types condemned to a lowly station, may capable of opposing it and eventually overcom-
be and has been used by the underlying popu- ing it.
lation to justify harsh criticism of ruling elites The opposite view—of the elite as impotent—
and to celebrate utopian images of a new elite is now quite popular among liberal-minded ob-
to come. servers. Far from being omnipotent, the elites
The moral conception of the elite, however, are thought to be so scattered as to lack any co-
is not always merely an ideology of the over- herence as a historical force. Their invisibility is
privileged or a counter-ideology of the under- not the invisibility of secrecy but the invisibility
privileged. It is often a fact: having controlled of the multitude. Those who occupy the for-
experiences and select privileges, many indi- mal places of authority are so check-mated—by
viduals of the upper stratum do come in due other elites exerting pressure, or by the public as
course to approximate the types of character they an electorate, or by constitutional codes—that,
claim to embody. Even when we give up—as we although there may be upper classes, there is
must—the idea that the elite man or woman is no ruling class; although there may be men of
born with an elite character, we need not dismiss power, there is no power elite; although there
the idea that their experiences and trainings de- may be a system of stratification, it has no effec-
velop in them characters of a specific type. . . . tive top. In the extreme, this view of the elite, as
weakened by compromise and disunited to the
point of nullity, is a substitute for impersonal
5 collective fate; for, in this view, the decisions of
These several notions of the elite, when appro- the visible men of the higher circles do not count
priately understood, are intricately bound up in history.
with one another, and we shall use them all in Internationally, the image of the omnipo-
this examination of American success. We shall tent elite tends to prevail. All good events and
study each of several higher circles as offering pleasing happenings are quickly imputed by the
 Mills—The Power Elite • 289

opinion-makers to the leaders of their own na- them is none the less useful: it makes us real-
tion; all bad events and unpleasant experiences ize more clearly the question of the structure of
are imputed to the enemy abroad. In both cases, power in the United States and the position of
the omnipotence of evil rulers or of virtuous the power elite within it.
leaders is assumed. Within the nation, the use of Within each of the most powerful institu-
such rhetoric is rather more complicated: when tional orders of modern society there is a gra-
men speak of the power of their own party or dation of power. The owner of a roadside fruit
circle, they and their leaders are, of course, im- stand does not have as much power in any area
potent; only ‘the people’ are omnipotent. But, of social or economic or political decision as the
when they speak of the power of their opponent’s head of a multi-million-dollar fruit corporation;
party or circle, they impute to them omnipo- no lieutenant on the line is as powerful as the
tence; ‘the people’ are now powerlessly taken in. Chief of Staff in the Pentagon; no deputy sheriff
More generally, American men of power carries as much authority as the President of the
tend, by convention, to deny that they are pow- United States. Accordingly, the problem of de-
erful. No American runs for office in order to fining the power elite concerns the level at which
rule or even govern, but only to serve; he does we wish to draw the line. By lowering the line,
not become a bureaucrat or even an official, but we could define the elite out of existence; by rais-
a public servant. And nowadays, as I have al- ing it, we could make the elite a very small cir-
ready pointed out, such postures have become cle indeed. In a preliminary and minimum way,
standard features of the public-relations pro- we draw the line crudely, in charcoal as it were:
grams of all men of power. So firm a part of the By the power elite, we refer to those political,
style of power-wielding have they become that economic, and military circles which as an in-
conservative writers readily misinterpret them as tricate set of overlapping cliques share decisions
indicating a trend toward an ‘amorphous power having at least national consequences. In so far
situation.’ as national events are decided, the power elite are
But the ‘power situation’ of America today those who decide them. . . .
is less amorphous than is the perspective of
those who see it as a romantic confusion. It is
less a flat, momentary ‘situation’ than a graded, 6
durable structure. And if those who occupy its It is not my thesis that for all epochs of human
top grades are not omnipotent, neither are they history and in all nations, a creative minority, a
impotent. It is the form and the height of the ruling class, an omnipotent elite, shape all histor-
gradation of power that we must examine if we ical events. Such statements, upon careful exam-
would understand the degree of power held and ination, usually turn out to be mere tautologies,6
exercised by the elite. and even when they are not, they are so entirely
If the power to decide such national issues as general as to be useless in the attempt to under-
are decided were shared in an absolutely equal stand the history of the present. The minimum
way, there would be no power elite; in fact, there definition of the power elite as those who decide
would be no gradation of power, but only a rad- whatever is decided of major consequence, does
ical homogeneity. At the opposite extreme as not imply that the members of this elite are al-
well, if the power to decide issues were absolutely ways and necessarily the history-makers; neither
monopolized by one small group, there would does it imply that they never are. We must not
be no gradation of power; there would simply confuse the conception of the elite, which we
be this small group in command, and below it, wish to define, with one theory about their role:
the undifferentiated, dominated masses. Ameri- that they are the history-makers of our time. To
can society today represents neither the one nor define the elite, for example, as ‘those who rule
the other of these extremes, but a conception of America’ is less to define a conception than to
290 • Part IV: The Ruling Class, Elites, and the Upper Class

state one hypothesis about the role and power of all of them add up in a way no man intended—
that elite. No matter how we might define the to history as fate. But not all epochs are equally
elite, the extent of its members’ power is sub- fateful. As the circle of those who decide is nar-
ject to historical variation. If, in a dogmatic way, rowed, as the means of decision are centralized
we try to include that variation in our generic and the consequences of decisions become enor-
definition, we foolishly limit the use of a needed mous, then the course of great events often rests
conception. If we insist that the elite be defined upon the decisions of determinable circles. This
as a strictly coordinated class that continually does not necessarily mean that the same circle of
and absolutely rules, we are closing off from our men follow through from one event to another
view much to which the term more modestly de- in such a way that all of history is merely their
fined might open to our observation. In short, plot. The power of the elite does not necessarily
our definition of the power elite cannot properly mean that history is not also shaped by a series
contain dogma concerning the degree and kind of small decisions, none of which are thought
of power that ruling groups everywhere have. out. It does not mean that a hundred small ar-
Much less should it permit us to smuggle into rangements and compromises and adaptations
our discussion a theory of history. may not be built into the going policy and the
During most of human history, historical living event. The idea of the power elite implies
change has not been visible to the people who nothing about the process of decision-making as
were involved in it, or even to those enacting it. such: it is an attempt to delimit the social areas
Ancient Egypt and Mesopotamia, for example, within which that process, whatever its charac-
endured for some four hundred generations with ter, goes on. It is a conception of who is involved
but slight changes in their basic structure. That in the process.
is six and a half times as long as the entire Chris- The degree of foresight and control of those
tian era, which has only prevailed some sixty who are involved in decisions that count may
generations; it is about eighty times as long as the also vary. The idea of the power elite does not
five generations of the United States’ existence. mean that the estimations and calculated risks
But now the tempo of change is so rapid, and upon which decisions are made are not often
the means of observation so accessible, that the wrong and that the consequences are sometimes,
interplay of event and decision seems often to indeed often, not those intended. Often those
be quite historically visible, if we will only look who make decisions are trapped by their own
carefully and from an adequate vantage point. inadequacies and blinded by their own errors.
When knowledgeable journalists tell us Yet in our time the pivotal moment does arise,
that ‘events, not men, shape the big decisions,’ and at that moment, small circles do decide or
they are echoing the theory of history as For- fail to decide. In either case, they are an elite of
tune, Chance, Fate, or the work of The Unseen power. The dropping of the A-bombs over Japan
Hand. For ‘events’ is merely a modern word for was such a moment; the decision on Korea was
these older ideas, all of which separate men from such a moment; the confusion about Quemoy
history-making, because all of them lead us to and Matsu, as well as before Dienbienphu were
believe that history goes on behind men’s backs. such moments; the sequence of maneuvers
History is drift with no mastery; within it there which involved the United States in World War
is action but no deed; history is mere happening II was such a ‘moment.’ Is it not true that much
and the event intended by no one.7 of the history of our times is composed of such
The course of events in our time depends moments? And is not that what is meant when
more on a series of human decisions than on any it is said that we live in a time of big decisions,
inevitable fate. The sociological meaning of ‘fate’ of decisively centralized power?
is simply this: that, when the decisions are innu- Most of us do not try to make sense of our age
merable and each one is of small consequence, by believing in a Greek-like, eternal recurrence,
 Mills—The Power Elite • 291

nor by a Christian belief in a salvation to come, in command of them have come into command
nor by any steady march of human progress. of instruments of rule quite unsurpassed in the
Even though we do not reflect upon such mat- history of mankind. And we are not yet at the
ters, the chances are we believe with Burckhardt climax of their development. We can no longer
that we live in a mere succession of events; that lean upon or take soft comfort from the histor-
sheer continuity is the only principle of history. ical ups and downs of ruling groups of previous
History is merely one thing after another; his- epochs. In that sense, Hegel is correct: we learn
tory is meaningless in that it is not the realiza- from history that we cannot learn from it.
tion of any determinate plot. It is true, of course,
that our sense of continuity, our feeling for the NOTES
history of our time, is affected by crisis. But we 1. Jacob Burckhardt, Force and Freedom (New
York: Pantheon Books, 1943), pp. 303 ff.
seldom look beyond the immediate crisis or the
2. Cf. Hans Gerth and C. Wright Mills, Charac-
crisis felt to be just ahead. We believe neither
ter and Social Structure (New York: Harcourt, Brace,
in fate nor providence; and we assume, without 1953), pp. 457 ff.
talking about it, that ‘we’—as a nation—can de- 3. The statistical idea of choosing some value and
cisively shape the future but that ‘we’ as individ- calling those who have the most of it an elite derives,
uals somehow cannot do so. in modern times, from the Italian economist, Pareto,
Any meaning history has, ‘we’ shall have who puts the central point in this way: ‘Let us as-
to give to it by our actions. Yet the fact is that sume that in every branch of human activity each
although we are all of us within history we do individual is given an index which stands as a sign of
not all possess equal powers to make history. To his capacity, very much the way grades are given in
the various subjects in examinations in school. The
pretend that we do is sociological nonsense and
highest type of lawyer, for instance, will be given 10.
political irresponsibility. It is nonsense because The man who does not get a client will be given 1—
any group or any individual is limited, first of reserving zero for the man who is an out-and-out
all, by the technical and institutional means of idiot. To the man who has made his millions—hon-
power at its command; we do not all have equal estly or dishonestly as the case may be—we will give
access to the means of power that now exist, nor 10. To the man who has earned his thousands we
equal influence over their use. To pretend that will give 6; to such as just manage to keep out of the
‘we’ are all history-makers is politically irrespon- poor-house, 1, keeping zero for those who get in. . . .
sible because it obfuscates any attempt to locate So let us make a class of people who have the highest
indices in their branch of activity, and to that class
responsibility for the consequential decisions of
give the name of elite.’ Vilfredo Pareto, The Mind
men who do have access to the means of power.
and Society (New York: Harcourt, Brace, 1935), par.
From even the most superficial examination 2027 and 2031. Those who follow this approach
of the history of the western society we learn that end up not with one elite, but with a number corre-
the power of decision-makers is first of all limited sponding to the number of values they select. Like
by the level of technique, by the means of power many rather abstract ways of reasoning, this one is
and violence and organization that prevail in a useful because it forces us to think in a clear-cut way.
given society. In this connection we also learn For a skillful use of this approach, see the work of
that there is a fairly straight line running upward Harold D. Lasswell, in particular, Politics: Who Gets
What, When, How (New York: McGraw-Hill, 1936);
through the history of the West; that the means
and for a more systematic use, H. D. Lasswell and
of oppression and exploitation, of violence and
Abraham Kaplan, Power and Society (New Haven:
destruction, as well as the means of production Yale University Press, 1950).
and reconstruction, have been progressively en- 4. The conception of the elite as members
larged and increasingly centralized. of a top social stratum, is, of course, in line with
As the institutional means of power and the the prevailing common-sense view of stratifica-
means of communications that tie them together tion. Technically, it is closer to ‘status group’ than
have become steadily more efficient, those now to ‘class,’ and has been very well stated by Joseph
292 • Part IV: The Ruling Class, Elites, and the Upper Class

A. Schumpeter, ‘Social Classes in an Ethically Ho- evaluated character-type is probably José Ortega y
mogeneous Environment,’ Imperialism and Social Gasset’s, The Revolt of the Masses, 1932 (New York:
Classes (New York: Augustus M. Kelley, Inc., 1951), New American Library, Mentor Edition, 1950), esp.
pp. 133 ff., especially pp. 137–47. Cf. also his Capi- pp. 91 ff.
talism, Socialism and Democracy, 3rd ed. (New York: 6. As in the case, quite notably, of Gaetano
Harper, 1950), Part II. For the distinction between Mosca, The Ruling Class (New York: McGraw-Hill,
class and status groups, see From Max Weber: Es- 1939). For a sharp analysis of Mosca, see Fritz
says in Sociology, trans. and ed. by Gerth and Mills Morstein Marx, ‘The Bureaucratic State,’ Review of
(New York: Oxford University Press, 1946). For an Politics, vol. I, 1939, pp. 457 ff. Cf. also Mills, ‘On
analysis of Pareto’s conception of the elite compared Intellectual Craftsmanship,’ April 1952, mimeo-
with Marx’s conception of classes, as well as data on graphed, Columbia College, February 1955.
France, see Raymond Aron, ‘Social Structure and 7. Cf. Karl Löwith, Meaning in History (Chicago:
Ruling Class,’ British Journal of Sociology, vol. I, nos. University of Chicago Press, 1949), pp. 125 ff. for
1 and 2 (1950). concise and penetrating statements of several leading
5. The most popular essay in recent years which philosophies of history.
defines the elite and the mass in terms of a morally

32 • Anthony Giddens

Elites and Power

It is certainly one of the most characteristic em- production, into an essentially political differen-
phases of the Marxian perspective that, in cap- tiation between those ‘who rule’ and those who
italism especially (but also, in a general sense, ‘are ruled’—a transmutation which was, indeed,
in the prior types of class system), the realm of in part made possible by Marx’s failure to specify
the ‘political’ is subordinate to that of the ‘eco- in a systematic fashion the modes whereby the
nomic’. What remains relatively obscure in Marx economic hegemony of the capitalist class be-
is the specific form of this dependence, and how comes ‘translated’ into the political domination
it is expressed concretely in the domination of of the ruling class. For if it is simply the case
the ruling class.1 The importance of this is not that economic control directly yields political
confined to the analysis of the social structure of power, the way is open for the assertion that,
capitalism, but bears directly upon the question in socialism, as in capitalism (indeed as in any
of the classless character of socialism. It relates, other conceivable type of complex society), who-
in addition, to issues brought to the forefront ever controls the means of production thereby
by the critique of the Marxian standpoint ad- achieves political domination as a ruling class.
vanced by the ‘elite theorists’ of the turn of the The movement of history from capitalism to
century. The substance of this critique, in the socialism then becomes conceived of as a mere
writings of such as Pareto and Mosca, may be ex- succession of ‘ruling classes’ (‘elites’), as in classi-
pressed as an attempt to transmute the Marxian cal ‘elite theory’, or more specifically as the emer-
concept of class, as founded in the relations of gence of the sort of ‘managerial’ or ‘technocratic’

Anthony Giddens. “Elites and Power,” in The Class Structure of the Advanced Societies, pp. 118–124. Copyright
© 1973 by Anthony Giddens. Reprinted by permission of HarperCollins and the author.
 Giddens—Elites and Power • 293

ruling class described in Burnham’s writings, and of scale of ‘fame’ or ‘achievement’, whereas the
more recently in some of the variants of the the- second usage may be taken to refer to persons
ory of the ‘technocratic society’.2 who are at the head of a specific social organi-
The points at issue between the Marxian sation which has an internal authority structure
standpoint and ‘elite theory’ have become fur- (the state, an economic enterprise, etc.). I shall
ther complicated in recent years by the use of use the term ‘elite group’ in this latter sense, to
concepts drawn from the latter, such as that of designate those individuals who occupy posi-
‘power elite’, as if they were synonymous with tions of formal authority at the head of a social
that of ‘ruling class’. It will be useful to clarify organisation or institution; and ‘elite’ very gen-
the usage of the terms ‘ruling class’, ‘elite’, ‘power erally, to refer either to an elite group or cluster
elite’, ‘governing class’, etc., which involves, in of elite groups.
part, looking more closely at the structuration In these terms, it can be said that a major
of the upper class. aspect of the structuration of the upper class
In the analysis which follows, I shall be in- concerns, first, the process of mobility into or
terested primarily in developing a set of formu- recruitment to, elite positions and, second, the
lations which illuminate significant conceptual degree of social ‘solidarity’ within and between
distinctions, rather than adhering to conven- elite groups. Mediate structuration thus concerns
tional terminological usage—if it can be said, in how ‘closed’ the process of recruitment to elite
any case, that there is any conventional practice positions is, in favour of those drawn from prop-
in a field in which there has been so much con- ertied backgrounds. Proximate structuration de-
fusion.3 I shall suggest that, given the distinc- pends primarily upon the frequency and nature
tions set out below, there can exist a ‘governing of the social contacts between the members of
class’ without it necessarily being a ‘ruling class’; elite groups. These may take various forms, in-
that there can exist a ‘power elite’ without there cluding the formation of marriage connections
necessarily being either a ‘ruling’ or a ‘governing or the existence of other kin ties, the prevalence
class’; that there can be what I shall call a system of personal ties of acquaintance or friendship,
of ‘leadership groups’ which constitutes neither etc. If the extent of social ‘integration’ of elite
a ‘ruling class’, ‘governing class’, nor ‘power groups is high, there is also likely to be a high de-
elite’; and that all of these social formations gree of moral solidarity characterising the elite as
are, in principle, compatible with the existence a whole and, probably, a low incidence of either
of a society which is ‘capitalist’ in its organisa- latent or manifest conflicts between them. There
tion. To begin with, a few elementary remarks has never been any elite, however solidary, which
are necessary about the notion of ‘elite’. As it is has been free of conflicts and struggles; but the
sometimes employed, ‘elite’ may refer to those degree and intensity of overt conflict has varied
who ‘lead’ in any given category of activity: to widely, and thus it is reasonable to speak broadly
actors and sportsmen as well as to political or of differentials in the solidarity of elite groups.
economic ‘leaders’. There is evidently a differ- Combining these two aspects of structura-
ence, however, between the first and the second, tion, we can establish a typology of elite forma-
in that the former ‘lead’ in terms of some sort tions [see diagram on this page].
294 • Part IV: The Ruling Class, Elites, and the Upper Class

A ‘uniform’ elite is one which shares the at- relationships in society. The first I shall call
tributes of having a restricted pattern of recruit- the institutional mediation of power; the other,
ment and of forming a relatively tightly knit the mediation of power in terms of control. By
unity. It hardly needs emphasising that the clas- the institutional mediation of power, I mean
sifications involved are not of an all-or-nothing the general form of state and economy within
character. The point has been made that even which elite groups are recruited and structured.
among traditional aristocracies there was never This concerns, among other things, the role of
a completely ‘closed’ pattern of recruitment, property in the overall organisation of economic
something which has only been approached by life, the nature of the legal framework defining
the Indian caste system—all elites open their economic and political rights and obligations,
ranks, in some degree, to individuals from the and the institutional structure of the state it-
lower orders, and may enhance their stability self. The mediation of control refers to the ac-
thereby. A relatively closed type of recruitment, tual (effective) power of policy-formation and
however, is likely to supply the sort of coherent decision-making held by the members of partic-
socialisation process producing a high level of ular elite groups: how far, for example, economic
solidarity between (and within) elite groups. leaders are able to influence decisions taken by
But it is quite feasible to envisage the existence politicians, etc. To express it another way, we
of instances which approximate more closely to can say that power has two aspects: a ‘collective’
the case of an ‘established’ elite, where there is a aspect, in the sense that the ‘parameters’ of any
relatively closed pattern of recruitment, but only concrete set of power relationships are contin-
a low level of integration between elite groups. gent upon the overall system of organisation of
A ‘solidary’ elite, as defined in the classification, a society; and a ‘distributive’ aspect, in the sense
might also appear to involve an unlikely combi- that certain groups are able to exert their will at
nation of elements, since it might seem difficult the expense of others.4 The mediation of control
to attain a high degree of integration among elite is thus expressed in terms of ‘effective’ power,
groups whose members are drawn from diverse manifest in terms of the capacity either to take or
class backgrounds. But, while this type of social to influence the taking of decisions which affect
formation is probably rare in capitalist societies, the interests of two or more parties differentially.
at least some of the state socialist countries fit We may conceptually separate two variable
quite neatly into this category: the Commu- factors in analysing effective power (that is to say,
nist Party is the main channel of access to elite power as differentiated from ‘formal authority’)
positions, and while it provides an avenue of in relation to types of elite formation. The first
mobility for individuals drawn in substantial concerns how far such power is ‘consolidated’ in
proportions from quite lowly backgrounds, at the hands of elite groups; the second refers to
the same time it ensures a high degree of sol- the ‘issue-strength’ of the power wielded by those
idarity among elite groups. An ‘abstract’ elite, in elite positions. While the former designates
involving both relatively open recruitment and a limitations upon effective power, deriving from
low level of elite solidarity, whatever its empiri- constraints imposed from ‘below’, the latter con-
cal reality, approximates closely to the picture of cerns how far that power is limited because it can
certain contemporary capitalist societies as these only be exercised in relation to a range of restricted
are portrayed in the writings of the theorists of issues. Thus it is often held to be characteristic of
so-called ‘pluralist democracy’. modern capitalist societies that there are quite
The distinguishing of different types of elite narrowly defined limitations upon the issues
formation does not, in itself, enable us to con- over which elite groups are able to exercise con-
ceptualise the phenomenon of power. As in the trol.5 By combining these two aspects of effective
case of class structuration itself, we may dis- power as exercised by elite groups, we can estab-
tinguish two forms of the mediation of power lish a classification of forms of power-structure
 Giddens—Elites and Power • 295

[see diagram at the top of this page]. Like the class structure [see diagram at the bottom of this
previous typology, this sets out an abstract com- page]. This makes possible a clarification of the
bination of possibilities; it goes almost without four concepts already mentioned—‘ruling class’,
saying that this is no more than an elementary ‘governing class’, ‘power elite’ and ‘leadership
categorisation of a very complex set of phe- groups’. It must be emphasised that these par-
nomena, and the labels applied here in no way tially cross-cut some of the existing usages in the
exhaust the variety of characteristics which are literature on class and elite theory. The Paretian
frequently subsumed under these terms. term ‘governing class’ is here not, as in Pareto’s
According to these definitions, the consolida- own writing, a replacement for the Marxian ‘rul-
tion of effective power is greatest where it is not ing class’; in this scheme, a governing class is ‘one
restricted to clearly defined limits in terms of its step down’, both in terms of elite formation and
‘lateral range’ (broad ‘issue-strength’), and where power-holding, from a ‘ruling class’.
it is concentrated in the hands of the elite, or an In this scheme, the ‘strongest’ case of a rul-
elite group. Power-holding is ‘oligarchic’ rather ing class is defined as that where a uniform elite
than ‘autocratic’ where the degree of centrali- wields ‘autocratic’ power; the weakest is where an
sation of power in the hands of elite groups is established elite holds ‘oligarchic’ power. Where
high, but where the issue-strength of that power a relatively closed recruitment pattern is linked
is limited. In the case of ‘hegemonic’ control, with the prevalence of defined restrictions upon
those in elite positions wield power which, while the effective power of elite groups, a governing
it is not clearly defined in scope and limited to a class exists, but not a ruling class. A governing
restricted range of issues, is ‘shallow’. A ‘demo- class borders upon being a ruling class where a
cratic’ order, in these terms, is one in which the uniform elite possesses ‘hegemonic’ power; and
effective power of elite groups is limited in both comes closest to being a system of leadership
respects. groups where an established elite holds ‘demo-
Finally, bringing together both classifications cratic’ power. Where a governing class involves
formulated above, we can set up an overall ty- a combination of an established elite and ‘hege-
pology of elite formations and power within the monic’ power, it stands close to being a power
296 • Part IV: The Ruling Class, Elites, and the Upper Class

elite. A power elite is distinguished from a rul- It is precisely one of the major tasks of the anal-
ing class in terms of pattern of recruitment, as ysis of elite formations to examine the media of
is a governing class from a system of leadership interchange which operate between elite groups
groups. The latter exists where elite groups only in any given society in order to determine what
hold limited power, and where, in addition, elite kinds of elite hierarchy exist.
recruitment is relatively open in character.
In terms of the mediation of control, this NOTES
classification leaves undefined the relative pri- 1. Most subsequent Marxist authors have either
been content with the most generalised assertions
macy of the power of any one elite group over
about this issue, or have wanted to have their cake
others. This can be conceptually expressed as re-
and eat it by insisting that capitalism is dominated
ferring to the nature of the hierarchy which exists by a ruling class who do not actually ‘rule’; cf. Nicos
among elite groups. A hierarchy exists among Poulantzas, Pouvoir politique et classes sociales de l’état
elite groups in so far as one such group holds capitaliste (Paris 1970), pp. 361ff.
power of broader issue-strength than others, and 2. James Burnham, The Managerial Revolution
is thereby able to exert a degree of control over (New York 1941).
decisions taken by those within them. Thus it 3. In this section of this chapter I have drawn
may be that the economic elite, or certain sectors upon part of my article ‘Elites in the British class
of the economic elite, are able to significantly structure’, Sociological Review 20, 1972.
4. cf. Talcott Parsons, ‘On the concept of political
condition political decisions through the use of
power’, Proceedings of the American Philosophical So-
‘influence’, ‘inducement’, or the ‘direct’ control ciety 107, 1963. The error in Parsons’ analysis, how-
of political positions—i.e., through the fact that ever, is to take insufficient account of the fact that
members of the economic elite are also incum- the ‘collective’ aspect of power is asymmetrical in its
bents of political positions. We may refer to all of consequences for the different groupings in society.
these modes of obtaining, or striving for, control 5. As in Keller’s ‘strategic elites’. See Suzanne
as the media of interchange between elite groups. Keller, Beyond the Ruling Class (New York 1963).
 Domhoff—Who Rules America? • 297

Contemporary Statements

33 • G. William Domhoff

Who Rules America?

Do corporations have far too much power in Partly because the owners and high-level
the United States? Three-fourths of Americans managers within the corporate community
have answered “yes” to that question for the past share great wealth and common economic in-
twenty years. This chapter attempts to resolve terests, but also due to the political opposition
a seeming paradox: How is it possible to have to their shared interests, they band together to
extreme corporate domination in a democratic develop their own social institutions—gated
country? This paradox is made all the more strik- neighborhoods, private schools, exclusive social
ing because in most other democratic countries, clubs, debutante balls, and secluded summer re-
corporations do not have as much power. The sorts. These social institutions create social co-
wealth and income differences between people hesion and a sense of group belonging, a “we”
at the top and the bottom are smaller and the feeling, and thereby mold the owners and top
safety net for those who are poor, ill, or elderly managers in the corporate community into
is stronger. a social upper class. In addition, the corporate
This chapter shows how the owners and owners and managers compensate for the fact
top-level managers of large companies work that they are few in number by financing and
together to maintain themselves as the core of directing a wide variety of nonprofit organiza-
the dominant power group. Their corporations tions, such as tax-free foundations, think tanks,
banks, and agribusinesses form a corporate com- and policy-discussion groups, to aid them in de-
munity that shapes the federal government on veloping policy alternatives that serve their in-
the policy issues of interest to it, issues that have terests. The highest-ranking employees in these
a major impact on the income, job security, and nonprofit organizations, those who serve on the
well-being of most other Americans. At the same organizations’ governing boards, become part of
time, there is competition for profit opportuni- the power elite, the leadership group for the cor-
ties within the corporate community; this can porate community and the upper class.
lead to highly visible policy conflicts among Although the corporate community focuses
rival corporate leaders, conflicts that are some- its complaints on taxes and government spend-
times fought out in Congress. Yet the corpo- ing, the deeper issue is power. In particular, its
rate community is cohesive on the policy issues leaders are wary of the federal government due
that affect its general welfare, which is often at to its capacity to aid average Americans by (1)
stake when organized workers, liberals, or strong creating government jobs for the unemployed,
environmentalists organize sustained political which might make people less likely to take
challenges. low-paying or dangerous positions in the private

G. William Domhoff. Who Rules America? Challenges to Corporate and Class Dominance, pp. xi–xx, 55–63,
83–84, 240–248, 252–253, 257–258. Copyright © 2010 by The McGraw-Hill Companies, Inc.
298 • Part IV: The Ruling Class, Elites, and the Upper Class

sector; (2) making health, unemployment, and and occasions that fit these specifications very
social security benefits more generous, which well. From the viewpoint of social psychology,
also might make people less willing to work the people who make up the upper class can be
in low-paying jobs; (3) helping employees seen as members of numerous small groups that
gain greater workplace rights and protections, meet at private schools, social clubs, retreats, re-
which would make it more difficult to control sorts, and social gatherings.
the workplace; and (4) investing tax dollars in
publicly controlled enterprises, which might
compete with private corporations. The corpo- Prepping for Power
rate community generally opposes all of these From infancy through young adulthood, mem-
possibilities on the grounds that they might in- bers of the upper class receive a distinctive ed-
crease taxes, impede economic growth, or limit ucation. This education begins early in life
freedom. Most of all, however, corporate lead- in preschools that sometimes are attached to
ers oppose any government support for unions a neighborhood church of high social status.
because unions are a potential organizational Schooling continues during the elementary
base for advocating a whole range of polices that years at a local private school called a day school.
threaten corporate power. In a phrase, control of There is a strong chance that at least one or two
labor markets is a crucial issue in the eyes of the years will eventually be spent away from home
corporate community. at a boarding school in a quiet rural setting.
Higher education is obtained at one of a small
number of prestigious private universities. Al-
The Corporate Community and the though some upper-class children may attend
Upper Class public high school if they live in a secluded
The corporate community and the upper class suburban setting, or go to a state university if
are closely intertwined. They are not quite two there is one of great esteem and tradition in their
sides of the same coin, but almost. The evi- home state, the system of formal schooling is so
dence presented in this chapter shows that (1) insulated that many upper-class students never
many super-wealthy stockholding families in see the inside of a public school in all their years
the upper class continue to be involved in the of education. This separate educational system
direction of major corporations through family is important evidence for the distinctiveness of
offices, investment partnerships, and holding the mentality and lifestyle that exists within the
companies; and (2) the professional managers upper class, because schools play a large role in
of middle-level origins are assimilated into the transmitting the class structure to their students
upper class both socially and economically, and (Cookson and Persell 1985).
share the values of upper-class owners. The linchpins in the upper-class educational
Social cohesion develops through the two system are the dozens of boarding schools de-
types of relationships found in a membership veloped in the last half of the nineteenth and
network: common membership in specific so- the early part of the twentieth centuries, coin-
cial institutions, and friendships based on social cident with the rise of a nationwide upper class
interactions within those institutions. Research whose members desired to insulate themselves
on small groups in laboratory settings suggests from an inner city that was becoming populated
that social cohesion is greatest when (1) the so- by lower-class immigrants. They become surro-
cial groups are seen to be exclusive and of high gate families that play a major role in creating
status; and (2) when the interactions take place an upper-class subculture on a national scale in
in relaxed and informal settings (Cartwright and America. The role of boarding schools in pro-
Zander 1968; Hogg 1992). Many of the social viding connections to other upper-class social
institutions of the upper class provide settings institutions is also important. As one informant
 Domhoff—Who Rules America? • 299

explained to a sociologist doing an interview The involvement of private school graduates

study of upper-class women: “Where I went to on boards of directors is demonstrated in a study
boarding school, the girls were from all over the of all alumni over the age of 45 from one of the
country, so I know people from all over. It’s help- most prestigious eastern boarding schools, St.
ful when you move to a new city and want to Paul’s. Using Poor’s Register of Corporations, Di-
get invited into the local social club” (Ostrander rectors and Executives, and Who’s Who in America
1984, p. 85). for 1980, it shows that 303 of these several thou-
It is within these several hundred schools sand men were working as officers or directors
that a unique style of life is inculcated through in corporations in general, and that 102 were
such traditions as the initiatory hazing of be- directors, of 97 corporations, in the Fortune
ginning students, the wearing of school blaz- 800. Their involvement was especially great in
ers or ties, and participation in esoteric sports the financial sector. Most striking of all, twenty
such as lacrosse, squash, and crew. Even a dif- one graduates of St. Paul’s were either officers or
ferent language is adopted to distinguish these directors at J. P. Morgan Bank (now JPMorgan
schools from public schools. The principal is a Chase after its merger in 2000 with another large
headmaster or rector, the teachers are sometimes bank). This finding suggests that the alumni of
called masters, and the students are in forms, particular schools may tend to cluster at specific
not grades. Great emphasis is placed upon the banks or corporations.
building of character. The role of the school in Due to special recruitment programs fi-
preparing the future leaders of America is em- nanced by wealthy individuals and corporations,
phasized through the speeches of the headmaster private schools have become a major educa-
and the frequent mention of successful alumni. tional launching pad for a small percentage of
Thus, boarding schools are in many ways the low-income African-American and Latino stu-
kind of highly effective socializing agent called dents who graduate from elite universities and
total institutions, isolating their members from go to work in the corporate world. The oldest
the outside world and providing them with a set of these programs, A Better Chance, founded in
of routines and traditions that encompass most the 1960s in response to the upheavals of the
of their waking hours. The end result is a feeling Civil Rights Movement, has graduated over
of separateness and superiority that comes from 11,000 students. As of 2004 it had 1,600 stu-
having survived a rigorous education. dents enrolled in 714 boarding schools and 699
Most private school graduates pursue ca- independent day schools. The Prep for Prep pro-
reers in business, finance, or corporate law, gram in New York City and the Steppingstone
which is evidence for the intertwining of the Foundation in Boston and Philadelphia, both of
upper class and the corporate community. Their more recent vintage, have developed programs
business-oriented preoccupations are demon- that identify high-achieving children of color in
strated in the greatest detail in a study of all grade school and help prepare them for private
those who graduated from Hotchkiss between schools with after-school, weekend, and summer
1940 and 1950. Using the school’s alumni files, instruction. Of the 609 Prep for Prep graduates
the researcher followed the careers of 228 grad- in college in 2001, 113 were at Wesleyan, 96 at
uates from their date of graduation until 1970. Harvard, 91 at Yale, 80 at Penn, 79 at Columbia,
Fifty-six percent of the sample became either and 63 at Brown (Zweigenhaft and Domhoff
bankers or business executives, with eighty of 2003, pp. 167–170).
the ninety-one businessmen serving as president,
vice president, or partner in their firms. Another
10 percent of the sample were lawyers, mostly as Social Clubs
partners in large firms closely affiliated with the Private social clubs are a major point of orien-
corporate community (Armstrong 1974). tation in the lives of upper-class adults. These
300 • Part IV: The Ruling Class, Elites, and the Upper Class

clubs also have a role in differentiating members groups. Several dozen men are in three or more
of the upper class from other members of soci- of the clubs, most of them very wealthy people
ety. The clubs of the upper class are many and who also sit on several corporate boards (Bona-
varied, ranging from family-oriented country cich and Domhoff 1981).
clubs and downtown men’s and women’s clubs The overlap of this club network with cor-
to highly specialized clubs for yachtsmen, sports- porate boards of directors provides further evi-
men, gardening enthusiasts, and fox hunters. dence for the intertwining of the upper class and
Downtown men’s clubs originally were places corporate community. In a study in the 1960s,
to have lunch and dinner, and occasionally to the club memberships of the chairpersons and
attend an evening performance or a weekend outside directors of the twenty largest industrial
party. As upper-class families deserted the city corporations were counted. The overlaps with
for large suburban estates, a new kind of club, upper-class clubs in general were ubiquitous,
the country club, gradually took over some of but the concentration of directors in a few clubs
these functions. The downtown club became al- was especially notable. At least one director from
most entirely a luncheon club, a site for holding twelve of the twenty corporations was a member
meetings, or a place to relax on a free afternoon. of the Links Club, which is the New York meet-
The country club, by contrast, became a haven ing ground of the national corporate establish-
for all members of the family. It offered social ment. Seven of General Electric’s directors at the
and sporting activities ranging from dances, par- time were members, as were four from Chrysler,
ties, and banquets to golf, swimming, and ten- four from Westinghouse, and three from IBM.
nis. Special group dinners were often arranged A majority of the top twenty-five corporations in
for all members on Thursday night, the tradi- every major sector of the economy had directors
tional maid’s night off across the United States. in at least one of these clubs, and several had
Initiation fees, annual dues, and expenses many more (Domhoff 1975).
vary from a few thousand dollars in downtown These clubs came under extreme criticism
clubs to $100,000 to $150,000 in some coun- as bastions of Christian white male privilege in
try clubs, but money is not the only barrier in the 1970s, first by wealthy Jewish members of
gaining membership to a club. Each club has a the corporate community who were incensed by
very rigorous screening process before accepting their exclusion, then by civil rights activists who
new members. Most require nomination by one decried the lack of any African-American mem-
or more active members, letters of recommenda- bers, and then by feminist groups, which pointed
tion from three to six members, and interviews out that the exclusion of women deprived women
with at least some members of the membership executives of opportunities to attend business
committee (Kendall 2008). luncheons and develop connections with execu-
Men and women, of the upper class often tives from outside their own workplace (Baltzell
belong to clubs in several cities, creating a na- 1964; Driscoll and Goldberg 1993; Zweigenhaft
tionwide pattern of overlapping memberships. and Domboff 1982). These criticisms made it
These overlaps provide evidence for social cohe- more difficult to obtain membership lists for up-
sion within the upper class. An indication of the dated studies. They also led to a decline in the
nature and extent of this overlapping is revealed listing of club membership in publicly available
in a study of membership lists for twenty clubs sources, such as Who’s Who in America, because
in several major cities across the country in the the information was being used to raise ques-
late 1960s. There is sufficient overlap among tions about the men’s fairness at confirmation
eighteen of the twenty clubs to form three re- hearings for government appointments. But one
gional groupings and a fourth group that pro- detailed study using Who’s Who in America for the
vides a bridge between the two largest regional years 1962, 1973, 1983, and 1995 showed that
 Domhoff—Who Rules America? • 301

corporate executives listed the same few clubs interests of one corporation or business sector, but
over the decades even though they were men- with such matters as the “investment climate,” the
tioned by a declining number of executives in “rate of profit,” and the overall “political climate.”
each decade (Barnes and Sweezes 2006). That is, they have a capitalist mentality.
With the exception of those few who join
the liberal-labor coalition or a leftist movement,
Class Awareness: A Capitalist members of the upper class also have a conserva-
Mentality tive outlook on issues that relate to the well-being
The institutions that establish the owners and of the corporate community as a whole. They
high-level executives of corporations as a na- may be liberal or conservative on social issues,
tional upper class transcend the presence or ab- and they vary somewhat on the degree to which
sence of any given person or family. Families can they are wary of government, but they are nearly
rise and fall in the class structure, but the insti- unanimous in their opposition to labor unions,
tutions of the upper class persist. Not everyone which they see as the major potential challeng-
in this nationwide upper class knows everyone ers to their wealth and power. This tendency to-
else, but everybody knows somebody who knows ward a general class perspective is utilized and
someone in other areas of the country, thanks to reinforced within the policy-planning network.
a common school experience, a summer at the The organizations in that network build upon
same resort, membership in the same social club, the structural economic power and the social co-
or membership on the same board of directors. hesion in reaching consensus on policy matters,
The upper class at any given historical moment where the potential for misunderstanding and
consists of a complex network of overlapping disagreement are great. Human beings are often
social circles knit together by the members they distrustful or egotistical, and there can be differ-
have in common and by the numerous signs of ences in needs between corporations in different
common cultural styles and values that emerge industries and of different sizes.
from a similar upbringing, education, and life- In other words, developing a common policy
style. Viewed from the standpoint of social outlook is not automatic even for the corporate
psychology, the upper class is made up of innu- community and upper class. Their leaders have
merable face-to-face small groups that are con- to create plans that satisfy as many different sec-
stantly changing in their composition as people tors of the corporate community as possible, or
move from one social setting to another. at least minimize the damage to those that can-
Involvement in these institutions usually in- not be fully accommodated. Then they have to
stills a class awareness that includes feelings of develop access to government. At the same time,
superiority, pride, and justified privilege. Deep they have to contend with the possible objections
down, most members of the upper class think of everyday working people who have little or
they are better than other people and therefore nothing except a job, a house, and the opportu-
fully deserving of their station in life. This class nity to obtain educational credentials that might
awareness is based in corporate ownership, but it help them move up the occupational ladder.
is reinforced by the shared social identities and
interpersonal ties created through participation REFERENCES
in the social institutions of the upper class. More Armstrong, Christopher, 1974. Privilege and pro-
ductivity: The cases of two private schools
important, the fact that the upper class is based in
and their graduates. Ph.D. diss., University of
the ownership and control of profit-producing in-
Pennsylvania.
vestments in stocks, bonds, and real estate shows Baltzell, E. Digby, 1964. The Protestant establish-
that it is a capitalist class as well as a social class. ment: Aristocracy and caste in America. New York:
Its members are not concerned simply with the Random House.
302 • Part IV: The Ruling Class, Elites, and the Upper Class

Barnes, Roy C. and Emily Sweezes, 2006. Bohemi- Driscoll, Dawn-Marie and Carol R. Goldberg, 1993.
ans and beyond: Social clubs and the corporate Members of the club: The coming of age of executive
elite. A paper presented to the annual meetings of women. New York: Free Press.
the Southern Sociological Society. New Orleans. Hogg, Michael. 1992. The social psychology of group
Bonacich, Phillip and G. William Domhoff. 1981. cohesiveness. New York: New York University
Latent classes and group membership. Social Net- Press.
works 3:175–196. Kendall, D. 2008. Members Only: Elite Clubs and the
Cartwright, Dorwin and Alvin Frederick Zander. process of Exclusion. Lanham, MD: Rowman and
1968. Group dynamics: Research and theory. New Littlefield.
York: Harper and Row. Ostrander, Susan A. 1984. Women of the upper class.
Cookson, Peter W. and Caroline Hodges Persell. Philadelphia: Temple University Press.
1985. Preparing for power: America’s elite boarding Zweigenhaft, Richard and G. William Domhoff.
schools. New York: Basic Books. 1982. Jews in the Protestant establishment. New
Domhoff, G. W. 1975. Social clubs, policy-planning York: Praeger.
groups, and corporations: A network study of ———. 2003. Blacks in the white elite: Will the
ruling-class cohesiveness. The Insurgent Sociolo- progress continue? Lanham, MD: Rowman and
gist 5:173–184. Littlefield.

34 • Alvin W. Gouldner

The Future of Intellectuals and the

Rise of the New Class

In all countries that have in the twentieth cen- be defined in terms of certain critical episodes.
tury become part of the emerging world so- What follows is only a synoptic inventory of
cioeconomic order, a New Class composed of some episodes decisive in the formation of the
intellectuals and technical intelligentsia—not New Class.
the same—enters into contention with the 1. A process of secularization in which most
groups already in control of the society’s econ- intelligentsia are no longer trained by, living
omy, whether these are businessmen or party within, and subject to close supervision by a
leaders. A new contest of classes and a new class churchly organization, and thus separated from
system is slowly arising in the third world of the everyday life of society.1
developing nations, in the second world of the Secularization is important because it de-
USSR and its client states, and in the first world sacralizes authority-claims and facilitates chal-
of late capitalism of North America, Western Eu- lenges to definitions of social reality made by
rope, and Japan. traditional authorities linked to the church.
The early historical evolution of the New Secularization is important also because it is
Class in Western Europe, its emergence into an infrastructure on which there develops the
the public sphere as a structurally differentiated modern grammar of rationality, or culture of
and (relatively) autonomous social stratum, may critical discourse, with its characteristic stress on

Alvin W. Gouldner. The Future of Intellectuals and the Rise of the New Class (1982), pp. 1–8, 18–20, 28–29,
83–85, 102–104, 106, 110, 112–113. Reprinted by permission of Oxford University Press, USA.
 Gouldner—The Future of Intellectuals and the Rise of the New Class • 303

self-groundedness—in Martin Heidegger’s sense 6. A sixth episode in the formation of the

of the “mathematical project.”2 New Class is the waning of the extended, patri-
2. A second episode in the emergence of the archical family system and its replacement by the
New Class is the rise of diverse vernacular lan- smaller, nuclear family. As middle class women
guages, the corresponding decline of Latin as the become educated and emancipated, they may
language of intellectuals, and especially of their increasingly challenge paternal authority and
scholarly production. Latin becomes a ritual, side with their children in resisting it. With de-
rather than a technical language. This develop- clining paternal authority and growing maternal
ment further dissolves the membrane between influence, the autonomy strivings of children
everyday life and the intellectuals—whether cler- are now more difficult to repress; hostility and
ical or secular. rebellion against paternal authority can become
3. There is a breakdown of the feudal and old more overt. There is, correspondingly, increasing
regime system of personalized patronage relations difficulty experienced by paternal authority in
between the old hegemonic elite and individual imposing and reproducing its social values and
members of the New Class as cultural producers; political ideologies in their children.
and 7. Following the French Revolution, there is
4. A corresponding growth of an anony- in many parts of Europe, especially France and
mous market for the products and services of Germany, a profound reformation and exten-
the New Class, thus allowing them to make an sion of public, non-church controlled, (relatively
independent living apart from close supervision more) multi-class education, at the lower levels as
and personalized controls by patrons. Along with well as at the college, polytechnical, and univer-
secularization, this means that the residence and sity levels. On the one hand, higher education
work of intellectuals are both now less closely in the public school becomes the institutional
supervised by others. basis for the mass production of the New Class
They may now more readily take personal of intelligentsia and intellectuals. On the other
initiatives in the public, political sphere, while hand, the expansion of primary and secondary
also having a “private” life. public school teachers greatly increases the jobs
5. The character and development of the available to the New Class.
emerging New Class also depended importantly As teachers, intellectuals come to be defined,
on the multi-national structure of European pol- and to define themselves, as responsible for and
ities. That Europe was not a single empire with “representative” of society as a whole,3 rather than
a central authority able to impose a single set of as having allegiance to the class interests of their
norms throughout its territory, but a system of students or their parents. As teachers, they are
competing and autonomous states with diverse not defined as having an obligation to reproduce
cultures and religions, meant that dissenting in- parental values in their children. Public teachers
tellectuals, scientists, and divines could and did supersede private tutors.
protect their own intellectual innovations by mi- 8. The new structurally differentiated edu-
grating from their home country when condi- cational system is increasingly insulated from
tions there grew insupportable and sojourning in the family system, becoming an important
foreign lands. Even the enforced travel of exiled source of values among students divergent from
intellectuals also enabled them to enter into a those of their families. The socialization of the
European-wide communication network. In an young by their families is now mediated by a
article (as yet unpublished), Robert Wuthnow semi-autonomous group of teachers.
has suggested that their often extensive travel led 9. While growing public education limits
many intellectuals to share a cosmopolitan iden- family influence on education, it also increases
tity transcending national limits and enhancing the influence of the state on education. The
their autonomy from local elites. public educational system thus becomes a major
304 • Part IV: The Ruling Class, Elites, and the Upper Class

cosmopolitanizing influence on its students, with itself becomes a technology to be pursued with
a corresponding distancing from localistic inter- “instrumental rationality.” The revolutionary or-
ests and values. ganization evolves from a ritualistic, oath-bound
10. Again, the new school system becomes a secret society into the modern “vanguard” party.
major setting for the intensive linguistic conver- When the Communist Manifesto remarks that
sion of students from casual to reflexive speech, Communists have nothing to hide,5 it is exactly
or (in Basil Bernstein’s terms) from “restricted” a proposed emergence into public life which is
linguistic codes to “elaborated” linguistic codes,4 implied. The Communist Manifesto was written
to a culture of discourse in which claims and as- by Marx and Engels for the “League of Commu-
sertions may not be justified by reference to the nists,” which was born of the “League of the Just”
speaker’s social status. This has the profound which, in turn, was descended from the “League
consequence of making all authority-referring of Outlaws.” This latter group of German emi-
claims potentially problematic. grants in Paris had a pyramidal structure, made a
11. This new culture of discourse often di- sharp distinction between upper and lower mem-
verges from assumptions fundamental to ev- bers, blindfolded members during initiation cer-
eryday life, tending to put them into question emonies, used recognition signs and passwords,
even when they are linked to the upper classes. and bound members by an oath.6 The vanguard
These school-inculcated modes of speech are, organization, however, de-ritualizes participation
also, (relatively) situation-free language variants. and entails elements of both the “secret society”
Their situation-freeness is further heightened and of the public political party. In the vanguard
by the “communications revolution” in general, organization, public refers to the public avail-
and by the development of printing technology, ability of the doctrine rather than the availability
in particular. With the spread of printed ma- of the organization or its membership to public
terials, definitions of social reality available to scrutiny. Here, to be “public” entails the orga-
intellectuals may now derive increasingly from nization’s rejection of “secret doctrines” known
distant persons, from groups geographically, cul- only to an elite in the organization—as, for in-
turally, and historically distant and even from stance, Bakunin’s doctrine of an elite dictatorship
dead persons, and may therefore diverge greatly of anarchists.7 The modern vanguard structure
from any local environment in which they are is first clearly encoded in Lenin’s What Is to Be
received. Definitions of social reality made by Done? Here it is plainly held that the proletariat
local elites may now be invidiously contrasted cannot develop a socialist consciousness by itself,
(by intellectuals) with definitions made in other but must secure this from a scientific theory de-
places and times. veloped by the intelligentsia.8 The “vanguard”
12. With the spread of public schools, literacy party expresses the modernizing and elite ambi-
spreads; humanistic intellectuals lose their exclu- tions of the New Class as well as an effort to over-
siveness and privileged market position, and now come its political limitations. Lenin’s call for the
experience a status disparity between their “high” development of “professional” revolutionaries, as
culture, as they see it, and their lower deference, the core of the vanguard, is a rhetoric carrying
repute, income and social power. The social posi- the tacit promise of a career-like life which invites
tion of humanistic intellectuals, particularly in a young members of the New Class to “normalize”
technocratic and industrial society, becomes more the revolutionary existence. . . .
marginal and alienated than that of the technical
intelligentsia. The New Class becomes internally There are several distinguishable conceptions
differentiated. of the New Class:
13. Finally, a major episode in the emergence 1. New Class as Benign Technocrats: Here the
of the modern intelligentsia is the changing form New Class is viewed as a new historical elite
of the revolutionary organization. Revolution already entrenched in institutional influence
 Gouldner—The Future of Intellectuals and the Rise of the New Class • 305

which it uses in benign ways for society; it is radicalized elements of the New Class played a
more or less inevitable and trustworthy: e.g., major leadership role in the key revolutions of
Galbraith,9 Bell,10 Berle and Means.11 our time. It greatly overemphasizes the com-
(Sed contra: This obscures the manner in mon interests binding the New and old class,
which the New Class egoistically pursues its own systematically missing the tensions between
special vested interests. Moreover, the power of them; it ignores the fact that elimination of the
the New Class today is scarcely entrenched. This old class is an historical option open to the New
view also ignores the limits on the rationality of Class. This static conception underestimates the
the New Class.) growth in the numbers and influence of the New
2. New Class as Master Class: Here the New Class. The view is also unexpectedly Marcusean
Class is seen as another moment in a long- in overstressing the prospects of old class con-
continuing circulation of historical elites, as a tinuity; it really sees the old class as having no
socialist intelligentsia that brings little new to effective opponents, either in the New Class or
the world and continues to exploit the rest of so- in the old adversary class, the proletariat. It thus
ciety as the old class had, but now uses education ends as seeing even less social change in prospect
rather than money to exploit others: Bakunin,12 than the Parsonian view [#3 above].)
Machajski.13 5. New Class as Flawed Universal Class
(Sed contra: The New Class is more histori- (my own view): The New Class is elitist and
cally unique and discontinuous than this seems; self-seeking and uses its special knowledge to ad-
while protecting its own special interests, it is not vance its own interests and power, and to control
bound by the same limits as the old class and, at its own work situation. Yet the New Class may
least transiently, contributes to collective needs.) also be the best card that history has presently
3. New Class as Old Class Ally: The New Class given us to play. The power of the New Class is
is here seen as a benign group of dedicated “pro- growing. It is substantially more powerful and
fessionals” who will uplift the old (moneyed) independent than Chomsky suggests, while still
class from a venal group to a collectivity-oriented much less powerful than is suggested by Gal-
elite and who, fusing with it, will forge a new, braith who seems to conflate present reality with
genteel elite continuous with but better than the future possibility. The power of this morally am-
past: Talcott Parsons.14 biguous New Class is on the ascendent and it
(Sed contra: Neither group is an especially holds a mortgage on at least one historical future.
morally bound agent; the old class is constrained In my own left Hegelian sociology, the New
to protect its profits, the New Class is cashing Class bearers of knowledge are seen as an em-
in on its education. Immersed in the present, bryonic new “universal class”—as the prefigured
this view misses the fact that each is ready to embodiment of such future as the working class
exploit the other, if need be, and shows little un- still has. It is that part of the working class which
derstanding of the profound [if different] limits will survive cybernation. At the same time, a left
imposed on the rationality and morality of each Hegelian sociology also insists that the New
of these groups, and of the important tensions Class is profoundly flawed as a universal class.
between them.) Moreover, the New Class is not some unified
4. New Class as Servants of Power: Here the subject or a seamless whole; it, too, has its own
New Class is viewed as subservient to the old internal contradictions. It is a class internally
(moneyed) class which is held to retain power divided with tensions between (technical) in-
much as it always did, and is simply using the telligentsia and (humanistic) intellectuals. No
New Class to maintain its domination of society: celebration, mine is a critique of the New Class
Noam Chomsky15 and Maurice Zeitlin.16 which does not view its growing power as inev-
(Sed contra: This ignores the revolution- itable, which sees it as morally ambivalent, em-
ary history of the twentieth century in which bodying the collective interest but partially and
306 • Part IV: The Ruling Class, Elites, and the Upper Class

transiently, while simultaneously cultivating its “business” or political interests. This autonomy
own guild advantage. . . . is said to be grounded in the specialized knowl-
edge or cultural capital transmitted by the edu-
cational system, along with an emphasis on the
The New Class as a Cultural obligation of educated persons to attend to the
Bourgeoisie welfare of the collectivity. In other words, the
1. The New Class and the old class are at first ideology of “professionalism” emerges.
undifferentiated; the New Class commonly 5. Professionalism is one of the public ideolo-
originates in classes with property advantages, gies of the New Class, and is the genteel subver-
that is, in the old class, or is sponsored by them. sion of the old class by the new. Professionalism
The New Class of intellectuals and intelligentsia is a phase in the historical development of the
are the relatively more educated counterpart— “collective consciousness” of the New Class.
often the brothers, sisters, or children—of the While not overtly a critique of the old class, pro-
old moneyed class. Thus the New Class contest fessionalism is a tacit claim by the New Class to
sometimes has the character of a civil war within technical and moral superiority over the old class,
the upper classes. It is the differentiation of the old implying that the latter lack technical creden-
class into contentious factions. To understand tials and are guided by motives of commercial
the New Class contest it is vital to understand venality. Professionalism silently installs the New
how the privileged and advantaged, not simply Class as the paradigm of virtuous and legitimate
the suffering, come to be alienated from the very authority, performing with technical skill and
system that privileges them. with dedicated concern for the society-at-large.
2. The “non-negotiable” objectives of the old Professionalism makes a focal claim for the legit-
moneyed class are to reproduce their capital, at imacy of the New Class which tacitly deautho-
a minimum, but, preferably, to make it accumu- rizes the old class.
late and to appropriate profit: M-C-M, as Marx On the one side, this is a bid for prestige
said. This is done within a structure in which within the established society; on the other, it
all of them must compete with one another. tacitly presents the New Class as an alternative to
This unrelenting competition exerts pressure to the old. In asserting its own claims to authority,
rationalize their productive and administrative professionalism in effect devalues the authority of
efforts and unceasingly to heighten efficiency. the old class.
(Marx called it, “revolutionizing” production.) 6. The special privileges and powers of the
But this rationalization is dependent increasingly New Class are grounded in their individual con-
on the efforts of the New Class intelligentsia and trol of special cultures, languages, techniques,
its expert skills. It is inherent in its structural sit- and of the skills resulting from these. The New
uation, then, that the old class must bring the Class is a cultural bourgeoisie who appropriates
New Class into existence. privately the advantages of a historically and
3. Much of the New Class is at first trained collectively produced cultural capital. Let us be
under the direct control of the old class’s firms or clear, then: the New Class is not just like the old
enterprises. Soon, however, the old class is sepa- class; its special culture is not just like capital. No
rated from the reproduction of the New Class by metaphor is intended. The special culture of the
the emergence and development of a public sys- New Class is a stock of capital that generates a
tem of education whose costs are “socialized.”17 stream of income (some of ) which it appropri-
4. The more that the New Class’s reproduc- ates privately.
tion derives from specialized systems of public 7. The fundamental objectives of the New
education, the more the New Class develops Class are: to increase its own share of the na-
an ideology that stresses its autonomy, its sepa- tional product; to produce and reproduce the
ration from and presumable independence of special social conditions enabling them to
 Gouldner—The Future of Intellectuals and the Rise of the New Class • 307

appropriate privately larger shares of the incomes and intelligentsia are commonly committed to a
produced by the special cultures they possess; to culture of critical discourse (CCD). CCD is the
control their work and their work settings; and latent but mobilizable infrastructure of modern
to increase their political power partly in order to “technical” languages.
achieve the foregoing. The struggle of the New 2. The culture of critical discourse is char-
Class is, therefore, to institutionalize a wage sys- acterized by speech that is relatively more
tem, i.e., a social system with a distinct principle situation-free, more context or field “inde-
of distributive justice: “from each according to pendent.” This speech culture thus values ex-
his ability, to each according to his work,” which pressly legislated meanings and devalues tacit,
is also the norm of “socialism.” Correspondingly, context-limited meanings. Its ideal is: “one
the New Class may oppose other social systems word, one meaning,” for everyone and forever.
and their different systems of privilege, for exam- The New Class’s special speech variant also
ple, systems that allocate privileges and incomes stresses the importance of particular modes of
on the basis of controlling stocks of money (i.e., justification, using especially explicit and articu-
old capital). The New Class, then, is prepared late rules, rather than diffuse precedents or tacit
to be egalitarian so far as the privileges of the features of the speech context. The culture of
old class are concerned. That is, under certain critical speech requires that the validity of claims
conditions it is prepared to remove or restrict the be justified without reference to the speaker’s
special incomes of the old class: profits, rents, societal position or authority. Here, good speech
interest. The New Class is anti-egalitarian, how- is speech that can make its own principles ex-
ever, in that it seeks special guild advantages— plicit and is oriented to conforming with them,
political powers and incomes—on the basis of its rather than stressing context-sensitivity and
possession of cultural capital. . . . context-variability. Good speech here thus has
theoreticity.19
Being pattern- and principle-oriented, CCD
The New Class as a Speech implies that that which is said may not be cor-
Community rect, and may be wrong. It recognizes that “What
1. The culture of critical discourse (CCD)18 is Is” may be mistaken or inadequate and is there-
an historically evolved set of rules, a grammar fore open to alternatives. CCD is also relatively
of discourse, which (1) is concerned to justify more reflexive, self-monitoring, capable of more
its assertions, but (2) whose mode of justifica- metacommunication, that is, of talk about talk;
tion does not proceed by invoking authorities, it is able to make its own speech problematic,
and (3) prefers to elicit the voluntary consent of and to edit it with respect to its lexical and gram-
those addressed solely on the basis of arguments matical features, as well as making problematic
adduced. CCD is centered on a specific speech the validity of its assertions. CCD thus requires
act: justification. It is a culture of discourse in considerable “expressive discipline,” not to speak
which there is nothing that speakers will on of “instinctual renunciation.”
principle permanently refuse to discuss or make 3. Most importantly, the culture of critical
problematic; indeed, they are even willing to talk speech forbids reliance upon the speaker’s per-
about the value of talk itself and its possible in- son, authority, or status in society to justify his
feriority to silence or to practice. This grammar claims. As a result, CCD de-authorizes all speech
is the deep structure of the common ideology grounded in traditional societal authority, while
shared by the New Class. The shared ideology of it authorizes itself, the elaborated speech variant
the intellectuals and intelligentsia is thus an ideol- of the culture of critical discourse, as the stan-
ogy about discourse. Apart from and underlying dard of all “serious” speech. From now on, per-
the various technical languages (or sociolects) sons and their social positions must not be visible
spoken by specialized professions, intellectuals in their speech. Speech becomes impersonal.
308 • Part IV: The Ruling Class, Elites, and the Upper Class

Speakers hide behind their speech. Speech nationalities and it is often a member of a tech-
seems to be dis-embodied, de-contextualized nical guild of international scope.
and self-grounded. (This is especially so for the 2. For all that, however, the New Class is
speech of intellectuals and somewhat less so for hardly the end of domination. While its ultimate
technical intelligentsia who may not invoke significance is the end of the old moneyed class’s
CCD except when their paradigms break down.) domination, the New Class is also the nucleus
The New Class becomes the guild masters of an of a new hierarchy and the elite of a new form of
invisible pedagogy. cultural capital.
4. The culture of critical discourse is the The historical limits of the New Class are
common ideology shared by the New Class, al- inherent in both the nature of its own charac-
though technical intelligentsia sometimes keep teristic rationality, and in its ambitions as a cul-
it in latency. The skills and the social conditions tural bourgeoisie. Its culture of critical discourse
required to reproduce it are among the common fosters a purely “theoretical” attitude toward the
interests of the New Class. Correspondingly, it world. Speakers are held competent to the degree
is in the common interest of the New Class to that they know and can say the rules, rather than
prevent or oppose all censorship of its speech just happening to follow them. The culture of
variety and to install it as the standard of good critical discourse thus values the very theoretic-
speech. The New Class thus has both a common ity that the “common sense” long suspected was
ideology in CCD and common interests in its cul- characteristic of intellectuals.
tural capital. . . . Intellectuals have long believed that those
who know the rule, who know the theory by
which they act, are superior because they lead
The Flawed Universal Class an “examined” life. They thus exalt theory over
1. The New Class is the most progressive force practice, and are concerned less with the success
in modern society and is a center of whatever of a practice than that the practice should have
human emancipation is possible in the fore- submitted itself to a reasonable rule. Since in-
seeable future. It has no motives to curtail the tellectuals and intelligentsia are concerned with
forces of production and no wish to develop doing things in the right way and for the right
them solely in terms of their profitability. The reason—in other words, since they value doc-
New Class possesses the scientific knowledge and trinal conformity for its own sake—they (we)
technical skills on which the future of modern have a native tendency toward ritualism and
forces of production depend. At the same time, sectarianism.
members of the New Class also manifest increas- 3. The culture of the New Class exacts still
ing sensitivity to the ecological “side effects” or other costs: since its discourse emphasizes the
distant diseconomies of continuing technical de- importance of carefully edited speech, this has
velopment. The New Class, further, is a center the vices of its virtues: in its virtuous aspect,
of opposition to almost all forms of censorship, self-editing implies a commendable circum-
thus embodying a universal societal interest in spection, carefulness, self-discipline and “se-
a kind of rationality broader than that invested riousness.” In its negative modality, however,
in technology. Although the New Class is at the self-editing also disposes toward an unhealthy
center of nationalist movements throughout the self-consciousness, toward stilted convoluted
world, after that phase is secured, the New Class speech, an inhibition of play, imagination and
is also the most internationalist and most uni- passion, and continual pressure for expressive
versalist of all social strata; it is the most cos- discipline. The new rationality thus becomes the
mopolitan of all elites. Its control over ordinary source of a new alienation.
“foreign” languages, as well as of technical so- Calling for watchfulness and self-discipline,
ciolects, enables it to communicate with other CCD is productive of intellectual reflexivity and
 Gouldner—The Future of Intellectuals and the Rise of the New Class • 309

the loss of warmth and spontaneity. Moreover, longer enough simply to be good. Now, one has
that very reflexivity stresses the importance of to explain it. The New Class is the universal class
adjusting action to some pattern of propriety. in embryo, but badly flawed.
There is, therefore, a structured inflexibility
when facing changing situations; there is a cer- NOTES
tain disregard of the differences in situations, 1. It is not my intention to suggest that modern
intellectuals are merely the secular counterpart of
and an insistence on hewing to the required rule.
clericals. Indeed, my own stress (as distinct, say, from
This inflexibility and insensitivity to the force
Edward Shils who does appear to view intellectuals
of differing contexts, this inclination to impose as priests manqués) is on the discontinuity of the two.
one set of rules on different cases also goes by the 2. For full development of this, see chapter 2,
ancient name of “dogmatism.” Set in the con- especially p. 42, of my Dialectic of Ideology and Tech-
text of human relationships, the vulnerability of nology (New York, 1976).
the New Class to dogmatism along with its very 3. Doubtless some will insist this is a “false con-
task-centeredness, imply a certain insensitivity to sciousness.” But this misses the point. My concern
persons, to their feelings and reactions, and open here is with their own definitions of their social
the way to the disruption of human solidarity. role, precisely because these influence the manner in
which they perform their roles. As W. I. Thomas and
Political brutality, then, finds a grounding in the
Florian Znaniecki long ago (and correctly) insisted,
culture of critical discourse; the new rationality a thing defined as real is real in its consequences.
may paradoxically allow a new darkness at noon. Moreover, the state who employs most of these
4. The paradox of the New Class is that it teachers is itself interested in having teachers consol-
is both emancipatory and elitist. It subverts all idate the tie between students and itself, rather than
establishments, social limits, and privileges, with the students’ parents.
including its own. The New Class bears a cul- 4. See Basil Bernstein, Class, Codes and Control,
ture of critical and careful discourse which is an vol. 1, Theoretical Studies Towards a Sociology of
historically emancipatory rationality. The new Language (London, 1971), vol. 2, Applied Studies
Towards a Sociology of Language (London, 1973),
discourse (CCD) is the grounding for a critique
vol. 3, Towards a Theory of Educational Transmission
of established forms of domination and pro-
(London, 1975). Bernstein’s theory is used here in
vides an escape from tradition, but it also bears a critical appropriation facilitated by the work of
the seeds of a new domination. Its discourse is Dell Hymes and William Labov. My own critique of
a lumbering machinery of argumentation that Bernstein emerges, at least tacitly, in the discussion
can wither imagination, discourage play, and of [the “Flawed Universal Class”] in the text. It is
curb expressivity. The culture of discourse of developed explicitly in my Dialectic of Ideology and
the New Class seeks to control everything, its Technology. While Labov has sharply criticized Bern-
topic and itself, believing that such domination stein, he himself also stresses the general importance
of self-monitored speech and of speech reflexivity
is the only road to truth. The New Class begins
in general (i.e., not only of careful pronunciation)
by monopolizing truth and by making itself its
thus converging with Bernstein’s focus on reflexiv-
guardian. It thereby makes even the claims of ity as characterizing the elaborated linguistic variant
the old class dependent on it. The New Class and distinguishing it from the restricted variant. See
sets itself above others, holding that its speech is William Labov, Sociolinguistic Patterns (Philadelphia,
better than theirs; that the examined life (their 1972), p. 208.
examination) is better than the unexamined 5. For example: “The Communists disdain
life which, it says, is sleep and no better than to conceal their views and aims. They openly de-
death. Even as it subverts old inequities, the New clare . . . ” (Communist Manifesto [Chicago, 1888],
authorized English edition edited by Engels, p. 58).
Class silently inaugurates a new hierarchy of the
6. See E. Hobsbawm, Primitive Rebels (Manches-
knowing, the knowledgeable, the reflexive and
ter, 1959), p. 167 ff.
insightful. Those who talk well, it is held, excel 7. A secret doctrine is one which, because it is
those who talk poorly or not at all. It is now no reserved only for the organization elite, can be made
310 • Part IV: The Ruling Class, Elites, and the Upper Class

known only after persons join organizations and 16. Maurice Zeitlin, “Corporate Ownership and
reach a certain membership position in it. A secret Control: The Large Corporations and the Capitalist
doctrine thus is never one which can have been a mo- Class,” American Journal of Sociology (March 1974),
tive for joining the organization in the first instance. pp. 1073–1119.
8. Lenin’s What Is to Be Done? was originally pub- 17. Cf. James O’Connor, Corporations and the
lished in 1902. State (New York, 1974), pp. 126–28 for the argu-
9. The New Industrial State (Boston, 1967). ment that government financing of R & D and ad-
10. The Coming of Post-Industrial Society (New vanced education constitute a socialization of part of
York, 1973). the costs of production whose net surplus is privately
11. The Modern Corporation and Private Property appropriated.
(New York, 1932). 18. This section is indebted to Basil Bernstein
12. “It stands to reason that the one who knows and is based on a critical appropriation of his “elab-
more will dominate the one who knows less,” M. orated and restricted linguistic codes,” which have
Bakounine, Oeuvres, Vol. 5 (Paris, 1911), p. 106. gone through various re-workings. That controver-
13. See V. F. Calverton, The Making of Society sial classic was published in J. J. Gumperz and D.
(New York, 1937). Hymes, Directions in Sociolinguistics (New York,
14. Talcott Parsons, The Social System (Glen- 1972). A recent re-working is to be found in Bern-
coe, 1951), chapter 10; Essays in Sociological Theory stein’s “Social Class, Language, and Socialization,” in
(Glencoe, 1954), chapter 18; “The Professions,” In- T. A. Sebeok, ed., Current Trends in Linguistics (The
ternational Encyclopedia of Social Sciences (New York, Hague, 1974). For full bibliographic and other de-
1968). tails see note 4 above.
15. While Chomsky’s position is exhibited in var- 19. Cf. Peter McHugh, “A Common-Sense Per-
ious of his writings, I shall rely here on his most re- ception of Deviance,” in H. P. Dreitzel, ed., Recent
cent statement in his Huizinga lecture, “Intellectuals Sociology, Number 2 (London, 1970), pp. 165 ff. For
and the State,” delivered at Leiden, 9 October 1977. good speech as “serious” speech see David Silverman,
Citations will be from the manuscript copy. Cf. N. “Speaking Seriously,” Theory and Society (Spring,
Chomsky, American Power and the New Mandarins 1974).
(New York, 1969).

35 • David Brooks

Bobos in Paradise
The New Upper Class and How They Got There

The Rise of the Educated Class Imagine how happy Stanley J. Kogan must have
I’m not sure I’d like to be one of the people fea- been, for example, when his daughter Jamie was
tured on the New York Times wedding page, but admitted to Yale. Then imagine his pride when
I know I’d like to be the father of one of them. Jamie made Phi Beta Kappa and graduated

David Brooks. Bobos in Paradise: The New Upper Class and How They Got There, pp. 13–14, 18–23, 25–26,
28–33, 35–37, 40–46, 51–53. Copyright © 2000 by David Brooks. Reprinted with the permission of Simon
& Schuster, Inc. All rights reserved.
 Brooks—Bobos in Paradise • 311

summa cum laude. Stanley himself is no slouch

in the brains department: he’s a pediatric urol- The Fifties
ogist in Croton-on-Hudson, with teaching po- The Times weddings page didn’t always pulse
sitions at the Cornell Medical Center and the with the accomplishments of the Résumé Gods.
New York Medical College. Still, he must have In the late 1950s, the page projected a calm
enjoyed a gloat or two when his daughter put on and more stately ethos. The wedding accounts
that cap and gown. of that era didn’t emphasize jobs or advanced
And things only got better. Jamie breezed degrees. The profession of the groom was only
through Stanford Law School. And then she sometimes mentioned, while the profession of
met a man—Thomas Arena—who appeared to the bride was almost never listed (and on the
be exactly the sort of son-in-law that pediatric rare occasions when the bride’s profession was
urologists dream about. He did his undergradu- noted, it was in the past tense, as if the mar-
ate work at Princeton, where he, too, made Phi riage would obviously end her career). Instead,
Beta Kappa and graduated summa cum laude. the Times listed pedigree and connections. An-
And he, too, went to law school, at Yale. After cestors were frequently mentioned. The ushers
school they both went to work as assistant U.S. were listed, as were the bridesmaids. Prep schools
attorneys for the mighty Southern District of were invariably mentioned, along with colleges.
New York. The Times was also careful to list the groom’s
These two awesome résumés collided at a clubs—the Union League, the Cosmopolitan
wedding ceremony in Manhattan, and given Club. It also ran down the bride’s debutante his-
all the school chums who must have attended, tory, where she came out, and whatever women’s
the combined tuition bills in that room must clubs she might be a member of, such as the Ju-
have been staggering. The rest of us got to read nior League. In short, the page was a galaxy of
about it on the New York Times weddings page. restricted organizations. . . .
The page is a weekly obsession for hundreds of The section from the late fifties evokes an en-
thousands of Times readers and aspiring Balzacs. tire milieu that was then so powerful and is now
Unabashedly elitist, secretive, and totally honest, so dated: the network of men’s clubs, country
the “mergers and acquisitions page” (as some of clubs, white-shoe law firms, oak-paneled Wall
its devotees call it) has always provided an ac- Street firms, and WASP patriarchs. Everybody
curate look at at least a chunk of the American has his or her own mental images of the old
ruling class. And over the years it has reflected Protestant Establishment: lockjaw accents, the
the changing ingredients of elite status. Social Register, fraternity jocks passing through
When America had a pedigreed elite, the page Ivy League schools, constant rounds of martinis
emphasized noble birth and breeding. But in and highballs, bankers’ hours, starched old men
America today it’s genius and geniality that en- like Averell Harriman, Dean Acheson, and John
able you to join the elect. And when you look at J. McCloy, the local bigwigs that appear in John
the Times weddings page, you can almost feel the Cheever and John O’Hara stories. . . .
force of the mingling SAT scores. It’s Dartmouth It really was possible to talk about an aristo-
marries Berkeley, MBA weds Ph.D., Fulbright cratic ruling class in the fifties and early sixties,
hitches with Rhodes, Lazard Frères joins with a national elite populated by men who had gone
CBS, and summa cum laude embraces summa to northeastern prep schools like Groton, An-
cum laude (you rarely see a summa settling for dover, Exeter, and St. Paul’s and then ascended
a magna—the tension in such a marriage would through old-line firms on Wall Street into the
be too great). The Times emphasizes four things boardrooms of the Fortune 500 corporations
about a person—college degrees, graduate degrees, and into the halls of Washington power. The
career path, and parents’ profession—for these are WASPs didn’t have total control of the coun-
the markers of upscale Americans today. . . . try or anything like it, but they did have the
312 • Part IV: The Ruling Class, Elites, and the Upper Class

hypnotic magic of prestige. As Richard Rovere people who happened to be born into it, the task
wrote in a famous 1962 essay entitled “The was to accept the duties that came along with its
American Establishment,” “It has very nearly un- privileges. . . .
challenged power in deciding what is and what
is not respectable opinion in this country.” . . .
This was the last great age of socially accept- The Hinge Years
able boozing. It was still an era when fox hunt- Then came the change. By 1960 the average ver-
ing and polo didn’t seem antiquarian. But the bal SAT score for incoming freshmen at Harvard
two characteristics of that world that strike us was 678, and the math score was 695—these
forcefully today are its unabashed elitism and its are stratospheric scores. The average Harvard
segregation. Though this elite was nowhere near freshman in 1952 would have placed in the bot-
as restrictive as earlier elites—World War II had tom 10 percent of the Harvard freshman class
exerted its leveling influence—the 1950s estab- of 1960. Moreover, the 1960 class was drawn
lishment was still based on casual anti-Semitism, from a much wider socioeconomic pool. Smart
racism, sexism, and a thousand other silent bar- kids from Queens or Iowa or California, who
riers that blocked entry for those without the wouldn’t have thought of applying to Harvard
correct pedigree. Wealthy Jewish and Protes- a decade earlier, were applying and getting ac-
tant boys who had been playing together from cepted. Harvard had transformed itself from a
childhood were forced to endure “The Great school catering mostly to the northeastern social
Division” at age 17, when Jewish and Gentile elite to a high-powered school reaching more of
society parted into two entirely separate orbits, the brightest kids around the country. And this
with separate debutante seasons, dance schools, transformation was replicated in almost all elite
and social secretaries. A Protestant business ex- schools. . . .
ecutive may have spent his professional hours History, as Pareto once remarked, is the
working intimately with his Jewish colleague, graveyard of aristocracies, and by the late fifties
but he never would have dreamed of putting and early sixties the WASP Establishment had
him up for membership in his club. When Sen- no faith in the code—and the social restric-
ator Barry Goldwater attempted to play golf at tions—that had sustained it. Maybe its mem-
the restricted Chevy Chase Club, he was told the bers just lost the will to fight for their privileges.
club was restricted. “I’m only half Jewish, so can’t As the writer David Frum theorizes, it had been
I play nine holes?” he is said to have replied. half a century since the last great age of fortune
The WASP elite was also genially making. The great families were into at least
anti-intellectual. Its members often spoke of their third genteel generation. Perhaps by then
“eggheads” and “highbrows” with polite disdain. there wasn’t much vigor left. Or perhaps it was
Instead, their status, as F. Scott Fitzgerald had the Holocaust that altered the landscape by dis-
pointed out a few decades before, derived from crediting the sort of racial restrictions that the
“animal magnetism and money.” By contrast Protestant Establishment was built on.
with today’s ruling class, they had relatively un- In any case, in 1964 Digby Baltzell astutely
complicated attitudes about their wealth. They perceived the crucial trends. “What seems to be
knew it was vulgar to be gaudy, they tended happening,” he wrote in The Protestant Establish-
toward thriftiness, but they seem not to have ment, “is that a scholarly hierarchy of campus
seen their own money as an affront to Ameri- communities governed by the values of admis-
can principles of equality. On the contrary, most sions committees is gradually supplanting the
took their elite status for granted, assuming that class hierarchies of local communities which are
such position was simply part of the natural and still governed by the values of parents. . . . Just as
beneficent order of the universe. There was al- the hierarchy of the Church was the main avenue
ways going to be an aristocracy, and so for the of advancement for the talented and ambitious
 Brooks—Bobos in Paradise • 313

youth from the lower orders during the medieval moral code, and to replace the old order with
period, and just as the business enterprise was a new social code that would celebrate spiritual
responsible for the nineteenth century rags-to- and intellectual ideals. The sixties radicals re-
riches dream (when we were predominantly an jected the prevailing definition of accomplish-
Anglo-Saxon country), so the campus commu- ment, the desire to keep up with the Joneses, the
nity has now become the principal guardian of prevailing idea of social respectability, the idea
our traditional opportunitarian ideals.” that a successful life could be measured by in-
The campus gates were thus thrown open come, manners, and possessions. . . .
on the basis of brains rather than blood, and
within a few short years the university landscape
was transformed. Harvard, as we’ve seen, was And Then Comes Money
changed from a school for the well-connected The hardest of the hard-core sixties radicals be-
to a school for brainy strivers. The remaining lieved the only honest way out was to reject the
top schools eliminated their Jewish quotas and notion of success altogether: drop out of the rat
eventually dropped their restrictions on women. race, retreat to small communities where real
Furthermore, the sheer numbers of educated human relationships would flourish. But that
Americans exploded. The portion of Amer- sort of utopianism was never going to be very
icans going to college had been rising steadily popular, especially among college grads. Mem-
throughout the 20th century, but between 1955 bers of the educated class prize human relation-
and 1974 the growth rate was off the charts. ships and social equality, but as for so many
Many of the new students were women. Between generations of Americans before them, achieve-
1950 and 1960 the number of female students ment was really at the core of the sixties grads’
increased by 47 percent. It then jumped by an value system. They were meritocrats, after all,
additional 168 percent between 1960 and 1970. and so tended to define themselves by their ac-
Over the following decades the student popula- complishments. Most of them were never going
tion kept growing and growing. In 1960 there to drop out or sit around in communes smelling
were about 2,000 institutions of higher learn- flowers, raising pigs, and contemplating poetry.
ing. By 1980 there were 3,200. In 1960 there Moreover, as time went by, they discovered that
were 235,000 professors in the United States. By the riches of the universe were lying at their feet.
1980 there were 685,000. . . . The rewards for intellectual capital have in-
creased while the rewards for physical capital
have not. That means that even liberal arts ma-
The Sixties jors can wake up one day and find themselves
The educated-class rebellion we call “the sixties” suddenly members of the top-income brackets.
was about many things, some of them impor- A full professor at Yale who renounced the cap-
tant and related to the Civil Rights movement italist rat race finds himself making, as of 1999,
and Vietnam, some of them entirely silly, and $113,100, while a professor at Rutgers pulls in
others, like the sexual revolution, overblown (ac- $103,700 and superstar professors, who become
tual sexual behavior was affected far more by the the object of academic bidding wars, now can
world wars than by the Woodstock era). But at rake in more than $300,000 a year. Congressio-
its core the cultural radicalism of the sixties was nal and presidential staffers top out at $125,000
a challenge to conventional notions of success. (before quintupling that when they enter the
It was not only a political effort to dislodge the private sector), and the journalists at national
establishment from the seats of power. It was a publications can now count on six-figure salaries
cultural effort by the rising members of the priv- when they hit middle age, not including lecture
ileged classes to destroy whatever prestige still fees. Philosophy and math majors head for Wall
attached to the WASP lifestyle and the WASP Street and can make tens of millions of dollars
314 • Part IV: The Ruling Class, Elites, and the Upper Class

from their quantitative models. America has al- bourgeois mainstream culture and the values of
ways had a lot of lawyers, and now the median the 1960s counterculture have merged. That cul-
income for that burgeoning group is $72,500, ture war has ended, at least within the educated
while income for the big-city legal grinds can class. In its place that class has created a third
reach seven figures. And super-students still culture, which is a reconciliation between the
flood into medicine—three-quarters of private previous two. The educated elites didn’t set out
practitioners net more than $100,000. Mean- to create this reconciliation. It is the product of
while, in Silicon Valley there are more million- millions of individual efforts to have things both
aires than people. . . . ways. But it is now the dominant tone of our
age. In the resolution between the culture and
the counterculture, it is impossible to tell who
The Anxieties of Abundance co-opted whom, because in reality the bohemi-
Those who want to win educated-class approval ans and the bourgeois co-opted each other. They
must confront the anxieties of abundance: how emerge from this process as bourgeois bohemi-
to show—not least to themselves—that even ans, or Bobos.
while climbing toward the top of the ladder they
have not become all the things they still profess
to hold in contempt. How to navigate the shoals The New Establishment
between their affluence and their self-respect. Today the New York Times weddings section is
How to reconcile their success with their spiritu- huge once again. In the early 1970s the young
ality, their elite status with their egalitarian ideals. rebels didn’t want to appear there, but now that
Socially enlightened members of the educated their own kids are in college and getting married,
elite tend to be disturbed by the widening gap they are proud to see their offspring in the Sun-
between rich and poor and are therefore made day paper. For a fee the Times will send you a re-
somewhat uncomfortable by the fact that their production of your listing, suitable for framing.
own family income now tops $80,000. Some of And the young people, the second-generation
them dream of social justice yet went to a col- Bobos, are willing to see their nuptials recorded.
lege where the tuition costs could feed an en- Look at the newlyweds on any given Sunday
tire village in Rwanda for a year. Some once had morning, beaming out at you from the pages of
“Question Authority” bumper stickers on their the Times. Their smiles seem so genuine. They
cars but now find themselves heading start-up all look so nice and approachable, not dignified
software companies with 200 people reporting or fearsome, the way some of the brides on the
to them. The sociologists they read in college 1950s pages did. Things are different but some-
taught that consumerism is a disease, and yet how similar. . . .
now they find themselves shopping for $3,000 Today’s establishment is structured differ-
refrigerators. They took to heart the lessons of ently. It is not a small conspiracy of well-bred
Death of a Salesman, yet now find themselves di- men with interlocking family and school ties who
recting a sales force. They laughed at the plastics have enormous influence on the levers of power.
scene in The Graduate but now they work for a Instead, this establishment is a large, amorphous
company that manufactures . . . plastic. group of meritocrats who share a consciousness
When you are amidst the educated upscalers, and who unself-consciously reshape institutions
you can never be sure if you’re living in a world to accord with their values. They are not con-
of hippies or stockbrokers. In reality you have fined to a few East Coast institutions. In 1962,
entered the hybrid world in which everybody is Richard Rovere could write, “Nor has the Estab-
a little of both. lishment ever made much headway in such fields
Marx told us that classes inevitably conflict, as advertising, television or motion pictures.”
but sometimes they just blur. The values of the Today’s establishment is everywhere. It exercises
 Brooks—Bobos in Paradise • 315

its power subtly, over ideas and concepts, and but who are you. Once you were established as
therefore pervasively. There are no sure-fire de- a Biddle or an Auchincloss or a Venderlip, your
mographic markers to tell who is a member of way was clear. But members of today’s educated
this establishment. Members tend to have gone class can never be secure about their own future.
to competitive colleges, but not all have. They A career crash could be just around the corner.
tend to live in upscale neighborhoods, such as In the educated class even social life is a series of
Los Altos, California, and Bloomfield, Michi- aptitude tests; we all must perpetually perform in
gan, and Lincoln Park, Illinois, but not all do. accordance with the shifting norms of propriety,
What unites them is their shared commitment ever advancing signals of cultivation. Reputa-
to the Bobo reconciliation. People gain entry tions can be destroyed by a disgraceful sentence,
into the establishment by performing a series of a lewd act, a run of bad press, or a terrible speech
delicate cultural tasks: they are prosperous with- at the financial summit at Davos.
out seeming greedy; they have pleased their el- And more important, members of the edu-
ders without seeming conformist; they have risen cated class can never be secure about their chil-
toward the top without too obviously looking dren’s future. The kids have some domestic and
down on those below; they have achieved success educational advantages—all those tutors and
without committing certain socially sanctioned developmental toys—but they still have to work
affronts to the ideal of social equality; they have through school and ace the SATs just to achieve
constructed a prosperous lifestyle while avoiding the same social rank as their parents. Compared
the old clichés of conspicuous consumption (it’s to past elites, little is guaranteed.
OK to hew to the new clichés). The irony is that all this status insecurity only
makes the educated class stronger. Its members
and their children must constantly be alert,
Class Rank working and achieving. Moreover, the educated
This has got to be one of the most anxious social class is in no danger of becoming a self-contained
elites ever. We Bobos are not anxious because caste. Anybody with the right degree, job, and
there is an angry mob outside the gates threaten- cultural competencies can join. Marx warned
ing to send us to the guillotine. There isn’t. The that “the more a ruling class is able to assimi-
educated elite is anxious because its members late the most prominent men [or women] of the
are torn between their drive to succeed and their dominated classes, the more stable and danger-
fear of turning into sellouts. Furthermore, we are ous its rule.” And in truth it is hard to see how
anxious because we do not award ourselves status the rule of the meritocrats could ever come to an
sinecures. Previous establishments erected social end. The WASP Establishment fell pretty eas-
institutions that would give their members se- ily in the 1960s. It surrendered almost without
curity. In the first part of the 20th century, once a shot. But the meritocratic Bobo class is rich
your family made it into the upper echelons of with the spirit of self-criticism. It is flexible and
society, it was relatively easy to stay there. You amorphous enough to co-opt that which it does
were invited on the basis of your connections to not already command. The Bobo meritocracy
the right affairs. You were admitted, nearly au- will not be easily toppled, even if some group
tomatically, to the right schools and considered of people were to rise up and conclude that it
appropriate for the right spouses. The pertinent should be.
question in those circles was not what do you do,
316 • Part IV: The Ruling Class, Elites, and the Upper Class

36 • Gil Eyal, Iván Szelényi, and Eleanor Townsley

Post-Communist Managerialism

The unique feature of making capitalism from the former communist nomenklatura knew by
the ruins of state socialism in Central Europe 1988 that the destruction of the old communist
is that it is happening without a propertied order was inevitable. They therefore designed a
bourgeoisie. In all other historical sites where scheme to convert political office into private
modern capitalism has developed, some form wealth, and attempted to transform themselves
of private property and some class of private into a new grande bourgeoisie. Indeed, many
proprietors—no matter how embryonic, and commentators on the Central European trans-
no matter how different from modern capitalist formation think that this is what happened after
entrepreneurs—already existed. In the classical 1989: communist officials used political office
case of transition, feudal landlords gradually to convert public goods into private individual
converted their property into private ownership wealth—de facto, they stole state property and
and began to be recruited into the new grande became a ‘kleptocracy’. . . .
bourgeoisie. Urban artisans and merchants were The second theory we confront is Erzsébet
busily accumulating capital, and were well posi- Szalai’s theory of ‘technocratic revolution,’ which
tioned to transform themselves from the third in some ways, can be viewed as a refinement of
estate of a feudal order into one of the fractions the general political capitalism thesis.2 Szalai’s
of the new dominant class in a capitalist mode argument is that the late state socialist nomen-
of production. Post-communism is the first sit- klatura was highly fragmented, and that the dy-
uation where the transition to private property namics of social change should be understood as
from a collective form of ownership is being at- an intense struggle between the bureaucratic and
tempted. Moreover, this project is being led by technocratic fractions of the old ruling estate. In
the second Bildungsbürgertum—by an uneasy al- this view, 1989 was a successful revolution of
liance between former communist apparatchiks, the late state-socialist technocracy against the
technocrats, managers and their former left-wing bureaucratic fraction of the communist ruling
critics, the dissident intellectuals. In short, capi- estate. . . . Rather than the nomenklatura as a
talism is being made by a coalition of property- whole grasping power, one of its fractions—the
less agents, who only yesterday outbid each other technocratic-managerial elite—appeared to have
in their anticapitalism. established itself as the new propertied class . . .
In 1988 two leading sociologists of the re- In 1990, we launched a survey in six East
gion, Jadwiga Staniszkis and Elemér Hankiss, and Central European countries—Russia, Po-
formulated a provocative hypothesis, which land, the Czech Republic, Slovakia, Hungary
[may be termed] the theory of political capi- and Bulgaria—to assess the empirical support
talism.1 According to Staniszkis and Hankiss, for [these] forecasts. In 1993, in each country,

Gil Eyal, Iván Szelényi, and Eleanor Townsley. “Post-Communist Managerialism,” in Making Capitalism With-
out Capitalists: The New Ruling Elites in Eastern Europe, pp. 14–15, 113–115, 117, 119–121, 151–152, 230,
243. Copyright © 1998 by Verso Press. Used by permission of Verso Press.
 Eyal, Szelényi, and Townsley—Post-Communist Managerialism • 317

Table 36.1. Occupational Destinations in 1993 of People Who Were in Nomenklatura Positions in 1988 by
Country

Occupation in 1993 Czech Republic Hungary Poland

All in position of authority 51.7 43.1 51.2

High political office 3.0 6.4 9.0
High manager—public 16.2 11.2 13.4
High manager—private 12.8 2.4 9.1
High cultural office 1.1 4.4 7.1
Low-level managers 12.6 13.0 8.6
Entrepreneurs 6.0 5.7 4.0
Professionals 12.2 19.9 13.9
Workers 12.6 5.5 9.5
Retired early (younger than 65) 15.4 19.1 17.2
Other retired and unemployed 8.1 12.6 8.2

All respondents 100% 100% 100%

(n) (468) (803) (849)

Note: Numbers may not add up to 100 percent due to rounding.

we interviewed 1,000 people who were mem- individuals who occupied nomenklatura posi-
bers of the 1988 nomenklatura, 1,000 people tions in 1988 in Hungary, the Czech Republic,
who belonged to the new economic, political, and Poland.3 While there is some variation across
and cultural elites at the time; we also conducted the countries, the main finding in the table is
personal life history interviews of 5,000 adults one of massive downward mobility among no-
randomly selected from the population. In this menklatura members during the first five years of
chapter we present data from three Central Eu- post-communism. Only half of those who occu-
ropean countries: the Czech Republic, Hungary, pied nomenklatura positions in 1988 were still in
and Poland. We ask: what happened to the old positions of authority in 1993, and this includes
nomenklatura? What are the social origins of rather minor positions in low-level management.
members of the new elite? Is there any evidence Indeed, the proportion of former nomenklatura
for the existence of a propertied bourgeoisie by members who occupied any authority position
1993? If so, how is this new class of domestic in 1993 was rather low. And—rather surpris-
proprietors constituted? How much, and what, ingly—in Hungary, which boasted the most
do they own? advanced market reform policies of the late com-
munist period, the loss of authority positions
among former nomenklatura members has been
Whatever Happened to the even more marked than it has been in the other
Nomenklatura? countries: only 43.1 percent of the Hungarian
Political capitalism theory expects to find that nomenklatura retained jobs in which they have
the old communist elite has turned itself into the subordinates . . .
new propertied bourgeoisie of post-communist If we dig deeper into the data, the problems
society. Its most general proposition is that peo- with the political capitalism thesis become even
ple who were in nomenklatura positions prior to more serious. In its original formulation, this
1989 were able to retain their power and privi- theory stated that political office had been used
lege through the post-communist transition by for the accumulation of private wealth in Central
converting their political capital into private Europe. In order to test the accuracy of this state-
economic wealth. Our data cast doubt on these ment, we need to disaggregate the nomenklatura
predictions. category further. There were very different kinds
Table 36.1 describes the 1993 post- of nomenklatura positions in the communist sys-
communist occupational destinations of those tem, and good reason to think that they were
318 • Part IV: The Ruling Class, Elites, and the Upper Class

divided among themselves. For this reason, a fair fraction of the nomenklatura was the least success-
test of political capitalism theory should investi- ful in weathering the post-communist transition.
gate whether or not any particular component of Only 39.3 percent of the political nomenklatura
the nomenklatura has successfully negotiated the retained positions of authority between 1988 and
post-communist transition and become private 1993, compared to 44.2 percent of the cultural
proprietors. After all, some of our nomenklatura elite and 70.7 percent of the economic elite. Sec-
members in 1988 were top managers of large ond, early retirement was also much more com-
firms—thus they were ‘technocrats,’ or what we mon among the political fraction (20.9 percent)
term the ‘economic elite’ of the late communist of the nomenklatura than among the economic
period. In their case, it is not obvious that be- and cultural fractions (14.7 and 11.1 percent
coming a manager of a privately owned firm respectively). This suggests that political capital
is a conversion of political capital into private was much less useful than cultural capital, and
wealth—managers and technocrats may simply particularly cultural capital in the form of man-
have used their human capital and their man- agerial expertise, for successfully navigating the
agerial experience to maintain senior economic pitfalls of post-communism. Third, Table 36.2
positions in the post-communist transition. suggests that former Communist Party and state
Other former nomenklatura members belonged officials who comprised the political fraction of
to the ‘cultural elite’—they were rectors of uni- the nomenklatura in 1988 had less success in en-
versities, Members of the Academies of Sciences, tering new private-sector positions than former
or editors of daily newspapers—and these jobs communist managers. Among the former mem-
in the cultural sector were not positions from bers of the economic nomenklatura, 24.6 percent
which one could easily generate vast amounts (5.4 plus 19.2 percent) either owned or managed
of private wealth. Thus, political capitalism a private business in 1993, compared to only 9.8
theory would not necessarily expect former percent (5.8 plus 4.0 percent) of members of the
members of the communist cultural elite to be political fraction. These findings indicate that it
those most able to convert political office into was more advantageous to be a manager than a
private wealth. Arguably, however, the political party or state official if one wanted to enter the
capitalism thesis should hold for members of new economic elite of post-communism. More-
the ‘political elite’. These were people who held over, these findings directly refute arguments by
positions in the Communist Party apparatus or Staniszkis and Hankiss that it was the political
in the civil service, and of all the members of fraction of the former communist elite who were
the nomenklatura they were the best placed to best placed to take advantage of post-communist
use their ‘office’ to enrich themselves through the market reforms; rather, we find that members of
mechanism of ‘spontaneous privatization’.4 With the economic nomenklatura were much bigger
this disaggregation of the nomenklatura into its beneficiaries of the post-communist transition
economic, cultural, and political components, than members of the political nomenklatura. . . .
we are now in a position to offer a crucial test of
Staniszkis’s and Hankiss’s version of the political
capitalism thesis by asking: to what extent has Diffuse Property Relations as the
the ‘bureaucratic’ fraction of the ruling estate— Context for Managerial Control
the ‘political elite’—benefited economically from In one respect, however, our findings diverge
the post-communist transition? from Szalai’s predictions. Staniszkis and Hankiss
Table 36.2 documents the 1993 occupational anticipated that former communists would use
destinations of former nomenklatura members their power to become the new corporate owners
for each component of the nomenklatura: the of post-communism, while Szalai argued more
economic fraction, the cultural fraction, and the specifically that the technocratic-managerial elite
political fraction. It shows, first, that the political was the most likely candidate to achieve this
 Eyal, Szelényi, and Townsley—Post-Communist Managerialism • 319

Table 36.2. Occupational Destinations in 1993 of People Who Were in Economic, Political, and Cultural
Nomenklatura Positions in 1988 in All Three Countries (Czech Republic, Hungary, and Poland)

Occupation in 1993 Economic Elite, 1988 Political Elite, 1988 Cultural Elite, 1988

All in position of authority 70.7 39.3 44.2

High political office 1.5 10.8 1.3
High manager—public 33.4 5.9 7.8
High manager—private 19.2 4.0 1.4
High cultural office 0.4 1.8 19.3
Low-level managers 10.8 11.0 11.8
Entrepreneurs 5.4 5.8 2.5
Professionals 4.7 16.9 27.4
Workers 2.2 13.4 3.3
Retired early (younger than 65) 14.7 20.9 11.1
Other retired and unemployed 7.6 9.4 14.1

All respondents 100% 100% 100%

(n) (536) (1,186) (398)

Note: Numbers may not add up to 100 percent due to rounding.

aim. Our data suggest that both these predic- of which have not led to ownership by iden-
tions miss the mark. Managerial ownership, or tifiable individuals. Our data on property also
the management buy-out of state-owned firms, document diffuse patterns of ownership in
is not the major story in post-communism. In- post-communist Central Europe, and with the
deed, the majority of corporate and industrial exception of foreign-owned firms (which are re-
managers have acquired no business property ally significant only in Hungary), it is not easy
at all [for details, see Eyal, Szelényi and Towns- to tell who the real owners are. Direct or indirect
ley5]. Furthermore, fully half of those who own public ownership, institutional cross-ownership,
businesses possess stakes not in the firms they ownership by banks that are owned by the gov-
manage, but in small subcontracting firms. Fi- ernment or state privatization agencies, and
nally, we find that those who own shares in the self-ownership (firms owning firms, which own
businesses they manage are likely to be managers them) are all typical. Together, this creates the
of smaller firms, and typically own only a small material base for the substantial autonomy and
fraction of the assets of these firms. In other power exercised by non-propertied technocrats
words, the former communist technocracy do and managers.
not hold ultimate economic decision-making Finally, while the big winners of the
power as owners, as Szalai predicted; rather, they post-communist transition are former commu-
exercise power as experts and managers. nist technocrats, we find that they cannot rule
While data available to us on ownership re- by themselves. They have been forced to create
lations in large firms are sketchy and may not a hegemonic power bloc together with the new
be sufficiently representative, the evidence at politocracy and the opinion-making intellectual
our disposal supports hypotheses put forward elite, and these two groups are composed largely
on the basis of ethnographic observations by of former dissident intellectuals.7 Immediately
David Stark and Larry King.6 Stark found that following the fall of communism, the new poli-
ownership in Central European corporations is tocracy and opinion-making intellectual elite
‘recombinant’, that is, it is neither private nor made an attempt to squeeze the former com-
public. King found firms with ‘recombinant’ munist technocracy out of power. They soon
property, too, but he also identified a number learned, however, that neither fraction of the
of alternative strategies of privatization, most intellectual elite could rule alone. During the
320 • Part IV: The Ruling Class, Elites, and the Upper Class

second post-communist elections, many former financial agencies. In the absence of a class
dissidents were dropped from the politocracy, of big private proprietors, the power of fi-
and the late communist pragmatists joined the nance managers is not a function of how
new political elite. These are strange bedfel- many shares they own in the banks they
lows indeed, who form the ‘unholy alliance’ of manage, or in the firms their banks manage.
post-communism. . . . Rather, their power is a form of ‘cultural
capital’; it is a function of their capacity
to appropriate the sacred knowledge of the
An Outline of a Theory of Managerial workings of the world capitalist system.
Capitalism Thesis 5. Even though Central European
On the basis of the analyses presented above managers are not limited by the power of a
[and elsewhere],8 we summarize our theory of propertied bourgeoisie, we would empha-
managerial capitalism in the following six theses. size that they do not exercise power in a
Thesis 1. Post-communist economies are vacuum. Rather, they occupy a historically
characterized by diffuse property relations. distinctive post-communist class context,
At the present time it is impossible to iden- formed by ongoing struggles over privatiza-
tify individuals or groups of individuals tion and the formation of new class actors.
with sufficient amounts of property who These struggles take place among members
are able to exercise anything even remotely of the power bloc—managers, technocrats,
similar to owners’ control of economic and intellectuals, in the first instance—and
decision-making. between this power bloc and former bureau-
Thesis 2. Ironically, it was precisely so-called crats, in the second. As we have argued [else-
‘privatization’ which created these diffuse where],9 no single class fraction has emerged
property relations. Privatization destroyed as the decisive victor in these struggles, and
redistributive control over state firms, but it as a consequence the power of managers
has not produced identifiable owners (yet). and their capacity to control semi-public
Thesis 3. The dispersion of property rights property comes to them by default.
is a universal phenomenon, but in market Thesis 6. Managerial strategies reflect their
capitalist economies with an established knowledge that the current balance of class
propertied bourgeoisie it faces strict limits forces is precarious. They understand that
which do not exist in post-communism. they exercise power by default. In order to
Post-communist managers do not have to survive, therefore, managers have developed
contend with a class of powerful capital- a diverse range of strategies to navigate the
ist proprietors; consequently, managerial political and economic uncertainties of
power and decision-making are visible con- post-communism. Probably the most preva-
tributions to the prestige and ‘distinction’ of lent managerial strategy was not managerial
the new power bloc. ‘buy-out’ but, rather, an attempt to stand
Thesis 4. Given the dispersion of property ‘on as many legs as possible’. During the
rights, the central representative of mana- process of privatization most managers ac-
gerial power in Central Europe is not the quired some property, but this was typically
manager of the industrial firm, but the fi- a relatively small stake, and not even nec-
nance manager. The most powerful people essarily in the firms they managed. Indeed,
of the post-communist era are bank man- as early as the late 1980s some members
agers, managers of investment funds, ex- of management teams were busy setting
perts at the Ministry of Finance, advisors up small subcontracting firms, owned by
at the IMF and the World Bank, and ex- themselves or by members of their fami-
perts working for foreign and international lies. They subcontracted the most lucrative
 Eyal, Szelényi, and Townsley—Post-Communist Managerialism • 321

activities of the state firms they managed 3. The population of nomenklatura members
to these companies, they even sold some of was defined as those individuals occupying the top
the more valuable assets of the parent firms 3,000–5,000 positions in these countries in 1988,
to these subcontracting units at underval- positions for which appointment usually required
the approval of some organ or official of the Central
ued prices. Still, it is probably the excep-
Committee of the Communist Party.
tion to the rule that these managers retired 4. In both Hungary and Poland during 1988,
from their main firm altogether, that they the regimes launched programs which were referred
‘jumped the boat’ to run the subcontract- to as ‘spontaneous privatization’. Communist elites
ing firms they own. The reason for their re- acknowledged the necessity of changing property
luctance to do so is clear: why should they rights, and privatizing publicly held assets, but they
swap a major managerial job for the posi- initiated this process in a rather unregulated way.
tion of owner-manager in a minor operation Under spontaneous privatization, firms could initi-
which employs only a handful of people? ate their own privatization and negotiate their own
terms with state organizations (in 1988, this still
On the other hand, managers also have an
meant the Communist Party). Indeed, the initial
interest in being more than only managers. hypotheses about political capitalism put forward by
In post-communist society, the managerial Staniszkis and Hankiss were formulated in reaction
elite is closely intertwined with the politoc- to these spontaneous privatization plans.
racy; hence post-communist managers are 5. Gil Eyal, Iván Szelényi and Eleanor Townsley
even more dependent upon politicians than (1998) Making Capitalism Without Capitalists: The
are capitalist managers in the West. State New Ruling Elites in Eastern Europe. London: Verso.
bureaucracies in East Central Europe often 6. David Stark (1996) ‘Recombinant Property in
have the power, through direct or indirect East European Capitalism’. American Journal of So-
ciology 101(4): 993–1027; Larry King (1997) Path-
state ownership of firms, to appoint and
ways from Socialism: The Transformation of Firms in
dismiss managers. As long as their position
Hungary, the Czech Republic and Slovakia. Ph.D dis-
can be threatened by the political elite, it sertation, Department of Sociology, UCLA.
seems to be wise for managers to have their 7. Erzsébet Szalai (1994) ‘The Power Structure in
own small private firm in the background. Hungary after the Political Transition’. Pp. 120–43
in The New Great Transformation, ed. Christopher
NOTES G.A. Bryant and Edmund Mokrzycki. London and
1. Jadwiga Staniszkis (1991) The Dynamics of New York: Routledge; Erzsébet Szalai (1994) Utelag-
Breakthrough. Berkeley: University of California azás: Hatalom és értelmiség az államszocializmus után
Press; Elemér Hankiss (1990) East European Alter- (At the crossroads: power and intellectuals after state
natives. Oxford: Clarendon Press. socialism). Budapest: Pesti Szalon Kiadó.
2. Erzsébet Szalai (1989) ‘The New Elite’. Across 8. Eyal, Szelényi and Townsley, Making Capital-
Frontiers 5(Fall-Winter): 25–31. [In Hungarian: ism Without Capitalists.
Beszélö 26, 1989]. 9. Ibid.
322 • Part IV: The Ruling Class, Elites, and the Upper Class

37 • Andrew G. Walder

China’s Evolving Oligarchy

Before the onset of its market reforms in the open recruitment through advancement within
1980s, the People’s Republic of China had a a broad party organization that recruited indi-
clearly defined elite typical of the single-party viduals at a young age and sponsored them for
hierarchy and command economy that it in- subsequent advancement (Li and Walder 2001).
herited from the Soviet Union. Career paths This was not, however, a social hierarchy based
into positions of power and privilege were con- on property and private wealth. Instead, rank
trolled by the Communist Party organization, determined access to public property and re-
access to which required demonstration of loy- sources. There was no significant private sector
alty through party membership and a record of before 1980. Significant private wealth could
loyal service. The only alternative career path not be accumulated, even at the highest levels
was through higher education, a credential that of the hierarchy. Families could not transmit
could lead without party membership to rela- wealth across generations, and assets could not
tively well-compensated professional positions be moved overseas. There was no career path
(Li and Walder 2001; Walder, Li, and Treiman outside the party hierarchy; a party career was
2000). Material privileges that determined a the only route to higher status, privilege, and
family’s style of life were determined by rank in authority (Walder 1995).
the hierarchy: housing standards, salary, access China’s political trajectory was different from
to scarce consumer items and higher-grade ser- those of many other state socialist regimes. Its
vices, the ability to travel domestically and inter- elite was attacked during the Cultural Revolu-
nationally, and the use of private automobiles. tion of 1966–1976, Mao Tse-tung’s effort to
Enjoyment of these privileges was premised on prevent the formation of an elitist, Soviet-style
continued good standing in the political hierar- bureaucracy. Many high-ranking officials lost
chy. Expulsion from the party or from leading their posts and were sent to perform manual
posts led to a loss of privileges (Walder 1992). labor. Their relatives’ opportunities were ham-
Because material living standards were so clearly pered by their family’s clouded political status.
linked to the political hierarchy, the elite could Austerity and frugal living standards were en-
be defined very precisely, with reference to the forced more stringently for those officials who
number of positions at different ranks. were left. Ranking political officials were un-
At the apex of this political hierarchy in 1981 able to transmit higher status to their offspring
were 1,625 party officials at the ministerial level (Walder and Hu 2009).
or above and 23,168 at the district level (Organi- This situation began to change fundamen-
zational History 2000: 1365, 1369). This was a tally in the early 1980s. Mao died in 1976, and
classic power elite as defined by Giddens (Chap- the post-Mao leadership gradually dismantled
ter 32): it ruled as an oligarchy, but had relatively the old Soviet-style planning system in favor

Original article prepared for this fourth edition of Social Stratification: Class, Race, and Gender in Sociological
Perspective, edited by David B. Grusky. Copyright © 2014 Westview Press.
 Walder—China’s Evolving Oligarchy • 323

of a more open, market-oriented economy some twenty-seven thousand with a net worth
with much greater scope for private enterprise. greater than US$1.25 million (Chinese News-
This occurred at the same time that previous net 2008). Although a great many private en-
policies hampering the careers and privileges trepreneurs have built fortunes outside the party
of ranking bureaucrats and their families were system, without becoming party members, the
abandoned. Just as a market economy was being party elite has by no means been isolated from
built, China’s officialdom was being liberated these developments, and the party system and its
from Mao-era political restrictions. Offspring of officeholders are closely integrated into networks
these elite households soon began to inherit their of private wealth.
parents’ high occupational and political status at The first and most obvious example of this
much greater rates (Walder and Hu 2009). Fear close integration is China’s system of state capi-
of the consolidation of an entrenched bureau- talism. In many postcommunist countries, state
cratic elite was no longer official party doctrine. enterprises were quickly privatized, either placed
Moreover, in an era that would soon see the under de facto managerial control or restruc-
collapse of communist-party states in the Soviet tured and sold to foreign investors. For almost
Union and almost everywhere else, China’s party two decades China resisted privatization and
hierarchy survived its own crisis in 1989, and propped up state enterprises with subsidies, until
its members and their families thrived during the debts proved unsustainable. Beginning in the
its market transition. In many postcommunist mid-1990s a drastic restructuring began, with
states, the issue was whether or not former mem- most of the smaller and less profitable state en-
bers of the communist-era elite could convert terprises being closed, downsized, or transferred
their prior elite status into wealth and power in to private owners at discounted rates. This did
the new economy (Eyal, Szelényi, and Townsley not mean the demise of the state sector. Larger
1998). In China the party remained in power, enterprises were also restructured and down-
maintaining an initially massive state-owned sized, but were then merged into even larger
economy and retaining enormous regulatory corporations and listed on domestic and foreign
authority. Almost unique among transitional stock exchanges. This occurred in a very wide
economies (except for Vietnam and Laos), the range of industries deemed “strategic” or “pillar”
Chinese market economy and private sector by the government: finance, mining, metallurgy,
grew up within the confines of a surviving com- chemicals, energy, telecommunications, civilian
munist party dictatorship. aviation, shipbuilding, tobacco, overseas ship-
After more than three decades, the outlines ping, machine building, railways, defense-related
of a vastly transformed elite are becoming clear. technology, construction, and power generation.
It is no longer the unitary, propertyless entity of In these sectors massive state enterprises have a
the former era. A party career and higher edu- monopolistic or oligopolistic position and are
cation are no longer the only routes to higher shielded from significant private and foreign
status. Enormous fortunes have been built on competition. Large state enterprises also remain
both private and public enterprises, and personal in alcoholic beverages, textiles, automobiles, and
wealth and private property are a central feature other sectors (Walter 2010).
of China’s new system of social stratification. By After more than a decade of restructur-
2008 one source documented approximately ing, massive state corporations have emerged
one thousand Chinese individuals with a net from a process that saw a rapid overall decline
worth of at least US$100 million (Hurun Re- in the proportion of state-owned enterprises,
port 2009); Web-based news sources in China, from some three-quarters in the late 1970s to
quoting internal government studies, reported about one-quarter in recent years. It is a more
more than thirty-two hundred people with a concentrated and powerful state sector than the
net worth greater than US$12.5 million and widely dispersed and localized state firms of the
324 • Part IV: The Ruling Class, Elites, and the Upper Class

previous era. Occupying a monopolistic posi- rely on collaboration between private and state
tion in the domestic market, given tax breaks capital to exploit resources controlled by govern-
and other indirect subsidies, and favored by a ments. The clearest such example is real estate
financial system that is still almost completely development, which has become one of the larg-
state owned and controlled, these corporations est and most important sectors driving domestic
have been highly profitable and are awash in economic growth. All land in China remains
capital; many have expanded through overseas state owned, but use rights can be sold to both
investment (Walter and Howie 2011). By 2009, public and private entities for extended periods.
thirty-four of these corporations were listed on At the outset of reform, very few occupants of
the Fortune Global 500, and eighty-nine were urban space had ownership claims, and even if
listed on the Forbes Global 2000. Chinese state they did, local governments could ignore them,
corporations were well represented in lists of the designate entire districts for redevelopment, and
ten largest steel, shipping, power generation, order the removal of households and organiza-
railway, and telecommunications enterprises, tions. Pent-up demand for urban housing and
and China’s state banks were among the largest rising demand for commercial space in a rapidly
in the world. Only a small number of privately growing economy have made this sector highly
controlled corporations were on these lists, pri- profitable. Real estate development companies
marily in technology sectors, and they were at typically have a mixed ownership structure, with
the bottom in terms of size (Walder 2011). capital from state corporations that find real es-
These enterprises are in essence owned and tate more profitable than their core businesses
controlled by the Chinese Communist Party joining with private firms that have expertise
(Walter and Howie 2011), even though almost in engineering, design, and marketing. Local
all list shares on domestic and foreign stock governments frequently take an equity stake in
exchanges. At least half of their shares are non- a project, with current officials and their family
tradable, ensuring state control, but state share members sometimes forming companies to par-
concentration often is at much higher levels. ticipate in projects as private partners. The real
All of the management personnel are part of estate sector breeds more private wealth than any
the state’s traditional nomenklatura system, the other. In 2009, 60 of China’s 129 documented
Soviet-inherited institution that ensures the par- US dollar billionaires were in real estate, some 40
ty’s control over appointments into positions of percent of the total. Of the 999 high-wealth in-
authority. The party secretaries (now known as dividuals identified that year, 326 were in real es-
CEOs) and other top management figures are tate, with average household wealth of US$703
routinely transferred across enterprises and into million. Eight of the ten wealthiest individuals
and out of government and party posts. The on the 2009 list were real estate magnates, with
commanding heights of the corporate economy, net wealth ranging from US$1.9 billion to $4
in other words, have become fully integrated billion (Hurun Report 2009).
with the structure of the party-state. The officials Party regulations formally prohibit officials
who work as corporate executives in this system from owning or operating private businesses,
do not receive the inexcusably large compensa- but there are no effective prohibitions against
tion packages and equity stakes that corporate their immediate family members, relatives, or
executives frequently lavish upon themselves in close associates doing so. Family ties facilitate
the United States, but they enjoy extensive cor- the accumulation of large fortunes through
porate perks (a private automobile with driver, both of the means just described. The relatives
subsidized homes and travel) and very high sal- of high officials are prominently represented in
aries that can run into millions of US dollars. the ranks of the top management of state-owned
The second example of close integration be- enterprises, a form of nepotism among the party
tween the party and the market is the sectors that elite that predates the market reforms. Greater
 Walder—China’s Evolving Oligarchy • 325

fortunes are reaped by family members of high These realities have created a large private
officials, who form their own private companies sector that has little of the influence or politi-
and benefit from the firm’s implied political cal independence that one might expect, given
backing, even without any direct involvement its size. The private corporate elite has almost
by the officials themselves. In some cases family none of the independence and influence that
members may form consulting or other com- authors like Mills (see Chapter 31) and Dom-
panies that participate directly in local projects hoff (see Chapter 33) have described for the
over which the related official has administrative United States. Enterprise owners who express
authority. These activities are separate from out- critical viewpoints or support independent citi-
right bribery and the solicitation of kickbacks, zens’ groups or lawsuits rightly fear retaliation in
phenomena that are no less prevalent in China the form of a corruption investigation, suddenly
than in other highly regulated market economies inflated tax liability, or even a charge of reveal-
in developing countries, and which have led to ing state secrets. A supportive and cooperative
highly publicized corruption cases against prom- relationship with the relevant political authori-
inent officials at the national as well as provin- ties is essential for business success, and private
cial and municipal levels. News sources within business owners have sought to build close and
China frequently claim that implausibly high cordial relations with officialdom as part of their
percentages of the wealthiest individuals are business strategy. The party has responded for
the children of high officials (Chinese Newsnet more than a decade by paternalistically recruit-
2008). Although it is impossible to verify such ing leading business owners into the party and
claims, there is little doubt that official families membership in largely honorary official institu-
have prospered. tions like People’s Congresses (Dickson 2008).
Despite these advantages, it is by no means Analysts have noted that China’s private business
true that the party elite monopolizes opportuni- class has none of the liberal or independent po-
ties for private enrichment. The majority of en- litical leanings that one might expect (Dickson
terprises are privately owned, and only a fraction 2003; Gallagher 2002). Research strongly sup-
have family or other connections to party-state ports the conclusion that private business own-
officials. The large private sector and its class of ers are so far largely supportive of the political
owners, however, are in many ways subordinate institutions within which they have prospered.
to, and integrated with, the party-state elite. Pri- Over the past thirty years China has devel-
vate enterprises are not permitted to grow to a oped a new kind of “dual elite.” One wing is
size that rivals state enterprises in the “strategic” firmly rooted in the party-state, the other in an
and “pillar” industries, and they cannot compete independent sector based on private ownership.
on equal terms with state enterprises in these But there are close ties between the two sectors,
sectors, if they are even permitted to exist. Pri- which overlap to some extent and have so far
vate firms that grow to rival state firms in pillar been mutually supportive. The party elite is by
industries are likely to be forced to merge with no means insulated from the wealth that is being
their state-owned competitors. Private firms in created in either sector. This dual elite is strik-
other sectors, both large and small, are allowed ingly different from that of the pre-reform era,
to flourish, but they must maintain good rela- when the two primary routes to higher social po-
tionships with the government agencies that are sition were party activism and higher education.
in charge of licensing, regulation, and taxation. The new elite has members with roots in the
The discretion exercised by these agencies can thriving party-state system, the separate private
be decisive in a firm’s success and even survival. business sector, or both. Both wings of the elite
China’s legal system is fully under party control, have generated extensive demands for expensive
so there is little effective recourse for firms that real estate, luxury automobiles, and imported
fall afoul of local or national authorities. prestige consumer goods. China has become
326 • Part IV: The Ruling Class, Elites, and the Upper Class

the world’s largest market for many of the most China’s power elite has grown and evolved so
expensive luxury automobiles. Whereas in the rapidly that it has yet to develop the clear signs
1980s and 1990s Chinese students could hope of social closure that Domhoff describes for the
to study abroad only on scholarships provided United States. Private clubs have only begun to
by either the Chinese government or foreign form, though gated communities are common in
universities, full tuition-paying students from many cities. There is a clear set of elite domestic
China are now common, even in the most ex- universities, but they are large and open to all
pensive private institutions, and they have begun through meritocratic standardized examinations,
to make visible inroads into private boarding and they are beginning to be overshadowed by
schools that cater to the international elite. Chi- the status of degrees from elite international uni-
nese investors now occupy the prominent posi- versities and prep schools. If in future years the
tion in international luxury real estate markets relative openness of the opportunity structure
that was held by Japanese investors in the 1980s. begins to decline, and if families that now dom-
China still has a clearly identifiable power inate political power and the state and private
elite in Giddens’s terms, still ruling as an oli- corporate hierarchies set themselves off from the
garchy. It remains relatively open through re- rest of society through the kinds of mechanisms
cruitment into the Communist Party, but it has described by Domhoff, China could evolve into
become even more open through the building of Giddens’s definition of a ruling class: a uniform
private-sector businesses by individuals unaffili- and established elite that rules as an oligarchy.
ated with the party. Although the private busi- At the outset of China’s market reforms it was
ness sector is still clearly subordinate to the party commonly believed that the party would never
(firms unaffiliated with the party-centered power let the private economy outstrip the state econ-
elite are not fully part of it), it is very large. The omy in overall size, and that if it did, the party
party-owned state corporations, which control would surely not survive. Neither expectation
strategic and pillar industries, dominate the has been fulfilled. Few expected the rapid rise
corporate economy with enterprises that dwarf of this form of authoritarian capitalism or pre-
those in the private sector in size and assets. dicted the radical restructuring of the state sector
The country’s trajectory is fundamentally dif- that has led to the consolidation of dominant
ferent from that of the postcommunist states of state-owned corporations with international
Central Europe analyzed by Eyal, Szelényi, and reach, fully integrated into the party-state. The
Townsley (Chapters 36 and 127). Because the unitary bureaucratic elite of the Mao era has
Communist Party has survived, there has been evolved and prospered, integrated into a thriving
none of the massive downward mobility of for- market economy through both the corporations
mer officials that they document; in many ways in the monopolistic state sector and family ties to
the reverse has occurred. Because the regime re- private enterprise. The rise of a separate private
mains a party-dominated oligarchy, former op- business elite has by no means threatened the
positional critics and technocrats have not risen bureaucracy’s position, and private entrepreneurs
into elite positions. The situation described by have been integrated into the state system in a
these authors parallels in many ways the fluid and politically subordinate role. Large percentages of
somewhat fragmented elite structure described China’s party elite have strong ties to both the
by Mills (Chapter 31). In Giddens’s terms, it ap- state and private sectors. China’s authoritarian
pears to be a somewhat inchoate governing class system is relatively well organized and unified
in formation. Because of the Communist Party’s and has so far proven durable in the face of
survival and its resurrection of state firms as a an extensive transformation of its Soviet-style
dominant set of monopolistic corporations fully economy. At the same time a striking rise in
integrated into the party-state structure, China income inequality, a new social hierarchy based
remains a power elite: solid, oligarchic, but with on private wealth and property ownership, and a
greatly increased private wealth. highly visible new elite that exercises both wealth
 Walder—China’s Evolving Oligarchy • 327

and power have evolved. Whether the new elite Mobility into the Chinese Administrative Elite.”
endures in its current form depends on both the American Journal of Sociology 106: 1371–1408.
future prosperity of the Chinese economy and Organizational History. 2000. Zhongguo gongchan-
the continued stability of the party system that dang zuzhishi ziliao 16 [Materials on the Organi-
zational History of the Chinese Communist Party,
has nurtured it for more than thirty years.
vol. 16]. Beijing: Zhonggong dangshi chubanshe.
Walder, Andrew G. 1992. “Property Rights and
REFERENCES
Stratification in Socialist Redistributive Econo-
Chinese Newsnet. 2008. “Zhongguo yiwan fujia 9
mies.” American Sociological Review 57: 524–539.
cheng yishang shi gaogan zidi” [“More Than 90
———. 1995. “Career Mobility and the Commu-
Percent of China’s Wealthiest Are Children of
nist Political Order.” American Sociological Review
High Officials”]. http://www.dwnews.com/gb/
60: 309–328.
MainNews/Forums/Backstage/2008.html (ac-
———. 2011. “From Control to Ownership: Chi-
cessed March 8, 2008).
na’s Managerial Revolution.” Management and
Dickson, Bruce J. 2003. Red Capitalists in China:
Organization Review 7: 19–38.
The Party, Private Entrepreneurs, and Prospects for
Walder, Andrew G., and Songhua Hu. 2009. “Rev-
Political Change. Cambridge, UK: Cambridge
olution, Reform, and Status Inheritance: Urban
University Press.
China, 1949–1996.” American Journal of Sociol-
———. 2008. Wealth into Power: The Communist
ogy 114: 1395–1427.
Party’s Embrace of China’s Private Sector. Cam-
Walder, Andrew G., Bobai Li, and Donald J. Trei-
bridge, UK: Cambridge University Press.
man. 2000. “Politics and Life Chances in a State
Eyal, Gil, Iván Szelényi, and Eleanor Townsley. 1998.
Socialist Regime: Dual Career Paths into the
Making Capitalism Without Capitalists: Class For-
Urban Chinese Elite.” American Sociological Re-
mation and Elite Struggles in Post-Communist
view 65: 191–209.
Central Europe. London: Verso.
Walter, Carl E. 2010. “The Struggle over Own-
Gallagher, Mary E. 2002. “‘Reform and Openness’:
ership: How the Reform of State Enterprises
Why China’s Economic Reforms Have Delayed
Changed China.” Copenhagen Journal of Asian
Democracy.” World Politics 54: 338–372.
Studies 28: 83–108.
Hurun Report. 2009. 2009 China Rich List. http://
Walter, Carl E., and Fraser J. T. Howie. 2011. Red
www.hurun.net/listen104,page1.aspx (accessed
Capitalism: The Fragile Financial Foundation of
September 23, 2009).
China’s Extraordinary Rise. Singapore: Wiley.
Li, Bobai, and Andrew G. Walder. 2001. “Career
Advancement as Party Patronage: Sponsored
 PART V

Poverty and the Underclass

The Experience of Poverty

38 Nickel and Dimed 330
39 Low-Income Urban Fathers and the “Package Deal” of Family Life 339

Poverty and the Economy

40 Being Poor, Black, and American 347
41 Poverty and the Great Recession 357

The Effects of Politics and Institutions

42 How Rich Countries Lift Up the Poor 365
43 Taxing the Poor: How Some States Make Poverty Worse 369

Neighborhoods and Segregation

44 American Apartheid: Segregation and the Making of the Underclass 376
45 Legacies of Inequality 386
46 Does Changing Neighborhoods Change Lives? 395
47 The Legacy of Multigenerational Disadvantage 403

How Important Is Early Childhood?

48 Skill Formation and the Economics of Investing in
Disadvantaged Children 412
49 The Long Reach of Early Childhood Poverty 417
50 Stressing Out the Poor 423

Incarceration and Poverty

51 Incarceration and Social Inequality 431
330 • Part V: Poverty and the Underclass

The Experience of Poverty

38 • Barbara Ehrenreich

Nickel and Dimed

At the beginning of June 1998 I leave behind of diversion, only the challenge of evading the
everything that normally soothes the ego and landlord’s Doberman pinscher. The big prob-
sustains the body—home, career, companion, lem with this place, though, is the rent, which
reputation, ATM card—for a plunge into the at $675 a month is well beyond my reach. . . .
low-wage workforce. There, I become another, So I decide to make the common trade-off
occupationally much diminished “Barbara Eh- between affordability and convenience, and go
renreich”—depicted on job-application forms for a $500-a-month efficiency thirty miles up a
as a divorced homemaker whose sole work ex- two-lane highway from the employment oppor-
perience consists of housekeeping in a few pri- tunities of Key West, meaning forty-five min-
vate homes. I am terrified, at the beginning, of utes if there’s no road construction and I don’t
being unmasked for what I am: a middle-class get caught behind some sun-dazed Canadian
journalist setting out to explore the world that tourists. . . .
welfare mothers are entering, at the rate of ap- I am not doing this for the anthropology.
proximately 50,000 a month, as welfare reform My aim is nothing so mistily subjective as to
kicks in. Happily, though, my fears turn out to “experience poverty” or find out how it “really
be entirely unwarranted: during a month of pov- feels” to be a long-term low-wage worker. I’ve
erty and toil, my name goes unnoticed and for had enough unchosen encounters with poverty
the most part unuttered. In this parallel universe and the world of low-wage work to know it’s not
where my father never got out of the mines and I a place you want to visit for touristic purposes;
never got through college, I am “baby,” “honey,” it just smells too much like fear. And with all my
“blondie,” and, most commonly, “girl.” real-life assets—bank account, IRA, health in-
My first task is to find a place to live. I figure surance, multiroom home—waiting indulgently
that if I can earn $7 an hour—which, from the in the background, I am, of course, thoroughly
want ads, seems doable—I can afford to spend insulated from the terrors that afflict the genu-
$500 on rent, or maybe, with severe economies, inely poor.
$600. In the Key West area, where I live, this No, this is a purely objective, scientific sort
pretty much confines me to flophouses and trailer of mission. The humanitarian rationale for wel-
homes—like the one, a pleasing fifteen-minute fare reform—as opposed to the more punitive
drive from town, that has no air-conditioning, and stingy impulses that may actually have mo-
no screens, no fans, no television, and, by way tivated it—is that work will lift poor women out

Barbara Ehrenreich. “Nickel-and-Dimed: On (Not) Getting by in America,” Harper’s Magazine, January 1999,
pp. 37–52. Copyright © 1999 by Barbara Ehrenreich. Reprinted by permission of International Creative
Management, Inc.
 Ehrenrich—Nickel and Dimed • 331

of poverty while simultaneously inflating their Most of the big hotels run ads almost con-
self-esteem and hence their future value in the tinually, just to build a supply of applicants to
labor market. Thus, whatever the hassles in- replace the current workers as they drift away or
volved in finding child care, transportation, etc., are fired, so finding a job is just a matter of being
the transition from welfare to work will end hap- at the right place at the right time and flexible
pily, in greater prosperity for all. Now there are enough to take whatever is being offered that
many problems with this comforting prediction, day. This finally happens to me at one of the big
such as the fact that the economy will inevitably discount hotel chains, where I go for housekeep-
undergo a downturn, eliminating many jobs. ing and am sent, instead, to try out as a wait-
Even without a downturn, the influx of a mil- ress at the attached “family restaurant,” a dismal
lion former welfare recipients into the low-wage spot with a counter and about thirty tables that
labor market could depress wages by as much as looks out on a parking garage and features such
11.9 percent, according to the Economic Policy tempting fare as “Pollish [sic] sausage and BBQ
Institute (EPI) in Washington, D.C. sauce” on 95-degree days. Phillip, the dapper
But is it really possible to make a living on young West Indian who introduces himself as
the kinds of jobs currently available to unskilled the manager, interviews me with about as much
people? Mathematically, the answer is no, as can enthusiasm as if he were a clerk processing me
be shown by taking $6 to $7 an hour, perhaps for Medicare, the principal questions being what
subtracting a dollar or two an hour for child shifts can I work and when can I start. I mutter
care, multiplying by 160 hours a month, and something about being woefully out of practice
comparing the result to the prevailing rents. as a waitress, but he’s already on to the uniform:
According to the National Coalition for the I’m to show up tomorrow wearing black slacks
Homeless, for example, in 1998 it took, on av- and black shoes; he’ll provide the rust-colored
erage nationwide, an hourly wage of $8.89 to polo shirt with Hearthside embroidered on it,
afford a one-bedroom apartment, and the Pre- though I might want to wear my own shirt to get
amble Center for Public Policy estimates that the to work, ha ha. At the word “tomorrow,” some-
odds against a typical welfare recipient’s landing thing between fear and indignation rises in my
a job at such a “living wage” are about 97 to 1. chest. I want to say, “Thank you for your time,
If these numbers are right, low-wage work is not sir, but this is just an experiment, you know, not
a solution to poverty and possibly not even to my actual life.”
homelessness. . . . So begins my career at the Hearthside, I shall
On the morning of my first full day of job call it, one small profit center within a global dis-
searching, I take a red pen to the want ads, which count hotel chain, where for two weeks I work
are auspiciously numerous. Everyone in Key from 2:00 till 10:00 p.m. for $2.43 an hour plus
West’s booming “hospitality industry” seems to tips. In some futile bid for gentility, the manage-
be looking for someone like me—trainable, flex- ment has barred employees from using the front
ible, and with suitably humble expectations as to door, so my first day I enter through the kitchen,
pay. I know I possess certain traits that might be where a red-faced man with shoulder-length
advantageous—I’m white and, I like to think, blond hair is throwing frozen steaks against the
well-spoken and poised—but I decide on two wall and yelling, “Fuck this shit!” “That’s just
rules: One, I cannot use any skills derived from Jack,” explains Gail, the wiry middle-aged wait-
my education or usual work—not that there are ress who is assigned to train me. “He’s on the rag
a lot of want ads for satirical essayists anyway. again”—a condition occasioned, in this instance,
Two, I have to take the best-paid job that is of- by the fact that the cook on the morning shift
fered me and of course do my best to hold it; no had forgotten to thaw out the steaks. For the
Marxist rants or sneaking off to read novels in next eight hours, I run after the agile Gail, ab-
the ladies’ room. . . . sorbing bits of instruction along with fragments
332 • Part V: Poverty and the Underclass

of personal tragedy. All food must be trayed, and Gail’s support—“It’s okay, baby, everyone does
the reason she’s so tired today is that she woke up that sometime”—because, to my total surprise
in a cold sweat thinking of her boyfriend, who and despite the scientific detachment I am doing
killed himself recently in an upstate prison. No my best to maintain, I care. . . .
refills on lemonade. And the reason he was in Sometimes I play with the fantasy that I am
prison is that a few DUIs caught up with him, a princess who, in penance for some tiny trans-
that’s all, could have happened to anyone. Carry gression, has undertaken to feed each of her sub-
the creamers to the table in a monkey bowl, jects by hand. But the non-princesses working
never in your hand. And after he was gone she with me are just as indulgent, even when this
spent several months living in her truck, peeing means flouting management rules—concerning,
in a plastic pee bottle and reading by candlelight for example, the number of croutons that can go
at night, but you can’t live in a truck in the sum- on a salad (six). “Put on all you want,” Gail whis-
mer, since you need to have the windows down, pers, “as long as Stu isn’t looking.” She dips into
which means anything can get in, from mosqui- her own tip money to buy biscuits and gravy for
toes on up. an out-of-work mechanic who’s used up all his
At least Gail puts to rest any fears I had of money on dental surgery, inspiring me to pick
appearing overqualified. From the first day on, up the tab for his milk and pie. . . .
I find that of all the things I have left behind, Ten days into it, this is beginning to look like
such as home and identity, what I miss the most a livable lifestyle. I like Gail, who is “looking
is competence. Not that I have ever felt utterly at fifty” but moves so fast she can alight in one
competent in the writing business, in which place and then another without apparently being
one day’s success augurs nothing at all for the anywhere between them. I clown around with
next. But in my writing life, I at least have some Lionel, the teenage Haitian busboy, and catch
notion of procedure: do the research, make a few fragments of conversation with Joan, the
the outline, rough out a draft, etc. As a server, svelte fortyish hostess and militant feminist who
though, I am beset by requests like bees: more is the only one of us who dares to tell Jack to
iced tea here, ketchup over there, a to-go box shut the fuck up. I even warm up to Jack when,
for table fourteen, and where are the high chairs, on a slow night and to make up for a particu-
anyway? Of the twenty-seven tables, up to six larly unwarranted attack on my abilities, or so
are usually mine at any time, though on slow I imagine, he tells me about his glory days as
afternoons or if Gail is off, I sometimes have the a young man at “coronary school”—or do you
whole place to myself. There is the touch-screen say “culinary”?—in Brooklyn, where he dated a
computer-ordering system to master, which is, knock-out Puerto Rican chick and learned ev-
I suppose, meant to minimize server-cook con- erything there is to know about food. I finish
tact, but in practice requires constant verbal up at 10:00 or 10:30, depending on how much
fine-tuning: “That’s gravy on the mashed, okay? side work I’ve been able to get done during the
None on the meatloaf,” and so forth—while the shift, and cruise home to the tapes I snatched up
cook scowls as if I were inventing these refine- at random when I left my real home—Marianne
ments just to torment him. Plus, something I Faithfull, Tracy Chapman, Enigma, King Sunny
had forgotten in the years since I was eighteen: Ade, the Violent Femmes—just drained enough
about a third of a server’s job is “side work” that’s for the music to set my cranium resonating but
invisible to customers—sweeping, scrubbing, hardly dead. Midnight snack is Wheat Thins and
slicing, refilling, and restocking. If it isn’t all Monterey Jack, accompanied by cheap white
done, every little bit of it, you’re going to face the wine on ice and whatever AMC has to offer. To
6:00 p.m. dinner rush defenseless and probably bed by 1:30 or 2:00, up at 9:00 or 10:00, read
go down in flames. I screw up dozens of times at for an hour while my uniform whirls around
the beginning, sustained in my shame entirely by in the landlord’s washing machine, and then
 Ehrenrich—Nickel and Dimed • 333

it’s another eight hours spent following Mao’s The other problem, in addition to the
central instruction, as laid out in the Little Red less-than-nurturing management style, is that
Book, which was: Serve the people. this job shows no sign of being financially vi-
I could drift along like this, in some dreamy able. You might imagine, from a comfortable
proletarian idyll, except for two things. One distance, that people who live, year in and year
is management. If I have kept this subject on out, on $6 to $10 an hour have discovered some
the margins thus far it is because I still flinch to survival stratagems unknown to the middle class.
think that I spent all those weeks under the sur- But no. It’s not hard to get my co-workers to talk
veillance of men (and later women) whose job it about their living situations, because housing, in
was to monitor my behavior for signs of sloth, almost every case, is the principal source of dis-
theft, drug abuse, or worse. Not that managers ruption in their lives, the first thing they fill you
and especially “assistant managers” in low-wage in on when they arrive for their shifts. After a
settings like this are exactly the class enemy. In week, I have compiled the following survey:
the restaurant business, they are mostly former • Gail is sharing a room in a well-known
cooks or servers, still capable of pinch-hitting in downtown flophouse for which she and
the kitchen or on the floor, just as in hotels they a roommate pay about $250 a week. Her
are likely to be former clerks, and paid a salary roommate, a male friend, has begun hit-
of only about $400 a week. But everyone knows ting on her, driving her nuts, but the rent
they have crossed over to the other side, which would be impossible alone.
is, crudely put, corporate as opposed to human. • Claude, the Haitian cook, is desperate to
Cooks want to prepare tasty meals; servers want get out of the two-room apartment he
to serve them graciously; but managers are there shares with his girlfriend and two other,
for only one reason—to make sure that money unrelated, people. As far as I can deter-
is made for some theoretical entity that exists far mine, the other Haitian men (most of
away in Chicago or New York, if a corporation whom only speak Creole) live in similarly
can be said to have a physical existence at all. . . . crowded situations.
Managers can sit—for hours at a time if • Annette, a twenty-year-old server who is
they want—but it’s their job to see that no one six months pregnant and has been aban-
else ever does, even when there’s nothing to do, doned by her boyfriend, lives with her
and this is why, for servers, slow times can be mother, a postal clerk.
as exhausting as rushes. You start dragging out • Marianne and her boyfriend are paying
each little chore, because if the manager on duty $170 a week for a one-person trailer.
catches you in an idle moment, he will give you • Jack, who is, at $10 an hour, the wealthiest
something far nastier to do. So I wipe, I clean, of us, lives in the trailer he owns, paying
I consolidate ketchup bottles and recheck the only the $400-a-month lot fee.
cheesecake supply, even tour the tables to make • The other white cook, Andy, lives on his
sure the customer evaluation forms are all stand- dry-docked boat, which, as far as I can tell
ing perkily in their places—wondering all the from his loving descriptions, can’t be more
time how many calories I burn in these strictly than twenty feet long. He offers to take me
theatrical exercises. When, on a particularly dead out on it, once it’s repaired, but the offer
afternoon, Stu finds me glancing at a USA Today comes with inquiries as to my marital sta-
a customer has left behind, he assigns me to vac- tus, so I do not follow up on it.
uum the entire floor with the broken vacuum • Tina and her husband are paying $60 a
cleaner that has a handle only two feet long, and night for a double room in a Days Inn.
the only way to do that without incurring ortho- This is because they have no car and the
pedic damage is to proceed from spot to spot on Days Inn is within walking distance of
your knees. . . . the Hearthside. When Marianne, one of
334 • Part V: Poverty and the Underclass

the breakfast servers, is tossed out of her until she ran out of money for estrogen pills. She
trailer for subletting (which is against the is supposed to be on the company plan by now,
trailer-park rules), she leaves her boyfriend but they claim to have lost her application form
and moves in with Tina and her husband. and need to begin the paperwork all over again.
• Joan, who had fooled me with her nu- So she spends $9 per migraine pill to control the
merous and tasteful outfits (hostesses wear headaches she wouldn’t have, she insists, if her
their own clothes), lives in a van she parks estrogen supplements were covered. Similarly,
behind a shopping center at night and Marianne’s boyfriend lost his job as a roofer be-
showers in Tina’s motel room. The clothes cause he missed so much time after getting a cut
are from thrift shops. on his foot for which he couldn’t afford the pre-
scribed antibiotic.
It strikes me, in my middle-class solipsism, My own situation, when I sit down to assess
that there is gross improvidence in some of these it after two weeks of work, would not be much
arrangements. When Gail and I are wrapping sil- better if this were my actual life. The seductive
verware in napkins—the only task for which we thing about waitressing is that you don’t have to
are permitted to sit—she tells me she is thinking wait for payday to feel a few bills in your pocket,
of escaping from her roommate by moving into and my tips usually cover meals and gas, plus
the Days Inn herself. I am astounded: How can something left over to stuff into the kitchen
she even think of paying between $40 and $60 drawer I use as a bank. But as the tourist busi-
a day? But if I was afraid of sounding like a so- ness slows in the summer heat, I sometimes leave
cial worker, I come out just sounding like a fool. work with only $20 in tips (the gross is higher,
She squints at me in disbelief, “And where am but servers share about 15 percent of their tips
I supposed to get a month’s rent and a month’s with the busboys and bartenders). With wages
deposit for an apartment?” I’d been feeling pretty included, this amounts to about the minimum
smug about my $500 efficiency, but of course wage of $5.15 an hour. Although the sum in the
it was made possible only by the $1,300 I had drawer is piling up, at the present rate of accu-
allotted myself for start-up costs when I began mulation it will be more than a hundred dollars
my low-wage life: $1,000 for the first month’s short of my rent when the end of the month
rent and deposit, $100 for initial groceries and comes around. Nor can I see any expenses to cut.
cash in my pocket, $200 stuffed away for emer- True, I haven’t gone the lentil-stew route yet, but
gencies. In poverty, as in certain propositions in that’s because I don’t have a large cooking pot,
physics, starting conditions are everything. pot holders, or a ladle to stir with (which cost
There are no secret economies that nourish about $30 at Kmart, less at thrift stores), not to
the poor; on the contrary, there are a host of spe- mention onions, carrots, and the indispensable
cial costs. If you can’t put up the two months’ bay leaf. I do make my lunch almost every day—
rent you need to secure an apartment, you end usually some slow-burning, high-protein combo
up paying through the nose for a room by the like frozen chicken patties with melted cheese on
week. If you have only a room, with a hot plate top and canned pinto beans on the side. Dinner
at best, you can’t save by cooking up huge len- is at the Hearthside, which offers its employees a
til stews that can be frozen for the week ahead. choice of BLT, fish sandwich, or hamburger for
You eat fast food, or the hot dogs and Styrofoam only $2. The burger lasts longest, especially if
cups of soup that can be microwaved in a con- it’s heaped with gut-puckering jalapeños, but by
venience store. If you have no money for health midnight my stomach is growling again.
insurance—and the Hearthside’s niggardly plan So unless I want to start using my car as a
kicks in only after three months—you go with- residence, I have to find a second, or alterna-
out routine care or prescription drugs and end tive, job. I call all the hotels where I filled out
up paying the price. Gail, for example, was fine housekeeping applications weeks ago—the
 Ehrenrich—Nickel and Dimed • 335

Hyatt, Holiday Inn, Econo Lodge, HoJo’s, But the chances of this are minuscule. She has
Best Western, plus a half dozen or so locally left the flophouse and her annoying roommate
run guesthouses. Nothing. Then I start mak- and is back to living in her beat-up old truck.
ing the rounds again, wasting whole mornings But guess what? she reports to me excitedly later
waiting for some assistant manager to show that evening: Phillip has given her permission to
up, even dipping into places so creepy that the park overnight in the hotel parking lot, as long as
front-desk clerk greets you from behind bul- she keeps out of sight, and the parking lot should
letproof glass and sells pints of liquor over the be totally safe, since it’s patrolled by a hotel secu-
counter. But either someone has exposed my rity guard! With the Hearthside offering benefits
real-life housekeeping habits—which are, shall like that, how could anyone think of leaving? . . .
we say, mellow—or I am at the wrong end of Management at Jerry’s is generally calmer
some infallible ethnic equation: most, but by and more “professional” than at the Hearthside,
no means all, of the working housekeepers I with two exceptions. One is Joy, a plump, blowsy
see on my job searches are African Americans, woman in her early thirties, who once kindly de-
Spanish-speaking, or immigrants from the Cen- voted several minutes to instructing me in the
tral European post-Communist world, whereas correct one-handed method of carrying trays
servers are almost invariably white and mono- but whose moods change disconcertingly from
lingually English-speaking. When I finally get shift to shift and even within one. Then there’s
a positive response, I have been identified once B. J., a.k.a. B.J.-the-bitch, whose contribution is
again as server material. Jerry’s, which is part of a to stand by the kitchen counter and yell, “Nita,
well-known national family restaurant chain and your order’s up, move it!” or, “Barbara, didn’t
physically attached here to another budget hotel you see you’ve got another table out there? Come
chain, is ready to use me at once. The prospect is on, girl!” Among other things, she is hated for
both exciting and terrifying, because, with about having replaced the whipped-cream squirt cans
the same number of tables and counter seats, with big plastic whipped-cream-filled baggies
Jerry’s attracts three or four times the volume that have to be squeezed with both hands—be-
of customers as the gloomy old Hearthside. . . . cause, reportedly, she saw or thought she saw
I start out with the beautiful, heroic idea of employees trying to inhale the propellant gas
handling the two jobs at once, and for two days from the squirt cans, in the hope that it might be
I almost do it: the breakfast/lunch shift at Jerry’s, nitrous oxide. On my third night, she pulls me
which goes till 2:00, arriving at the Hearthside at aside abruptly and brings her face so close that
2:10, and attempting to hold out until 10:00. In it looks as if she’s planning to butt me with her
the ten minutes between jobs, I pick up a spicy forehead. But instead of saying, “You’re fired,”
chicken sandwich at the Wendy’s drive-through she says, “You’re doing fine.” The only trouble
window, gobble it down in the car, and change is I’m spending time chatting with customers:
from khaki slacks to black, from Hawaiian to “That’s how they’re getting you.” Furthermore
rust polo. There is a problem, though. When I am letting them “run me,” which means ha-
during the 3:00 to 4:00 p.m. dead time I finally rassment by sequential demands: you bring the
sit down to wrap silver, my flesh seems to bond ketchup and they decide they want extra Thou-
to the seat. I try to refuel with a purloined cup sand Island; you bring that and they announce
of soup, as I’ve seen Gail and Joan do dozens of they now need a side of fries; and so on into
times, but a manager catches me and hisses “No distraction. Finally she tells me not to take her
eating!” though there’s not a customer around to wrong. She tries to say things in a nice way, but
be offended by the sight of food making contact you get into a mode, you know, because every-
with a server’s lips. So I tell Gail I’m going to thing has to move so fast. . . .
quit, and she hugs me and says she might just I make friends, over time, with the other
follow me to Jerry’s herself. “girls” who work my shift: Nita, the tattooed
336 • Part V: Poverty and the Underclass

twenty-something who taunts us by going boyfriend seems to have a second job: Nita does
around saying brightly, “Have we started mak- something at a computer eight hours a day; an-
ing money yet?” Ellen, whose teenage son cooks other welds. Without the forty-five-minute com-
on the graveyard shift and who once managed mute, I can picture myself working two jobs and
a restaurant in Massachusetts but won’t try having the time to shower between them.
out for management here because she prefers So I take the $500 deposit I have coming
being a “common worker” and not “ordering from my landlord, the $400 I have earned to-
people around.” Easy-going fiftyish Lucy, with ward the next month’s rent, plus the $200 re-
the raucous laugh, who limps toward the end served for emergencies, and use the $1,100 to
of the shift because of something that has gone pay the rent and deposit on trailer number 46 in
wrong with her leg, the exact nature of which the Overseas Trailer Park, a mile from the cluster
cannot be determined without health insur- of budget hotels that constitute Key West’s ver-
ance. We talk about the usual girl things—men, sion of an industrial park. Number 46 is about
children, and the sinister allure of Jerry’s choco- eight feet in width and shaped like a barbell in-
late peanut-butter cream pie—though no one, side, with a narrow region—because of the sink
I notice, ever brings up anything potentially and the stove—separating the bedroom from
expensive, like shopping or movies. As at the what might optimistically be called the “living”
Hearthside, the only recreation ever referred to area, with its two-person table and half-sized
is partying, which requires little more than some couch. The bathroom is so small my knees rub
beer, a joint, and a few close friends. Still, no one against the shower stall when I sit on the toilet,
here is homeless, or cops to it anyway, thanks and you can’t just leap out of the bed, you have
usually to a working husband or boyfriend. All to climb down to the foot of it in order to find a
in all, we form a reliable mutual-support group: patch of floor space to stand on. Outside, I am
If one of us is feeling sick or overwhelmed, an- within a few yards of a liquor store, a bar that
other one will “bev” a table or even carry trays advertises “free beer tomorrow,” a convenience
for her. If one of us is off sneaking a cigarette store, and a Burger King—but no supermarket
or a pee, the others will do their best to con- or, alas, laundromat. By reputation, the Overseas
ceal her absence from the enforcers of corporate park is a nest of crime and crack, and I am hop-
rationality. . . . ing at least for some vibrant, multicultural street
I make the decision to move closer to Key life. But desolation rules night and day, except
West. First, because of the drive. Second and for a thin stream of pedestrian traffic heading
third, also because of the drive: gas is eating for their jobs at the Sheraton or 7-Eleven. There
up $4 to $5 a day, and although Jerry’s is as are not exactly people here but what amounts
high-volume as you can get, the tips average to canned labor, being preserved from the heat
only 10 percent, and not just for a newbie like between shifts. . . .
me. Between the base pay of $2.15 an hour and When my month-long plunge into poverty
the obligation to share tips with the busboys and is almost over, I finally land my dream job—
dishwashers, we’re averaging only about $7.50 housekeeping. I do this by walking into the per-
an hour. Then there is the $30 I had to spend sonnel office of the only place I figure I might
on the regulation tan slacks worn by Jerry’s serv- have some credibility, the hotel attached to Jer-
ers—a setback it could take weeks to absorb. (I ry’s, and confiding urgently that I have to have
had combed the town’s two downscale depart- a second job if I am to pay my rent and, no, it
ment stores hoping for something cheaper but couldn’t be front-desk clerk. “All right,” the per-
decided in the end that these marked-down sonnel lady fairly spits, “So it’s housekeeping,”
Dockers, originally $49, were more likely to and she marches me back to meet Maria, the
survive a daily washing.) Of my fellow serv- housekeeping manager, a tiny, frenetic Hispanic
ers, everyone who lacks a working husband or woman who greets me as “babe” and hands me
 Ehrenrich—Nickel and Dimed • 337

a pamphlet emphasizing the need for a positive meaning the things a worker has to do just so
attitude. The hours are nine in the morning till she’ll be ready to work again. . . .
whenever, the pay is $6.10 an hour, and there’s Then it comes, the perfect storm. Four of my
one week of vacation a year. I don’t have to ask tables fill up at once. Four tables is nothing for
about health insurance once I meet Carlotta, the me now, but only so long as they are obligingly
middle-aged African-American woman who will staggered. As I bev table 27, tables 25, 28, and
be training me. Carla, as she tells me to call her, 24 are watching enviously. As I bev 25, 24 glow-
is missing all of her top front teeth. ers because their bevs haven’t even been ordered.
On that first day of housekeeping and last Twenty-eight is four yuppyish types, meaning
day of my entire project—although I don’t yet everything on the side and agonizing instruc-
know it’s the last—Carla is in a foul mood. We tions as to the chicken Caesars. Twenty-five is a
have been given nineteen rooms to clean, most middle-aged black couple, who complain, with
of them “checkouts,” as opposed to “stay-overs,” some justice, that the iced tea isn’t fresh and the
that require the whole enchilada of bed-stripping, tabletop is sticky. But table 24 is the meteoro-
vacuuming, and bathroom-scrubbing. When one logical event of the century: ten British tourists
of the rooms that had been listed as a stay-over who seem to have made the decision to absorb
turns out to be a checkout, Carla calls Maria to the American experience entirely by mouth.
complain, but of course to no avail. “So make Here everyone has at least two drinks—iced
up the motherfucker,” Carla orders me, and I do tea and milk shake, Michelob and water (with
the beds while she sloshes around the bathroom. lemon slice, please)—and a huge promiscuous
For four hours without a break I strip and re- orgy of breakfast specials, mozz sticks, chicken
make beds, taking about four and a half minutes strips, quesadillas, burgers with cheese and with-
per queen-sized bed, which I could get down to out, sides of hash browns with cheddar, with
three if there were any reason to. We try to avoid onions, with gravy, seasoned fries, plain fries,
vacuuming by picking up the larger specks by banana splits. Poor Jesus (the cook)! Poor me!
hand, but often there is nothing to do but drag Because when I arrive with their first tray of
the monstrous vacuum cleaner—it weighs about food—after three prior trips just to refill bevs—
thirty pounds—off our cart and try to wrestle it Princess Di refuses to eat her chicken strips with
around the floor. Sometimes Carla hands me the her pancake-and-sausage special, since, as she
squirt bottle of “BAM” (an acronym for some- now reveals, the strips were meant to be an ap-
thing that begins, ominously, with “butyric”; petizer. Maybe the others would have accepted
the rest has been worn off the label) and lets me their meals, but Di, who is deep into her third
do the bathrooms. No service ethic challenges Michelob, insists that everything else go back
me here to new heights of performance. I just while they work on their “starters.” Meanwhile,
concentrate on removing the pubic hairs from the yuppies are waving me down for more decaf
the bathtubs, or at least the dark ones that I can and the black couple looks ready to summon the
see. . . . NAACP.
When I request permission to leave at about Much of what happened next is lost in the
3:30, another housekeeper warns me that no one fog of war. Jesus starts going under. The little
has so far succeeded in combining housekeeping printer on the counter in front of him is spew-
at the hotel with serving at Jerry’s: “Some kid did ing out orders faster than he can rip them off,
it once for five days, and you’re no kid.” With much less produce the meals. Even the invincible
that helpful information in mind, I rush back to Ellen is ashen from stress. I bring table 24 their
number 46, down four Advils, shower, stooping reheated main courses, which they immediately
to fit into the stall, and attempt to compose my- reject as either too cold or fossilized by the mi-
self for the oncoming shift. So much for what crowave. When I return to the kitchen with their
Marx termed the “reproduction of labor power,” trays (three trays in three trips), Joy confronts me
338 • Part V: Poverty and the Underclass

with arms akimbo: “What is this?” She means In one month, I had earned approximately
the food—the plates of rejected pancakes, hash $1,040 and spent $517 on food, gas, toiletries,
browns in assorted flavors, toasts, burgers, sau- laundry, phone, and utilities. If I had remained
sages, eggs. “Uh, scrambled with cheddar,” I in my $500 efficiency, I would have been able
try, “and that’s . . . ” “NO,” she screams in my to pay the rent and have $22 left over (which is
face. “Is it a traditional, a super-scramble, an $78 less than the cash I had in my pocket at the
eye-opener?” I pretend to study my check for a start of the month). During this time I bought
clue, but entropy has been up to its tricks, not no clothing except for the required slacks and
only on the plates but in my head, and I have to no prescription drugs or medical care (I did fi-
admit that the original order is beyond recon- nally buy some vitamin B to compensate for the
struction. “You don’t know an eye-opener from a lack of vegetables in my diet). Perhaps I could
traditional?” she demands in outrage. All I know, have saved a little on food if I had gotten to a
in fact, is that my legs have lost interest in the supermarket more often, instead of convenience
current venture and have announced their inten- stores, but it should be noted that I lost almost
tion to fold. I am saved by a yuppie (mercifully four pounds in four weeks, on a diet weighted
not one of mine) who chooses this moment to heavily toward burgers and fries.
charge into the kitchen to bellow that his food is How former welfare recipients and single
twenty-five minutes late. Joy screams at him to mothers will (and do) survive in the low-wage
get the hell out of her kitchen, please, and then workforce, I cannot imagine. Maybe they will
turns on Jesus in a fury, hurling an empty tray figure out how to condense their lives—includ-
across the room for emphasis. ing child-raising, laundry, romance, and meals—
I leave. I don’t walk out, I just leave. I don’t into the couple of hours between full-time jobs.
finish my side work or pick up my credit-card Maybe they will take up residence in their vehi-
tips, if any, at the cash register or, of course, ask cles, if they have one. All I know is that I couldn’t
Joy’s permission to go. And the surprising thing hold two jobs and I couldn’t make enough
is that you can walk out without permission, that money to live on with one. And I had advantages
the door opens, that the thick tropical night air unthinkable to many of the long-term poor—
parts to let me pass, that my car is still parked health, stamina, a working car, and no children
where I left it. There is no vindication in this to care for and support. . . .
exit, no fuck-you surge of relief, just an over- The thinking behind welfare reform was that
whelming, dank sense of failure pressing down even the humblest jobs are morally uplifting and
on me and the entire parking lot. I had gone psychologically buoying. In reality they are likely
into this venture in the spirit of science, to test a to be fraught with insult and stress. But I did dis-
mathematical proposition, but somewhere along cover one redeeming feature of the most abject
the line, in the tunnel vision imposed by long low-wage work—the camaraderie of people who
shifts and relentless concentration, it became a are, in almost all cases, far too smart and funny
test of myself, and clearly I have failed. . . . and caring for the work they do and the wages
When I moved out of the trailer park, I gave they’re paid. The hope, of course, is that some-
the key to number 46 to Gail and arranged for day these people will come to know what they’re
my deposit to be transferred to her. She told me worth, and take appropriate action.
that Joan is still living in her van and that Stu
had been fired from the Hearthside. . . .
 Edin, Nelson, and Reed—Low-Income Urban Fathers and the “Package Deal” of Family Life • 339

39 • Kathryn Edin, Timothy Nelson, and Joanna Miranda Reed

Low-Income Urban Fathers and the

“Package Deal” of Family Life

Economically disadvantaged fathers are far less than a third of such couples are still together by
likely to marry before having children than the time the child turns 5 (Center for Research
middle-class fathers are, and they have them on Child Wellbeing 2007). Low-income couples
far earlier (Nock 2007). When they do marry, who marry before having children are fragile as
they are more likely to divorce (Martin 2004). well—much more so than middle-class married
In the absence of a marital tie, the government couples are—but they still function as partners
assigns them financial obligations, which most for a considerable period of time (Martin 2004;
do not satisfy fully (U.S. Census Bureau 2007). McLanahan 2004).
Thus, such men’s fathering behavior attracts a We offer the reader a portrait of the romantic
good deal of attention from both scholars and partnerships of such men. [Our sample] is of a
policymakers. relatively large group of very economically dis-
Little attention is paid, however, to these advantaged white and black men (with earnings
men’s roles as romantic partners. Qualitative below the poverty line for a family of four in the
studies have been an exception to this trend formal economy over the prior year) who live
(i.e., Drake and Cayton 1945; Liebow 1967; in poor and struggling working-class neighbor-
Nelson, Clampet-Lundquist, and Edin 2002; hoods throughout the Philadelphia metropolitan
Furstenberg 2001). As these studies have repeat- area and have biological children, most of them
edly shown, economically disadvantaged men do outside of marriage. All were fathers of at least
engage in romantic relationships; this is the con- one minor child outside of a marital tie when we
text into which most of their children are born interviewed them.
(though some children are the product of non- By exploring in depth the texture of their
relationships, i.e., one-night stands) (Augustine, romantic lives and worldviews, we show that
Nelson, and Edin 2009). New survey research the function of the romantic tie for the fa-
reveals that fully eight in ten nonmarital children ther role departs radically from the traditional
now enter the world with a mother and father 1950s “package deal” conception of family life.
who describe themselves as “romantically in- Furstenberg and Cherlin (1991), and more re-
volved”; up to half of those parents live together, cently Nicholas Townsend (2002), point to the
and at least 70 percent of both mothers and fa- family behaviors of men who came of age in an
thers say there is at least a 50–50 chance they’ll earlier generation, arguing that for these men
marry each other. Yet it is also true that fewer fatherhood flowed through, and was contingent

Kathryn Edin, Timothy Nelson, and Joanna Miranda Reed. “Daddy, Baby; Momma, Maybe: Low Income
Urban Fathers and the ‘Package Deal’ of Family Life,” in Social Class and Changing Families in an Unequal
America, edited by Marcia Carlson and Paula England (Stanford: Stanford University Press, 2011), pp. 85–96,
102–107. Copyright © 2011 by the Board of Trustees of the Leland Stanford Jr. University. All rights reserved.
Used with the permission of Stanford University Press, www.sup.org.
340 • Part V: Poverty and the Underclass

upon, men’s relationship with the children’s formal economy. In the course of our conversa-
mother. Furstenberg and Cherlin coined the tions with these men, we collected detailed life
term “package deal” to explain the very low rates histories. Our data come from these life histories
of father involvement among the divorced fa- (e.g., “Tell me the whole story of your relation-
thers they observed. In this view, the tie between ship from the time you first met until now”), and
the mother and father is central and serves to from their answers to questions formulated to
bind men to their obligations to their children— capture their worldview (e.g., “What should fa-
obligations they would otherwise ignore (e.g., in thers do for their children?” “In your view, what
the case of divorce). makes for a good father?”).
Two of us (Edin and Nelson) and a multira-
cial team of graduate students spent seven years
observing and interviewing low-income fathers Findings
residing in high-poverty neighborhoods in Phil- In PCFS, we asked each father to tell us “the
adelphia and its poorest inner suburb, Camden, whole story” of how he got together with the
New Jersey. mothers of each of his children, and how these
For the men we have studied, the father-child relationships developed over time. Typically, the
relationship is typically what is viewed as cen- pre-pregnancy narrative was noticeably succinct:
tral, and is what binds men to couple rela- the couple met, began to “affiliate,” and then
tionships—relationships that might not have “came up pregnant.” Men seldom even men-
otherwise formed or been maintained (see also tioned, much less discussed, any special quali-
Edin et al. 2007; Reed 2007). This is not to say ties of their partners or any common tastes or
that such men are child-centric to the degree values that drew the two together. Usually, the
that mothers are (Edin and Kefalas 2005), only woman lived on his block, hung out on his cor-
that in the realm of family relations, “Daddy, ner, worked at the same job, was a friend of his
baby; momma, maybe” is a fair representation sister, or the girlfriend of a friend, and she was
of the worldview of many of the fathers we simply willing to “affiliate” with him. Hanging
studied. out on the stoop, an occasional outing to a bar
or a club, a window-shopping trip to a hot venue
such as the downtown Gallery or the popular
The Philadelphia/Camden Fathers South Street strip, and fantasizing about shared
Study (PCFS): Method children is what constituted romance (see also
We began the Philadelphia/Camden Fathers Townsend 2002: 42).
Study with two and a half years of participant In the case of couples with a first birth to-
observation in one of the eight low-income gether, the length of the courtship before con-
communities (census tract clusters where at least ception was usually exceedingly brief—typically
20 percent of the population lived in poverty well under a year. Only rarely did such couples
in 1990) selected for the study. Based on this “fall in love,” get engaged, or get married before
fieldwork, which began in 1995, we constructed conceiving a child together, though some did so
an interview guide that we administered in later. Instead, they meet, they “associate,” “affili-
systematic, repeated, in-depth interviews with ate,” “communicate,” begin to “kick it,” “talk to
110 white and African American men between each other,” “get with each other,” or “end up
1997 and 2002. We sampled equal numbers of together.” Then, “one thing leads to another.”
blacks and non-Hispanic whites. To offer more Planned pregnancies were rare, yet the contra-
of a life-course view, roughly half of our fathers ceptive practices that couples usually engaged in
were under 30 and the rest were older. During initially seldom continued for long (see also Edin
the prior year, the earnings of all of them were and Kefalas 2005; Edin et al. 2007; Augustine,
below the poverty line for a family of four in the Nelson, and Edin 2009). Then the inevitable
 Edin, Nelson, and Reed—Low-Income Urban Fathers and the “Package Deal” of Family Life • 341

occurred: the woman “comes up pregnant” (see recalled that they did very much want children
also Davis, Gardner, and Gardner 1941: 127). and fairly soon, even if not right then. Yet, the
John, a 24-year-old white father of one, de- partnering process was far more haphazard than
scribed the sequencing of his relationship with discriminating.
his child’s mother in this way: “Actually she was Bruce was a white father of 2-year-old twins.
dating a friend of mine and somehow . . . she At 42, Bruce met a “new girl,” Debbie. He didn’t
wanted me. . . . Eventually, I just got stuck with use protection because “Every time I had any
her for a little while.” John didn’t feel that he had kind of relationship there is no babies born so I
found the ideal match—he “got stuck” with his didn’t believe in safe sex. Next thing I knew, this
baby’s mother “for a little while.” No language of girl Debbie, she was pregnant.” Debbie made the
love or even attraction (except her attraction to announcement that she was seven weeks preg-
him) entered into this narrative, although there nant after the two had been together for only
may well have been attraction involved. Nor four months. Bruce told her, ‘“I am shooting
does the phrase “a little while” indicate much blanks, you can’t be pregnant! . . .’ Then we went
commitment. for a DNA test and that was when she found out
Thirty-nine-year-old Amin, a black father of that I was the father and she was the mother.”
two, described the development of his relation- Here a “family” was formed through a pregnancy
ship with his youngest son’s mother, a coworker brought to term in a relationship that was nei-
in the dietary department of a local hospital, ther casual nor serious. Yet for Bruce, and most
in this way: “She was attractive to me when I other men in our study, this was not viewed as
first saw her and I made my approach and we a problem.
begin to socialize and communicate and then Children, while usually desired, were only
from there we began to affiliate at some point rarely explicitly planned, according to the nar-
and we became intimate and our son was born.” ratives of men in PCFS (see Augustine, Nelson,
As Amin told it, attraction, affiliation, and inti- and Edin 2009). Yet contraception and other
macy quite naturally—and inevitably—led to a attempts to avoid pregnancy faded quickly as
son being born. the couple began “affiliating.” Once she became
Despite the vague and bureaucratic language pregnant and decided to take the pregnancy
often used to describe these relationships (e.g., to term (this decision is generally ceded to the
“affiliation”), almost no father had much trouble woman), the bond between the two typically
pinpointing when his relationship with his ba- coalesced into more of a “relationship,” though
by’s mother began; they knew the point at which often in dramatic fits and starts (see also Edin
they got “together” with their baby’s mother and Kefalas 2005).
(though a small number of pregnancies do occur David, a black 30-year-old, was the father
outside of relationships). Being together gener- of five children by three women. In the months
ally means that the couple is spending regular just before conceiving his fifth child, he was both
time together and defines the relationship as ‘‘with” Deborah and “seeing” Kathy on the side.
something more than a casual encounter or a He went to Kathy whenever he and Deborah
one-night stand. argued (though Deborah didn’t know about
The verbs “affiliate,” “associate,” and “get Kathy). Which woman he should choose became
with” suggest a bond that is more than a casual a dilemma. However, when Deborah ended up
liaison, but not exactly a boyfriend/girlfriend re- pregnant, he decided to “do what was right” and
lationship. Few of these men were consciously chose her:
“courting” or searching for a life partner. Indeed,
there is little evidence that they were even at- [When I was first with Deborah] I had a girl-
tempting to discriminate much based on who friend on the side too. Kathy. She’s somebody
would be a suitable mother for their child. Many that I met at an NA meeting. We got close
342 • Part V: Poverty and the Underclass

and we were helping each other [with our ad- But the arrival of a child introduces a sharp con-
dictions]. One thing led to another, and we tradiction. On the one hand, fathers’ narratives
got intimate. . . . Me and Deborah would get offer evidence that they are making some effort
into an argument, she’d tell me to leave, I’d go to sustain a relationship with their babies’ moth-
stay with Kathy. (So how did you end it with ers—often motivated by the desire to live with
Kathy?) Deborah got pregnant, and I had to or be intensely involved with the children. But
do what was right, stand by Deborah. they also recognize, at least in part, that they are
trying live out the dream of being part of a real
For Jack, a 33-year-old white father of two, “family unit” against almost impossible odds.
his babies’ mother was just one of several “girls” Thus, while working to solidify these relation-
he met on a weekend home from college. She ships, they are also often deeply fatalistic about
was already married to someone else, but left her their chances of staying together over the long
husband for Jack that very day. Three months term. This fatalism is fueled by men’s fears about
later she was pregnant, yet Jack saw nothing their ongoing ability to provide and their utter
remarkable about the process by which he first conviction that their babies’ mothers will leave
became a father. them if they fail, often rendered men’s relational
efforts half-hearted.
[After high school] I went to college. . . . Most fathers [in our sample] emphatically
My grades weren’t great but I was getting said that their relationship with their children
through. I was going back home every other ought not be contingent on their relationship
weekend. . . . Met some girls. In turn, met with their child’s mother—and outright rejected
my [baby’s mother]. Shortly afterwards, she the package deal. Yet they nonetheless realized
became pregnant, so I quit school, got a job. that due to normative and legal practices govern-
ing the custody of nonmarital children, as well
Kahlid, a 28-year-old black father of one, said as their own limited economic prospects, their
that it was just a few months into the relation- relationship to the mother was their conduit to
ship with his child’s mother that the pair con- the child.
ceived. Despite the fact that the outside observer Across the income distribution, most cou-
might read his narrative as a classic example of ples now stay together, at least in part, because
putting the cart before the horse, Kahlid, like they share common interests and values. Owing
Jack, viewed the sequence of events leading up to the lightning speed of the courtship period,
to the pregnancy as unexceptional. disadvantaged couples who bear children to-
gether often find they have little in common, as
(How long were you and your girlfriend to- we show below. But childbirth offers a vital new
gether before she got pregnant?) Six, seven shared interest—a child—and this often brings
months. (What went through your mind?) I some measure of emotional closeness. Consider
was happy! I came [over after] work and she the story of 38-year-old Tony, father of one.
said that she was getting symptoms and I was
like, “What you talking about?” Morning (How do you think the birth of Lyssa affected
sickness, throwing up, like this. So she went your relationship?) I think in the beginning it
to the hospital. She took the test and she says brought us a lot closer together. (Why would
she was pregnant . . . and I was excited. It was it bring you closer together?) It just—being
my first child. And she was going to keep the with this little baby That’s just a part of both
baby, and I was happy. I knew I had to keep a of us—it was amazing.
job and take care of my responsibilities.
Bucket, a black father of two children by the
The glories of the delivery room are the high same woman, described how the birth of his first
point of many fathers’ life histories in PCFS. child made things even a “little better.”
 Edin, Nelson, and Reed—Low-Income Urban Fathers and the “Package Deal” of Family Life • 343

(How’d you all get along during the rest of the committed to really being with her” (see also
pregnancy?) We had a good time man. While Waller 2008). Reading between the lines of his
she was pregnant, I couldn’t go nowhere. narrative, we can assume that spotty employ-
Shoot! She wanted me to do this, wanted ment and drug use were probably also a cause.
me to do that. I was like a puppy anyway. After their breakup, Donald’s ex-partner went
I waited on her. I did certain things that she on to have two additional children by two other
wanted me to do. I was glad man. (How did men and had just married a third man, who was
the birth of your child affect your relationship trying to play the role of father to his 14-year-old
with her?) It didn’t affect it at all. See like, daughter while her mother was trying to push
when the baby was born and we had a baby, it Donald out of their lives. Listen to the deep cyn-
seemed like things got a little better and stuff. icism in Donald’s view of women:

Yet men like Bucket have powerful reasons not Yeah, their whole thing is, “What can you give
to invest in the relationship as well. The ethno- me?” and “What can I get from you and how
graphic literature typically focuses on how much fast can I get it?” More or less money-wise.
women mistrust men. What is less well known is There is no commitment to a real relation-
how little men trust women (for exceptions, see ship. Particularly in black women, their whole
Waller 2008; Hannertz 1969: 100–102; Liebow goal is let me see how much I can get for it and
1967: 137–60; Rainwater 1970: 209). The chief how good I can look and you know. . . . To
source of power (outside of physical violence) in be honest, yeah, I do [have a theory], a lot of
these relationships is control over the child. In them are caught up in, “I want to look good.”
the context of a nonmarital birth, women have “I want to be independent.” “I don’t want to
much greater power in this regard. Because men be dependent upon you.” “I want to be able
view children as their most precious resource to be with you, but not like that.” There is
(Nelson and Edin, forthcoming), they are, on no source of commitment in a relationship.
the one hand, eager to hang on, and expect to [It is not] just men [that aren’t committed],
be actively involved in parenting the child. Yet but anybody [nowadays]. In a relationship it
they are often very apprehensive about their is about, “We are in it 50–50 commitment,”
ability to satisfy the economic demands they also and so forth. They are not here. Their thing
know their children’s mothers will place on them is, “When things go bad, I am out of here.”
over time, and most are convinced that a wom- To me, that was my experience and I am not
an’s love for a man will come to an abrupt halt taking it [anymore].
the minute he fails to provide (see Drake and
Cayton 1945: 564–99; Rainwater 1970: 216; Amin was similarly convinced that money
Furstenberg 2011). was necessary to sustain a relationship and that
Men on the economic edge, even those in “situations”—such as unemployment—would
multi-year partnerships with several children, almost certainly bring a relationship to an end.
often obsessed about the younger guy with the Because he lacked confidence that he could re-
nicer car who had a better job and might turn main stably employed, Amin felt that he couldn’t
their partner’s head. Most were convinced that, even contemplate marriage, and worried he
for women, “there is no source of commitment in might not even be able to sustain a long-term
a relationship.” Donald was certain that a woman relationship.
would dump any man if he failed to provide.
This black 37-year-old found himself always When a woman is a woman and she accepts a
“feuding and fightin’” with his child’s mother man for who he is and not necessarily what he
during their eight-year on-again-off-again rela- does and doesn’t have, then she will stand by
tionship, which he blamed on the fact that she a man regardless. But when a woman starts to
was too controlling, plus the fact that “I was not allow the fact that he is not bringing as much
344 • Part V: Poverty and the Underclass

money into the home affect her relationship collect unemployment because I wasn’t on the
and her attitude towards him then that is a books. . . . She was already pregnant with the
problem. . . . And that is another reason why second baby and that is when I couldn’t deal
I have not considered marriage because . . . with her anymore and we couldn’t get along.
there is not too many women out there that is
really ready themselves to honestly fulfill the
covenant of the words that you recite when Conclusions
you are at the podium. Once mothers learned they were pregnant, shar-
ing the news with the father-to-be and deciding
Jeff, a black 47-year-old father of two, also what to do about it were important defining
believed a woman’s love was contingent on moments for the couple. The pregnancy often
money. He also saw this mercenary strain in his prompted many of the men [in our sample] to
daughter. think of themselves as a couple for the first time,
or to reconsider a partner who had fallen out of
[My daughter’s mother taught me that] love favor. The decision whether to have the baby was
is like running water. It turns off and on. I re- usually left to the mother, with fathers typically
ally believe that behind the fact that they can agreeing to “support what she wants to do.”
love you when you’re doing, but when you Pregnancy is a challenging time for most
don’t do, they don’t love. . . . My daughter fathers’ romantic relationships. Economic pres-
had told me on numerous occasions that this sures were great, such troubles were likely to
[or that] individual does more for her than I break the couple up before the birth. Many
do and I felt hurt. Regardless if this person couples’ troubles momentarily disappeared after
is doing something for you or not, he can’t the “magic moment” of their child’s birth, as the
fill my shoes. I’m still your father regardless. fact of a shared child brought some measure of
If I give you a million dollars or I give you a cohesion. Fathers who were still romantically
penny, I’m still your father. involved, even if things were rough during the
pregnancy, saw the baby as sufficient reason
Tom, a white father of three, believed that to renew their efforts to stay together with the
when a man is “down and out” a woman could mother.
easily be attracted away by a guy who was doing While entry into a “real relationship,” some-
better or “has a nicer car.” In his case, a slow- times symbolized by the move to cohabitation,
down at work while the mother of his younger may have indicated a growing level of commit-
two children was pregnant the second time led ment to the romantic partner, it also reflected a
to unbearable conflict. father’s wish to be an involved parent. Usually,
among fathers, the impulse to parent is at least
I was working every day and I was paying the as strong as, if not stronger than, the desire to
bills. I did everything that I possibly could. I remain partnered for the sake of the relation-
came home, I took care of the kids, I would ship. Work done by Edin and Kefalas (2005) on
put the kids to bed, just so we would have a companion sample of mothers in the PCFS
some (alone) time. . . . Her father had got me communities shows that mothers share the desire
a job doing the roofing. . . . We saved money for fathers to be involved, but view his involve-
and we moved into the house and things were ment as a desirable, yet optional, complement to
good and we were splitting the rent. Then I their mothering activities.
lost the job and I couldn’t afford them pay- In sum, for this group of mostly disadvan-
ments anymore. . . . Yeah, it got real slow and taged men parenting children outside of mar-
the winter time came and there was just no riage, multiple and serious problems such as
work, so I didn’t know what to do. I couldn’t infidelity, sexual mistrust, substance abuse,
 Edin, Nelson, and Reed—Low-Income Urban Fathers and the “Package Deal” of Family Life • 345

with Children: The Unfolding Lives of New Un-

domestic violence, criminal activity, and incar-
married Urban Parents, edited by Paula England
ceration abound in their narratives about their and Kathryn Edin. New York: Russell Sage
experiences as romantic partners. Despite their Foundation.
desire to form strong relationships with their Edin, Kathryn, and Maria B. Kefalas. 2005. Promises
children, men found that in the aftermath of the I Can Keep: Why Poor Women Put Motherhood be-
breakup, staying in regular contact with their fore Marriage. Berkeley: University of California
progeny was surprisingly difficult. Press.
What is evident from the narratives presented Grail, Timothy S. 2007. “Custodial Mothers and Fa-
thers and Their Child Support: 2005.” Current
here is that the haphazard nature of the way that
Population Reports P60–234. Washington, DC:
these families are formed places enormous pres-
U.S. Census Bureau.
sure on already disadvantaged young people ———. 2011. “The Recent Transformation of the
who don’t know each other very well, but who American Family: Witnessing and Exploring So-
nonetheless attempt to form “real” relationships, cial Change.” In Families in an Unequal Society,
sometimes via cohabitation. Although preg- edited by Marcia Carlson and Paula England.
nancy and birth do typically prompt relationship Palo Alto, CA: Stanford University Press.
investment among men, their precarious eco- Furstenberg, Frank F., Jr., and Andrew B. Cherlin.
nomic situations and their mistrust of women 1991. Divided Families: What Happens to Chil-
prompt disinvestment. Men’s behavior is usually dren When Parents Part. Cambridge, MA: Har-
vard University Press.
the most proximate cause of a relationship’s de-
Furstenberg, Frank F., Jr. 2001. “The Fading Dream:
mise. Another significant factor is the ambigu- Prospects for Marriage in the Inner City.” In
ous nature of the relationships themselves, where Problem of the Century, edited by Elijah Ander-
expectations are seldom revealed until they are son and Douglas Massey, pp. 224–47. New York:
violated. And though men may have believed Russell Sage Foundation.
they could have a direct relationship with their Martin, Steven P. 2004. “Growing Evidence for a Di-
children with or without the mothers—“daddy vorce Divide? Education and Marital Dissolution
baby, momma maybe”-—accomplishing this feat Rates in the United States since the 1970’s.” New
is far more difficult than most had imagined. York, NY: Russell Sage Foundation Working Pa-
pers: Series on Social Dimensions’ of Inequality.
McLanahan, Sara. 2004. “Diverging Destinies: How
REFERENCES
Children Fare under the Second Demographic
Augustine, Jennifer March, Timothy Nelson, and
Transition.” Demography 41(4): 607–27.
Kathryn Edin. 2009. “Why Do Poor Men Have
Nelson, Timothy, and Kathryn Edin. Forthcoming.
Children?” Annals of the American Academy of
Fragile Fatherhood: What Being a Daddy Means in
Political and Social Science 624 (July): 99–117.
the Lives of Low-Income Men. New York: Russell
Center for Research on Child Wellbeing. 2007.
Sage Foundation.
“Parents’ Relationship Status Five Years after a
Nelson, Timothy J., Susan Clampet-Lundquist,
Non Marital Birth.” Princeton, NJ: Center for
and Kathryn Edin. 2002. “Sustaining Fragile
research on Child Wellbeing, Research Brief 39
Fatherhood: How Low-Income, Non-Custodial
(August).
Fathers in Philadelphia Talk about Their Fam-
Davis, Allison, Burleigh B. Gardner, and Mary B.
ilies.” In The Handbook of Father Involvement:
Gardner. 1941. Deep South: A Social Anthropolog-
Multidisciplinary Perspectives, edited by Catherine
ical Study of Caste and Class. Chicago: University
Tamis-LeMonda and Natasha Cabrera, pp. 525–
of Chicago Press.
53. Mahwah, NJ: Lawrence Erlbaum Associates.
Drake, St. Clair, and Horace R. Cayton. 1945. Black
Nock, Stephen. 2007. “Marital and Unmarried
Metropolis: A Study of Negro Life in a Northern
Births to Men.” Department of Health and
City. New York: Harcourt Brace.
Human Services Publication (PHS) 2006–1978.
Edin, Kathryn, Paula England, Emily Fitzsimmons
Hyattsville, MD: U.S. Department of Health
Shafer, and Joanna Reed. 2007. “Forging Frag-
and Human Services.
ile Families: Are the Pregnancies Planned, Un-
Reed, Joanna. 2007 “Anatomy of the Break-Up:
planned, or In-Between?” In Unmarried Couples
How and Why Do Unmarried Parents Break
346 • Part V: Poverty and the Underclass

Up?” In Unmarried Couples with Children, edited Waller, Maureen R. 2002. My Baby’s Father: Unmar-
by Paula England and Kathryn Edin. New York: ried Parents and Paternal Responsibility. Ithaca,
Russell Sage Foundation. NY: Cornell University Press.
Townsend, Nicholas W. 2002. The Package Deal: ———. 2008. “How Do Disadvantaged Parents
Marriage, Work and Fatherhood in Men’s Lives. View Tensions in Their Relationships? Insights
Philadelphia: Temple University Press. for Relationship Longevity among At-Risk Cou-
ples.” Family Relations 57(2): 128–43.
 Wilson—Being Poor, Black, and American • 347

Poverty and the Economy

40 • William Julius Wilson

Being Poor, Black, and American

Through the second half of the 1990s and in tragic results. And through Katrina, the nation’s
to the early years of the 2lst century, public at- attention became riveted on these poor urban
tention to the plight of poor black Americans neighborhoods.
seemed to wane. There was scant media attention If television cameras had focused on the
to the problem of concentrated urban poverty urban poor in New Orleans, or in any inner-city
(neighborhoods that fall beneath the federally ghetto, before Katrina, I believe the initial re-
designated poverty line), little or no discussion action to descriptions of poverty and poverty
of inner-city challenges by mainstream political concentration would have been unsympathetic.
leaders, and even an apparent quiescence on the Public opinion polls in the United States rou-
part of ghetto residents themselves. This was tinely reflect the notion that people are poor
dramatically different from the 1960s, when and jobless because of their own shortcomings
the transition from legal segregation to a more or inadequacies, in other words, few people
racially open society was punctuated by social would have reflected on how the larger forces in
unrest that sometimes expressed itself in violent society—including segregation, discrimination,
terms, as seen in the riots that followed the assas- a lack of economic opportunity, and failing
sination of Dr. Martin Luther King, Jr. public schools—adversely affect the inner-city
But in 2005, Hurricane Katrina exposed con- poor. However, because Katrina was clearly a
centrated poverty in New Orleans. When televi- natural disaster that was beyond the control of
sion cameras focused on the flooding, the people the inner-city poor, Americans were much more
trapped in houses and apartments, and the vast sympathetic. In a sense, Katrina turned out to
devastation, many Americans were shocked to be something of a cruel natural experiment,
see the squalid living conditions of the poor. Of wherein better-off Americans could readily see
course, the devastation of Katrina was broadly the effects of racial isolation and chronic eco-
visited upon the residents of New Orleans, black nomic subordination.
and white, rich and poor, property owner and Despite the lack of national public aware-
public housing tenant alike. But while many ness of the problems of the urban poor prior to
residents were able to flee, the very poor, lack- Katrina, social scientists have rightly devoted
ing automobiles or money for transportation considerable attention to concentrated poverty,
and lodging, stayed to wait out the storm with because it magnifies the problems associated

William Julius Wilson. “Being Poor, Black, and American,” American Educator (Spring 2011), pp. 10–23,
46. Reprinted with permission from the Spring 2011 issue of American Educator, the quarterly journal of the
American Federation of Teachers, AFL-CIO.
348 • Part V: Poverty and the Underclass

with poverty in general: joblessness, crime, de- Americans now associate with black neighbor-
linquency, drug trafficking, broken families, and hoods. This action was clearly motivated by ra-
dysfunctional schools. Neighborhoods of highly cial bias, and it was not until the 1960s that the
concentrated poverty are seen as dangerous, and FHA discontinued mortgage restrictions based
therefore they become isolated, socially and eco- on the racial composition of the neighborhood.2
nomically, as people go out of their way to avoid Subsequent policy decisions worked to trap
them.1 blacks in these increasingly unattractive inner
In this chapter, I provide a political, eco- cities. Beginning in the 1950s, the suburbaniza-
nomic, and cultural framework for under- tion of the middle class, already under way with
standing the emergence and persistence of government-subsidized loans to veterans, was
concentrated urban poverty. I pay particular at- aided further by federal transportation and high-
tention to poor inner-city black neighborhoods, way policies that included the building of free-
which have the highest levels of concentrated way networks through the hearts of many cities,
poverty. I conclude this article by suggesting a which had a devastating impact on the neighbor-
new agenda for America’s ghetto poor, based on hoods of black Americans. These developments
the analysis I put forth in the following sections. not only spurred relocation from the cities to the
suburbs among better-off residents, the freeways
themselves also “created barriers between the sec-
Political Forces tions of the cities, walling off poor and minority
Since 1934, with the establishment of the Fed- neighborhoods from central business districts.”3
eral Housing Administration (FHA), a program For instance, a number of studies have revealed
necessitated by the massive mortgage foreclo- how Richard J. Daley, the former mayor of Chi-
sures during the Great Depression, the U.S. cago, used the Interstate Highway Act of 1956
government has sought to enable citizens to be- to route expressways through impoverished Af-
come homeowners by underwriting mortgages. rican American neighborhoods, resulting in even
In the years following World War II, however, greater segregation and isolation.4 A lasting leg-
the federal government contributed to the early acy of that policy is the 14-lane Dan Ryan Ex-
decay of inner-city neighborhoods by withhold- pressway, which created a barrier between black
ing mortgage capital and making it difficult for and white neighborhoods.5
these areas to retain or attract families who were At the same time, government policies such
able to purchase their own homes. The FHA as mortgages for veterans and mortgage interest
selectively administered the mortgage program tax exemptions for developers enabled the quick,
by formalizing a process that excluded certain cheap production of massive amounts of tract
urban neighborhoods using empirical data that housing6 and drew middle-class whites into the
suggested a probable loss of investment in these suburbs.7 The homes in these neighborhoods
areas. “Redlining,” as it came to be known, was were manufactured on a large scale, using an
assessed largely on racial composition. assembly line model of production, and were ar-
Although many neighborhoods with a con- ranged in carefully engineered suburban neigh-
siderable number of European immigrants were borhoods that included many public amenities,
redlined, virtually all black neighborhoods were such as shopping centers and space for public
excluded. Homebuyers hoping to purchase a schools. These neighborhoods represented an
home in a redlined neighborhood were uni- ideal alternative for people who were seeking to
versally denied mortgages, regardless of their escape cramped city apartments, and were often
financial qualifications. This severely restricted touted as “utopian communities” that enabled
opportunities for building or even maintaining people to live out the “suburban dream.”
quality housing in the inner city, which in many Explicit racial policies in the suburbs rein-
ways set the stage for the urban blight that many forced this segregation by allowing suburbs to
 Wilson—Being Poor, Black, and American • 349

Figure 40.1. Child Poverty Rate by Race/Ethnicity

Source: A Call for Change, Figure 1.9 (Data from Kidscount; Population Reference Bureau, Analysis of Data from the U.S.
Census Bureau, 2008 American Community Survey).

separate their financial resources and municipal inner-city racial minorities to access these areas
budgets from those of the cities. In the 19th and because these practices were effectively used to
early 20th centuries, strong municipal services screen out residents on the basis of race.
in cities were very attractive to residents of small As separate political jurisdictions, suburbs
towns and suburbs; as a result, cities tended also exercised a great deal of autonomy through
to annex suburbs and surrounding areas. But covenants and deed restrictions. In the face of
the relations between cities and suburbs in the mounting pressure for integration in the 1960s,
United States began to change following the “suburbs chose to diversify by race rather than
Great Depression; the century-long influx of class. They retained zoning and other restrictions
poor migrants who required expensive services that allowed only affluent blacks (and in some
and paid relatively little in taxes could no longer instances Jews) to enter, thereby intensifying the
be profitably absorbed into the city economy. concentration and isolation of the urban poor.”8
Annexation largely ended in the mid-20th cen- Although these policies clearly had racial conno-
tury as suburbs began to successfully resist in- tations, they also reflected class bias and helped
corporation. Suburban communities also drew reinforce the exodus of white working-class and
tighter boundaries through the use of zoning middle-class families from urban neighborhoods
laws, discriminatory land-use controls, and and the growing segregation of low-income
site selection practices that made it difficult for blacks in inner-city neighborhoods.
350 • Part V: Poverty and the Underclass

Federal public housing policy contributed to the construction of public housing. And the
the gradual growth of segregated black ghettos as federal government acquiesced to opposition to
well. The federal public housing program’s poli- the construction of public housing in the neigh-
cies evolved in two stages that represented two dis- borhoods of organized white groups in the city.
tinct styles. The Wagner-Steagall Housing Act of Thus, units were overwhelmingly concentrated
1937 initiated the first stage. Concerned that the in the over-crowded and deteriorating inner-city
construction of public housing might depress pri- ghettos—the poorest and least-powerful sections
vate rent levels, groups such as the U.S. Building of cities and metropolitan areas. In short, public
and Loan League and the National Association of housing became a federally funded institution
Real Estate Boards successfully lobbied Congress that isolated families by race and class, resulting
to require, by law, that for each new unit of public in high concentrations of poor black families in
housing erected, one “unsafe or unsanitary” unit inner-city ghettos.10
of public housing must be destroyed. In the last quarter of the 20th century, one of
The early years of the public housing pro- the most significant changes in these neighbor-
gram produced positive results. Initially, the pro- hoods was the outmigration of middle-income
gram mainly served intact families temporarily blacks. Before the 1970s, African American fam-
displaced by the Depression or in need of hous- ilies faced extremely strong barriers when they
ing after the end of World War II. For many of considered moving into white neighborhoods.
these families, public housing was the first step Not only did many experience overt discrimina-
on the road toward economic recovery. Their tion in the housing market, some were violently
stays in the projects were relatively brief because attacked. Although even today fair-housing
they were able to accumulate sufficient economic audits continue to reveal the existence of dis-
resources to move on to private housing. crimination in the housing market, fair-housing
The passage of the Housing Act of 1949 legislation has reduced the strength of these bar-
marked the beginning of the second policy stage. riers. At the same time, middle-income African
It instituted and funded the urban renewal pro- Americans have increased their efforts to move
gram, designed to eradicate urban slums, and from areas with concentrated black poverty to
therefore was seemingly nonracial. However, the more desirable neighborhoods throughout met-
public housing that it created “was now meant ropolitan areas, including white neighborhoods.11
to collect the ghetto residents left homeless by In addition, beginning in 1980, when Ron-
the urban renewal bulldozers.”9 A new, lower ald Reagan became president, sharp spending
income ceiling for public housing residency cuts in direct aid to cities dramatically reduced
was established by the Federal Public Housing budgets for general revenue sharing (unrestricted
Authority, and families with incomes above that funds that can be used for any purpose), urban
ceiling were evicted, thereby restricting access to mass transit, economic development assistance,
public housing to only the most economically urban development action grants, social service
disadvantaged segments of the population. block grants, local public works, compensatory
This change in federal housing policy coin- education, public service jobs, and job training.
cided with the Second Great Migration4 of Afri- Many of these programs were designed to help
can Americans from the rural South to the cities disadvantaged individuals gain some traction
of the Northeast and Midwest, which lasted 30 in attaining financial security.12 It is telling that
years—from 1940 to 1970. As the black urban the federal contribution was 17.5 percent of the
population in the North grew, pressure mounted total city budgets in 1977, but only 5.4 percent
in white communities to keep blacks out. Sub- by 2000.13 These cuts were particularly acute for
urban communities, with their restrictive cove- older cities in the East and Midwest that largely
nants and special zoning laws, refused to permit depended on federal and state aid to fund social
 Wilson—Being Poor, Black, and American • 351

Figure 40.2. Gaps in Mathematics and Reading Among Forth Graders (2009)

Note: Large cities data include students from all cities in the nation with populations of 250,000 or more. National data
include students attending public schools across the nation.
Source: A Call for Change, Figures 2.5 and 2.23 (Data from the U.S. Department of Education, National Assessment of
Educational Progress 2009 Reading and Mathematics Assessments).

services for their poor population and to main- social services as well as high levels of neighbor-
tain aging infrastructure. hood joblessness—have reinforced the percep-
The decline in federal support for cities since tion that cities are dangerous places to live and
1980 coincided with an increase in the immigra- have perpetuated the exodus of working- and
tion of people from poorer countries—mainly middle-class residents. Thus, while poverty and
low-skilled workers from Mexico—and whites joblessness, and the social problems they gener-
steadily moving to the suburbs. With minorities ate, remain prominent in ghetto neighborhoods,
displacing whites as a growing share of the pop- many cities have fewer and fewer resources with
ulation, the implications for the urban tax base which to combat them.
were profound. . . . Finally, policymakers have indirectly contrib-
This financial crisis left many cities uted to concentrated poverty in inner-city neigh-
ill-equipped to handle three devastating public borhoods with decisions that have decreased the
health problems that emerged in the 1980s and attractiveness of low-paying jobs and accelerated
disproportionately affected areas of concentrated the relative decline in the wages of low-income
poverty: first, the prevalence of drug trafficking workers. In particular, in the absence of an effec-
and associated violent crime; second, the ac- tive labor market policy, policymakers have tol-
quired immunodeficiency syndrome (AIDS) erated industry practices that undermine worker
epidemic and its escalating public health costs; security—including the erosion of benefits and
and third, the rise in the homeless population, the rise of involuntary part-time employment.
including not only individuals, but entire fam- In sum, federal government policies, even
ilies as well.14 A number of fiscally strapped those that are not explicitly racial, have had a
cities have watched helplessly as these prob- profound impact on inner-city neighborhoods.
lems—aggravated by the reduction of citywide These impacts have been felt in many cities
352 • Part V: Poverty and the Underclass

across the country, but they perhaps have been yet job growth in these older central cities did
felt more in the older central cities of the Mid- not exceed 3 percent.19
west and Northeast—the traditional Rust Belt— With the decline in manufacturing employ-
where depopulated, high-poverty areas have ment in many of the nation’s central cities, most
experienced even greater problems. of the jobs for lower-skilled workers are now
in retail and service industries (for example,
store cashiers, customer service representatives,
Economic Forces fast-food servers, and custodial work). Whereas
Older urban areas were once the hubs of eco- jobs in manufacturing industries typically were
nomic growth and activity, and were therefore unionized, relatively stable, and carried higher
major destinations for people in search of eco- wages, those for workers with low to modest lev-
nomic opportunity. However, the economies els of education in the retail and service industries
of many of these cities have since been eroded tend to provide lower wages, be unstable, and
by complex economic transformations and lack the benefits and worker protections—such
shifting patterns in metropolitan development. as workers’ health insurance, medical leave, re-
These economic forces are typically considered tirement benefits, and paid vacations—typically
non-racial—in the sense that their origins are offered through unionization. This means that
not the direct result of actions, processes, or workers relegated to low-wage service and retail
ideologies that explicitly reflect racial bias. Nev- firms are more likely to experience hardships as
ertheless, they have accelerated neighborhood they struggle to make ends meet. In addition,
decline in the inner city and widened gaps in the local economy suffers when residents have
race and income between cities and suburbs.15 fewer dollars to spend in their neighborhoods.20
Since the mid-20th century, the mode of pro- Beginning in the mid-1970s, the employ-
duction in the United States has shifted dramat- ment balance between central cities and suburbs
ically from manufacturing to one increasingly shifted markedly to the suburbs. Since 1980,
fueled by finance, services, and technology. This over two-thirds of employment growth has oc-
shift has accompanied the technological revolu- curred outside the central city: manufacturing is
tion, which has transformed traditional indus- now over 70 percent suburban, and wholesale
tries and brought about changes that range from and retail trade is just under 70 percent.21 The
streamlined information technology to biomed- suburbs of many central cities, developed orig-
ical engineering.16 inally as bedroom localities for commuters to
In the last several decades, almost all im- the central business and manufacturing districts,
provements in productivity have been associated have become employment centers in themselves.
with technology and human capital, thereby For example, in Baltimore, Detroit, and Phila-
drastically reducing the importance of physical delphia, less than 20 percent of the jobs are now
capital.17 With the increased globalization of located within three miles of the city center.22
economic activity, firms have spread their oper- Accompanying the rise of suburban and exur-
ations around the world, often relocating their ban economies has been a change in commuting
production facilities to developing nations that patterns. Increasingly, workers completely bypass
have dramatically lower labor costs.18 the central city by commuting from one suburb
These global economic transformations have to another. “In the Cleveland region, for exam-
adversely affected the competitive position of ple, less than one-third of workers commute to a
many U.S. Rust Belt cities. For example, Bal- job in the central city and over half (55 percent)
timore, Cleveland, Detroit, Philadelphia, and begin and end in the suburbs.”23
Pittsburgh perform poorly on employment Sprawl and economic stagnation reduce
growth, an important traditional measure of inner-city residents’ access to meaningful eco-
economic performance. Nationally, employment nomic opportunities and thereby fuel the eco-
increased by 25 percent between 1991 and 2001, nomic decline of their neighborhoods. For
 Wilson—Being Poor, Black, and American • 353

Figure 40.3. Unemployment Rate for White and Black Males Age 20 and Older (2nd Quarter 2010)

Source: A Call for Change, Figure 6.2 (Data from the U.S. Department of Labor, U.S. Bureau of Labor Statistics).

example, in Cleveland, although entry-level many employers indicated they felt that, unlike
workers are concentrated in inner-city neigh- women and immigrants (who have recently ex-
borhoods, 80 percent of the entry-level jobs are panded the labor pool for service-sector jobs),
located in the suburbs,24 and there is little pub- inner-city black males lack these qualities.27 In-
lic transportation between these neighborhoods stead, low-skilled black males are perceived as
and jobs. dangerous or threatening. In the past, all black
In addition to the challenges in learning men had to demonstrate was a strong back and
about and reaching jobs, there is persistent racial muscles for heavy lifting and physical labor in a
discrimination in hiring practices, especially for factory, at a construction site, or on an assem-
younger and less-experienced minority work- bly line. They did not have to interact with cus-
ers.25 This racial factor affects black males espe- tomers. Today, they have to search for work in
cially seriously. Today, most of the new jobs for the service sector, and employers are less likely
workers with limited education and experience to hire them because they have to come into
are in the service sector, which includes jobs that contact with the public. Consequently, black
tend to be held by women, such as waitstaff, sales male job-seekers face rising rates of rejection.
clerks, and nurse’s aides. Indeed, “employment This may well account for the higher dropout
rates of young black women now exceed those rate and lower academic achievement of black
of young black men, even though many of these males in comparison with black females. Black
women must also care for children.”26 The shift males are far less likely than black females to see
to service jobs has resulted in a greater demand a strong relationship between their schooling
for workers who can effectively serve and re- and postschool employment.
late to the consumer. In an extensive study in With the departure of higher-income families,
Chicago that my colleagues and I conducted, the least upwardly mobile in society—mainly
354 • Part V: Poverty and the Underclass

low-income people of color—are left behind in grouping students of similar capability for in-
neighborhoods with high concentrations of pov- struction, not only tends to segregate African
erty and deteriorating physical conditions. These American students but often results in placing
neighborhoods offer few jobs and typically lack some black students in lower-level classes, even
basic services and amenities, such as banks, gro- though they have the cultural capital—requisite
cery stores and other retail establishments, parks, skills for learning—to compete with students in
and quality transit.28 Typically, these communi- higher-level classes.30
ties also suffer from substandard schools, many However, there has emerged a form of what
with run-down physical plants. Two of the most some scholars refer to as “laissez faire racism,” a
visible indicators of neighborhood decline are perception that blacks are responsible for their
abandoned buildings and vacant lots. According own economic predicament and therefore are
to one recent report, there are 60,000 abandoned undeserving of special government support.31
and vacant properties in Philadelphia, 40,000 in The idea that the federal government “has a
Detroit, and 26,000 in Baltimore.29 special obligation to help improve the living
standards of blacks” because they “have been
discriminated against for so long” was supported
Cultural Forces by only one in five whites in 2001, and has not
In addition to racial and nonracial political and exceeded support by more than one in four since
economic forces, cultural forces may also con- 1975. Significantly, the lack of white support for
tribute to or reinforce racial inequality. Two types this idea is not related to background factors
of cultural forces are in play: (1) national views such as level of education or age.
and beliefs on race, and (2) cultural traits— The vast majority of social scientists agree
shared outlooks, modes of behavior, traditions, that as a national cultural frame, racism, in its
belief systems, worldviews, values, skills, prefer- various forms, has had harmful effects on Afri-
ences, styles of self-presentation, etiquette, and can Americans as a group. Indeed, considerable
linguistic patterns—that emerge from patterns research has been devoted to the effects of racism
of intragroup interaction in settings created by in American society. Distinct cultural frames in
discrimination and segregation and that reflect the inner city have not only been shaped by race
collective experiences within those settings. and poverty, but in turn often shape responses to
Racism has historically been one of the poverty, including responses that may contribute
most prominent American cultural frames and to the perpetuation of poverty.
has played a major role in determining how I have researched how several factors deter-
whites perceive and act toward blacks. At its mine the extent to which communities, as areas
core, racism is an ideology of racial domina- bounded by place, differ in outlook and behav-
tion with two key features: (1) beliefs that one ior.32 These factors include the degree to which
race is either biologically or culturally inferior the community is socially isolated from the
to another, and (2) the use of such beliefs to broader society; the material assets or resources
rationalize or prescribe the way members of the controlled by members of the community; the
“inferior” race should be treated as well as to benefits and privileges the community members
explain their social position as a group and their derive from these resources; their accumulated
collective accomplishments. Today, there is no cultural experiences from current as well as his-
question that the more categorical forms of rac- torical, political, and economic arrangements;
ist ideology—in particular, those that assert the and the influence members of the community
biogenetic inferiority of blacks—have declined wield because of these arrangements.
significantly, even though they still may be em- Culture is closely intertwined with social
bedded in institutional norms and practices. relations in the sense of providing tools (skills,
For example, school tracking, the practice of habits, and styles) and creating constraints
 Wilson—Being Poor, Black, and American • 355

(restrictions on behavior or outlooks) in patterns time invested in underground work reduces time
of social interaction.33 These constraints include devoted to accumulating skills or contacts for le-
cultural frames (shared visions of human behav- gitimate employment.
ior) developed over time through the processes of For those committed to fighting inequality,
ingroup meaning making (shared views on how especially those involved in multiracial coalition
the world works) and decision making (choices politics, the lesson from this discussion of key
that reflect shared definitions of how the world social, political, economic, and cultural forces is
works)—for example, in the inner-city ghetto to fashion a new agenda that gives more scru-
cultural frames define issues of trust/street smarts tiny to both racial and nonracial policies. Given
and “acting black” or “acting white”—that lead our devastating recent recession and slow, jobless
to observable group characteristics. recovery, it is especially important to scrutinize
One of the effects of living in racially fiscal, monetary, and trade policies that may have
segregated neighborhoods is exposure to long-term consequences for our national and re-
group-specific cultural traits (cultural frames, gional economies. We must ameliorate the primary
orientations, habits, and worldviews as well problem feeding concentrated poverty: inner-city
as styles of behavior and particular skills) that joblessness. The ideal solution would be economic
emerged from patterns of racial exclusion and policies that produce a tight labor market—that
that may not be conducive to social mobility. For is, one in which there are ample jobs for all ap-
example, research has found that some groups in plicants. More than any other group, low-skilled
the inner city put a high value on “street smarts,” workers depend upon a strong economy, partic-
the behaviors and actions that keep them safe in ularly a sustained tight labor market.
areas of high crime.34 Street smarts may be an This new agenda should also include an even
adaptation to living in unsafe neighborhoods. sharper focus on traditional efforts to fight pov-
In this environment, it is wise to avoid eye con- erty, to ensure that the benefits from any eco-
tact with strangers and keep to yourself. This nomic upturn are widely shared among the poor
mindset may also lead someone to approach and that they become less vulnerable to down-
new situations with a certain level of skepticism ward swings in the economy. I refer especially to
or mistrust. Although such an approach is log- the following:
ical and smart in an unsafe neighborhood, the • combating racial discrimination in em-
same behavior can be interpreted as antisocial in ployment, which is especially devastating
another setting. Moreover, this street-smart be- during slack labor markets;
havior may, in some cases, prevent individuals • revitalizing poor urban neighborhoods, in-
from performing well on a job interview, creat- cluding eliminating abandoned buildings
ing a perception that they are not desirable job and vacant lots to make them more attrac-
candidates. tive for economic investment that would
Elijah Anderson finds that for some young help improve the quality of life and create
men, the draw of the street is so powerful that jobs in the neighborhood;
they cannot avail themselves of legitimate em- • promoting job training programs to
ployment opportunities when they become enhance employment opportunities for
available. Likewise, Sudhir Venkatesh maintains ghetto residents;
that adherence to the code of shady dealings im- • improving public education to prepare
pedes social mobility.35 The “underground econ- inner-city youngsters for higher-paying
omy enables people to survive but can lead to and stable jobs in the new economy; and
alienation from the wider world,” he states.36 For • strengthening unions to provide the higher
example, none of the work experience accrued in wages, worker protections, and benefits
the informal economy can be listed on a resume typically absent from low-skilled jobs in
for job searches in the formal labor market, and retail and service industries.
356 • Part V: Poverty and the Underclass

and Ruth D. Peterson (Stanford, CA: Stanford Uni-

In short, this new agenda would reflect a
versity Press, 1995), 37–54.
multipronged approach that attacks inner-city 7. Katz, “Reframing the ‘Underclass’ Debate.”
poverty on various levels, an approach that rec- 8. Katz, “Reframing the ‘Underclass’ Debate,”
ognizes the complex array of factors that have 461–462. On the history of suburbs in America, see
contributed to the crystallization of concen- Kenneth T. Jackson, Crabgrass Frontier: The Subur-
trated urban poverty and limited the life chances banization of the United States (New York: Oxford
of so many inner-city residents. University Press, 1985). For a good discussion of the
effects of housing discrimination on the living con-
NOTES ditions, education, and employment of urban mi-
1. Paul A. Jargowsky, “Ghetto Poverty among norities, see John Yinger, Closed Doors, Opportunities
Blacks in the 1980s,” Journal of Policy Analysis and Lost: The Continuing Costs of Housing Discrimination
Management 13, no. 2 (Spring 1994): 288–310. (New York: Russell Sage Foundation, 1995).
2. Michael B. Katz, “Reframing the ‘Underclass’ 9. Mark Condon, “Public Housing, Crime and
Debate,” in The “Underclass” Debate: Views from the Urban Labor Market: A Study of Black Youth in
History, ed. Michael B. Katz (Princeton, NJ: Princ- Chicago,” Working Paper Series (Cambridge, MA:
eton University Press, 1993), 440–478; David W. Malcolm Wiener Center, John F. Kennedy School of
Bartelt, “Housing the ‘Underclass,’” in The “Un- Government, Harvard University, 1991), 4.
derclass” Debate: Views from History, ed. Michael B. 10. Sampson and Wilson, “Toward a Theory
Katz (Princeton, NJ: Princeton University Press, of Race.” Also see Bartelt, “Housing the ‘Under-
1993), 118–157; Thomas J. Sugrue, “The Structure class’”; Kelley, “The Black Poor and the Politics of
of Urban Poverty: The Reorganization of Space and Opposition”; Sugrue, “The Structure of Urban Pov-
Work in Three Periods of American History,” in The erty”; Hirsch, Making the Second Ghetto; and John
“Underclass” Debate: Views from History, ed. Michael F. Bauman, Norman P. Hummon, and Edward K.
B. Katz (Princeton, NJ: Princeton University Press, Muller, “Public Housing, Isolation, and the Urban
1993), 85–117; and Robin D. G. Kelley, “The Black Underclass,” Journal of Urban History 17, no. 3 (May
Poor and the Politics of Opposition in a New South 1991): 264–292.
City, 1929–1970,” in The “Underclass” Debate: Views 11. Lincoln Quillian, “Migration Patterns and
from History, ed. Michael B. Katz (Princeton, NJ: the Growth of High-Poverty Neighborhoods, 1970–
Princeton University Press, 1993), 293–333. 1990,” American Journal of Sociology 105, no. 1 (July
3. Katz, “Reframing the ‘Underclass’ Debate,” 1999): 1–37.
462. Also see Bartelt, “Housing the ‘Underclass’”; 12. See Demetrios Caraley, “Washington Aban-
Sugrue, “The Structure of Urban Poverty”; and dons the Cities,” Political Science Quarterly 107, no.
Martin Anderson, The Federal Bulldozer: A Critical 1 (Spring 1992): 1–30.
Analysis of Urban Renewal, 1949–1962 (Cambridge, 13. Bruce A. Wallin, Budgeting for Basics: The
MA: MIT Press, 1964). Changing Landscape of City Finances (Washington,
4. Raymond A. Mohl, “Planned Destruction: DC: Brookings Institution Metropolitan Policy Pro-
The Interstates and Central City Housing,” in From gram, August 2005).
Tenements to the Taylor Homes: In Search of an Urban 14. Caraley, “Washington Abandons the Cities.”
Housing Policy in Twentieth-Century America, ed. 15. Radhika K. Fox and Sarah Treuhaft, Shared
John F. Bauman, Roger Biles, and Kristin M. Szyl- Prosperity, Stronger Regions: An Agenda for Rebuild-
vian (University Park, PA: Pennsylvania State Uni- ing America’s Older Core Cities (Oakland, CA: Pol-
versity Press, 2000), 226–245; Adam Cohen and icyLink, 2006).
Elizabeth Taylor, American Pharaoh: Mayor Richard 16. Fox and Treuhaft, Shared Prosperity; and
J. Daley; His Battle for Chicago and the Nation (Bos- Bill Joy, “Why the Future Doesn’t Need Us,” Wired
ton: Little, Brown, 2000); and Arnold R. Hirsch, (April 2000): 238–262.
Making the Second Ghetto: Race and Housing in Chi- 17. William Julius Wilson, When Work Disap-
cago, 1940–1960 (Cambridge, MA: Cambridge Uni- pears: The World of the New Urban Poor (New York:
versity Press, 1983). Alfred A. Knopf, 1996).
5. Cohen and Taylor, American Pharaoh. 18. Fox and Treuhaft, Shared Prosperity.
6. Robert J. Sampson and William Julius Wil- 19. Fox and Treuhaft, Shared Prosperity.
son, “Toward a Theory of Race, Crime, and Urban 20. Fox and Treuhaft, Shared Prosperity; and Wil-
Inequality,” in Crime and Inequality, ed. John Hagan son, When Work Disappears.
 Danziger, Chavez, and Cumberworth—Poverty and the Great Recession • 357

21. U.S. Department of Housing and Urban De- 30. For a review of the literature on school
velopment, The State of the Cities. tracking, see Janese Free, “Race and School Track-
22. Fox and Treuhaft, Shared Prosperity. ing: From a Social Psychological Perspective” (paper
23. Fox and Treuhaft, Shared Prosperity, 32. presented at the annual meeting of the American
24. Fox and Treuhaft, Shared Prosperity. Sociological Association, San Francisco, CA, August
25. See, for example, Wilson, When Work Disap- 14, 2004).
pears; Joleen Kirschenman and Kathryn M. Neck- 31. Lawrence Bobo, James R. Kluegel, and Ryan
erman, “‘We’d Love to Hire Them, But . . . ’: The A. Smith, “Laissez-Faire Racism: The Crystallization
Meaning of Race for Employers,” in The Urban of a Kinder, Gentler, Antiblack Ideology,” in Racial
Underclass, ed. Christopher Jencks and Paul E. Pe- Attitudes in the 1990s: Continuity and Change, ed.
terson (Washington, DC: Brookings Institution, Steven A. Tuch and Jack K. Martin (Westport, CT:
1991), 203–234; Kathryn M. Neckerman and Praeger, 1997), 15–44.
Joleen Kirschenman, “Hiring Strategies, Racial Bias, 32. Wilson, When Work Disappears.
and Inner-City Workers,” Social Problems 38, no. 4 33. Charles Tilly, Durable Inequality (Berkeley,
(November 1991): 433–447; and Harry J. Holzer, CA: University of California Press, 1998).
What Employers Want: Job Prospects for Less-Educated 34. Elijah Anderson, Code of the Street: Decency,
Workers (New York: Russell Sage, 1995). Violence, and the Moral Life of the Inner City (New
26. Harry J. Holzer, Paul Offner, and Elaine York: W. W. Norton, 1999).
Sorensen, “What Explains the Continuing Decline 35. Anderson, Code of the Street; and Sudhir
in Labor Force Activity among Young Black Men?” Alladi Venkatesh, Off the Books: The Underground
(paper presented for Color Lines Conference, Har- Economy of the Urban Poor (Cambridge, MA: Har-
vard University, August 30, 2003). vard University Press, 2006).
27. Wilson, When Work Disappears. 36. Venkatesh, Off the Books, 385. For another
28. William Julius Wilson, The Truly Disadvan- excellent study of how activities in the underground
taged: The Inner City, the Underclass, and Public Policy economy can adversely affect inner-city residents, see
(Chicago: University of Chicago Press, 1990); Wil- Loïc Wacquant, “Inside the Zone: The Art of the
son, When Work Disappears; and Fox and Treuhaft, Hustler in the Black American Ghetto,” Theory, Cul-
Shared Prosperity. ture, and Society 15, no. 2 (1998): 1–36.
29. Fox and Treuhaft, Shared Prosperity.

41 • Sheldon Danziger, Koji Chavez, and Erin Cumberworth

Poverty and the Great Recession

Most Americans believe that “a rising tide lifts rapidly for both less-educated and more edu-
all boats,” meaning that economic growth raises cated workers, family incomes rose rapidly for
the incomes of the poor as well as the incomes every quintile in the income distribution, and
of the middle class and the rich. The economy poverty fell rapidly. Indeed, the official poverty
did operate like this from the end of World for all persons fell from 22.4% in 1959 to 11.1 %
War II to the early 1970s. During this period, in 1973. This period may, at least in retrospect,
inflation-adjusted annual earnings increased be regarded as a “golden age” of rising family

Sheldon Danziger, Koji Chavez, and Erin Cumberworth. “Poverty and the Great Recession,” Recession Trends
Report, pp. 1–5. Copyright © 2012 by the Russell Sage Foundation, 112 East 64th Street, New York, NY
10065. Reprinted with permission.
358 • Part V: Poverty and the Underclass

incomes and falling poverty rates. The purpose officially lasted from December 2007 through
of this chapter is to show how poverty rates have June 2009, but monthly unemployment rates
evolved as the golden age came to a close and as remained above 9% for more than two years
the Great Recession played out. after the official start of the current economic
recovery. What is surprising is that poverty did
not increase even more than it did. Poverty
After the Golden Age reached 15% in both the recessions of the early
The “golden age” came to an end in the early 1980s (1982 and 1983) and of the early 1990s
1970s as the economy entered an era of slow real (1993). Each of these recessions was shorter and
wage growth and rising inequalities that persists the subsequent recoveries were more rapid than
today. A growing economy with low unemploy- has been the case in the recovery from the Great
ment rates continues to be necessary, but is no Recession.
longer sufficient, to reduce poverty in the United Poverty did not rise even more through
States. The official poverty rate has never fallen 2010 primarily because the American Recovery
below the 1973 low achieved at the end of the and Reinvestment Act of 2009 (i.e., the “stim-
“golden age.” It has increased during recessions ulus bill”) dramatically increased government
and fallen during economic recoveries, but even spending on unemployment insurance and
in recoveries it has remained well above that 1973 other safety net programs and provided funds
low. It came closest to meeting that low after the to state and local governments to prevent lay-
long economic boom of the 1990s, when the un- offs, as Brookings’ Gary Burtless has shown.
employment rate fell to 4.0% in 2000 and there The stimulus thus offset a greater percentage
was a “labor shortage” in many labor markets. of the recession-induced income losses than
The poverty rate fell to 11.3% in that year. was the case in recent recessions. However,
Why did the golden age end? Poverty did most of the safety net aspects of the stimulus
not fall between 1973 and 2000, even though bill have expired and Congress has refused to
Gross National Product per capita increased sub- extend them and has made reducing the deficit
stantially, because a series of economic changes a higher priority than protecting the poor and
meant that workers no longer captured the ben- the unemployed.
efits of rising labor market productivity. These
changes include labor-saving technological
advancements, the globalization of labor and Who Bears the Brunt of Poverty?
product markets, immigration of workers with These trends and fluctuations in the official
very low educational attainment, the declining poverty rate for all persons mask important
real value of the minimum wage, and declining differences in poverty rates and in the effects
union influence. of the Great Recession on various demographic
The first decade of the 21st century has been groups. Which groups have poverty rates that
one of slow economic growth and rising poverty. are above average during normal economic
Although the recession of 2001 was mild, there times and which experienced the largest poverty
was no wage growth for less-educated workers rate increases due to the recession? We address
or even for the median worker, as the economy this question by considering in turn trends by
recovered. Even though the annual unemploy- age, education, race, and family type. Groups
ment rate for 2007 fell to 4.6% as the economy that have the highest poverty rates in normal
recovered, the poverty rate was 12.5%. economic times also experienced the largest in-
creases over the course of the recession.
We begin with trends by age in Figure 41.1
The Great Recession which shows poverty rates for those at each end
It is not surprising that the Great Recession in- of the age distribution—children under age 18
creased poverty to 15.1% in 2010. The recession and the elderly above age 65—as well as for
 Danziger, Chavez, and Cumberworth—Poverty and the Great Recession • 359

Figure 41.1. Poverty Rates for Children, Adults, and the Elderly, 1966–2010

Source: U.S. Census Bureau Historical Poverty tables, http://www.census.gov/hhes/www/poverty/data/historical

/people.html.

nonelderly adults. According to the official mea- each year for inflation and Medicare and Med-
sure, the poverty rate increased most for children icaid guarantee health insurance coverage. The
both since the early 1980s and during the Great “golden age” in the labor market marked the
Recession. The gap between the adult and child end of an era of relatively low poverty rates for
poverty rates increased over time because (a) chil- children and adults; however, the elderly have
dren live with younger adults whose economic for the most part been protected against reces-
status varies more over the business cycle than sions, including the Great Recession, for the last
that of mature adults, (b) children live more with four decades. Poverty for the elderly in 1966 was
single-parents and less with two parents and (c) 28.5%, almost double the rate for all persons,
the erosion of the poverty-reducing impacts of 14.7%; in 2010, the elderly poverty rate was 9%,
income support programs. From 2007 to 2010, 6.1 percentage points lower than the national
the poverty rate among children increased 4.0 average.
percentage points (from 18 to 22%), compared Working aged adults, ages 18 to 64, benefit
to a 2.6 percentage point increase (from 12.5 to most during economic booms and are most neg-
15.1 %) in the rate for all persons. atively affected during recessions. The poverty
The official poverty rate among the elderly rate for younger cohorts of non-elderly adults is
has steadily decreased since 1967 and was unaf- higher and even more responsive to economic
fected by the Great Recession. The elderly have changes. As Figure 41.2 shows, the poverty rates
had a European-style safety net since a series of for individuals ages 18 to 24 and ages 25 to 34
policy changes enacted in the decade following have been more responsive to economic cycles
President Johnson’s 1964 declaration of War on in recent decades than those of other adults.
Poverty: Social Security benefits and Supple- In 2007, the poverty rates for those ages 18 to
mental Security Income benefits are indexed 24 and ages 25 to 34 were 17.3% and 12.3%,
360 • Part V: Poverty and the Underclass

Figure 41.2. Poverty Rates for Nonelderly Adults by Age Cohort, 1968–2010

Source: Stanford Center on Poverty and Inequality using March CPS data downloaded from IPUMS

respectively. The 2007 poverty rates for the than a high school degree, manufacturing and
three “mature” age groups ranged from 8.5% to construction jobs, experienced the largest em-
9.5%. From 2007 to 2010, the Great Recession ployment losses. The gap between the poverty
dramatically increased poverty rates for each rate for those between the ages of 25 and 64
age group. For 18 to 24 year olds, the poverty with no more than a high school degree and
rate increased 4.7 percentage points, to 22%. those with at least a college degree has been
The poverty rates for 25 to 34, 35 to 44, and widening for decades. Between 2007 and 2010,
45 to 54 year olds increased by a more mod- the poverty rate for those with less than a high
est three percentage points. The poverty rate school degree increased 5.5 percentage points,
for those ages 55 to 64 increased by only 1.5 from 28.1% to 33.6%; for high school gradu-
percentage points. The 2010 poverty rates were ates, the rate increased 3.3 percentage points,
higher than at any time over the last four de- from 11.7% to 15%. In contrast, the rate for
cades for all adult age cohorts, except those ages those with at least a college degree increased
55 to 64. Thus, the Great Recession, similar only from 3.4% to 4.3%.
to other recent recessions, has had a dispropor- Popular discourse on the effects of the Great
tionate impact on younger adults, as employers Recession on young adults often focuses on the
tend to follow a “last-hired, first-fired” pattern plight of college graduates.1 However, the Great
for mass layoffs. Recession’s effect on poverty among the college
The labor market prospects for less-educated educated is much smaller than its effects on
workers have been difficult since the early 1970s young adults with less education.
(see Figure 41.3). They were also hit the hard- The Great Recession also had disparate effects
est by the Great Recession, particularly because on the poverty rates of white non-Hispanics, black
some of the best jobs for workers with no more non-Hispanics, and Hispanic adults between the
 Danziger, Chavez, and Cumberworth—Poverty and the Great Recession • 361

Figure 41.3. Poverty Rates by Educational Attainment, Persons Ages 25–64

Source: Stanford Center on Poverty and Inequality using March CPS data downloaded from IPUMS

ages of 18 and 64. As Figure 41.4 shows, the pov-

The Great Recession and the
erty rates for black non-Hispanics and Hispanics
Working Poor
have been substantially higher and more respon-
sive to the business cycle than the poverty rate The welfare reforms of the Clinton era moved
for white non-Hispanics. During the economic many single mothers from being non-working
boom of the 1990s, both black non-Hispanics welfare recipients to being workers who often re-
and Hispanics experienced decreasing poverty mained poor, even when their earnings exceeded
rates; between 1993 and 2000, the rates for both the welfare benefits they lost. As shown in Figure
groups fell from about 25% to about 18%; the 41.5, there was a substantial uptick during the
rate for white non-Hispanic adults fell only from 1990s in the proportion of poor families and
about 8.5 to 7% during this boom. poor single-mother families in which at least one
The Great Recession led to greater poverty family member worked. The simple implication:
increases for black non-Hispanics and His- The poverty of the 1990s was increasingly gen-
panics than for white non-Hispanics. Between erated by low wages rather than nonwork, espe-
2007 and 2010, the poverty rates for black cially for single mothers.
non-Hispanic adults increased by 3.7 percent- Did the Great Recession reverse this trend? It
age points (from 19.6 to 23.3%), by 4.5 points did—as Figure 41.6 shows, between 2007 and
for Hispanics (from 17.9 to 22.4%), but by only 2010, the percent of individuals living in poor
2.2 percentage points for white non-Hispanics families in which at least one family member
(from 7.7% to 9.9%). As with groups classified worked decreased by 3.1 percentage points from
by educational attainment, the Great Recession 66.6% to 63.5%. For those in poor single-mother
disproportionately increased poverty among families, the percent with a worker decreased
groups with higher-than-average poverty rates. 6.7 percentage points from 61.7% to 55%. In
362 • Part V: Poverty and the Underclass

Figure 41.4. Poverty Rates by Race/Ethnicity, Persons Ages 18–64

Source: “White Non-Hispanic” and “Hispanic any race” are a combination of the Census Bureau’s published poverty
figures (Historical Poverty Tables—People, Table 3, http://www.census.gov/hhes/www/poverty/data/historical/people.
html) and estimates produced by the Stanford Center on Poverty and Inequality using March CPS microdata from
IPUMS. For “White Non-Hispanics,” estimates for years 1974–2010 are from the Census Bureau, and estimates for
1970–1973 are produced by the Stanford Center on Poverty and Inequality For “Hispanic any race,” estimates for
years 1976–2010 are from the Census Bureau, and estimates for 1970–1975 are produced by the Stanford Center on
Poverty and Inequality.
The poverty rate for Black Non-Hispanics is not available from the Census Bureau, and was produced by Stanford
Center on Poverty and Inequality using March CPS data from IPUMS.

contrast, there was a smaller decline in employ- Tax Credit). It also does not subtract payroll
ment among all American families. Between and income taxes or work-related expenses and
2007 and 2010, the percentage of individuals out-of pocket medical expenses that reduce the
in all families with at least one worker decreased resources available to poor families to pay for ne-
only by 1.8 percentage points to 87.7%. cessities such as food, clothing, shelter and util-
ities. Lastly, the official measure does not vary
geographically, although living costs are much
Improved Poverty Measurement higher in large cities than in rural areas. (For a
The official poverty rate, adopted in the late more extensive treatment of these issues, see the
1960s, has been subject to many criticisms. 1995 National Academy of Sciences report on
The official measure is based on money income measuring poverty.)
before taxes; it does not count a variety of gov- In November 2011, the U.S. Census Bureau
ernment benefits that raise the resources avail- released its “Supplemental Poverty Measure,” or
able to poor families to make ends meet (e.g., SPM, which modernizes the poverty measure to
noncash benefits, such as food stamps, and re- respond to a number of flaws, including the ones
fundable tax credits, such as the Earned Income discussed above. For 2010, the SPM reports a
 • 363
Figure 41.5. Poverty Rates for People Living in Specified Family Types, 1980–2010

Source: The poverty rates for people living in families with a female householder (no husband present) and unrelated
individuals are from the Census Bureau Historical Poverty Tables, http://www.census.gov/hhes/www/poverty/data/
historical/people.html, Table 2.
The rates for people living in married-couple families are produced by the Stanford Center on Poverty and Inequality
using March CPS data downloaded from IPUMS.

Figure 41.6. Percent of Poor Persons Living in Families with at Least One Worker

Source: The percentage of people living in working families comes from the Census Bureau Detailed Poverty
Tables, http://www.census.gov/hhes/www/cpstables/032011/pov/toc.htm.
364 • Part V: Poverty and the Underclass

poverty rate of 16.0 percent for all persons, prospects for reducing poverty over the next sev-
only slightly higher than the 15.1% official rate. eral years are dismal. The Federal Reserve Board
However, the addition of non-cash benefits like and the Congressional Budget Office project that
food stamps and the Earned Income Tax Credit the unemployment rate is unlikely to fall below
(EITC) lowers the poverty rate for children 6% until late in 2015. Given these economic
from an official rate of 22.5% to an SPM rate forecasts and an assumption that antipoverty
of 18.2%. In contrast, because the elderly have policies are constant, Emily Monea and Isabel
high out-of-pocket medical expenses, their SPM Sawhill of the Brookings Institution project that
poverty rate, 15.9%, is much higher than their poverty will hardly fall at all during the slow eco-
official rate, 9.0%. The SPM tells us that pov- nomic recovery—it will be 14.5% in 2016 even
erty in 2010 was lower than it otherwise would if we avoid a double-dip recession. Poverty will
have been because parts of the stimulus pack- remain high unless greater attention is given to
age—for example increases in food stamps and expanding public policies that have been shown
the EITC—are counted as reducing poverty by to reduce poverty.
the SPM, but not by the official measure. As the
Center for Budget and Policy Priorities’ Arloc NOTES
Sherman has documented, because the SPM 1. “The Curse of the Class of 2009,” Wall Street
Journal (May 9, 2009); “College Degree No Shield
counts these benefits, it shows a smaller increase
as More Jobs Are Slashed,” The Washington Post (Jan-
in poverty between 2008 and 2010 than does the
uary 4, 2009).
official measure.

The Future of Poverty

The last time poverty was as high as it was in
2010 was in the early 1980s. Unfortunately, the
 Kenworthy—How Rich Countries Lift Up the Poor • 365

The Effects of Politics and Institutions

42 • Lane Kenworthy

How Rich Countries Lift Up the Poor

As a nation grows richer, the living standards of Switzerland—it looks more like the American
its least well-off ought to rise. But one of the and Canadian one. This raises the question:
most striking features of late-industrial develop- Why do the poor only sometimes reap some of
ment is that the fruits of growth have—in some the fruits of a growing economy?
countries—been very unequally shared. Which
countries have succeeded in lifting up the poor?
And which have failed? Government Policy Matters
Figure 42.1 shows what happened from the We often think of economic growth as a
late 1970s to the mid-2000s in four countries, “trickle-down” process in which rising earnings
the United States, Canada, the United Kingdom, are secured via more work hours and higher
and Sweden. The charts show the degree to which wages. But in almost all of these countries (Ire-
household income at the tenth percentile of the land and the Netherlands are exceptions) the
income distribution (vertical axis) improves earnings of low-end households increased little,
as GDP per capita (horizontal axis) increases. if at all, over time. Instead, as Figure 42.2 sug-
Each of these countries experienced economic gests, increases in net government transfers—
growth, moving to the right along the horizontal transfers received minus taxes paid—tended to
axis. But they varied markedly in the degree to drive increases in incomes when they occurred.
which that growth reached the poor. In Canada, Governments in some of these nations did
there was little or no rise in household income at more to pass the fruits of economic growth on
the tenth percentile. In the United States, there to the poor. For the most part, this didn’t en-
was very little, and it occurred only in the late tail increasing the share of GDP allocated to
1990s. The United Kingdom did much better, public transfers. Such increases were common
though also mainly during a particular period, in the 1960s and 1970s. But in most of these
the mid-1990s to the mid-2000s. In Sweden, affluent nations—even the most generous ones,
economic growth yielded a consistent improve- such as Denmark and Sweden—increases in the
ment in the incomes of those at the bottom. share of GDP allocated to public transfers largely
In most advanced democracies—Austria, stopped after the 1970s. In recent decades, the
Belgium, Denmark, Finland, France, Ireland, distinction has been between countries that kept
the Netherlands, Norway, and Spain—the transfers rising in line with GDP versus those
pattern during these years resembles Sweden’s. that did not. Sometimes doing so requires no
But in others—Australia, Germany, Italy, and explicit policy change, as benefit levels tend to

Lane Kenworthy. “How Rich Countries Lift Up the Poor,” Pathways (Fall 2011), pp. 28–32. Used by permis-
sion of the Center on Poverty and Inequality, Stanford University.
366 • Part V: Poverty and the Underclass

Figure 42.1 Where Have Low-End Household Incomes Risen with Economic Growth?
United States Canada United Kingdom Sweden

13,000
P10 income ($)

2005
2004
2004 2004
1981
1979
1981
1979

4,000

$10 $37 $10 $37 $10 $37 $10 $37

GDP per capita ($ thousands)

Note: Low-end household income by GDP per capita, 1979–2005. P10 income: Posttransfer-posttax household income,
adjusted for household size, at the tenth percentile of the distribution. Household income and per capita GDP are in
year-2000 U.S. dollars. The axis values represent the full range for the 17 nations mentioned in the text.
Source: Author’s calculations using Luxembourg Income Study and OECD data.

Figure 42.2 Government Transfers and Taxes Have Been the Chief Mechanism Through Which Eco-
nomic Growth Reaches the Poor

Group 1: earnings

Group 1: net govt transfers

Group 2: earnings

Group 2: net govt transfers

1979 2005
0 5,500
Average in bottom-income-decile households ($)

Note: Average income in the bottom decile of the posttranster-posttax income distribution. Household income, adjusted
for household size, in year-2000 U.S. dollars. Group 1: Denmark, Finland, Ireland, Netherlands, Norway, Sweden,
United Kingdom. Group 2: Australia, Canada, Germany, Switzerland, United States. This calculation can’t be done for
Austria, Belgium, France, Italy, and Spain. Actual years vary depending on the country.
Source: Author’s calculations using Luxembourg income Study data.

rise automatically as the economy grows. This automatically increase when the economy grows.
happens when, for instance, pensions, unem- Social Security retirement benefits are indexed to
ployment compensation, and related benefits average wages, so they have tended to rise more
are indexed to average wages. Increases in other or less in concert with GDP. Unemployment
transfers, such as social assistance, typically re- benefit levels are determined by state govern-
quire periodic policy updates. That’s true also of ments. In many instances, the benefit level is a
tax reductions for low-income households. “replacement rate,” which means the payment
In the United States, only one of the main is a certain fraction of the unemployed per-
government transfer programs, Social Security, son’s former wage or salary. Because real wages
is structured in such a way that benefit levels in the bottom half of the distribution have not
 Kenworthy—How Rich Countries Lift Up the Poor • 367

increased in the past several decades, unem- can do better at helping able adults get into (or
ployment benefits for Americans in low-wage back into) employment, but we shouldn’t pre-
jobs have failed to keep up with growth in the tend that paid work is a realistic route to guar-
economy. Other programs, such as the Earned anteeing rising incomes for everyone.
Income Tax Credit (EITC), the Supplemental Income isn’t a perfect measure of the mate-
Nutritional Assistance Program (SNAP, formerly rial well-being of low-end households. We need
called Food Stamps), Social Security Disability to supplement it with information on actual
Insurance (SSDI), and Supplemental Security living conditions, and researchers and govern-
Income (SSI), are indexed to prices. This means ments now routinely collect such data. Unfor-
they keep up with inflation, but not with eco- tunately, those data aren’t available far enough
nomic growth. Temporary Assistance for Needy back in time to give us a reliable comparative
Families (TANF, formerly AFDC) payments picture of changes. For that, income remains our
are determined by state policymakers; there best guide. What the income data tell us is that
is no automatic increase, not even for prices. the United States has done less well by its poor
AFDC-TANF benefit levels have fallen steadily than many other affluent nations, because we’ve
in inflation-adjusted terms over the past several failed to keep government supports for the least
decades. well-off rising in sync with our GDP.
If most of the poorest Americans were el-
derly Social Security recipients, the U.S. pattern
in Figure 42.1 probably would look more like Tradeoffs?
Sweden’s. But in the United States, as in many It often is said that there is no free lunch, that
other countries, most of those in the bottom 10 generosity comes at a cost. If we commit to im-
percent are not retirees. Many elderly Americans provement in the absolute living standards of the
have no income from earnings, but Social Secu- least well-off, must we sacrifice other desirable
rity benefits, payments from employer-based re- outcomes?
tirement programs (company pension or 401k), Here, too, the experiences of rich nations
and other income (from the sale of a house, over the past several decades can offer some in-
for instance) combine to keep them above the sight. I begin with a measure of “progress for the
bottom decile. The fact that most of our other poor”: the slope of each country’s line in charts
government transfers have only kept up with in- such as those shown in Figure 42.1. This is an
flation rather than with the economy, coupled indicator of the degree to which low-end house-
with the decline in AFDC-TANF benefits, ac- hold incomes rise as the society gets richer. Fig-
counts for the slow income growth at the bottom ure 42.3 displays a number of scatterplot graphs,
in the United States. each of which has this measure on the horizontal
Should we bemoan the fact that employment axis. Countries positioned to the right have been
and earnings aren’t the key trickle-down mech- more successful at boosting the incomes of poor
anism? No. At higher points in the income dis- households. On the vertical axes are indicators
tribution, they do play more of a role. But for of economic health, liberty, mobility, happiness,
the bottom 10 percent there are limits to what and fiscal discipline. These are measured at the
employment can accomplish. Some people have end of the period, around the year 2007 (before
psychological, cognitive, or physical conditions the economic crash). Each of the outcome mea-
that limit their earnings capability. Others are sures is arrayed so that it is better to be higher on
constrained by family circumstances. At any the vertical axis. Evidence suggestive of a tradeoff
given point in time, some will be out of work would therefore appear in the form of a nega-
due to structural or cyclical unemployment. And tively sloped line.
in all rich countries, a large and growing number The conclusion from these charts is straight-
of households are headed by retirees. We surely forward: There is little or no indication that
368 • Part V: Poverty and the Underclass

Figure 42.3 Are There Tradeoffs Between Progress for the Poor and Other Desirable Outcomes?
National wealth Employment Economic liberty

Ease of doing business (rank)

41 NO
79 CH 1
US
NO DK
CA

Employment rate (%)

US SE UK
NL DK
GDP per capita

IE CA ASI
ASI UK NO
(US$ 000s)

US AU
CH IE
FI l SE
SE NL DE IE
CA AU FI
ASI FI DK CH
BE UK ES BE
DE
DE FR NL
FR AU
IT BE FR
ES
24 ES 59 IT 20 IT

Low High Low High Low High

Intergenerational mobility Subjective well-being Fiscal discipline

.10 8.2 IE
DK 15 ASI
DK NO
FI IE
Father-son earnings

CA DK
CH

Government debt
ASI SE
Life satisfaction

US
AU NL FI ES

(% of GDP)
correlation

CH UK SE
(0 to 10)

DEASI SE NL
NO ES
US NO
CA FI DE CA AU
UK UK
FR
BE

DE BE
FR

FR
IT
.45 US 6.7 IT 110 IT

Low High Low High Low High

Note: For more outcomes and for details on the measures Progress
and data for sources,
the poor see Lane Kenworthy, Progress
Progress
forforthe
thePoor,
poor

Oxford University Press, 2011.

improvement in the incomes of the poor entails was one of the largest in any of the rich countries
a sacrifice of other valued outcomes. for which reliable data are available.
Unfortunately, apart from a few exceptions
such as the EITC, movement in this direction
Prospects for Progress in America here in the United States has been halting. In
Modest, regularized increases in the most other cases, the politics of helping Ameri-
inflation-adjusted benefit levels of existing so- ca’s poor have proved quite difficult.
cial programs—the Earned Income Tax Credit, Is public opinion the obstacle? Most Ameri-
unemployment compensation, social assistance cans support capitalism and business. Many be-
(TANF and SNAP), housing assistance, and lieve hard work, rather than luck or help from
disability benefits—would yield significant im- others, is the key to success. Many feel they have
provements in the incomes of America’s least opportunity to get ahead. At a general level,
well-off. many are skeptical about the government’s abil-
Recent developments just across the pond ity to help. Yet many believe income inequality is
have shown us the way. One of the most success- too high and that high inequality is not necessary
ful recent antipoverty efforts in affluent coun- for the country’s prosperity. There is only limited
tries was that of the New Labour governments support for enhanced redistribution as a remedy
in the United Kingdom from the late 1990s for high inequality, but Americans do support
through the late 2000s. Though Tony Blair and increased government spending on programs
Gordon Brown’s governments focused much of perceived to enhance opportunity and economic
their rhetoric and policy reform on improving security. And a majority consistently favors in-
employment and economic opportunity, they creased government expenditure on the poor.
also increased net government transfers to low Social scientists’ research on the determinants
earners, single parents, and pensioners. Benefit of social policy generosity tells us that what mat-
and tax changes between 1997 and 2005 in- ters most are institutions. Given America’s polit-
creased real disposable income for lowest-income ical institutions—the lack of a social democratic
households by about 20 percent. This increase political party, a privatized system of campaign
 Newman and O’Brien—Taxing the Poor • 369

financing, a majoritarian electoral system, a fed- (AFDC) and 1970s (food stamps); create an
eral government structure, extensive separation employment-conditional earnings subsidy in
of power across the three branches of govern- the 1970s (the EITC) and expand it in ensuing
ment, a bicameral legislature, and the filibuster years; implement severe time limits on receipt
practice in the Senate—it is not surprising that of a key social assistance benefit (TANF) in the
we are a laggard among the rich countries in 1990s; or fail to adopt government support for
public safety net generosity. near-universal health care coverage in the 1970s
Yet the world of social policy is not a deter- and 1990s but then pass it in 2010.
ministic one. Structures and institutions con- The possibilities for American social policy
strain, but they don’t dictate outcomes. For surely are not endless, but neither are they as
instance, over the past century, center-right limited as a focus on America’s political structure
Christian democratic parties have been nearly might lead us to presume. Over the course of the
as important as social democratic ones in pro- past century, U.S. policymakers sometimes have
moting generous social programs. Government been able, even at unlikely moments, to fashion
support for child care and early education in compromises that helped boost the incomes and
continental Belgium and France rivals that in material well-being of America’s low-end house-
social-democratic Denmark, Norway, and Swe- holds. When new or expanded programs have
den. In recent years, EITC-type policies have worked reasonably well, Americans have tended
been implemented and expanded in widely di- to like them. They then become difficult to re-
verse institutional settings and by governments move. This staying power is aided by the array
at all ends of the partisan spectrum. of veto points in the U.S. policy-making process.
It was not foreordained that the United States The trajectory of American social policy has
would institute public health insurance pro- therefore tended to be one of advance—slow
grams for its elderly and its poor in the 1960s and halting advance, but advance nonetheless.
and enhance them in subsequent decades; ex- To me this suggests reason for optimism about
pand its social assistance programs in the 1960s prospects for the future.

43 • Katherine S. Newman and Rourke L. O’Brien

Taxing the Poor

How Some States Make Poverty Worse

In the fall of 2007, Alabama Arise, a coalition of much as 12% to consumers’ grocery bills. State
congregations and organizations based in Mont- taxes of this kind hit everyone, rich and poor
gomery, Alabama, mounted a campaign to per- alike, in the pocketbook. But Alabama citizens
suade the state legislature to cancel the sales tax at the bottom of the ladder, who live at the very
on food for home consumption, a levy that— edge of survival to begin with, were finding
between state and local taxation—was adding as themselves unable to feed their families at the

Katherine S. Newman and Rourke L. O’Brien. “Taxing the Poor: How Some States Make Poverty Worse,”
Pathways (Summer 2011), pp. 22–26. Used by permission of the Center on Poverty and Inequality, Stanford
University.
370 • Part V: Poverty and the Underclass

end of the month. Looking to stretch the dollar that policymakers have at their disposal for al-
or the allotment of food stamps, poor families leviating poverty—the tax system—remains in
were going without or switching to cheap food the shadows. Divergent state policies determine
that fills the stomach, but leads to obesity and all how much money is left in the hands of poor
the damaging consequences that follow. families, a fact that is not well understood in an-
Twenty senators had promised their support. tipoverty circles. The argument that we lay out
One more was needed to pull the measure over in the following pages is that in the last three de-
the threshold. Arise was convinced they had a cades, states and localities have pursued sharply
winning strategy. Instead of going after compre- divergent tax policies that have a direct impact
hensive reform of income, sales, and property on the resources poor households hold at the end
taxes, they proposed a modest plan: eliminating of the day.
the state portion of the grocery tax, expanding Our purpose here is to spotlight taxation’s
family-friendly income tax deductions, and cap- unrecognized impacts on the nation’s poor. We
ping a lopsided deduction that benefited those argue that much of the action in these impacts
at high incomes. The requisite three-fifths vote lies at the state and local level, rather than at the
seemed within reach; only one more senator was federal level where most social science research
needed. To the lasting disappointment of the on taxation has focused. In particular, we exam-
reformers, none of the opponents broke ranks. ine the role of regressive taxation and argue that
The measure failed once again, leaving Alabama, overreliance on sales taxes has had a punishing
one of the poorest states in the country, with the impact on the poor in many states. Particular re-
dubious distinction of being one of only two gions of the country have, for historical reasons,
states to exact the full sales tax on food. For the moved in that direction over time. Although the
poor and the near-poor, the consequences are origins of the divergence lie in the distant past,
dire. this is not a historical artifact; the regional diver-
Alabama and Mississippi exact the highest gence in tax regimes has actually grown over the
tax levies from the poor, but they are not alone. last 25 years. Hence, it makes a big difference for
Most of the Southern states rely on regressive a poor household to be located in the South—
taxation of this kind and have done so for de- and increasingly the West—as opposed to the
cades. Property taxes are a mirror image in that Northeast, even after controlling for the cost of
part of the country; they are very low and have living, the racial composition of the states, pov-
been that way for a long time. Accordingly, com- erty levels, or state expenditures. As such, these
pared to states in the Northeast, wealthy people states offer a poignant lesson in “what not to do”
are able to escape tax burdens while the poor are when it comes to alleviating poverty through the
burdened with them, to their detriment in terms tax system.
of longevity, health, education, and family struc-
ture. While these outcomes are familiar to most
researchers interested in inequality and poverty, Diverging Destinies
the role that regressive taxation plays in their dis- Southern states have long favored the use of sales
tribution and magnitude is a little-studied aspect tax for funding the public sector, in sharp con-
of fiscal sociology. We argue it is time to “bring trast to most northern states, which rely more
tax back in.” heavily on progressive property and income
Although it is commonplace to point to the taxes. Indeed, this story begins at the end of
states as “laboratories of democracy” and cruci- Radical Reconstruction when the Deep South
bles of policy experimentation, the policy do- first shifted to sales tax and then began to impose
mains that attract attention tend to be limited supermajority rules and constitutional limits on
to education, housing, and welfare policy. Yet spending to limit the use of any other kind of
in some respects, the most critical instrument taxation. Because of these divergent trajectories,
 Newman and O’Brien—Taxing the Poor • 371

Figure 43.1. State and Local Income Tax Liability for a Family of Three at the Poverty Line, by Region,
1982–2008
200
0
Constant Dollars
-200
-400
-600

1980 1990 2000 2010

Year

Northeast Midwest South West

the states entered the modern era with markedly Income Tax Credit (EITC) or other refundable
different tax regimes. credit. Figure 43.2 shows the trend in the sales
Estimating the tax burden on the poor is a tax; here we see that the Northeastern and Mid-
complicated endeavor (a process we discuss at western states are relatively flat, whereas the
length in Taxing the Poor). Briefly, using data West and particularly the South have increased
drawn from state income tax returns, adminis- the sales tax burden on the poor. Summarizing
trative data on sales tax rates, and information across all these data (Figure 43.3) unearths a re-
on patterns of household consumption, we es- markable trend: Since 1982, the total state and
timated the income and sales tax burden for a local tax liability (income and sales) for a family
hypothetical family of three for every state. We of three at the poverty line has increased in the
then repeated this exercise using data for every Southern and, to a lesser extent, Western states
year from 1982 to 2008. This provides us with while the burden has declined in the Midwest-
a picture for the taxes paid by the “Jones family” ern and, most dramatically, in the Northeastern
both across states and over time. states.
Figure 43.1 displays the average state income
tax paid by our hypothetical family in each re-
gion from 1982 to 2008. Here we see that over Taxation and Poverty-Related
the past 25 years, many Southern and West- Outcomes
ern states increased income taxes on the poor But does this increasing regional divergence
while most Northeastern and Midwestern states and—in the South and West—increasing tax
significantly reduced the tax burden on those burden on the poor help us to understand re-
below the poverty line. By 2008, most poor gional variation in the poverty-related outcomes
families in the Northeast actually had a negative we care about?
state income tax liability, that is, they actually The poor in the Southern region are at a
received a rebate in the form of a State Earned greater disadvantage than their counterparts
372 • Part V: Poverty and the Underclass

in other parts of the country because the state assess. That tax policy is making a bad situation
and local tax burdens they face make them even considerably worse. Our statistical models show
poorer. A particularly pernicious driver of these that across time, states that increase taxes on the
differences lies in the sales taxes the poor must poor do considerably worse on aggregate mea-
pay (alongside the non-poor), especially the food sures of health (mortality) and crime (aggregate
taxes that many Southern states and localities property and violent crime rates) as well as social

Figure 43.2. State and Local Sales Tax Liability for a Family of Three at the Poverty Line, by Region,
1982–2008
400
300
Constant Dollars
200
100
0

1980 1990 2000 2010

Year

Northeast Midwest South West

Figure 43.3. Total State and Local Tax Liability for a Family of Three at the Poverty Line, by Region,
1982–2008
400
300
Constant Dollars
200
0
-400 -200

1980 1990 2000 2010

Year

Northeast Midwest South West

 Newman and O’Brien—Taxing the Poor • 373

indicators (high school completion and out-of-

wedlock births). What Is to Be Done?
Why should taxation make such a difference? At the federal level, the poor have fared relatively
Well, money matters. When the poor lose more well over the last 25 years. The advent of the
of their income to taxation, it weakens their al- Federal EITC has had a salutary impact on the
ready vulnerable position, which has repercus- nation’s low-income households, dropping their
sions for aggregate measures of crime, health, federal tax burden by nearly 200 percent. To the
family formation, and educational attainment. extent that social scientists focus on taxation,
But we were also interested in exploring if there this is the story we know, and it is both positive
were any knock-on consequences to the type of and universal (for working poor families). The
tax used to extract revenue from the poor. As we real action in terms of divergence is to be found
noted above, many Southern states are unique in the states. There we see profound differences
in that they tax food for home consumption that hit Southern and increasingly Western
(Figure 43.4). We know there is a connection pocketbooks much harder.
between the price of food and obesity; when The reliance on regressive taxation—under-
faced with a limited budget, low-income fam- girded by supermajority rules and limits on tax-
ilies typically opt for cheaper, high-calorie, ation and spending—has hammered the poor
low-quality foodstuffs over relatively more ex- both by taking money from the pockets that can
pensive, healthful, and fresh products. By in- least afford it and by stripping states of the reve-
creasing the cost of each item, a sales tax may nue they need to run first-class institutions that
therefore lead some low-income families to con- could potentially equalize or at least take a stab
sume less nutritious food in an effort to stretch at improving the public services that support
their budget. Sales taxes on food, therefore, may better life chances for the poor. This strategy
be one explanation for the fact that obesity rates is self-defeating; it is costing these states more
are higher in the South than in the rest of the every year in lives lost prematurely, young peo-
country. We conducted another series of statis- ple descending into poverty in greater numbers
tical analyses to test this proposition and found than they should, and crime, which takes a toll
that indeed, high sales taxes on food is related on everyone. It will take a monumental effort to
to higher rates of obesity in the population of change course and place the South on a trajec-
Southern states. tory that is dependent on the federal government

Figure 43.4. Tax Treatment of Food for Home Consumption, 2008

Food exempt from sales tax

Food tax with credit or rebate

Food taxed at reduced rate

(range 1–4%)

Food subject to local tax

Food subject to full state/local tax

374 • Part V: Poverty and the Underclass

and better able to support the infrastructure and Assuming this is not genuinely on the table,
human capital requirements of its citizens. The we note that it is the refundability of the Fed-
cost of doing otherwise is simply too high for eral EITC that makes it so important. It actu-
the people of the Southern states, and it may be- ally puts much-needed dollars in the hands of
come so in the West as well. low-income families. Twenty-four states have
What, then, are the alternatives? We offer recognized the wisdom of the approach and have
two ways of thinking about policy directions. enacted statutes of their own, but they are not all
One emphasizes redressing some of the most created equal. Some are more generous than oth-
regressive aspects of existing sales tax, without ers in that they send families checks because the
contemplating its elimination. The other takes liability falls below zero. Encouraging (and even
into account the challenge of reversing course at rewarding) the other 26 states to enact their own
the state level and hence focuses instead on re- refundable EITCs would also be a boon, particu-
ducing the influence of states altogether on the larly in the Southern states. Spreading childcare
fundamental social policies which every Ameri- tax credits and making them refundable as well
can should be entitled to, regardless of their res- would have similar positive consequences.
idential location. Still, making tax systems more progressive will
Most states make use of sales taxes, but not not solve the central problem facing the South
all of them are as punishing to the poor as the and increasingly the Western states, which, fol-
Southern states. Some achieve a more equitable lowing California’s Proposition 13, are becoming
solution by exempting basic necessities like food ever more reliant on regressive taxation. In the
for home consumption. We should mount a na- South, the situation is more egregious because
tional campaign to follow suit in the Southern there is too much need and too few resources.
states and any other region of the country where A more progressive taxation scheme may be able
basic foodstuffs are taxed and continue with an to generate slightly more revenue—and will en-
initiative designed to eliminate sales taxes on sure that the poor are able to hold on to more
medicine and clothing. of their earnings—but it will not be enough to
Another means to redress regressivity is to fund social programs and education on par with
follow the lead of states that rebate sales tax the rest of the country without significant federal
on a means-tested basis—tilting heavily to- intervention.
ward low-income families—or refund money As long as major social policies remain in the
through earned income credits. At a minimum, hands of the states, we are likely to see interre-
one could use the annual consumer expenditure gional inequality persist. This is only partly be-
survey to calculate the amount the Jones family cause the states that are least generous are also
would need to pay for a healthy diet and rebate the most conservative. The current system re-
at least that much to households that are cur- quires the poorest states to provide for the poor-
rently paying into the system. est citizens by generating revenue for programs
The Federal EITC is by far the most effective like Temporary Assistance for Needy Families
way to put resources in the hands of working (TANF) and Medicaid from the weakest tax base.
poor families. Some recent deficit-reduction What are the prospects for shifting some of the
proposals have proposed to eliminate this federal power to set eligibility and benefit levels federally?
tax provision on the grounds of “shared pain.” America’s federal structure has resulted in
We should recognize, first and foremost, what a 50 distinct welfare states—each with the re-
catastrophe any such move would represent. It sponsibility of providing for its poorest citizens.
would instantly plunge poor families deeper into Wealthier states, blessed with either a deep tax
deprivation by removing one of the most impor- base or fewer needy citizens, or both, can afford
tant and effective redistributive mechanisms we to provide much more than those states bur-
have. dened with the double whammy of poor citizens
 Newman and O’Brien—Taxing the Poor • 375

and, consequently, a shallow tax base. American living. But the basic principle, that all American
social policy in the twenty-first century is largely families are entitled to safety nets of equivalent
a federal story, with Washington playing an in- value, should be made real by taking states out
creasingly central—and equalizing—role in the of the equation. The long history of Social Se-
financing of education, welfare, and health care. curity and the GI Bill, to name two major social
But, as students of welfare reform can attest, policies that have had durable effects on mobility
states continue to play a central role. and economic stability for millions of American
We think that needs to change. Specifically, families, tells us why this is so important. It took
we believe that the major safety net programs, decades to redress the racial inequalities that
particularly TANF and Medicaid, should be emerged in the administration of these critical
regulated and financed at the federal level, just programs because they were left in the hands
like food stamps, Supplemental Security In- of the states (a requirement Southern senators
come (SSI), Medicare, and Social Security. We insisted on if they were not to torpedo the cen-
can follow the advice of the National Research tral provisions of the New Deal). Leaving these
Council’s recommendations, included in a re- decisions in the hands of states and localities in-
port on changing the way we calculate the na- troduces inequalities that punish the poor if they
tional poverty line, and adjust payments to take happen to live in states that are unwilling or less
into account regional differences in the cost of able to address their needs.
376 • Part V: Poverty and the Underclass

Neighborhoods and Segregation

44 • Douglas S. Massey and Nancy A. Denton

American Apartheid
Segregation and the Making of the Underclass

It is quite simple. As soon as there is a group area then all your uncer-
tainties are removed and that is, after all, the primary purpose of this
bill [requiring racial segregation in housing].
—Minister of the Interior, Union of South Africa
legislative debate on the Group Areas Act of 

During the 1970s and 1980s a word disappeared ongoing institutional arrangements and contem-
from the American vocabulary.1 It was not in the porary individual actions. They view segregation
speeches of politicians decrying the multiple ills as an unfortunate holdover from a racist past,
besetting American cities. It was not spoken by one that is fading progressively over time. If ra-
government officials responsible for administer- cial residential segregation persists, they reason,
ing the nation’s social programs. It was not men- it is only because civil rights laws passed during
tioned by journalists reporting on the rising tide the 1960s have not had enough time to work or
of homelessness, drugs, and violence in urban because many blacks still prefer to live in black
America. It was not discussed by foundation ex- neighborhoods. The residential segregation of
ecutives and think-tank experts proposing new blacks is viewed charitably as a “natural” out-
programs for unemployed parents and unwed come of impersonal social and economic forces,
mothers. It was not articulated by civil rights the same forces that produced Italian and Polish
leaders speaking out against the persistence of neighborhoods in the past and that yield Mexi-
racial inequality; and it was nowhere to be found can and Korean areas today.
in the thousands of pages written by social sci- But black segregation is not comparable to
entists on the urban underclass. The word was the limited and transient segregation experi-
segregation. enced by other racial and ethnic groups, now
Most Americans vaguely realize that urban or in the past. No group in the history of the
America is still a residentially segregated society, United States has ever experienced the sustained
but few appreciate the depth of black segrega- high level of residential segregation that has been
tion or the degree to which it is maintained by imposed on blacks in large American cities for

Douglas S. Massey and Nancy A. Denton. “The Missing Link,” in American Apartheid: Segregation and the
Making of the Underclass, pp. 1–9, 12–16, 239–241. Copyright © 1993 by the President and Fellows of Har-
vard College. Reprinted by permission of the publisher, Harvard University Press, Cambridge, Massachusetts.
 Massey and Denton—American Apartheid • 377

the past fifty years. This extreme racial isolation private achievements. For the past twenty years
did not just happen; it was manufactured by this fundamental fact has been swept under the
whites through a series of self-conscious actions rug by policymakers, scholars, and theorists of
and purposeful institutional arrangements that the urban underclass. Segregation is the missing
continue today. Not only is the depth of black link in prior attempts to understand the plight of
segregation unprecedented and utterly unique the urban poor. As long as blacks continue to be
compared with that of other groups, but it shows segregated in American cities, the United States
little sign of change with the passage of time or cannot be called a race-blind society.
improvements in socioeconomic status.
If policymakers, scholars, and the public have
been reluctant to acknowledge segregation’s per- The Forgotten Factor
sistence, they have likewise been blind to its The present myopia regarding segregation is all
consequences for American blacks. Residential the more startling because it once figured prom-
segregation is not a neutral fact; it systematically inently in theories of racial inequality. Indeed,
undermines the social and economic well-being the ghetto was once seen as central to black sub-
of blacks in the United States. Because of racial jugation in the United States. In 1944 Gunnar
segregation, a significant share of black Amer- Myrdal wrote in An American Dilemma that
ica is condemned to experience a social envi- residential segregation “is basic in a mechanical
ronment where poverty and joblessness are the sense. It exerts its influence in an indirect and
norm, where a majority of children are born out impersonal way: because Negro people do not
of wedlock, where most families are on welfare, live near white people, they cannot . . . associate
where educational failure prevails, and where so- with each other in the many activities founded
cial and physical deterioration abound. Through on common neighborhood. Residential segrega-
prolonged exposure to such an environment, tion . . . becomes reflected in uni-racial schools,
black chances for social and economic success hospitals, and other institutions” and creates “an
are drastically reduced. artificial city . . . that permits any prejudice on
Deleterious neighborhood conditions are the part of public officials to be freely vented on
built into the structure of the black community. Negroes without hurting whites.”2
They occur because segregation concentrates Kenneth B. Clark, who worked with Gunnar
poverty to build a set of mutually reinforcing Myrdal as a student and later applied his research
and self-feeding spirals of decline into black skills in the landmark Brown v. Topeka school
neighborhoods. When economic dislocations integration case, placed residential segregation
deprive a segregated group of employment and at the heart of the U.S. system of racial oppres-
increase its rate of poverty, socioeconomic depri- sion. In Dark Ghetto, written in 1965, he argued
vation inevitably becomes more concentrated that “the dark ghetto’s invisible walls have been
in neighborhoods where that group lives. The erected by the white society, by those who have
damaging social consequences that follow from power, both to confine those who have no power
increased poverty are spatially concentrated as and to perpetuate their powerlessness. The dark
well, creating uniquely disadvantaged environ- ghettos are social, political, educational, and—
ments that become progressively isolated—geo- above all—economic colonies. Their inhabitants
graphically, socially, and economically—from are subject peoples, victims of the greed, cruelty,
the rest of society. insensitivity, guilt, and fear of their masters.”3
The effect of segregation on black well-being Public recognition of segregation’s role in per-
is structural, not individual. Residential segrega- petuating racial inequality was galvanized in the
tion lies beyond the ability of any individual to late 1960s by the riots that erupted in the na-
change; it constrains black life chances irrespec- tion’s ghettos. In their aftermath, President Lyn-
tive of personal traits, individual motivations, or don B. Johnson appointed a commission chaired
378 • Part V: Poverty and the Underclass

by Governor Otto Kerner of Illinois to identify theoretical scrutiny on everything from culture
the causes of the violence and to propose poli- to family structure, to institutional racism, to
cies to prevent its recurrence. The Kerner Com- federal welfare systems. Few people spoke of ra-
mission released its report in March 1968 with cial segregation as a problem or acknowledged its
the shocking admonition that the United States persisting consequences. By the end of the 1970s
was “moving toward two societies, one black, residential segregation became the forgotten fac-
one white—separate and unequal.”4 Prominent tor in American race relations.10
among the causes that the commission identified While public discourse on race and poverty
for this growing racial inequality was residential became more acrimonious and more focused on
segregation. divisive issues such as school busing, racial quo-
In stark, blunt language, the Kerner Commis- tas, welfare, and affirmative action, conditions
sion informed white Americans that “discrimina- in the nation’s ghettos steadily deteriorated.11
tion and segregation have long permeated much By the end of the 1970s, the image of poor mi-
of American life; they now threaten the future nority families mired in an endless cycle of un-
of every American.”5 “Segregation and poverty employment, unwed childbearing, illiteracy, and
have created in the racial ghetto a destructive dependency had coalesced into a compelling and
environment totally unknown to most white powerful concept: the urban underclass.12 In the
Americans. What white Americans have never view of many middle-class whites, inner cities
fully understood—but what the Negro can never had come to house a large population of poorly
forget—is that white society is deeply implicated educated single mothers and jobless men—
in the ghetto. White institutions created it, white mostly black and Puerto Rican—who were un-
institutions maintain it, and white society con- likely to exit poverty and become self-sufficient.
dones it.”6 In the ensuing national debate on the causes for
The report argued that to continue present this persistent poverty, four theoretical explana-
policies was “to make permanent the division tions gradually emerged: culture, racism, eco-
of our country into two societies; one, largely nomics, and welfare.
Negro and poor, located in the central cities; the Cultural explanations for the underclass can
other, predominantly white and affluent, located be traced to the work of Oscar Lewis, who iden-
in the suburbs.”7 Commission members rejected tified a “culture of poverty” that he felt promoted
a strategy of ghetto enrichment coupled with patterns of behavior inconsistent with socioeco-
abandonment of efforts to integrate, an approach nomic advancement.13 According to Lewis, this
they saw “as another way of choosing a perma- culture originated in endemic unemployment
nently divided country.”8 Rather, they insisted and chronic social immobility, and provided an
that the only reasonable choice for America was ideology that allowed poor people to cope with
“a policy which combines ghetto enrichment feelings of hopelessness and despair that arose
with programs designed to encourage integra- because their chances for socioeconomic success
tion of substantial numbers of Negroes into the were remote. In individuals, this culture was
society outside the ghetto.”9 typified by a lack of impulse control, a strong
America chose differently. Following the present-time orientation, and little ability to
passage of the Fair Housing Act in 1968, the defer gratification. Among families, it yielded
problem of housing discrimination was declared an absence of childhood, an early initiation into
solved, and residential segregation dropped off sex, a prevalence of free marital unions, and a
the national agenda. Civil rights leaders stopped high incidence of abandonment of mothers and
pressing for the enforcement of open housing, children.
political leaders increasingly debated employ- Although Lewis explicitly connected the
ment and educational policies rather than hous- emergence of these cultural patterns to struc-
ing integration, and academicians focused their tural conditions in society, he argued that once
 Massey and Denton—American Apartheid • 379

the culture of poverty was established, it became attention by focusing on a third possible cause of
an independent cause of persistent poverty. This poverty: government welfare policy. According
idea was further elaborated in 1965 by the Har- to Charles Murray, the creation of the underclass
vard sociologist and then Assistant Secretary of was rooted in the liberal welfare state.19 Federal
Labor Daniel Patrick Moynihan, who in a con- antipoverty programs altered the incentives gov-
fidential report to the President focused on the erning the behavior of poor men and women,
relationship between male unemployment, fam- reducing the desirability of marriage, increasing
ily instability, and the intergenerational trans- the benefits of unwed childbearing, lowering the
mission of poverty, a process he labeled a “tangle attractiveness of menial labor, and ultimately re-
of pathology.”14 He warned that because of the sulted in greater poverty.
structural absence of employment in the ghetto, A slightly different attack on the welfare state
the black family was disintegrating in a way that was launched by Lawrence Mead, who argued
threatened the fabric of community life. that it was not the generosity but the permis-
When these ideas were transmitted through siveness of the U.S. welfare system that was at
the press, both popular and scholarly, the con- fault.20 Jobless men and unwed mothers should
nection between culture and economic structure be required to display “good citizenship” before
was somehow lost, and the argument was pop- being supported by the state. By not requiring
ularly perceived to be that “people were poor anything of the poor, Mead argued, the welfare
because they had a defective culture.” This po- state undermined their independence and com-
sition was later explicitly adopted by the conser- petence, thereby perpetuating their poverty.
vative theorist Edward Banfield, who argued that This conservative reasoning was subsequently
lower-class culture—with its limited time hori- attacked by liberal social scientists, led princi-
zon, impulsive need for gratification, and psy- pally by the sociologist William Julius Wilson,
chological self-doubt—was primarily responsible who had long been arguing for the increasing
for persistent urban poverty.15 He believed that importance of class over race in understand-
these cultural traits were largely imported, aris- ing the social and economic problems facing
ing primarily because cities attracted lower-class blacks.21 In his 1987 book The Truly Disad-
migrants. vantaged, Wilson argued that persistent urban
The culture-of-poverty argument was strongly poverty stemmed primarily from the structural
criticized by liberal theorists as a self-serving ide- transformation of the inner-city economy.22 The
ology that “blamed the victim.”16 In the ensuing decline of manufacturing, the suburbanization
wave of reaction, black families were viewed not of employment, and the rise of a low-wage ser-
as weak but, on the contrary, as resilient and vice sector dramatically reduced the number of
well-adapted survivors in an oppressive and ra- city jobs that paid wages sufficient to support
cially prejudiced society.17 Black disadvantages a family, which led to high rates of joblessness
were attributed not to a defective culture but among minorities and a shrinking pool of “mar-
to the persistence of institutional racism in the riageable” men (those financially able to support
United States. According to theorists of the un- a family). Marriage thus became less attractive
derclass such as Douglas Glasgow and Alphonso to poor women, unwed childbearing increased,
Pinkney, the black urban underclass came about and female-headed families proliferated. Blacks
because deeply imbedded racist practices within suffered disproportionately from these trends
American institutions—particularly schools and because, owing to past discrimination, they were
the economy—effectively kept blacks poor and concentrated in locations and occupations par-
dependent.18 ticularly affected by economic restructuring.
As the debate on culture versus racism Wilson argued that these economic changes
ground to a halt during the late 1970s, con- were accompanied by an increase in the spa-
servative theorists increasingly captured public tial concentration of poverty within black
380 • Part V: Poverty and the Underclass

neighborhoods. This new geography of poverty, We join earlier scholars in rejecting the view
he felt, was enabled by the civil rights revolution that poor urban blacks have an autonomous
of the 1960s, which provided middle-class blacks “culture of poverty” that explains their failure
with new opportunities outside the ghetto.23 to achieve socioeconomic success in American
The out-migration of middle-class families from society. We argue instead that residential seg-
ghetto areas left behind a destitute community regation has been instrumental in creating a
lacking the institutions, resources, and values structural niche within which a deleterious set
necessary for success in post-industrial society. of attitudes and behaviors—a culture of segre-
The urban underclass thus arose from a complex gation—has arisen and flourished. Segregation
interplay of civil rights policy, economic restruc- created the structural conditions for the emer-
turing, and a historical legacy of discrimination. gence of an oppositional culture that devalues
Theoretical concepts such as the culture of work, schooling, and marriage and that stresses
poverty, institutional racism, welfare disincen- attitudes and behaviors that are antithetical and
tives, and structural economic change have all often hostile to success in the larger economy.
been widely debated. None of these explanations, Although poor black neighborhoods still contain
however, considers residential segregation to be many people who lead conventional, productive
an important contributing cause of urban pov- lives, their example has been overshadowed in
erty and the underclass. In their principal works, recent years by a growing concentration of poor,
Murray and Mead do not mention segregation welfare-dependent families that is an inevitable
at all;24 and Wilson refers to racial segregation result of residential segregation.
only as a historical legacy from the past, not as an We readily agree with Douglas, Pinkney, and
outcome that is institutionally supported and ac- others that racial discrimination is widespread
tively created today.25 Although Lewis mentions and may even be institutionalized within large
segregation sporadically in his writings, it is not sectors of American society, including the labor
assigned a central role in the set of structural fac- market, the educational system, and the wel-
tors responsible for the culture of poverty, and fare bureaucracy. We argue, however, that this
Banfield ignores it entirely. Glasgow, Pinkney, view of black subjugation is incomplete without
and other theorists of institutional racism men- understanding the special role that residential
tion the ghetto frequently, but generally call not segregation plays in enabling all other forms of
for residential desegregation but for race-specific racial oppression. Residential segregation is the
policies to combat the effects of discrimination institutional apparatus that supports other ra-
in the schools and labor markets. In general, cially discriminatory processes and binds them
then, contemporary theorists of urban poverty together into a coherent and uniquely effective
do not see high levels of black-white segregation system of racial subordination. Until the black
as particularly relevant to understanding the un- ghetto is dismantled as a basic institution of
derclass or alleviating urban poverty.26 American urban life, progress ameliorating ra-
The purpose of this book is to redirect the cial inequality in other arenas will be slow, fitful,
focus of public debate back to issues of race and and incomplete.
racial segregation, and to suggest that they should We also agree with William Wilson’s basic
be fundamental to thinking about the status of argument that the structural transformation of
black Americans and the origins of the urban un- the urban economy undermined economic sup-
derclass. Our quarrel is less with any of the pre- ports for the black community during the 1970s
vailing theories of urban poverty than with their and 1980s.27 We argue, however, that in the
systematic failure to consider the important role absence of segregation, these structural changes
that segregation has played in mediating, exac- would not have produced the disastrous social
erbating, and ultimately amplifying the harmful and economic outcomes observed in inner cities
social and economic processes they treat. during these decades. Although rates of black
 Massey and Denton—American Apartheid • 381

poverty were driven up by the economic dislo- among blacks in response to the economic dislo-
cations Wilson identifies, it was segregation that cations of the 1970s and 1980s, so did the use of
confined the increased deprivation to a small welfare programs. Because of racial segregation,
number of densely settled, tightly packed, and however, the higher levels of welfare receipt were
geographically isolated areas. confined to a small number of isolated, all-black
Wilson also argues that concentrated poverty neighborhoods. By promoting the spatial con-
arose because the civil rights revolution allowed centration of welfare use, therefore, segregation
middle-class blacks to move out of the ghetto. created a residential environment within which
Although we remain open to the possibility that welfare dependency was the norm, leading to the
class-selective migration did occur,28 we argue intergenerational transmission and broader per-
that concentrated poverty would have hap- petuation of urban poverty.
pened during the 1970s with or without black
middle-class migration. Our principal objection
to Wilson’s focus on middle-class out-migration Coming to Terms with American
is not that it did not occur, but that it is misdi- Apartheid
rected: focusing on the flight of the black middle Our fundamental argument is that racial segre-
class deflects attention from the real issue, which gation—and its characteristic institutional form,
is the limitation of black residential options the black ghetto—are the key structural factors
through segregation. responsible for the perpetuation of black poverty
Middle-class households—whether they are in the United States. Residential segregation is
black, Mexican, Italian, Jewish, or Polish—al- the principal organizational feature of American
ways try to escape the poor. But only blacks society that is responsible for the creation of the
must attempt their escape within a highly seg- urban underclass. . . . It can be shown that any
regated, racially segmented housing market. Be- increase in the poverty rate of a residentially seg-
cause of segregation, middle-class blacks are less regated group leads to an immediate and auto-
able to escape than other groups, and as a result matic increase in the geographic concentration
are exposed to more poverty. At the same time, of poverty. When the rate of minority poverty is
because of segregation no one will move into increased under conditions of high segregation,
a poor black neighborhood except other poor all of the increase is absorbed by a small number
blacks. Thus both middle-class blacks and poor of neighborhoods. When the same increase in
blacks lose compared with the poor and mid- poverty occurs in an integrated group, the added
dle class of other groups: poor blacks live under poverty is spread evenly throughout the urban
unrivaled concentrations of poverty and affluent area, and the neighborhood environment that
blacks live in neighborhoods that are far less ad- group members face does not change much.
vantageous than those experienced by the middle During the 1970s and 1980s, therefore, when
class of other groups. urban economic restructuring and inflation
Finally, we concede Murray’s general point drove up rates of black and Hispanic poverty in
that federal welfare policies are linked to the rise many urban areas, underclass communities were
of the urban underclass, but we disagree with created only where increased minority poverty
his specific hypothesis that generous welfare coincided with a high degree of segregation—
payments, by themselves, discouraged employ- principally in older metropolitan areas of the
ment, encouraged unwed childbearing, under- northeast and the midwest. Among Hispanics,
mined the strength of the family, and thereby only Puerto Ricans developed underclass com-
caused persistent poverty.29 We argue instead munities, because only they were highly seg-
that welfare payments were only harmful to the regated; and this high degree of segregation is
socioeconomic well-being of groups that were directly attributable to the fact that a large pro-
residentially segregated. As poverty rates rose portion of Puerto Ricans are of African origin.
382 • Part V: Poverty and the Underclass

The interaction of intense segregation and such as those observed during the 1970s can
high poverty leaves black neighborhoods ex- only lead to a concentration of housing aban-
tremely vulnerable to fluctuations in the urban donment, crime, and social disorder, pushing
economy, because any dislocation that causes poor black neighborhoods beyond the threshold
an upward shift in black poverty rates will also of stability.
produce a rapid change in the concentration of By building physical decay, crime, and social
poverty and, hence, a dramatic shift in the social disorder into the residential structure of black
and economic composition of black neighbor- communities, segregation creates a harsh and
hoods. The concentration of poverty, for exam- extremely disadvantaged environment to which
ple, is associated with the wholesale withdrawal ghetto blacks must adapt. In concentrating pov-
of commercial institutions and the deterioration erty, moreover, segregation also concentrates
or elimination of goods and services distributed conditions such as drug use, joblessness, welfare
through the market. dependency, teenage childbearing, and unwed
Neighborhoods, of course, are dynamic and parenthood, producing a social context where
constantly changing, and given the high rates of these conditions are not only common but the
residential turnover characteristic of contempo- norm. By adapting to this social environment,
rary American cities, their well-being depends to ghetto dwellers evolve a set of behaviors, attitudes,
a great extent on the characteristics and actions and expectations that are sharply at variance with
of their residents. Decisions taken by one actor those common in the rest of American society.
affect the subsequent decisions of others in the As a direct result of the high degree of racial
neighborhood. In this way isolated actions affect and class isolation created by segregation, for ex-
the well-being of the community and alter the ample, Black English has become progressively
stability of the neighborhood. more distant from Standard American English,
Because of this feedback between individual and its speakers are at a clear disadvantage in
and collective behavior, neighborhood stability U.S. schools and labor markets. Moreover, the
is characterized by a series of thresholds, beyond isolation and intense poverty of the ghetto pro-
which various self-perpetuating processes of vides a supportive structural niche for the emer-
decay take hold. Above these thresholds, each gence of an “oppositional culture” that inverts
actor who makes a decision that undermines the values of middle-class society. Anthropolo-
neighborhood well-being makes it increasingly gists have found that young people in the ghetto
likely that other actors will do the same. Each experience strong peer pressure not to succeed
property owner who decides not to invest in up- in school, which severely limits their prospects
keep and maintenance, for example, lowers the for social mobility in the larger society. Quanti-
incentive for others to maintain their properties. tative research shows that growing up in a ghetto
Likewise, each new crime promotes psycholog- neighborhood increases the likelihood of drop-
ical and physical withdrawal from public life, ping out of high school, reduces the probability
which reduces vigilance within the neighbor- of attending college, lowers the likelihood of em-
hood and undermines the capacity for collective ployment, reduces income earned as an adult,
organization, making additional criminal activ- and increases the risk of teenage childbearing
ity more likely. and unwed pregnancy.
Segregation increases the susceptibility Segregation also has profound political con-
of neighborhoods to these spirals of decline. sequences for blacks, because it so isolates them
During periods of economic dislocation, a rising geographically that they are the only ones who
concentration of black poverty is associated with benefit from public expenditures in their neigh-
the simultaneous concentration of other nega- borhoods. The relative integration of most ethnic
tive social and economic conditions. Given the groups means that jobs or services allocated to
high levels of racial segregation characteristic of them will generally benefit several other groups
American urban areas, increases in black poverty at the same time. Integration thus creates a basis
 Massey and Denton—American Apartheid • 383

for political coalitions and pluralist politics, and enforcement provisions as its price of enactment,
most ethnic groups that seek public resources yielding a Fair Housing Act that was structurally
are able to find coalition partners because other flawed and all but doomed to fail. As documen-
groups can anticipate sharing the benefits. That tation of the law’s defects accumulated in multi-
blacks are the only ones to benefit from resources ple congressional hearings, government reports,
allocated to the ghetto—and are the only ones and scholarly studies, little was done to repair the
harmed when resources are removed—makes it situation until 1988, when a series of scandals
difficult for them to find partners for political and political errors by the Reagan administration
coalitions. Although segregation paradoxically finally enabled a significant strengthening of fed-
makes it easier for blacks to elect representatives, eral antidiscrimination law.
it limits their political influence and marginalizes Yet even more must be done to prevent the
them within the American polity. Segregation permanent bifurcation of the United States into
prevents blacks from participating in pluralist black and white societies that are separate and
politics based on mutual self-interest. unequal. As of 1990, levels of racial segregation
Because of the close connection between were still extraordinarily high in the nation’s
social and spatial mobility, segregation also per- large urban areas, particularly those of the north.
petuates poverty. One of the primary means by Segregation has remained high because fair hous-
which individuals improve their life chances— ing enforcement relies too heavily on the private
and those of their children—is by moving to efforts of individual victims of discrimination.
neighborhoods with higher home values, safer Whereas the processes that perpetuate segrega-
streets, higher-quality schools, and better ser- tion are entrenched and institutionalized, fair
vices. As groups move up the socioeconomic housing enforcement is individual, sporadic,
ladder, they typically move up the residential and confined to a small number of isolated cases.
hierarchy as well, and in doing so they not only As long as the Fair Housing Act is enforced
improve their standard of living but also enhance individually rather than systemically, it is un-
their chances for future success. Barriers to spa- likely to be effective in overcoming the structural
tial mobility are barriers to social mobility, and arrangements that support segregation and sus-
by confining blacks to a small set of relatively tain the ghetto. Until the government throws its
disadvantaged neighborhoods, segregation con- considerable institutional weight behind efforts
stitutes a very powerful impediment to black to dismantle the ghetto, racial segregation will
socioeconomic progress. persist. . . .
Despite the obvious deleterious consequences Ultimately, however, dismantling the ghetto
of black spatial isolation, policymakers have not and ending the long reign of racial segregation
paid much attention to segregation as a contrib- will require more than specific bureaucratic re-
uting cause of urban poverty and have not taken forms; it requires a moral commitment that
effective steps to dismantle the ghetto. Indeed, white America has historically lacked. The seg-
for most of the past two decades public policies regation of American blacks was no historical
tolerated and even supported the perpetuation accident; it was brought about by actions and
of segregation in American urban areas. Al- practices that had the passive acceptance, if not
though many political initiatives were launched the active support, of most whites in the United
to combat discrimination and prejudice in the States. Although America’s apartheid may not be
housing and banking industries, each legislative rooted in the legal strictures of its South African
or judicial act was fought tenaciously by a pow- relative, it is no less effective in perpetuating ra-
erful array of people who believed in or benefited cial inequality, and whites are no less culpable
from the status quo. for the socioeconomic deprivation that results.
Although a comprehensive open hous- As in South Africa, residential segrega-
ing bill finally passed Congress under unusual tion in the United States provides a firm basis
circumstances in 1968, it was stripped of its for a broader system of racial injustice. The
384 • Part V: Poverty and the Underclass

geographic isolation of Africans within a nar- 5. Ibid.

rowly circumscribed portion of the urban en- 6. Ibid., p. 2.
vironment—whether African townships or 7. Ibid., p. 22.
American ghettos—forces blacks to live under 8. Ibid.
9. Ibid.
extraordinarily harsh conditions and to endure
10. A few scholars attempted to keep the Kerner
a social world where poverty is endemic, infra- Commission’s call for desegregation alive, but their
structure is inadequate, education is lacking, voices have largely been unheeded in the ongoing
families are fragmented, and crime and violence debate. Thomas Pettigrew has continued to assert
are rampant.30 Moreover, segregation confines the central importance of residential segregation,
these unpleasant by-products of racial oppres- calling it the “linchpin” of American race relations;
sion to an isolated portion of the urban geog- see “Racial Change and Social Policy,” Annals of
raphy far removed from the experience of most the American Academy of Political and Social Science
whites. Resting on a foundation of segregation, 441 (1979):114–31. Gary Orfield has repeatedly
pointed out segregation’s deleterious effects on black
apartheid not only denies blacks their rights as
prospects for education, employment, and socioeco-
citizens but forces them to bear the social costs nomic mobility; see “Separate Societies: Have the
of their own victimization. Kerner Warnings Come True?” in Fred R. Harris
Although Americans have been quick to crit- and Roger W. Wilkins, eds., Quiet Riots: Race and
icize the apartheid system of South Africa, they Poverty in the United States (New York: Pantheon
have been reluctant to acknowledge the conse- Books, 1988), pp. 100–122; and “Ghettoization
quences of their own institutionalized system of and Its Alternatives,” in Paul E. Peterson, ed., The
racial separation. The topic of segregation has New Urban Reality (Washington, D.C.: Brookings
virtually disappeared from public policy debates; Institution, 1985), pp. 161–96.
11. See Thomas B. Edsall and Mary D. Edsall,
it has vanished from the list of issues on the civil
Chain Reaction: The Impact of Race, Rights, and Taxes
rights agenda; and it has been ignored by social
on American Politics (New York: Norton, 1991).
scientists spinning endless theories of the under- 12. For an informative history of the evolution of
class. Residential segregation has become the for- the concept of the underclass, see Michael B. Katz,
gotten factor of American race relations, a minor The Undeserving Poor: From the War on Poverty to the
footnote in the ongoing debate on the urban War on Welfare (New York: Pantheon, 1989), pp.
underclass. Until policymakers, social scientists, 185–235.
and private citizens recognize the crucial role of 13. Oscar Lewis, La Vida: A Puerto Rican Family
America’s own apartheid in perpetuating urban in the Culture of Poverty—San Juan and New York
(New York: Random House, 1965); “The Culture
poverty and racial injustice, the United States
of Poverty,” Scientific American 215 (1966): 19–25;
will remain a deeply divided and very troubled
“The Culture of Poverty,” in Daniel P. Moynihan,
society.31 ed., On Understanding Poverty: Perspectives from the
Social Sciences (New York: Basic Books, 1968), pp.
NOTES 187–220.
1. Epigraph from Edgar H. Brookes, Apartheid: A 14. The complete text of this report is reprinted
Documentary Study of Modern South Africa (London: in Lee Rainwater and William L. Yancey, The Moyni-
Routledge and Kegan Paul, 1968), p. 142. han Report and the Politics of Controversy (Cambridge:
2. Gunnar Myrdal, An American Dilemma, vol. 1 MIT Press, 1967), pp. 39–125.
(New York: Harper and Brothers, 1944), p. 618; see 15. Edward C. Banfield, The Unheavenly City
also Walter A. Jackson, Gunnar Myrdal and America’s (Boston: Little, Brown, 1970).
Conscience (Chapel Hill: University of North Caro- 16. William Ryan, Blaming the Victim (New
lina Press, 1990), pp. 88–271. York: Random House, 1971).
3. Kenneth B. Clark, Dark Ghetto: Dilemmas of 17. Carol Stack, All Our Kin: Strategies of Survival
Social Power (New York: Harper and Row, 1965), in a Black Community (New York: Harper and Row,
p. 11. 1974).
4. U.S. National Advisory Commission on Civil 18. Douglas C. Glasgow, The Black Underclass:
Disorders, The Kerner Report (New York: Pantheon Poverty, Unemployment, and Entrapment of Ghetto
Books, 1988), p. 1.
 Massey and Denton—American Apartheid • 385

Youth (New York: Vintage, 1981), p. 11; Alphonso 28. The evidence on the extent of middle-class
Pinkney, The Myth of Black Progress (Cambridge: out-migration from ghetto areas is inconclusive. Be-
Cambridge University Press, 1984), pp. 78–80. cause racial segregation does not decline with rising
19. Charles Murray, Losing Ground: American socioeconomic status, out-movement from poor
Social Policy, 1950–1980 (New York: Basic Books, black neighborhoods certainly has not been to white
1984). areas. When Kathryn P. Nelson measured rates of
20. Lawrence M. Mead, Beyond Entitlement: The black out-migration from local “zones” within forty
Social Obligations of Citizenship (New York: Free metropolitan areas, however, she found higher
Press, 1986). rates of out-movement for middle- and upper-class
21. William Julius Wilson, The Declining Signifi- blacks compared with poor blacks; but her “zones”
cance of Race: Blacks and Changing American Institu- contained more than 100,000 inhabitants, making
tions (Chicago: University of Chicago Press, 1978). them considerably larger than neighborhoods (see
22. William Julius Wilson, The Truly Disadvan- “Racial Segregation, Mobility, and Poverty Concen-
taged: The Inner City, the Underclass, and Public Policy tration,” paper presented at the annual meetings of
(Chicago: University of Chicago Press, 1987), pp. the Population Association of America, Washing-
1–108. ton, D.C., March 19–23, 1991). In contrast, Ed-
23. Ibid., pp. 49–62. ward Gramlich and Deborah Laren found that poor
24. The subject indices of Losing Ground and and middle-class blacks displayed about the same
Beyond Entitlement contain no references at all to likelihood of out-migration from poor census tracts
residential segregation. (see “Geographic Mobility and Persistent Poverty,”
25. The subject index of The Truly Disadvan- Department of Economics, University of Michigan,
taged contains two references to pre-1960s Jim Crow Ann Arbor, 1990).
segregation. 29. See Eggers and Massey, “A Longitudinal
26. Again with the exception of Thomas Petti- Analysis of Urban Poverty.”
grew and Gary Orfield. 30. See International Defense and Aid Fund
27. We have published several studies document- for Southern Africa, Apartheid: The Facts (London:
ing how the decline of manufacturing, the subur- United Nations Centre against Apartheid, 1983),
banization of jobs, and the rise of low-wage service pp. 15–26.
employment eliminated high-paying jobs for manual 31. We are not the first to notice the striking
workers, drove up rates of black male unemploy- parallel between the institutionalized system of ra-
ment, and reduced the attractiveness of marriage to cial segregation in U.S. cities and the organized,
black women, thereby contributing to a prolifera- state-sponsored system of racial repression in South
tion of female-headed families and persistent pov- Africa. See John H. Denton, Apartheid American
erty. See Mitchell L. Eggers and Douglas S. Massey, Style (Berkeley, Calif.: Diablo Press, 1967); James A.
“The Structural Determinants of Urban Poverty,” Kushner, “Apartheid in America: An Historical and
Social Science Research 20 (1991): 217–55; Mitchell Legal Analysis of Contemporary Racial Residential
L. Eggers and Douglas S. Massey, “A Longitudinal Segregation in the United States,” Howard Law Jour-
Analysis of Urban Poverty: Blacks in U.S. Metropol- nal 22 (1979):547–60.
itan Areas between 1970 and 1980,” Social Science
Research 21 (1992): 175–203.
386 • Part V: Poverty and the Underclass

45 • Robert Sampson

Legacies of Inequality

Daniel Patrick Moynihan became famous by call- and dislocations that Moynihan highlighted are
ing attention to the racialized configuration of the undeniably clustered for many of those in soci-
American city. With an unflinching eye, Moyni- ety who lack the resources to escape communi-
han’s view of history led him to forecast key social ties of disadvantage. My reading is thus that he
trends and identify multiple challenges confront- was pointing to a neighborhood tangle of inequal-
ing urban America in the decade following Brown ity—one inextricably tied to race—and that he
vs. Board of Education. Nothing in his view was emphasized its durability without government in-
more salient than the “tangle of pathology” in tervention.3 Moynihan was insistent on this point,
the black ghetto, by which he meant the strong and in the most important chapter of his famous
clustering of numerous social problems.1 The lan- (but never published and often unread) report
guage was blunt and ignited a firestorm of protest “The Case for National Action” he argued against
that was not soon forgotten.2 To this day the term the “variable” approach and called for a more ho-
“pathology” is avoided by social scientists. listic approach to both explanation and policy:
Yet many of the underlying facts Moynihan
confronted remain unchanged today. This chap- It is our view that the problem is so interre-
ter will show that while specific neighborhoods lated, one thing with another, that any list of
have shifted or traded places, with poverty mov- program proposals would necessarily be in-
ing outward from the inner city, the general force complete, and would distract attention from
of ecological concentration and neighborhood the main point of interrelatedness (emphasis
racial stratification continues to have a strong added). We have shown a clear relation be-
grip on the city. tween male employment, for example, and
My goal in this chapter is to get beyond the the number of welfare dependent children,
pathology debate and focus on the larger issues employment in turn reflects educational
that motivated Moynihan. My take is that Moyni- achievement, which depends in large part on
han wanted social policy to focus primarily on family stability, which reflects employment.
the inequality that resided at the structural and Where we should break into this cycle, and
social-ecological level and not just the individual how are the most difficult domestic questions
or family. That is a core idea that runs counter to facing the United States.4
much of the individual reductionism (or behav-
iorism) that characterizes contemporary social sci- More important, Moynihan coupled inter-
ence and public policy. A host of social problems connectedness with durability and suggested

Robert Sampson. “Legacies of Inequality,” in Great American City: Chicago and the Enduring Neighborhood
Effect (Chicago: University of Chicago Press, 2012), pp. 97–120, 445–448, 493, 495–500, 504, 506, 508–509,
511–512, 515–521, 523–524. Copyright © 2012 by The University of Chicago. Used by permission of the
University of Chicago Press.
 Sampson—Legacies of Inequality • 387

the idea of reinforcing cycles of disadvantage or and causality in the social world. Let me now
what might better be called poverty traps.5 He lay out the essential facts of spatially inscribed
noted in the beginning of his report that “so inequality.
long as this situation persists, the cycle of pov-
erty and disadvantage will continue to repeat it-
self.” In the final policy chapter, he warned that Things Go Together
“three centuries of injustice have brought about What are often considered sources of compro-
deep-seated structural distortions in the life of mised well-being—such as unemployment, seg-
the Negro American” and that the “present tan- regation, poverty, and family disruption—are
gle of pathology is capable of perpetuating itself equally clustered in space. The basics of these
without assistance from the white world. The patterns are common to many cities and extend
cycle can be broken only if these distortions are across multiple ecological units of analysis rang-
set right.”6 Otherwise, he argued that once set ing from census tracts to metropolitan areas and
in motion, racially linked poverty is a trap that even states.7 I illustrate and update this phenom-
reinforces itself and can be broken only by struc- enon here by considering empirical patterns for
tural interventions. a core set of socioeconomic indicators in both
This structural logic emerges that implies Chicago and the U.S. I argue that disadvantage
three broad ideas or theses that are interlinked: is not encompassed in a single characteristic but
1. The “tangle of pathology,” or what today we rather is a synergistic composite of social factors
would call social dislocations or social prob- that mark the qualitative aspects of growing up
lems, has a deep neighborhood structure in severely disadvantaged neighborhoods. My
and connection to concentrated inequality. colleagues and I investigated this idea by examin-
2. Neighborhood social disadvantage has ing six characteristics of census tracts nationwide,
durable properties and trends to repeat taken from the 1990 and 2000 censuses, to cre-
itself, and because of racial segregation is ate a measure of concentrated disadvantage: wel-
most pronounced in the black commu- fare receipt, poverty, unemployment, female-headed
nity. I would add a related implication or households, racial composition (percentage black),
subthesis: black children are singularly ex- and density of children. These indicators all
posed to the cumulative effects of structural loaded on a single principal component we called
disadvantage in ways that reinforce the “concentrated disadvantage” in both decades
cycle. and both Chicago and the rest of the U.S. (alto-
3. The “poverty trap” cycle can be broken only gether some sixty-five thousand census tracts).8
with structural interventions of the sort that The main difference of note between the U.S.
government or other large organizational and Chicago neighborhoods is that the expo-
units (e.g., foundations) are equipped to sure of children under eighteen years of age to
carry out. concentrated economic disadvantage and racial
segregation is more pronounced in Chicago. The
I emphasize the big picture by showing both data thus confirm that neighborhoods that are
stability and change in neighborhood stratifi- both black and poor, and that are characterized
cation and by showing that Chicago is not as by high unemployment and female-headed fam-
unique as some claim. Inequality is durable ilies, are ecologically distinct, a characteristic that
and multiplex but not inevitable or natural, is not simply the same thing as low economic
generating direct implications for theories of status. In this pattern Chicago is not alone.
community-level processes, the social repro- To probe the implications of this point in
duction of inequality, individual selection a different but more concrete way, I calculated
bias, neighborhood interventions (thesis 3), the per capita income in the year 2000 in black
388 • Part V: Poverty and the Underclass

compared to white neighborhoods in Chicago the unprecedented rate of 197 inmates per one
(defined here as census tracts with 75 percent or hundred thousand persons in 1990 and the
more of each group). The result was that not one previously unimaginable rate of 504 inmates
white community experiences what is most typ- per one hundred thousand persons in 2008.10
ical for those residing in segregated black areas Scholars have broadly described this national
with respect to the basics of income—the en- phenomenon as mass incarceration.11
tire distribution for white communities (mean = Yet, in fact, mass incarceration has a local
$42,508) sits to the right of the mean per capita concentration. Obscured by a focus on national
income of black communities ($12,276). Trying trends are profound variations in incarceration
to estimate the effect of concentrated disadvan- rates by communities within cities, especially by
tage on whites is thus tantamount to estimating their racial composition. Like the geographically
a phantom reality. Latino and mixed-race com- concentrated nature of criminal offending by
munities are better off, but even there the mean individuals, a small proportion of communities
is well below white areas. The bottom line is that bear the disproportionate brunt of U.S. crime
the racial stratification of Chicago’s urban land- policy’s experiment with mass incarceration. In
scape precludes a simple estimation of a single Chicago, we can see this by calculating the rate
causal effect of disadvantage for all racial groups. of incarceration for each census tract in Chicago
It follows from these data and the spatial logic and comparing it to the level of concentrated
of the “tangle of pathology” thesis that the knot disadvantage and percent black.12 The correla-
of inequality is far greater in the black commu- tion of incarceration rates in 1990–95 with con-
nity than the white community. Another way to centrated disadvantage and percent black at the
test this thesis is to examine how the relationship neighborhood level in 1990 is 0.82 and 0.75,
between the unemployment rate and poverty respectively. The corresponding correlations for
varies by the racial status of the neighborhood. 2000 disadvantage and percent black predicting
I find that there is no significant relationship incarceration in 2000–2005 are 0.80 and 0.74.
between unemployment rates and economic de- Because of this ecological concentration, large
pendence in white communities—surprisingly, swaths of the city, especially in the southwest
in fact, a simple flat line emerges. Yet in mixed and northwest, are relatively untouched by the
and minority areas the correlation is strongly imprisonment boom no matter which period
positive and significant.9 This finding demon- we examine, with almost no one sent to prison
strates the much tighter connection among in some areas. By contrast, there is a dense and
economic-related indicators in minority areas, a spatially contiguous cluster of areas in the near
process that contributes to the synergistic inter- west and south central areas of Chicago that have
section of racial segregation with concentrated rates of incarceration many times higher. This
disadvantage. pattern of concentration results in a racialized
configuration of the city—the correlation be-
tween concentrated disadvantage and the incar-
Concentrated Incarceration: A New ceration rate in white areas is no different than
“Pathology”? zero, but in black, Latino, and mixed areas the
There is a new kind of social distortion that has correlation is over 0.6 (p < 0.01).
come to characterize the national scene. From Once again, then, “things go together,” but
the 1920s to the early 1970s, the incarceration the strength of the connection, especially among
rate in the United States averaged 110 inmates social dislocations like poverty, crime, infant
per one hundred thousand persons. Beginning mortality, low birth weight, incarceration, and
in the mid-1970s, the incarceration rate in the unemployment, is decidedly stronger in commu-
United States accelerated dramatically, reaching nities of color.
 Sampson—Legacies of Inequality • 389

Figure 45.1. Persistence and Change in Poverty Across Forty Years

How stable are social processes over time? What

Poverty Traps: Social
predicts neighborhood change? Does it take a
Transformations Revisited
planned intervention to bring about true struc-
The continued concentration of poverty, crime, tural change? This section examines stability and
incarceration, and general disadvantage is an change over important periods of Chicago’s his-
intriguing puzzle, because residential moves en- tory from the 1960s to the present.
sure that different individuals make up the same Figure 45.1 extends the examination of the
neighborhoods over time. How much stability or social transmission of concentrated poverty
change in concentrated neighborhood inequality from 1960—the dawn of the civil rights era—to
is generated over time given the dynamic flows the census of 2000. It shifts the analysis to the
of people? The question is not how individu- community-area level and gives names to places
als change but whether and to what extent the (e.g., Hyde Park, Grand Boulevard, South Shore,
geographic concentration of poverty became and Lincoln Park) that serve as markers of dif-
increasingly entrenched in certain urban neigh- ference and symbolic value.13 Community areas
borhoods. We can extend this line of reasoning have political and social reputations that continue
to other dimensions as well, perhaps even those to this day.14 It is also of relevance that Wilson’s
that can change quickly. For example, how te- thesis of concentration effects was developed on
nacious is the hold of high rates of violence on data from the community areas of Chicago.
select urban neighborhoods? When crime goes Figure 45.1 shows that concentrated pov-
down, or up, are the trends similar for all places? erty is surprisingly stable over the long period
390 • Part V: Poverty and the Underclass

Figure 45.2. Stable Asymmetry of Racial Change Across Forty Years

of forty years. In a time of rapid social change, The link of race to poverty suggests a logical
riots, crime-rate swings, racial change, economic next step. To further elaborate the persistence
recessions, and gentrification at the end of the of racial segregation I examine changes in racial
twentieth century, concentrated poverty is stub- composition across the same four-decade period,
born in its concentration. Note the high correla- plotting in Figure 45.2 the percentage of a neigh-
tion between poverty in 1960 and 2000 of 0.78. borhood’s population that was black in 2000
For over forty years, Riverdale and communities against percent black in 1960. This graph shows
in the “black belt” on the South Side stand out as that where racial change is abundant it is struc-
“poverty carriers” in contrast to the largely white tured in an asymmetric way, yielding four types
and well-off communities such as Norwood and of neighborhoods. The first two reflect durable
Edison Park in the upper west corner of the city. segregation—those that are stably black (upper
Beverly on the Southwest Side remained eco- right) and those that are stably white (lower
nomically healthy despite its proximity to the left). The third type reflects transitional neigh-
black belt and increasing racial heterogeneity. borhoods that went either from all white to black
Overall, then, the data reveal that if we know (“white flight”) or partially black to segregated
the poverty level of a neighborhood at one point black (up the left side and across the top). That is
in time it is possible to quite accurately forecast not a surprise, although it may be hard for people
its outcome up to four decades later. It is not in Chicago to imagine that communities such
obvious that this should be possible given that as West Englewood and Auburn Gresham were
people move in and out of neighborhoods that once white. The fourth type is the small number
are constantly in flux. How do we account for of integrated communities that have remained
this stability amid change? so, such as Hyde Park and the Near West Side.
 Sampson—Legacies of Inequality • 391

What is startling is the missing type—in the emphasized and that was documented again
third-largest city in America, not one neighbor- in Wilson’s When Work Disappears.17 This can
hood transitioned from predominantly black to pre- be demonstrated by considering the prediction
dominantly white. The lack of observations in the of neighborhood unemployment rates in 2000
lower right quadrant of the graph indicates that from unemployment rates in 1990 for predom-
none of the areas that had large percentages of inantly black neighborhoods compared to pre-
black population in 1960 lost significant shares. dominantly white neighborhoods. In Chicago,
In fact, there appears to be a threshold effect of the correlation of unemployment over time is
around 50 percent black, above which all neigh- relatively high (0.64) and significant in the black
borhoods either maintained or increased their community but nonexistent in the white com-
share of black population. Only one or two com- munity (-0.05). The distributions are virtually
munities maintained a reasonable black-white incomparable—most white neighborhoods sit to
integration over this period.15 Figure 45.2 thus the left of where the black distribution starts.18
tells a story of change within a relatively sta- The data thus confirm that blacks and whites
ble social-ecological structure: there were great live in different social worlds of work.
shifts in neighborhood racial composition from So too does imprisonment, a new “social
1960–2000, but neighborhoods that were ini- distortion” that did not exist in Moynihan’s
tially black stayed that way over time, while at time to anywhere near the extent that it does
the same time many areas of the city remained today. There has always been incarceration, of
off limits to blacks. In light of this pattern, Chi- course, but in the post-Moynihan era the U.S.
cago has the distinction of being not just one has witnessed a form of what many call mass in-
of the most racially segregated cities in America, carceration. I argued earlier, however, that mass
but it appears to be durably so.16 . . . incarceration is “placed” in certain communities,
a pattern that I now show visually in Figure 45.3.
The stability of incarceration’s spatial distribu-
Racial Inequality tion over time is virtual unity (correlation of
Another implication of Moynihan’s reasoning 0.98), but what is most striking in Figure 45.3 is
is that the grip of neighborhood inequality— the incommensurability of mass incarceration in
the “tangle”—is amplified, more durable, and black compared to white communities. There is
qualitatively distinct in the black community. no overlap in the distribution and a clear gap be-
The expectation is that we should see more per- tween the “highest incarcerated white commu-
sistence of disadvantage in minority or black nity” and the “lowest” for black communities.
communities than white ones and the durability West Garfield Park and East Garfield Park on
of nonoverlapping neighborhood distributions. the city’s West Side stand out as the epicenter of
Chicago, with its segregated urban structure, the modern incarceration regime. West Garfield
affords an opportunity to assess this notion Park has a rate forty-two times higher than the
within relatively homogeneous subgroups. I dis- highest-ranked white community on incarcera-
aggregated the city into four race/ethnic strata— tion, Clearing (4,226 vs. 103 per 100,000). This
predominantly white, black, Latino, and other is a staggering differential.
(mixed). In minority areas the continuity of In a similar vein, the data reveal that the rate
concentrated disadvantage during the 1990s is of male unemployment is much more predic-
much higher, but the result is especially strik- tive of incarceration in the black communities
ing when comparing segregated black neighbor- of Figure 45.3 than white communities. The
hoods, where the correlation is 0.83, to white correlation in white communities is nonsignif-
neighborhoods, where the stability correlation is icant, whereas for black communities the male
only 0.24. unemployment rate in 2000 predicts later in-
This divergence is even more evident for the carceration (0.53, p < 0.01). This interaction
durability of unemployment that Moynihan suggests that incarceration is part of the cycle of
392 • Part V: Poverty and the Underclass

Figure 45.3. Incarceration in Black and White: Persistence of Spatially Concentrated Incarceration by the
Racial Status of Chicago Communities, 1990–2005

disadvantage behind the “poverty traps,” traps question is the hardest of all, and there are no
that find their most intense manifestation in easy answers. This is especially true in a culture
segregated and racially isolated communities. in which freedom of choice to move anywhere
There is likely a reciprocal feedback, in which one can afford is highly valued.
imprisonment removes males from the commu- Although the evidence is insufficient at this
nity while at the same time unemployed males point to support a strong causal claim, it is no-
drive the incarceration “input,” thus reinforcing table that communities undergoing robust in-
a vicious cycle. This modern-day distortion im- terventions saw unexpected declines in the rate
plicates, and complicates, policy operations of of concentrated poverty as predicted from 1990
the government. . . . levels alone. Looking again at Figure 45.1, and
considering changes from 1990 to 2000 in par-
ticular, communities below the regression line
Structural Interventions: Breaking of continuity share a similar profile. Places like
the Trap? Douglas, Oakland, and the Near South Side were
Inequality is stubbornly persistent even though mainly poor black communities that witnessed
change is constantly taking place. How can structural changes from the outside, including
the durable inequality I have documented be the dismantling of segregated high-rise public
changed in possibly transformative ways? This housing by the Chicago Housing Authority and
 Sampson—Legacies of Inequality • 393

considerable investment by the city and institu-

tions such as the University of Chicago in both Conclusion
the physical infrastructure and educational sys- Glasses half full are of course also half empty.
tem (e.g., charter schools). The Near West Side How then should we view the twin processes of
also saw considerable investment by the city and stability and change? The first lesson is that cities
economic developers, and along with the Near and neighborhoods are constantly changing, and
South and Near North sides, saw a relative im- in no inherent direction. Chicago lost population
provement in median income in the very recent in the 1970s, but reversed itself in the 1990s, for
past. example. Many other cities are back from the
Individual choice in migration decisions gloom-and-doom years of the ’70s and ’80s as
should not be overlooked either. There has been well, with many high-crime neighborhoods see-
a return to cities across the country, and Chi- ing a turn to increased safety. Yet we have simul-
cago is no exception. The city saw growth in its taneously seen inequality increasing in society at
population overall and a number of formerly large and evidence of neighborhood social repro-
poor areas became attractive to bohemians and duction that disproportionately affects minority
the so-called creative class. Gentrification may groups.19 Poverty and its correlates are persistent
be overstated in some cases, but it is real. Other in terms of neighborhood concentration, espe-
communities, for example, Wicker Park, Buck- cially for black areas. Despite urban social trans-
town, and the Near South Side, saw gentrifica- formation in the post-Moynihan era, in other
tion that was not driven by large-scale policy words, most neighborhoods remained stable in
intervention. The West Loop and Near West their relative economic standing even with the
Side have also been hot, with trendy restaurants inflow and outflow of individual residents. The
sprouting like wildflowers, Oprah’s studio, and second lesson is thus that there is an enduring
considerable gentrification that has crept into vulnerability to certain neighborhoods that is not
the heart of the Latino Pilsen district. simply a result of the current income of residents.
In brief, the data suggest that in an ongoing Although not studied directly in this chapter,
process with durable or self-reinforcing proper- neighborhoods possess reputations that, when
ties, macro interventions of the sort government coupled with certain residential selection deci-
is uniquely suited to mount, or support through sions, reproduce existing patterns of inequality.
zoning changes, need to be at the top of our pol- The change that does occur reveals strong
icy considerations and evaluation efforts. More- patterns of asymmetry by race and class, sug-
over, from the mid to late 1990s to the present gesting that once a neighborhood is beyond a
was an especially propitious time for structural certain threshold or “tipping point” of either
interventions because a natural change has been percent black or percent poor—but especially
taking place across much of the country—crime the former—further change is in the direction
and violence rates have been falling sharply. Re- of greater racial homogeneity and more poverty.
cent times have been less plagued by crime and Not one neighborhood in Chicago more than
a renewed appreciation for the viability of cities 50 percent black in 1960 became predominantly
has emerged. In places like Chicago and cities white forty years later (Figure 45.2). By contrast,
nationwide, neighborhood renewal is occurring a large number of white neighborhoods turned
in the heart of some of what were previously black even as the polar extremes (all-black and
high-crime, very poor, and often devastated all-white neighborhoods) remained the dominant
areas. No one factor can be credited with this pattern. Neighborhoods in general also tended to
change, but it appears that smart local decisions stay in the same poverty category or move to a
that capitalize on secular changes are benefiting higher-poverty category over time. Therefore the
cities even though the benefits remain spatially third lesson I have shown is that upgrading (or
concentrated. gentrification) does not topple stability of the
394 • Part V: Poverty and the Underclass

ecological structure even in the 1990–2007 pe- 7. RJ Sampson, JD Morenoff, and T

riod. Whether we consider concentrated disad- Gannon-Rowley. 2002. “Assessing neighborhood
vantage in Chicago or the U.S. as a whole, this effects: Social processes and new directions in
pattern holds—Chicago is not an outlier. research.”
8. RJ Sampson, P Sharkey, and SW Raudenbush.
These findings resonate with the earlier con-
2008. “Durable effects of concentrated disadvantage
cerns introduced at the outset by Moynihan and on verbal ability among African-American children.”
[other] scholars. The pattern may not be encour- p. 848, table 1.
aging, but it does signal a distressing reality that 9. The correlation is 0.59 (p<.01) in mixed areas.
some forty-five years since the original Moyni- In black areas it is even larger, at 0.78 (p<.01).
han report, the language has changed but the 10. Bureau of Justice Statistics. 2010. “Prisoners
same questions must be raised again. The jury is in 2008.” http://bis.ojp.usdoj.gov/content/glance/
still out, but the preliminary data presented here incrt.cfm.
suggest a glimmer of hope that cycles of pov- 11. Western, Bruce. 2006. Punishment and
Inequality in America. New York: Russell Sage
erty can be broken, providing examples of poor
Foundation.
communities that are repositioning themselves 12. We measure the imprisonment rate by the
on an upward trajectory. But how pervasive the number of unique Chicago-residing felony defen-
phenomenon is remains to be seen, and research dants sentenced to the Illinois Department of Cor-
so far has not systematically examined the conse- rections from the Circuit Court of Cook County
quences of change in one community on changes divided by the number of Chicago inhabitants ages
in other, perhaps even distant communities that eighteen to sixty-four in the 2000 census.
may be receiving new burdens. 13. Wilson, Wiliam J., and Richard P. Taub.
2006. There Goes the Neighborhood. New York: Ran-
NOTES dom House, Inc.
1. Moynihan, Daniel P. 1965. The Criminal Area. 14. Suttles, Gerald D. 1990. The Man-Made
London: Routledge and Kegan Paul Ltd. City: The Land-Use Confidence Game in Chicago.
2. Massey, Douglas, and Robert J. Sampson. Chicago: University of Chicago Press.
2009. “Moynihan Redux: Legacies and Lessons.” 15. RP Taub, DG Taylor, and JD Dunham.
The Annals of the American Academy of Political and 1987. Paths of Neighborhood Change: Race and Crime
Social Science. in Urban America. Chicago: University of Chicago
3. Sampson, Robert J. 2009. “Neighborhood Press.
Social Capital as Differential Social Organization 16. DS Massey, GA Condran, and NA Denton.
Resident and Leadership Dimensions.” American 1987. “The effect of residential segregation on black
Behavioral Scientist. social and economic well-being.”
4. Moynihan, Daniel P. 1965. The Criminal Area. 17. Wilson, William J. 1996. When Work Disap-
London: Routledge and Kegan Paul Ltd. Chapter 3, pears: The World of the New Urban Poor. New York:
page 1. Knopf.
5. Bowels, Samuel, Steven Durlauf, and Karla 18. Sampson, Robert J. 2009. “Neighborhood
Hoff. 2006. Poverty Traps. New York: Russell Sage Social Capital as Differential Social Organization
Foundation. Resident and Leadership Dimensions.” American
6. Moynihan, Daniel P. 1965. The Criminal Area. Behavioral Scientist.
London: Routledge and Kegan Paul Ltd. Chapter 5, 19. Reardon, Sean F., and Kendra Bischoff. 2011.
page 1. “Income Inequality and Income Segregation.” Amer-
ican Journal of Sociology 116(4):1092–1153.
 DeLuca and Rosenbaum—Does Changing Neighborhoods Change Lives? • 395

46 • Stefanie DeLuca and James E. Rosenbaum

Does Changing Neighborhoods Change Lives?

Neighborhood effects on the lives of families and can have devastating consequences for young
young people have long been a focus of sociolog- people’s chances of graduating from high school
ical research (Mayer and Jencks 1990; Sampson (Wodtke, Harding, and Elwert 2011). Several
2012). Theoretically, neighborhoods are impor- other studies suggest that one pathway through
tant contexts for socialization and development, which neighborhoods affect children’s cognitive
as well as places where we find some of the struc- outcomes is exposure to violent crime (Sharkey
tural mechanisms that determine inequality and 2010; Burdick-Will et al. 2011). Finally, the
opportunity. Such theoretical expectations have ten-year experimental evaluation of the MTO
led many social scientists to surmise that neigh- program showed that escaping high-poverty
borhoods are important determinants of life neighborhoods not only improved mental health
chances. for women, but also reduced extreme obesity
Until recently the prevailing view has none- and diabetes (Ludwig et al. 2011, 2012). This
theless been that, however plausible neighbor- new evidence has led to a shift in social science
hood effects might be, the evidence for them sensibilities and suggests that we should revisit
is inconclusive. The mixed findings from the policies and interventions designed to improve
Moving to Opportunity (MTO) experimen- neighborhood context.
tal demonstration were especially important in The debates about neighborhood effects has
creating skepticism about the effects of com- raged for so long because of two types of meth-
munity-level factors and the housing policy to odological and empirical complications. First,
improve them. despite relatively high levels of residential mobil-
But that was then. In recent years cutting-edge ity in the United States, we see little variation in
methodological advances, groundbreaking data the types of communities that low-income, mi-
collection efforts, and new long-term findings nority families inhabit. Poor families are trapped
from the MTO program have produced con- in dangerous neighborhoods, and their children
vincing evidence that the consequences of liv- are stuck in poor schools; recent work finds that
ing in high-poverty, violent neighborhoods are most black children who grow up in the most
significant, just as has long been assumed. For disadvantaged communities will remain in these
example, we now know that living in a neighbor- communities as adults (South and Deane 1993;
hood of concentrated disadvantage significantly South and Crowder 1997; Massey and Denton
reduces the verbal ability of African American 1993; Sharkey 2008). Therefore, we do not nor-
children, to a degree equivalent to missing a year mally get the chance to observe how a different
of school (Sampson, Sharkey, and Raudenbush environment might affect their life chances.
2008). The negative effects of living in poor The second research challenge is that families
communities also accumulate over time and choose neighborhoods, and the characteristics

Original article prepared for this fourth edition of Social Stratification: Class, Race, and Gender in Sociological
Perspective, edited by David B. Grusky. Copyright © 2014 by Westview Press.
396 • Part V: Poverty and the Underclass

of families that lead them to choose certain fortunes of the families who moved as part of
neighborhoods are also likely to affect family this intervention.
and child well-being. This leads to the selection
problem (i.e., “endogeneity”), which plagues our
attempts to recover causal estimates of environ- The Gautreaux Program
mental effects. Recent advances in quantitative As a result of a 1976 Supreme Court decision,
models have provided some ways to reduce such Hills v. Gautreaux 425 U.S. 284, the Gautreaux
challenges in the estimation of neighborhood program allowed low-income, black public
effects (Harding 2003; Sampson, Sharkey, and housing residents in Chicago to receive Section
Raudenbush 2008; Sharkey 2010). 8 housing certificates (or vouchers) and move to
However, there have been opportunities to private-sector apartments either in mostly white
study what happens when parents and children suburbs or within the city (Polikoff 2006). Be-
experience moderate to radical changes in their tween 1976 and 1998 more than seven thousand
neighborhood or schooling environments. Res- families participated; more than half moved to
idential mobility programs, in which poor fam- suburban communities. Because of its design,
ilies relocate to opportunity-rich communities the Gautreaux program presents an unusual op-
by using housing vouchers, provide one way to portunity, allowing us to examine whether indi-
examine variation in neighborhood conditions. vidual outcomes change when low-income black
Here we describe one particularly important families move to safer neighborhoods with better
mobility plan, Chicago’s Gautreaux program, labor markets and higher quality schools.
the predecessor to the MTO experiment. The Gautreaux participants circumvented the
early results of Gautreaux were promising, typical barriers to living in suburbs, not by their
showing gains in education, neighborhood jobs, finances, or values, but by being accepted
quality, and employment for families who into the program and their quasi-random as-
moved to more affluent suburban areas, as de- signment to the suburbs. The program provided
scribed in detail below. It has been argued, how- housing subsidy vouchers and housing support
ever, that the original design of Gautreaux was services, but not employment or transporta-
lacking in rigor and that a randomized field trial tion assistance. Unlike in the usual situation of
in other cities would be a better test of the early working-class blacks living in working-class sub-
results. Indeed, such skepticism was warranted: urbs, Gautreaux permitted low-income blacks
that is, Gautreaux generally showed larger ef- to live in middle- and upper-income white
fects on both individual and neighborhood suburbs. Participants moved to more than 115
outcomes than MTO (cf. DeLuca et al. 2010; suburbs throughout the six counties surround-
Rosenbaum and Zuberi 2010). In part this is ing Chicago. Suburbs with a population that was
what led some scholars to conclude that neigh- more than 30 percent black were excluded by
borhood effects did not exist beyond the effects the consent decree. Few very high-rent suburbs
of family background. In light of the recent re- were excluded by funding limitations of Section
search cited previously, much of which shows 8 certificates.
significant and moderate to large neighborhood
effects on children’s educational outcomes and
mothers’ mental health, it now makes sense to Early Findings
revisit the results of this earlier housing inter- Early research on Gautreaux had shown sig-
vention. After all, it seems that some of the nificant relationships between placement
original conclusions of the Gautreaux research neighborhoods and subsequent gains in em-
might have been on the mark, and accordingly ployment and education for families who used
a close look at Gautreaux is warranted. Here a voucher to move to the suburbs. A study of
we examine a decade of research following the 330 Gautreaux mothers in the early 1990s found
 DeLuca and Rosenbaum—Does Changing Neighborhoods Change Lives? • 397

that suburban movers had higher employment neighborhood (DeLuca and Rosenbaum 2003),
than city movers, but not higher earnings, and and mothers continued to live in areas with
the employment difference was especially large much lower poverty rates and higher household
for adults who were unemployed prior to the incomes (Keels et al. 2005).
move (Rosenbaum 1995). Another study found Individual-level economic outcomes, such as
that as young adults, Gautreaux children who welfare receipt, employment, and earnings, were
moved to the suburbs were more likely than city also influenced by the income and racial char-
movers to graduate from high school, attend acteristics of placement neighborhoods. Women
college, and attend four-year colleges (versus who moved to racially mixed or predominantly
two-year colleges); if they were not in college, white neighborhoods with higher levels of socio-
they were more likely to be employed and have economic resources did better than their counter-
jobs with better pay and benefits (Rosenbaum parts in areas with low resources and more black
1995). Analyses indicated that children mov- residents (DeLuca and Rosenbaum 2003, Men-
ing to suburbs were just as likely to interact denhall, DeLuca, and Duncan 2006). Research
with neighbors as city movers, but the suburb on the children of the original Gautreaux families
movers interacted with white children, whereas has demonstrated that the neighborhoods where
city movers interacted mostly with black chil- they resided in the late 1990s were substantially
dren. The program seems to have been effective more integrated than their overwhelmingly mi-
at integrating low-income black children into nority origin neighborhoods (Keels 2008a). How-
middle-class white suburbs. Although the cur- ever, relocating to lower poverty, more integrated
riculum level of suburban schools was often far areas had mixed effects on the delinquent behav-
ahead of that in city schools, mothers reported iors and arrest rates of boys and girls. Suburban
that suburban teachers often made extra efforts boys were much less likely to become involved
to help their children catch up with the class. in the criminal justice system, whereas girls who
Initial concerns that these children would not moved to the suburbs were more likely to be con-
be accepted were unsupported by the evidence. victed of criminal offenses (Keels 2008b).

Recent Research How Did Gautreaux “Work”?

To improve upon the design and data quality The findings described here focus on the ad-
of the earlier work, more recent research used vances made in recent quantitative work. We
administrative data to locate recent addresses had employed techniques to approximate the as-
for a 50 percent random sample of Gautreaux sessment of Gautreaux as a “treatment”: a social
movers who had relocated before 1990, as well intervention with effects we might measure with
as to track economic outcomes for mothers.1 In statistical corrections and design comparisons.
addition, multiple Census measures were used However, the stories Gautreaux participants tell
to characterize neighborhoods, and a more com- about their experiences can contribute greatly to
prehensive accounting for preprogram charac- our understanding. The long-term family out-
teristics was employed in the regression models. comes we observed appear to be significantly
The use of administrative records permitted us linked to the mobility program and the charac-
to locate 1,504 of 1,507 families, and we found teristics of the placement neighborhoods. How-
that 66 percent of suburban families remained ever, administrative data cannot show us how
in the suburbs an average of fifteen years after these outcomes occurred or the mechanisms
placement. After controlling for premove indi- through which neighborhoods have their im-
vidual and neighborhood attributes, the racial pact. This is a problem common to neighbor-
composition of the placement neighborhood hood research, which makes improving mobility
predicted the racial composition of the current programs especially difficult. However, in several
398 • Part V: Poverty and the Underclass

qualitative studies (Rosenbaum, Reynolds, and behaving to comply with the social norms of the
DeLuca 2002; Rosenbaum, DeLuca, and Tuck new neighborhoods. Several noted initial difficul-
2005) we analyzed interviews with mothers who ties in adjusting to suburban norms, which were
described how these neighborhoods helped im- unfamiliar and intolerant of some of their prior
prove their lives and those of their children. Was behaviors. These mothers, who had lived all their
the important factor just increasing access to lives in housing projects where these norms did
better resources, or was it necessary to interact not exist, saw benefits to complying with these
with neighbors to obtain the full benefit of these expectations, and they decided to adopt them.
new resources? Some of these normative constraints, such as
We analyzed interviews with 150 Gautreaux low tolerance for drugs and parties, were actually
mothers and found that after the move, they de- liberating, because the trade-off was community
scribed a new sense of efficacy and control over safety. This meant that mothers did not have to
their lives and stated that the major changes in spend all their time watching their children and
their environments helped them see that they could give them more freedom.
had the ability to make improvements in their Similarly, mothers reported social respon-
lives. Certain features of the new suburban siveness from their neighbors. They received the
neighborhoods changed their perception of what benefits of reciprocal relations in child care and
was possible. Specifically, the women reported in neighbors’ concern and watchfulness, pro-
that they felt better about having an address in moting the safety of their children, their prop-
the suburbs and not having to put down a pub- erty, and themselves. They also received favors in
lic housing location on job applications. Other terms of transportation and some acts of charity.
women noted that by moving to areas with more It is remarkable that these new residents, who
white residents, they and their children got to generally differed in race and class from their
know more white people, and racial stereotypes neighbors, were awarded this collective generos-
were debunked. One child whose only exposure ity, and the interviews suggest that it may have
to white people had been those she saw on TV been conditional on their showing a willingness
reported that after moving, she discovered that to abide by community norms.
not all whites looked like TV actors. Most important, the new suburban social
Social interactions with whites allowed some contexts provided a form of capital that en-
of these women to feel that they had more social hanced people’s capabilities. Some mothers
and cultural know-how and to feel much less reported that they could count on neighbors
intimidated by future contexts in which they if their children misbehaved or seemed at risk
might have to interact with whites. In addition, of getting into trouble; their children were sick
working through some of the initial difficulties and couldn’t attend school; or there was some
of the transition to the suburbs allowed these threat to their children, apartments, or them-
women to realize that they could handle man- selves. This was not just interpersonal support;
ageable challenges along the way to better jobs it was systemic and enabled these mothers to
and more schooling. In comparison, the drugs take actions and make commitments that oth-
or gang violence in their old city neighborhoods erwise would have been difficult or risky. For
seemed to be forces too large for them to con- example, some mothers reported willingness
trol and therefore permanent impediments to to take jobs because they could count on a
the improvements they were trying to make in neighbor to watch their children if they were
their lives. These findings suggest to us that one’s late getting home from work. We believe it is
repertoire of capabilities can vary depending on through these mechanisms—some social, some
the type of neighborhood in which one lives and psychological—that the Gautreaux families were
works. able to permanently escape the contexts and
Many of the mothers we interviewed also consequences of segregated poverty and unsafe
noted that they had to change their way of inner-city neighborhoods.
 DeLuca and Rosenbaum—Does Changing Neighborhoods Change Lives? • 399

More recent interviews with Gautreaux few did so, because they were unlikely to ever get
mothers suggest that some aspects of the another.2 As a result, participants’ preferences for
city-suburban divide were also important for placement neighborhoods had relatively little to
shaping how the placement community affected do with where they ended up moving, providing
their children’s behavior (Keels 2008b; Menden- a degree of exogenous variability in neighborhood
hall 2004). City movers placed in both mod- placement that undergirds Gautreaux research.
erate- and low-poverty neighborhoods found Few significant differences were found between
that although their immediate neighborhood suburban and city movers’ individual character-
was safe, the larger community to which their istics, but premove neighborhood attributes show
children had easy access continued to be dan- statistically significant differences on two of nine
gerous. In comparison, children placed in the comparisons. This may indicate selection bias,
suburbs had less direct neighborhood expo- although MTO’s random assignment studies
sure to drugs and illegal activities and attended also find some substantial differences (Goering
higher-performing public schools with greater and Feins 2003, Table 7.1). Several papers have
financial and teacher resources. Interviews re- discussed these issues at length and examine
vealed that affluent suburban neighborhoods multiple neighborhood-level indicators, detailed
also provided substantially fewer opportunities preprogram neighborhood differences, and in-
for involvement in delinquent criminal activities tergenerational effects (DeLuca and Rosenbaum
and gangs. 2003; Keels et al. 2005; Mendenhall, DeLuca,
and Duncan 2006; DeLuca et al. 2010; Keels
2008a and 2008b).
Was Gautreaux a Social Experiment? In contrast MTO, sponsored by the Depart-
Methodologically, we often rely on observational ment of Housing and Urban Development,
data and regression analyses to estimate the “ef- was an experiment, in which low-income fam-
fect” of neighborhood contexts and interventions. ilies were randomly assigned to an experimen-
These approaches have their weaknesses; it is tal group (who moved to low-poverty census
complicated, if not impossible, to recover causal tracts), an open-choice conventional housing
effects when we know that there are unobservable voucher group, or a “no move” control group.
characteristics of families that lead not only to This program was developed to formally test the
their selection of neighborhood, but also to the Gautreaux findings, with more rigorous design
outcomes of interest. As a result, there has been an and pre/postmove data collection. Unfortu-
increased push to employ experimental designs to nately, although MTO was based on a stronger
control for and assign social and economic “treat- design, it was in some respects a weaker “neigh-
ments,” be they neighborhoods, school programs, borhood change treatment.” The Gautreaux pro-
or income subsidies (DeLuca and Dayton 2009). gram moved families an average of twenty-five
Along these lines, the Gautreaux program resem- miles away from their original neighborhoods,
bled a quasi-experiment. Although the program to radically different labor markets, where nearly
was not designed as an experiment, and families all children attended schools with above-average
were not formally randomly assigned to condi- achievement and were too far away to interact
tions, aspects of the program administration with prior friends. In comparison, MTO moved
break the link between family preferences and most families less than ten miles, mostly in the
neighborhood placement. In principle, partici- city; most children attended schools with very
pants could choose where they moved. In prac- low achievement; and many children contin-
tice, qualifying rental units were secured by rental ued interacting with old friends. In addition,
agents working for the Gautreaux program and MTO occurred in the hot labor market of the
offered to families according to their position on late 1990s, and large numbers of families in the
a waiting list, regardless of their locational prefer- control group moved out of high-rise housing
ence. Although participants could refuse an offer, projects through the federal HOPE VI program,
400 • Part V: Poverty and the Underclass

so the control group was experiencing unusual feelings of safety. The MTO program also
benefits and atypical circumstances that may not showed significant reductions in depression and
generalize to more ordinary times (Rosenbaum obesity among mothers and daughters (but no
and Zuberi 2010). difference for sons) (Orr et al. 2003; Ludwig et
Whereas early Gautreaux analyses showed al. 2011). Gautreaux studied neither of these
that suburban children attended much better outcomes.
schools and enjoyed more improvements in edu- When comparing the two programs, it is cru-
cational outcomes than the city movers, those in cial to understand the nature of the comparisons
the MTO program did not attend significantly being made. What policy makers need to know
better schools, and perhaps unsurprisingly, did is how a family fares when given the opportu-
not have better educational outcomes. Com- nity to move to a neighborhood that has lower
pared to the control group, the MTO treatment poverty or is more integrated, compared to what
group showed no difference in test scores, school would have happened to that family had it not
dropout, or self-reported measures of school en- been given that option. Gautreaux research stud-
gagement an average of five years after random ies can only compare subgroups of families that
assignment (Sanbonmatsu et al. 2006). This was moved in conjunction with the program and
so partly because many MTO experimental fam- experienced variation in neighborhood contexts;
ilies sent their children to schools in the same there is no comparison group of similar families
school district (often the same schools), and even who did not move as part of the program. The
when they changed schools, the new ones were evaluation design of MTO is much stronger, be-
not much better than the original ones. This cause it tracked the fortunes of a randomly as-
practice of staying in the same district was driven signed control group of families who expressed
in part by the belief that it would be safer and interest in the program but, owing to the luck
less disruptive to keep children in more familiar of the draw, were not assigned to the move (al-
settings (DeLuca and Rosenblatt 2010). though some managed to move on their own).
Whereas Gautreaux was associated with gains At the same time, however, Gautreaux offered a
in mothers’ employment, the MTO treatment stronger program, which can inform us about
group showed no impact compared with the what happens when families move long distances
control group: both groups showed large gains to radically different neighborhoods, moves that
of comparable magnitude. However, MTO out- changed their social context in many ways: racial
comes were measured in the late 1990s, when a integration, poverty, school quality, labor market
strong labor market and strong welfare reform strength, and safety. Though both interventions
were present, so although MTO found no differ- indicate how powerful the effects of residential
ence between groups, it did find a large employ- moves can be for some families, the differences
ment gain for the control group. One possible in findings indicate the importance of program
interpretation is that virtually everyone who design features, in the influence of historical
could work was doing so, and residential moves context, and concurrent policy effects. For exam-
had no additional effect for that reason. Recent ple, alternative forms of mobility, such as those
qualitative work also suggests that although ex- under the involuntary conditions of HOPE VI,
perimental movers relocated to neighborhoods may have different results.
with higher employment rates, they lacked the Many policy reforms have tried to improve
qualifications for these jobs and found or kept individuals’ education or employability while
better employment prospects in the city (Turney they remain in the same poor schools or labor
et al. 2006). markets, but these reforms have often failed.
Despite the shorter moves and less change in Such policies may be fighting an uphill battle as
social environment, both Gautreaux and MTO long as families remain in the same social con-
found large effects on mothers’ and children’s texts and opportunity structures. In contrast,
 DeLuca and Rosenbaum—Does Changing Neighborhoods Change Lives? • 401

Gautreaux findings suggest that housing policy is dangerous and disadvantaged communities, and
one possible lever to assist poor families, by mov- additional supports may be required to help
ing them into better neighborhoods with much them overcome the challenges they still face as
better schools and labor markets. The initial they navigate labor markets and schools.
gains in neighborhood quality that many of the
Gautreaux families achieved persisted for at least NOTES
one to two decades. Recent research on a new 1. The collection and preparation of the data
used for the more recent studies executed from 2000
residential mobility program in Baltimore shows
to 2005 were initiated by James E. Rosenbaum and
some similarly encouraging results. The Balti-
Stefanie DeLuca and extended in a larger project by
more Mobility Program, born from the remedy Stefanie DeLuca, Greg J. Duncan, James E. Rosen-
of the Thompson v. HUD (2005–2006) deseg- baum, Ruby Mendenhall, and Micere Keels. Various
regation lawsuit, also helps former and current authors collaborated on each study, as cited in the
public housing residents move to low-poverty, references.
mostly white neighborhoods. Counselors help 2. Although only about 20 percent of the eligible
families find housing, prepare them for mov- applicants ended up moving through the program,
ing into suburban areas, and support them after self-selection does not appear to have played a major
they have relocated. New studies of this program role in program take-up (Peterson and Williams
1995). Rather than opting out of the program, most
show that most of the families who relocate to
families that did not move were not offered a hous-
higher-opportunity neighborhoods throughout ing unit and thus not given the chance to participate.
the Baltimore metropolitan area remain in such Housing counselors were forbidden by the consent
neighborhoods over time, and that these neigh- decree from making offers selectively among eligible
borhoods have significantly higher-performing families, and there is no evidence that they did so.
schools than the city neighborhoods families
were living in when they signed up for the pro- REFERENCES
gram (DeLuca and Rosenblatt 2009, 2011). Burdick-Will, Julia, Jens Ludwig, Stephen Rauden-
Both the Gautreaux program and the more bush, Robert Sampson, Lisa Sanbonmatsu, and
Patrick Sharkey. 2011. “Converging Evidence for
recent Baltimore program suggest that it is pos-
Neighborhood Effects on Children’s Test Scores:
sible for low-income black families to escape
An Experimental, Quasi-Experimental, and Ob-
from neighborhoods with concentrated racial servational Comparison.” In Whither Opportu-
segregation, crime, and poverty. These findings nity? Rising Inequality, Schools, and Children’s Life
are of considerable importance, given that most Chances, edited by Greg J. Duncan and Richard
poor minority families rarely acquire housing in Murnane, 255–277. New York: Russell Sage
higher-opportunity neighborhoods, even those Foundation.
receiving traditional housing vouchers (DeLuca, DeLuca, Stefanie, and Elizabeth Dayton. 2009.
Garboden, and Rosenblatt 2013). However, we “Switching Social Contexts: The Effects of
Housing Mobility and School Choice Programs
also find that it can be more difficult to trans-
on Youth Outcomes.” Annual Review of Sociology
late the benefits of sustained residence in better
35: 457–491.
neighborhoods into large-scale economic and DeLuca, Stefanie, Greg Duncan, Ruby Mendenhall,
social success for families and children. Time and Micere Keels. 2010. “Gautreaux Mothers
will tell if the Baltimore program will yield dif- and Their Children: An Update.” Housing Policy
ferent results, and a large mixed-methods study Debate 20: 7–25.
is under way to examine the educational effects DeLuca, Stefanie, Philip Garboden, and Peter
of the housing remedy. It may be that housing Rosenblatt. 2013. “Segregating Shelter: How
mobility is a necessary but insufficient lever for Housing Policies Shape the Residential Locations
improving the lives of poor families trapped in of Low Income Minority Families.” Annals of the
American Academy of Political and Social Science
inner-city neighborhoods. These families have
647: 268–299.
spent years struggling in some of our most
402 • Part V: Poverty and the Underclass

DeLuca, Stefanie, and James Rosenbaum. 2003. “If ———. 2012. “Neighborhood Effects on the
Low-Income Blacks Are Given a Chance to Live Long-Term Well-Being of Low-Income Adults.”
in White Neighborhoods, Will They Stay? Exam- Science 337: 1505–1510.
ining Mobility Patterns with Quasi-Experimental Massey, Douglas, and Nancy Denton. 1993. Amer-
Data. Housing Policy Debate 14: 305–345. ican Apartheid. Cambridge, MA: Harvard Uni-
DeLuca, Stefanie, and Peter Rosenblatt. 2009. versity Press.
“Walking Away from The Wire: Residential Mo- Mayer, Susan, and Christopher Jencks. 1990.
bility and Opportunity in Baltimore.” Paper pre- “Growing Up in Poor Neighborhoods: How
sented at the Annual Meeting of the American Much Does It Matter?” Science 243: 1441.
Sociological Association, San Francisco, CA. Mendenhall, Ruby. 2004. “Black Women in
———. 2010. “Does Moving to Better Neighbor- Gautreaux’s Housing Desegregation Program:
hoods Lead to Better Schooling Opportunities? The Role of Neighborhoods and Networks in
Parental School Choice in an Experimental Economic Independence.” PhD diss., North-
Housing Voucher Program.” Teachers College Re- western University, Evanston, IL.
cord 112 (5): 1441–1489. Mendenhall, Ruby, Stefanie DeLuca, and Greg
———. 2011. “Increasing Access to High Perform- Duncan. 2006. “Neighborhood Resources and
ing Schools in an Assisted Housing Voucher Economic Mobility: Results from the Gautreaux
Program.” Report prepared for the Depart- Program” Social Science Research 35: 892–923.
ment of Housing and Urban Development and Orr, Larry, Judith D. Feins, Robin Jacob, Erik
the Department of Education report Finding Beecroft, Lisa Sanbonmatsu, Lawrence F. Katz,
Common Ground: Coordinating Housing and Jeffrey B. Liebman, and Jeffrey R. Kling. 2003.
Education Policy to Support Racial and Eco- Moving to Opportunity Interim Impacts Evalua-
nomic Integration. http://www.prrac.org/pdf/ tion. Washington, DC: US Department of Hous-
HousingEducationReport-October2011.pdf. ing and Urban Development, Office of Policy
Goering, John, and Judith Feins. 2003. Choosing a Development and Research.
Better Life? Washington, DC: The Urban Insti- Peterson, George, and Kale Williams. 1995. “Hous-
tute Press. ing Mobility: What Has It Accomplished and
Harding, David J. 2003. “Counterfactual Models What Is Its Promise?” In Housing Mobility:
of Neighborhood Effects: The Effect of Neigh- Promise or Illusion? edited by Alexander Polikoff,
borhood Poverty on Dropping Out and Teenage 7–102. Washington, DC: Urban Institute Press.
Pregnancy.” American Journal of Sociology 109: Polikoff, Alexander. 2006. Waiting for Gautreaux:
676–719. A Story of Segregation, Housing, and the Black
Keels, M. 2008a. “Residential Attainment of Ghetto. Chicago: Northwestern University Press.
Now-Adult Gautreaux Children: Do They Gain, Rosenbaum, James E. 1995. “Housing Mobility
Hold, or Lose Ground in Neighborhood Ethnic Strategies for Changing the Geography of Op-
and Economic Segregation?” Housing Studies 23: portunity.” Housing Policy Debate 6 (1): 231–270.
541–564. Rosenbaum, James E., Stefanie DeLuca, and Tammy
———. 2008b. “Second-Generation Effects of Chi- Tuck. 2005. “Crossing Borders and Adapting:
cago’s Gautreaux Residential Mobility Program Low-Income Black Families in Suburbia.” In
on Children’s Participation in Crime.” Journal of The Geography of Opportunity: Race and Housing
Research on Adolescence 18: 305–352. Choice in Metropolitan America, edited by Xavier
Keels, Micere, Greg J. Duncan, Stefanie DeLuca, de Souza Briggs, 150–175. Washington, DC:
Ruby Mendenhall, and James E. Rosenbaum. Brookings Institution.
2005. “Fifteen Years Later: Can Residential Mo- Rosenbaum, James E., Lisa Reynolds, and Stefanie
bility Programs Provide a Permanent Escape from DeLuca. 2002. “How Do Places Matter? The
Neighborhood Crime and Poverty?” Demography Geography of Opportunity, Self-Efficacy, and a
42 (1): 51–73. Look Inside the Black Box of Residential Mobil-
Ludwig, Jens, et al. 2011. “Neighborhoods, Obesity, ity.” Housing Studies 17: 71–82.
and Diabetes—A Randomized Social Experi- Rosenbaum, James E., and Anita Zuberi. 2010.
ment.” New England Journal of Medicine 365 “Comparing Residential Mobility Programs: De-
(15): 1509–1519. sign Elements, Neighborhoods, and Outcomes in
 Sharkey and Elwert—The Legacy of Multigenerational Disadvantage • 403

MTO and Gautreaux.” Housing Policy Debate 20 Neighborhood Effects on Cognitive Abil-
(1): 27–41. ity.” American Journal of Sociology 116 (6):
Sampson, Robert J. 2012. Great American City: Chi- 1934–1981.
cago and the Enduring Neighborhood Effect. Chi- South, Scott, and Kyle Crowder. 1997. “Escaping
cago: University of Chicago Press. Distressed Neighborhoods: Individual, Com-
Sampson, Robert J., Patrick Sharkey, and Stephen munity and Metropolitan Influences.” American
W. Raudenbush. 2008. “Durable Effects of Con- Journal of Sociology 102: 1040–1084.
centrated Disadvantage on Verbal Ability Among South, Scott, and Glenn Deane. 1993. “Race and
African-American Children.” Proceedings of the Residential Mobility: Individual and Structural
National Academy of Sciences 105 (3): 845–852. Determinants.” Social Forces 72: 147–167.
Sanbonmatsu, Lisa, Jeffrey R. Kling, Greg J. Dun- Thompson v. Department of Housing and Urban
can, and Jeanne Brooks-Gunn. 2006. “Neigh- Development, Civ. Action no. MJG-95–309
borhoods and Academic Achievement: Results (2005–2006).
from the Moving to Opportunity Experiment.” Turney, K., S. Clampet-Lundquist, K. Edin, J. R.
Journal of Human Resources 41 (4): 649–691. Kling, and G. Duncan. 2006. “Neighborhood
Sharkey, Patrick. 2008. “The Intergenerational Effects on Barriers to Employment: Results from
Transmission of Context.” American Journal of a Randomized Housing Mobility Experiment
Sociology 113 (4): 931–969. in Baltimore.” Brookings-Wharton Papers on
———. 2010. “The Acute Effect of Local Homi- Urban Affairs, 137–187.
cides on Children’s Cognitive Performance.” Pro- Wodtke, Geoffrey T., David J. Harding, and Felix El-
ceedings of the National Academy of Sciences 107: wert. 2011. “Neighborhood Effects in Temporal
11733–11738. Perspective: The Impact of Long-Term Exposure
Sharkey, Patrick, and Felix Elwert. 2011. “The to Concentrated Disadvantage on High School
Legacy of Disadvantage: Multigenerational Graduation.” American Sociological Review 76
(5): 713–736.

47 • Patrick Sharkey and Felix Elwert

The Legacy of Multigenerational Disadvantage

Research on the relationship between neighbor- instantaneous effects on children, a life course
hoods and child development has frequently perspective on neighborhood inequality shifts
overlooked a crucial dimension of neighborhood attention toward continuity and change in the
stratification: time. Whereas much research on environment over time and across generations
the impact of neighborhoods implicitly treats and considers the role that neighborhoods play
the environment as a static feature of a child’s in altering or structuring individuals’ or families’
life and assumes that the neighborhood has trajectories.

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Patrick Sharkey and Felix Elwert, “The Legacy of Disadvantage: Multigenerational
Neighborhood Effects on Cognitive Ability,” American Journal of Sociology 116:6 (March 2011), pp. 1934–
1981, published by the University of Chicago Press. The article printed here was originally prepared by Patrick
Sharkey and Felix Elwert for this fourth edition of Social Stratification: Class, Race, and Gender in Sociological
Perspective, edited by David B. Grusky. Copyright © 2014 by Westview Press.
404 • Part V: Poverty and the Underclass

Recent research demonstrates the complex re- available for the child, and the neighborhood in
lationships between exposure to disadvantaged which he raises his child.
neighborhood environments and child devel- Illuminating the relationships that link par-
opmental outcomes (Sampson, Sharkey, and ents’ and children’s residential environments
Raudenbush 2008; Wheaton and Clarke 2003). to children’s cognitive outcomes is not only a
But what if the child’s caregivers were also raised theoretical problem, but also poses consider-
in similarly disadvantaged environments? Could able methodological challenges. Virtually all
a parent’s childhood neighborhood environment previous observational studies of neighborhood
have an influence on the next generation? In this effects used regression techniques (or some
study we add to the recent line of research that variant, such as propensity score matching),
incorporates time into the literature on neigh- in which a set of family background measures
borhood inequality, but we reach further back in is controlled. The same techniques are not ap-
time than other studies and ask how the neigh- propriate for investigating multigenerational
borhood environment, experienced over multi- effects if the dimensions of family background
ple generations of a family, influences children’s that influence neighborhood location in the sec-
cognitive abilities. ond generation are influenced by neighborhood
Our focus on multigenerational disadvan- conditions in the first generation. In essence,
tage is motivated by recent research on the per- controlling for family background would block
sistence of neighborhood economic status across the indirect pathways by which first-generation
generations, which demonstrates that neighbor- neighborhood characteristics may influence de-
hood inequality in one generation is commonly velopmental outcomes a generation later, thus
transmitted to the next. For example, more underestimating the importance of parents’
than 70 percent of African American children neighborhoods for children’s outcomes.
who grow up in the poorest quarter of Ameri- The theoretical problem of multigenera-
can neighborhoods remain in the poorest quar- tional relationships thus becomes a method-
ter of neighborhoods as adults (Sharkey 2008). ological problem, arising in any scenario in
The persistence of neighborhood disadvantage which confounders are potentially endogenous
across generations adds considerable complex- to treatments experienced at an earlier point
ity to how researchers approach the relationship in time—or in an earlier generation. Instead
between neighborhoods and child development, of conventional regression models, we draw on
because it forces one to consider direct and indi- newly developed methods designed to generate
rect pathways by which neighborhood exposures unbiased treatment effects in such situations,
in both the parents’ and children’s generations under assumptions that we specify in the follow-
may influence children’s trajectories. A child’s ing discussion. In a series of articles introduc-
own neighborhood may influence her cogni- ing marginal structural models and the method
tive ability through, for example, the quality of of inverse probability of treatment weighting,
her schooling experience or the influence of her Robins and colleagues (Robins 1998, 1999b;
peers. But there also may be pathways by which Hernán, Brumback, and Robins 2000; Robins,
a parent’s childhood neighborhood, experienced Hernán, and Brumback 2000) show that treat-
a generation earlier, continues to exert a linger- ment effect bias can be addressed by using a
ing influence on his child’s cognitive ability. It is model that weights each subject by the inverse of
plausible that the parent’s childhood neighbor- the predicted probability that the subject receives
hood may influence his own schooling experi- a given treatment at a given point conditional on
ence, his experiences in the labor market, and prior treatment history and prior confounders
even his mental health. All of these aspects of a (both time varying and time invariant). Here we
parent’s life may, in turn, influence the resources adapt this method for treatments received across
available to him for childrearing—including the generations to estimate multigenerational neigh-
quality of the home environment, the resources borhood effects on children’s cognitive abilities.
 Sharkey and Elwert—The Legacy of Multigenerational Disadvantage • 405

influence their cognitive development (Alexan-

Intergenerational Pathways of der, Entwisle, and Olson 2007; Downey, von
Influence Hippel, and Broh 2004; Winship and Koren-
The theory underlying a multigenerational per- man 1997). This literature provides a sense of the
spective argues that there are numerous possible multiple ways in which different aspects of the
pathways, observed and unobserved, by which home or family environment may be linked with
the neighborhood environment in one genera- children’s cognitive development. The presence
tion may be linked with a child’s cognitive ability of numerous possible pathways linking parents’
in the next generation. If childhood neighbor- own childhood environments to their children’s
hoods affect any dimension of adult social or development a generation later provides the the-
economic status, health, or family life, then oretical basis for a multigenerational analysis.
disadvantages experienced during childhood in
one generation may linger and affect cognitive
ability in the next. The fact that several studies Data
have found strong childhood neighborhood ef- To assess the multigenerational effects of neigh-
fects on specific aspects of adult attainment, such borhood poverty on cognitive ability, we draw on
as education and mental health (Harding 2003; data from the Panel Study of Income Dynamics
Kaufman and Rosenbaum 1992; Wheaton and (PSID) (Hill and Morgan 1992), a nationally
Clarke 2003), lends credence to the hypothesis representative survey that began with a sample
that neighborhood effects may extend across of roughly five thousand families in 1968 and
generations. The presence of neighborhood ef- has followed the members of these families over
fects in studies examining various additional time. We match families to their census tract of
dimensions of adult life strengthens this hypoth- residence through the PSID restricted-use geo-
esis considerably, even if these studies produce code file, which contains tract identifiers for sam-
inconsistent results. ple families from 1968 through 2003. Finally, we
For such indirect effects to exist, however, utilize data on cognitive ability from the 2002
we must make the additional assumption that Child Development Supplement (CDS) (Hof-
aspects of family background and the social ferth et al. 1997; Mainieri 2004), a follow-up
environments in which children spend their survey of a sample of PSID parents with chil-
childhoods have an influence on their cogni- dren ages five to eighteen who were originally
tive development. This assumption taps into a assessed in the 1997 CDS at ages zero to twelve.
long-standing debate on the malleability of cog- The CDS was designed to supplement the core
nitive ability (Heckman 1995; Herrnstein and PSID interview with information on child devel-
Murray 1994; Jacoby and Glauberman 1995; opment and details about the home, school, and
Neisser et al. 1996). Although there is little neighborhood environments, as well as familial
doubt that cognitive ability—whether conceived and social relationships. The final sample com-
as intelligence, IQ, or simply performance on prises 1,556 parent-child pairs, including 730
tests of cognitive skills—has a genetic compo- African American pairs, 792 white pairs, and 34
nent, there is also near consensus that develop- pairs of all other racial and ethnic groups.
ment is sensitive to the family, school, and social The outcomes under study represent two
environment. Empirically, children’s cognitive dimensions of child and adolescent cognitive
development has been linked with parents’ ed- ability measured using the Woodcock-Johnson
ucation, alcohol use, mental health, social and Psycho-Educational Battery-Revised (WJ-R)
economic status, and parenting practices, as well (Woodcock and Johnson 1989): broad reading
as various aspects of the home environment (Guo scores and applied problems scores. The broad
and Harris 2000; Shonkoff and Phillips 2000). reading score measures reading ability and com-
These same characteristics of parents may also af- bines results from two subscales, the letter-word
fect the schooling experiences of children, which assessment and the passage comprehension
406 • Part V: Poverty and the Underclass

assessment. To measure ability in math, we use the child’s observed cognitive ability, and the po-
the applied problems score. Raw results from tential outcome AN is the child’s cognitive ability
each subtest are normalized to reflect the child’s that would be observed if her family had expe-
abilities relative to the national average for the rienced the multigenerational neighborhood
child’s age (Mainieri 2004). regime N = {NP, NO}, consisting of the neigh-
The treatment is defined as living in a borhood environments in the parent generation,
high-poverty neighborhood during childhood; it NP, and the neighborhood environment in the
is measured for children in the three survey years child’s generation, NO. The difference between
prior to the CDS and for their parents when they the two potential outcomes, į = AN–AN*, defines
were ages fifteen to seventeen. Specifically, we de- the causal effect of experiencing the multigen-
fine high-poverty neighborhoods as those where erational residential history N rather than some
the poverty rate is at least 20 percent. However, other residential history, N*. In each generation,
mindful of severe racial discrepancies in expo- we classify neighborhoods as either poor or non-
sure to high-poverty neighborhoods, Sharkey poor, yielding four possible multigenerational
and Elwert (2011) conduct an additional set of regimes: N = {(poor, poor), (poor, nonpoor),
race-specific analyses in which the definition of (nonpoor, poor), (nonpoor, nonpoor)}. We are
the treatment is allowed to vary by race. In these primarily interested in two causal contrasts: (1)
analyses they use a 40 percent poverty threshold E[AN={poor,poor}]–E[AN={nonpoor,nonpoor}], which defines
to define high-poverty neighborhoods for blacks the average causal effect if both parents and chil-
and a 10 percent threshold for whites. Parental dren had grown up in poor rather than nonpoor
neighborhood poverty is measured as the average neighborhoods, which we term the joint, or mul-
poverty rate in the census tracts where parents tigenerational, causal effect; and (2) E[AN={poor,fix-
}
lived during the three survey years from ages fif- ]–E[AN=={nonpoor,fix}], which defines the average
teen to seventeen. Child neighborhood poverty is causal effect if the parents had grown up in a
measured as the average poverty rate in children’s poor rather than a nonpoor neighborhood and
neighborhoods in the three survey years prior their children had grown up in a neighborhood
to the 2002 CDS: 1997, 1999, and 2001. All of fixed type (either poor or nonpoor), which we
models include extensive control variables from term the direct causal effect of parents’ exposure
each generation, including demographic charac- on children’s cognitive abilities. Note that this
teristics of the head of household and the child direct effect captures the portion of parental ex-
(age, gender); various dimensions of family back- posure that does not operate through influencing
ground (family income, occupational status, edu- the children’s own neighborhood of residence;
cational attainment); employment characteristics instead, it may operate via parents’ education,
of the head of the household or family (disability income, parenting style, or other characteristics
status, welfare receipt, annual hours worked); of the parents or the home environment that are
and other measured attitudes and characteristics influenced by the parents’ childhood neighbor-
of the family, such as home ownership and atti- hood context.
tudes about the future.

Two Difficulties of Estimating

Defining Multigenerational Effects Multigenerational Causal Effects
of Neighborhood Disadvantage The causal effect of multigenerational neighbor-
The primary objective of this study is to estimate hood poverty on children’s cognitive abilities,
the joint multigenerational effects of parents’ A, can be identified from observational data if
and children’s exposure to neighborhood poverty neighborhood of residence in each generation,
on child cognitive ability, in order to capture the NP and NO, is statistically independent of the po-
effect of enduring disadvantage. Formally, A is tential outcomes, AN, given observed covariates
 Sharkey and Elwert—The Legacy of Multigenerational Disadvantage • 407

Figure 47.1. Directed Acyclic Graph Displaying Possible Direct and Indirect Causal Pathways Link-
ing Neighborhood Exposure (N) and Confounding Variables (C), Which Determine Neighborhood of
Residence in the Parent (P) and Child (O) Generations, to Child Cognitive Outcomes (A). The Vector U
Represents Unobserved Factors

and previous treatments. This assumption is multigenerational effect of N on A is identifiable

known in the technical literature as sequential from the observed data.
ignorability or unconfoundedness of treatment Simply conditioning on the measured co-
assignment (Robins 1986, 1999b). It implies founders CP and CO in a conventional regression
that there are no unobserved variables that model, such as ordinary least squares (OLS),
jointly affect the treatment (NP or N O) and the however, will not provide an unbiased estimate
outcome. for the multigenerational effect of N on A. To
Even under this assumption, however, tradi- understand why, we consider how the confound-
tional regression (as well as conventional propen- ing variable CO should best be handled. On the
sity score) methods generally cannot estimate the one hand, C O must be controlled to avoid bias
multigenerational effect of neighborhood disad- from confounding, as CO causes both NO and A.
vantage, because neighborhood disadvantage in On the other hand, conditioning on CO creates
the second generation is endogenous to neigh- two endogeneity problems. First, note that CO is
borhood disadvantage in the first generation in on the causal pathway from NP to A. Controlling
two distinct ways. for CO may thus “control away” part of the effect
Figure 47.1 illustrates these two endogene- of parental neighborhood poverty on the child’s
ity problems. The figure shows a directed acy- cognitive outcome and consequently produce
clic graph (Pearl 1995, 2000) representing the bias. Second, even if NP had no direct or indirect
causal relationships among neighborhood pov- causal effect on A (i.e., if there are no pathways
erty, child cognitive outcomes, and other vari- to be “controlled away”), conditioning on C O
ables. All arrows between the temporally ordered will induce a noncausal association between NP
variables represent direct causal effects, and the and A because of the unobserved variable U—
absence of an arrow indicates the absence of a conditioning on C O will induce an association
direct causal effect. The variables in Figure 47.1 between N P and U (because conditioning on
are defined as before, with the addition of U the common effect of two variables always does;
representing unmeasured variables causing both Pearl 1995) and thus between NP and A, induc-
C O and A. For expositional clarity, we assume ing endogenous selection bias that would make
without loss of generality that Figure 47.1 con- it impossible to reject the causal null hypothesis
tains all variables in the system. We note that of no direct causal effect of NP on A, even if the
treatments in Figure 47.1 are sequentially ignor- null hypothesis were true (Elwert and Winship,
able, because there are no arrows from any un- forthcoming). In sum, the analyst is thus obliged
observed variables into N P or NO. Therefore, the to simultaneously condition on CO to control for
408 • Part V: Poverty and the Underclass

confounding of NO and not to condition on CO advantaged neighborhoods, and the sum of F +

to avoid controlling away part of the effect of NP G1 + G2 estimates the child’s mean cognitive abil-
on A and inducing endogenous selection bias. ity if all parents and children had grown up in
Conventional regression models, however, can- disadvantaged neighborhoods. The sum of the
not simultaneously control and not control for two slopes G1 + G2 thus gives the joint causal effect
the same variable. But so-called marginal struc- of multigenerational neighborhood disadvantage
tural models (MSMs) can. on children’s cognitive outcomes. The slope G1
by itself gives the direct causal effect of parental
neighborhood disadvantage.
Marginal Structural Models for
Multigenerational Neighborhood
Effects Results from Regression and
We use MSMs with inverse probability of treat- Marginal Structural Models
ment (IPT) weighting to estimate the multigen- Table 47.1 shows estimates for the causal effects
erational effects of neighborhood disadvantage of multigenerational neighborhood poverty on
on children’s cognitive outcomes. The models children’s cognitive abilities from three models:
are well suited for the task because they are more one unadjusted for any covariates, a conventional
powerful than conventional regression models regression-adjusted model, and an IPT-weighted
in at least two senses. First, they are designed MSM. Prior to accounting for nonrandom selec-
to resolve the core endogeneity problems of tion into neighborhoods, neighborhood poverty
time-varying (here, multigenerational) expo- in each generation is strongly and negatively as-
sure (Robins 1998, 1999a, 1999b; Robins et al. sociated with children’s broad reading scores (col-
2000; Hernán et al. 2000), and second, they can umn 1). Standard regression adjustments reduce
do so making fewer assumptions than traditional the apparent direct effect of parental and child
regression models do.1 neighborhood poverty substantially; however, the
The MSMs have two steps. In the first step, conventional regression model includes all of the
every family is weighted by the inverse of the parents’ covariates, including their educational
estimated probability of experiencing the se- attainment, income, and so forth. If the influence
quence of neighborhood conditions that they of parents’ childhood neighborhoods is mediated
did in fact experience. This weighting removes by these or other aspects of parents’ adult lives,
all confounding by observed variables CP and CO. then the regression estimates will “control away”
In the second step, the weighted data are ana- all of these indirect pathways of influence and
lyzed with a (weighted) regression model. With hence underestimate the true effect.
observed confounders taken care of by weight- Column 3 presents estimates from an MSM,
ing, there is no longer any need to control for which accounts for observed selection into treat-
observed confounders on the causal pathway in ment status in both generations through IPT
the outcomes equation, which therefore does not weighting (which, unlike conventional regres-
“control away” part of the effect of interest. The sion models, does not “control away” mediated
outcomes equation is effects). The multigenerational effect of coming
from a family residing in poor neighborhoods
Eweighted[A| NP, NO] = F + NPG1 + in two successive generations rather than from
NOG2 (1) a family living in nonpoor neighborhoods is
a reduction of 9.27 points in a child’s broad
Under the assumption of no sequential ig- reading scores. This multigenerational effect of
norability, the parameters in Equation (1) have neighborhood disadvantage on reading scores is
a causal interpretation. The model intercept, substantively large (more than half a standard
F, estimates the child’s mean cognitive abil- deviation in broad reading scores) and statisti-
ity if all parents and children had grown up in cally significant at the (p-value < 0.05). Results
 Sharkey and Elwert—The Legacy of Multigenerational Disadvantage • 409

Table 47.1. Estimated Effects of Multigenerational Exposure to Neighborhood Poverty on Children’s

Cognitive Ability

Neighborhood poverty: >=20% poor

Broad reading score Applied problems score
Unadjusted Regression Marginal Unadjusted Regression Marginal
estimates adjusted struct. model estimates adjusted struct. model

Parent neighborhood -8.83*** -2.85** -5.07** -9.68*** -2.40** -5.97***

poverty only, ȕ1 (1.16) (1.21) (2.38) (0.97) (1.12) (1.85)

Child neighborhood -5.98*** -1.73 -4.20** -5.66*** -0.84 -2.39

poverty only, ȕ2 (1.23) (1.10) (2.00) (1.00) (1.02) (1.79)

Multigenerational - - -9.27*** - - -8.36***

exposure, ȕ1+ȕ2 - - (1.68) - - (1.69)

Notes: *** = significant at p<.01; ** = significant at p<.05; * = significant at p<.10

Standard errors account for clustering at the family level.

for applied problem-solving scores are similar impact of neighborhood change arising from
to results for broad reading scores. For this out- residential mobility, as well as observational re-
come, the joint causal effect of multigenerational search on social contexts and child development.
exposure to neighborhood poverty is substan- First we consider the experimental and
tively large and statistically significant, reducing quasi-experimental evidence available from
children’s applied problem scores by 8.36 points, residential mobility programs, including the
more than half a standard deviation (p-value < Gautreaux program in Chicago, the Moving
0.01). A formal sensitivity analysis (not shown) to Opportunity experiment, and other similar
in Sharkey and Elwert (2011) demonstrates that programs (Briggs 1997). In all such programs,
these results are quite robust to possible unob- participants (typically low-income families living
served selection bias. in public housing) are given the chance to move
to less disadvantaged environments, frequently in
the same city or within the metropolitan area.
Discussion Research based on these programs exploits exog-
The evidence presented here suggests that a enous variation in the destinations of participants
multigenerational perspective is crucial to un- in the programs to estimate how a change in the
derstanding the relationship between neigh- neighborhood environment impacts children’s
borhood environments and cognitive ability. A and adults’ social outcomes. Although this type
family’s exposure to neighborhood poverty over of study provides sound evidence on the causal ef-
two consecutive generations is found to reduce fect of contemporary neighborhood exposure due
the average child’s cognitive ability by more than to a change in the neighborhood environment
half a standard deviation. This empirical result arising from a residential move, by design these
points to a revised, broader conceptualization of studies do not capture the lagged or cumulative
how the neighborhood environment influences effects of previous neighborhood environments.
cognitive ability, and furthermore suggests a re- This focus on contemporary neighborhood
vised theoretical and empirical perspective on circumstances has been questioned in recent re-
the influence of social contexts on child devel- search on youth in Chicago, which shows that the
opment. We argue that this revised perspective impact of living in severely disadvantaged neigh-
should inform interpretations of experimental borhoods continues to be felt years later (Samp-
and quasi-experimental research assessing the son et al. 2008). The challenge is strengthened
410 • Part V: Poverty and the Underclass

considerably when one considers the possibility the neighborhood as competing developmental
of generation-lagged effects or cumulative, multi- contexts.
generational effects. A change in a family’s neigh- Our theory and results indicate that the
borhood may bring about an abrupt and radical family and the neighborhood environments
change in the social environment surrounding are closely intertwined, combining to influence
children, but this change may be a short-term the developmental trajectories of individuals in
departure from a familial history of life in dis- ways that extend across generations. As we have
advantaged environments. The shift in context shown, a multigenerational perspective is essen-
may improve the opportunities available to tial to understanding inequality in cognitive abil-
adults and children, the child’s peers and school ity. Our findings support other studies showing
environment, and the parent’s mental health, but a link between the neighborhood environment
it may not undo the lingering influence of the and children’s cognitive ability, but we extend
parent’s childhood environment. In short, a tem- this literature by calling attention to the history
porary change of scenery may not disrupt the of social environments occupied by family mem-
effects of a family history of disadvantage. bers over generations. This approach reflects the
This assessment should not be taken as a broader implication of this chapter, which is that
criticism of the residential mobility literature, to understand inequality in cognitive ability and
but as a lens with which to interpret it. Evalu- other developmental domains, it is not sufficient
ations of residential mobility programs provide to focus on a single point in a child’s life, or even
powerful evidence for policy makers interested on a single generation of a family. Instead, we
in designing programs to move families into must understand the history of disadvantages
areas that may improve adults’ mental health or experienced by generations of family members.
children’s life chances. But these programs tell us
little about the cumulative disadvantages facing a NOTES
family living in America’s poorest neighborhoods 1. MSMs make fewer assumptions than conven-
tional regression models in that MSMs can handle
over long periods of time.
(as illustrated in Figure 47.1) unobserved variables
Next we consider the extensive literature on
U that affect the confounders C of treatment N,
neighborhood effects based on observational whereas conventional regression cannot handle the
data. The most common analytical approach presence of U.
in this literature involves estimating neighbor-
hood effects while controlling for a set of family REFERENCES
background measures. A common claim made Alexander, Karl L., Doris R. Entwisle, and Linda
in reviews of these studies is that the family en- Steffel Olson. 2007. “Lasting Consequences of
vironment is more important for child develop- the Summer Learning Gap.” American Sociolog-
ment than the neighborhood environment (Ellen ical Review 72: 167–180.
Briggs, Xavier de Souza. 1997. “Moving Up Versus
and Turner 1997; Leventhal and Brooks-Gunn
Moving Out: Neighborhood Effects in Housing
2000). A multigenerational perspective suggests
Mobility Programs.” Housing Policy Debate 8:
that such a conclusion is misleading. Aspects of 195–234.
family background that are linked with child de- Downey, Douglas B., Paul T. von Hippel, and Beck-
velopmental outcomes, such as parental income ett A. Broh. 2004. “Are Schools the Great Equal-
or education, may be endogenous to neighbor- izer? Cognitive Inequality During the Summer
hood conditions in the prior generation. Parents’ Months and the School Year.” American Sociolog-
educational attainment, economic position, and ical Review 69: 613–635.
health are better thought of as partial outcomes Ellen, Ingrid Gould, and Margery Austin Turner.
1997. “Does Neighborhood Matter? Assessing Re-
of their own earlier residential circumstances.
cent Evidence.” Housing Policy Debate 8: 833–866.
In this sense, individuals and families insepa-
Elwert, Felix, and Christopher Winship. Forthcom-
rably embody neighborhood histories, and it is ing. “Endogenous Selection Bias.” In Annual Re-
therefore a mistake to think of the family and view of Sociology.
 Sharkey and Elwert—The Legacy of Multigenerational Disadvantage • 411

Guo, Guang, and Kathleen Mullan Harris. 2000. Pearl, Judea. 1995. “Causal Diagrams for Empirical
“The Mechanisms Mediating the Effects of Pov- Research.” Biometrika 82: 669–710.
erty on Children’s Intellectual Development.” ———. 2009. Causality: Models, Reasoning, and In-
Demography 37: 431–447. ference. Cambridge, UK: Cambridge University
Harding, David. 2003. “Counterfactual Models of Press.
Neighborhood Effects: The Effect of Neighbor- Robins, James M. 1986. “A New Approach to Causal
hood Poverty on Dropping Out and Teenage Inference in Mortality Studies with Sustained Ex-
Pregnancy.” American Journal of Sociology 109: posure Periods—Application to Control of the
676–719. Healthy Worker Survivor Effect.” Mathematical
Heckman, James J. 1995. “Lessons from the Bell Modelling 7: 1393–1512.
Curve.” The Journal of Political Economy 103: ———. 1998. “Marginal Structural Models.” 1997
1091–1120. Proceedings of the American Statistical Associa-
Hernán, Miguel Ángel, Babette Brumback, and tion, Section on Bayesian Statistical Science.
James M. Robins. 2000. “Marginal Structural ———. 1999a. “Association, Causation, and Mar-
Models to Estimate the Causal Effect of Zid- ginal Structural Models.” Synthese 121: 151–179.
ovudine on the Survival of HIV-Positive Men.” ———. 1999b. “Marginal Structural Models versus
Epidemiology 11: 561–570. Structural Nested Models as Tools for Causal
Herrnstein, Richard, and Charles Murray. 1994. Inference.” In Statistical Models in Epidemiol-
The Bell Curve: Intelligence and Class Structure in ogy, edited by E. Halloran, 95–134. New York:
American Life. New York: Free Press. Springer-Verlag.
Hill, Martha S., and James N. Morgan. 1992. The Robins, James M., Miguel Hernán, and Babette
Panel Study of Income Dynamics: A User’s Guide. Brumback. 2000. “Marginal Structural Models
Newbury Park, CA: Sage Publications. and Causal Inference in Epidemiology.” Epide-
Hofferth, Sandra, Pamela E. Davis-Kean, Jean Davis, miology 11: 550–560.
and Jonathan Finkelstein. 1997. The Child De- Sampson, Robert J., Patrick Sharkey, and Stephen
velopment Supplement to the Panel Study of Income W. Raudenbush. 2008. “Durable Effects of Con-
Dynamics: 1997 User Guide. Ann Arbor, MI: In- centrated Disadvantage on Verbal Ability among
stitute for Social Research. African-American Children.” Proceedings of the
Jacoby, Russess, and Naomi Glauberman, eds. 1995. National Academy of Sciences 105 (3): 845–853.
The Bell Curve Debate: History, Documents, Opin- Sharkey, Patrick. 2008. “The Intergenerational
ions. New York: Random House. Transmission of Context.” American Journal of
Kaufman, Julie E., and James Rosenbaum. Sociology 113: 931–969.
1992. “The Education and Employment of Shonkoff, Jack P., and Deborah A. Phillips, eds.
Low-Income Black Youth in White Suburbs.” 2000. From Neurons to Neighborhoods: The Sci-
Educational Evaluation & Policy Analysis 14: ence of Early Childhood Development. Washing-
229–240. ton, DC: National Academy Press.
Leventhal, Tama, and Jeanne Brooks-Gunn. 2000. Wheaton, Blair, and Philippa Clarke. 2003. “Space
“The Neighborhoods They Live in: The Effects Meets Time: Integrating Temporal and Con-
of Neighborhood Residence on Child and Ad- textual Influences on Mental Health in Early
olescent Outcomes.” Psychological Bulletin 126: Adulthood.” American Sociological Review 68:
309–337. 680–706.
Mainieri, Tina. 2004. The Panel Study of Income Winship, Christopher, and Sanders Korenman.
Dynamics Child Development Supplement: User 1997. “Does Staying in School Make You
Guide for CDS-II. Ann Arbor, MI: Institute for Smarter?” In Intelligence, Genes, and Success:
Social Research. Scientists Respond to the Bell Curve, edited by B.
Neisser, Ulric, Gwyneth Boodoo, Thomas J. Bouch- Devlin, S. E. Fienberg, D. P. Resnick and K. Ro-
ard Jr., A. Wade Boykin, Nathan Brody, Stephen eder, 215–234. New York: Springer-Verlag.
J. Ceci, Diane F. Halpern, John C. Loehlin, Woodcock, Richard W., and M. Bonner Johnson.
Robert Perloff, Robert J. Sternberg, and Susana 1989. Tests of Achievement, Standard Battery
Urbina. 1996. “Intelligence: Knowns and Un- (Form B). Chicago: Riverside Publishing.
knowns.” American Psychologist 51: 77–101.
412 • Part V: Poverty and the Underclass

How Important Is Early Childhood?

48 • James J. Heckman

Skill Formation and the Economics of Investing in

Disadvantaged Children

Four core concepts important to devising and extending throughout the early childhood
sound social policy toward early childhood have years” (2). This principle stems from two charac-
emerged from decades of independent research teristics that are intrinsic to the nature of learn-
in economics, neuroscience, and developmen- ing: (i) early learning confers value on acquired
tal psychology (1). First, the architecture of skills, which leads to self-reinforcing motivation
the brain and the process of skill formation are to learn more, and (ii) early mastery of a range
influenced by an interaction between genetics of cognitive, social, and emotional competencies
and individual experience. Second, the mastery makes learning at later ages more efficient and
of skills that are essential for economic success therefore easier and more likely to continue.
and the development of their underlying neural Early family environments are major predic-
pathways follow hierarchical rules. Later attain- tors of cognitive and noncognitive abilities. Re-
ments build on foundations that are laid down search has documented the early (by ages 4 to 6)
earlier. Third, cognitive, linguistic, social, and emergence and persistence of gaps in cognitive
emotional competencies are interdependent; all and noncognitive skills (3, 4). Environments
are shaped powerfully by the experiences of the that do not stimulate the young and fail to cul-
developing child; and all contribute to success in tivate these skills at early ages place children at
the society at large. Fourth, although adaptation an early disadvantage. Disadvantage arises more
continues throughout life, human abilities are from lack of cognitive and noncognitive stimu-
formed in a predictable sequence of sensitive pe- lation given to young children than simply from
riods, during which the development of specific the lack of financial resources.
neural circuits and the behaviors they mediate This is a source of concern because family en-
are most plastic and therefore optimally recep- vironments have deteriorated. More U.S. chil-
tive to environmental influences. dren are born to teenage mothers or are living
A landmark study concluded that “virtually in single parent homes compared with 40 years
every aspect of early human development, from ago (5). Disadvantage is associated with poor
the brain’s evolving circuitry to the child’s capac- parenting practices and lack of positive cognitive
ity for empathy, is affected by the environments and noncognitive stimulation. A child who falls
and experiences that are encountered in a cumu- behind may never catch up. The track records
lative fashion, beginning in the prenatal period for criminal rehabilitation, adult literacy, and

James J. Heckman. “Skill Formation and the Economics of Investing in Disadvantaged Children,” Science 312,
issue 5782 (June 2006), pp. 1900–1902. Reprinted with permission from AAAS.
 Heckman—Skill Formation and the Economics of Investing in Disadvantaged Children • 413

public job training programs for disadvan- Table 48.1. Economic Benefits and Costs of the Perry
taged young adults are remarkably poor (3). Preschool Program
Disadvantaged early environments are power-
Perry Preschool
ful predictors of adult failure on a number of Child care $986
social and economic measures. Earnings $40,537
Many major economic and social prob- K–12 $9184
lems can be traced to low levels of skill and College/adult $–782
ability in the population. The U.S. will add Crime $94,065
Welfare $355
many fewer college graduates to its workforce
Abuse/neglect $0
in the next 20 years than it did in the past 20 Total benefits $144,345
years (6, 7 ). The high school dropout rate, Total costs $16,514
properly measured with inclusion of individ- Net present value $127,831
uals who have received general educational Benefits-to-costs ratio 8.74
development (GED) degrees, is increasing at All values are discounted at 3% and are in 2004 dollars. Earnings,
a time when the economic return of school- Welfare, and Crime refer to monetized value of adult outcomes
ing has increased (8). It is not solely a phe- (higher earnings, savings in welfare, and reduced costs of crime).
nomenon of unskilled immigrants. Over 20% K–12 refers to the savings in remedial schooling. College/adult
refers to tuition costs. (27)
of the U.S. workforce is functionally illiterate,
compared with about 10% in Germany and morning programs at school and afternoon vis-
Sweden (9). Violent crime and property crime its by the teacher to the child’s home. The Perry
levels remain high, despite large declines in re- intervention group had IQ scores no higher
cent years. It is estimated that the net cost of than the control group by age 10. Yet, the Perry
crime in American society is $1.3 trillion per treatment children had higher achievement test
year, with a per capita cost of $4,818 per year scores than the control children because they
(10). Recent research documents the importance were more motivated to learn. In followups to
of deficits in cognitive and noncognitive skills age 40, the treated group had higher rates of
in explaining these and other social pathologies high school graduation, higher salaries, higher
(11). percentages of home ownership, lower rates
of receipt of welfare assistance as adults, fewer
out-of-wedlock births, and fewer arrests than
Noncognitive Skills and Examples of the controls (13). The economic benefits of the
Successful Early Interventions Perry Program are substantial (Table 48.1). Rates
Cognitive skills are important, but noncogni- of return are 15 to 17% (14). (The rate of return
tive skills such as motivation, perseverance, and is the increment in earnings and other outcomes,
tenacity are also important for success in life. suitably valued, per year for each dollar invested
Much public policy, such as the No Child Left in the child.) The benefit-cost ratio (the ratio of
Behind Act, focuses on cognitive test score out- the aggregate program benefits over the life of
comes to measure the success of interventions in the child to the input costs) is over eight to one.
spite of the evidence on the importance of non- Perry intervened relatively late. The Abe-
cognitive skills in social success. Head Start was cedarian program, also targeted toward disad-
deemed a failure in the 1960s because it did not vantaged children, started when participants
raise the intelligence quotients (IQs) of its par- were 4 months of age. Children in the treatment
ticipants (12). Such judgments are common but group received child care for 6 to 8 hours per
miss the larger picture. Consider the Perry Pre- day, 5 days per week, through kindergarten
school Program (13), a 2-year experimental in- entry; nutritional supplements, social work ser-
tervention for disadvantaged African-American vices, and medical care were provided to con-
children initially ages 3 to 4 that involved trol group families. The program was found to
414 • Part V: Poverty and the Underclass

Figure 48.1. Average Percentile Rank on Peabody Individual Achievement Test–Math Score by
Age and Income Quartile

Income quartiles are computed from average family income between the ages of 6 and 10. Adapted from (3)
with permission from MIT Press.

permanently raise the IQ and the noncognitive gaps in test scores across socioeconomic groups
skills of the treatment group over the control are stable by age, suggesting that later schooling
group. However, the Abecedarian program was and variations in schooling quality have little
intensive, and it is not known whether it is the effect in reducing or widening the gaps that ap-
age of intervention or its intensity that contrib- pear before students enter school (4, 24). Figure
uted to its success in raising IQ (15–17). 48.1 plots gaps in math test scores by age across
Reynolds et al. present a comprehensive re- family income levels. The majority of the gap at
view of early childhood programs directed to- age 12 appears at the age of school enrollment.
ward disadvantaged children and their impact Carneiro and Heckman performed a cost-benefit
(18). Similar returns are obtained for other early analysis of classroom size reduction on adult
intervention programs (19, 20), although more earnings (3). Although smaller classes raise the
speculation is involved in these calculations be- adult earnings of students, the earnings gains
cause the program participants are in the early received by students do not offset the costs of
stages of their life cycles and do not have long hiring additional teachers. The student-teacher
earnings histories. achievement ratio (STAR) randomized trial of
classroom size in Tennessee shows some effect of
reduced classroom size on test scores and adult
Schools and Skill Gaps performance, but most of the effect occurs in the
Many societies look to the schools to reduce earliest grades (25, 26). Schools and school qual-
skills gaps across socioeconomic groups. Because ity at current levels of funding contribute little to
of the dynamics of human skill formation, the the emergence of test score gaps among children
abilities and motivations that children bring to or to the development of the gaps.
school play a far greater role in promoting their
performance in school than do the traditional
inputs that receive so much attention in pub- Second Chance Programs
lic policy debates. The Coleman Report (21) as America is a second chance society. Our educa-
well as recent work (22, 23) show that families tional policy is based on a fundamental optimism
and not schools are the major sources of inequal- about the possibility of human change. The dy-
ity in student performance. By the third grade, namics of human skill formation reveal that later
 Heckman—Skill Formation and the Economics of Investing in Disadvantaged Children • 415

Figure 48.2. Rates of Return to Human Capital Invest- from funds if they were invested for purposes
ment in Disadvantaged Children unrelated to disadvantaged children.

Conclusions
Investing in disadvantaged young children is a
rare public policy initiative that promotes fair-
ness and social justice and at the same time
promotes productivity in the economy and in
society at large. Early interventions targeted to-
ward disadvantaged children have much higher
returns than later interventions such as reduced
pupil-teacher ratios, public job training, convict
rehabilitation programs, tuition subsidies, or
The declining figure plots the payout per year per dollar expenditure on police. At current levels of re-
invested in human capital programs at different stages of the life
cycle for the marginal participant at current levels of spending.
sources, society overinvests in remedial skill in-
The opportunity cost of funds (r) is the payout per year if the vestments at later ages and underinvests in the
dollar is invested in financial assets (e.g., passbook savings) early years.
instead. An optimal investment program from the point of Although investments in older disadvantaged
view of economic efficiency equates returns across all stages of
the life cycle to the opportunity cost. The figure shows that, individuals realize relatively less return overall,
at current levels of funding, we overinvest in most schooling such investments are still clearly beneficial. In-
and post-schooling programs and underinvest in preschool deed, the advantages gained from effective early
programs for disadvantaged persons. Adapted from (3) with
interventions are sustained best when they are
permission from MIT Press.
followed by continued high-quality learning
compensation for deficient early family environ- experiences. The technology of skill formation
ments is very costly (4). If society waits too long shows that the returns on school investment and
to compensate, it is economically inefficient to postschool investment are higher for persons
invest in the skills of the disadvantaged. A serious with higher ability, where ability is formed in
trade-off exists between equity and efficiency for the early years. Stated simply, early investments
adolescent and young adult skill policies. There must be followed by later investments if maxi-
is no such trade-off for policies targeted toward mum value is to be realized.
disadvantaged young children (28).
The findings of a large literature are captured REFERENCES AND NOTES
in Figure 48.2. This figure plots the rate of re- 1. E. I. Knudsen, J. J. Heckman, J. Cameron, J.
P. Shonkoff, Proc. Natl. Acad. Sci. U.S.A., in press.
turn, which is the dollar flow from a unit of in-
2. J. P. Shonkoff, D. Phillips, From Neurons to
vestment at each age for a marginal investment
Neighborhoods: The Science of Early Child Develop-
in a disadvantaged young child at current levels ment (National Academies Press, Washington, DC,
of expenditure. The economic return from early 2000).
interventions is high, and the return from later 3. P. Carneiro, J. J. Heckman, in Inequality in
interventions is lower. Remedial programs in America: What Role for Human Capitol Policies? J. J.
the adolescent and young adult years are much Heckman, A. B. Krueger, B. Friedman, Eds. (MIT
more costly in producing the same level of skill Press, Cambridge, MA, 2003), ch. 2, pp. 77–237.
attainment in adulthood. Most are economically 4. F. Cunha, J. J. Heckman, L. J. Lochner, D. V.
inefficient. This is reflected in Fig. 48.2 by the Masterov, in Handbook of the Economics of Education,
E. A. Hanushek, F. Welch, Eds. (North Holland,
fact that a segment of the curve lies below the
Amsterdam, 2006).
opportunity cost of funds (the horizontal line 5. J. J. Heckman, D. V. Masterov, “The pro-
fixed at r). The opportunity cost is the return ductivity argument for investing in young children”
416 • Part V: Poverty and the Underclass

(Working Paper No. 5, Committee on Economic 20. L. N. Masse, W. S. Barnett, A Benefit Cost
Development, Washington, DC, 2004). Analysis of the Abecedarian Early Childhood Interven-
6. J. B. DeLong, L. Katz, C. Goldin, in Agenda tion (Rutgers University, National Institute for Early
for the Nation, H. Aaron, J. Lindsay, P. Nivola, Eds. Education Research, New Brunswick, NJ, 2002).
(Brookings Institution Press, Washington, DC, 21. J. S. Coleman, Equality of Educational Op-
2003), pp. 17–60. portunity (U.S. Department of Health, Education,
7. D. T. Ellwood, in The Roaring Nineties: Can and Welfare, Office of Education, Washington, DC,
Full Employment Be Sustained? A. Krueger, R. Solow, 1966).
Eds. (Russell Sage Foundation, New York, 2001), 22. S. W. Raudenbush, “Schooling, statistics
pp. 421–489. and poverty: Measuring school improvement and
8. J. J. Heckman, P. LaFontaine, J. Lab. Econ., improving schools” (Inaugural Lecture, Division of
in press. Social Sciences, University of Chicago, Chicago, IL,
9. International Adult Literacy Survey, 2002: User’s 22 February 2006).
Guide, Statistics Canada, Special Surveys Division, 23. J. J. Heckman, M. I. Larenas, S. Urzua, un-
National Literacy Secretariat, and Human Resources published data.
Development Canada (Statistics Canada, Ottawa, 24. D. A. Neal, in Handbook of Economics of Ed-
Ontario, 2002). ucation, E. Hanushek, F. Welch, Eds. (Elsevier, Am-
10. D. A. Anderson, J. Law Econ. 42, 611 (1999). sterdam, 2006).
11. J. J. Heckman, J. Stixrud, S. Urzua, J. Lab. 25. B. Krueger, D. M. Whitmore, Econ. J., 111,
Econ., in press. 1 (2001).
12. Westinghouse Learning Corporation and 26. B. Krueger, D. M. Whitmore, in Bridging
Ohio University, The Impact of Head Start: An Eval- the Achievement Gap, J. E. Chubb, T. Loveless, Eds.
uation of the Effects of Head Start on Children’s Cogni- (Brookings Institution Press, Washington, DC,
tive and Affective Development, vols. 1 and 2 (Report 2002).
to the Office of Economic Opportunity, Athens, 27. W. S. Barnett, Benefit-Cost Analysis of Pre-
OH, 1969). school Education, 2004 (http://nieer.org/resources/
13. L. J. Schweinhart et al., Lifetime Effects: The files/BarnettBenefits.ppt).
High/Scope Perry Preschool Study Through Age 40 28. F. Cunha, J. J. Heckman, J. Hum. Resour.,
(High/Scope, Ypsilanti, MI, 2005). in press.
14. A. Rolnick, R. Grunewald, “Early childhood 29. This paper was generously supported by
development: Economic development with a high NSF (grant nos. SES-0241858 and SES-0099195),
public return” (Tech. rep., Federal Reserve Bank of National Institute of Child Health and Human
Minneapolis, Minneapolis, MN, 2003). Development (NIH grant no. R01HD043411),
15. C. T. Ramey, S. L. Ramey, Am. Psychol. 53, funding from the Committee for Economic De-
109 (1998). velopment, with a grant from the Pew Charitable
16. C. T. Ramey, S. L. Ramey, Prev. Med. 27, Trusts and from the Partnership for America’s Eco-
224 (1998). nomic Success. This research was also supported by
17. C. T. Ramey et al., Appl. Dev. Sci. 4, 2 (2000). the Children’s Initiative project at the Pritzker Fam-
18. A. J. Reynolds, M. C. Wang, H. J. Walberg, ily Foundation and a grant from the Report to the
Early Childhood Programs for a New Century (Child Nation of America’s Promise. The views expressed in
Welfare League of America Press, Washington, DC, this paper are those of the author and not necessarily
2003). those of the sponsoring organizations. See our Web
19. L. A. Karoly et al., Investing in Our Children: site (http://jenni.uchicago.edu/econ_neurosci) for
What We Know and Don’t Know About the Costs and more information.
Benefits of Early Childhood Interventions (RAND,
Santa Monica, CA, 1998).
 Duncan and Magnuson—The Long Reach of Early Childhood Poverty • 417

49 • Greg J. Duncan and Katherine Magnuson

The Long Reach of Early Childhood Poverty

Using a poverty line of about $22,000 for a on deep and persistent poverty occurring very
family of four, the Census Bureau counted early in the childhoods of the poor.
more than 15 million U.S. children living in
poor families in 2009. Poor children begin
school well behind their more affluent age mates American Poverty and Its
and, if anything, lose ground during the school Consequences for Children
years. On average, poor kindergarten children If we were to draw the poverty line at 50 percent
have lower levels of reading and math skills and of median disposable income (about $29,000 for
are rated by their teachers as less well behaved a family of three in today’s dollars), as is com-
than their more affluent peers (see Figure 49.1). mon in much cross-national research on pov-
Children from poor families also go on to com- erty, nearly one-quarter of U.S. children would
plete less schooling, work less, and earn less than be classified as poor (Figure 49.2). Comparing
others. across countries, the U.S. fares badly, though
Social scientists have been investigating links not too much worse than countries like the
between family poverty and subsequent child UK, Canada, and Poland. More striking are
outcomes for decades. Yet, careful thought the cross-country differences when the poverty
about the timing of economic hardship across threshold is set at a more spartan 40 percent of
childhood and adolescence is almost universally median disposable income (about $23,000). In
neglected. Emerging research in neuroscience this instance, the 15 percent U.S. childhood
and developmental psychology suggests that poverty rate is more than half again as high as
poverty early in a child’s life may be particularly any country other than Poland. Clearly, deep
harmful because the astonishingly rapid devel- poverty is considerably more pervasive for chil-
opment of young children’s brains leaves them dren in the U.S. than among children in most
sensitive (and vulnerable) to environmental Western industrialized countries.
conditions. What are the consequences of growing up
After a brief review of possible mechanisms in a poor household? Economists, sociologists,
and the highest quality evidence linking poverty developmental psychologists, and neuroscien-
to negative childhood outcomes, we highlight tists emphasize different pathways by which
emerging research linking poverty occurring as poverty may influence children’s development.
early as the prenatal year to adult outcomes as Economic models of child development focus
far as the fourth decade of life. Based on this evi- on what money can buy. They view families
dence, we discuss how policy might better focus with greater economic resources as being better

Greg J. Duncan and Katherine Magnuson. “The Long Reach of Early Childhood Poverty,” Pathways (Winter
2011), pp. 22–27. Used by permission of the Center Poverty and Inequality, Stanford University.
418

Figure 49.1. Rates of Kindergarten Proficiencies for Poor, Near Poor, and Middle-Class Children

19%
Recognizing letters 41%
72%
13%
Beginning word sounds 22%
39%

79%
Counting and basic shapes 86%
94%
29%
Relative size 43%
65%
4%
Ordinality, sequence 10%
27%

0% 20% 40% 60% 80% 100%

Poor Near Poor Middle Class

Source: Authors’ calculations from the Early Childhood Longitudinal Survey–Kindergarten Cohort

Figure 49.2. Poverty Rates for Young Children

24
21 20
18
15 17
13
20
9 9
7 7 7
7
4 2 4
3
es

nd

k
y
da

en

d
UK

l
ar
an

ea

n
at

na

la

ed

la
nm

Isr
rm
St

er

Po
Ca

Sw
itz

De
Ge
d
ite

Sw
Un

40% of Median– 50% of Median–

$23,000 for a family of 3 $29,000 for a family of 3

Source: Gornick, J. and Jantti, M. (forthcoming). “Child poverty in upper-income countries: Lessons from the Luxem-
bourg Income Study.” In S. B. Kamerman, S. Phipps, and A. Ben-Arieh (Eds.), From Child Welfare to Child Well-Being: An
International Perspective on Knowledge in the Service of Making Policy. A Special Volume in Honor of Alfred J. Kahn. Springer
Publishing Company.
 Duncan and Magnuson—The Long Reach of Early Childhood Poverty • 419

able to purchase or produce important “inputs” between socioeconomic status and various as-
into their young children’s development (e.g., pects of brain function in young children.
nutritious meals; enriched home learning en- Intensive programs aimed at providing early
vironments and child care settings outside the care and educational experiences for high-risk
home; and safe and stimulating neighborhood infants and toddlers also support the idea that
environments), and higher-quality schools and children’s early years are a fruitful time for in-
post-secondary education for older children. The tervention. The best known of these are the
cost of the inputs and family income constraints Abecedarian program, which provided a full-day,
are therefore the key considerations for under- center-based, educational program for chil-
standing poverty’s effects on children. dren who were at high risk for school failure,
Psychologists and sociologists point to the starting in early infancy and continuing until
quality of family relationships to explain pover- school entry, and the Perry Preschool program,
ty’s detrimental effects on children. These the- which provided one or two years of intensive
oretical models point out that higher incomes center-based education for preschoolers. Both of
may improve parents’ psychological well-being these programs have been shown to generate im-
and their ability to engage in positive family pressive longterm improvements in subsequent
processes, in particular high-quality parental education and employment. Perry also produced
interactions with children. A long line of re- large reductions in adult crime.
search has found that low-income parents are
more likely than others to use an authoritarian
and punitive parenting style and less likely to A Causal Story?
provide their children with stimulating learn- Regardless of the timing of low income, isolating
ing experiences in the home. Poverty and eco- its causal impact on children’s well-being is diffi-
nomic insecurity take a toll on parents’ mental cult. Poverty is associated with other experiences
health, which may be an important cause of of disadvantage (such as poor schools or being
low-income parents’ non-supportive parenting. raised by a single parent), making it difficult
Depression and other forms of psychological to know for certain whether it is poverty per se
distress can profoundly affect parents’ interac- that really matters or other related experiences.
tions with their children. But as we argue below, The best method for identifying the extent to
it is not just the fact that these relationships exist which income really matters would be an ex-
that matters, but when. periment that compares families who receive
some additional money to similar parents who
do not receive such money. The only large-scale
Why Early Poverty May Matter Most randomized interventions to alter family income
It is not solely poverty that matters for children’s directly were the Negative Income Tax Experi-
outcomes, but also the timing of child poverty. ments, which were conducted between 1968
For some outcomes later in life, particularly and 1982 with the primary goal of identifying
those related to achievement skills and cognitive the influence of a guaranteed income on parents’
development, poverty early in a child’s life may labor force participation. Researchers found that
be especially harmful. Emerging evidence from elementary school children whose families en-
both human and animal studies highlights the joyed a 50 percent boost in family income from
critical importance of early childhood in brain the program exhibited higher levels of early aca-
development and for establishing the neural demic achievement and school attendance than
functions and structures that shape future cog- children who did not. No test score differences
nitive, social, emotional, and health outcomes. were found for adolescents, although youth
There is also clear evidence emerging from neu- who received the income boost did have higher
roscience that demonstrates strong correlations rates of high school completion and educational
420 • Part V: Poverty and the Underclass

attainment. This suggests that higher income policy change. A second, Canadian study found
may indeed cause higher achievement, although similar results when researchers took advantage
even in this case it is impossible to distinguish of variation across Canadian provinces in the
the effects of income from the possible benefits generosity of Canada’s National Child Benefit
to children from the reductions in parental work program to estimate income impacts on child
effort that accompanied the income increases. achievement. Thus, the weight of the evidence
According to newer experimental welfare suggests that increases in income for poor fami-
reform evaluations in the 1990s, though, pro- lies are causally related to improvements in chil-
viding income support to working poor parents dren’s outcomes.
through wage supplements does improve chil-
dren’s achievement. One study analyzed data
from seven random-assignment welfare and anti- The Long Reach
poverty policies. All of these policies increased None of this past income literature has been
parental employment, while only some in- able to examine family income early in a child’s
creased family income. These analyses indicated life in relation to that child’s adult attainments.
improved academic achievement for preschool This limitation comes largely from the lack of
and elementary school children by programs data on both early childhood income and later
that boosted both income and parental employ- adult outcomes. Recent research by Duncan and
ment, but not by programs that only increased his colleagues, however, has now made this link
employment. using recently released data from the Panel Study
These experimental findings suggest that of Income Dynamics, which has followed a na-
income plays a causal role in boosting younger tionally representative sample of U.S. families
children’s achievement, although here it should and their children since 1968. The study is based
be kept in mind that the beneficial welfare-to- on children born between 1968 and 1975, for
work programs increased both income and pa- whom adult outcomes were collected between
rental employment. However, combining these ages 30 and 37.
results with those from the 1970s experiments, Measures of income were available in every
we note that both kinds of programs increased year of a child’s life from the prenatal period
income but produced opposing impacts on work through age 15. This enabled Duncan and his
hours. This suggests that the income boost may colleagues to measure poverty across several dis-
have been the most active ingredient in promot- tinct periods of childhood, distinguishing in-
ing children’s achievement. come early in life (prenatal through age 5) from
Non-experimental studies that take care to income in middle childhood and adolescence.
ensure they are comparing families who differ The simple associations between income early in
in terms of income, but who are otherwise sim- life and adult outcomes are striking (Table 49.1).
ilar, can also provide strong evidence. One such Compared with children whose families had in-
study took advantage of an increase in the max- comes of at least twice the poverty line during
imum Earned Income Tax Credit for working their early childhood, poor children completed
poor families with more than two children by two fewer years of schooling, earned less than
more than $2,000 between the years of 1993 half as much money, worked 451 fewer hours
and 1997. This generous increase in tax bene- per year, received $826 per year more in food
fits enabled researchers to compare the school stamps, and are nearly three times as likely to
achievement of children in otherwise similar— report poor overall health. Poor males are more
and even the same—working families before and than twice as likely to be arrested. For females,
after the increase in the tax credit. And indeed, poverty is associated with a more than five fold
improvements in low-income children’s achieve- increase in the likelihood of bearing a child out
ment in middle childhood coincided with the of wedlock prior to age 21.
 Duncan and Magnuson—The Long Reach of Early Childhood Poverty • 421

Table 49.1. Adult Outcomes by Poverty Status Between the Prenatal Year and Age Five
Income below the official Income between one and Income more than
U.S. poverty line two times the poverty line twice the poverty line
Mean or % Mean or % Mean or %
Completed schooling 11.8 yrs 12.7 yrs 14.0 yrs
Earnings ($10,000) $17.9 $26.8 $39.7
Annual work hours 1,512 1,839 1,963
Food stamps $896 $337 $70
Poor health 13% 13% 5%
Arrested (men only) 26% 21% 13%
Nonmarital birth
50% 28% 9%
(women only)

Note: Earnings and food stamp values are in 2005 dollars.

None of these simple comparisons, however, association between overall childhood income
considered the various factors that go along with and health and non-marital birth seems to be
growing up in poverty that also might explain largely attributable to income during adoles-
poorer adult outcomes (e.g., single parenthood cence, rather than earlier in childhood.
or lack of motivation). To account for this, we More detailed analyses show that for fami-
also adjusted for an extensive set of background lies with average early childhood incomes below
control variables, all of which were measured $25,000, a $3,000 annual boost to family in-
either before or near the time of birth. This ef- come is associated with a 17 percent increase in
fort to separate income from other related dis- adult earnings (Figure 49.3). Results for work
advantages and characteristics of poor children hours are broadly similar to those for earnings.
produces smaller correlations than in the absence In this case, a $3,000 annual increase in the pre-
of these statistical controls. This suggests that a natal to age 5 income of low-income families
substantial portion of the simple correlation is associated with 135 additional work hours
between childhood income and most adult out- per year after age 25. In contrast, increments
comes can be accounted for by the disadvan- to early-childhood income for higher-income
tageous conditions associated with birth into a children were not significantly associated with
low-income household. higher adult earnings or work hours. The im-
But what about the timing of poverty? To plication is clear: If we are hoping that giving
better understand whether poverty in early child- parents extra income will bolster their children’s
hood is particularly important, Duncan and col- chances for success, early childhood is the time
leagues replaced the average childhood income to do it.
measure with three stage-specific measures of
income. As before, adjustments are made for
the effects of the extensive array of background Refashioning Income Supports
conditions. Early childhood is a particularly sensitive period
In the case of adult earnings and work hours, in which economic deprivation may compro-
early childhood income appears to matter much mise children’s life achievement and employment
more than later income. For some measures, like opportunities. Research continues to confirm a
work hours, there appears to even be a negligi- remarkable sensitivity (and growing number) of
ble role for income beyond age 5. Early income developing brain structures and functions that are
also appears to matter for completed schooling, related to growing up in an impoverished home.
but in this case adolescent family income seems We also have convincing evidence linking
to matter even more. In contrast, the strong early poverty with both child achievement and
422 • Part V: Poverty and the Underclass

Figure 49.3. Percentage Increase in Adult Earnings Associated with a $3,000 Annual Increase in Child-
hood Income

20

15 20

10

5
4 1
2 0 0
0
Prenatal Age 6–10 Age 11–15
to age 5

< $25,000 $25,000 +

adult employment. The achievement studies em- of welfare policies, we should take care to ensure
ploy unusually rigorous methods for estimating that sanctions and other regulations do not deny
causal relationships between income early in life benefits to families with very young children.
and achievement test scores as children age. The Not only do young children appear to be most
effect sizes estimated in these studies are broadly vulnerable to the consequences of deep poverty,
similar. An annual income increase of $3,000 but mothers with very young children are also
sustained for several years appears to boost chil- least able to support themselves through employ-
dren’s achievement by roughly one-fifth of a ment in the labor market.
standard deviation. In the early grades, children’s A more generous, and perhaps smarter, ap-
achievement increases by nearly one standard de- proach would be enacting income transfer pol-
viation per year, so 20 percent of a standard devi- icies that provide more income to families with
ation amounts to about two months’ advantage young children. In the case of work support pro-
in school. grams like the Earned Income Tax Credit, this
Very recent research has linked poverty early might mean extending more generous credits to
in childhood to adult earnings and work hours. families with young children. In the case of child
Although non-experimental, the study’s key tax credits, this could mean making the credit
finding—that income early in childhood appears refundable and also providing larger credits to
to matter much more than income later in child- families with young children.
hood for a range of employment outcomes—is Interestingly, several European countries gear
strikingly consistent with the achievement time-limited benefits to the age of children. In
studies. Germany, a modest parental allowance is avail-
Taken together, this research suggests that able to a mother working fewer than 20 hours
greater policy attention should be given to reme- per week until her child is 18 months old. France
diating situations involving deep and persistent guarantees a modest minimum income to most
poverty occurring early in childhood. In the case of its citizens, including families with children of
 Evans, Brookes-Gunn, and Klebanov—Stressing Out the Poor • 423

all ages. Supplementing this basic support is the is the only policy path worth pursuing. Obvi-
Allocation de Parent Isolé (API) program for sin- ously investments later in life, including those
gle parents with children under age 3. In effect, that provide direct services to children and fam-
the API program acknowledges a special need ilies, may also be well-advised. Economic logic
for income support during this period, especially requires a comparison of the costs and benefits
if a parent wishes to care for very young chil- of the various programs that seek to promote the
dren and forgo income from employment. The development of disadvantaged children through-
state-funded child care system in France that be- out the life course. In this context, expenditures
gins at age 3 alleviates the problems associated on income-transfer and service-delivery pro-
with a parent’s transition into the labor force. grams should be placed side by side and judged
In emphasizing the potential importance of by their costs and benefits, with the utmost goal
policies to boost income in early childhood, we of making our social investments as profitable
do not mean to imply that focusing on this area as possible.

50 • Gary W. Evans, Jeanne Brooks-Gunn, and Pamela Kato Klebanov

Stressing Out the Poor

It is well known that economic deprivation early risks, in turn, build upon one another to ele-
in life sets children on a trajectory toward dimin- vate levels of chronic (and toxic) stress within
ished educational and occupational attainment. the body. And this toxic stress directly hinders
But why is early-childhood poverty so harmful? poor children’s academic performance by com-
If we can’t answer that question well, our reform promising their ability to develop the kinds of
efforts are reduced to shots in the dark. skills necessary to perform well in school.
In this article, we offer a new perspective on We will unpack this new Risk–Stress Model
this question. We suggest that childhood poverty in the balance of our article. However, before
is harmful, in part, because it exposes children doing so, it’s useful to first go over the evidence
to stressful environments. Low-income children regarding the relation between poverty and
face a bewildering array of psychosocial and achievement and then to present some of the
physical demands that place much pressure on well-known pathways through which this re-
their adaptive capacities and appear to be toxic lationship is generated. With that background
to the developing brain. Although poor children in place, we can then describe the Risk–Stress
are disadvantaged in other ways, we focus our Model, as represented in Figure 50.1.
analysis here on the new, underappreciated path-
way depicted in Figure 50.1. As shown in this
figure, children growing up in poverty demon- Poverty and Achievement
strate lower academic achievement because of It is well known that children born into
their exposure to a wide variety of risks. These low-income families lag behind their middle- and

Gary W. Evans, Jeanne Brooks-Gunn, and Pamela Kato Klebanov. “Stressing Out the Poor: Chronic Psycho-
logical Stress and the Income-Achievement Gap,” Pathways (Winter 2011), pp. 16–21. Used by permission of
the Center for the Study of Poverty and Inequality, Stanford University.
424 • Part V: Poverty and the Underclass

Figure 50.1. A New Pathway to Account for the Income-Achievement Gap

POVERTY CUMULATIVE RISK EXPOSURE CHRONIC STRE SS LOW ACHIEVEMENT

Figure 50.2. Average Percentile Rank on Peabody Individual Achievement Math

65

60 Highest income quartile

SCORE PERCENTILE

55
Third income quartile
50

45
Second income quartile
40
Lowest income quartile
35
6 8 10 12
AGE OF CHILD

Source: James J. Heckman. “Skill Formation and the Economics of Investing in Disadvantaged Children.” Science,
312(5782): 1900–1902.

upper-income counterparts on virtually all in- expectations, to less rigorous curricula, and to
dices of achievement. To provide one example, less capable peers, all of which further disadvan-
a national study of elementary school children tage them and generate ever more substantial
shows that children in the poorest quarter of between-group gaps. The Risk–Stress Model, to
American households begin kindergarten nearly which we turn later, suggests that such splaying
10 percent behind their middle-income and af- may also be attributed to the cognitive deficits
fluent classmates in math (Figure 50.2). Six years and poorer health that chronic stress generates.
later, as they are about to enter middle school, Both cognitive deficits and ill health then repeat-
the poorest quarter of American children have edly disadvantage poverty-stricken children in
fallen even further behind, with the gap between one educational setting after another.
themselves and their most affluent schoolmates
nearly doubling.
The splaying pattern revealed here, a gen- Parenting Practices
eral one that holds across various outcomes, What types of forces have social scientists
may be attributed to the tendency for advan- conventionally understood as explaining the
tage and disadvantage to accumulate over time. achievement gaps illustrated in Figure 50.2?
This accumulation occurs in various ways; for One reason poor children lag behind their more
example, children who score poorly at age six affluent peers is that their parents interact with
may be tracked into low-achievement school them in ways that aren’t conducive to achieve-
groups, which in turn exposes them to lower ment. For example, psychologist Kathryn Grant
 Evans, Brookes-Gunn, and Klebanov—Stressing Out the Poor • 425

and her colleagues have documented a strong Children from impoverished households face a
and consistent relation between socioeconomic wide array of physical and psychosocial stress-
disadvantage and harsh, unresponsive parenting. ors. Their homes, schools, and neighborhoods
In one national dataset, 85 percent of American are much more chaotic than the settings in
parents above the poverty line were shown to which middle- and upper-income children grow
be responsive, supportive, and encouraging to up. Such conditions can, in turn, produce toxic
their children during infancy and toddlerhood, stress capable of damaging areas of the brain
whereas only 75 percent of low-income parents known to underlie cognitive processes—such as
had the same achievement-inducing parenting attention, memory, and language—that all com-
style. While most low-income parents (i.e., 75 bine to undergird academic success. In the pages
percent) do provide adequate levels of support that remain, we document each of the steps in
and encouragement, these data reveal, then, a the Risk–Stress Model.
nontrivial difference across income levels in the
chances that children will experience a prob-
lematic parenting style. There is considerable Poverty and Cumulative Risk
evidence that at least a portion of the cognitive Exposure
developmental consequences of early childhood The stressors that poor children face take both
poverty is due to this difference. a physical and psychosocial form. The physical
form is well documented; poor children are ex-
posed to substandard environmental conditions
Cognitive Stimulation including toxins, hazardous waste, ambient
It’s also well known that children from air and water pollution, noise, crowding, poor
low-income households tend to receive less cog- housing, poorly maintained school buildings,
nitive stimulation and enrichment. For example, residential turnover, traffic congestion, poor
a child from a low-income family who enters neighborhood sanitation and maintenance, and
first grade has been exposed on average to just crime. The psychosocial form is also well doc-
25 hours of one-on-one picture book reading, umented; poor children experience significantly
whereas an entering middle-income child has higher levels of family turmoil, family separa-
been exposed on average to more than 1,000 tion, violence, and significantly lower levels of
hours of such reading. Likewise, during the structure and routine in their daily lives.
first three years of life, a child with professional An important aspect of early, disadvantaged
parents will be exposed to three times as many settings may be exposure to more than one risk
words as a child with parents on welfare. factor at a time. A powerful way to capture expo-
And it’s not just simple parental effects that sure to such multiple sources of stress and strain
account for the achievement deficit. If a child is is the construct of cumulative risk. Although
born into a high-income family, he or she may there are various ways to quantify cumulative
also benefit from high-quality stimulation and risk, one common approach is to simply count
enrichment from extended family, from sib- the number of physical or psychosocial risks
lings and friends, and from more formal care to which a child has been exposed. In one UK
providers. All of this redounds to the benefit of study, the authors counted how often children
higher-income children while further handicap- were exposed to such stresses as (a) living with
ping low-income children. a single parent, (b) experiencing family discord,
So much for the well-known pathways by (c) experiencing foster or some other form of in-
which disadvantage is transmitted. We turn stitutional care, (d) living in a crowded home,
now to another and less-appreciated aspect of and (e) attending a school with high turnover
low-income environments that may also harm of both classmates and teachers. It was found in
cognitive development. The key concern here: this study that inner-city children experienced
426 • Part V: Poverty and the Underclass

Figure 50.3. Cumulative Risk Exposure Among Low- and Middle-Income Rural Nine-Year-Olds

60
Poverty
PERCENTAGE OF SCHOOL CHILDREN

50
Middle-Income
40
EXPOSED

30

20

10

0
0 1 2 3 4 5 6
NUMBER OF CUMULATIVE RISKS

Note: Cumulative risks include family turmoil, violence, child separation from family, noise, crowding, and housing quality.
Source: Gary W. Evans and Kimberly English. 2002. “The Environment of Poverty: Multiple Stressor Exposure, Psycho-
physiological Stress, and Socioemotional Adjustment.” Child Development, 73(4):1238–48.

far more of these stresses than did the better-off simple answer is that it does. In cross-sectional
working-class children. The same result holds in analyses of 9- and 13-year-old children, Evans
the United States (see Figure 50.3). In rural New and colleagues found that the risk exposure
England, only 12 percent of middle-income described in Figure 50.3 elevated baseline, rest-
nine-year-olds experienced three or more physi- ing blood pressure as well as overnight indices
cal and psychosocial risk factors, whereas nearly of such stress hormones as cortisol. At age 13,
50 percent of low-income children crossed this when challenged by mental arithmetic prob-
same threshold (of three risk factors). lems, children with higher levels of cumulative
In a national U.S. sample of premature risk exposure did not show a typical healthy re-
and low birth weight infants, Brooks-Gunn sponse, instead exhibiting a muted rise in blood
and colleagues similarly found that infants pressure. These same children also didn’t recover
born into low-income families experienced as successfully from the mental challenge posed
nearly three times more risk factors than their by these arithmetic problems (as indexed by the
middle-income counterparts by the time they longer time it took their blood pressure to return
were toddlers. These same low-income toddlers to pre-stressor baseline levels). The evidence thus
were seven times more likely than their affluent suggests that children exposed to high levels of
counterparts to experience a very high number cumulative risk are less efficient both in mobiliz-
of risk factors (≥ 6). The pattern is overwhelm- ing and then shutting off physiological activity.
ingly clear: Being born into early poverty often The Risk–Stress Model, as represented in Fig-
means exposure to many more physical and psy- ure 50.1, implies that the effect of family poverty
chosocial risk factors. on stress is mediated by risk exposure. Although
one would ideally like to test that mediation, it’s
also important to simply document the associa-
Cumulative Risk Exposure and tion between poverty and stress (thereby ignoring
Chronic Stress the mediating factor). Many investigators have
But does such differential exposure indeed result indeed documented that disadvantaged children
in higher stress levels among poor children? The have higher chronic physiological stress levels, as
 Evans, Brookes-Gunn, and Klebanov—Stressing Out the Poor • 427

Figure 50.4. Resting Blood Pressure in Nine-Year-Old, White Rural Children

diastolic systolic

61 102.5
60.5 102
60 101.5
59.5 101
MM/HG

59
100.5
58.5
58 100
57.5 99.5
57 99
POVERTY MIDDLE INCOME POVERTY MIDDLE INCOME

Source: Gary W. Evans and Kimberly English. 2002. “The Environment of Poverty: Multiple Stressor Exposure, Psycho-
physiological Stress, and Socioemotional Adjustment.” Child Development, 73(4):1238–48.

Figure 50.5. Overnight Stress Hormones in Nine-Year-Old, White Rural Children

cortisol epinephrine norepinephrine

.035 6 33
UG/MG CREATINE

NG/MG CREATINE

NG/MG CREATINE

.03 5
.025 32
4
.02
3 31
.015
.01 2
30
.005 1
0 0 29
POVERTY MIDDLE POVERTY MIDDLE POVERTY MIDDLE
INCOME INCOME INCOME

Source: Gary W. Evans and Kimberly English. 2002. “The Environment of Poverty: Multiple Stressor Exposure, Psycho-
physiological Stress, and Socioemotional Adjustment.” Child Development, 73(4):1238–48.

indicated by elevated resting blood pressure. A Program (IHDP) is a representative sample of

smaller number of studies have also uncovered low birth weight (≤ 2500 grams) and premature
higher levels of chronic stress hormones, such (≤ 37 weeks gestational age) babies born in 1985
as cortisol, among disadvantaged children. To at eight medical centers throughout the country.
provide just a few examples, Figures 50.4 and This sample of nearly 1,000 babies is racially and
50.5 show elevated resting blood pressure as well economically diverse (52 percent Black, 37 per-
as higher overnight urinary stress hormones in a cent White, 11 percent Hispanic).
sample of nine-year-old rural children. We assessed resting blood pressure and child’s
The foregoing data, which pertain to height and weight at 24, 30, 36, 48, 60, and 78
nine-year-olds, don’t tell us when such stress months of age. The collection of physical health
symptoms emerge. Do poverty-stricken chil- data at such young ages and over time provided
dren show evidence of elevated stress early on in us with an unprecedented opportunity to ex-
their lives? Or do such symptoms only emerge amine the early trajectories of chronic stress
later? We sought to answer this question by rean- among a high-risk sample of babies. Both base-
alyzing a national data set of very young at-risk line blood pressure levels and Body Mass Index
children. The Infant Health and Development (BMI) reflect wear and tear on the body and
428

Figures 50.6–50.7. Developmental Trajectories in Chronic Stress in Relation to Neighborhood Poverty

BMI from 24 to 78 Months by Neighborhood Poverty

17
BMI

16

15
24 36 48 60 72
AGE IN MONTHS

Affluent Neighbors Near Poor & Poor Neighbors

Diastolic Blood Pressure from 24 to 78 Months by Neighborhood Poverty

(with controls for blood pressure method & child’s state)

70

65
DBP

60

55

50

24 36 48 60 72
AGE IN MONTHS

Affluent Neighbors Near Poor & Poor Neighbors

 Evans, Brookes-Gunn, and Klebanov—Stressing Out the Poor • 429

are precursors of lifelong health problems. The the brain appear vulnerable to early childhood
former is indicative of cardiovascular health and deprivation. Using batteries of neurocognitive
the latter of metabolic equilibrium. BMI, which tests of brain function and brain imaging stud-
reflects fat deposition, is measured as height di- ies, Farah and other neuroscientists can map the
vided by weight (kg/m2). areas of the brain that are recruited by neuro-
We sought to assess whether these two mea- cognitive tasks. As shown in Figure 50.8, among
sures of stress are elevated in poverty-stricken the areas of the brain most sensitive to childhood
neighborhoods. Low-income neighborhoods, SES are language, long-term memory (LTM),
as defined in our study, have median household working memory (WM), and executive con-
incomes below $30,000 (in 1980 dollars), while trol. What the graph depicts is the separation,
middle-income neighborhoods have median in- in standard deviation units, between a low- and
come levels exceeding $30,000 per household. middle-SES sample of 11-year-old Black chil-
As is evident in Figures 50.6 and 50.7, babies dren from Philadelphia. For this sample, one
growing up in low-income neighborhoods have standard deviation represents about one-fifth of
health trajectories indicative of elevated chronic the total distribution of scores. Samples differing
stress. Additional statistical controls for infant by 3.5 or more standard deviations are virtually
birth weight, health, and demographic charac- non-overlapping. Given that the samples differ
teristics did not alter these trajectories. These by about 3.5 standard deviations for all four areas
figures also reveal, even more importantly, that of brain functioning, this means that there is vir-
elevated stress emerges very early for children tually no overlap between poor and middle-class
growing up in low-income neighborhoods. Black children when it comes to language,
BMI, for example, proves to be unusually low long-term memory, working memory, or exec-
among poor children under five years old, but it utive control. Eleven-year-old Black children
then takes off as these children grow older. The from lower SES families reveal dramatic deficits
blood pressure measure, by contrast, registers in multiple, basic cognitive functions critical to
high among low-income children from almost learning and eventual success in society. These
the very beginning of our measurements (i.e., results reveal the starkly cognitive foundation to
24 months). This research confirms, then, that the poor performance of low-income children.
low-income children are more likely than others But is this achievement gap attributable to
to develop dangerous stress trajectories very early cumulative risk and chronic stress? With a re-
on in their childhood. As we discuss below, this cent follow-up of the sample depicted in Figures
has profound consequences for their likelihood 50.4 and 50.5, Evans and colleagues have now
of success in school and beyond. provided the first test of the final link in the
Risk–Stress Model. The baseline finding from
their research is that working memory in early
Chronic Stress and the adulthood (i.e., age 17) deteriorated in direct re-
Achievement Cap lation to the number of years the children lived
The next and final step in our chain model in poverty (from birth through age 13). If, in
pertains to the effects of chronic stress on other words, a child lived in poverty continu-
achievement. Here we turn to an important ously, his or her working memory was greatly
longitudinal program on poverty and the brain compromised. The main result of interest, how-
at the University of Pennsylvania conducted by ever, was that such deterioration occurred only
Martha Farah and her colleagues. In a series of among poverty-stricken children with chron-
studies with multiple samples drawn from lower- ically elevated physiological stress (as measured
and middle-class Black families in Philadelphia, between ages 9 and 13). That is, chronic early
Farah and colleagues show that several areas of childhood poverty did not lead to working
430 • Part V: Poverty and the Underclass

Figure 50.8. Effect Sizes Measured in Standard Deviations of Separation Between Low- and Middle-SES
10-to-12-Year-Old, African American Children

4
3.5
3
2.5
SD SEPARATOR

2
1.5
1
.5
0
LANGUAGE LTM WM EXECUTIVE
CONTROL

Source: Martha J. Farah, David M. Shera, Jessica H. Savage, Laura Betancourt, Joan M. Giannetta, Nancy L. Brodsky,
Elsa K. Malmud, and Hallam Hurt. 2006. “Childhood Poverty: Specific Associations with Neurocognitive Development.”
Brain Research, 1110(1): 166–74.

memory deficits among children who somehow early in life. Even young babies growing up in
avoided experiencing the stress that usually ac- low-income neighborhoods already evidence el-
companies poverty. evated chronic stress. This stress then accounts
for a significant portion of the association be-
tween poverty and working memory, a critical
Conclusion cognitive skill involved in language and reading
Childhood socioeconomic disadvantage leads acquisition.
to deficits in academic achievement and occu- The Risk–Stress Model suggests that the
pational attainment. It’s long been argued that poverty-achievement link can be broken by ad-
such deficits arise because poor children are ex- dressing (a) the tendency of poverty to be asso-
posed to inadequate cognitive stimulation and ciated with physical or psychosocial risks (e.g.,
to parenting styles that don’t encourage achieve- environmental toxins, family turmoil), (b) the
ment. We don’t dispute the important role of effects of such risks on stress, and (c) the effects
these two variables. But we have outlined here of stress on achievement. If this model bears up
evidence for a new, complementary pathway that under further testing, it would be useful to ex-
links early childhood poverty to high levels of plore which of these pathways is most amenable
exposure to multiple risks, which in turn elevates to intervention.
chronic toxic stress. This cascade can begin very
 Western and Pettit—Incarceration and Social Inequality • 431

Incarceration and Poverty

51 • Bruce Western and Becky Pettit

Incarceration and Social Inequality

In the last few decades, the institutional contours main reasons: it is invisible, it is cumulative, and
of American social inequality have been trans- it is intergenerational. The inequality is invisible
formed by the rapid growth in the prison and in the sense that institutionalized populations
jail population.1 America’s prisons and jails have commonly lie outside our official accounts of
produced a new social group, a group of social economic well-being. Prisoners, though drawn
outcasts who are joined by the shared experience from the lowest rungs in society, appear in no
of incarceration, crime, poverty, racial minority, measures of poverty or unemployment. As a re-
and low education. As an outcast group, the men sult, the full extent of the disadvantage of groups
and women in our penal institutions have lit- with high incarceration rates is underestimated.
tle access to the social mobility available to the The inequality is cumulative because the social
mainstream. Social and economic disadvantage, and economic penalties that flow from incarcer-
crystallizing in penal confinement, is sustained ation are accrued by those who already have the
over the life course and transmitted from one weakest economic opportunities. Mass incarcer-
generation to the next. This is a profound in- ation thus deepens disadvantage and forecloses
stitutionalized inequality that has renewed race mobility for the most marginal in society. Fi-
and class disadvantage. Yet the scale and empir- nally, carceral inequalities are intergenerational,
ical details tell a story that is largely unknown. affecting not just those who go to prison and jail
Though the rate of incarceration is histor- but their families and children, too.
ically high, perhaps the most important social
fact is the inequality in penal confinement. This The scale of incarceration is measured by a rate
inequality produces extraordinary rates of incar- that records the fraction of the population in
ceration among young African American men prison or jail on an average day. From 1980 to
with no more than a high school education. For 2008, the U.S. incarceration rate climbed from
these young men, born since the mid-1970s, 221 to 762 per 100,000. In the previous five de-
serving time in prison has become a normal life cades, from the 1920s through the mid-1970s,
event. the scale of punishment in America had been
The influence of the penal system on social stable at around 100 per 100,000. Though
and economic disadvantage can be seen in the the incarceration rate is now nearly eight times
economic and family lives of the formerly incar- its historic average, the scale of punishment
cerated. The social inequality produced by mass today gains its social force from its unequal
incarceration is sizable and enduring for three distribution.

Bruce Western and Becky Pettit. “Incarceration & Social Inequality,” Daedalus 139, issue 3 (Summer 2010),
pp. 8–19. Copyright © 2010 by Massachusetts Institute of Technology.
432 • Part V: Poverty and the Underclass

Like criminal activity, prisons and jails are educated. State prisoners average just a tenth
overwhelmingly a male affair. Men account grade education, and about 70 percent have no
for 90 percent of the prison population and a high school diploma.3
similar proportion of those in local jails. The Disparities of race, class, gender, and age have
social impact of mass incarceration lies in the produced extraordinary rates of incarceration
gross asymmetry of community and family at- among young African American men with lit-
tachment. Women remain in their communities tle schooling. Figure 51.1 shows prison and jail
raising children while men confront the possi- incarceration rates for men under age thirty-five
bility of separation through incarceration.2 Age in 1980, at the beginning of the prison boom,
intensifies these effects: incarceration rates are and in 2008, after three decades of rising incar-
highest for those in their twenties and early thir- ceration rates. The figure reports incarceration
ties. These are key years in the life course, when separately for whites, Latinos, and African Amer-
most men are establishing a pathway through icans and separately for three levels of education.
adulthood by leaving school, getting a job, and Looking at men with a college education, we
starting a family. These years are important not see that incarceration rates today have barely
just for a man’s life trajectory, but also for the increased since 1980. Incarceration rates have
family and children that he helps support. increased among African Americans and whites
The profound race and class disparities in who have completed high school. Among young
incarceration produce the new class of social African American men with high school diplo-
outsiders. African Americans have always been mas, about one in ten is in prison or jail.
incarcerated at higher rates than whites, at least Most of the growth in incarceration rates is
since statistics were available from the late nine- concentrated at the very bottom, among young
teenth century. The extent of racial disparity, men with very low levels of education. In 1980,
however, has varied greatly over the past cen- around 10 percent of young African American
tury, following a roughly inverse relationship to men who dropped out of high school were in
the slow incorporation of African Americans as prison or jail. By 2008, this incarceration rate
full citizens in American society. In the late nine- had climbed to 37 percent, an astonishing level
teenth century, U.S. Census data show that the of institutionalization given that the average in-
incarceration rate among African Americans was carceration rate in the general population was
roughly twice that of whites. The demographic 0.76 of 1 percent. Even among young white
erosion of Jim Crow through the migration of dropouts, the incarceration rate had grown re-
Southern African Americans to the North in- markably, with around one in eight behind bars
creased racial disparity in incarceration through by 2008. The significant growth of incarceration
the first half of the twentieth century. (Racial rates among the least educated reflects increasing
disparities in incarceration have always been class inequality in incarceration through the pe-
higher in the North than the South.) By the late riod of the prison boom.
1960s, at the zenith of civil rights activism, the These incarceration rates provide only a
racial disparity had climbed to its contemporary snapshot at a point in time. We can also exam-
level, leaving African Americans seven times ine the lifetime chance of incarceration—that
more likely to be in prison or jail than whites. is, the chance that someone will go to prison at
Class inequalities in incarceration are re- some point in his or her life. This cumulative
flected in the very low educational level of those risk of incarceration is important if serving time
in prison and jail. The legitimate labor market in prison confers an enduring status that affects
opportunities for men with no more than a high life chances after returning to free society. The
school education have deteriorated as the prison lifetime risk of imprisonment describes how
population has grown, and prisoners them- many people are at risk of these diminished life
selves are drawn overwhelmingly from the least chances.
 Western and Pettit—Incarceration and Social Inequality • 433

Figure 51.1. Percentage of Men Aged Twenty to Thirty-Four in Prison or Jail, by Race/Ethnicity and
Education, 1980 and 2008

Source: Becky Pettit, Bryan Sykes, and Bruce Western, “Technical Report on Revised Population Estimates and NLSY79
Analysis Tables for the Pew Public Safety and Mobility Project” (Harvard University, 2009).

We calculated the cumulative chance of im- likelihood of serving prison time. For the older
prisonment for two birth cohorts, one born just post-war cohort who reached their mid-thirties
after World War II, from 1945 to 1949, and an- at the end of the 1970s, about one in ten Afri-
other born from 1975 to 1979 (Table 51.1). For can American men served time in prison. For
each cohort, we calculated the chances of impris- the younger cohort born from 1975 to 1979,
onment, not jail incarceration. Prisons are the the lifetime risk of imprisonment for African
deep end of the criminal justice system, now in- American men had increased to one in four.
carcerating people for an average of twenty-eight Prison time has become a normal life event for
months for a felony conviction. While there are African American men who have dropped out
about ten million admissions to local jails each of high school. Fully 68 percent of these men
year—for those awaiting trial or serving short born since the mid-1970s have prison records.
sentences—around seven hundred thousand The high rate of incarceration has redrawn the
prisoners are now admitted annually to state and pathway through young adulthood. The main
federal facilities. sources of upward mobility for African Ameri-
These cumulative chances of imprisonment can men—namely, military service and a college
are calculated up to age thirty-four. For most degree—are significantly less common than a
of the population, this represents the lifetime prison record. For the first generations growing
434 • Part V: Poverty and the Underclass

Table 51.1. Cumulative Risk of Imprisonment by Age Thirty to Thirty-Four for Men Born from 1945 to
1949 and 1975 to 1979, by Educational Attainment and Race/Ethnicity
All High School Dropouts High School/ GED* College
– cohort
White 1.4 3.8 1.5 0.4
Black 10.4 14.7 11.0 5.3
Latino 2.8 4.1 2.9 1.1
– cohort
White 5.4 28.0 6.2 1.2
Black 26.8 68.0 21.4 6.6
Latino 12.2 19.6 9.2 3.4
* Denotes completed high school or equivalency.
Source: Becky Pettit, Bryan Sykes, and Bruce Western, “Technical Report on revised Population Estimates and NLSY79
Analysis Tables for the Pew Public Safety and Mobility Project” (Harvard University, 2009).

up in the post–civil rights era, the prison now the population. The segregation and social con-
looms as a significant institutional influence on centration of incarceration thus help conceal its
life chances. effects. This fact is particularly important for
public policy because in assessing the social and
The ubiquity of penal confinement in the lives economic well-being of the population, the in-
of young African American men with little carcerated fraction is frequently overlooked, and
schooling is historically novel, emerging only in inequality is underestimated as a result.
the last decade. However, this new reality is only The idea of invisible inequality is illustrated
half the story of understanding the significance by considering employment rates as they are
of mass incarceration in America. The other half conventionally measured by the Current Popu-
of the story concerns the effects of incarcera- lation Survey, the large monthly labor force sur-
tion on social and economic inequality. Because vey conducted by the Census Bureau. For groups
the characteristic inequalities produced by the that are weakly attached to the labor market, like
American prison boom are invisible, cumulative, young men with little education, economic status
and intergenerational, they are extremely endur- is often measured by the employment-to-popu-
ing, sustained over lifetimes and passed through lation ratio. This figure, more expansive than the
families. unemployment rate, counts as jobless those who
Invisible Inequality. The inequality created by have dropped out of the labor market altogether.
incarceration is often invisible to the mainstream The Current Population Survey is drawn on a
of society because incarceration is concentrated sample of households, so those who are institu-
and segregative. We have seen that steep racial tionalized are not included in the survey-based
and class disparities in incarceration have pro- description of the population.
duced a generation of social outliers whose col- Figure 51.2 shows the employment-to-popu-
lective experience is wholly different from the rest lation ratio for African American men under age
of American society. The extreme concentration thirty-five who have not completed high school.
of incarceration rates is compounded by the ob- Conventional estimates of the employment rate
viously segregative function of the penal system, show that by 2008, around 40 percent of African
which often relocates people to far-flung facili- American male dropouts were employed. These
ties distant from their communities and families. estimates, based on the household survey, fail to
As a result, people in prison and jail are discon- count that part of the population in prison or
nected from the basic institutions—households jail. Once prison and jail inmates are included
and the labor market—that dominate our in the population count (and among the job-
common understanding and measurement of less), we see that employment among young
 Western and Pettit—Incarceration and Social Inequality • 435

Figure 51.2. Employment-to-Population Ratio, African American Men Aged Twenty to Thirty-Four with
Less than Twelve Years of Schooling, 1980 to 2008

Source: Becky Pettit, Bryan Sykes, and Bruce Western, “Technical Report on Revised Population Estimates and NLSY79
Analysis Tables for the Pew Public Safety and Mobility Project” (Harvard University, 2009).

African American men with little schooling fell incarcerated respondents in prison. The NLSY
to around 25 percent by 2008. Indeed, by 2008 began in 1979, when its panel of respondents
these men were more likely to be locked up than was aged fourteen to twenty-one; it completed
employed. its latest round of interviews in 2006. Matching
Cumulative Inequality. Serving time in prison our population estimates of incarceration, one
or jail diminishes social and economic oppor- in five African American male respondents in
tunities. As we have seen, these diminished op- the NLSY has been interviewed at some point
portunities are found among those already most between 1979 and 2006 while incarcerated,
socioeconomically disadvantaged. A burgeon- compared to 5 percent of whites and 12 percent
ing research literature examining the economic of Latino respondents. Analysis of the NLSY
effects of incarceration finds that incarcera- showed that serving time in prison was associ-
tion is associated with reduced earnings and ated with a 40 percent reduction in earnings and
employment.4 with reduced job tenure, reduced hourly wages,
We analyzed panel data from the National and higher unemployment.
Longitudinal Survey of Youth (NLSY), one The negative effects of incarceration, even
of the few surveys that follows respondents among men with very poor economic oppor-
over a long period of time and that interviews tunities to begin with, are related to the strong
436 • Part V: Poverty and the Underclass

Figure 51.3. Twenty-Year Earnings Mobility Among Men in the Bottom Quintile of Earnings Distribu-
tion in 1986, National Longitudinal Survey of Youth (NLSY) Men

AFQT stands for Armed Forces Qualifying Test.

Source: Becky Pettit, Bryan Sykes, and Bruce Western, “Technical Report on Revised Population Estimates and NLSY79
Analysis Tables for the Pew Public Safety and Mobility Project” (Harvard University, 2009).

negative perceptions employers have of job counterparts. The stigma of a criminal convic-
seekers with criminal records. Devah Pager’s ex- tion may also repel employers who prefer job ap-
perimental research has studied these employer plicants with clean records. Pager’s audit study
perceptions by sending pairs of fake job seekers offers clear evidence for the negative effects of
to apply for real jobs.5 In each pair, one of the criminal stigma. Employers, fearing legal liability
job applicants was randomly assigned a resume or even just unreliability, are extremely reluctant
indicating a criminal record (a parole officer is to hire workers with criminal convictions.
listed as a reference), and the “criminal” appli- A simple picture of the poor economic op-
cant was instructed to check the box on the job portunities of the formerly incarcerated is given
application indicating he had a criminal record. by the earnings mobility of men going to prison
A criminal record was found to reduce callbacks compared to other disadvantaged groups. The
from prospective employers by around 50 per- NLSY data can be used to study earnings mo-
cent, an effect that was larger for African Amer- bility over several decades. We calculated the
icans than for whites. chances that a poor man in the lowest fifth of
Incarceration may reduce economic oppor- the earnings distribution in 1986 would move
tunities in several ways. The conditions of im- up and out of the lowest fifth by 2006. Among
prisonment may promote habits and behaviors low-income men who are not incarcerated,
that are poorly suited to the routines of regu- nearly two-thirds are upwardly mobile by 2006
lar work. Time in prison means time out of the (Figure 51.3). Another group in the NLSY has
labor force, depleting the work experience of the very low levels of cognitive ability, scoring in the
incarcerated compared to their nonincarcerated bottom quintile of the Armed Forces Qualifying
 Western and Pettit—Incarceration and Social Inequality • 437

Test, the standardized test used for military ser- first of all they don’t like to bring to reality
vice. Among low-income men with low scores that you’re in prison; they don’t wanna think
on the test, only 41 percent are upwardly mo- about that . . . Or some of ’em just don’t care.
bile. Upward mobility is even less common So the male’s kinda like wiped out of there, so
among low-income high school dropouts. Still, that puts all the burden on the woman.7
we observe the least mobility of all among men
who were incarcerated at some point between Partly because of the burdens of incarceration
1986 and 2006. For these men, only one in four on women who are left to raise families in free
rises out of the bottom quintile of the earnings society, incarceration is strongly associated with
distribution. divorce and separation. In addition to the forced
Intergenerational Inequality. Finally, the ef- separation of incarceration, the post-release ef-
fects of the prison boom extend also to the fam- fects on economic opportunities leave formerly
ilies of those who are incarcerated. Through the incarcerated parents less equipped to provide fi-
prism of research on poverty, scholars find that nancially for their children. New research also
the family life of the disadvantaged has become shows that the children of incarcerated parents,
dramatically more complex and unstable over particularly the boys, are at greater risk of devel-
the last few decades. Divorce and nonmarital opmental delays and behavioral problems.8
births have contributed significantly to rising Against this evidence for the negative effects
rates of single parenthood, and these changes of incarceration, we should weigh the gains to
in American family structure are concentrated public safety obtained by separating violent or
among low-income mothers. As a consequence, otherwise antisocial men from their children and
poor children regularly grow up, at least for partners. Domestic violence is much more com-
a time, with a single mother and, at different mon among the formerly incarcerated compared
times, with a variety of adult males in their to other disadvantaged men. Survey data indi-
households. cate that formerly incarcerated men are about
High rates of parental incarceration likely four times more likely to assault their domestic
add to the instability of family life among poor partners than men who have never been incar-
children. Over half of all prisoners have children cerated. Though the relative risk is very high,
under the age of eighteen, and about 45 percent around 90 percent of the partners of formerly
of those parents were living with their children incarcerated report no domestic violence at all.
at the time they were sent to prison. About The scale of the effects of parental incarcer-
two-thirds of prisoners stay in regular contact ation on children can be revealed simply by sta-
with their children either by phone, mail, or vis- tistics showing the number of children with a
itation.6 Ethnographer Megan Comfort paints a parent in prison or jail. Among white children
vivid picture of the effects of men’s incarceration in 1980, only 0.4 of 1 percent had an incarcer-
on the women and families in their lives. She ated parent; by 2008 this figure had increased
quotes a prisoner at San Quentin State Prison to 1.75 percent. Rates of parental incarceration
in California: are roughly double among Latino children, with
3.5 percent of children having a parent locked
Nine times out of ten it’s the woman [main- up by 2008. Among African American children,
taining contact with prisoners]. Why? Be- 1.2 million, or about 11 percent, had a parent
cause your homeboys, or your friends, if incarcerated by 2008 (Figure 51.4).
you’re in that lifestyle, most the time they’re
gonna be sittin’ right next to your ass in The spectacular growth in the American penal
prison . . . The males, they don’t really par- system over the last three decades was concen-
ticipate like a lot of females in the lives of the trated in a small segment of the population,
incarcerated . . . They don’t deal with it, like among young minority men with very low levels
438 • Part V: Poverty and the Underclass

Figure 51.4. Number of Children under Eighteen with a Parent in Prison or Jail, 1980 to 2008

Source: Becky Pettit, Bryan Sykes, and Bruce Western, “Technical Report on Revised Population Estimates and NLSY79
Analysis Tables for the Pew Public Safety and Mobility Project” (Harvard University, 2009).

of education. By the early 2000s, prison time life, for all but those whose incarceration rates are
was a common life event for this group, and highest, prisons are exotic institutions unknown
today more than two-thirds of African American to the social mainstream. Our national data sys-
male dropouts are expected to serve time in state tems, and the social facts they produce, are struc-
or federal prison. These demographic contours tured around normative domestic and economic
of mass imprisonment have created a new class life, systematically excluding prison inmates.
of social outsiders whose relationship to the state Thus we define carceral inequalities as invis-
and society is wholly different from the rest of ible, cumulative, and intergenerational. Because
the population. they are so deeply concentrated in a small disad-
Social marginality is deepened by the inequal- vantaged fraction of the population, the social
ities produced by incarceration. Workers with and economic effects of incarceration create a
prison records experience significant declines discrete social group whose collective experience
in earnings and employment. Parents in prison is so distinctive yet unknown that their disad-
are likely to divorce or separate, and through the vantage remains largely beyond the apprehen-
contagious effects of the institution, their chil- sion of public policy or public conversation.
dren are in some degree “prisonized,” exposed to The redrawing of American social inequality by
the routines of prison life through visitation and mass incarceration amounts to a contraction of
the parole supervision of their parents. Yet much citizenship—a contraction of that population
of this reality remains hidden from view. In social that enjoys, in T. H. Marshall’s words, “full
 Western and Pettit—Incarceration and Social Inequality • 439

membership in society.”9 Inequality of this kind young men in high-incarceration communities,

threatens to be self-sustaining. Socioeconomic a prison record now carries little stigma; incen-
disadvantage, crime, and incarceration in the tives to commit to the labor market and family
current generation undermine the stability of life have been seriously weakened.
family life and material support for children. As To say that prison reduces crime (perhaps
adults, these children will be at greater risk of di- only in the short run) is a spectacularly modest
minished life chances and criminal involvement, claim for a system that now costs $70 billion an-
and at greater risk of incarceration as a result. nually. Claims to the crime-reducing effects of
Skeptics will respond that these are false prison, by themselves, provide little guidance for
issues of social justice: the prison boom sub- policy because other approaches may be cheaper.
stantially reduced crime, and criminals should Measures to reduce school dropout, increase
forfeit their societal membership in any case. human capital, and generally increase employ-
The crime-reducing effects of incarceration are ment among young men seem especially prom-
hotly debated, however. Empirical estimates of ising alternatives. Results for programs for very
the effects of incarceration on crime vary widely, young children are particularly striking. Evalu-
and often they turn on assumptions that are ations of early childhood educational programs
difficult to test directly. Researchers have fo- show some of their largest benefits decades later
cused on the sharp decline in U.S. crime rates in reduced delinquency and crime.11 For adult
through the 1990s, studying the influence of ris- men now coming out of prison, new evaluations
ing prison populations. Conservative estimates show that jobs programs reduce recidivism and
attribute about one-tenth of the 1990s crime increase employment and earnings.12 The demo-
decline to the growth in imprisonment rates.10 graphic concentration of incarceration accom-
Though the precise impact of incarceration on panies spatial concentration. If some portion
crime is uncertain, there is broad agreement that of that $70 billion in correctional expenditures
additional imprisonment at high rates of incar- were spent on improving skills and reducing un-
ceration does little to reduce crime. The possibil- employment in poor neighborhoods, a sustain-
ity of improved public safety through increased able and socially integrative public safety may be
incarceration is by now exhausted. produced.
Studies of the effects of incarceration on Much of the political debate about crime pol-
crime also focus only on the short term. Indeed, icy ignores the contemporary scale of criminal
because of the negative effects of incarceration punishment, its unequal distribution, and its
on economic opportunities and family life, in- negative social and economic effects. Our analy-
carceration contributes to crime in the long run sis of the penal system as an institution of social
by adding to idleness and family breakdown stratification, rather than crime control, high-
among released prisoners. Scale matters, too. If lights all these neglected outcomes and leaves
the negative effects of incarceration were scat- us pessimistic that widespread incarceration can
tered among a small number of serious criminal sustainably reduce crime. The current system is
offenders, these effects may well be overwhelmed expensive, and it exacerbates the social problems
by reduction in crime through incapacitation. it is charged with controlling. Our perspective,
Today, however, clear majorities of the young focused on the social and economic inequali-
men in poor communities are going to prison ties of American life, suggests that social policy
and returning home less employable and more improving opportunity and employment, for
detached from their families. In this situation, young men in particular, holds special promise
the institutions charged with public safety have as an instrument for public safety.
become vitally implicated in the unemployment Our perspective on inequality points to a
and the fragile family structure characteristic of broader view of public safety that is not pro-
high-crime communities. For poorly educated duced by punishment alone. Robust public
440 • Part V: Poverty and the Underclass

safety grows when people have order and pre- Young Workers,” in Do Prisons Make Us Safer? ed.
dictability in their daily lives. Crime is just one Steven Raphael and Michael A. Stoll (New York:
danger, joining unemployment, poor health, and Russell Sage Foundation, 2009).
family instability along a spectrum of threats to 5. Devah Pager, Marked: Race, Crime, and Find-
ing Work in an Era of Mass Incarceration (Chicago:
an orderly life. Public safety is built as much on
University of Chicago Press, 2007).
the everyday routines of work and family as it 6. Christopher Murnola, “Incarcerated Parents
is on police and prisons. Any retrenchment of and Their Children” (Washington, D.C.: Bureau of
the penal system therefore must recognize how Justice Statistics, 2000).
deeply the prison boom is embedded in the 7. Megan Comfort, “In the Tube at San Quentin:
structure of American social inequality. Amelio- The ‘Secondary Prisonization’ of Women Visiting
rating these inequalities will be necessary to set Inmates,” Journal of Contemporary Ethnography 32
us on a path away from mass incarceration and (1) (2003): 82; emphasis original.
toward a robust, socially integrative public safety. 8. Christopher Wildeman, “Paternal Incarcera-
tion and Children’s Physically Aggressive Behaviors:
Evidence from the Fragile Families and Child Well-
NOTES
being Study.” Working paper 2008-02-FF (Fragile
1. We gratefully acknowledge Bryan Sykes, Deir-
Families and Child Wellbeing, 2008).
dre Bloome, and Chris Muller, who helped conduct
9. T. H. Marshall, Citizenship and Social Class
the research reported in this paper. This research was
(Concord. Mass.: Pluto Press, 1992).
supported in part by a gift from The Elfenworks
10. Western, Punishment and Inequality in Amer-
Foundation.
ica, chap. 7.
2. In her essay for this issue, Candace Kruttschnitt
11. Pedro Carneiro and James J. Heckman,
shows that women’s incarceration has pronounced
“Human Capital Policy,” in James J. Heckman and
effects by separating mothers from their children.
Alan B. Krueger, Inequality in America: What Role
The continued growth of women’s incarceration
for Human Capital Policies? (Cambridge, Mass.: MIT
rates threatens to have large effects on family life.
Press, 2005).
3. Bruce Western, Punishment and Inequality in
12. Cindy Redcross, Dan Bloom, and Gilda
America (New York: Russell Sage Foundation, 2006);
Azurdia, “Transitional Jobs for Ex-prisoners: Imple-
Caroline Wolf Harlow, Education and Correctional
mentation, Two-Year Impacts, and Costs of the Cen-
Populations (Washington, D.C.: Bureau of Justice
ter for Employment Opportunities (CEO) Prisoner
Statistics, 2003).
Reentry Program” (MDRC, 2009).
4. Marry J. Holzer, “Collateral Costs: Effects of
Incarceration on Employment and Earnings Among
 PART VI

Who Gets Ahead?

Class Mobility
52 A Refined Model of Occupational Mobility 443
53 Trends in Class Mobility: The Post-War European Experience 453
54 Social Mobility in Europe 464
55 It’s a Decent Bet That Our Children Will Be Professors Too 480

Income Mobility
56 Intergenerational Income Mobility 496
57 Advantage in Comparative Perspective 501

Classic Models of Status Attainment

58 The Process of Stratification 506
59 Inequality: A Reassessment of the Effect of Family and Schooling
in America 517

Education and Reproduction

60 Explaining Educational Differentials 524
61 The Widening Academic Achievement Gap Between the
Rich and the Poor 536
62 Nonpersistent Inequality in Educational Attainment 551
63 Determined to Succeed 562
64 Towards a Theory of Inequality in Higher Education 569
65 Does College Still Have Equalizing Effects? 578
66 Who Benefits Most from College? 587

Social Psychological Models

67 The Educational and Early Occupational Attainment Process 596
68 Ain’t No Makin’ It: Leveled Aspirations in a
Low-Income Neighborhood 608
69 A New Social Psychological Model of Educational Attainment 621
 Labor Markets
70 The Dual Labor Market: Theory and Implications 629
71 An Outline of a Theory of the Matching of Persons to Jobs 632
72 The Rise of Precarious Work 640
73 Little Labor: How Union Decline Is Changing
the American Landscape 645

Social Capital, Networks, and Attainment

74 The Strength of Weak Ties 653
75 Social Networks and Status Attainment 657
76 Structural Holes 659
77 Networks, Race, and Hiring 663
78 When Do Social Networks Increase Inequality? 671
 Featherman and Hauser—A Rekned Model of Occupational Mobility • 443

Class Mobility

52 • David L. Featherman and Robert M. Hauser

A Rekned Model of Occupational Mobility

In this chapter we describe and apply a log-linear (1977) have critically reviewed these and related
model of the mobility table. . . . The model per- works in light of British mobility data collected
mits us to locate groups or clusters of cells in the in 1972. It would be easy to identify our present
classification that share similar chances of mo- analytic interests with those of the class theorists,
bility or immobility, freed of the confounding but we think such an inference unwarranted.
influences of the relative numbers of men in each Although we are attempting a description of
origin or destination category and of changes in the mobility regime that is free of the distribu-
those relative numbers between origin and desti- tions of occupational origins and destinations,
nation distributions. we believe with Goldthorpe and Llewellyn
By modeling the mobility table in this way that the class theorists are attempting to inter-
we obtain new insights into the process of mo- pret what [might be] termed the gross flows of
bility, changes in that process, and the interac- manpower. For the American case we have al-
tions of the mobility process with changes in ready described those flows [see Featherman and
the occupational structure within one mobility Hauser (1978: Chapter 3)], and our interest now
classification or between two or more mobility centers on the net or underlying patterns of asso-
classifications. For example, we take a fresh look ciation in the mobility table.
at the differing tendencies toward immobility in We have approached the mobility table with-
the several occupational strata, at the existence of out strong theoretical presuppositions about
“class” boundaries limiting certain types of mo- affinities among occupational strata. Like Blau
bility, at differences in upward and downward and Duncan, we have worked inductively, but
exchanges between occupational strata, and at our more refined analytic tools have led to sub-
differences among strata in the dispersion of stantively different conclusions than theirs about
recruitment and supply. In these purposes our the major features of the mobility process in the
analysis parallels Blau and Duncan’s treatment of United States.
manpower flows (1967: Chap. 2; also, see Blau, Some readers may find the following discus-
1965). sion excessively technical, but we have tried to
Several sociologists have recently drawn at- minimize the presentation of methodological
tention to relationships between occupational detail. We have tried to avoid describing the
mobility and class formation, for example, Gid- methods by which empirical specifications of the
dens (1973), Parkin (1971), and Westergaard mobility table may be explored, although we be-
and Resler (1975). Goldthorpe and Llewellyn lieve these are interesting in their own right. We

David L. Featherman and Robert M. Hauser. “A Refined Model of Occupational Mobility,” in Opportunity and
Change, pp. 139–141, 145–148, 150–153, 155–157, 179–180, 549–551, 553–562, 564. Copyright © 1978
by Academic Press. Used with permission from Elsevier.
444 • Part VI: Who Gets Ahead?

have focused on the rationale and interpretation models and methods are less appealing. Good-
of our model, including comparisons with other man (1965, 1969a) proposed that the analyst ig-
ways of looking at the mobility table that seem nore or “blank out” those cells where interaction
likely to elucidate the properties of the model. was greatest (where frequencies were thought to
be especially dense or especially sparse) and at-
tempt to fit a modified model of statistical inde-
Mobility Models pendence, termed “quasi-independence,” to the
The record of sociological mobility studies is remaining frequencies in the table. In the case
paralleled by a history of statistical analysis in where only diagonal cells were blanked out in a
which occupational mobility has often served as mobility table, Goodman called the model one
stimulus, object, or illustration of statistical ideas of “quasi-perfect mobility,” after the term “per-
(for example, see Pearson, 1904; Chessa, 1911; fect mobility,” which had earlier been applied to
Rogoff, 1953; Glass, 1954; Goodman, 1961, the model of statistical independence in a mobil-
1968, 1969a, 1972c; Tyree, 1973; White, 1963, ity table. For an early application of this model
1970a; Singer and Spilerman, 1976). Indeed, to a large (17 × 17) table see Blau and Duncan
it is consistent with the historical pattern that (1967: 64–67). Goodman (1965, 1968, 1969a)
sociologists were introduced to the method of noted that quasi-independence might hold over
path analysis primarily by way of its successful all cells in a table whose entries were not ig-
application in studies of occupational mobility nored, or it might hold within, but not between
(Duncan and Hodge, 1963; Blau and Duncan, certain subsets of cells whose entries were not
1967). Devices for the statistical analysis of mo- ignored. . . .
bility data range from simple descriptive mea- Models of quasi-independence have provided
sures to complex analytic schemes. We make no important insights into the structure of mobility
systematic effort to review these measures and tables. Aside from Goodman’s expository papers,
models, for there are several recent and compre- they have been applied in cross-national, inter-
hensive reviews (Boudon, 1973; Pullum, 1975; urban, and cross-temporal analyses (Iutaka et al.,
Bibby, 1975). We focus almost exclusively on 1975; Featherman et al., 1975; Pullum, 1975;
multiplicative (loglinear) representations of the Hauser et al., 1975; Ramsøy, 1977; Goldthorpe
occupational mobility table. In so doing we do et al., 1978). Goodman (1969a) also has shown
not intend to suggest that other methods and how related ideas may be supplied to test any
approaches are inferior, but to exploit features specific hypothesis about the pattern of associa-
of the loglinear model that seem interesting and tion in a mobility table.
fruitful. . . . At the same time the application of
In a series of papers, Goodman (1963, 1965, quasi-independence models in mobility analy-
1968, 1969a, 1969b, 1972c) developed and ex- sis has been less than satisfying in some ways.
posited methods for the analysis of contingency Even where large numbers of cells are blocked,
tables (including mobility tables) in which the quasi-independence models do not fit large tables
significant interactions were localized in speci- very well (Pullum, 1975; Hauser et al., 1975).
fied cells or sets of cells in the table (also, see That is, when mobility data are not highly aggre-
Pullum, 1975). For example, in the case of gated, it appears that association is not limited to
highly aggregated (3 × 3 or 5 × 5) mobility tables the small number of cells on or near the main di-
Goodman showed that most of the interaction agonal. The larger the number of entries blocked
pertained to cells on or near the main diagonal (or fitted exactly) before a good fit is obtained,
(when the occupation categories of origin and the less substantively appealing is the model
destination were listed in order of increasing sta- of quasi-independence. Moreover, by treating
tus). White (1963, 1970b) has made essentially departures from quasi-independence in the
the same suggestion, but some aspects of his blocked or ignored cells as parameters or indices
 Featherman and Hauser—A Rekned Model of Occupational Mobility • 445

of mobility and departures in the unblocked Hauser (1978) has applied this approach in an
cells as error, the quasi-independence model at- analysis of the classic British mobility table, and
taches too much theoretical importance to occu- Baron (1977) has used it in a reanalysis of Ro-
pational inheritance (Hope, 1976). Of course, goff’s (1953) Indianapolis data.
occupational inheritance is always defined by
reference to a given classification of occupations,
and the problem is exacerbated by the fact that A Refined Multiplicative Model of
the model of quasi-independence fits best when the Mobility Table 1
the mobility table is based on broad occupation Let xij be the observed frequency in the ijth cell
groups. The model is of greatest validity in the of the classification of men by their own occupa-
measurement of immobility in classifications tions (j = 1, . . . ,J) and their own occupations or
where the concept of occupational inheritance fathers’ occupations at an earlier time (i = 1, . . .
becomes vague. ,I). In the context of mobility analysis the same
The focus on fit on or near the main diago- categories will appear in rows and columns, and
nal follows a traditional sociological interest in the table will be square with I = J. For k = 1, . . .
occupational inheritance, but it also draws our ,K, let Hk be a mutually exclusive and exhaustive
attention away from other aspects of association partition of the pairs (i, j) in which
in the table. For example, one might hypothesize
that certain types of mobility are as prevalent as E[xij] = mij = ĮȕiȖjįij, (1)
other types of mobility or immobility. More gen-
erally, one might wish to construct a parametric where įij = įk for (i, j) ‫ א‬Hk, subject to the nor-
model of mobility and immobility for the full malization ∏iȕi = ∏iȖj = ∏i∏jįij = 1. The normal-
table that would recognize the somewhat arbi- ization of parameters is a matter of convenience,
trary character of occupational inheritance and and we choose the value of one so that it will
the possible gradations of association throughout hold. Note that, unlike the usual set-up, the in-
the table. teraction effects are not constrained within rows
Goodman’s (1972c) general multipli- or columns although the marginal frequencies
cative model of mobility tables and other are fixed. The model says the expected frequen-
cross-classifications substantially advanced the cies are a product of an overall effect (Į), a row
sophistication and precision of mobility analysis. effect (ȕi), a column effect (Ȗj), and an interac-
For example, Goodman proposed and applied tion effect (įij). The row and column parameters
to the classic British and Danish mobility data a represent conditions of occupational supply and
number of alternative specifications, all but two demand; they reflect demographic replacement
of which—the simple independence model and processes and past and present technologies and
that of quasi-perfect mobility—assumed ordi- economic conditions. The cells (i, j) are assigned
nality in the occupational categories. The models to K mutually exclusive and exhaustive levels,
incorporated combinations of parameters for up- and each of those levels shares a common inter-
ward and downward mobility, for the number of action parameter įk. Thus, aside from total, row,
boundaries crossed, and for barriers to crossing and column effects, each expected frequency is
particular categoric boundaries. Many of these determined by only one parameter, which re-
models—as well as problems in comparing their flects the level of mobility or immobility in that
goodness of fit—are reviewed by Bishop et al. cell relative to that in other cells in the table.
(1975: Chaps. 5, 8, 9), and some of the same The interaction parameters of the model cor-
models are discussed by Haberman (1974: Chap. respond directly to our notions of variations in
6). Applying Goodman’s (1972c) general model the density of observations (White, 1963:26).
we take a slightly different approach in devel- Unlike several models fitted by Goodman
oping models of the mobility table. Elsewhere, (1972c), this model does not assume ordinal
446 • Part VI: Who Gets Ahead?

measurement of occupations. Of course, the as- group. Mobility from lower to upper nonman-
sumption of ordinality may help us interpret re- ual occupations (Level 3) is more likely than the
sults, or our findings may be used to explore the opposite movement, and the latter is as likely as
metric properties of our occupational classifica- stability in the lower nonmanual category (Level
tion. For the model to be informative, the distri- 4). Moreover, the densities of immobility in the
bution of levels across the cells of the table must lower nonmanual category and of downward
form a meaningful pattern, and one in which mobility to it are identical to those in the sec-
the parameters are identified (Mason et al., ond zone of relatively high density, which occurs
1973; Haberman, 1974: 217). Furthermore, the at the lower end of the occupational hierarchy.
number of levels (K ) should be substantially less The second zone includes movements from the
than the number of cells in the table. These latter farm to the lower manual group and back as well
properties are partly matters of substantive and as immobility in the lower manual group. Last,
statistical interpretation and judgment, rather there is a broad zone of relatively low density
than characteristics of the model or of the data. (Level 5) that includes immobility in the upper
We have found it difficult to interpret models manual category, upward and downward mobil-
where the number of levels is much greater than ity within the manual stratum, mobility between
the number of categories recognized in the occu- upper manual and farm groups, and all move-
pational classification. . . . ments between nonmanual and either manual
or farm groups. The design says that an upper
manual worker’s son is equally likely to be im-
Mobility to First Jobs: An Illustration mobile or to move to the bottom or top of the
Table 52.1 gives frequencies in a classification of occupational distribution; obversely, it says that
son’s first, full-time civilian occupation by father’s an upper manual worker is equally likely to have
(or other family head’s) occupation at the son’s been recruited from any location in the occupa-
sixteenth birthday among American men who tional hierarchy, including his own. Also, it is
were ages 20–64 in 1973 and were not currently worth noting that four of the five density levels
enrolled in school.2 Table 52.2 gives the design recognized in the model occur along the main
matrix of a model for the data of Table 52.1. diagonal, and two of these (Levels 4 and 5) are
Each numerical entry in the body of the table assigned both to diagonal and off-diagonal cells.
gives the level of Hk to which the corresponding With a single exception the design is sym-
entry in the frequency table was assigned. For- metric. That is, the upward and downward
mally, the entries are merely labels, but, for con- flows between occupations are assigned to the
venience in interpretation, the numerical values same density levels, except mobility from lower
are inverse to the estimated density of mobility to upper nonmanual strata (Level 3) exceeds that
or immobility in the cells to which they refer. from upper to lower nonmanual strata (Level 4).
On this understanding the design says that, This asymmetry in the design is striking because
aside from conditions of supply and demand, it suggests the power of upper white-collar fam-
immobility is highest in farm occupations (Level ilies to block at least one type of status loss and
1) and next highest in the upper nonmanual cat- because it is the only asymmetry in the design.
egory (Level 2). If we take the occupation groups For example, Blau and Duncan (1967: 58–67)
as ranked from high to low in the order listed, we suggest that there are semipermeable class
may say that there are zones of high and almost boundaries separating white-collar, blue-collar,
uniform density bordering the peaks at either and farm occupations, which permit upward
end of the status distribution. There is one zone mobility but inhibit downward mobility. The
of high density that includes upward or down- only asymmetry in the present design occurs
ward movements between the two nonmanual within one of the broad classes delineated by
groups and immobility in the lower nonmanual Blau and Duncan.
 Featherman and Hauser—A Rekned Model of Occupational Mobility • 447

Table 52.1. Frequencies in a Classification of Mobility from Father’s (or Other Family Head’s)
Occupation to Son’s First Full-Time Civilian Occupation: U.S. Men Aged 20–64 in March 1973
p p g
Son’s occupation
Upper Lower Upper Lower
Father’s occupation nonmanual nonmanual manual manual Farm Total
Upper nonmanual 1414 521 302 643 40 2920
Lower nonmanual 724 524 254 703 48 2253
Upper manual 798 648 856 1676 108 4086
Lower manual 756 914 771 3325 237 6003
Farm 409 357 441 1611 1832 4650
Total 4101 2964 2624 7958 2265 19,912
Note: Frequencies are based on observations weighted to estimate population counts and compensate for departures
of the sampling design from simple random sampling (see Featherman and Hauser [1978: Appendix B]). Broad
occupation groups are upper nonmanual: professional and kindred workers, managers and officials, and non-retail sales
workers; lower nonmanual: proprietors, clerical and kindred workers, and retail sales workers; upper manual: craftsmen,
foremen and kindred workers; lower manual: service workers, operatives and kindred workers, and laborers, except
farm; farm: farmers and farm managers, farm laborers and foremen.

Overall, the design resembles a river valley in as Ȥ2 with 16 df. With the model of Table 52.2
which two broad plains are joined by a narrow as null hypothesis we obtain G 2 = 66.5 with 12
strip of land between two great peaks. The con- df, since we lose 4 df in creating the five cate-
tours of the peaks differ in that the one forming gories of H. Clearly the model does not fit, if
one side of the valley is both taller and more we take the probability associated with the test
nearly symmetric than that forming the other statistic as our only guide. On the other hand
side. This representation appears in Figure 52.1. the model does account for 98.9% of the asso-
In some respects, this design matrix parallels ciation in the data, that is, of the value of G 2
Levine’s (1967: Chap. 4) description of the sur- under independence. Given the extraordinarily
face of the British mobility table as a saddle (also large sample size we might expect small depar-
see Levine, 1972). However, our interpretation tures from frequencies predicted by the model
is more extreme, since the density reaches an ab- to be statistically significant. . . .
solute minimum in the center of the table, not Table 52.2. Asymmetric 5-Level Model of
merely a minimum among the diagonal cells. In Mobility from Father’s Occupation to First Full-
this way our model for the American 5 × 5 table Time Civilian Occupation
p
is closer to Goodman’s (1969a: 38, 1969b: 846)
Son’s occupation
conclusion that a British 5 × 5 table shows “sta-
tus disinheritance” in the middle category. We Father’s occupation (1) (2) (3) (4) (5)
show elsewhere (Featherman and Hauser, 1978) 1. Upper nonmanual 2 4 5 5 5
that Levine’s interpretation of the British data is 2. Lower nonmanual 3 4 5 5 5
3. Upper manual 5 5 5 5 5
based on a confounding of marginal effects and
4. Lower manual 5 5 5 4 4
interactions which parallels that entailed in the 5. Farm 5 5 5 4 1
use of mobility ratios, even though Levine did
Note: Broad occupation groups are upper nonmanual:
not use mobility ratios. professional and kindred workers, managers and
The model of Table 52.2 provides less than officials, and non-retail sales workers; lower nonmanual:
a complete description of the mobility data in proprietors, clerical and kindred workers, and retail sales
workers; upper manual: craftsmen, foremen and kin-
Table 52.1. Under the model of statistical inde-
dred workers; lower manual: service workers, operatives
pendence we obtain a likelihood-ratio statistic, and kindred workers, and laborers, except farm; farm:
G 2 = 6167.7, which is asymptotically distributed farmers and farm managers, farm laborers and foremen.
448 • Part VI: Who Gets Ahead?

Figure 52.1. Volume of Mobility from Father’s Occupation to First Full-Time Civilian
Occupation: U.S. Men Aged 20–64 in March 1973

Upper
Farm Nonmanual
Lower Lower
Nonmanual
FA Manual
TH Upper
ER Upper Manual I ON
’S AT
OC Manual
Lower Lower C UP
CU C
PA Nonmanual Upper
TI
Manual ’ SO
ON Nonmanual
Farm
S ON

The base is a unit square, and the total volume under the surface is one. Length and breadth can be read
as probabilities, and height is proportionate to probability. The vertical scale has been compressed by a
factor of 10.

The measures of fit we have examined have the model of Table 52.2 for mobility in the 1973
told us nothing about the several parameters of data from father’s (or head’s) occupation at son’s
the model. That is, we have not shown that our sixteenth birthday to son’s first full-time civil-
suggested interpretation of the design matrix ian occupation. The parameters are expressed in
(Table 52.2) is substantively appealing, or even additive form, that is, they are effects on logs
that the design correctly sorts the cells of the mo- of frequencies under the model of Eq. (1). The
bility table into zones of high and low density. row and column parameters clearly show an
Certainly, we want to look at the way in which intergenerational shift out of farming and into
the model fits and interprets the data as well as white-collar or lower blue-collar occupations.
at deviations from fitted values. Of course these parameters reflect a number of
The upper panel of Table 52.3 shows the row, factors, including temporal shifts in the distri-
column, and level parameters estimated under bution of the labor force across occupations,
 Featherman and Hauser—A Rekned Model of Occupational Mobility • 449

Table 52.3. Parameters and Residuals (in Additive Form) from Main, Row, and Column Effects in the
Model of Table 52.2: Mobility from Father’s (or Other Family Head’s) Occupation to Son’s First Full-
Time Civilian Occupation, U.S. Men Aged 20–64 in March 1973
p , g 7
A. Additive parameters
Category of row, column, or level
Design factor (1) (2) (3) (4) (5)
Rows (father’s occupation) –.466 –.451 .495 .570 –.148
Columns (son’s occupation) .209 .190 .240 1.020 –1.660
Levels (density) 3.044 1.234 .549 .243 –.356
Grand mean = 6.277

B. Level parameter plus residual (log R*ij )

Son’s occupation
Father’s occupation (1) (2) (3) (4) (5)
1. Upper nonmanual 1.23 .25 –.34 –.36 –.45
2. Lower nonmanual .55 .25 –.53 –.29 –.29
3. Upper manual –.30 –.49 –.26 –.37 –.43
4. Lower manual –.43 –.22 –.44 .24 .28
5. Farm –.32 –.44 –.28 .24 3.04
Note: See text for explanation.

differential fertility, and life cycle differences in In any event the parameters do show a sharp
occupational positions. The level parameters density gradient across the levels of the design.
show very large differences in mobility and im- The smallest difference, between Levels 3 and 4,
mobility across the several cells of the classifi- indicates a relative density e .549–.243 = e .306 = 1.36
cation, and these differences closely follow our times as great at Level 3 as at Level 4. The het-
interpretation of the design matrix. Differences erogeneity of Level 5 is indicated by the fact that
between level parameters may readily be inter- the difference in density between Levels 3 and 4
preted as differences in logs of frequencies, net is about as large as the range of residuals within
of row and column effects. For example, the esti- Level 5. Immobility in farm occupations and in
mates say that immobility in farm occupations is upper nonmanual occupations is quite distinct
3.40 = 3.044 – (–.356) greater (in the metric of from densities at other levels, but also immobil-
logged frequencies) than the estimated mobility ity in the farm occupations is e 3.044–1.234 = e1.810 =
or immobility in cells assigned to Level 5 in the 6.11 times as great as in the upper nonmanual
design matrix. In multiplicative terms, immo- occupations.
bility in farm occupations is e3.40 = 29.96 times We can write the sample counterpart of Eq.
greater than mobility or immobility at Level 5. (1) as
It would be incorrect to attach too much im-
portance to the signs of the level parameters as ˆ
m̂ij = ˆ ˆj ˆ
i   ij. (2)
reported in Table 52.3, for they simply reflect
our normalization rule that level parameters sum Recalling that
to zero (in the log-frequency metric) across the
cells of the table. For example, while the param- eij = xij / m̂ij, (3)
eters for Levels 4 and 5 each reflect relatively low
densities, it is not clear that either parameter in- we substitute Eq. (2) into (3) and rearrange
dicates “status disinheritance” in the diagonal terms to obtain
cells to which it pertains (compare Goodman,
1969a, 1969b). xij = ˆ ˆj ˆ
i   ij eij . (4)
450 • Part VI: Who Gets Ahead?

We divide both sides of Eq. (4) by the first three occupational career. In doing so, of course, we
terms on the right-hand side to obtain do some injustice to details reported in the pre-
ceding analysis.
xij First, there is great immobility at the top and
R*ij = =ˆ
 ij eij . (5) at the bottom of the occupational hierarchy, here
ˆ
ˆ i 
ˆj represented by upper nonmanual and by farm
occupations, respectively. This immobility is far
We shall call R*ij the new mobility ratio, or, sim- more extreme than has heretofore been supposed
ply, the mobility ratio. In the case of diagonal by most students of the mobility process; it may
cells R ij* is equivalent to the new immobility ratio even be consistent with the beliefs of the more
proposed by Goodman (1969a,b, 1972c; also, extreme critics of rigidity in the American class
see Pullum, 1975: 7–8), but we suggest the ratio structure.
be computed for all cells of the table as an aid Second, surrounding the extremes of the oc-
both to substantive interpretation and to the cupational hierarchy at both top and bottom are
evaluation of model design. transitional zones, within which there are rela-
The lower panel of Table 52.3 gives logs of the tively homogeneous chances of immobility and
new mobility ratios for the model of Table 52.2 of exchange with adjacent extreme strata.
fitted to the classification of mobility to first jobs. Third, taken as aggregates the extreme and
While the entries in this panel depend on our transitional zones of the occupational hierar-
specification of the model, that specification need chy are relatively closed both to upward and to
not rigidly govern our interpretation of the rel- downward movements. That is, there are sharp
ative densities. Obviously, the pattern of relative breaks between the density of observations
densities does conform substantially to our earlier within the extreme and transitional zones and
description of the design. The fit is good enough the density of mobility beyond those zones. In
so there is no overlap in densities across levels this sense (but not in others) we may say that the
recognized in the design, and all of the negative data suggest the existence of barriers to move-
entries are neatly segregated in Level 5 of the de- ment across class boundaries.
sign. If immobility among skilled workers—in Fourth, once the boundaries of the transi-
cell (3, 3)—is high relative to mobility in other tional zones have been crossed, no social dis-
cells at Level 5, it is still clear that the immobil- tance gradient seems to underlie variations in
ity in that category is substantially less than the long-distance mobility chances. These are sur-
immobility in any other occupation group. . . . prisingly uniform, and observed variations in
them show no consistent pattern.
Fifth, if immobility is very great at the ex-
Mobility Chances: A New tremes of the occupational hierarchy, it is al-
Perspective most nonexistent in the middle of the hierarchy.
As an alternative to the Blau-Duncan interpre- Contrary to widespread belief, men of upper
tation, we think our multiplicative models yield blue-collar origin are about as likely to end up
a cogent and parsimonious description of occu- anywhere higher or lower in the occupational hi-
pational mobility among American men. Unlike erarchy as in their stratum of origin. Obversely,
its precursor, our description does not reflect the upper blue-collar workers are about as likely to
shape of occupational distributions of origin or have originated anywhere higher or lower in the
destination, but only the underlying patterns occupational hierarchy as in their stratum of
of immobility and exchange between occupa- destination. Those who would find their beliefs
tional strata. It may be useful here to review the about “class” rigidity confirmed by our estimates
major features of this description that appear in of immobility at the extremes of the occupa-
mobility between generations and within the tional hierarchy must reconcile these with our
 Featherman and Hauser—A Rekned Model of Occupational Mobility • 451

finding that between generations immobility in Bishop et al. (1975, especially pp. 206–211, 225–
upper manual occupations is no more prevalent 228, 282–309, 320–324) is valuable. Our model is
than most types of extreme, long-distance mo- a special case of Goodman’s (1972c) general model.
bility. There is no evidence of “class” boundaries 2. The reported frequencies are based on a com-
plex sampling design and have been weighted to
limiting the chances of movement to or from the
estimate population counts while compensating for
skilled manual occupations. certain types of survey nonresponse. The estimated
Sixth, there is a rough equality in the pro- population counts have been scaled down to reflect
pensities to move in one direction or the other underlying sample frequencies, and an additional
between occupational strata. There are several downward adjustment was made to compensate for
exceptions to this symmetric mobility pattern, departures of the sampling design from simple ran-
some of which may be quite important, but dom sampling (see Featherman and Hauser [1978:
none suggests a dominant tendency toward up- Appendix B]). The frequency estimates in Table 52.1
ward relative to downward mobility across or have been rounded to the nearest integer, but our
computations have been based on unrounded fig-
within class boundaries.
ures. We treat the adjusted frequencies as if they had
Last, from a methodological perspective, our been obtained under simple random sampling.
description of the mobility regime is extremely
simple. In broad outline it might be fitted to a REFERENCES
5 × 5 table with the expenditure of as few as 2 Baron, James N. 1977. “The structure of intergenera-
df. None of our analyses of American mobility tional occupational mobility: Another look at the
tables required the expenditure of more than 6 Indianapolis mobility data.” Unpublished Mas-
of the 16 df left unused by the model of simple ter’s thesis, University of Wisconsin–Madison.
statistical independence. Bibby, John. 1975. “Methods of measuring mobil-
ity.” Quality and Quantity 9 (March):107–136.
We reemphasize that the present description
Bishop, Yvonne M. M., and Stephen E. Fienberg.
of relative mobility chances does not conflict
1969. “Incomplete two-dimensional contingency
in any way with our earlier description (Feath- tables.” Biometrics 25:119–128.
erman and Hauser 1978: Chapter 3) of occu- Bishop, Yvonne M. M., Stephen E. Fienberg, and
pational inflow and outflow patterns. Rather, Paul W. Holland. 1975. Discrete Multivariate
relative mobility chances are components of Analysis: Theory and Practice. Cambridge: M.I.T.
inflows and outflows, but the latter are also af- Press.
fected by distributions of occupational origins Blau, Peter M. 1965. “The flow of occupational
and destinations. Major features of the inflow supply and recruitment.” American Sociological
and outflow tables, like the pervasiveness of up- Review 30 (August):475–490.
Blau, Peter M., and Otis Dudley Duncan. 1967.
ward mobility between generations and within
The American Occupational Structure. New York:
the occupational career, are absent from the Wiley.
present account because they are functions of Boudon, Raymond. 1973. Mathematical Structure of
changing distributions of occupational origins Social Mobility. San Francisco: Jossey-Bass.
and destinations. Chessa, Frederico. 1911. La Trasmissione Erediteria
delle Professioni. Torino: Fratelli Bocca.
NOTES Duncan, Otis Dudley, and Robert W. Hodge. 1963.
1. We assume the familiarity of the reader with “Education and occupational mobility.” Ameri-
loglinear models for frequency data. Fienberg can Journal of Sociology 68 (May):629–644.
(1970a, 1977) and Goodman (1972a,b) give useful Featherman, David L., and Robert M. Hauser. 1978.
introductions, as does the comprehensive treatise by Opportunity and Change. New York: Academic
Bishop et al. (1975). We rely heavily on methods Press.
for the analysis of incomplete tables, which have Featherman, David L., F. Lancaster Jones, and Rob-
been developed by Goodman (1963, 1965, 1968, ert M. Hauser. 1975. “Assumptions of social
1969a,b, 1971, 1972c), Bishop and Fienberg (1969), mobility research in the United States: the case
Fienberg (1970b, 1972), and Mantel (1970); again,
452 • Part VI: Who Gets Ahead?

of occupational status.” Social Science Research variables.” American Sociological Review 37

4:329–360. (February):28–46.
Fienberg, Stephen E. 1970a. “The analysis of mul- ———. 1972c. “Some multiplicative models for the
tidimensional contingency tables.” Ecology analysis of cross-classified data.” Pp. 649–696 in
51:419–433. Proceedings of the Sixth Berkeley Symposium on
———. 1970b. “Quasi-independence and max- Mathematical Statistics and Probability. Berkeley:
imum likelihood estimation in incomplete University of California Press.
contingency tables.” Journal of the American Sta- Haberman, Shelby J. 1974. The Analysis of Frequency
tistical Association 65:1610–1616. Data. Chicago: University of Chicago Press.
———. 1972. “The analysis of incomplete mul- Hauser, Robert M. 1978. “A structural model of
tiway contingency tables.” Biometrics 23 the mobility table.” Social Forces 56 (March):
(March):177–202. 919–953.
———. 1977. The Analysis of Cross-Classified Cate- Hauser, Robert M., Peter J. Dickinson, Harry P. Tra-
gorical Data. Cambridge: M.I.T. Press. vis, and John M. Koffel. 1975. “Temporal change
Giddens, Anthony. 1973. The Class Structure of the in occupational mobility: evidence for men in the
Advanced Societies. New York: Harper and Row. United States.” American Sociological Review 40
Glass, D. B. 1954. Social Mobility in Britain. Lon- (June):279–297.
don: Routledge and Kegan Paul. Hope, Keith. 1976. Review of Thomas W. Pullum’s
Goldthorpe, John W., and Catriona Llewellyn. Measuring Occupational Inheritance. American
1977. “Class mobility in modern Britain: three Journal of Sociology 82 (November):726–730.
theses examined.” Sociology 11 (May):257–287. Iutaka, S., B. F. Bloomer, R. E. Burke, and O.
Goldthorpe, John W., Clive Payne, and Catriona Wolowyna. 1975. “Testing the quasi-perfect mo-
Llewellyn. 1978. “Trends in class mobility.” So- bility model for intergenerational data: interna-
ciology 12 (September):441–468. tional comparison.” Economic and Social Review
Goodman, Leo A. 1961. “Statistical methods for the 6:215–236.
mover-stayer model.” Journal of the American Sta- Levine, Joel Harvey. 1967. Measurement in the
tistical Association 56 (December):841–868. study of intergenerational status mobility. Un-
———. 1963. “Statistical methods for the prelimi- published doctoral dissertation. Department of
nary analysis of transaction flows.” Econometrica Social Relations. Harvard University.
31 (January):197–208. ———. 1972. “A two-parameter model of inter-
———. 1965. “On the statistical analysis of mo- action in father-son status mobility.” Behavioral
bility tables.” American Journal of Sociology 70 Science 17 (September):455–465.
(March):564–585. Mantel, Nathan. 1970. “Incomplete contingency
———. 1968. “The analysis of cross-classified data: tables.” Biometrics 26:291–304.
independence, quasi-independence, and interac- Mason, Karen Oppenheim, William M. Mason,
tion in contingency tables with or without miss- Halliman H. Winsborough, and Kenneth W.
ing entries.” Journal of the American Statistical Poole. 1973. “Some methodological issues in
Association 63 (December):1091–1131. cohort analysis of archival data.” American Socio-
———. 1969a. “How to ransack social mobility ta- logical Review 38 (April):242–258.
bles and other kinds of cross-classification tables.” Parkin, Frank. 1971. Class Inequality and Political
American Journal of Sociology 75 (July):1–39. Order: Social Stratification in Capitalist and Com-
———. 1969b. “On the measurement of social mo- munist Societies. New York: Praeger.
bility: an index of status persistence.” American Pearson, Karl. 1904. “On the theory of contingency
Sociological Review 34 (December):831–850. and its relation to association and normal cor-
———. 1971. “A simple simultaneous test proce- relation.” Reprinted, 1948, in Karl Pearson’s Early
dure for quasi-independence in contingency ta- Papers. Cambridge: Cambridge University Press.
bles.” Applied Statistics 20:165–177. Pullum, Thomas. 1975. Measuring Occupational In-
———. 1972a. “A general model for the analysis heritance. New York: Elsevier.
of surveys.” American Journal of Sociology 77 Ramsøy, Natalie. 1977. Social Mobilitet i Norge (So-
(May):1035–1086. cial Mobility in Norway). Oslo: Tiden Forlag.
———. 1972b. “A modified multiple regres- Rogoff, Natalie. 1953. Recent Trends in Occupational
sion approach to the analysis of dichotomous Mobility. Glencoe, Illinois: Free Press.
 Erikson and Goldthorpe—Trends in Class Mobility • 453

Singer, Burton, and Seymour Spilerman. 1976. “The White, Harrison C. 1963. “Cause and effect in social
representation of social processes by Markov mod- mobility tables.” Behavioral Science 8:14–27.
els.” American Journal of Sociology 82 (July):1–54. ———. 1970a. Chains of Opportunity: System Mod-
Tyree, Andrea. 1973. “Mobility ratios and associ- els of Occupational Mobility in Organizations.
ation in mobility tables.” Population Studies 27 Cambridge: Harvard University Press.
(July):577–588. ———. 1970b. “Stayers and movers.” American
Westergaard, John, and Henrietta Resler. 1975. Class Journal of Sociology 76 (September):307–324.
in a Capitalist Society: A Study of Contemporary
Britain. New York: Basic Books.

53 • Robert Erikson and John H. Goldthorpe

Trends in Class Mobility

The Post-War European Experience

The issue of trends in class mobility in industrial aims at establishing the distinctive properties of
societies is one characterised by a wide-ranging industrial societies in terms of the essential pre-
dissensus which, unfortunately, extends to mat- requisites for, or necessary consequences of, the
ters of fact as well as of interpretation. We do technical and economic rationality that is seen as
not suppose that in this paper we will be able to their defining characteristic. What is implied so
resolve all the disagreements that are apparent. far as social mobility is concerned may be put in
We do, however, believe that we can address the the form of the following three-part proposition.
issue on the basis of comparative mobility data In industrial societies, in comparison with
of a distinctively higher quality than those pre- preindustrial ones,
viously utilised, and that the results we report (i) absolute rates of social mobility are generally
have significant consequences—positive or neg- high, and moreover upward mobility—i.e.
ative—for most of the rival positions that have from less to more advantaged positions—
been taken up.1 predominates over downward mobility;
From the 1960s onwards, perhaps the dom- (ii) relative rates of mobility—or, that is, mo-
inant view on mobility trends has been that de- bility opportunities—are more equal, in the
rived from what we will refer to as the ‘liberal sense that individuals of differing social ori-
theory’ of industrialism, as developed by various gins compete on more equal terms to attain
American authors (Kerr et al., 1960, 1973; Kerr, (or to avoid) particular destinations; and
1969, 1983; Dunlop et al., 1975; cf. also Par- (iii) both the level of absolute rates of mobility
sons, 1960: chs. 3 and 4; 1967: chs. 4 and 15; and the degree of equality in relative rates
1971). This theory is a functionalist one which tend to increase over time.

This chapter, from a paper originally prepared for the European Research Conference (“European Society or
European Societies,” Gausdal, Norway, 1991), relies on material also used in a subsequent publication (Robert
Erikson and John H. Goldthorpe, The Constant Flux: A Study of Class Mobility in Industrial Societies [Oxford:
Clarendon Press, 1992]). Used by permission of the authors.
454 • Part VI: Who Gets Ahead?

To explain why these contrasts between through which particular individuals are allo-
mobility in pre-industrial and industrial so- cated to different positions within the division
ciety should arise, a number of arguments are of labour. Most fundamentally, rational proce-
deployed which have, moreover, been elabo- dures of social selection require a shift away from
rated and extended in the specialist literature ascription and towards achievement as the leading
by authors generally sympathetic to the liberal criterion: what counts is increasingly what in-
position (see esp. Blau and Duncan, 1967: ch. dividuals can do, and not who they are. More-
12; Treiman, 1970). While all the arguments in over, the growing demand for highly qualified
question take on a functionalist form, one may personnel promotes the expansion of education
usefully distinguish between those relating to and training, and also the reform of educational
three different kinds of effect—structural, proces- institutions so as to increase their accessibility to
sual and compositional. individuals of all social backgrounds. Human
First, it is held that within industrial society resources cannot be wasted; talent must be fully
the dynamism of a rationally developed technol- exploited wherever it is to be found. Thus, as
ogy calls for continuous, and often rapid, change within a society of widening educational provi-
in the structure of the social division of labour, sion ‘meritocratic’ selection comes to predomi-
which also tends to become increasingly differ- nate, the association between individuals’ social
entiated. High rates of mobility thus follow as origins and their eventual destinations tends
from generation to generation, and in the course steadily to weaken and the society takes on a
of individual lifetimes, the redistribution of the more ‘open’ character. And at the same time var-
active population is required: that is, among ious other features of industrialism also serve to
economic sectors—first, from agriculture to reduce the influence of social origins on indi-
manufacturing and then from manufacturing viduals’ future lives. For example, urbanisation
to services—and, in turn, among industries and and greater geographical mobility loosen ties of
among a growing diversity of occupations. Fur- kinship and community; mass communications
thermore, the overall tendency is for advancing spread information, enlarge horizons and raise
technology to upgrade levels of employment. Al- aspirations; and a greater equality of condition—
though some skills are rendered obsolete, new that is, in incomes and living standards—means
ones are created and the net effect is a reduction that the resources necessary for the realisation of
in the number of merely labouring and routine ambition are more widely available.
occupations and a rising demand for technically Thirdly, it is argued that the foregoing effects
and professionally qualified personnel. At the interact with each other, in that the emphasis
same time, both the increasing scale of produc- on achievement as the basis for social selection
tion, dictated by economic rationality, and the will be strongest within the expanding sectors
expansion of the services sector of the econ- of the economy—that is, the more technolog-
omy promote the growth of large bureaucratic ically advanced manufacturing industries and
organisations in which managerial and admin- services—and within the increasingly dominant
istrative positions also multiply. Industrial so- form of large-scale bureaucratic organisation.
cieties become increasingly ‘middle class’ or at Conversely, ascriptive tendencies will persist
least ‘middle-mass’ societies. Consequently, up- chiefly within declining sectors and organisa-
ward mobility is more likely than downward in tional forms—for example, within agriculture
both intergenerational and worklife perspective. or small-scale, family-based business enterprise.
Under industrialism, the chances of ‘success’ are In other words, compositional effects on mobil-
steadily improved for all. ity occur in that, once a society begins to indus-
Secondly, it is further claimed that as well as trialise, the proportion of its population that is
thus reshaping the objective structure of oppor- subject to the new ‘mobility regime’ character-
tunity, industrialism transforms the processes istic of industrialism not only increases as that
 Erikson and Goldthorpe—Trends in Class Mobility • 455

regime imposes itself, but further as those areas generally regarded as empirically untenable, their
and modes of economic activity that are most suggestion that a historic upward shift in such
resistant to it become in any event ever more rates tends to occur at some—perhaps quite
marginal. early—stage in the industrialisation process has
One reason that may then be suggested for not been similarly disconfirmed. Again, it is not
the degree of dominance exerted by the liberal part of Lipset and Zetterberg’s case that the high
position is the coherent way in which the un- mobility that they see as characteristic of indus-
derlying theory has been developed. Another is trial societies is the result of a tendency towards
the manifest failure of the main attempt made greater openness. Rather, they place the empha-
directly to controvert it. That is, the revision and sis firmly on the effects of structural change, and
extension of the Marxist theory of proletarian- in turn they are at pains to point out (1959: 27)
isation (cf. Braverman, 1974; Carchedi, 1977; that ‘the fact that one country contains a greater
Wright and Singelmann, 1982; Crompton and percentage of mobile individuals than another
Jones, 1984) which sought to show the necessity does not mean that that country approximates a
for the systematic ‘degrading’, rather than up- model of equal opportunity more closely’.
grading, of labour under the exigencies of late Secondly, a yet more radical challenge to
capitalism—with the consequence of large-scale the liberal view may be derived from the pio-
downward mobility of a collective kind. This un- neering work of Sorokin (1927/1959). Taking a
dertaking lacked from the start any secure empir- synoptic view, as much dependent on historical
ical foundation, and the accumulation of results and ethnographic evidence as on contemporary
incompatible with the new theory resulted in its social research, Sorokin was led to the conclu-
eventual abandonment even by those who had sion that in modern western societies mobility
been among its most resourceful supporters (see was at a relatively high level, and he was further
e.g., Singelmann and Tienda, 1985; Wright and ready to acknowledge the possibility that, from
Martin, 1987). However, various other positions the eighteenth century onwards, mobility rates
can still be identified that to a greater or lesser had in general shown a tendency to rise. How-
extent come into conflict with the liberal view ever, he was at the same time much concerned
and that continue to merit serious attention. to reject the idea that what was here manifested
First of all, it should be noted that the the- was in effect ‘the end of history’ and the start
ory of mobility in industrial society advanced by of a ‘perpetual and “eternal” increase of vertical
Lipset and Zetterberg (1956, 1959) and some- mobility’. Rather, Sorokin argued, the present
times simply assimilated to the liberal theory situation represented no more than a specific his-
(see e.g., Kerr, 1983: 53) does in fact differ from torical phase; in some societies in some periods
it in crucial respects. For example, Lipset and mobility increased, while in other periods it de-
Zetterberg do not seek to argue that mobility clined. Overall, no ‘definite perpetual trend’ was
steadily increases with industrial development: to be seen towards either greater or less mobility,
indeed, they remark that among industrial so- but only ‘trendless fluctuation’. Those who were
cieties no association is apparent between mo- impressed by the distinctiveness of the modern
bility rates and rates of economic growth. What era knew too little about historical societies and
they propose (1959: 13) is, rather, some kind of their diversity: ‘What has been happening is only
‘threshold’ effect: ‘our tentative interpretation an alternation—the waves of greater mobility su-
is that the social mobility of societies becomes perseded by the cycles of greater immobility—
relatively high once their industrialization, and and that is all’ (1959: 152–4).
hence their economic expansion, reaches a cer- It might from the foregoing appear that So-
tain level’. And although Lipset and Zetterberg’s rokin’s position was merely negative. But, in fact,
claim that (absolute) mobility rates in industrial underlying his denial of developmental trends in
societies become uniformly high would now be mobility and his preference for a cyclical view, at
456 • Part VI: Who Gets Ahead?

least the elements of a theory can be discerned. hypothesis that across industrial societies rates
In arguing against the supposition that rates of of social mobility display a basic similarity. This
mobility in the modern period are quite un- hypothesis cannot stand if expressed in terms of
precedented, one of the points Sorokin most absolute rates but, they argue, becomes far more
stresses is that while certain barriers to mobil- plausible if applied, rather, to relative rates.
ity have been largely removed—for example, When mobility is considered at the ‘phenotyp-
juridical and religious ones—it is important to ical’ level of absolute rates cross-national simi-
recognise that other barriers have become more larity can scarcely be expected. This is because
severe or have been newly introduced: for exam- these rates are greatly influenced by the struc-
ple, those represented by systems of educational tural context of mobility and, in turn, by effects
selection and occupational qualification (1959: deriving from a range of economic, technolog-
153–4, 169–79). This, moreover, is what must ical and demographic circumstances which are
always be expected: the forms of social stratifi- known to vary widely and which, so far as partic-
cation which provide the context for mobility ular individuals and families are concerned, must
are themselves structures expressing differential be regarded as ‘exogenously determined’. When,
power and advantage, and thus possess impor- however, mobility is considered net of all such
tant self-maintaining properties. Those who effects, or that is, at the ‘genotypical’ level of rel-
hold privileged positions will not readily cede ative rates, the likelihood of cross-national sim-
them and, in the nature of the case, can draw ilarity being found is much greater. For at this
on superior resources in their defence. Indeed, level only those factors are involved that bear on
Sorokin remarks that if he had to believe in the the relative chances of individuals of differing so-
existence of a permanent trend in mobility, it cial origin achieving or avoiding, in competition
would be in a declining one, since social strata with each other, particular destination positions
are often observed to become more ‘closed’ over among those that are structurally given. And
time as the cumulative result of those in superior there is reason to suppose that in modern soci-
positions using their power and advantage to re- eties the conditions under which such ‘endog-
strict entry from below (1959: 158–60). How- enous mobility regimes’ operate—for example,
ever, this propensity for closure—which we may the degree of differentiation in occupational hi-
understand as being endogenous to all forms of erarchies and in job rewards and requirements—
stratification—is not the only influence on mo- may not be subject to substantial variation.
bility rates. A further point that Sorokin several For present purposes, then, the chief signif-
times makes (see e.g. 1959: 141–152, 466–72) icance of the FJH hypothesis lies in the rather
is that in periods of both political and economic comprehensive challenge that it poses to the
upheaval—associated, say, with revolution or claims of liberal theorists. On the one hand,
war or with rapid commercial, industrial and so far as absolute mobility rates are concerned,
technological change—marked surges in mobil- it implies a basic scepticism, essentially akin
ity are typically produced as the social structure to that of Sorokin, about the possibility of any
as a whole, including the previously existing dis- long-term, developmentally driven trend; while,
tribution of power and advantage, is disrupted. on the other hand, it stands directly opposed
In other words, increased mobility here results to the proposition that under industrialism a
from the impact of factors that are exogenous to steady increase occurs in the equality of mobility
the stratification order. chances. Although some initial developmental ef-
Thirdly and finally, one may note a more re- fect in this direction early in the industrialisation
cently developed position which, however, has process might be compatible with the hypothe-
evident affinities with that of Sorokin. Feather- sis, any continuing change in relative mobility
man, Jones and Hauser (1975) aim at present- rates is clearly precluded (cf. Grusky and Hauser,
ing a reformulation of Lipset and Zetterberg’s 1984: 20). Once societies can be deemed to have
 Erikson and Goldthorpe—Trends in Class Mobility • 457

Table 53.1. National Inquiries Used as Data Sources

become industrial, their mobility regimes should of the century, in which the oldest respondents
stabilise in some approximation to the common within our national samples were born.
pattern that the FJH hypothesis proposes, and
should not thereafter reveal any specific or per-
sistent tendencies, whether towards convergence Relative Rates
on greater openness or otherwise. No forces are The reader is referred to Erikson & Goldthorpe
recognised as inherent in the functional dynam- (1992, Chapter 3) for a detailed account of ab-
ics of industrialism that work systematically to solute rates. We instead focus here on trends in
expand mobility opportunities. intergenerational class mobility from the stand-
The divergent arguments concerning mobil- point of relative rates, which we shall treat in
ity trends that we have reviewed in this section terms of odds ratios. That is, ratios which show
will then provide the context within which we the relative odds of individuals in two different
present our empirical analyses of data for Euro- classes of origin being found in one rather than
pean nations. The details of the surveys utilised another of two different classes of destination;
are provided in Table 53.1 (and see Erikson and or, alternatively, one could say, which show the
Goldthorpe 1992 for an elaborated discussion). degree of net association that exists between the
Although we here concentrate on the experience classes of origin and of destination involved. Two
of European nations, that by no means implies further points concerning such ratios may here
that we will be treating questions of mobility be noted.
trends within an unduly limited context. To the First, the total set of odds ratios that can be
contrary, we have the advantage that while these calculated within a mobility table may be taken
nations can supply us with high quality data (far as representing the ‘endogenous mobility re-
better, for example, than those usually available gime’ or ‘pattern of social fluidity’ that the table
from Third World nations), they do also display embodies. And since odds ratios are ‘margin in-
a remarkably wide range of variation in their lev- sensitive’ measures, it is then possible for this un-
els and patterns of industrial development—and derlying regime or pattern to remain unaltered
even if we consider only that time-span to which as between two or more mobility tables even
our mobility data have some reference: that is, though their marginal distributions differ and,
from the 1970s back to the first two decades thus, all absolute rates that can be derived from
458 • Part VI: Who Gets Ahead?

them. Further, we might add, it would be possi- analyses throughout on the five-class version of
ble for relative rates as expressed by odds ratios our schema (Table 53.2).
to differ from table to table in some systematic As regards relative rates, there is, as we have
way without this being readily apparent from an noted, an obvious opposition between the expec-
inspection of absolute rates. tations that follow from the liberal theory and
Secondly, odds ratios are the elements of log- those that may be derived from the FJH hypoth-
linear models; and, thus, where these ratios are esis. According to the former, a tendency should
taken as the measure of relative mobility rates, be found in the course of the development of
hypotheses about the latter can be presented and industrial societies for relative rates to become
formally tested through the application of such more equal—or, one could say, for all odds ratios
models. This is the approach that we shall here to move closer to the value of 1, which signifies
follow. the complete independence of class origins and
In examining relative rates, we shall attempt destinations or ‘perfect mobility.’ According to
to make inferences about the extent of changes the latter, relative rates will be basically the same
over time from the mobility experience of suc- across all societies that have market economies
cessive birth-cohorts within our national sam- and (at least) nuclear family systems, whatever
ples. However, instead of here working with stage their level of industrial development may
yearly cohorts, we distinguish four ten-year have reached; and thus, when examined over
birth-cohorts, which can then be regarded, given time within particular industrial societies, rela-
the closeness of the dates of the inquiries from tive rates should reveal little change at all.
which the samples derive, as having more or less We may then start off from an attempt at
comparable locations within the broad sweep of evaluating these rival positions, and, to this end,
recent European economic history. we introduce a rather simple loglinear model
The first—that is, the earliest—of these co- which is, however, able to provide a direct rep-
horts comprises men aged 55 to 64, who were resentation of expectations under the FJH hy-
thus mostly born in the first two decades of the pothesis, at least if this is taken stricto sensu. This
century, and who entered employment before model, which we have earlier labelled the ‘con-
or during the inter-war depression years. The stant social fluidity’ (CnSF) model (Goldthorpe,
second is of men aged 45 to 54, the majority 1980, 1987: ch. 3; Erikson, Goldthorpe and
of whom were born in the 1920s and entered Portocarero, 1983) may, for present purposes,
employment in the later 1930s or the war years. be written as
The third is that of men aged 35 to 44, who were
born between the late 1920s and the early 1940s, logFijk = ȝ + ȜOi + ȜDj + ȜCk (1)
and whose working lives have fallen very largely + ȜOC
ik
+ ȜDC
jk
+ ȜODij
within the post-war period. And finally the
fourth cohort is that of men aged 25 to 34, who where Fijk is the expected frequency in cell ijk of
were born from the end of the 1930s onwards a three-way table comprising class of origin (O),
and who mostly entered employment while class of destination (D) and cohort (C) and, on
the long boom was in train. This last cohort is the right-hand side of the equation, μ is a scale
clearly made up of respondents who could not factor, ȜOi, ȜDj and ȜCk represent the ‘main’ effects
be generally assumed to have reached a stage of of the distribution of individuals over origins,
occupational maturity, and this we must recall destinations and cohorts respectively, and the re-
where it might be relevant to the interpretation maining terms represent the effects for the three
of our results. possible two-way associations.
When we thus divide our national samples Thus, the model entails a number of substan-
into cohorts, we again have a potential problem tive propositions, most of which are unprob-
of unduly low cell counts in the mobility tables lematic: for example, that an association exists
for each cohort. To obviate this, we base our between class of origin and class of destination;
 Erikson and Goldthorpe—Trends in Class Mobility • 459

Table 53.2. The Class Schema

and that further associations exist between class We can then consider our nine nations sepa-
of origin and cohort and between class of des- rately, and in each case fit the above CnSF model
tination and cohort—in other words, men in to a three-way table which comprises the five
different cohorts have different origin and des- classes of origin, five classes of destination and
tination distributions. It is, however, a further four cohorts that we propose to distinguish. The
proposition that is critical. Since no three-way results of so doing are presented in Table 53.3.
association is provided for in the model (the In this table we also report the results of ap-
Ȝ term does not appear), it is also entailed plying a model that represents the hypothesis of
that the level of association between class of the (conditional) independence of class origins
origin and of destination is constant across co- and destinations; that is, the CnSF model minus
horts or, one could alternatively say, that over the Ȝ term. We do not expect this indepen-
the mobility tables for successive cohorts all cor- dence model to fit the data—and, as can be
responding relative rates, as measured by odds seen, in no case does it—but it serves as a useful
ratios, are identical. baseline, by reference to which we can assess,
460 • Part VI: Who Gets Ahead?

through the rG2 statistic in the fourth column differences in sample size. When one examines
of the table, how much of the total association the G 2(S) values in the final column of the table,
between class of origin and class of destination one finds that these are in fact contained within
the CnSF model is able to account for. a rather narrow range, and moreover that in no
Also in addition to the more usual ‘goodness case do they exceed the 65 mark, thus imply-
of fit’ statistics, we give in the last column of ing, with df = 48, p values of less than 0.05. In
Table 53.3 values for the statistic G 2(S). This other words, if we were restricted throughout
we introduce here to attempt to deal with a to a sample size of 1,746, such as that we have
difficulty arising from the large variation in the for Ireland, we would find it difficult to reject
sizes of our national samples. In consequence of the CnSF model for any nation—although the
this variation, our national mobility tables dif- FRG would have to be regarded as a borderline
fer quite widely in their capacity to show up as case.
statistically significant relatively small deviations What could therefore be claimed on the
from models that we fit to them. It is as if we basis of the foregoing is that while significant
were looking at slides through microscopes of deviations from the CnSF model do have to be
greatly differing power: we have the possibility recognised in at least some of our nations, such
of seeing far more detail in some cases than in deviations would not appear to be at all substan-
others. Thus, there is the evident danger that we tial. In this connection, it is of further relevance
do not evaluate a model in an evenhanded way to note that in all cases but one the CnSF model
from nation to nation. We could, for example, accounts for more than 90 per cent of the total
be led to reject a model in the Polish case on ac- association existing between class of origin and
count of deviations which, were they present also of destination—the exception being Sweden,
in the case of, say, Ireland, we would simply not where the independence model fits least badly;
observe. Thus, we evidently need some measure and again, that within the different national mo-
of goodness of fit that is standardised by sam- bility tables the CnSF model leads to the mis-
ple size. One possibility would be to take G 2/N. classification of, at most, only a little over 5 per
However, we prefer, as a more refined measure, cent of all cases.
Schwartz’s suggestion of G 2(S) which is given by
((G 2–df )/N) x K + df, where K is the sample size
which is to be taken as standard. We will follow Conclusions
the conservative practice of setting K equal to We have sought in this chapter to use data from
the size of the smallest of the national samples European nations in order to evaluate various
with which we are concerned—and thus, in arguments concerning mobility trends within
Table 53.3, at 1,746. To help remind the reader industrial societies. The liberal theory is under-
of the hypothetical nature of G 2(S)—that it is mined because, instead of the anticipated trend
the 2 value that we would expect from a sample of change, in the direction of greater equality, we
of size K, all other things being equal—we report find evidence of an essential stability. Although
it only to the nearest integer. shifts in relative rates can in some cases be de-
What, then, can we learn from the content tected (see Erikson and Goldthorpe 1992: 90–
of Table 53.3? It would in fact appear that the 101), these are not only ones which go in various
CnSF model performs fairly well. It is true that directions but, more importantly, ones which are
the p values reported indicate that in only four of very limited magnitude—so that one might
of the nine nations—England, Ireland, North- wish to speak more of ‘oscillation’ than of fluc-
ern Ireland and Sweden—would one retain this tuation. In other words, the liberal theory would
model, taken as the null hypothesis, according here appear to fail because the logic of indus-
to the conventional 0.05 criterion. However, trialism has not in fact automatically generated
it is also evident that much of the variation in the changes within processes of social selection
the G 2 and p values returned is attributable to which were expected of it, and through which
 Erikson and Goldthorpe—Trends in Class Mobility • 461

Table 53.3. Results of Fitting the CnST Model to Intergenerational Class Mobility
for Four Birth-Cohorts

Notes: * O = origin class; D = destination class; C = cohort.

†
rG2 shows the percentage reduction in the G2 for a model taken as baseline (here the condi-
tional independence model) that is achieved by a more complex model (here the CnSF model).
‡
Δ is the dissimilarity index, showing the percentage of all cases in the table analysed that are
misclassified—that is, allocated to the wrong cell—by a particular model.

a steady increase in fluidity and openness would largely in terms of the functional exigencies of
be promoted. industrialism (1970: 218). However, in a re-
It is in this connection of interest to note that cent paper (with Yip), he puts much stronger
of late exponents of the liberal theory appear to emphasis on the part that is played in creating
have modified their position in regard to relative greater openness and equality of opportunity
rates quite significantly. Thus, for example, Trei- by the more proximate factor of greater equal-
man initially sought to provide the hypothesis of ity of condition—that is, by a greater equality in
a trend towards greater openness with a rationale the economic, cultural and social resources that
462 • Part VI: Who Gets Ahead?

families possess. And while it is still maintained which indicate that some variation in fluidity pat-
that this increase in equality of condition itself terns does in fact occur among nations—indeed,
ultimately derives from the development of in- more than within nations over time—and also
dustrialism, it is at the same time accepted that that this variation shows no tendency to dimin-
‘industrialization and inequality do not move in ish (see Erikson and Goldthorpe 1992: 90–101).
perfect concert’ and, further, that other factors, Thus, expectations of convergence are not met.
especially political ones—for example, whether However, neither would cross-national variation
a nation has a socialist regime—may also affect appear to be increasing; and, more importantly,
the degree of inequality that exists (Treiman and as we have elsewhere sought to show (Erikson
Yip, 1989: 376–7). That is to say, it would here and Goldthorpe 1987, 1992: ch. 5), it could not
seem to be recognised that even in cases where a be reckoned as sufficiently wide to rule out the
trend towards greater fluidity may be empirically possibility that the ‘basic’ similarity in relative
established, this cannot be regarded as simply a rates that the FJH hypothesis claims is the major
matter of developmental necessity but must source of the temporal stability that we have ob-
rather be explained as the contingent outcome of served; or, that is, the possibility that constancy
quite complex patterns of social action (cf. also above all reflects commonality.
Ganzeboom, Luijkx and Treiman, 1989; Sim- In other words, in so far as the degree of
kus et al., 1990). And conversely, this revised, similarity proposed by the FJH hypothesis is
and evidently much weaker, position is then of established, we may think of temporal shifts in
course able to accommodate the alternative pos- fluidity within nations as being no more than
sibility that, in particular instances, no trend of oscillations occurring around the standard pat-
this kind is observed—because countervailing tern that the hypothesis implies or, at all events,
forces have in fact proved too strong. as being restricted in their frequency and extent
The stability in relative rates that we have by whatever set of effects it is that generates this
shown gains in significance, we may add, not pattern. And in this regard, then, the ultimate
only because the period that our data cover task becomes that of understanding these effects;
comprised decades of unprecedented economic or, that is, of seeking to explain not variance, to
growth but also because it was, of course, one which, as Lieberson (1987) has observed, analyt-
of major political upheavals, in which, in the ical strategies within macrosociology have thus
train of war and revolution, national frontiers far been chiefly oriented, but rather a lack of
were redrawn and massive shifts of population variance—for which, unfortunately, appropriate
occurred. The fact that the relative rates under- strategies remain largely to be devised.
lying the mobility experience of cohorts within
our national samples should then reveal so little NOTES
change—whether directional or otherwise—be- 1. This paper is based on chapter 3 of Robert
Erikson and John H. Goldthorpe, The Constant
comes all the more remarkable. While we have
Flux: A Study of Class Mobility in Industrial Soci-
not been able to support the claim of a sustained
eties, The Clarendon Press, Oxford, 1992. The re-
developmental trend, we have, it appears, found search on which this book reports was carried out
indications of something of no less sociological under the auspices of the CASMIN-Projekt, based
interest: that is, of a constancy in social pro- at the Institüt für Sozialwissenschaften of the Uni-
cess prevailing within our several nations over versity of Mannheim and funded by grants from
decades that would in general have to be char- the Stiftung Volkswagenwerk, Hanover. Readers are
acterised in terms of the transformation and tur- referred to the above work (chapter 2 esp.) for full
bulence that they witnessed. details of the comparative methodology followed in
Furthermore, this finding is, as we have indi- research.
cated, one which may be related to a larger so-
REFERENCES
ciological argument, namely, that represented by
Blau, P.M. and O.D. Duncan (1967): The American
the FJH hypothesis. We have presented analyses Occupational Structure. New York: Wiley.
 Erikson and Goldthorpe—Trends in Class Mobility • 463

Braverman, H. (1974): Labor and Monopoly Capital- ——— (1984): ‘Eras of Social Mobility in 19th and
ism. New York: Monthly Review Press. 20th Century Europe’. Journal of Social History
Carchedi, G. (1977): On the Economic Identification 17.
of Classes. London: Routledge. Kerr, C. (1969): Marshall, Marx and Modern Times.
Clark, C. (3rd ed., 1957): The Conditions of Eco- Cambridge: Cambridge University Press.
nomic Progress. London: Macmillan. ——— (1983): The Future of Industrial Societies.
Crompton, R. and G. Jones (1984): White-Collar Cambridge, Mass.: Harvard University Press.
Proletariat: Deskilling and Gender in Clerical Kerr, C., J.T. Dunlop, F.H. Harbison and C.A.
Work. London: Macmillan. Myers (1960, 2nd ed. 1973): Industrialism and
Dunlop, J.T., F.H. Harbison, C. Kerr and C.A. Industrial Man. Cambridge, Mass.: Harvard Uni-
Myers (1975): Industrialism and Industrial Man versity Press.
Reconsidered. Princeton: Inter-University Study Kocka, J. (1980): ‘The Study of Social Mobility and
of Human Resources in National Development. the Formation of the Working Class in the 19th
Erikson, R. and J.H. Goldthorpe (1987): ‘Common- Century’. Le mouvement social 111.
ality and Variation in Social Fluidity in Industrial Kuznets, S. (1966): Modern Economic Growth. New
Nations. Part I: A Model for Evaluating the “FJH Haven: Yale University Press.
Hypothesis”; Part II: The Model of Core Social Lieberson, S. (1985, 2nd ed. 1987): Making It
Fluidity Applied’. European Sociological Review 3. Count. Berkeley: University of California Press.
——— (1992): The Constant Flux: A Study of Class de Lillo, A. and A. Schizzerotto (1985): La valutazi-
Mobility in Industrial Societies. Oxford: Claren- one sociale delle occupazioni. Bologna: II Mulino.
don Press. Lipset, S.M. and H.L. Zetterberg (1956): ‘A The-
Erikson, R., J.H. Goldthorpe and L. Portocarero ory of Social Mobility’. Transactions of the Third
(1983): ‘Intergenerational Class Mobility and World Congress of Sociology, vol. 3. London: Inter-
the Convergence Thesis’. British Journal of So- national Sociological Association.
ciology 34. ——— (1959): ‘Social Mobility in Industrial Socie-
Featherman, D.L., F.L. Jones and R.M. Hauser ties’ in S.M. Lipset and R. Bendix, Social Mobil-
(1975): ‘Assumptions of Social Mobility Research ity in Industrial Society. Berkeley: University of
in the US: The Case of Occupational Status’. So- California Press.
cial Science Research 4. Mendels, F.F. (1976): ‘Social Mobility and Phases
Gagliani, G. (1985): ‘Long-Term Changes in the of Industrialisation’. Journal of Interdisciplinary
Occupational Structure’. European Sociological History 7.
Review 1. Parsons, T. (1960): Structure and Process in Modern
Ganzeboom, H., R. Luijkx and D.J. Treiman (1989): Societies. Glencoe: Free Press.
‘Intergenerational Class Mobility in Comparative ——— (1967): Sociological Theory and Modern Soci-
Perspective’. Research in Social Stratification and ety. New York: Free Press.
Mobility 8. ——— (1971): The System of Modern Societies. En-
Goldthorpe, J.H. (with Catriona Llewellyn and glewood Cliffs: Prentice-Hall.
Clive Payne) (1980, 2nd ed. 1987): Social Mobil- Priebe, H. (1976): ‘The Changing Role of Agricul-
ity and Class Structure in Modern Britain. Oxford: ture, 1920–1970’ in C.M. Cipolla, ed., Fontana
Clarendon Press. Economic History of Europe, vol. 5 (ii). London:
Goldthorpe, J.H. (1985): ‘On Economic Devel- Fontana.
opment and Social Mobility’. British Journal of Rostow, W.W. (1960): The Stages of Economic
Sociology 36. Growth: A Non-Communist Manifesto. Cam-
Grusky, D.B. and R.M. Hauser (1984): ‘Compara- bridge: Cambridge University Press.
tive Social Mobility Revisited: Models of Conver- Schwartz, J.E. (1985): ‘Goodman’s Coefficient of
gence and Divergence in 16 Countries’. American Multiple Determination: Why It Is Not Anal-
Sociological Review 49. ogous to R 2’. Stockholm: Swedish Institute for
Hauser, R.M., P.J. Dickinson, H.P. Travis and J.M. Social Research.
Koffel (1975): ‘Temporal Change in Occupa- Simkus, A.A., R. Andorka, J. Jackson, K-B. Yip and
tional Mobility: Evidence for Men in the United D.J. Treiman (1990): ‘Changes in Social Mobil-
States’. American Sociological Review 40. ity in Two Societies in the Crux of Transition: A
Kaelble, H. (1981): Historical Research on Social Mo- Hungarian-Irish Comparison, 1943–1973’. Re-
bility. London: Croom Helm. search in Social Stratification and Mobility 9.
464 • Part VI: Who Gets Ahead?

Singelmann, J. (1978): From Agriculture to Services: Stratification: Research and Theory for the 1970s.
The Transformation of Industrial Employment. Indianapolis: Bobbs Merrill.
Beverly Hills: Sage. Treiman, D.J. and K-B. Yip (1989): ‘Educational
Singelmann, J. and M. Tienda (1985): ‘The Process and Occupational Attainment in 21 Countries’
of Occupational Change in a Service Society: in M.L. Kohn, ed., Cross-National Research in
The Case of the United States, 1960–1980’ in B. Sociology. Newbury Park: Sage.
Roberts, R. Finnegan and D. Gallie, eds., New Wright, E.O. and B. Martin (1987): ‘The Transfor-
Approaches to Economic Life. Manchester: Man- mation of the American Class Structure, 1960–
chester University Press. 1980’. American Journal of Sociology 93.
Sorokin, P.A. (1927, 2nd ed. 1959): Social and Cul- Wright, E.O. and J. Singelmann (1982): ‘Proletari-
tural Mobility. Glencoe: Free Press. anization in the Changing American Class Struc-
Treiman, D.J. (1970): ‘Industrialisation and So- ture’. American Journal of Sociology 88.
cial Stratification’ in E.O. Laumann, ed., Social

54 • Richard Breen

Social Mobility in Europe

I discuss here the findings of a project that looks social class according to the seven-class CASMIN
at social mobility in 11 European countries over scheme (see Table 54.2). This identifies classes I
a period of almost 30 years (see Breen 2004 for + II (the service class); III (routine non-manual
details). I want to summarize some of the main class); IVab (petty bourgeoisie with and without
results and discuss how they relate to previous employees); IVc (farmers); V + VI (skilled manual
research in this area; and I also want to make workers, technicians and supervisors of manual
some comments about the consequences of this workers); VIIa (unskilled manual workers not in
research and of these findings for the study of agriculture); and VIIb (farm workers). One con-
social mobility. But I will begin by briefly ex- sequence of choosing this categorization is that
plaining what the project is all about. it allows our results to be compared with those
of The Constant Flux (Erikson and Goldthorpe
1992), where the same categories were used.
Data
The data used in the project comprise 117 mo-
bility surveys covering the period 1970 to 2000. Methodological Issues
The sources of the data are shown in Table 54.1. As Table 54.1 showed, the 11 countries con-
The number of tables per country ranges from tribute rather different numbers of mobility ta-
two in Israel and Italy to 35 in the Netherlands. bles to our cross-national analyses. Sweden, for
In general the age range of the respondents example, has a table for every year from 1976
in our mobility tables is 25 to 64, and we coded to 1999, whereas Poland and Ireland have only

The ideas, issues and theories summarized in this brief piece are examined in greater detail in the following
publication: Richard Breen, Social Mobility in Europe, published by Oxford University Press in 1992. The
article printed here was originally prepared by Richard Breen for The Inequality Reader: Contemporary and
Foundational Readings in Race, Class, and Gender, First Edition, edited by David B. Grusky, pp. 405–422.
Copyright © 2007 by Westview Press.
 465

Table 54.1. Sources of Data

Country No. of Sources of Data Years for Which Data Are

Tables Included
Germany 22 Zumabus 1976–77 1979(2) 1980 1982
Allbus 1980 1982 1984 1986 1988
1990–92 1994 1996 1998
Politik in der BRD 1978 1980
Wohlfahrtssurvey 1978
German socio-economic panel 1986 1999
France 4 Formation-qualification professionnelle Insee surveys 1970 1977 1985 1993
Italy 2 National survey on social mobility 1985
Italian household longitudinal survey 1997
Ireland 3 Survey of the determinants of occupational 1973
status and mobility
Survey of income distribution and poverty 1987
Living in Ireland survey 1994
Great Britain 15 General household survey 1973 1975–76 1979–1984
1987–1992
Sweden 24 Annual surveys of living conditions (ULF) 1976–1999
Norway 3 Colbjørnsen et al. 1987 1982
Moen et al. 1996 1994
Level of Living Survey 1995
Poland 3 Zagorski 1976 1972
Slomczynski 1989 1988
Treiman/ Szelényi 1994
Hungary 4 Social mobility and life history survey 1973 1983 1992
Way of Life and Time Use Survey 2000
(Hungarian Central Statistical Office)
Israel 2 Matras and Weintraub, 1977 1974
Kraus and Toren 1992 1991
Netherlands 35 Parliamentary Election Study 1970 1971 1977 1981
1982 1986 1994 1998
Political Action Survey I 1974 1979
Justice of Income Survey 1976
CBS Life Situation Survey 1977 1986
National Labour Market Survey 1982
National Prestige and Mobility Survey 1982 1985 1988 1990
1992 1994 1996
Strategic Labour Market Survey 1998
Cultural Changes [ISSP] 1987
Justice of Income Survey 1987
Primary and Social Relationships 1987
Social and Cultural Trends 1990
Justice of Income Survey [ISJP] 1991
Family Survey I, 1992–93 1992
Households in the Netherlands pilot 1994
Households in the Netherlands 1995
Social Inequality in the Netherlands 1996
National Crime Study 1996
Social and Economic Attitudes 1998
Netherlands Family Survey II 1998
Use of Information Technology 1999
466 • Part VI: Who Gets Ahead?

Table 54.2. The Goldthorpe Class Schema

Goldthorpe Description Employment Relations CASMIN Version

Class
I Higher-grade professionals, Employer or service I + II Service class
administrators and officials; relationship
managers in large industrial
establishments; large proprietors
II Lower-grade professionals, Service relationship
administrators and officials; higher-
grade technicians; managers in small
industrial establishments; supervisors
of non-manual employees
IIIa Routine non-manual employees, Intermediate III Routine
higher-grade (administration non-manual
and commerce)
IIIb Routine non-manual employees, Labour contract
lower-grade (sales and services)
IVa Small proprietors, artisans, etc., Employer IVab Non-farm
with employees petty-bourgeoisie
IVb Small proprietors, artisans, etc., Self-employed
without employees
IVc Farmers and smallholders; Employer or self- IVc Farmers and
other self-employed workers employed other self-employed
in primary production workers in primary
production
V Lower-grade technicians; Intermediate V + VI Technicians,
supervisors of manual workers supervisors and
skilled manual
VI Skilled manual workers Labour contract workers
VIIa Semi- and unskilled manual workers Labour contract VIIa Semi- and unskilled
(not in agriculture, etc.) manual workers
(not in agriculture)
VIIb Semi- and unskilled manual Labour contract VIIb Semi- and
workers in agriculture unskilled manual
workers in agriculture

three tables each, covering the years between equal we must, as a consequence, attach more
the early 1970s and 1994. The amount of infor- credence to results about temporal trends drawn
mation we possess regarding change over time, from countries with a larger number of observa-
and the reliability of the conclusions based on tions (Sweden, the Netherlands, Great Britain
this information, will vary between countries. and Germany).
If we have a small number of observations, any Furthermore, the data that we use are never
one of them may be very influential in deter- free of error, and differences in data quality may
mining whether or not the data display a trend easily be mistaken for substantive differences.
(as we shall see) and this will inevitably lead to We have used the best-quality data available
uncertainty in the conclusions we draw. All else from each of our 11 countries, but we still
 Breen—Social Mobility in Europe • 467

need to be aware of the potential for differen- confidence in the extent to which they allow the
tial reliability and validity to induce spurious specifics of the pattern of social fluidity to be
cross-national variation and temporal change. compared across either time or countries. This
As far as the differences between countries are consideration has then dictated our choice of
concerned, the fieldwork for the surveys we use models. Rather than seeking to develop detailed
was in all cases carried out according to inter- models of the fluidity regime we prefer instead
nationally accepted procedures, and the sub- to fit rather general models and to assess their
sequent coding of the variables—notably class adequacy using several measures (including the
origins and destinations—followed a common conventional chi-squared goodness of fit test
and widely implemented procedure. Neverthe- and the index of dissimilarity).
less, while adherence to such norms is some re-
assurance that the data attain high standards of
quality, the surveys in the various countries were Absolute Mobility and Class
carried out independently of each other, and so Structure
we should be cautious about what we infer from In contemporary studies of social mobility a key
them concerning cross-national differences. As distinction is drawn between observed patterns
far as change within countries is concerned, of social mobility, sometimes referred to as ‘ab-
we can have more faith in our findings when solute mobility’, and social fluidity (or ‘relative
the various surveys have been administered in mobility’). Absolute mobility is concerned with
a consistent fashion. In three cases the data al- patterns and rates of mobility, where mobility is
ways come from the same survey series: these understood simply as movement between class
are France (the FQP—Formation-Qualification origins (the social class in which someone was
Professionnelle—surveys), Britain (the Gen- brought up) and class destinations (the class
eral Household Survey) and Sweden (the ULF they occupy at the time of the survey). Social
series). In a further five countries the data sets fluidity concerns the relationship between class
come from highly comparable sources: these are origins and current class position: specifically it
Ireland (where the three surveys were all carried is based on the comparison, between people of
out by the same fieldwork organization), Hun- different class origins, of their chances of being
gary (where the four surveys were all fielded by found in one destination class rather than an-
the Hungarian Central Statistical Office), Italy other. If these chances were the same regardless
(where a number of the same academics were of origins then the mobility table would display
involved in the design and execution of the perfect mobility: the class in which a respon-
two surveys), Germany and Israel. But in the dent was found would not depend on (would
remaining three cases—Norway, Poland and be independent of) the class in which she or he
the Netherlands—the data come from various was brought up. Social fluidity is often inter-
sources within each country and thus the pos- preted as an index of equality in the chances of
sibility that variations in data quality might be access to more or less advantageous social po-
mistaken for temporal change is greatest here. sitions between people coming from different
We believe that more reliance can be placed social origins (in other words, as an index of
on estimates of trends within countries than societal openness),1 and contemporary research
measures of differences between them: thus on social mobility accordingly pays much more
our discussion, later in this chapter, of which attention to social fluidity than to absolute mo-
countries are more or less open in their mobility bility. I will follow that precedent here and so
regime, should be interpreted with some cau- will make only a few comments about absolute
tion. Finally, while the data that we have are mobility.
probably adequate for presenting a picture of In fact, our results concerning absolute mo-
broad trends and differences, we would have less bility can be summarised quite easily. There has
468 • Part VI: Who Gets Ahead?

Table 54.3a. Aggregate Class Structures (All Men) in the 11 Countries by Decade
gg g
Class 1970s 1980s 1990s
I + II 23.1 28.6 30.8
III 8.7 9.0 10.1
IVa + b 7.9 8.6 10.4
IVc 8.6 5.7 4.0
V + VI 27.7 27.6 27.1
VIIa 20.6 18.3 15.7
VIIb 3.5 2.3 2.0
Note: Each country is weighted equally in computation of aggregate class structure.

Table 54.3b. Aggregate Class Structures (Women in the Labour Force) in the 10
Countries by Decade
Class 1970s 1980s 1990s
I + II 22.1 30.5 34.6
III 32.8 32.3 35.1
IVa + b 6.5 6.0 6.1
IVc 8.6 4.4 2.3
V + VI 6.1 6.3 7.1
VIIa 21.1 18.6 13.7
VIIb 2.8 1.7 1.2
Note: Each country is weighted equally in computation of aggregate class structure.

been a marked convergence in the class struc- upward and of downward mobility—and,
tures of European countries and in their patterns again, this is evident among both sexes, as
of absolute mobility, and these things are true for Tables 54.4 and 54.5 show. But, further, the
both men and women. distribution of people in the mobility tables
A. Convergence in class structures has been of the different countries has also grown
driven by some internationally consistent more similar. If we calculate the index of
trends, such as the growth in the service dissimilarity (Δ) from comparisons, be-
class, I + II, and the decline in manual work, tween all pairs of countries, of their entire
particularly of the unskilled kind. Among mobility tables, we find that the average Δ
women, increased rates of labour force par- (the average difference between countries)
ticipation have been associated with a re- falls from 43 per cent in the 1970s, to 33
duction in international variation as more in the 1980s and 30 in the 1990s, among
and more of them enter occupations in the women, with the comparable figures for
white collar classes, I + II and III. But the men being 39, 30 and 30 per cent. And the
single biggest cause of this convergence has variance around these means has also de-
been the declining significance of the farm clined: from 163.2 to 62.6 to 41.6 among
classes, IVc and VIIb, in those countries women and from 137.5 to 62.9 to 56.1
(such as Poland, Ireland, and Hungary) among men. Although European coun-
where a large farm sector persisted until the tries continue to show differences in their
last quarter of the 20th century. absolute mobility flows, these have become
B. This trend towards convergence in class smaller.
structures has occurred together with de-
creasing variation between countries in Absolute mobility concerns the observed
their rates of overall mobility, of vertical, of rates and patterns of flows between origin and
Table 54.4. Percentage Mobile by Country in Each Decade (All Men)
g y y
Total Mobility
Germany France Italy Ireland Great Sweden Norway Poland Hungary Israel N’lands Mean Variance
Britain
1970s 61.6 66.6 — 56.7 63.0 70.8 — 59.4 77.5 74.4 66.3 66.3 48.0
1980s 62.1 67.5 69.5 61.3 61.8 71.4 71.9 61.0 74.9 — 67.7 66.9 25.8
1990s 60.3 67.0 72.1 66.1 60.8 71.0 68.1 67.4 71.6 74.3 65.7 67.7 19.9

Vertical Mobility
1970s 44.1 43.8 — 39.9 50.7 54.0 — 40.9 53.0 43.7 50.6 46.7 28.5
1980s 45.8 45.9 40.8 42.6 50.8 54.7 55.2 42.9 55.8 — 54.1 48.9 34.6
1990s 46.3 46.3 46.3 45.5 50.7 55.2 52.1 45.9 53.7 50.4 54.0 49.7 13.9

Upward Mobility
1970s 31.7 25.9 — 21.6 32.8 35.1 — 22.1 26.9 20.1 36.1 28.0 37.1
1980s 33.6 29.1 29.0 27.9 33.1 35.3 39.3 24.8 34.7 — 38.9 32.6 22.9
1990s 33.3 29.9 35.9 31.4 31.7 36.6 34.2 26.3 35.9 35.0 37.7 33.4 11.4

Downward Mobility
1970s 12.4 17.9 — 18.4 17.9 19.0 — 18.8 26.2 23.5 14.5 18.7 17.3
1980s 12.2 16.8 11.8 14.7 17.7 19.4 15.9 18.0 21.1 — 15.2 16.3 8.7
1990s 13.0 16.4 10.4 14.1 19.0 18.6 17.9 19.6 17.8 15.4 16.3 16.2 8.0
469
 470

Table 54.5. Percentage Mobile by Country in Each Decade (Women in the Labour Force)

Total Mobility
Germany France Italy Great Sweden Norway Poland Hungary Israel N’lands Mean Variance
Britain
1970s 74.0 71.4 — 78.8 73.1 — 50.8 81.0 76.5 74.0 72.5 86.4
1980s 75.6 77.6 74.3 76.3 73.6 76.2 66.3 79.5 — 73.9 74.8 13.7
1990s 72.6 77.2 75.0 73.9 73.2 77.4 76.2 76.5 82.2 72.3 75.7 8.8

Vertical Mobility
1970s 48.6 41.7 — 52.1 55.4 — 34.0 54.1 44.9 51.4 47.8 52.2
1980s 48.8 45.7 51.0 52.6 56.4 54.1 48.5 58.2 — 51.4 51.9 15.7
1990s 47.3 46.0 47.9 53.2 57.9 53.0 50.3 55.7 53.5 53.6 51.8 14.8

Upward Mobility
1970s 25.8 27.8 — 27.5 23.9 — 19.5 23.3 26.0 30.9 25.6 11.7
1980s 29.6 32.9 38.5 29.0 27.5 34.4 31.7 38.8 — 33.6 32.9 15.7
1990s 32.2 33.2 36.7 30.6 33.5 37.1 34.1 42.0 39.0 34.8 35.3 11.6

Downward Mobility
1970s 22.8 13.9 — 24.6 31.5 — 14.4 30.8 19.0 20.5 22.2 44.2
1980s 19.2 12.8 12.5 23.7 28.9 19.8 16.8 19.4 — 17.8 19.0 26.0
1990s 15.2 12.8 11.3 22.5 24.4 15.9 16.2 13.7 14.5 18.8 16.5 17.6
 Breen—Social Mobility in Europe • 471

destination classes and, in mobility analysis, is the recent rapidity of the transition out of ag-
is treated as the consequence of social fluidity riculture. Similarly, we saw in our comparative
(the relative chances of people from each origin analysis that the shift towards a concentration of
being found in each destination class) operating women in the white-collar classes has been more
within fixed origin and destination distributions. rapid in countries such as Hungary and Poland
A model in which origins and destinations are where the class distribution in the 1970s differed
independent, given the observed distributions of most from this. The result has been the growing
these two in each country and at each point in similarity in destination distributions that we
time, correctly classifies over 80 percent of cases, have already remarked upon. But because coun-
while a model which also assumes a common tries embarked on this developmental path long
level and pattern of social fluidity correctly clas- before the first of our surveys was fielded, there
sifies around 95 per cent of cases. It is evident, is also decreasing variation in class origins. The
therefore, that changes over time, and differ- mean value of the ǻ between class origins for
ences between countries, in absolute mobility are each pair of countries fell from 33 percent in the
driven by variation in the origin and destination 1970s to 23 in the 1980s and 24 in the 1990s.2
distributions rather than in social fluidity. Absolute mobility flows converged because their
Can such variation be said to follow a pat- main determinants did.
tern? We believe that the answer, in very broad This convergence chiefly occurred between
terms, is yes. We might imagine societies fol- the 1970s and 1980s and whether the trend will
lowing a developmental path that incorporates persist, or even strengthen, is, of course, difficult
two major transitions: from an agricultural to to say. Clearly, if the working classes continue
an industrial society, and from an industrial to a to decline in those countries where the decline
post-industrial society. The consequences, for the has begun, and if this extends from VIIa to V +
class structure, of the former transition are a de- VI, then further convergence will be inevitable
cline in the proportions in classes IVc and VIIb as men, like women, come to be heavily con-
and a growth in the remaining classes, especially centrated in classes I + II and III. Recent his-
(among men) the manual working classes V + torical experience of the location of industrial
VI and VIIa. The transition to a post-industrial production would suggest that we can expect
society sees the decline of V + VI and VIIa and further convergence; in any event, it seems un-
the growth of I + II and III. Everywhere the de- likely that any of these countries will display a
cline in agriculture is either more or less com- growth in classes V + VI and VIIa, while some at
plete (Britain, Germany, Sweden, Israel, the least will experience a decline. As for the coun-
Netherlands) or well under way while, in eight tries in which these classes have not yet begun
of our 11 countries (Ireland, Poland and Hun- to decline (Ireland, Poland and Hungary), the
gary being the exceptions), between the 1970s outlook seems less certain. In Ireland the growth
and 1990s, the class structure saw a steady fall in of classes I + II and III has outstripped that of
the proportion of men in classes V + VI and VIIa V + VI and VIIa over this period, but this is not
and a consistent increase in the proportion in I true of the male class structure in Poland and
+ II and III. Among women the pattern was ex- Hungary. On the other hand, among women
actly the same. These differences mean that some in Poland and Hungary there has been a steady
countries display a post-industrial class structure growth in classes I + II and III and an increase,
with a heavy concentration of people in classes I then a decline, in V + VI and VIIa, suggesting
+ II and III: this is particularly true of the male that the second transition may be under way.
class structure in Britain and the Netherlands Much here depends on the nature of economic
and it is true of the female class structure in sev- development. Foreign direct investment in man-
eral countries. But the important thing, from the ufacturing, as in the Irish case, is one mechanism
point of view of the study of absolute mobility, by which the size of the working class may be
472 • Part VI: Who Gets Ahead?

Figure 54.1. Annual LmSF (or Unidiff ) Coefficients by Country (All Men)
Germany France

1.1
0.9
0.7
0.5 Ireland Great Britain

1.1
0.9
0.7
0.5 Sweden Poland

1.1
0.9
0.7
0.5 Hungary Netherlands
1.1
0.9
0.7
0.5
70

75

80

85

90

95

00

70

75

80

85

90

95

00
19

19

19

19

19

19

20

19

19

19

19

19

19

20
Year

sustained and the rate of convergence conse- (three), and Hungary (four), and the lack of a
quently slowed. consistent pattern of change in these countries,
must leave some room for uncertainty. But in
contrast to absolute mobility we see no evidence
Social Fluidity of convergence among countries in their social
In our comparative analysis we found that trends fluidity. Figures 54.1 (for men) and 54.2 (for
in social fluidity are very similar among men and women) show these within-country trends in the
women, showing a widespread tendency towards form of the annual ȕ coefficients from an LmSF
greater fluidity. Britain is the sole clear exception model3 with common local association among
to this: here there has been little or no change. all the yearly tables of a given country.
In other cases—notably Germany—there is no Figure 54.3 (for men) and Figure 54.4 (for
statistically significant change, though the trend, women) show the LmSF ȕ parameters from a
at least for men, is towards a weaker association model, applied to decade data from each coun-
between origins and destinations. Elsewhere—in try, which assumes common local association
France, Ireland, Sweden, Poland, Hungary and in each table, varying only by ȕ. The value for
the Netherlands—there is a statistically signifi- Britain in the 1980s is set to unity. Among men,
cant increase in fluidity, though the small num- Figure 54.3 shows that, in the 1970s, levels of
ber of observations for Ireland (three), Poland social fluidity were lowest in Germany, France,
 Breen—Social Mobility in Europe • 473

Figure 54.2. Annual LmSF (or Unidiff ) Coefficients by Country (Women in the Labour Force)
Germany
1.9
1.5
1.1
.7
.3
France Great Britain
1.9
1.5
1.1
.7
.3
Sweden Poland
1.9
1.5
1.1
.7
.3
Hungary Netherlands
1.9
1.5
1.1
.7
.3
70

75

80

85

90

95

00

70

75

80

85

90

95

00
19

19

19

19

19

19

20

19

19

19

19

19

19

Year 20

Italy, Ireland, Hungary and the Netherlands regimes: for example, the within decade variance
and highest in Britain, Sweden, Norway, Poland of the ȕs shown in Figure 54.3 is largest for the
and Israel. Fluidity increased in France, Sweden 1980s and the ǻ for CnSF across countries in
and the Netherlands and possibly in Ireland, each decade is larger for the 1990s than for the
Hungary and Poland too. The increases in the 1980s.
Netherlands and Hungary were particularly The picture among women (Figure 54.4)
marked. These different trends have left several is very similar. Once again, the points for the
countries—Sweden, Norway, Poland, Hungary 1970s are above those for the 1980s, which are
and the Netherlands—with, as far as we can above those for the 1990s, indicating a general
tell, rather similar rates of fluidity, followed by tendency for fluidity to increase, with Britain
Britain (where the absence of change has led to being an exception. The average ȕ falls from
a shift in its relative position), Ireland, France, 1.28 in the 1970s to 1.14 in the 1980s and 1.05
Italy and Germany, which remains the coun- in the 1990s. France and Germany are the least
try with the strongest association between class fluid societies; Britain, Sweden, Poland and, by
origins and class destinations. At the other ex- the 1990s, the Netherlands are the most fluid.
treme, Israel is consistently more open than any Hungary presents a different picture for women
other country. Overall, however, we can find no than men, the former showing much lower flu-
convincing evidence of convergence in fluidity idity, compared with other countries, than the
474 • Part VI: Who Gets Ahead?

Figure 54.3. LmSF (or Unidiff ) Coefficients per Decade, per Country (All Men)

1.3

I I
N
N N
L N
1.1 N
I
L
L
I
N
I
L
.9
N N
1970s
I N I L
I 1980s
L L L
L 1990s
I
.7
DE FR IE GB SE PL HU NE
Country

latter. In Israel the values are 0.84 in the 1970s direction. For women the picture is very similar.
and 0.71 in the 1990s. Taken together with the Of 18 decade-to-decade changes, two of them
results for men this is evidence of the exception- show a decline in fluidity (Germany between the
ally fluid nature of Israeli society. 1980s and 1990s, Britain between the 1970s and
Overall, the results from our 11 countries 1980s) while 14 show an increase.
point to a clear conclusion: there is a widespread
tendency for social fluidity to increase, even
though this might not be a statistically signifi- Previous Research
cant trend in every case. Among men, fluidity is These results differ quite substantially from those
greater at the end of the period than at the start of the previous major comparative study of social
in every country except Britain and Israel (where mobility, namely The Constant Flux, in which
the values remain the same). Furthermore, of the Erikson and Goldthorpe (1993) argue strongly
20 decade-to-decade changes in fluidity reported in favour of the so-called FJH hypothesis of a
in Breen (2004), we find that in 16 of them flu- basic similarity in social fluidity in all industrial
idity increased and it declined in three—in Ire- societies ‘with a market economy and a nuclear
land and Britain between the 1970s and 1980s family system’ (Featherman, Jones and Hauser
and in Norway between the 1980s and 1990s. 1975: 340) and they also claim (1992: 367) that
There is just one further notable case in which ‘relative rates possess a high degree of tempo-
fluidity fell (but which is obscured by the use ral stability’. Indeed, our results are somewhat
of aggregated decade data) and that is Hungary, closer to those of the slightly earlier compara-
where fluidity declined significantly between the tive study by Ganzeboom, Luijkx and Treiman
1992 and 2000 observations. Although there are (1989). They used 149 mobility tables for men
some cases (such as Sweden) where we cannot be drawn from 35 countries spanning the period
unequivocal about an increase in fluidity, we can 1947–86. They claim that they show that ‘al-
say with confidence that nowhere (with the pos- though . . . there is a basic similarity in mobility
sible exception of post-Communist Hungary) patterns . . . at the same time there are substantial
is there any evidence of a trend in the opposite cross-national and cross-temporal differences in
 Breen—Social Mobility in Europe • 475

Figure 54.4. LmSF (or Unidiff ) Coefficients per Decade, per Country (Women in the Labour Force)
1.7

N N
1.5 N
I
1.3 L L
I I N
L
1.1 N I

I N
N L
N I
1970s I L
.9
I 1980s L L
L 1990s
.7

DE FR GB SE PL HU NE
Country

the extent of mobility.’ Furthermore, ‘a smaller class destination in mobility tables, Breen and
but significant part of the variance in mobility Jonsson (1997) found that reliability was lower
regimes can be explained by the trend towards for origin information from older age groups,
increasing openness over time’ (Ganzeboom, Lu- implying that the common practice of using age
ijkx and Treiman 1989: 47). However, although groups to draw conclusions about cohort change
our results and those of Ganzeboom et al. are over time in mobility regimes may be unsound.
rather similar, it should be noted that the latter More simply, one cannot properly make infer-
have been heavily criticised, notably by Wong ences about change from cross-sectional data.
(1994: 138), who reanalyses their data and finds An important reason why we observe a trend
that ‘the model of temporal invariance cannot be towards increasing social fluidity and Erikson
rejected for a majority of countries . . . Hungary and Goldthorpe do not is that our data, by and
and Sweden are the only countries giving irrefut- large, refer to later-born cohorts. Although our
able evidence of temporal variation’. analyses, like those of the overwhelming major-
Why do we find evidence for change and ity of mobility studies, are based on period data,
cross-national difference when Erikson and there is, I believe, good reason to suppose that
Goldthorpe did not? As far as temporal change social fluidity is driven by cohort, rather than
is concerned, because Erikson and Goldthorpe period, effects.
have only one mobility table per country, their Erikson and Goldthorpe’s data represent co-
claim that there is little systematic temporal vari- horts born between, approximately, 1900 and
ation in patterns of social fluidity within coun- 1945, whereas our data extend this to cohorts
tries is based on analyses in which age groups are born around 1970. This means that our data
taken to represent different birth cohorts. But contain a larger share of people who benefited
as Breen and Jonsson (2003) have pointed out, from what has been called the Golden Age of
this approach confounds lifecycle and cohort Welfare Capitalism—in other words, the long
effects and makes no allowance for either selec- post-war economic boom and the generally
tive mortality or recall errors. In an assessment more egalitarian educational and social welfare
of the reliability of measures of class origin and policies that followed in its wake. Work that I
476 • Part VI: Who Gets Ahead?

have done with Janne Jonsson on Sweden clearly fit to the data than CnSF, but this incremental
shows that here the major reduction in class in- improvement to any goodness of fit measure is
equalities in educational attainment took place small. For example, when we analyzed our data
among cohorts of people born between 1910 according to decade, a very small index of dis-
and 1950 and that since then there has been no similarity was returned by a model that allowed
further equalization. But this means that, as we for no temporal or cross-national variation in
compare period data for the 1970s, 1980s and social fluidity (3.95 per cent for men and 3.81
1990s, we gradually lose the older, less fluid co- for women) and allowing for such variation only
horts, who are replaced by younger, more fluid improved ǻ by, at most, two percentage points.
cohorts. In the case of Sweden such a process This compares with a ǻ of around 15 per cent
will continue until the second decade of the 21st in models in which origins and destinations are
century, after which we expect to see no further independent. Much the same picture emerged
increase in period social fluidity. But it seems when we used annual data, and arguments like
likely that the specific cohorts which benefited this usually lead to the conclusion that most so-
from the Golden Age will have differed between cial fluidity is common and invariant over time.
countries, so leading to variation between them Sometimes the same point is made using the de-
in the fluidity they display in particular periods. viance, rather than ǻ, as the yardstick, and here
Whereas we argue that countries display con- the result is even more extreme. For example, 90
siderable variation in social fluidity, Erikson and per cent of the deviance returned by the model
Goldthorpe argue the contrary. But in their own of independence disappears when we add com-
analyses (Erikson and Goldthorpe 1992: 381) mon (among countries and over decades) social
the logarithm of each odds ratio in their mo- fluidity, and a model allowing fluidity to change
bility table from Czechoslovakia (which is the over time and differ between countries improves
country in which they find the greatest fluidity) it only by a further seven per cent. Taken to-
is two-thirds as strong as that from Scotland (the gether, the decompositions of the deviance and
country with the least fluidity). So, for example, of ǻ would seem to indicate that more than 85
given an odds ratio of 3 in Czechoslovakia, the per cent of social fluidity is common over na-
comparable odds ratio would be 5.3 in Scotland. tions and time.
On this basis it is difficult to argue for common- This then seems to conflict with our finding
ality in the strength of social fluidity, and, in- of large variation between countries in their odds
deed, these results are quite similar to ours: in ratios. The problem with using arguments based
our data the logarithm of the odds ratios in Israel on ǻ or the deviance to support the argument
is around half those in Germany. So, an odds that social fluidity is largely invariant is that vari-
ratio of 3 in Israel would be an odds ratio of 7 in ation in fluidity is assessed using measures of the
Germany. Given this, one might ask why Erik- consequences of fluidity for the whole mobility
son and Goldthorpe, like many others, conclude regime (i.e. for absolute mobility), and, as we
that there is little variation in fluidity. The reason have seen, these consequences are unanimously
is the following. All comparative mobility stud- agreed to be quite minor. An analogy may help
ies find that the difference in the goodness of fit to make the point. In a linear regression, Y = a +
between a model in which social fluidity is com- bX; X (analogous to social fluidity) may display a
mon across all countries (the model of common lot of variation, but it will have little impact on Y
social fluidity or CnSF) and a model in which (analogous to overall mobility) if the coefficient,
fluidity can differ between them is small, rela- b, is close to zero. But simply because b is small
tive to the difference between the CnSF model we could not then claim that there was little vari-
and one in which origins and destinations are ation in X. Measures such as the change in ǻ or
independent. It is certainly true that the model in G 2 might be said to capture the strength of
of variation in fluidity is a significantly better effect of fluidity on overall mobility, but they do
 Breen—Social Mobility in Europe • 477

not measure the variation in fluidity itself, and it its Gini coefficient. Overall we could discern no
is therefore mistaken to conclude, on this basis, tendencies towards either convergence or diver-
that social fluidity is common and invariant. It gence in fluidity, and thus the hypothesis that, as
is empirically the case that, between countries nations have come to follow different policy tra-
or over time, large variations in social fluidity jectories—particularly in economic policy—so
can be found which nevertheless have little im- we might see growing differences between them
pact on the overall mobility regime. To illustrate in fluidity, receives no support. There is some
this: if we take the fluidity pattern from the 1997 indication, however, that fluidity is greater in
Italian men’s table in our data and insert it into state socialist (Poland and Hungary) and social
the 1991 Israeli men’s table, while preserving democratic (Norway and Sweden) countries, and
the Israeli marginal distributions, the ǻ between the argument for such a political explanation
the real and the constructed Israeli tables is 6 receives additional support from the finding of
per cent.4 When we consider that the Israeli and declining fluidity in Hungary during the 1990s.
Italian mobility regimes are close to the extremes But, on the other hand, we observe very high
of the range of fluidity found in our data, this fluidity in Israel and data from the General So-
suggests that 6 per cent represents the maximum cial Survey (made available to us by Mike Hout)
impact of differences in fluidity on the distri- show that fluidity is high in the United States.
bution of individuals in the mobility table. The This leads to the conclusion that direct political
conclusion to be drawn from these apparently intervention of the kinds associated with state
contradictory measures of the variation in flu- socialist and social democratic societies may be
idity is not that it is common or invariant, but, one means by which a society can reach relatively
rather, that even quite substantial differences in high rates of fluidity, but it is not the only one.
fluidity have little impact on the distribution of One major difficulty in devising theories to
cases over the mobility table—i.e. on observed, explain variation in mobility or fluidity is that
absolute mobility flows. mobility tables, especially period-based tables,
reflect a large number of underlying processes—
artefactual, contingent and substantive. For one
Explanation thing, this aggregation of processes renders it
Thus far I have said nothing about how we might difficult to explain variations in fluidity; for an-
explain patterns of social mobility or social flu- other, it may also be the case that some of the
idity, aside from some rather vague remarks commonality that has often been observed in
about the impact of policies pursued during the comparisons of social fluidity derives from the
Golden Age of Welfare Capitalism. In broad mixing together in the mobility tables of pro-
terms, explanations of variations in social flu- cesses that, when investigated separately, might
idity are usually pitched at a macro-sociological show greater and more systematic societal and
level: that is, fluidity is assumed to be related to temporal differences.
other, societal level measures. The very influen- While it is reasonable to suppose that fluid-
tial ‘liberal theory of industrialism’, for example, ity in a nation is shaped by government policy,
argues that fluidity will increase with economic the education system, the workings of the labour
development. In our case, the stage of economic market and the like, it is also the case that what
development of our countries varies rather little, we observe in a mobility table may also reflect
but, even so, we could find no evident link be- purely artefactual sources of variation arising
tween their ranking in fluidity terms and their from differences in the way that the data them-
GDP per capita. Nor did we find any support for selves represent the underlying phenomenon of
Erikson and Goldthorpe’s argument that fluidity interest. Furthermore, what we might call con-
is related to income inequality: there is no signif- tingent factors, which are usually omitted from
icant association between a country’s fluidity and any theoretical discussion of social fluidity, may
478 • Part VI: Who Gets Ahead?

play an important role in shaping what we ob- description rather than sociological explanation.
serve. Walter Müller and Reinhard Pollak (see But while this might be a valid concern if we
Breen 2004) attribute the high fluidity they conceive of class structures as macro-sociological
find among Germans born in the 1920s to the phenomena, it may be less so, and may leave
massive migration from the eastern part of Ger- open the possibility of sociological explanation,
many that occurred following the Second World if we were to turn our attention to the detailed
War. The measured class origins of this cohort evolution of businesses and firms and of the jobs
are thus their pre-migration origins, which had that constitute classes.
very little relevance in shaping their subsequent Furthermore, it now seems to me that a pe-
mobility patterns: the physical detaching of a riod perspective on social mobility may not be
large share of the cohort from their true origins the most appropriate if, as I have suggested,
led to higher measured social fluidity. The same social fluidity is driven by factors related to co-
argument may explain the high level of fluidity horts, rather than periods. The cohort perspec-
in Israel, a country in which a very large share tive on social mobility has not been entirely
of the population is comprised of immigrants. neglected, but it has certainly played a very
secondary role to the usual period view. Under
the cohort approach, change in the association
Conclusions between origins and destinations is believed to
The experience of this project should lead us occur in specific cohorts, while, from a period
to question the balance that mobility research point of view, this association is considered
has struck between social fluidity and absolute to change among all cohorts at similar histor-
mobility. The emphasis has lain heavily on the ical points in time. The importance, for social
former but, insofar as we are concerned with the change, of the replacement of older by younger
mobility regime, this now seems inappropriate. cohorts has been stressed by several sociologists.
This is by no means to deny that social fluidity In the specific field of mobility research, Erikson
tells us important things about the prevailing and Goldthorpe (1992), for example, argue that
degree of inequality in the chances of attaining equality of condition is essential for equality of
one class position rather than another, and may opportunity, and that changes in the latter will
be indicative of other characteristics of society. then predominantly take effect during a person’s
Nevertheless, although one would not want to upbringing and schooling. As a result, the over-
say that fluidity can never make a difference all level of inequality of opportunity in society
(since we can easily construct examples in which will change mainly through cohort replacement,
extreme patterns of fluidity will be highly con- with younger cohorts experiencing a different
sequential for the distribution of cases in a mo- degree of inequality of educational attainment
bility table), within the advanced industrial and than older ones. It has long been education that
post-industrial societies, the range of fluidity that is the major channel of social mobility: if this is
we observe is relatively inconsequential in deter- so, then social fluidity should respond to changes
mining variation in mobility flows and in the in the level of class inequality in educational at-
life chances of individuals and families as these tainment, which, because educational inequality
are captured in measures of class position. Many is itself a characteristic of cohorts, implies a co-
previous authors (such as Grusky and Hauser hort explanation of change in fluidity.
1984; Goldthorpe 1985) have called for more Though rarely spelled out, there is a tension
attention to be paid to structural change, but, between the cohort interpretation of change in
as Erikson and Goldthorpe (1992: 104, 189) social fluidity and the belief that period effects
suggest, it is not clear how such change should drive variation in the origin-destination associ-
be explained nor, indeed, whether it might not ation over the lifecycle through labour market
be better approached as a matter of historical changes that simultaneously affect different
 Breen—Social Mobility in Europe • 479

birth cohorts. We believe that there are strong 4. We use the observed Italian fluidity pattern,
arguments for adopting a cohort rather than, or and thus the magnitude of the difference that we
in addition to, a period perspective. It is widely report does not depend on the adequacy of any par-
agreed that changes in an individual’s social class ticular model of fluidity.
position are relatively rare after the age of about
REFERENCES
35 (Goldthorpe 1980: 51–2, 69–71; Erikson
Breen, Richard (ed.). 2004. Social Mobility in Europe.
and Goldthorpe 1992: 72). Such stability im- Oxford: Oxford University Press.
plies that, except in unusual circumstances in Breen, Richard, and Jan O. Jonsson. “Period and Co-
which this stability is disrupted, period effects hort Change in Social Fluidity: Sweeden 1976–
will be a less important source of change in flu- 1999.” Unpublished paper.
idity when compared with cohort replacement Erikson, Robert and John H. Goldthorpe. 1992. The
and within-cohort change that occurs during the Constant Flux: A Study of Class Mobility in Indus-
early years in the labour force. trial Societies. Oxford: Oxford University Press.
Featherman, David L., F. L. Jones, and Robert M.
NOTES Hauser. 1975. “Assumptions of Mobility Re-
1. The terms association and social fluidity can search in the United States: The Case of Occupa-
be used interchangeably: greater social fluidity im- tional Status.” Social Science Research 4: 329–60.
plies lower origin–destination association. In the Ganzeboom, Harry B. G., Ruud Luijkx, and Donald
log-linear modeling context in which this paper is J. Treiman. 1989. “Intergenerational Class Mo-
situated, this association is captured by odds ratios. bility in Comparative Perspective.” Research in
2. These figures are for men. For women the fig- Social Stratification and Mobility 8: 3–84.
ures are 36, 24 and 24 per cent. The slight differ- Goldthorpe, John H. 1980. Social Mobility and Class
ences arise because our samples of women include Structure in Modern Britain. Oxford: Clarendon
only those in the labour force and we have no data Press.
for women in Ireland. ———. 1985. “On Economic Development and
3. LmSF means ‘log-multiplicative social fluidity’ Social Mobility.” British Journal of Sociology 36:
(by analogy with CnSF—‘common social fluidity’). 549–573.
We use LmSF to refer to the unidiff (Erikson and Grusky, David B. and Robert M. Hauser. 1984.
Goldthorpe 1992) or log-multiplicative layer ef- “Comparative Social Mobility Revisited: Models
fect model (Xie 1992) when the local association is of Convergence and Divergence in 16 Coun-
modeled in a completely unspecified way (as it is in tries.” American Sociological Review 49: 19–38.
CnSF). Wong, Raymond S. 1994. “Postwar Mobility Trends
in Advanced Industrial Societies.” Research in So-
cial Stratification and Mobility 13: 121–44.
480 • Part VI: Who Gets Ahead?

55 • Jan O. Jonsson, David B. Grusky, Matthew Di Carlo,

and Reinhard Pollak

It’s a Decent Bet That Our Children Will Be

Professors Too

Are children born into privilege very likely to either a categorical form that has parents passing
end up privileged themselves? Are children born on a big-class position (e.g., manager, profes-
into less privileged families likewise fated to re- sional, craft worker) to their children or a gra-
main in their social class of origin? We care about dational form that has parents passing on their
such questions for many reasons but perhaps pri- socioeconomic standing to their children. We
marily because they speak to whether the com- argue here that these standard approaches ignore
petition for money, power, and prestige is fairly the important role that detailed occupations play
run. For many people, the brute facts of extreme in reproducing inequality.
poverty or inequality are not in themselves prob- The conventional wisdom about how to mea-
lematic or objectionable, and what really matters sure mobility was codified a half century ago. The
is simply whether the competition for riches is a study of mobility bifurcated at that time into one
fair one in which everyone, no matter how ad- camp that represented social structure in grada-
vantaged or disadvantaged their parents may be, tional terms (e.g., Svalastoga 1959) and another
has an equal chance to win. This commitment that represented it in big-class terms (e.g., Carls-
to a fair competition motivates a quite extensive son 1958; Glass 1954). These competing repre-
research literature on how much mobility there sentations of social structure were subsequently
is, whether some countries have more of it than attached to competing understandings of how
others, and whether opportunities for mobility inequality is reproduced: The class scholar as-
are withering away. sumed that parents pass on their social class to
The purpose of this chapter is to ask whether children, while the gradational scholar assumed
conventional methods of monitoring mobility that parents pass on their occupational prestige
are adequate for the task. We’re concerned that or socioeconomic standing to their children.
they’re not and that, in particular, such methods Under both approaches, detailed occupations
may overlook some of the most important forms were usually treated as the appropriate starting
and sources of rigidity. The long-standing con- point in representing the underlying structure of
vention in the field, and one that we regard as inequality, but they were transformed either by
problematic, has been to assume that intergen- aggregating them into big social classes (i.e., the
erational reproduction takes one of two forms, class approach) or by scaling them in terms of

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
materials appearing in: Jonsson et al. 2009, Grusky et al. 2008, and Jonsson et al. 2011. The article printed
here was originally prepared by Jan O. Jonsson, David B. Grusky, Matthew Di Carlo, and Reinhard Pollak,
in The Inequality Reader: Contemporary and Foundational Readings in Race, Class, and Gender, Second Edition,
edited by David B. Grusky pp. 499–516. Copyright © 2011 by Westview Press.
 Jonsson et al.—It’s a Decent Bet That Our Children Will Be Professors Too • 481

their socioeconomic status or prestige (i.e., the acquaintances about their income, it’s not just
gradational approach). The study of mobility has because doing so is viewed as too intrusive and
in this sense been reduced to the study of either personal but also because we suspect that query-
class or socioeconomic mobility, yet quite strik- ing about occupation will yield more in the way
ingly these simplifying assumptions have come of useful information.
to be adopted with little in the way of evidence
that they adequately characterize the structure of
opportunity. Mechanisms of Reproduction
Is it possible that both class and gradational If our main argument, therefore, is that occupa-
representations are incomplete and obscure im- tions are an important conduit for reproduction,
portant rigidities in the mobility regime? We this is obviously not to suggest that inequality
suggest that indeed these simplifying represen- is reproduced exclusively through occupations.
tations provide only partial accounts and that Rather, there’s good reason to believe that, while
the structure of inequality is best revealed by much reproduction occurs through occupations,
supplementing them with a third representation the more frequently studied big-class and socio-
that treats detailed occupations as fundamental economic mechanisms are also doing impor-
conduits of reproduction. Because the social, tant reproductive work. We suggest below that
cultural, and economic resources conveyed to a comprehensive mobility model should exam-
children depend so fundamentally on the de- ine at once reproduction at the socioeconomic,
tailed occupations of their parents, one might big-class, and microclass levels. In most mobility
expect such occupations to play a featured role in analyses, the three levels are confounded, and
intergenerational reproduction, but in fact this conclusions about the structure of mobility may
role has gone largely unexplored in most mobil- conceal possible differences in how these forms
ity analyses. of reproduction play out. We develop this ar-
It’s not just that detailed occupations serve gument below by reviewing each of these three
as a main conduit for reproduction. In addi- mechanisms of reproduction in turn.1
tion, they index the main communities and Gradational regime: The gradational (or socio-
identities of workers, and as such they should economic) approach to studying mobility has in-
be understood as a powerful omnibus indicator equality taking on a simple unidimensional form
of the social world within which individuals in which families are arrayed in terms of either
are located. At a dinner party, we tend to ask a income or occupational status. The life chances
new acquaintance “What do you do?” because of children growing up within such systems are a
the response, almost invariably conveyed in the function, then, of their standing within this uni-
form of a detailed occupation, provides at once dimensional queue of families. When children
evidence about life chances and capacities (skills are born high in the queue, they tend to secure
and credentials, earnings capacity, networks), high-status and highly rewarded occupations by
honor and esteem (prestige, socioeconomic sta- virtue of (1) their privileged access to the eco-
tus), and the social and cultural world within nomic resources (e.g., wealth, income) needed
which interactions occur (consumption prac- either to purchase training for the best occupa-
tices, politics, and attitudes). We care, in other tions (e.g., an elite education) or to “purchase”
words, about occupations because they are preg- the jobs themselves (e.g., a proprietorship), (2)
nant with information on the life chances, social their privileged access to social networks pro-
standing, and social world of their incumbents viding information about and entrée to the best
(see Weeden and Grusky 2005). The (largely un- occupations, and (3) their privileged access to
tested) bias in this regard is that occupation is far cultural resources (e.g., socialization) that mo-
more strongly correlated with these many vari- tivate them to acquire the best jobs and provide
ables than is income. If we tend to avoid asking them with the cognitive and interactional skills
482 • Part VI: Who Gets Ahead?

(e.g., culture of critical discourse) to succeed tastes and aspirations develop in class-specific
in them. Under the gradational model, it is ways (e.g., the children of proprietors develop
the total amount of resources that matters, and tastes for autonomy, and the children of routine
children born into privileged circumstances are nonmanuals develop tastes for stability), because
privileged because they have access to so many human capital is cultivated and developed in
resources (e.g., Hout and Hauser 1992). The class-specific ways (e.g., the children of pro-
imagery here is accordingly that of two unidi- prietors develop entrepreneurial skills, and the
mensional hierarchies, one for each generation, children of routine nonmanuals develop bureau-
smoothly joined together through the mediating cratic skills), because social capital is distributed
mechanism of total resources (economic, social, in class-specific ways (e.g., the children of propri-
or cultural). In Figure 55.1a, an ideal-typical etors are apprised of entrepreneurial opportuni-
gradational regime is depicted by projecting a ties, and the children of routine nonmanuals are
detailed cross-classification of occupational or- apprised of routine nonmanual opportunities),
igins and destinations onto a third dimension, and because the tangible physical capital (e.g., a
one that represents the densities of mobility and shop, business) passed on to children of propri-
immobility. This graph, which orders origin etors motivates them to remain proprietors. By
and destination occupations by socioeconomic virtue of these processes, children do not have
score, shows the characteristic falloff in mobility generic access to all occupations of comparable
chances as the distance between origin and des- standing (as gradationalists would have it), but
tination scores increases. instead are especially well positioned to assume
Big-class regime: The big-class regime, by con- occupations that align with the culture, training,
trast, has inequality taking the form of mutually contacts, and capital that their class origins en-
exclusive and exhaustive classes. These classes tail. We represent an ideal-typical class regime of
are often assumed to convey a package of con- this sort in Figure 55.1b. Because we are focus-
ditions (e.g., employment relations), a resulting ing on reproduction, we have assumed here (and
social environment that structures behavior and in Figure 55.1c) that all off-diagonal cells have
decision making, and a culture that may be un- the same density, save for random noise.
derstood as an adaptation (or maladaptation) to Microclass regime: The occupational, or “mi-
this environment. For our purposes, the relevant croclass,” approach shares with the big-class
feature of this formulation is that all children model the presumption that contemporary
born into the same class will have largely the labor markets are balkanized into discrete cate-
same mobility chances, even though their par- gories, but such balkanization is assumed to take
ents may hold different occupations with dif- principally the form of institutionalized occupa-
ferent working conditions and socioeconomic tions (e.g., doctor, plumber, postal clerk) rather
standing. The logic of the class situation is as- than institutionalized big classes (e.g., routine
sumed, then, to be overriding and to determine nonmanuals). By implication, the occupations
the life chances of the children born into it. Ob- constituting big classes will have differing pro-
versely, two big classes of similar status will not pensities for mobility and immobility, a het-
necessarily convey to their incumbents identical erogeneity that obtains because the distinctive
mobility chances, as they may differ on various occupational worlds into which children are
nonstatus dimensions that have implications for born have consequences for the aspirations they
mobility. For example, even though proprietors develop, the skills they value and to which they
and routine nonmanuals are roughly similar in have access, and the networks upon which they
socioeconomic status, the children of proprietors can draw (see Table 55.1). The children of car-
will tend to become proprietors and the children penters, for example, may be especially likely to
of routine nonmanuals will tend to become rou- become carpenters because they are exposed to
tine nonmanuals. This pattern arises because carpentry skills at home, socialized in ways that
 483

Figure 55.1a.
Gradational
Regime

Figure 55.1b.
Big-Class Regime

Figure 55.1c.
Microclass
Regime

Note: The base of each figure indexes occupational origin and destination, while
the vertical dimension indexes densities of mobility and immobility for each
possible combination of origin and destination.
484 • Part VI: Who Gets Ahead?

Table 55.1. Mechanisms of Intergenerational Reproduction

Type of reproduction
Type of resources Big-class Micro-class
Human capital General or abstract skills Occupation-specific skills
(e.g., cognitive or verbal abilities) (e.g., acting skills, carpentry skills)
Cultural capital Abstract culture and tastes Occupation-specific culture and tastes
(e.g., “culture of critical discourse”) (e.g., aspirations to become a medical doctor)
Social networks Classwide networks (typically Occupation-specific networks (typically
developed through neighborhood- developed through on-the-job interactions)
or job-related interactions)
Economic resources Liquid resources Fixed resources
(e.g., stocks, bonds, income) (e.g., business, farm)

render them especially appreciative of carpentry a special interest in understanding “how things
as a vocation, and embedded in social networks work.” In the aftermath of the World Trade
that provide them with information about how Center collapse, we can imagine the engineer’s
to become carpenters and how to secure jobs in family talked mainly about why the building
carpentry. Although a microclass regime again failed structurally, whereas the sociologist’s fam-
assumes a lumpy class form, the lumpiness is ily talked mainly about why there is terrorism.
much finer than big-class analysts would allow The transmission of occupation-specific
(see Figure 55.1c). The strong big-class repro- human capital is likely to occur outside the
duction that we long thought was revealed in professional sector as well. The mechanic is es-
mobility tables may instead be artifactual and pecially likely to spend time at home engaging
express little more than the tendency for repro- in repairs, may take her or his children into the
duction at the detailed occupational level. repair shop, and may otherwise encourage an
We have referred above to the occupational interest in taking things apart and fixing them
skills, culture, and networks that parents trans- (i.e., a “practical” engineer). Likewise, the seam-
mit to their children. The transmission of skills stress may talk frequently about fashion at home,
should, however, be particularly stressed and take her or his children to fashion shows, and
may well be especially important in understand- train them in sewing and designing clothes.
ing why occupations are passed on. The con- These examples make the simple point that the
ventional view would have it that the ongoing occupational commitments of parents can affect
separation of home and workplace has made what they discuss or practice at home, how they
it more difficult for parents to transmit such spend time with their children, and hence the
occupational human capital. We agree that its skills that they impart to their children.
transmission may now be weakened, but this It would be possible to presume that repro-
obviously does not mean that it’s altogether pre- duction takes on an exclusively gradational,
cluded. The sociologist, for example, may well big-class, or microclass form and build a mobil-
talk shop with her or his children at the dinner ity model that then capitalizes on the imagery
table, litter the home with books and magazines underlying that particular form. The field has
that betray a sociological orientation, and in all indeed often proceeded in just that way: That is,
other ways inculcate a sociological perspective big-class analysts have often insisted on build-
in the natural course of everyday child rearing. ing purist big-class models, while gradation-
The engineer, by contrast, may bring home toys alists have in turn insisted on building purist
that involve building things, focus conversation gradational models. The model that we develop
and inquiry on the world of things, and impart will, by contrast, combine all three forms (big
 Jonsson et al.—It’s a Decent Bet That Our Children Will Be Professors Too • 485

class, microclass, gradational) and thereby make advanced by Goldthorpe, that “a reliance on
it possible to tease out the net contribution of occupationally specific factors, which are likely
each. We apply this approach to ask (1) whether themselves to be quite variable over time and
the mobility regime contains pockets of extreme space, would seem especially inadequate” in ex-
microclass rigidity that are concealed when mi- plaining class reproduction (2007, 144).
croclasses are aggregated into big classes and (2) In the present analysis, we will not be explor-
whether such microclass reproduction is the ing the structure of cross-national variation in re-
main mechanism through which big classes are production, and instead we’ll be presenting the
reproduced. If the answer to these questions is in shared features that hold in approximate form
the affirmative, it will follow that there is more in all countries. We refer the reader elsewhere
microclass rigidity than is consistent with the (Jonsson et al. 2009, 2011) for a discussion of
practice of ignoring it and less big-class rigidity cross-national variability in microclass mobility.
than is consistent with the practice of building
our analyses exclusively around it.
We suspect that a microclass foundation to The Structure of Contemporary
reproduction is a generic feature of late industri- Mobility
alism rather than something idiosyncratic to the The analyses presented here will be carried out
United States. The mechanisms that we’ve laid with data sets that provide information on the
out are, after all, in play to a greater or lesser ex- father’s occupation, the child’s occupation and
tent in all countries (see Table 55.1). The relative age, and other variables that aid in occupational
strength of big-class or microclass reproduction and big-class coding (e.g., employment status,
in any given society will be affected by the pre- branch of industry). Because our analyses are
vailing mix of institutional forms, some support- pitched at the occupational level, our father-by-
ing big-class structuration (e.g., trade unions) son mobility tables will have many cells, and
and others supporting microclass structuration large data sets for each country are needed. We
(e.g., state-supported occupational closure). We meet this requirement by drawing on multiple
have chosen to analyze four countries (Germany, surveys in all countries save Sweden. For Swe-
the United States, Sweden, and Japan) that, by den, the occupational data for the children come
virtue of this different mix of institutional forms, from the 1990 Census, and the occupational
have mobility regimes that support reproduction data for the parents are recovered by linking
of different types. to the 1960 and 1970 Censuses (Erikson and
How might mobility vary by country? Jonsson 1993). The data from the remaining
Whereas Germany and the United States are countries come from the sources listed in Table
often understood as the home ground of occu- 55.2. For this chapter, we’re forced to focus on
pationalization, Sweden has a long tradition of the mobility of men, as we’ve found that women
big-class organization, and Japan is typically as- experience more complicated mobility processes
sumed to be stratified more by family and firm that are not as readily summarized in such a
than by big class or occupation. We have sought short treatment. We have discussed the mobility
in this fashion to explore the reach of micro- of women in Jonsson et al. (2009).
class mechanisms into labor markets, like those We have worked hard to render the data as
of Sweden and Japan, that have not historically comparable as possible. Given our need for large
been regarded as taking a microclass form. If a data sets, some compromises nonetheless had to
microclass mechanism nonetheless emerges as be made, most notably pertaining to the period
fundamental in Sweden or Japan, the case for covered and the age of the respondents. The data
building that mechanism more systematically from the United States, for example, are drawn
into our models is strengthened. This design disproportionately from earlier time periods, al-
allows us to assess the strong claim, as recently though more recent data from the United States
 486

Table 55.2. Micro-Classes Nested in Manual-Nonmanual Classes, Macro Classes, and Meso Classes
1. NONMANUAL CLASS 2. MANUAL CLASS
1. Professional-managerial 2. Proprietors 3. Routine nonmanual 4. Manual 5. Primary
1. Classical professions 1. Proprietors 1. Sales 1. Craft Fisherman
1. Fishermen
1. Jurists 1. Real estate agents 1. Craftsmen, n.e.c 2. Farmers
2. Health professionals 2. Agents, n.e.c. 2. Foremen 3. Farm laborers
3. Professors and instructors 3. Insurance agents 3. Electronics service
and repair
4. Natural scientists 4. Cashiers 4. Printers and related
workers
5. Statistical and social scientists 5. Sales workers 5. Locomotive operators
6. Architects 2. Clerical 6. Electricians
7. Accountants 1. Telephone operators 7. Tailors and related workers
8. Authors and journalists 2. Bookkeepers 8. Vehicle mechanics
9. Engineers 3. Office workers 9. Blacksmiths and machinists
2. Managers and officials 4. Postal clerks 10. Jewelers
1. Officials, government and
non-profit organizations 11. Other mechanics
2. Other managers 12. Plumbers and pipe-fitters
3. Commercial managers 13. Cabinetmakers
4. Building managers and proprietors 14. Bakers
3. Other professions 15. Welders
1. Systems analysts and programmers 16. Painters
2. Aircraft pilots and navigators 17. Butchers
3. Personnel and labor relations workers 18. Stationary engine operators
4. Elementary and secondary teachers 19. Bricklayers and carpenters
5. Librarians 20. Heavy machine operators
(continues)
Table 55.2. (continued)
1. NONMANUAL CLASS 2. MANUAL CLASS
1. Professional-managerial 2. Proprietors 3. Routine nonmanual 4. Manual 5. Primary
3. Other professions (continued) 2. Lower manual
6. Creative artists 1. Truck drivers
7. Ship officers 2. Chemical processors
8. Professional and technical, n.e.c. 3. Miners and related workers
9. Social and welfare workers 4. Longshoremen
10. Workers in religion 5. Food processing workers
11. Nonmedical technicians 6. Textile workers
12. Health semiprofessionals 7. Sawyers
13. Hospital attendants 8. Metal processors
14. Nursery school teachers and aides 9. Operatives and kindred, n.e.c.
10. Forestry workers
3. Service workers
1. Protective service workers
2. Transport conductors
3. Guards and watchmen
4. Food service workers
5. Mass transportation operators
6. Service workers, n.e.c.
7. Hairdressers
8. Newsboys and deliverymen
9. Launderers
10. Housekeeping workers
11. Janitors and cleaners
12. Gardeners
487
 488

Table 55.3. Surveys for Intergenerational Mobility Analysis

Survey Period Ages Birth Cohorts Occup. 1 Sample1Size

Scheme
Sample Size
1. Occupational Changes
in a Generation I (OCG I) 1962 30–64 1898–1932 1960 SOC 17,544
2. Occupational Changes
in a Generation II (OCG II) 1973 30-64 1909–1943 1960–70 SOC 18,856
3. General Social Survey (GSS) 1972–2003 30–64 1908–1970 1970–80 SOC 9,685
4. Survey of Social Stratification
& Mobility (SSM) 1955–1995 30–64 1891–1970 Japanese SCO 6,703
5. Japan General Social Survey
(JGSS) 2000–2002 30–64 1936–1972 Japanese SCO 1,917
6. German Social Survey2 (ALLBUS) 1980–2002 30–64 1916–1972 ISCO-68, ISCO-88 5,647
7. German Socioeconomic Panel
(GSOEP) 1986, 1999, 2000 30–64 1922–1970 ISCO-68, ISCO-88 2,886
8. German Life History Study
LV I-III 1981–1989
1981-1989 30–64 1921–1959 ISCO-68 1,234
9. ZUMA-Standarddemographie
Survey 1976–1982 30–64
0–64 1912–1952 ISCO-68 2,929
10. 1990 Swedish Census (linked
to 1960 and 1970 Censuses) 1990 30–47 1943–1960 NYK80 184,451
1
SOC = Standard Occupational Classification; SCO = Standard Classification of Occupations; ISCO = International Standard Classification of Occupations; NYK = Nor-
disk yrkesklassificering.
2
German data exclude respondents from East Germany (GDR). If a respondent was not gainfully employed at the time of the survey, his last occupation was used.
 Jonsson et al.—It’s a Decent Bet That Our Children Will Be Professors Too • 489

are used as well (see Table 55.3 for details). Ad- next identify three “macroclasses” in the non-
ditionally, the Swedish data set covers only re- manual category (i.e., professional-managerial,
spondents between thirty and forty-seven years proprietor, routine nonmanual) and another
old, whereas all other data sets cover respondents two macroclasses in the manual category (i.e.,
between thirty and sixty-four years old. We have manual, primary). Within three of these mac-
elsewhere shown that such differences in cover- roclasses, we then allow further “mesoclass” dis-
age don’t affect our results much (Jonsson et al. tinctions to emerge: the professional-managerial
2009). class is divided into classical professions, manag-
The starting point for all of our analyses is ers and officials, and other professions; the rou-
the detailed microclass coding scheme repre- tine nonmanual class is divided into sales workers
sented in Table 55.2. The microclass category and clerks; and the manual class is divided into
may be defined as “a grouping of technically craft, lower manual, and service workers. The re-
similar jobs that is institutionalized in the labor sulting scheme, which embodies three layers of
market through such means as (a) an association big-class distinctions (i.e., manual-nonmanual,
or union, (b) licensing or certification require- macroclass, and mesoclass), may be understood
ments, or (c) widely diffused understandings . . . as a nondenominational hybrid of conventional
regarding efficient or otherwise preferred ways schemes that assembles in one scheme many of
of organizing production and dividing labor” the contrasts that have historically been empha-
(Grusky 2005, 66). The scheme used here in- sized by big-class scholars.
cludes eighty-two microclasses and captures These distinctions are introduced in our
many of the boundaries in the division of labor mobility models as a nested set of contrasts
that are socially recognized and defended. These (see Jonsson et al. 2009). This approach not
microclasses were then scaled in terms of the in- only allows us to tease out the net residue of
ternational socioeconomic scale (Ganzeboom, reproduction at the mesoclass, macroclass, and
de Graaf, and Treiman 1992). We have applied manual-nonmanual levels but also allows for
this scheme to model an 82 × 82 mobility table patterns of exchange that are more complicated
formed by cross-classifying the father’s and off- than those conventionally allowed. The stylized
spring’s occupation in data pooled from the parent-to-child mobility table in Figure 55.2
United States, Sweden, Germany, and Japan (for depicts these three sets of overlapping big-class
details, see Jonsson et al. 2009). The distinctive parameters and shows how they capture quite
feature of the resulting analysis is that microclass complicated affinities off the microclass diag-
effects, represented on the main diagonal of Fig- onal, off the mesoclass diagonal, and even off
ure 55.2, are layered over more conventional the macroclass diagonal. If we had instead pro-
big-class effects. ceeded by fitting mesoclass effects alone (as is
Given our suspicion that net big-class effects conventional), we could absorb excess densities
may be weak, it is clearly important to adopt a in the dark-gray regions of Figure 55.2 but not
big-class scheme that fully captures such big-class the surrounding light-gray regions. The cells in
effects as can be found, as otherwise any possible the white zones of Figure 55.2 are in fact the
shortfall in big-class explanatory power might be only ones that index mobility with respect to
attributed to poor operationalization. We have all class levels. Moreover, even the cells in these
accordingly proceeded by fitting a set of nested zones will be modeled with a gradational term,
big-class contrasts that capture the many and a parameter that allows us to estimate the extent
varied big-class distinctions that scholars have to which short-distance moves occur more fre-
identified. As shown in Table 55.2, we begin quently than long-distance ones.
by distinguishing the manual and nonmanual This gradational term captures the tendency
classes, a big-class distinction so important that of children to assume occupations that are so-
early class scholars often focused on it alone. We cioeconomically close to their origins. If the
490 • Part VI: Who Gets Ahead?

Figure 55.2. Overlapping Inheritance Terms in Mobility Model

Note: The Y axis pertains to occupational origins and the X axis to occupational destina-
tions. The unlabeled microdiagonal squares represent occupational immobility. The size of
each big-class category represents the number of microclass categories it encompasses (not
the number of workers within the class).

apparent clustering at the microclass, mesoclass, socioeconomic effect, μi (origin) and μj (desti-
macroclass, or manual-nonmanual levels reflects nation) are socioeconomic scale values assigned
nothing more than this gradational tendency, to each of the eighty-two microclasses, and įA,
then the inheritance parameters represented in įB, įC, and įM refer to manual-nonmanual,
Figure 55.2 will become insignificant when the macroclass, mesoclass, and microclass immo-
gradational parameter is included. The big-class bility effects, respectively. The latter parameters
and microclass parameters, taken together, thus are fitted simultaneously and therefore capture
speak to the extent to which the mobility regime net effects. The manual-nonmanual parameter,
is lumpy rather than gradational, while the rel- for example, indexes the average density across
ative size of these parameters speaks to whether those cells pertaining to manual or nonmanual
conventional big-class analyses have correctly inheritance after purging the additional residue
represented the main type of lumpiness. The of inheritance that may obtain at the macroclass,
following model is therefore yielded: mesoclass, and microclass levels.

The Structure of Mobility

where i indexes origins, j indexes destinations, When this model is applied to our pooled
mij refers to the expected value in the ijth cell, Į four-nation sample, the microclass and big-class
refers to the main effect, ȕi and Ȗj refer to row parameters take on the form represented in Fig-
and column marginal effects, ij refers to the ure 55.3. Although cross-national variations of
 Jonsson et al.—It’s a Decent Bet That Our Children Will Be Professors Too • 491

Figure 55.3. The Contours of Class Reproduction for Men

Note: The base indexes occupational origins and destinations, while the vertical dimension indexes den-
sities of mobility and immobility (for each possible combination of origin and destination). 1 = classical
professions; 2 = managers and officials; 3 = other professions; 4 = proprietors; 5 = sales; 6 = clerical; 7 =
craft; 8 = lower manual; 9 = service; 10 = primary sector.

interest have emerged in our analyses (see Jonsson coefficients, the two largest are for proprietors
et al. 2009), Figure 55.3 is based on pooled data and primary-sector workers, but even these two
that smooth out such variation and thus represent are smaller than all but the very smallest micro-
the cross-nationally shared features of mobility. class coefficients. It also bears noting that both
The most striking feature of this figure is the of these big classes are big classes in name only.
microdiagonal clustering that appears as a pali- That is, because the proprietor class comprises
sade protecting occupational positions from in- only shopkeepers, it is not the usual amalgam
truders. This palisade bespeaks very substantial of many occupations, and there is accordingly
departures from equality of opportunity. For good reason to regard proprietors as effectively
example, children born into the classical pro- a microclass. Likewise, the primary sector is
fessions are, on average, 4.2 times more likely not much of an amalgam, dominated as it is by
to remain in their microclass of origin than to farmers (see Table 55.2). The remaining twelve
move elsewhere within their mesoclass, while big-class effects, all of which pertain to true amal-
the corresponding coefficients for children born gams, are comparatively weak. The strongest of
into managerial, craft, and service occupations these effects, those for classical professions, sales
are 4.6, 7.9, and 5.6, respectively. Although the work, clerical work, and the manual-nonmanual
interior regions of the class structure are typically strata, range in size from 1.3 to 1.4 (in multipli-
represented as zones of fluidity (e.g., Featherman cative form).
and Hauser 1978), we find here substantial mi-
croclass reproduction throughout the class struc-
ture, even among the “middle classes.” Is Big-Class Reproduction a Myth?
How do the microclass and big-class co- The foregoing results raise the possibility that the
efficients compare? Of the fourteen big-class big-class inheritance showing up in generations
492 • Part VI: Who Gets Ahead?

Figure 55.4. Immobility Coefficients With and Without Microclass Controls

Note: For convenience in presentation, the two primary-sector coefficients are each divided by two.

of mobility studies is largely microclass inher- have motivated generations of mobility scholars
itance in disguise. Have conventional mobility to regard big-class reproduction as a powerful
studies indeed created the false impression that force.
big-class reproduction is the dominant form The results from our full model nonetheless
of reproduction? We can address this question imply that this conclusion is somewhat mis-
by examining whether the big-class effects that leading. When microclass effects are allowed,
appear in conventional mobility analyses are some of the big-class effects are greatly reduced
much reduced in size when microclass effects are in strength (i.e., classical professions, sales, cler-
overlaid on them. It’s useful to proceed by reesti- ical), while others disappear altogether or be-
mating our model after omitting the microclass come quite small (i.e., managers and officials,
inheritance terms. The relevant estimates from other professionals, craft workers, service work-
this trimmed model, which represents a con- ers, lower manual workers). It follows that con-
ventional big-class analysis, are shown in Figure ventional big-class analyses have generated the
55.4. appearance of big-class reproduction because it is
We begin by noting that the mesoclass effects confounded with microclass reproduction. This
under this trimmed model are indeed strong is not to suggest that all big-class reproduction is
and consistent with the effects secured in con- just microclass reproduction in disguise. Clearly,
ventional mobility analyses. The coefficient for some big-class reproduction persists even in the
managers, for example, implies that children presence of microclass controls, a result that was
born into the managerial class are 1.62 times also revealed in Figure 55.3.
more likely to remain in that class than to exit We may conclude on the basis of these re-
it (i.e., e.48 % 1.62). The corresponding inheri- sults that the big-class reproduction appearing in
tance coefficients for craft workers, lower manual conventional analyses is largely generated by the
workers, and service workers are 1.40, 1.63, and tendency for children to inherit their microclass.
1.93, respectively. It is coefficients such as these, The practical implication of this result is that
all of which are net of gradational effects, that big-class reproduction may not be easily reduced
 Jonsson et al.—It’s a Decent Bet That Our Children Will Be Professors Too • 493

without interventions that take on inheritance at There are two main ways in which con-
the occupational level. We return to this issue in ventional models misrepresent the structure of
the concluding discussion. opportunity: (1) The most extreme pockets of ri-
gidity are concealed when analysis is carried out
exclusively at the big-class level, and (2) the main
Conclusion rigidities in the big-class mobility table have been
The main intellectual backdrop to this analy- taken as evidence of big-class reproduction when
sis is the ongoing sociological debate about the in fact occupational reproduction is the princi-
types of social groupings that have taken hold in pal underlying mechanism. These results suggest
contemporary industrialism. Throughout much that the big-class mobility table, long a fixture in
of the twentieth century, sociologists were fasci- the discipline, obscures important mechanisms
nated, arguably obsessed, with theorizing about behind intergenerational reproduction.
the conditions under which big classes might Why are occupations such an important
form, an understandable fascination insofar as conduit for social reproduction? In all countries,
individual life chances and even collective out- parents accumulate much occupation-specific
comes (e.g., revolutions) were believed to de- capital, identify with their occupation, and ac-
pend on class processes. At the same time, class cordingly “bring home” their occupation in ways,
analysts viewed occupations as mere technical both direct and indirect, that then make it salient
positions in the division of labor (rather than to their children and lead them to invest in it. It
meaningful social groups), while scholars in follows that children develop a taste for occupa-
the occupations and professions of literature tional reproduction, are trained by their parents
focused narrowly on individual occupations in occupation-specific skills, have access to oc-
and how they developed under conditions of cupational networks that facilitate occupational
professionalization or proletarianization. The reproduction, and use those skills and networks
occupational form was not understood within to acquire more occupation-specific training
either of these traditions as a critical source of outside the home. If children are risk averse and
inequality and social reproduction (see Grusky oriented principally to avoiding downward mo-
2005). At best, occupations were described as bility, the safest path to realizing this objective
the “backbone” of the inequality system (e.g., may well be to use their occupation-specific re-
Parkin 1971), but such a characterization served sources on behalf of occupational reproduction.
principally as an impetus for then reducing oc- Indeed, even in the absence of any intrinsic
cupations to gradational scores (e.g., Hauser interest in occupational reproduction, children
and Warren 1997; Ganzeboom, de Graaf, and may still pursue it because it is the best route
Treiman 1992) or using them as aggregates to big-class reproduction (Erikson and Jonsson
in constructing big classes (e.g., Erikson and 1996). The son of an embalmer, for example,
Goldthorpe 1992). may not have any particular interest in becoming
These characteristic representations of the an embalmer but may decide it’s foolhardy to fail
form of mobility have been treated as assump- to exploit the in-house training that is available
tions rather than amenable to evidence. The to him.
main objective of our research has been to con- It might be tempting to take the position that
sider whether, when treated as empirical matters, the extreme microclass inequalities uncovered
these conventional representations of the struc- here are not all that objectionable. Should we re-
ture of mobility are incomplete. We have found ally care, for example, that the child of the truck
that occupations are an important conduit for driver has a special propensity to become a truck
reproduction and that incorporating this con- driver while the child of a gardener has a special
duit into mobility models can improve our un- propensity to become a gardener? Must we truly
derstanding of the mobility process. commit ourselves to equal access to truck driving
494 • Part VI: Who Gets Ahead?

and gardening? If pressed, we would argue that provide some insight, then, into why contempo-
all ascriptive constraints on choice, even those rary efforts to equalize opportunity have under-
pertaining to purely horizontal inequalities, are performed, they do not necessarily lead us to any
inconsistent with a commitment to an open wholesale rethinking of those efforts.
society. By this logic, all types of origin-by-des-
tination association are problematic because NOTES
they imply that human choice has been circum- 1. We will often refer to occupations as “micro-
classes” because they have many of the features and
scribed, a circumscription that is wholly deter-
characteristics that are often attributed to big classes.
mined by the accident of birth. We care, in other
words, that the truck driver is fated to become a
REFERENCES
truck driver at birth because that amounts to a Carlsson, G. 1958. Social Mobility and Class Struc-
stripping away of choice, and most of us would ture. Lund: Gleerups.
embrace an open society in which choices are Erikson, Robert, and John H. Goldthorpe. 1992.
expanded, not stripped away. Although our illus- The Constant Flux: A Study of Class Mobility in
trative nonchoice (i.e., being a truck driver ver- Industrial Societies. Oxford: Clarendon Press.
sus being a gardener) may not have implications Erikson, Robert, and Jan O. Jonsson. 1993. Ur-
for total rewards (of the sort that are consensually sprung och utbildning. SOU 1993:85. Stockholm:
Fritzes.
valued), it is nonetheless a fateful nonchoice that
———. 1996. “Explaining Class Inequality in Ed-
determines the texture and content of a human
ucation: The Swedish Test Case.” In Can Educa-
life. It is this commitment to an open society, tion Be Equalized? edited by Robert Erikson and
sometimes left quite implicit, that underlies the Jan O. Jonsson, 1–64. Boulder: Westview.
discipline’s long-standing interest in monitoring Featherman, David L., and Robert M. Hauser. 1978.
marital homogamy, occupational sex segrega- Opportunity and Change. New York: Academic
tion, and many other forms of ascription that Press.
are hybrids of vertical and horizontal processes. Ganzeboom, Harry B. G., Paul de Graaf, and Don-
It bears emphasizing, however, that such ald J. Treiman. 1992. “A Standard International
Socio-Economic Index of Occupational Status.”
an argument need not be pursued in the pres-
Social Science Research 21: 1–56.
ent case, given that the horizontal inequalities
Glass, D. V. 1954. Social Mobility in Britain. Lon-
uncovered here contribute directly to the per- don: Routledge and Kegan Paul.
petuation of vertical ones. That is, we should Goldthorpe, John H. 2007. On Sociology. Stanford:
care about the immobility of truck drivers and Stanford University Press.
gardeners not just because truck driving and Grusky, David B. 2005. “Foundations of a
gardening imply different styles of life (i.e., “hor- Neo-Durkheimian Class Analysis.” In Approaches
izontal” inequality), but also because microclass to Class Analysis, edited by Erik Olin Wright,
immobility of this sort is the principal mecha- 51–81. Cambridge: Cambridge University Press.
Grusky, David B., Yoshimichi Sato, Jan O. Jonsson,
nism ensuring that the working class reproduces
Satoshi Miwa, Matthew Di Carlo, Reinhard Pol-
itself. The results from our models make it clear
lak, and Mary C. Brinton. 2008. “Social Mobil-
that big-class reproduction arises largely because ity in Japan: A New Approach to Modeling Trend
children frequently remain within their micro- in Mobility.” In Intergenerational Mobility and
class of origin. Intragenerational Mobility, edited by Tsutomu
We are left with the conclusion that, insofar Watanabe, 3:1–25. Sendai, Japan: SSM Research
as microclass reproduction could be eliminated, Project Series.
real declines in big-class reproduction would be Hauser, Robert M., and John R. Warren. 1997. “So-
observed. It is troubling in this regard that mi- cioeconomic Indexes for Occupations: A Review,
croclass reproduction is deeply rooted in family Update, and Critique.” Sociological Methods 27:
177–298.
dynamics and may require unacceptably intru-
Hout, Michael, and Robert M. Hauser. 1992. “Hier-
sive policy to root it out. Although our results archy and Symmetry in Occupational Mobility.”
 Jonsson et al.—It’s a Decent Bet That Our Children Will Be Professors Too • 495

European Sociological Review 8 (December): edited by Robert Erikson, Markus Jannti, and
239–266. Timothy Smeeding. New York: Russell Sage
Jonsson, Jan O., David B. Grusky, Matthew Di Foundation.
Carlo, Reinhard Pollak, and Mary C. Brinton. Parkin, Frank. 1971. Class Inequality and Political
2009. “Micro-Class Mobility: Social Reproduc- Order: Social Stratification in Capitalist and Com-
tion in Four Countries.” American Journal of So- munist Societies. New York: Praeger.
ciology (January). Svalastoga, Kaare. 1959. Prestige, Class, and Mobility.
Jonsson, Jan O., David B. Grusky, Reinhard Pollak, Copenhagen: Gyldendal.
and Matthew Di Carlo. 2011. “Occupations Weeden, Kim A., and David B. Grusky. 2005. “The
and Social Mobility: Gradational, Big Class, and Case for a New Class Map.” American Journal of
Micro-Class Reproduction in Comparative Per- Sociology 111: 141–212.
spective.” In Persistence, Privilege, and Parenting,
496 • Part VI: Who Gets Ahead?

Income Mobility

56 • Gary Solon

Intergenerational Income Mobility

Intergenerational income mobility is the degree causal processes underlying the observed extent
to which position in the income distribution of intergenerational mobility.
persists or changes from one generation to the
next.
For example, a society in which individuals’ Theory
adult income is altogether independent of their The classic theoretical analysis by Becker and
parents’ income is a highly mobile society. A Tomes (1979) encompasses a multitude of rea-
society in which one’s percentile in the income sons that relative income status is correlated
distribution is surely identical to one’s parents’ across generations. In a recent variant of that
percentile is completely immobile. Neither ex- analysis, Solon (2004) adopts the functional
treme is ideal, and neither corresponds to the form assumptions consistent with the log-linear
intergenerational patterns typically observed in intergenerational mobility regression commonly
actual societies. Which intergenerational associ- estimated by empirical researchers. In this model,
ation between the extremes is desirable depends an individual parent divides her income between
on the processes generating it. In any case, rea- her own consumption and investment in an in-
sonable and well-informed observers are likely dividual child’s human capital so as to maximize
to disagree about the optimal level of intergen- a Cobb-Douglas utility function in which the
erational mobility because of differences in their two goods are the parent’s consumption and the
values, such as feelings about equity/efficiency child’s adult income. The mapping from the par-
tradeoffs. See Jencks and Tach (forthcoming) for ent’s investment in her child’s human capital to
a discussion of some of the normative issues. the child’s subsequent income as an adult oper-
Because understanding intergenerational ates through two functions.
mobility is important for understanding the First, a semi-logarithmic human capital
nature of income inequality, intergenerational production function relates the child’s level of
mobility has received a great deal of attention human capital to the logarithm of the sum of
from both theoretical and empirical researchers. the parent’s investment and public investment
Over the last two decades, empirical research de- (e.g., publicly supported education and health
scribing the extent of intergenerational mobil- care for children) plus a variable representing
ity has made considerable progress. Much work the human capital endowments children receive
remains to improve our understanding of the regardless of the investment choices of their

Gary Solon. “Intergenerational Income Mobility,” in The New Palgrave Dictionary of Economics, Second Edition,
edited by Steven N. Durlauf and Lawrence E. Blume. Published in 2008 by Palgrave Macmillan. Reproduced
with permission of Palgrave Macmillan.
 Solon—Intergenerational Income Mobility • 497

families and the government. These more me- and Ȝ is an index of the progressivity of public
chanically determined endowments to children investment in children’s human capital.
are, in Becker and Tomes’s words, “determined This result implies that the intergenerational
by the reputation and ‘connections’ of their income elasticity increases with the responsive-
families, the contribution to the ability, race, ness of earnings to human capital investment
and other characteristics of children from the and with the heritability of income-generating
genetic constitutions of their families, and the traits, and it decreases with the progressivity of
learning, skills, goals, and other ‘family com- public investment in children’s human capital.
modities’ acquired through belonging to a par- Cross-country differences in intergenerational
ticular family culture.” The transmission of these mobility could arise from differences in any of
endowments is assumed to follow a first-order these factors. So could changes over time in a
autoregressive process across generations. Thus, particular country’s intergenerational mobil-
intergenerational transmission occurs both be- ity. Finally, it can be shown that the same fac-
cause higher-income parents have greater where- tors that increase the intergenerational income
withal to invest in the human capital of their elasticity also increase the equilibrium level of
children and because of the genetic and cultural cross-sectional income inequality within a gen-
heritability of human capital. Second, the map- eration. Thus, we should not be surprised if so-
ping to the child’s income is completed by a cieties with particularly high income inequality
semi-logarithmic earnings function that relates also exhibit high intergenerational persistence of
the child’s log earnings to her level of human income status.
capital.
This simple model leaves out some impor-
tant aspects of intergenerational transmission. Empirical Evidence
For example, it assumes that the parent cannot If lifetime income data were available for both
borrow against the child’s prospective earnings the parents’ and children’s generations in a na-
and does not bequeath financial assets to the tionally representative sample, estimation of
child. See Becker and Tomes (1986) and Mul- the intergenerational income elasticity ȕ could
ligan (1997) for analyses that relax this assump- be performed simply by applying least squares
tion. Also, the model’s single-parent/single-child to the regression of the children’s log lifetime
structure ignores the role of assortative mating, income on the parents’ log lifetime income. In
which is discussed by Lam and Schoeni (1994) most countries, however, the ideal data are not
and Chadwick and Solon (2002). Nevertheless, available. As of the 1980s, data constraints had
the model is rich enough to illustrate some key forced most of the then-small empirical litera-
aspects of the intergenerational transmission ture to rely on short-term income measures,
process. These are embodied in the following such as annual income in only one year, for pe-
result concerning the intergenerational income culiarly homogeneous samples. As summarized
elasticity ȕ, which is the coefficient in the regres- in Becker and Tomes (1986), the resulting esti-
sion of the child’s log income on the parent’s log mates suggested that “a 10% increase in father’s
income: earnings (or income) raises son’s earnings by
less than 2%.” As discussed in detail in Solon
(1– ) +
⬵ (1) (1989, 1992), however, these estimates were bi-
1 + (1– ) ased substantially downward. The “right-sided”
measurement error from using short-term pa-
where ș is the elasticity of earnings with respect rental income measures to proxy for parents’
to human capital investment, Ȝ is the autore- lifetime income can serve as a good classroom
gressive parameter representing the genetic and example for the econometrics textbook analysis
cultural heritability of income-generating traits, of the attenuation bias resulting from “noisy”
498 • Part VI: Who Gets Ahead?

measurement of an explanatory variable. And data sets. And, as discussed in Reville (1995) and
when the estimates were based on relatively ho- Haider and Solon (2005), many of the newer es-
mogeneous parent samples, this bias was aggra- timates have been biased by “left-sided” measure-
vated by the diminished “signal variance” in the ment error. At the time researchers began to use
explanatory variable. intergenerational data from the PSID and NLS,
By the 1990s, empirical researchers in the the offspring were only about thirty years old or
United States had the benefit of better data. By even younger. For workers in their twenties, the
that time, two longitudinal surveys initiated in log of current income as a proxy for log lifetime
the late 1960s, the Panel Study of Income Dy- income is subject to “mean-reverting” measure-
namics (PSID) and the National Longitudinal ment error, instead of the classical measurement
Surveys (NLS) of labor market experience, had error typically analyzed in econometrics text-
generated new data with an intergenerational books. The mean-reversion occurs because the
span. Because these surveys used national prob- workers who eventually will have high lifetime
ability samples, they were less subject to the earnings typically experience steeper earnings
problems from homogeneous samples. And growth. As a result, the early-career gap in cur-
because the longitudinal surveys repeatedly col- rent earnings between workers with high and
lected income information at each reinterview, low lifetime earnings tends to understate their
they enabled exploration of the impact of using lifetime gap. This sort of mean-reverting mea-
longer-term income measures. Many of the new surement error in a dependent variable is still an-
studies, surveyed in Solon (1999), treated the other source of attenuation bias. Once all these
errors-in-variables issue by averaging the paren- downward biases in the estimation of the inter-
tal income measured over several years. A typical generational elasticity are considered, it becomes
finding was that, in a regression of son’s log earn- plausible that the intergenerational elasticity in
ings on a multi-year measure of father’s log earn- the United States may well be as large as 0.5 or
ings, the estimated slope coefficient was about 0.6.
0.4—that is, double the 0.2 value that previously In recent years, researchers have estimated
had been described as an upper bound for the intergenerational elasticities for many other
intergenerational elasticity. A few studies treated countries, sometimes with much larger sam-
the errors-in-variables problem by performing ples than are available from the U.S. surveys. As
instrumental variables estimation with parental summarized in Solon (2002), the elasticity esti-
characteristics like education or occupation used mates for the United States and United King-
to instrument for measured parental income. dom are towards the high end among developed
That approach usually produced somewhat countries, with considerably smaller estimates
higher intergenerational elasticity estimates, but, appearing for Canada, Sweden, Finland, and
as explained in Solon (1992), the consistency of Norway. Some new estimates for developing
such instrumental variables estimation depends countries in Latin America (Dunn, 2004; Fer-
on the “excludability” of the instruments from reira and Veloso, 2004; Grawe, 2004) are even
the model for children’s income. higher than the U.S. and U.K. estimates. By and
Even the new estimates based on multi-year large, these cross-country comparisons accord
parental income data probably were too low. with the theoretical prediction of greater in-
As emphasized in Solon (1992) and Mazumder tergenerational income persistence in countries
(2005), averaging parental income over several with greater income inequality, higher returns
years reduces but does not eliminate attenuation to human capital, and less progressive public in-
bias. vestment in children’s human capital. A related
Nonrandom attrition from the longitudinal question is whether the changes in income in-
surveys probably generated a weaker version of equality experienced by many countries over the
the sample homogeneity that had plagued earlier last quarter-century have been accompanied by
 Solon—Intergenerational Income Mobility • 499

changing intergenerational elasticities. In most NOTES

of the time-trends research conducted so far, the 1. A related literature (Solon, Page, and Duncan,
time spans and sample sizes have been too lim- 2000; Page and Solon, 2003a, 2003b; Oreopou-
los, 2003; Raaum, Salvanes, and Sorensen, 2006)
ited to permit strong conclusions.
has compared sibling correlations and correlations
Cross-country comparisons have only begun
among unrelated children who grew up in the same
to illuminate why intergenerational income as- neighborhood. The typical finding that the sibling
sociations are as large (and as small) as they are. correlations are considerably larger than the neigh-
To what extent does intergenerational transmis- bor correlations suggests that family influences loom
sion occur because higher-income parents invest larger than neighborhood influences in accounting
more in their children’s human capital? What for the effects of origins on socioeconomic outcomes.
are the roles of genetic and cultural heritability?
One intriguing line of research seeks clues from REFERENCES
comparisons of relatives with varying degrees of Becker, G.S. and Tomes, N. 1979. An equilibrium
theory of the distribution of income and inter-
genetic and environmental relatedness. Siblings
generational mobility. Journal of Political Econ-
studies of this type (Taubman, 1976; Bjork-
omy 87, 1153–1189.
lund, Jantti, and Solon, 2005) have compared ———. 1986. Human capital and the rise and fall
correlations in socioeconomic status among of families. Journal of Labor Economics 4, S1–S39.
monozygotic twins, dizygotic twins, non-twin Bjorklund, A., Jantti, M., and Solon, G. 2005.
full siblings, half-siblings, and biologically unre- Influences of nature and nurture on earnings
lated adoptive siblings and also have compared variation: a report on a study of various sibling
biological siblings reared together and apart.1 types in Sweden. In S. Bowles, H. Gintis, and
Intergenerational studies (Bjorklund, Lindahl, M. Osborne Groves, Unequal Chances: Family
Background and Economic Success, 145–164. New
and Plug, 2004; Plug, 2004; Sacerdote, 2004)
York: Russell Sage Foundation.
have compared parent-child outcome associa-
Bjorklund, A., Lindahl, M., and Plug, E. 2004. In-
tions in biological and adoptive families. Some tergenerational effects in Sweden: what can we
empirical patterns consistent with an impor- learn from adoption data? Discussion Paper No.
tant role for genetic transmission are: outcome 1194, IZA.
correlations are particularly high among mono- Chadwick, L. and Solon, G. 2002. Intergenerational
zygotic twins; correlations for dizygotic twins income mobility among daughters. American
and non-twin full siblings exceed those for Economic Review 92, 335–344.
half-siblings and adoptive siblings; correlations Dunn, C.E. 2004. The intergenerational transmis-
for biological siblings are positive even when sion of earnings: evidence from Brazil. Doctoral
dissertation, University of Michigan.
the siblings are reared apart; intergenerational
Ferreira, S.G. and Veloso, F.A. 2004. Intergenera-
associations are higher for biologically related tional mobility of wages in Brazil. Unpublished
parents and children; and adoptive children’s manuscript.
outcomes are positively associated with those of Grawe, N.D. 2004. Intergenerational mobility for
their biological parents (even after controlling whom? The experience of high- and low-earning
for the adoptive parents’ outcomes). Empirical sons in international perspective. In M. Corak
patterns consistent with important environmen- (ed.), Generational Income Mobility in North
tal factors are: outcome correlations are positive America and Europe, 58–89. Cambridge: Cam-
among biologically unrelated adoptive siblings; bridge University Press.
Haider, S. and Solon, G. 2005. Life-cycle variation
correlations among biological siblings tend to
in the association between current and lifetime
be higher when the siblings are reared together; earnings. Unpublished manuscript.
and adoptive children’s outcomes are positively Jencks, C. and Tach, L. Forthcoming. Would equal
associated with those of their adoptive parents opportunity mean more mobility? In S.L. Mor-
(even after controlling for the biological parents’ gan, D.B. Grusky, and G.S. Fields (eds.), Mobility
outcomes). and Inequality: Frontiers of Research from Sociology
500 • Part VI: Who Gets Ahead?

and Economics. Stanford: Stanford University Sacerdote, B. 2004. What happens when we ran-
Press. domly assign children to families? Working
Lam, D. and Schoeni, R.F. 1994. Family ties and Paper No. 10894, National Bureau of Economic
labor markets in the United States and Brazil. Research.
Journal of Human Resources 29, 1235–1258. Solon, G. 1989. Biases in the estimation of inter-
Mazumder, B. 2005. Fortunate sons: new estimates generational earnings correlations. Review of Eco-
of intergenerational mobility in the U.S. using nomics and Statistics 71, 172–174.
Social Security earnings data. Review of Economics ———. 1992. Intergenerational income mobility in
and Statistics 87, 235–255. the United States. American Economic Review 82,
Mulligan, C.B. 1997. Parental Priorities and Eco- 393–408.
nomic Inequality. Chicago: University of Chicago ———. 1999. Intergenerational mobility in the
Press. labor market. In O. C. Ashenfelter and D. Card
Oreopoulos, P. 2003. The long-run consequences of (eds.), Handbook of Labor Economics, vol. 3A,
living in a poor neighborhood. Quarterly Journal 1761–2000. Amsterdam: North-Holland.
of Economics 118, 1533–1575. ———. 2002. Cross-country differences in inter-
Page, M.E. and Solon, G. 2003a. Correlations be- generational earnings mobility. Journal of Eco-
tween brothers and neighboring boys in their nomic Perspectives 16, 59–66.
adult earnings: the importance of being urban. ———. 2004. A model of intergenerational mobil-
Journal of Labor Economics 21, 831–855. ity over time and place. In M. Corak (ed.), Gen-
———. 2003b. Correlations between sisters and erational Income Mobility in North America and
neighbouring girls in their subsequent income Europe, 38–47. Cambridge: Cambridge Univer-
as adults. Journal of Applied Econometrics 18, sity Press.
545–562. Solon, G., Page, M.E., and Duncan, G.J. 2000. Cor-
Plug, E. 2004. Estimating the effect of mother’s relations between neighboring children in their
schooling on children’s schooling using a sam- subsequent educational attainment. Review of
ple of adoptees. American Economic Review 94, Economics and Statistics 82, 383–392.
358–368. Taubman, P. 1976. The determinants of earnings:
Raaum, O., Salvanes, K., and Sorensen, E.O. 2006. genetics, family, and other environments; a study
The neighbourhood is not what it used to be. of white male twins. American Economic Review
Economic Journal. 66, 858–870.
Reville, R.T. 1995. Intertemporal and life cycle vari-
ation in measured intergenerational earnings mo-
bility. Unpublished manuscript.
 Ermisch et al.—Advantage in Comparative Perspective • 501

57 • John Ermisch, Markus Jäntti, Timothy Smeeding,

and James A. Wilson

Advantage in Comparative Perspective

Of all the potential consequences of rising eco- less chance of escaping poverty or low socioeco-
nomic inequality, none is more worrisome, or nomic status (SES).
more difficult to study, than the possibility that Investments in children are even broader than
rising inequality will have the long-term effect of this discussion suggests. An investment is a diver-
reducing equality of opportunity and intergener- sion of current resources, such as time or money,
ational mobility. The reasoning underlying this from use for immediate consumption of goods
worry is straightforward. Families clearly have a and services we value, to activities that pay off in
strong interest in investing in the future social the future in terms of additional resources, in-
and economic well-being of their children. Al- cluding those that benefit our children. A prime
though some of these investments may not re- example is of course education, but many activi-
quire financial resources, many others obviously ties that parents carry out on behalf of their chil-
do—among them, paying for quality child care dren are investments in a similar sense. Some of
and early childhood education, buying books them may involve a low monetary cost, but re-
and computers, living in higher-priced neigh- quire an investment of time, such as undertaking
borhoods with access to good public schools, many different types of activities with children
assisting with college costs, and providing sup- (for example, teaching them to swim or read-
port for young adults to help them get started in ing to them). In engaging in such activities, par-
their independent economic lives once their ed- ents increase their own enjoyment and current
ucation is completed. As financial resources have well-being as well as benefiting their children in
become more unequal in a number of countries later life. Other child-related activities can be
over the last three decades, the differences in the quite costly, such as paying university tuition.
capacities of rich and poor families to invest in And some activities may benefit children in a
their children also have become more unequal. different dimension than the initial investment.
This change is occurring in a period when rela- For example, in addition to aiding their cognitive
tively more educational investment is needed to development, parents and schools help socialize
meet ongoing labor market changes (Goldin and children, teach them to behave courteously, pro-
Katz 2008). It follows that unless these inequities vide motivations, and work in a variety of ways
are offset by public policies designed to moderate to aid their socioemotional development. These
their effects, the children of the rich will have traits may not only pay off in the social and be-
a relatively better chance of staying rich in the havioral dimension, but ready them for school
future, and the children of the poor will have so that their cognitive development is enhanced

John Ermisch, Markus Jäntti, Timothy Smeeding, and James A. Wilson. “Advantage in Comparative Perspec-
tive,” in From Parents to Children: The Intergenerational Transmission of Advantage (New York: Russell Sage
Foundation, 2012), pp. 3–7, 29–31. Copyright © 2012 by the Russell Sage Foundation, 112 East 64th Street,
New York, NY 10065. Reprinted with permission.
502 • Part VI: Who Gets Ahead?

as well. Social and behavioral traits may also be Of course, it is also possible that the pre-
more important for future earnings and jobs as diction that high inequality leads to low mo-
employers may highly value such traits. In eco- bility is simply wrong. But one compelling
nomic parlance, there is complementarity be- reason to doubt this is the recent discovery
tween investments in the social (socioemotional) that the predicted relationship does show up
and cognitive dimensions. in cross-national comparisons. Figure 57.1 pre-
Further, all such investments take place in in- sents the relationship between income inequal-
stitutional contexts that provide leeway for par- ity (measured by the Gini coefficient for the
ents and governments to influence how effective parents’ generation) and the intergenerational
such investments may be. For instance, universal income elasticity—a measure of the strength of
early childhood education for all children might the relationship between the incomes of parents
be especially beneficial for the lowest-SES chil- and the incomes of their grown children. Mo-
dren if all such programs had comparable re- bility is measured as the inverse of the elasticity
sources. However, to the extent that the quality in Figure 57.1, hence the lower the elasticity the
of preschools and teachers is subject to neigh- greater the mobility. Indeed, most measures of
borhood effects, as in elementary and secondary mobility are actually measures of persistence of
schooling in many nations, low-SES children are the younger generation’s place in the order of
likely to be excluded from the best preschools outcomes compared to their parents. So when
and thereby lessen the equalizing effect of early elasticities are high, the parent–adult child rela-
childhood education. Other childhood invest- tionship is strongest. This plot includes eleven
ments may also be subject to institutional con- industrialized countries where both measures are
straints, nepotism, ability to pay and co-funding, now available and demonstrates wide variance in
including tuition for colleges and universities intergenerational mobility across those countries
(for example, on U.S.-Canadian differences in (Björklund and Jäntti 2009).
financial aid and tertiary school completion, see As Figure 57.1 shows, the relationship be-
Belley, Frenette, and Lochner 2010). tween inequality and intergenerational elasticity
Although there is evidence that parental is moderately positive. Higher levels of inequality
investments in children have become more are associated with lower rates of mobility—the
unequal over the past thirty years in some coun- rank order correlation is 0.62. Although we can-
tries (Kaushal, Magnuson, and Waldfogel 2011), not lean too heavily on a regression based on only
analysis of the best multi-generational data avail- eleven data points, there are multiple estimates
able in the United States (from the Panel Study of both inequality and mobility rates in most of
of Income Dynamics) does not show a clear de- these nations, adding credence to the estimates
cline in intergenerational mobility between chil- shown in Figure 57.1 (Blanden 2011). What is
dren born in the 1950s and those born in the most interesting here is that these countries seem
late 1970s, just before inequality began to rise to vary a great deal in the degree to which they
(Lee and Solon 2009). Part of the problem may manage to attenuate the estimated relationship
be measurement error. The individuals in the between inequality and intergenerational mobil-
cohort born during the period of rising inequal- ity. Some countries lie alongside the least squares
ity are only in their early thirties, still a bit too regression line indicating levels of mobility close
young to provide reliable estimates of lifetime to what their levels of inequality might predict
income. Another possibility is that the gradual, (for example, Norway, Germany, and the United
thirty-year rise in inequality in the United States Kingdom). Sweden and Finland are low inequal-
and smaller increases elsewhere are still too small ity countries that lie slightly above the regression
to have the types of negative effects suggested by line, with slightly less mobility than their levels of
increased economic inequality. inequality predict. Denmark shows intermediate
 Ermisch et al.—Advantage in Comparative Perspective • 503

Figure 57.1. Estimates of Intergenerational Income Elasticities for Fathers and Sons, Early 1980s

Source: Authors’ calculations based on data from Björklund and Jäntti (2009, Figure 20.1).

levels of inequality but stands out with much social welfare, such as health or education. This
higher rates of mobility than expected. Canada could make a difference if expenditures in some
and Australia tend to fall between intermedi- areas are more effective than others in promoting
ate and high levels of inequality, but like Den- mobility. Finally, the effectiveness of institutions
mark, also show higher levels of mobility than designed to promote mobility may depend in
expected. A final group of countries (Italy, the part on the amount of inequality they have to
United States, and France) generally have high cope with. For example, a universal preschool
levels of inequality and lower levels of intergen- program may be effective in countries where
erational mobility than one would predict. differences in the private resources available to
If this pattern is real, and not just a matter families are modest. But where family differences
of random variation around the plotted regres- are great, they may swamp even a well-designed,
sion line, it suggests that there may be signifi- well-funded preschool program.
cant differences in the types and effectiveness of Inequality has increased over time relative to
public and private investments and institutions the levels of income inequality shown in Figure
that different countries deploy in their efforts to 57.1. This is the case in all the countries shown
equalize opportunities across the income distri- here, except for France, where it has fallen over
bution. These differences may be due to institu- the past twenty-five years (Smeeding, Erikson,
tional design. For example, some countries may and Jäntti 2011; Brandolini and Smeeding
intervene earlier in the lives of disadvantaged in- 2009). Most countries investigated here have
dividuals, and early intervention may be partic- higher inequality now than at any time in the
ularly effective, as many believe (Knudsen et al. past, but the rank order of countries by levels
2006). Or, countries may differ in the sheer size of income inequality is about the same now
of their social welfare expenditures or in the dis- as it was for prior generations. The amount of
tribution of expenditures across various areas of income available to low income families with
504 • Part VI: Who Gets Ahead?

children is also important in determining life one might expect growing economic inequality
chances as high child poverty means less paren- to affect intergenerational mobility.
tal economic resources. Child poverty rates for
these countries in the most recent Luxemburg NOTES
Income Study (LIS) year generally follow the 1. Based on LIS key figures, at http://www.lis-
datacenter.org/lis-ikf-webapp/app/search-ikf-figures
same patterns as do current measures of inequal-
(accessed November 28, 2011), and Gornick and
ity, where low-inequality nations in Scandina-
Jäntti (2010).
via have low child poverty rates (5 percent or
below at half the median income poverty level), REFERENCES
middle-inequality central European countries Belley, Philippe, Marc Frenette, and Lance Lochner.
have poverty rates of 10 to 11 percent, Australia 2010. “Post-Secondary Attendance by Paren-
and the United Kingdom are at about 14 per- tal Income: Comparing the U.S. and Canada.”
cent and the highest rates in Canada, Italy, and CIBC Centre for Human Capital and Produc-
the United States are 17 percent or above.1 This tivity Working Paper 2010-3. London, Canada:
makes it all the more interesting to know how University of Western Ontario.
Björklund, Anders, and Markus Jäntti. 2009. “In-
countries like Canada, with both above average
tergenerational Income Mobility and the Role of
inequality and above average child poverty rates,
Family Background.” In The Oxford Handbook of
do so well on mobility outcomes. If the ability Economic Inequality, edited by Wiemer Salverda,
to invest more in children increases among the Brian Nolan, and Timothy M. Smeeding. New
rich and declines among the poor as inequalities York: Oxford University Press.
increase, greater inequality may lead to even less Blanden, Jo. 2011. “Cross-Country Rankings in In-
mobility. If intergenerational mobility is driven tergenerational Mobility: A Comparison of Ap-
by cumulative forces of advantage and disadvan- proaches for Economics and Sociology.” Journal
tage over the life course, mobility outcomes may of Economic Surveys. DOI: 10.1111/j.467-6419.2
011.00690.X.
become worse for the current generation of chil-
Brandolini, Andrea, and Timothy M. Smeeding.
dren because of increasingly higher inequality
2009. “Income Inequality in Richer and OECD
(DiPrete and Eirich 2006). Countries.” In The Oxford Handbook of Economic
If the children of higher-SES parents do well Inequality, edited by Wiemer Salverda, Brian
in school, they are more likely to attend and Nolan, and Timothy Smeeding. New York: Ox-
graduate from college, are better able to rely ford University Press.
on parental help as they establish careers, and Cunha, Flavio, and James J. Heckman. 2009. “The
ultimately, will earn more income as the wage Economics and Psychology of Inequality and
distribution rewards higher-skilled workers with Human Development.” Journal of the European
better earnings. This pattern of development is Economic Association 7(2–3): 320–64.
DiPrete, Thomas A., and Gregory M. Eirich. 2006.
a form of “cumulative advantage” where success
“Cumulative Advantage as a Mechanism for In-
begets more success within generations. But equality: A Review of Theoretical and Empirical
there is also the possibility that greater inequality Developments.” Annual Review of Sociology 32:
increases the SES gradient from one generation 271–97.
to the next, and if so, this process of cumulative Goldin, Claudia, and Lawrence F. Katz. 2008. The
advantage might make an even larger difference Race Between Education and Technology. Cam-
across generations. Patterns of cumulative advan- bridge, Mass.: Harvard University Press.
tage within generations can be established only Gornick, Janet C., and Markus Jäntti. 2010. “Child
if SES-related differences are followed across Poverty in Upper-Income Countries: Lessons
from the Luxembourg Income Study.” In From
children’s lives. To establish between-generation
Child Welfare to Child Wellbeing: An Interna-
cumulative advantage requires observations from tional Perspective on Knowledge in the Service of
a minimum of three generations. These require- Making Policy, edited by Shelia B. Kamerman,
ments are generally beyond currently available Shelly Phipps, and Asher Ben-Arieh. New York:
data, but offer an appealing framework for how Springer.
 Ermisch et al.—Advantage in Comparative Perspective • 505

Kaushal, Neeraj, Katherine Magnuson, and Jane of the National Academy of the Sciences 103(27):
Waldfogel. 2011. “How Is Family Income Re- 10155–62. Available at: http://jenni.uchicago
lated to Investments in Children’s Learning?” In .edu/papers/Knudsenetal_PNAS_vl03n27_2006.
Whither Opportunity: Rising Inequality and the pdf (accessed November 29, 2011).
Uncertain Life Chances of Low-Income Children, Lee, Chulin, and Gary Solon. 2009. “Trends in In-
edited by Greg J. Duncan and Richard Murnane. tergenerational Income Mobility.” The Review of
New York: Russell Sage Foundation. Economics and Statistics 91(4): 766–72.
Knudsen, Eric I., James J. Heckman, Judy L. Cam- Smeeding, Timothy M., Robert Erikson, and Markus
eron, and Jack P. Shonkoff. 2006. “Economic, Jäntti. 2011. Persistence, Privilege, and Parenting:
Neurobiological, and Behavioral Perspectives on The Comparative Study of Intergenerational Mobil-
Building America’s Future Workforce.” Proceedings ity. New York: Russell Sage Foundation.
506 • Part VI: Who Gets Ahead?

Classic Models of Status Attainment

58 • Peter M. Blau and Otis Dudley Duncan,

with the collaboration of Andrea Tyree

The Process of Stratikcation

Stratification systems may be characterized in achievement. Some ascriptive features of the

various ways. Surely one of the most important system may be regarded as vestiges of an earlier
has to do with the processes by which individu- epoch, to be extirpated as rapidly as possible.
als become located, or locate themselves, in po- Public policy may emphasize measures designed
sitions in the hierarchy comprising the system. to enhance or to equalize opportunity—hope-
At one extreme we can imagine that the circum- fully, to overcome ascriptive obstacles to the full
stances of a person’s birth—including the per- exercise of the achievement principle.
son’s sex and the perfectly predictable sequence The question of how far a society may re-
of age levels through which he is destined to alistically aspire to go in this direction is hotly
pass—suffice to assign him unequivocally to a debated, not only in the ideological arena but
ranked status in a hierarchical system. At the in the academic forum as well. Our contribu-
opposite extreme his prospective adult status tion, if any, to the debate will consist largely in
would be wholly problematic and contingent submitting measurements and estimates of the
at the time of birth. Such status would become strength of ascriptive forces and of the scope of
entirely determinate only as adulthood was opportunities in a large contemporary society.
reached, and solely as a consequence of his own The problem of the relative importance of the
actions taken freely—that is, in the absence of two principles in a given system is ultimately
any constraint deriving from the circumstances a quantitative one. We have pushed our inge-
of his birth or rearing. Such a pure achievement nuity to its limit in seeking to contrive relevant
system is, of course, hypothetical, in much quantifications.
the same way that motion without friction is The governing conceptual scheme in the
a purely hypothetical possibility in the physi- analysis is quite a commonplace one. We think
cal world. Whenever the stratification system of of the individual’s life cycle as a sequence in time
any moderately large and complex society is de- that can be described, however partially and
scribed, it is seen to involve both ascriptive and crudely, by a set of classificatory or quantitative
achievement principles. measurements taken at successive stages. Ide-
In a liberal democratic society we think ally we should like to have under observation a
of the more basic principle as being that of cohort of births, following the individuals who

Peter M. Blau and Otis Dudley Duncan, with the collaboration of Andrea Tyree. “The Process of Stratifi-
cation,” in The American Occupational Structure, pp. 163–177, 199–201, 203. Copyright © 1967 by Peter
M. Blau and Otis Dudley Duncan. Reprinted with the permission of The Free Press, a division of Simon &
Schuster, Inc. All rights reserved.
 Blau and Duncan—The Process of Stratikcation • 507

make up the cohort as they pass through life. among these variables leads to a formulation of
As a practical matter we resorted to retrospective a basic model of the process of stratification.
questions put to a representative sample of sev-
eral adjacent cohorts so as to ascertain those facts
about their life histories that we assumed were A Basic Model
both relevant to our problem and accessible by To begin with, we examine only five variables.
this means of observation. For expository convenience, when it is necessary
Given this scheme, the questions we are con- to resort to symbols, we shall designate them
tinually raising in one form or another are: how by arbitrary letters but try to remind the reader
and to what degree do the circumstances of birth from time to time of what the letters stand for.
condition subsequent status? and, how does sta- These variables are:
tus attained (whether by ascription or achieve- V: Father’s educational attainment
ment) at one stage of the life cycle affect the X: Father’s occupational status
prospects for a subsequent stage? The questions U: Respondent’s educational attainment
are neither idle nor idiosyncratic ones. Current W: Status of respondent’s first job
policy discussion and action come to a focus in a Y: Status of respondent’s occupation in 1962
vaguely explicated notion of the “inheritance of
poverty.” Thus a spokesman for the Social Secu- Each of the three occupational statuses is
rity Administration writes: scaled by the [socioeconomic] index described
[elsewhere],3 ranging from 0 to 96. The two edu-
It would be one thing if poverty hit at ran- cation variables are scored on the following arbi-
dom and no one group were singled out. It is trary scale of values (“rungs” on the “educational
another thing to realize that some seem des- ladder”) corresponding to specified numbers of
tined to poverty almost from birth—by their years of formal schooling completed:
color or by the economic status or occupation 0: No school
of their parents.1 1: Elementary, one to four years
2: Elementary, five to seven years
Another officially sanctioned concept is that 3: Elementary, eight years
of the “dropout,” the person who fails to gradu- 4: High school, one to three years
ate from high school. Here the emphasis is not 5: High school, four years
so much on circumstances operative at birth but 6: College, one to three years
on the presumed effect of early achievement on 7: College, four years
subsequent opportunities. Thus the “dropout” is 8: College, five years or more (i.e., one or
seen as facing “a lifetime of uncertain employ- more years of postgraduate study)
ment,”2 probable assignment to jobs of inferior
status, reduced earning power, and vulnerability Actually, this scoring system hardly differs
to various forms of social pathology. from a simple linear transformation, or “cod-
In this study we do not have measurements ing,” of the exact number of years of school
on all the factors implicit in a full-blown con- completed. In retrospect, for reasons given [else-
ception of the “cycle of poverty” nor all those where],4 we feel that the score implies too great
variables conceivably responding unfavorably to a distance between intervals at the lower end of
the achievement of “dropout” status. . . . This the scale; but the resultant distortion is minor in
limitation, however, is not merely an analytical view of the very small proportions scored 0 or 1.
convenience. We think of the selected quantita- A basic assumption in our interpretation of
tive variables as being sufficient to describe the regression statistics—though not in their calcu-
major outlines of status changes in the life cycle lation as such—has to do with the causal or tem-
of a cohort. Thus a study of the relationships poral ordering of these variables. In terms of the
508 • Part VI: Who Gets Ahead?

father’s career we should naturally assume pre- schooling to enter regular employment. (These
cedence of V (education) with respect to X (oc- “inconsistent” responses include men giving 19
cupation when his son was 16 years old). We are as their age at first job and college graduation
not concerned with the father’s career, however, or more as their education; 17 or 18 with some
but only with his statuses that comprised a con- college or more; 14, 15, or 16 with high-school
figuration of background circumstances or ori- graduation or more; and under 14 with some
gin conditions for the cohorts of sons who were high school or more.) When the two variables
respondents in the Occupational Changes in a are studied in combination with occupation of
Generation (OCG) study. Hence we generally first job, a very clear effect is evident. Men with a
make no assumption as to the priority of V with given amount of education beginning their first
respect to X; in effect, we assume the measure- jobs early held lower occupational statuses than
ments on these variables to be contemporaneous those beginning at a normal or advanced age for
from the son’s viewpoint. The respondent’s ed- the specified amount of education.
ucation, U, is supposed to follow in time—and Despite the strong probability that the U-W
thus to be susceptible to causal influence from— sequence is reversed for an appreciable minority
the two measures of father’s status. Because we of respondents, we have hardly any alternative
ascertained X as of respondent’s age 16, it is true to the assumption made here. If the bulk of the
that some respondents may have completed men who interrupted schooling to take their first
school before the age to which X pertains. Such jobs were among those ultimately securing rela-
cases were doubtlessly a small minority and in tively advanced education, then our variable W
only a minor proportion of them could the fa- is downwardly biased, no doubt, as a measure
ther (or other family head) have changed status of their occupational status immediately after
radically in the two or three years before the re- they finally left school for good. In this sense,
spondent reached 16. the correlations between U and W and between
The next step in the sequence is more prob- W and Y are probably attenuated. Thus, if we
lematic. We assume that W (first job status) fol- had really measured “job after completing edu-
lows U (education). The assumption conforms cation” instead of “first job,” the former would
to the wording of the questionnaire, which stip- in all likelihood have loomed somewhat larger
ulated “the first full-time job you had after you as a variable intervening between education and
left school.” In the years since the OCG study 1962 occupational status. We do not wish to
was designed we have been made aware of a fact argue that our respondents erred in their reports
that should have been considered more carefully on first job. We are inclined to conclude that
in the design. Many students leave school more their reports were realistic enough, and that it
or less definitively, only to return, perhaps to a was our assumption about the meaning of the
different school, some years later, whereupon responses that proved to be fallible.
they often finish a degree program.5 The OCG The fundamental difficulty here is concep-
questionnaire contained information relevant to tual. If we insist on any uniform sequence of the
this problem, namely the item on age at first job. events involved in accomplishing the transition
Through an oversight no tabulations of this item to independent adult status, we do violence to
were made for the present study. Tables prepared reality. Completion of schooling, departure from
for another study6 using the OCG data, however, the parental home, entry into the labor market,
suggest that approximately one-eighth of the re- and contracting of a first marriage are crucial
spondents report a combination of age at first steps in this transition, which all normally occur
job and education that would be very improb- within a few short years. Yet they occur at no
able unless (a) they violated instructions by re- fixed ages nor in any fixed order. As soon as we
porting a part-time or school-vacation job as the aggregate individual data for analytical pur-
first job, or (b) they did, in fact, interrupt their poses we are forced into the use of simplifying
 Blau and Duncan—The Process of Stratikcation • 509

Table 58.1. Simple Correlations for Five Status Variables

p
Variable
Variable Y W U X V
Y: 1962 occ. status — .541 .596 .406 .322
W: First-job status — .538 .417 .332
U: Education — .438 .453
X: Father’s occ. status — .516
V: Father’s education —

assumptions. Our assumption here is, in effect, In proposing this sequence we do not over-
that “first job” has a uniform significance for look the possibility of what Carlsson calls “de-
all men in terms of its temporal relationship to layed effects,”8 meaning that an early variable
educational preparation and subsequent work may affect a later one not only via intervening
experience. If this assumption is not strictly variables but also directly (or perhaps through
correct, we doubt that it could be improved by variables not measured in the study).
substituting any other single measure of initial In translating this conceptual framework into
occupational status. (In designing the OCG quantitative estimates the first task is to establish
questionnaire, the alternative of “job at the time the pattern of associations between the variables
of first marriage” was entertained briefly but in the sequence. This is accomplished with the
dropped for the reason, among others, that un- correlation coefficient. Table 58.1 supplies the
married men would be excluded thereby.) correlation matrix on which much of the subse-
One other problem with the U-W transition quent analysis is based. In discussing causal in-
should be mentioned. Among the younger men terpretations of these correlations, we shall have
in the study, 20 to 24 years old, are many who to be clear about the distinction between two
have yet to finish their schooling or to take up points of view. On the one hand, the simple cor-
their first jobs or both—not to mention the men relation—given our assumption as to direction
in this age group missed by the survey on ac- of causation—measures the gross magnitude of
count of their military service.7 Unfortunately, the effect of the antecedent upon the consequent
an early decision on tabulation plans resulted in variable. Thus, if rYW = .541, we can say that an
the inclusion of the 20 to 24 group with the increment of one standard deviation in first job
older men in aggregate tables for men 20 to 64 status produces (whether directly or indirectly)
years old. We have ascertained that this results an increment of just over half of one standard
in only minor distortions by comparing a vari- deviation in 1962 occupational status. From
ety of data for men 20 to 64 and for those 25 another point of view we are more concerned
to 64 years of age. Once over the U-W hurdle, with net effects. If both first job and 1962 status
we see no serious objection to our assumption have a common antecedent cause—say, father’s
that both U and W precede Y, except in regard occupation—we may want to state what part of
to some fraction of the very young men just the effect of W on Y consists in a transmission
mentioned. of the prior influence of X. Or, thinking of X as
In summary, then, we take the somewhat ide- the initial cause, we may focus on the extent to
alized assumption of temporal order to represent which its influence on Y is transmitted by way of
an order of priority in a causal or processual se- its prior influence on W.
quence, which may be stated diagrammatically We may, then, devote a few remarks to the
as follows: pattern of gross effects before presenting the
apparatus that yields estimates of net direct and
(V,X) – (U) – (W) – (Y). indirect effects. Since we do not require a causal
510 • Part VI: Who Gets Ahead?

ordering of father’s education with respect to his generation, we should have to assume that for
occupation, we may be content simply to note the fathers, likewise, education and occupation
that rXV = .516 is somewhat lower than the cor- are correlated not only because one affects the
responding correlation, rYU = .596, observed other but also because common causes lie behind
for the respondents themselves. The difference both, which we have not measured. The bidirec-
suggests a heightening of the effect of education tional arrow merely serves to sum up all sources
on occupational status between the fathers’ and of correlation between V and X and to indicate
the sons’ generations. Before stressing this inter- that the explanation thereof is not part of the
pretation, however, we must remember that the problem at hand.
measurements of V and X do not pertain to some The straight lines running from one mea-
actual cohort of men, here designated “fathers.” sured variable to another represent direct (or net)
Each “father” is represented in the data in pro- influences. The symbol for the path coefficient,
portion to the number of his sons who were 20 such as PYW, carries a double subscript. The first
to 64 years old in March 1962. subscript is the variable at the head of the path,
The first recorded status of the son himself is or the effect; the second is the causal variable.
education (U). We note that rUV is just slightly (This resembles the convention for regression
greater than rUX. Apparently both measures on coefficients, where the first subscript refers to
the father represent factors that may influence the “dependent” variable, the second to the “in-
the son’s education. dependent” variable.)
In terms of gross effects there is a clear order- Finally, we see lines with no source indicated
ing of influences on first job. Thus rWU > rWX carrying arrows to each of the effect variables.
> rWV. Education is most strongly correlated These represent the residual paths, standing for
with first job, followed by father’s occupation, all other influences on the variable in question,
and then by father’s education. including causes not recognized or measured, er-
Occupational status in 1962 (Y ) apparently rors of measurement, and departures of the true
is influenced more strongly by education than by relationships from additivity and linearity, prop-
first job; but our earlier discussion of the first-job erties that are assumed throughout the analysis.
measure suggests we should not overemphasize An important feature of this kind of causal
the difference between rYW and rYU. Each, scheme is that variables recognized as effects of
however, is substantially greater than rYX, which certain antecedent factors may, in turn, serve as
in turn is rather more impressive than rYV. causes for subsequent variables. For example, U
Figure 58.1 is a graphic representation of the is caused by V and X, but it in turn influences
system of relationships among the five variables W and Y. The algebraic representation of the
that we propose as our basic model. The
numbers entered on the diagram, with the Figure 58.1. Path Coefficients in Basic Model of the
exception of rXV, are path coefficients, Process of Stratification
the estimation of which will be explained
shortly. First we must become familiar with
the conventions followed in constructing
this kind of diagram. The link between V
and X is shown as a curved line with an
arrowhead at both ends. This is to distin-
guish it from the other lines, which are
taken to be paths of influence. In the case
of V and X we may suspect an influence
running from the former to the latter. But
if the diagram is logical for the respondent’s
 Blau and Duncan—The Process of Stratikcation • 511

scheme is a system of equations, rather than the partial exception to the rule that all causes must
single equation more often employed in multiple be explicitly represented in the diagram is the
regression analysis. This feature permits a flexi- unmeasured variable that can be assumed to op-
ble conceptualization of the modus operandi of erate strictly as an intervening variable. Its inclu-
the causal network. Note that Y is shown here sion would enrich our understanding of a causal
as being influenced directly by W, U, and X, but system without invalidating the causal scheme
not by V (an assumption that will be justified that omits it. Sociologists have only recently
shortly). But this does not imply that V has no begun to appreciate how stringent are the logical
influence on Y. V affects U, which does affect Y requirements that must be met if discussion of
both directly and indirectly (via W ). Moreover, causal processes is to go beyond mere impres-
V is correlated with X, and thus shares in the sionism and vague verbal formulations.9 We are
gross effect of X on Y, which is partly direct and a long way from being able to make causal infer-
partly indirect. Hence the gross effect of V on Y, ences with confidence, and schemes of the kind
previously described in terms of the correlation presented here had best be regarded as crude first
rYV, is here interpreted as being entirely indi- approximations to adequate causal models.
rect, in consequence of V ’s effect on intervening On the empirical side, a minimum test of the
variables and its correlation with another cause adequacy of a causal diagram is whether it satis-
of Y. factorily accounts for the observed correlations
among the measured variables. In making such
a test we employ the fundamental theorem in
Path Coefficients path analysis, which shows how to obtain the
Whether a path diagram, or the causal scheme it correlation between any two variables in the sys-
represents, is adequate depends on both theoreti- tem, given the ipath coefficients and correlations
cal and empirical considerations. At a minimum, entered on the diagram.10 Without stating this
before constructing the diagram we must know, theorem in general form, we may illustrate its
or be willing to assume, a causal ordering of the application here. For example,
observed variables (hence the lengthy discussion
of this matter earlier in this chapter). This in- rYX = pYX + pYU rUX + pYW rWX
formation is external or a priori with respect to
the data, which merely describe associations or and
correlations. Moreover, the causal scheme must
be complete, in the sense that all causes are ac- rWX = pWX + pWU rUX.
counted for. Here, as in most problems involv-
ing analysis of observational data, we achieve We make use of each path leading to a given
a formal completeness of the scheme by repre- variable (such as Y in the first example) and the
senting unmeasured causes as a residual factor, correlations of each of its causes with all other
presumed to be uncorrelated with the remaining variables in the system. The latter correlations,
factors lying behind the variable in question. If in turn, may be analyzed; for example, rWX,
any factor is known or presumed to operate in which appeared as such in the first equation, is
some other way it must be represented in the broken down into two parts in the second. A
diagram in accordance with its causal role, even complete expansion along these lines is required
though it is not measured. Sometimes it is pos- to trace out all the indirect connections between
sible to deduce interesting implications from the variables; thus,
inclusion of such a variable and to secure use-
ful estimates of certain paths in the absence of rYX = pYX + pYU pUX + pYU pUV rVX + pYW pWX
measurements on it, but this is not always so. A + pYW pWU pUX + pYW pWU pUV rVX .
512 • Part VI: Who Gets Ahead?

Now, if the path coefficients are properly es- regressions of both W and Y on V were so small
timated, and if there is no inconsistency in the as to be negligible. Hence V could be disregarded
diagram, the correlations calculated by a formula as a direct influence on these variables without
like the foregoing must equal the observed cor- loss of information. The net regression of Y on X
relations. Let us compare the values computed was likewise small but, as it appears, not entirely
from such a formula with the corresponding ob- negligible. Curiously, this net regression is of the
served correlations: same order of magnitude as the proportion of
occupational inheritance in this population—
rWV = pWX rXV + pWU rUV about 10 percent, as discussed [elsewhere].12 We
might speculate that the direct effect of the fa-
= (.224)(.516) + (.440)(.453) ther’s occupation on the occupational status of a
mature man consists of this modest amount of
= .116 + .199 = .315, strict occupational inheritance. The remainder
of the effect of X on Y is indirect, inasmuch as
which compares with the observed value of .332; X has previously influenced U and W, the son’s
and education and the occupational level at which
he got his start. For reasons noted [elsewhere]13
rYV = pYU rUV + pYX rXV + pYW rWV we do not assume that the full impact of the
tendency to take up the father’s occupation is
= (.394)(.453) + (.115)(.516) + registered in the choice of first job.
(.281)(.315) With the formal properties of the model in
mind we may turn to some general problems
= .326 confronting this kind of interpretation of our
results. One of the first impressions gained from
(using here the calculated rather than the ob- Figure 58.1 is that the largest path coefficients
served value of rWV), which resembles the actual in the diagram are those for residual factors, that
value, .322. Other such comparisons—for rYX, is, variables not measured. The residual path is
for example—reveal, at most, trivial discrepan- merely a convenient representation of the extent
cies (no larger than .001). to which measured causes in the system fail to
We arrive, by this roundabout journey, at the account for the variation in the effect variables.
problem of getting numerical values for the path The residual is obtained from the coefficient of
coefficients in the first place. This involves using determination; if R2Y(WUX) is the squared multiple
equations of the foregoing type inversely. We correlation of Y on the three independent vari-
have illustrated how to obtain correlations if the ables, then the residual for Y is 1-R 2Y (WUX ) .
path coefficients are known, but in the typical Sociologists are often disappointed in the size
empirical problem, we know the correlations (or of the residual, assuming that this is a measure
at least some of them) and have to estimate the of their success in “explaining” the phenome-
paths. For a diagram of the type of Figure 58.1 non under study. They seldom reflect on what
the solution involves equations of the same form it would mean to live in a society where nearly
as those of linear multiple regression, except perfect explanation of the dependent variable
that we work with a recursive system of regres- could be secured by studying causal variables like
sion equations11 rather than a single regression father’s occupation or respondent’s education. In
equation. such a society it would indeed be true that some
Table 58.2 records the results of the re- are “destined to poverty almost from birth . . . by
gression calculations. It can be seen that some the economic status or occupation of their par-
alternative combinations of independent vari- ents” (in the words of the reference cited in end-
ables were studied. It turned out that the net note 1). Others, of course, would be “destined”
 Blau and Duncan—The Process of Stratikcation • 513

Table 58.2. Partial Regression Coefficients in Standard Form (Beta Coefficients) and Coefficients of
Determination, for Specified Combinations of Variables
p
Dependent Independent Variables a Coefficient of
Variablesa W U X V Determination (R2)
Ub — — .279 .310 .26
W — .433 .214 .026 .33
Wb — .440 .224 — .33
Y .282 .397 .120 -.014 .43
Yb .281 .394 .115 — .43
Y .311 .428 — — .42
a
V: father’s education; X: father’s occ. status; U: respondent’s education; W: first-job status; Y: 1962 occ. status.
b
Beta coefficients in these sets were taken as estimates of path coefficients for Figure 58.1.

to affluence or to modest circumstances. By no list of unmeasured variables that someone has

effort of their own could they materially alter the alleged to be crucial to the process under study.
course of destiny, nor could any stroke of for- But the mere existence of such variables is al-
tune, good or ill, lead to an outcome not already ready acknowledged by the very presence of the
in the cards. residual. It would seem to be part of the task
Thinking of the residual as an index of the of the critic to show, if only hypothetically, but
adequacy of an explanation gives rise to a serious specifically, how the modification of the causal
misconception. It is thought that a high mul- scheme to include a new variable would disrupt
tiple correlation is presumptive evidence that or alter the relationships in the original diagram.
an explanation is correct or nearly so, whereas His argument to this effect could then be ex-
a low percentage of determination means that a amined for plausibility and his evidence, if any,
causal interpretation is almost certainly wrong. studied in terms of the empirical possibilities it
The fact is that the size of the residual (or, if one suggests.
prefers, the proportion of variation “explained”) Our supposition is that the scheme in Fig-
is no guide whatever to the validity of a causal ure 58.1 is most easily subject to modification
interpretation. The best-known cases of “spu- by introducing additional measures of the same
rious correlation”—a correlation leading to an kind as those used here. If indexes relating to
egregiously wrong interpretation—are those in socioeconomic background other than V and
which the coefficient of determination is quite X are inserted we will almost certainly estimate
high. differently the direct effects of these particular
The relevant question about the residual is variables. If occupational statuses of the re-
not really its size at all, but whether the unob- spondent intervening between W and Y were
served factors it stands for are properly repre- known we should have to modify more or less
sented as being uncorrelated with the measured radically the right-hand portion of the diagram.
antecedent variables. We shall entertain [else- Yet we should argue that such modifications
where]14 some conjectures about unmeasured may amount to an enrichment or extension
variables that clearly are not uncorrelated with of the basic model rather than an invalidation
the causes depicted in Figure 58.1. It turns out of it. The same may be said of other variables
that these require us to acknowledge certain pos- that function as intervening causes. In theory, it
sible modifications of the diagram, whereas other should be possible to specify these in some de-
features of it remain more or less intact. A deli- tail, and a major part of the research worker’s
cate question in this regard is that of the burden task is properly defined as an attempt at such
of proof. It is all too easy to make a formidable specification. In the course of such work, to be
514 • Part VI: Who Gets Ahead?

sure, there is always the possibility of a discovery path. Having traced all such possible compound
that would require a fundamental reformulation, paths, we obtain the entirety of indirect effects
making the present model obsolete. Discarding as their sum.
the model would be a cost gladly paid for the Let us consider the example of effects of ed-
prize of such a discovery. ucation on first job, U on W. The gross or total
Postponing the confrontation with an altered effect is rWU = .538. The direct path is pWU = .440.
model, the one at hand is not lacking in interest. There are two indirect connections or com-
An instructive exercise is to compare the mag- pound paths: from W back to X, then forward
nitudes of gross and net relationships. Here we to U; and from W back to X, then back to V, and
make use of the fact that the correlation coeffi- then forward to U. Hence we have:
cient and the path coefficient have the same di-
mensionality. The correlation rYX = .406 (Table rWU = pWU + pWX pUX + pWX rXV pUV
58.1) means that a unit change (one standard
deviation) in X produces a change of 0.4 unit (gross) (direct) (indirect)
in Y, in gross terms. The path coefficient, pYX =
.115 (Figure 58.1), tells us that about one-fourth or, numerically,
of this gross effect is a result of the direct influ-
ence of X on Y. (We speculated above on the role 538 = .440 + (.224)(.279) + (.224)(.516)
of occupational inheritance in this connection.) (.310)
The remainder (.405–.115 = .29) is indirect,
via U and W. The sum of all indirect effects, = .440 + .062 + .036
therefore, is given by the difference between the
simple correlation and the path coefficient con- = .440 + .098.
necting two variables. We note that the indirect
effects on Y are generally substantial, relative to In this case all the indirect effect of U on W
the direct. Even the variable temporally closest derives from the fact that both U and W have X
(we assume) to Y has “indirect effects”—actually, (plus V ) as a common cause. In other instances,
common antecedent causes—nearly as large as when more than one common cause is involved
the direct ones. Thus rYW = .541 and pYW = .281, and these causes are themselves interrelated, the
so that the aggregate of “indirect effects,” which complexity is too great to permit a succinct ver-
in this case are common determinants of Y and bal summary.
W that spuriously inflate the correlation between A final stipulation about the scheme had
them, is .26. best be stated, though it is implicit in all the
To ascertain the indirect effects along a given previous discussion. The form of the model
chain of causation we must multiply the path itself, but most particularly the numerical esti-
coefficients along the chain. The procedure is mates accompanying it, are submitted as valid
to locate on the diagram the dependent variable only for the population under study. No claim
of interest, and then trace back along the paths is made that an equally cogent account of the
linking it to its immediate and remote causes. In process of stratification in another society could
such a tracing we may reverse direction once but be rendered in terms of this scheme. For other
only once, following the rule “first back, then populations, or even for subpopulations within
forward.” Any bidirectional correlation may be the United States, the magnitudes would almost
traced in either direction. If the diagram contains certainly be different, although we have some
more than one such correlation, however, only basis for supposing them to have been fairly con-
one may be used in a given compound path. In stant over the last few decades in this country.
tracing the indirect connections no variable may The technique of path analysis is not a method
be intersected more than once in one compound for discovering causal laws but a procedure for
 Blau and Duncan—The Process of Stratikcation • 515

giving a quantitative interpretation to the man- quotation we are not interested in splitting hairs
ifestations of a known or assumed causal system or in generating a polemic. It merely serves as
as it operates in a particular population. When a convenient point of departure for raising the
the same interpretive structure is appropriate for questions of what is specifically meant by “vi-
two or more populations there is something to cious circle,” what are the operational criteria
be learned by comparing their respective path for this concept, and what are the limits of its
coefficients and correlation patterns. We have usefulness.
not yet reached the stage at which such compara- To begin with, there is the question of fact—
tive study of stratification systems is feasible. . . . or, rather, of how the quantitative facts are to
be evaluated. How “difficult” is it, in actuality,
“for individuals to modify their status” (presum-
The Concept of a Vicious Circle ably reference is to the status of the family of
Although the concept of a “cycle of poverty” has orientation)? We have found that the father-son
a quasi-official sanction in U.S. public policy correlation for occupational status is of the order
discussion, it is difficult to locate a systematic ex- of .4. (Assuming attenuation by errors of mea-
plication of the concept. As clear a formulation surement, this should perhaps be revised slightly
as any that may be found in academic writing is upward.) Approaching the measurement prob-
perhaps the following:15 lem in an entirely different way, we find that the
amount of intergenerational mobility between
Occupational and social status are to an im- census major occupation groups is no less than
portant extent self-perpetuating. They are seven-eighths as much as would occur if there
associated with many factors which make it were no statistical association between the two
difficult for individuals to modify their status. statuses whatsoever, or five-sixths as much as the
Position in the social structure is usually asso- difference between the “minimum” mobility in-
ciated with a certain level of income, educa- volved in the intergenerational shift in occupa-
tion, family structure, community reputation, tion distributions and the amount required for
and so forth. These become part of a vicious “perfect” mobility.17 Evidently a very consider-
circle in which each factor acts on the other in able amount of “status modification” or occu-
such a way as to preserve the social structure pational mobility does occur. (There is nothing
in its present form, as well as the individual in the data exhibited by Lipset and Bendix to
family’s position in that structure. . . . The cu- indicate the contrary.) If the existing amount of
mulation of disadvantages (or of advantages) modification of status is insufficient in terms of
affects the individual’s entry into the labor some functional or normative criterion implic-
market as well as his later opportunities for itly employed, the precise criterion should be
social mobility. made explicit: How much mobility must occur to
contradict the diagnosis of a “vicious circle”?
The suspicion arises that the authors in pre- Next, take the postulate that occupational
paring this summary statement were partly cap- status (of origin) is “associated with many fac-
tured by their own rhetoric. Only a few pages tors” and that “each factor acts on the other” so
earlier they had observed that the “widespread as “to preserve . . . the individual family’s po-
variation of educational attainment within sition.” Here the exposition virtually cries out
classes suggests that one’s family background for an explicit quantitative causal model; if not
plays an enabling and motivating rather than one of the type set forth in the first section of
a determining role.”16 But is an “enabling and this chapter, then some other model that also
motivating role” logically adequate to the func- takes into account the way in which several
tion of maintaining a “vicious circle”? In focus- variables combine their effects. Taking our own
ing closely on the precise wording of the earlier earlier model, for want of a better alternative, as
516 • Part VI: Who Gets Ahead?

representative of the situation, what do we learn factors, each of which has a significant relation-
about the “associated factors”? Family “position” ship to subsequent achievement, the total effect
is, indeed, “associated with . . . education,” and of origin on achievement is materially enhanced.
education in turn makes a sizable difference in Here, in other words, the concept of “cumula-
early and subsequent occupational achievement. tion” appears to refer to the intercorrelations of
Yet of the total or gross effect of education (U ) a collection of independent variables. But the ef-
on Y, occupational status in 1962 (rYU = .596), fect of such intercorrelations is quite opposite to
only a minor part consists in a transmission of what the writers appear to suppose. They are not
the prior influence of “family position,” at least alone in arguing from a fallacious assumption
as this is indicated by measured variables V (fa- that was caustically analyzed by Karl Pearson
ther’s education) and X (father’s occupation). . . . half a century ago.18 The crucial point is that if
A relevant calculation concerns the compound the several determinants are indeed substantially
paths through V and X linking Y to U. Using intercorrelated with each other, then their com-
data for men 20 to 64 years old with nonfarm bined effect will consist largely in redundancy,
background, we find: not in “cumulation.” This circumstance does not
relieve us from the necessity of trying to under-
pYX pUX = .025 stand better how the effects come about (a point
also illustrated in a less fortunate way in Pear-
pYX rXV pUV = .014 son’s work). It does imply that a refined estimate
of how much effect results from a combination
pYX pWX pUX = .014 of “associated factors” will not differ greatly from
a fairly crude estimate based on the two or three
pYW pWX rXV pUV = .008 most important ones.

Sum = .061 NOTES

1. Mollie Orshansky, “Children of the Poor,” So-
This is the entire part of the effect of educa- cial Security Bulletin, 26(July 1963).
2. Forrest A. Bogan, “Employment of High
tion that has to do with “perpetuating” the “fam-
School Graduates and Dropouts in 1964,” Special
ily’s position.” By contrast, the direct effect is pYU Labor Force Report, No. 54 (U.S. Bureau of Labor
= .407 and the effect via W (exclusive of prior Statistics, June 1965), p. 643.
influence of father’s education and occupation 3. Peter M. Blau and Otis Dudley Duncan, The
on respondent’s first job) is pYW pWU = .128, for a American Occupational Structure, New York: The
total of .535. Far from serving in the main as a Free Press, 1967, ch. 4.
factor perpetuating initial status, education oper- 4. Ibid.
ates primarily to induce variation in occupational 5. Bruce K. Eckland, “College Dropouts Who
status that is independent of initial status. The Came Back,” Harvard Educational Review, 34(1964),
402–420.
simple reason is that the large residual factor for
6. Beverly Duncan, Family Factors and School
U is an indirect cause of Y. But by definition it is Dropout: 1920–1960, U.S. Office of Education,
quite uncorrelated with X and V. This is not to Cooperative Research Project No. 2258, Ann Arbor:
gainsay the equally cogent point that the degree Univ. of Michigan, 1965.
of “perpetuation” (as measured by rYX) that does 7. Blau and Duncan, op. cit., Appendix C.
occur is mediated in large part by education. . . . 8. Gösta Carlsson, Social Mobility and Class
Our model also indicates where the “vicious Structure, Lund: CWK Gleerup, 1958, p. 124.
circle” interpretation is vulnerable. In the passage 9. H. M. Blalock, Jr., Causal Inferences in Non-
on the vicious circle quoted, there seems to be an experimental Research, Chapel Hill: Univ. of North
Carolina Press, 1964.
assumption that because of the substantial inter-
10. Sewall Wright, “Path Coefficients and Path
correlations between a number of background
Regressions,” Biometrics, 16(1960), 189–202; Otis
 Jencks et al.—Inequality • 517

Dudley Duncan, “Path Analysis,” American Journal 16. Ibid., p. 190.

of Sociology, 72(1966), 1–16. 17. U.S. Bureau of the Census, “Lifetime Oc-
11. Blalock, op. cit., pp. 54 ff. cupational Mobility of Adult Males: March 1962,”
12. Blau and Duncan, op. cit., ch. 4. Current Population Reports, Series P-23, No. 11 (May
13. Ibid., ch. 3. 12, 1964), Table B.
14. Ibid., ch. 5. 18. Karl Pearson, “On Certain Errors with Re-
15. Seymour M. Lipset and Reinhard Bendix, gard to Multiple Correlation Occasionally Made by
Social Mobility in Industrial Society, Berkeley: Univ. Those Who Have Not Adequately Studied This Sub-
of California Press, 1959, pp. 198–199. ject,” Biometrika, 10(1914), 181–187.

59 • Christopher Jencks, Marshall Smith, Henry Acland,

Mary Jo Bane, David Cohen, Herbert Gintis, Barbara Heyns,
and Stephan Michelson

Inequality
A Reassessment of the Effect of Family and Schooling in America

Most Americans say they believe in equality. But still believe in what they often call “equal oppor-
when pressed to explain what they mean by this, tunity.” By this they mean that the rules deter-
their definitions are usually full of contradic- mining who succeeds and who fails should be
tions. Many will say, like the Founding Fathers, fair. People are, of course, likely to disagree about
that “all men are created equal.” Many will also precisely what is “fair” and what is “unfair.” Still,
say that all men are equal “before God,” and that the general principle of fair competition is al-
they are, or at least ought to be, equal in the most universally endorsed.
eyes of the law. But most Americans also believe During the 1960s, many reformers devoted
that some people are more competent than oth- enormous effort to equalizing opportunity. More
ers, and that this will always be so, no matter specifically, they tried to eliminate inequalities
how much we reform society. Many also believe based on skin color, and to a lesser extent on eco-
that competence should be rewarded by success, nomic background. They also wanted to elimi-
while incompetence should be punished by fail- nate absolute deprivation: “poverty,” “ignorance,”
ure. They have no commitment to ensuring that “powerlessness,” and so forth. But only a handful
everyone’s job is equally desirable, that everyone of radicals talked about eliminating inequality per
exercises the same amount of political power, or se. Almost none of the national legislation passed
that everyone receives the same income. during the 1960s tried to reduce disparities in
But while most Americans accept inequality adult status, power, or income in any direct way.
in virtually every sphere of day-to-day life, they There was no significant effort, for example, to

Christopher Jencks, Marshall Smith, Henry Acland, Mary Jo Bane, David Cohen, Herbert Gintis, Barbara
Heyns, and Stephan Michelson. Inequality: A Reassessment of the Effect of Family and Schooling in America,
pp. 3–11. Copyright © 1972 by Basic Books, Inc. Reprinted by permission of Basic Books, a member of the
Perseus Books Group.
518 • Part VI: Who Gets Ahead?

make taxation more progressive, and very little condition. Having assumed this, they all showed
effort to reduce wage disparities between highly steady progress toward the elimination of pov-
paid and poorly paid workers. Instead, attention erty, since fewer and fewer people had incomes
focused on helping workers in poorly paid jobs below the official “poverty line.”
to move into better paid jobs. Nor was there Yet despite all the official announcements
much effort to reduce the social or psychologi- of progress, the feeling that lots of Americans
cal distance between high- and low-status occu- were poor persisted. The reason was that most
pations. Instead, the idea was to help people in Americans define poverty in relative rather than
low-status occupations leave these occupations absolute terms. Public opinion surveys show, for
for more prestigious ones. Even in the political example, that when people are asked how much
arena, “maximum feasible participation” implied money an American family needs to “get by,”
mainly that more “leaders” should be black and they typically name a figure about half what the
poor, not that power should be equally distrib- average American family actually receives.1 This
uted between leaders and followers. has been true for the last three decades, despite
Because the reforms of the 1960s did not the fact that real incomes (i.e. incomes adjusted
tackle the problem of adult inequality directly, for inflation) have doubled in the interval.
they accomplished only a few of their goals. Political definitions of poverty have reflected
Equalizing opportunity is almost impossible these popular attitudes. During the Depres-
without greatly reducing the absolute level of sion, the average American family was living on
inequality, and the same is true of eliminating about $30 a week. A third of all families were
deprivation. living on less than half this amount, i.e. less than
Consider the case of equal opportunity. $15 a week. This made it natural for Franklin
One can equalize the opportunities available to Roosevelt to speak of “one third of a nation” as
blacks and whites without equalizing anything ill-housed, ill-clothed, and ill-fed. One third of
else, and considerable progress was made in this the nation was below what most people then re-
direction during the late 1960s. But equalizing garded as the poverty line.
the opportunities available to different children By 1964, when Lyndon Johnson declared war
of the same race is far more difficult. If a soci- on poverty, incomes had risen more than five-
ety is competitive and rewards adults unequally, fold. Even allowing for inflation, living standards
some parents are bound to succeed while oth- had doubled. Only about 10 percent of all fam-
ers fail. Successful parents will then try to pass ilies had real incomes as low as the bottom third
along their advantages to their children. Unsuc- had had during the Depression. But popular
cessful parents will inevitably pass along some of conceptions of what it took to “get by” had also
their disadvantages. Unless a society completely risen since the Depression. Mean family income
eliminates ties between parents and children, in- was about $160 a week, and popular opinion
equality among parents guarantees some degree now held that it took $80 a week for a family
of inequality in the opportunities available to of four to make ends meet. About a quarter of
children. The only real question is how serious all families were still poor by this definition. As
these inequalities must be. a matter of political convenience, the Adminis-
Or consider the problem of deprivation. tration set the official poverty line at $60 a week
When the war on poverty began in late 1963, for a family of four rather than $80, ensuring
it was conceived as an effort to raise the living that even conservatives would admit that those
standards of the poor. The rhetoric of the time below the line were poor. But by 1970 inflation
described the persistence of poverty in the midst had raised mean family income to about $200 a
of affluence as a “paradox,” largely attributable to week, and the National Welfare Rights Organi-
“neglect.” Official publications all assumed that zation was rallying liberal support for a guaran-
poverty was an absolute rather than a relative teed income of $100 a week for a family of four.
 Jencks et al.—Inequality • 519

These political changes in the definition of to make their beds, tend their gardens, and drive
poverty were not just a matter of “rising expec- their cars. These are not privileges that become
tations” or of people’s needing to “keep up with more widely available as people become more af-
the Joneses.” The goods and services that made fluent. If all workers’ wages rise at the same rate,
it possible to live on $15 a week during the De- the highly paid professional will have to spend a
pression were no longer available to a family constant percentage of his income to get a maid,
with the same “real” income (i.e. $40 a week) a gardener, or a taxi. The number of people who
in 1964. Eating habits had changed, and many are “rich,” in the sense of controlling more than
cheap foods had disappeared from the stores. their share of other people’s time and effort, will
Most people had enough money to buy an auto- therefore remain the same, even though con-
mobile, so public transportation had atrophied, sumption of yachts and filet mignon is rising.
and families without automobiles were much If the distribution of income becomes more
worse off than during the Depression. The labor equal, as it did in the 1930s and 1940s, the
market had also changed, and a person without number of people who are “rich” in this sense
a telephone could not get or keep many jobs. A of the term will decline, even though absolute
home without a telephone was more cut off so- incomes are rising. If, for example, the wages of
cially than when few people had telephones and domestic servants rise faster than the incomes of
more people “dropped by.” Housing arrange- their prospective employers, fewer families will
ments had changed, too. During the Depression, feel they can afford full-time servants. This will
many people could not afford indoor plumbing lower the living standards of the elite to some ex-
and “got by” with a privy. By the 1960s, priv- tent, regardless of what happens to consumption
ies were illegal in most places. Those who could of yachts and filet mignon.
not afford an indoor toilet ended up in build- This same logic applies not only to income
ings which had broken toilets. For this they paid but to the cognitive skills taught in school.
more than their parents had paid for privies. Young people’s performance on standardized
Examples of this kind suggest that the “cost tests rose dramatically between World War I and
of living” is not the cost of buying some fixed World War II, for example. But the level of com-
set of goods and services. It is the cost of partic- petence required for many adult roles rose too.
ipating in a social system. The cost of participa- When America was a polyglot nation of immi-
tion depends in large part on how much other grants, all sorts of jobs were open to those who
people habitually spend to participate. Those could not read English. Such people could, for
who fall far below the norm, whatever it may example, join the army, drive a truck, or get a
be, are excluded. It follows that raising the in- job in the construction industry. Today, when
comes of the poor will not eliminate poverty if almost everyone can read English, the range of
the incomes of other Americans rise even faster. choices open to nonreaders has narrowed. The
If people with incomes less than half the national military no longer takes an appreciable number
average cannot afford what “everyone” regards as of illiterates, a driver’s license requires a written
“necessities,” the only way to eliminate poverty examination, and apprenticeships in the con-
is to make sure everyone has an income at least struction trades are restricted to those who can
half the average. pass tests. Those who cannot read English are at
This line of reasoning applies to wealth as a disadvantage, simply because they are atypical.
well as poverty. The rich are not rich because America is not organized with their problems in
they eat filet mignon or own yachts. Millions mind. The same thing applies to politics. If the
of people can now afford these luxuries, but average citizen’s vocabulary expands, the vocab-
they are still not “rich” in the colloquial sense. ulary used by politicians and newspapers will ex-
The rich are rich because they can afford to buy pand too. Those with very limited vocabularies
other people’s time. They can hire other people relative to their neighbors will still have trouble
520 • Part VI: Who Gets Ahead?

following events, even though their vocabulary attend the same schools as middle-class chil-
is larger than, say, their parents’ vocabulary was. dren, by giving them extra compensatory
Arguments of this kind suggest that it makes programs in school, by giving their parents
more sense to think of poverty and ignorance as a voice in running their schools, or by some
relative than as absolute conditions. They also combination of all three approaches.
suggest that eliminating poverty and ignorance,
at least as these are usually defined in America, So far as we can discover, each of these as-
depends on eliminating, or at least greatly reduc- sumptions is erroneous.
ing, inequality. This is no simple matter. Since 1. Poverty is not primarily hereditary. While
a competitive system means that some people children born into poverty have a higher-
“succeed” while others “fail,” it also means that than-average chance of ending up poor,
people will end up unequal. If we want to re- there is still an enormous amount of eco-
duce inequality, we therefore have two options. nomic mobility from one generation to
The first possibility is to make the system less the next. Indeed, there is nearly as much
competitive by reducing the benefits that derive economic inequality among brothers raised
from success and the costs paid for failure. The in the same homes as in the general popu-
second possibility is to make sure that everyone lation. This means that inequality is recre-
enters the competition with equal advantages ated anew in each generation, even among
and disadvantages. people who start life in essentially identical
The basic strategy of the war on poverty circumstances.
during the 1960s was to try to give everyone en- 2. The primary reason some people end up
tering the job market or any other competitive richer than others is not that they have
arena comparable skills. This meant placing great more adequate cognitive skills. While chil-
emphasis on education. Many people imagined dren who read well, get the right answers
that if schools could equalize people’s cognitive to arithmetic problems, and articulate their
skills this would equalize their bargaining power thoughts clearly are somewhat more likely
as adults. In such a system nobody would end up than others to get ahead, there are many
very poor—or, presumably, very rich. other equally important factors involved.
This strategy rested on a series of assumptions Thus there is almost as much economic
which went roughly as follows: inequality among those who score high on
1. Eliminating poverty is largely a matter of standardized tests as in the general popu-
helping children born into poverty to rise lation. Equalizing everyone’s reading scores
out of it. Once families escape from poverty, would not appreciably reduce the number
they do not fall back into it. Middle-class of economic “failures.”
children rarely end up poor. 3. There is no evidence that school reform can
2. The primary reason poor children do not substantially reduce the extent of cognitive
escape from poverty is that they do not inequality, as measured by tests of verbal
acquire basic cognitive skills. They cannot fluency, reading comprehension, or math-
read, write, calculate, or articulate. Lack- ematical skill. Neither school resources nor
ing these skills, they cannot get or keep a segregation has an appreciable effect on ei-
well-paid job. ther test scores or educational attainment.
3. The best mechanism for breaking this vi-
cious circle is educational reform. Since Our work suggests, then, that many popular
children born into poor homes do not ac- explanations of economic inequality are largely
quire the skills they need from their parents, wrong. We cannot blame economic inequality
they must be taught these skills in school. primarily on genetic differences in men’s capac-
This can be done by making sure that they ity for abstract reasoning, since there is nearly
 Jencks et al.—Inequality • 521

as much economic inequality among men with Pursued with vigor, such a strategy would make
equal test scores as among men in general. We “poverty” (i.e. having a living standard less than
cannot blame economic inequality primarily half the national average) virtually impossible.
on the fact that parents pass along their disad- It would also make economic “success,” in the
vantages to their children, since there is nearly sense of having, say, a living standard more than
as much inequality among men whose parents twice the national average, far less common than
had the same economic status as among men in it now is. The net effect would be to make those
general. We cannot blame economic inequality with the most competence and luck subsidize
on differences between schools, since differences those with the least competence and luck to a
between schools seem to have very little effect far greater extent than they do today.
on any measurable attribute of those who attend This strategy was rejected during the 1960s
them. for the simple reason that it commanded rela-
Economic success seems to depend on varie- tively little popular support. The required leg-
ties of luck and on-the-job competence that are islation could not have passed Congress. Nor
only moderately related to family background, could it pass today. But that does not mean it
schooling, or scores on standardized tests. The was the wrong strategy. It simply means that
definition of competence varies greatly from until we change the political and moral premises
one job to another, but it seems in most cases to on which most Americans now operate, poverty
depend more on personality than on technical and inequality of opportunity will persist at
skills. This makes it hard to imagine a strategy pretty much their present level.
for equalizing competence. A strategy for equal- At this point the reader may wonder whether
izing luck is even harder to conceive. trying to change these premises is worthwhile.
The fact that we cannot equalize luck or Why, after all, should we be so concerned about
competence does not mean that economic in- economic equality? Is it not enough to ensure
equality is inevitable. Still less does it imply that equal opportunity? And does not the evidence
we cannot eliminate what has traditionally been we have described suggest that opportunities are
defined as poverty. It only implies that we must already quite equal in America? If economic op-
tackle these problems in a different way. Instead portunities are relatively equal, and if the lucky
of trying to reduce people’s capacity to gain a and the competent then do better for themselves
competitive advantage on one another, we would than the unlucky and incompetent, why should
have to change the rules of the game so as to re- we feel guilty about this? Such questions cannot
duce the rewards of competitive success and the be answered in any definitive way, but a brief
costs of failure. Instead of trying to make every- explanation of our position may help avoid
one equally lucky or equally good at his job, we misunderstanding.
would have to devise “insurance” systems which We begin with the premise that every indi-
neutralize the effects of luck, and income-sharing vidual’s happiness is of equal value. From this it
systems which break the link between vocational is a short step to Bentham’s dictum that society
success and living standards. should be organized so as to provide the greatest
This could be done in a variety of ways. Em- good for the greatest number. In addition, we as-
ployers could be constrained to reduce wage sume that the law of diminishing returns applies
disparities between their best- and worst-paid to most of the good things in life. In economic
workers.2 The state could make taxes more pro- terms this means that people with low incomes
gressive, and could provide income supplements value extra income more than people with high
to those who cannot earn an adequate living incomes.3 It follows that if we want to maximize
from wages alone. The state could also provide the satisfaction of the population, the best way
free public services for those who cannot afford to divide any given amount of money is to make
to buy adequate services in the private sector. everyone’s income the same. Income disparities
522 • Part VI: Who Gets Ahead?

(except those based on variations in “need”) will from society. We do not think it can or should be
always reduce overall satisfaction, because indi- the sole consideration.
viduals with low incomes will lose more than When we turn from the distribution of in-
individuals with high incomes gain. come to the distribution of other things, our
The principal argument against equalizing commitment to equality is even more equivocal.
incomes is that some people contribute more to We assume, for example, that occupational pres-
the general welfare than others, and that they are tige resembles income in that those who have
therefore entitled to greater rewards. The most low-prestige occupations usually value additional
common version of this argument is that unless prestige more than those who have high-prestige
those who contribute more than their share are occupations. Insofar as prestige is an end in it-
rewarded (and those who contribute less than self, then, the optimal distribution is again egal-
their share punished) productivity will fall and itarian. But occupational prestige derives from
everyone will be worse off. A more sophisticated a variety of factors, most of which are more dif-
version is that people will only share their in- ficult to redistribute than income. We cannot
comes on an equal basis if all decisions that affect imagine a social system in which all occupations
these incomes are made collectively. If people are have equal prestige, except in a society where all
left free to make decisions on an individual basis, workers are equally competent. Since we do not
their neighbors cannot be expected to pay the see any likelihood of equalizing competence, we
entire cost of their mistakes. regard the equalization of occupational prestige
We accept the validity of both these argu- as a desirable but probably elusive goal.
ments. We believe that men need incentives to When we turn from occupational prestige
contribute to the common good, and we prefer to educational attainment and cognitive skills,
monetary incentives to social or moral incen- the arguments for and against equality are re-
tives, which tend to be inflexible and very co- versed. If schooling and knowledge are thought
ercive. We believe, in other words, that virtue of strictly as ends in themselves, it is impossible
should be rewarded, and we assume that there to make a case for distributing them equally. We
will be considerable variation in virtue from one can see no reason to suppose, for example, that
individual to another. This does not, however, people with relatively little schooling value addi-
mean that incomes must remain as unequal as tional schooling more than people who have al-
they are now. Even if we assume, for example, ready had a lot of schooling. Experience suggests
that the most productive fifth of all workers ac- that the reverse is the case. Insofar as schooling
counts for half the Gross National Product, it is an end in itself, then, Benthamite principles
does not follow that they need receive half the imply that those who want a lot should get a
income. A third or a quarter might well suffice to lot, and those who want very little should get
keep both them and others productive. very little. The same is true of knowledge and
Most people accept this logic to some ex- cognitive skills. People who know a lot generally
tent. They believe that the rich should pay more value additional knowledge and skills more than
taxes than the poor, although they often disagree those who know very little. This means that in-
about how much more. Conversely, they believe sofar as knowledge or skill is valued for its own
that the poor should not starve, even if they con- sake, an unequal distribution is likely to give
tribute nothing to the general welfare. They be- more satisfaction to more people than an equal
lieve, in other words, that people should not be distribution.
rewarded solely for their contribution to the gen- The case for equalizing the distribution of
eral welfare, but that other considerations, such schooling and cognitive skill derives not from
as need, should also be taken into account. Our the idea that we should maximize consumer
egalitarianism is simply another way of saying satisfaction, but from the assumption that
that we think need should play a larger role than equalizing schooling and cognitive skill is nec-
it now does in determining what people get back essary to equalize status and income. This puts
 Jencks et al.—Inequality • 523

egalitarians in the awkward position of trying to a forthcoming study of the social meaning of low
impose equality on people, even though the nat- income.
ural demand for both cognitive skill and school- 2. Lester C. Thurow and Robert E.B. Lucas, in
ing is very unequal. Since we have found rather “The American Distribution of Income” [Washing-
ton, D.C.: U.S. Government Printing Office, March
modest relationships between cognitive skill and
17, 1972], discuss the possibility of such constraints
schooling on the one hand and status and in- in some detail. The principal virtue of this approach
come on the other, we are much less concerned is that it reduces the incomes of the rich before they
than most egalitarians with making sure that are defined as “income” rather than afterwards. This
people end up alike in these areas. means that the recipient is less conscious of what he
Our commitment to equality is, then, neither is giving up and less likely to feel he is being cheated
all-embracing nor absolute. We do not believe of his due.
that everyone can or should be made equal to 3. If everyone had equal earning power we could
everyone else in every respect. We assume that assume that people “chose” their incomes volun-
tarily and that those with low incomes were those
some differences in cognitive skill and vocational
who were maximizing something else (e.g. leisure,
competence are inevitable, and that efforts to autonomy, etc.). But as we note [elsewhere], people’s
eliminate such differences can never be 100 per- concern with income as against other objectives has
cent successful. But we also believe that the dis- no apparent effect on their actual income, at least
tribution of income can be made far more equal while they are young [see Christopher Jencks, Mar-
than it is, even if the distribution of cognitive shall Smith, Henry Acland, Mary Jo Bane, David
skill and vocational competence remains as un- Cohen, Herbert Gintis, Barbara Heyns, and Stephan
equal as it is now. We also think society should Michelson, Inequality: A Reassessment of the Effect of
get on with the task of equalizing income rather Family and Schooling in America, New York: Harper
and Row, 1972, ch. 7, note 64]. Thus we infer that
than waiting for the day when everyone’s earning
income differences derive largely from differences in
power is equal.
earning power and luck.

NOTES
1. This material has been collected and analyzed
by Lee Rainwater at Harvard University, as part of
524 • Part VI: Who Gets Ahead?

Education and Reproduction

60 • Richard Breen and John H. Goldthorpe

Explaining Educational Differentials

Introduction not brought their take-up rates of more ambi-

In the light of recent research in the sociology tious educational options closer to those of their
of education, which has involved extensive more advantaged counterparts.
over-time and cross-national analyses (see esp. It has, though, to be recognized that this latter
Shavit and Blossfeld 1993; Erikson and Jonsson generalization is not entirely without exception.
1996b), it would seem that the following em- In one national case at least, that of Sweden,
pirical generalizations can reliably be made and there can be little doubt that class differentials
constitute explananda that pose an evident the- in educational attainment have indeed declined
oretical challenge. over several decades (Erikson and Jonsson 1993);
Over the last half-century at least, all eco- and, while some conflict of evidence remains, a
nomically advanced societies have experienced similar decline has been claimed for The Neth-
a process of educational expansion. Increasing erlands (De Graaf and Ganzeboom 1993) and
numbers of young people have stayed on in for Germany (Müller and Haun 1994; Jonsson,
full-time education beyond the minimum school Mills and Müller 1996). Thus, any theory that is
leaving age, have taken up more academic sec- put forward in order to explain the more typical
ondary courses, and have entered into some persistence of class differentials should be one
form of tertiary education. that can at the same time be applied mutatis mu-
Over this same period, class differentials in tandis to such ‘deviant’ cases.
educational attainment, considered net of all ef- It would in addition be desirable that such a
fects of expansion per se, have tended to display theory should be capable of yet further extension
a high degree of stability, i.e. while children of all in order to account for a third regularity that has
class backgrounds have alike participated in the emerged from the research referred to.
process of expansion, the pattern of association Over a relatively short period—in effect,
between class origins and the relative chances of from the 1970s onwards—gender differentials
children staying on in education, taking more in levels of educational attainment, favouring
academic courses or entering higher education males over females, have in nearly all advanced
has, in most societies, been rather little altered. societies declined sharply and, in some instances,
Children of less advantaged class origins have have been virtually eliminated or even reversed.

Richard Breen and John H. Goldthorpe. “Explaining Educational Differentials: Towards a Formal Rational
Action Theory,” in Rationality and Society 9:3 (August 1997), pp. 275–287, 293–300, 302–305. Copyright ©
1997 by Sage Publications, Inc. Reprinted by permission of Sage.
 Breen and Goldthorpe—Explaining Educational Differentials • 525

In other words, while the process of educational in character. It is, rather, secondary effects that
expansion has not in the main led to children for us play the crucial role. These are effects that
from less advantaged family backgrounds catch- are expressed in the actual choices that children,
ing up with those from more advantaged back- together perhaps with their parents, make in the
grounds in their average levels of attainment, in course of their careers within the educational
families across the class structures of contempo- system—including the choice of exit. Some ed-
rary societies daughters have tended rather rap- ucational choices may of course be precluded to
idly to catch up with sons. some children through the operation of primary
In an earlier paper (Goldthorpe 1996), a effects: i.e. because these children lack the re-
theory of persisting class differentials in edu- quired level of demonstrated ability. But, typ-
cational attainment, sensitive to the further re- ically, a set of other choices remains, and it is
quirements previously indicated, was developed further known that the overall patterns of choice
from a ‘rational action’ standpoint. In the pres- that are made, are in themselves—over and
ent [chapter], our aim is to refine this theory above primary effects—an important source of
and to express it in a formal model. In this way class differentials in attainment.
we would hope to clarify its central arguments We then further assume that, in their central
and in turn the wider implications that it car- tendencies, these patterns of educational choice
ries. Since such attempts at the formalization of reflect action on the part of children and their
theory are still not very common in sociology, parents that can be understood as rational, i.e.
the [chapter] may also serve to stimulate dis- they reflect evaluations made of the costs and
cussion of the merits or demerits of this kind benefits of possible alternatives—e.g. to leave
of endeavour. Readers interested in the more school or to stay on, to take a more academic or
general problematik in the context of which the a more vocational course—and of the probabil-
theory was initially conceived are referred to ities of different outcomes, such as educational
the earlier paper. In the remainder of this intro- success or failure. These evaluations, we further
ductory section we set out certain ‘background’ suppose, will be in turn conditioned by differ-
assumptions of our subsequent exposition that ences in the typical constraints and opportuni-
will not be further discussed. The more specific ties that actors in different class positions face
assumptions on which our model rests will be and in the level of resources that they command.
introduced, and their significance considered, as However, what we seek to dispense with is any
the [chapter] proceeds. assumption that these actors will also be subject
We assume, to begin with, that class differ- to systematic influences of a (sub)cultural kind,
entials in educational attainment come about whether operating through class differences in
through the operation of two different kinds of values, norms or beliefs regarding education
effect which, following Boudon (1974), we label or through more obscure ‘subintentional’ pro-
as ‘primary’ and ‘secondary’. Primary effects are cesses. Not only do we thus gain in theoretical
all those that are expressed in the association that parsimony, but we would in any event regard
exists between children’s class origins and their the ‘culturalist’ accounts of class differentials
average levels of demonstrated academic ability. in educational attainment that have so far been
Children of more advantaged backgrounds are in advanced, as in various ways unsatisfactory (see
fact known to perform better, on average, than further Goldthorpe 1996).
children of less advantaged backgrounds on stan- Finally, two other assumptions, regarding
dard tests, in examinations, etc. Primary effects, the structural context of action, should also be
as will be seen, enter into our model but, fortu- spelled out. On the one hand, we do of course
nately, in such a way that we need not take up suppose the existence of a class structure, i.e.
the vexed and complex question of the extent to a structure of positions defined by relations in
which they are genetic, psychological or cultural labour markets and production units. And, in
526 • Part VI: Who Gets Ahead?

addition, we need to assume that within this the entire range of decisions that young people
structure classes are in some degree hierarchi- may be required to make over the course of their
cally ordered in terms of the resources associated educational careers as regards leaving or staying
with, and the general desirability of the positions on or as regards which educational option to
they comprise. On the other hand, we suppose pursue. However, in the interests of simplicity,
an educational system—i.e. a set of educational we will here set out the model as it would apply
institutions that serve to define the various op- just to the choice of leaving or continuing in ed-
tions that are open to individuals at successive ucation. The salient elements of the exposition
stages in their educational careers. And here, are shown in Figure 60.1 by means of a decision
too, we have a more specific requirement. That tree. Here we assume that pupils must choose
is, that this system should possess a diversified whether to continue in education—i.e. follow
structure that provides options not just for more the ‘stay’ branch of the tree—to the comple-
or less education but also for education of dif- tion of a further level (as, say, in the decision
fering kinds, and that in turn entails individuals of whether or not to continue to A-level after
making choices at certain ‘branching points’ that GCSE) or to leave and enter the labour market,
they may not be able later to modify, or at least i.e. follow the ‘leave’ branch. Continuing in edu-
not in a costless way. It might be thought that cation has two possible outcomes, which we take
this latter requirement will tend to limit the ap- to be success or failure. Because remaining at
plicability of our model to educational systems school often leads to an examination, we equate
of the more traditional European, rather than, success with passing such an examination. This is
say, the American variety, i.e. to ones where the indicated by the node labelled P in Figure 60.1,
type of school attended is likely to be more con- while failing the examination is indicated by the
sequential than the total number of years spent node labelled F. Leaving is then the third educa-
in education. However, we would argue that, on tional outcome in our model, that is, in addition
examination, educational systems such as that of to those of staying in education and passing and
the USA turn out to be more diversified than is staying and failing, and is indicated by node L.
often supposed, so that children do in fact face In deciding whether to continue in education
educational choices that involve considerations or leave, parents and their children, we suppose,
that go beyond simply ‘more’ or ‘less’: for exam- take into account three factors. The first of these
ple, in the American case, with the choice at sec- is the cost of remaining at school. Continuing
ondary level between academic and vocational in full-time education will impose costs on a
tracks. It is further of interest to note how two family which they would not have to meet were
American authors have specified in this regard their child to leave school: these include the di-
the divergence between assumptions that we rect costs of education and earnings forgone. We
and they would share and those of most econo- can therefore express these costs relative to the
mists working within the ‘human capital’ para- costs of leaving by setting the latter to be zero
digm. While for the latter education appears as and the former as c > 0. The second factor is
a ‘fungible linear accumulation, like a financial the likelihood of success if a pupil continues in
investment’, a more realistic view would be that education. Since we distinguish only between
educational systems, the American included, success and failure, subjective beliefs about the
‘offer an array of choices and constraints that chances of success at the next stage of education
defy . . . simple linear formulations’ (Arum and can be captured in our model by a single param-
Hout 1995, 1). eter, which we label π. This parameter measures
the subjective conditional probability of passing
the relevant examination given continuation.
A Model of Educational Decisions The third factor is then the value or utility that
The model that we present is intended to be ge- children and their families attach to the three ed-
neric: that is, as one applicable in principle to ucational outcomes represented by P, F and L in
 Breen and Goldthorpe—Explaining Educational Differentials • 527

Figure 60.1. In our model this factor is expressed Figure 60.1. Single Decision Tree
in terms of beliefs about the chances of access
that each outcome affords to three possible des- S*
tination classes. α
For the purposes of our exposition, we take
P
these classes as being the service class or salariat W*
of professionals, administrators and managers 1–α
(S* in Figure 60.1), the working class (W*) and π
the underclass, (U*)—the class, say, of those
with only a precarious place in the labour market
and in only the lowest grades of employment if
not unemployed. However, it should be empha-
sized that nothing of significance attaches to this
S*
choice of classes, except that, as earlier noted, we Stay
1–π β1
need to have a hierarchical ordering. Thus, the
service class is regarded as comprising the most F β2
advantaged and most desirable positions and the W*
underclass the least advantaged and least desir-
1 – β1 – β2
able, with the working class falling in-between.
This ranking of classes is, moreover, assumed to U*
be universally recognized or, at all events, not to
vary across the population in any socially struc- Leave
tured way.
As we have said, each of the three possible ed-
ucational outcomes in our model has attached to
it subjective probabilities of access to each of the S*
three possible destination classes. So, as Figure γ1
60.1 shows, for pupils who remain at school and L γ2
pass their examination, node P, the probability of W*
access to the service class is given by Į. There is
1 – γ1 – γ2
no path linking this educational outcome to the
underclass. This means that anyone who reaches U*
this particular outcome is believed to be certain
to avoid this class. It follows, therefore, that the
probability of entering the working class, con-
ditional on having been educationally success- Į, ȕ and Ȗ parameters reflect people’s beliefs,
ful, is given by 1–Į. At the other two outcome in this case, about the returns to various edu-
nodes, F and L, there is a positive probability of cational outcomes conceptualized in terms of
entering all three destination classes. So, for the access to more or less desirable locations in the
outcome F (remaining at school and failing) the class structure. In principle, therefore, these pa-
probability of access to the service class is given rameters could vary widely across individuals
by ȕ1, the probability of access to the working and families. Again, though, we assume a soci-
class by ȕ2 and the probability of access to the etal consensus in regard to a set of beliefs that
underclass by 1 – ȕ1 – ȕ2. For the L outcome the then serve as conditions on the parameters in
corresponding probabilities are then given by the question and that may be stated as follows:
Ȗ parameters. i. Į > ȕ1 and Į > Ȗ1. It is generally believed
We repeat that these are all subjective proba- that remaining at school and succeeding af-
bilities. Just as with π, the values for our various fords a better chance of access to the service
528 • Part VI: Who Gets Ahead?

class than does remaining at school and decision rather than another. For ease of exposi-
failing or leaving school. Our model does tion here we consider only two classes of origin,
not require that we make any assumptions the service class, S, and the working class, W. In
about the relative magnitude of ȕ1 and Ȗ1. It all of what follows we assume that these classes
could, for example, be the case that a young differ in only two ways. First (and it is here that
person’s chances of access to the service class we give recognition to ‘primary’ effects) children
are improved simply by acquiring more of the two classes differ in their average ability.
years of education, even if this does not lead Ability is taken to be normally distributed within
to examination success. Alternatively, such each class with means as > aw and variance given
time spent in education may be wasted in by ı2a. Secondly, the two classes have different
the sense that leaving school and embarking levels of resources, r, which they can use to meet
earlier on a career will yield a better chance c, the costs of education. Resources are taken to
of access to the service class. have a logistic distribution with mean values rs
ii. Ȗ1 + Ȗ2 > ȕ1 + ȕ2. Remaining at school and > rw for the two classes and a common disper-
failing increases the chances of entering the sion parameter, ı2r. Throughout, we make no
underclass. This means that there is a risk other assumptions about differences between
involved in choosing to continue to the the classes. In particular, and as earlier noted, we
next level of education. do not suppose any class-specific cultural values
iii. Ȗ2/Ȗ1 > 1; (Ȗ2/Ȗ1) ≥ (ȕ2/ȕ1). Those who leave or social norms nor any class differences in the
school immediately have a better chance of subjective Į, ȕ and Ȗ parameters of our model.
entry to the working class than to the ser- We then propose three mechanisms through
vice class. This may or may not be the case which class differentials in educational attain-
among those who remain at school and fail ment may arise at the level of ‘secondary’ effects.
though, if it is, their odds of entering the Of these three, we would wish to stress the par-
working class rather than the service class ticular importance of the first, since this provides
are no greater than for those who leave an account of how these differentials may be cre-
school immediately. ated and sustained through the apparently ‘free’
iv. Į > 0.5. Staying on at school and passing choices made by those in less advantaged classes.
the examination makes entry to the service Our second and third mechanisms can be un-
class more likely than entry to the working derstood as accentuating the differing patterns of
class.1 choice that derive from this initial source.

In the interests of realism, especially as re-

gards (ii) and (iii) above, it ought to be noted Relative Risk Aversion
that ‘leaving’ and entering the labour market We begin with an assumption regarding aspira-
need not in most educational systems be equated tions: that is, that families in both classes alike
with a definitive ending of the individual’s edu- seek to ensure, so far as they can, that their chil-
cational career. Taking this option could in fact dren acquire a class position at least as advanta-
lead to further vocational courses pursued in geous as that from which they originate or, in
conjunction with employment. other words, they seek to avoid downward social
mobility. This means that the educational strat-
egy pursued by parents in the service class is to
The Generation of Class Differentials maximize the chances of their children acquiring
Given the model previously outlined, we can a position in this class. In terms of our model
now turn to the question of explaining why dif- their strategy is to maximize the probability of
ferences exist across classes in the proportions of access to S*. For working-class parents the impli-
young people who make one kind of educational cation is that they should seek for their children
 Breen and Goldthorpe—Explaining Educational Differentials • 529

a place in either the working or the service class, Taking the first term on the left hand side of (3)
since either meets the criterion of being at least we have
as good as the class from which they originate. pa p p
In terms of our model their strategy is then to = 1 7 (3a)
g1 g
a 1 g1 + g2
maximize the probability of access to S* or W*,
which is the same as minimizing the probability by conditions (3) and (4). Taking the second
of access to U*. This establishes families in both term on the left hand side of (3) we have
classes as having identical relative risk aversion: 1 1 - p 2 b1 1 1 - p 2 1 b1 + b2 2
they want to avoid, for their children, any posi- g1
*
1 g1 + g2 2 (3b)
tion in life that is worse than the one from which
they start. by conditions (ii) and (iii). Together (3a) and
To see the consequence of these two strategies, (3b) imply (3) which in turn implies ps > pw as
maximize pr(S*) for those of service-class origins required.
and minimize pr(U*) for those of working-class This result establishes that if continuing in
origins—assume, for the moment, that continu- education is costless and there are no class dif-
ing in education is costless (c = 0). Then we find ferences in the subjective probability parameters
that whether or not a pupil believes it to be in Į, ȕ, Ȗ and π, children from middle-class back-
his or her best interests to continue in education grounds will more strongly ‘prefer’ (in the sense
rather than leave depends on the value pi (where of perceiving it to be in their best interests) to
i indicates the ith pupil) given by remain in school to a further level of education
pi a + 1 1 - pi 2 b1
rather than leave.
piS = (1) The proportions in each class who prefer to
pi a + 1 1 - pi 2 b1 + g1
stay are derived as follows. Assume that p has an
for the ith service-class pupil and unspecified distribution with means in each class
ps and pw and dispersion parameters ıpS and ıpW.
pi + 1 1 - pi 2 1 b1 + b2 2
piW = Because piS > piW for any common value of πi,
pi + 11 - pi 2 1 b1 + b2 2 + 1 g1 + g2 2
and assuming, for the moment, no class differ-
(2) ence in the distribution of π, it follows that ps >
pw. Then, given that only those pupils for whom
for the ith working-class pupil. Here we have p exceeds one-half prefer to stay at school, the
allowed π to vary between pupils but we have proportions in each class preferring this outcome
assumed the values of Į, ȕ and Ȗ to be common are given by the area under the unspecified dis-
to all. If p takes a value greater than one-half this tribution function above the point
indicates that the expected returns to remaining 1
2 - ps
at school exceed those of leaving. Thus, without zs =
spS
taking account, as yet, of the costs of pursuing
the former strategy, pupils for whom pi > 0.5, for the service class and analogously for the
can be said to prefer to remain in education. working class.
Even if subjective expectations of future success,
as captured by π, do not differ between the two
classes it will nevertheless be the case that, given Differences in Ability and
conditions (1) to (4), piS > piW for any value of π Expectations of Success
less than one.2 Thus far we have been assuming that the op-
Proof: piS > piW ‫ ׊‬π ≤ 1 if and only if tion of continuing in education is open to all
pupils. But, of course, this is often not the case
pa + 11 - p2 b1 p + 1 1 - p 2 1 b1 + b2 2
g1
7
1 g1 + g2 2
(3) and successive levels of education may only be
open to those who meet some criterion, such as
530 • Part VI: Who Gets Ahead?

a given level of performance in a previous exam- future examination performance to wish to con-
ination. Let us assume, for the sake of simplicity, tinue in education—then ability differences will
that this criterion can be expressed directly in be wholly captured in differences in the sub-
terms of ability, so that, for example, a pupil may jective parameter π. This will cause the average
only continue in education if his or her ability value of π to be lower among working- than
level exceeds some threshold, k: i.e. we impose service-class pupils because of the class difference
the condition that ai must be greater than k. Re- in average ability levels.
calling our assumption regarding primary effects
that the mean level of ability is higher in the
service class than in the working class but that Differences in Resources
both have the same variance in ability, it follows Thus far we have assumed education to be cost-
that the proportion of service-class children who less. If we relax this assumption we need to take
meet this condition exceeds the proportion of account of class differences in the resources that
working-class children. families in different locations in the class struc-
However, we might also suppose that pupils’ ture can devote to their children’s education.
own knowledge of their ability helps shape the Assume, therefore, that pupils can continue
subjective probability they attach to being suc- in education if and only if ri > c where ri is the
cessful in the next stage of education, which we level of resources available for children’s educa-
labelled πi. So we can write πi = g(ai), where g tion in the ith family. Given that service-class
indicates that π is a function of a. If we then families have, on average, greater resources than
denote by π*s and π*w the required minimum sub- working-class families (rs > rw) and that resources
jective probabilities compatible with continuing have the same dispersion within each class, it fol-
in education (these are the smallest values of πi lows that the proportion of service-class pupils
for which pi > 0.5) we can write the probability for whom this resource requirement is met will
of continuing in education as exceed the proportion of working-class pupils.
We have now suggested three mechanisms
pr(ai > k)pr(ʌi > ʌ* | ai > k) which, taken together, give rise to class differen-
tials in the proportions of children who choose
= pr(ai > k, ʌi > ʌ*) to stay on in education. Our first mechanism
shows how, solely because of the relative risk
= pr(ai > k, g(ai) > ʌ* (4) aversion that is seen as being common across
classes, there will be a stronger preference among
If service-class than working-class pupils for re-
maining in education given that no costs attach
pr(g(ai) > ʌ*) ” pr(ai > k) to doing so. Our second mechanism then allows
for class differences in average ability levels and
then (4) reduces to in turn in expectations of success. The effect of
this is to introduce class differences in the values
pr(ʌi > ʌ*). of π (the subjective probability of future educa-
tional success), which further widen class differ-
If pupils’ expectations about how well they ences in the value of p and thus in the strength
will perform at the next level of education are of the preference for staying on in education. Fi-
upwardly bounded by how well they have per- nally, our third mechanism takes account of the
formed in their most recent examination—for costs of continuing in education and allows for
example, if there are no pupils who, although a further source of class differentials, the average
they have failed to exceed the threshold k are resource levels available to meet these costs. The
nevertheless sufficiently optimistic about their effect of this is to promote class differences via
 Breen and Goldthorpe—Explaining Educational Differentials • 531

the proportion of families in each class whose odds ratio between the service and the working
resources exceed the costs of their children con- class is equal to
tinuing in education or, more simply, who can
afford to allow their children to continue. . . . FS > 11 - FS 2
FW > 11 - FW 2
Explaining Empirical
Generalizations where we use ࢥS to mean the proportion of
We may now seek to apply our model to the ex- service-class pupils who remain in education
planation of the empirical generalizations that and similarly for ࢥW . It is then possible to show
were set out in the Introduction, beginning with (see Breen and Goldthorpe 1997, 300–2) that,
that of the widely observed persistence of class given a decline over time in c, together with an
differentials in educational attainment in the increase in the proportion of both service- and
context of an overall increase in educational par- working-class pupils who consider it in their
ticipation rates. To account for the latter trend best interests to remain in education, the odds
is fairly easy: the relative costs of education have of continuing in education increase by a roughly
declined over time in all economically advanced constant amount for each class, and so preserve
countries. As the period of compulsory school- a similar constancy in the odds ratio. This tells
ing has been extended, the costs of successively us that, under these circumstances, a uniform
higher levels of education have been reduced decline in the costs of education, i.e. uniform
through the abolition of fees, the introduction of across classes, will result in the odds for children
maintenance grants, soft loans, etc. In our model of all classes choosing to continue being mul-
this change is treated via mechanism (iii)—class tiplied by something like a common factor. So
differences in resources—and is captured in a if, for example, some level of education is made
decline in the size of the parameter c. This will free of charge (in the sense that fees are no lon-
lead to an increase in the proportions of children ger levied) class differences in participation (as
from both service- and working-class origins measured by odds ratios) at this level will remain
continuing in education, providing, of course, more or less unchanged even though the overall
that the preference for continuing (given by our participation rate will increase.
p parameter) does not decline. However, far from Our model also sheds some new light on the
pi declining over time it is more plausible to be- concept of ‘maximally maintained inequality’
lieve that there is a widespread increase in the in education (Raftery and Hout 1990; Hout,
desire to remain in full-time education as edu- Raftery and Bell 1993). These authors argue
cational credentials come to take on increasing that class differences in educational attainment
importance in the labour market and in securing will only begin to decline when participation in
a relatively advantaged class position. Indeed, in a given level of education of children of more
so far as education is regarded as a ‘positional’ advantaged backgrounds reaches saturation. In
good, pi could be expected to rise steadily simply our model, such a reduction will occur once c
as a consequence of educational expansion itself. declines to the value at which all members of
At the same time our model can provide the service class have resources that exceed it. At
an explanation of how, within a context of ed- this point, all service-class families will possess
ucational expansion, class differentials may resources that exceed the costs of remaining in
nonetheless persist. To see this, recall that class education and thus the proportion in this class
differences in educational attainment are usu- who choose to continue in education will be
ally measured by odds ratios which compare the equal to the proportion who perceive it to be in
odds of continuing in education versus leaving their interests (i.e. for whom pi > 0.5). Further
for pairs of origin classes. Under our model, the reductions in c will then have no influence on
532 • Part VI: Who Gets Ahead?

the numbers of service-class children who choose evidence (for reviews, see Erikson and Jonsson
to continue but will still increase the proportion 1996a; Goldthorpe 1996) that in this same pe-
of working-class children who do so. Under riod the average income levels of different classes
these conditions, the relevant odds ratio could in Sweden became more equal, while the degree
be expected to move towards unity.3 However, of economic insecurity experienced by members
it should be recognized that, as understood in of the working class was steadily reduced. And
terms of our model, maximally maintained in- through time-series analysis, correlations can in
equality does not imply that a decline in class fact be established between these latter tenden-
differentials can only commence at the point at cies and the growing equality in educational out-
which all children of more advantaged class or- comes that are at all events consistent with the
igins continue in education. Rather, this effect hypothesis of a causal influence (Erikson 1996).
occurs once all such children whose pi is greater As against the constancy in class differentials
than one-half continue, in other words, once all in educational attainment, to which exceptions
those who perceive it to be in their best inter- are few, the decline in gender differentials that
ests to continue are able to act accordingly. It has occurred in virtually all advanced societies
is true that in some instances the achievement since the 1970s must appear as rather dramatic.
of this latter condition will, in fact, give rise to Because gender differentials arise within, rather
100 percent continuation among children of than between, families, neither changes in the
more advantaged classes.4 But further declines costs of education nor in inequalities in resources
in c, even if they lead to riW > c for all members among families are appropriate to explaining
of the working class, will not lead to equality in their reduction. In the light of our model, this
the proportions continuing in education in each may rather be seen as resulting from shifts in the
class so long as there still remains a class differ- perception of educational returns that have been
ence in the proportion who prefer to continue. prompted by changes in women’s labour market
It further follows from our model that class participation. It would be a fair comment to say
differentials in educational attainment will also that the pattern of returns to different educa-
respond to changes in the costs of education tional decisions that we have thus far envisaged
which, rather than being uniform, have a vari- would, for most of the 20th century, be more
able impact across classes. Such changes could applicable to young men than to young women.
be brought about directly through the selective Until quite recently, it is likely that educational
subsidization of young people according to decisions in the case of girls were shaped in the
their class of origin, as occurred, for example, in main by the expectation that their primary social
some post-war Communist societies. However, roles would be those of wife and mother, and
essentially the same effect could follow from a that their class positions would therefore be de-
general reduction in inequality of condition be- termined more by whom they married than by
tween classes. Specifically, if class differences in how they themselves fared in the labour market.
resources, r, become smaller, our model would In so far as this were the case, then the relative
predict that differentials in educational attain- returns to education for women would be some-
ment, as measured by odds-ratios, would in turn what different to those we have supposed in the
decline. exposition of our model: at all events, the re-
It is then in this way that the model may be turns associated with any particular educational
seen as applying to the national case that most decision would be less highly differentiated
obviously deviates from the typical pattern of than for men. So, for example, young women
persisting class inequalities in education, i.e. that of service-class origins could be thought best
of Sweden, in which, as earlier noted, a narrow- able to retain their position in this class through
ing of such inequalities over the post-war decades marriage; but to meet young service-class men
is well attested. There is indeed further extensive did not necessitate that they themselves should
 Breen and Goldthorpe—Explaining Educational Differentials • 533

acquire the educational qualifications that led to effects: gender differentials in educational attain-
a service-class occupation. Rather, their qualifi- ment should decline, as indeed they have, and at
cations had to be such as to provide them with the same time the magnitude of class differences
employment that would bring them into con- among women should increase.
tact with potential service-class husbands, and
this requirement might be met through only rel-
atively modest levels of educational attainment, Conclusions: Theoretical
leading to a job as, say, a secretary or nurse. And Implications of the Model
within both the home and the educational sys- As regards the theoretical implications of our
tem alike, as much emphasis was indeed placed model, we would see these as being of main sig-
on the acquisition of social and domestic skills nificance in their bearing on explanatory strat-
as on skills that would have value in the labour egy. The model represents children and their
market. families as acting in a (subjectively) rational
Such a flatter ‘gradient’ in the returns to dif- way, i.e. as choosing among the different edu-
ferent educational pathways would, if incorpo- cational options available to them on the basis
rated into our model, have two consequences. of evaluations of their costs and benefits and of
First, the proportion of women choosing to re- the perceived probabilities of more or less suc-
main in education at each decision point would cessful outcomes. It then accounts for stability,
be smaller than the proportion of men; and, or change, in the educational differentials that
second, class differentials would tend to be less ensue by reference to a quite limited range of
among women than among men. The former situational features. For example, in the case of
result follows from the lesser incentive to con- persisting class differentials, the explanatory em-
tinue in successively higher levels of education phasis falls on similarly persisting inequalities in
that would be held out to women of all class or- the resources that members of different classes
igins alike; the latter comes about because the can command in the face of the constraints and
magnitude of the class differences among those opportunities that their class positions typically
choosing to remain in education (for given val- entail. Class differences in demonstrated aca-
ues of ability and resources) is directly propor- demic ability are also recognized, but not—as
tional to the differences in returns associated we have emphasized—class differences of a (sub)
with the various possible educational outcomes. cultural character.
If we consider equations (1) and (2) shown ear- To the extent, then, that our model holds
lier, then as the difference between, say, Į1, ȕ1 good, i.e. that it can provide an adequate account
and Ȗ1 diminishes, so the difference between piS of the regularities we have considered and that its
and piW will also diminish. further empirical implications are not rejected—
Over the past 20 years, we would suggest, the the relatively parsimonious strategy of the ratio-
pattern of returns to education for women has nal action approach is supported; and, we might
drawn closer to that for men, as rates of wom- add, in an area in which ‘culturalist’ theories of
en’s labour market participation and, especially one kind or another have hitherto enjoyed great
rates for married women, have increased and as popularity—even if not great explanatory suc-
a woman’s own employment has taken on greater cess (see Goldthorpe 1996). In turn, the case for
significance in determining the standard of living attempting to pursue this strategy in other areas
enjoyed by her family and further, perhaps, her of sociological enquiry is strengthened.
own class position. In other words, our model as Finally, though, we would wish to allude to
expressed in Figure 60.1 has come increasingly certain theoretical implications that might be
to apply to women: the ‘gradient’ in their re- regarded as following from our model but that
turns to education has steepened. According to do not in fact do so. To begin with, we are not
our model, then, such a change should have two required to suppose that, in making educational
534 • Part VI: Who Gets Ahead?

choices, children and their parents in fact go as the relation of norms to rational action is con-
through the processes of ratiocination that the cerned (see Elster 1991). However, we do not in
model might appear to attribute to them. We this regard seek what Popper (1972: Ch. 8) has
do take it to be the case that the actors in ques- criticized as reduction by fiat, but only reduction
tion have some knowledge of how their society insofar as it is warranted by the empirical sup-
works, have some concern for their own, or for port that our theoretical arguments can obtain
family interests, and seek to use the former to in the particular area in which they have been
promote the latter. But we can at the same time applied.5
accept that the decisions they make may only
rarely result from any entirely explicit proce- NOTES
dures, rather than, say, ‘emerging’ over a period 1. Strictly speaking the mathematics of our model
require a slightly weaker condition, namely that Į ≥
of time and, in all probability, reflecting also
Ȗ1 / (Ȗ1 + Ȗ2). This imposes a condition on the mag-
various non-rational influences. What under-
nitude of the difference in the chances of access to
lies our approach is the idea that it is rational the service class as between remaining at school and
considerations that are, not the only, but the passing the examination and leaving immediately.
main common factor at work across individual The conditional probability of access to the service
instances, and that will therefore shape patterns class for those who leave immediately should not be
of educational choices in aggregate and, in turn, greater than Ȗ1 + Ȗ2 times the conditional probability
the regularities that constitute our explananda. of access to the service class for those who remain at
Our model then aims to represent these consid- school and pass the examination. However, because
erations in an ‘idealized’ way, so as to capture the of condition (iii), condition (iv) will always be met
if Į > 0.5.
key generative processes involved, rather than to
2. Note that, whereas piW can take any value be-
represent decision-making as it actually occurs at tween zero and one depending on the value of π, if
the level of particular families. ȕ1 ≥ Ȗ1 then piS will exceed one-half for all values of π.
Further, while we do not in explaining class 3. Though empirically this will be observed only if
differentials in education invoke systematic vari- the proportion of service-class children who consider
ation in values or derived norms, this does not it in their best interests to remain in education does
mean that we have to deny their very existence. not change for other reasons. For example, given an
Thus, in so far as class-specific norms may be increase over time in the importance of educational
identified—which is an empirical issue—we qualifications in obtaining jobs we might see changes
in the relative values of the Į, ȕ and Ȗ parameters
could recognize them as serving as guides to ra-
causing the proportion for whom pi > 0.5 to increase
tional action that have evolved over time out of
in both classes. Under these conditions a narrowing
distinctive class experience and that may substi- of the odds-ratio will not necessarily follow.
tute for detailed calculation when educational 4. In our model this will be the case for the ser-
choices arise. Understood in this way, such vice class if (in addition to conditions [i] to [iv]) ȕ1
norms could conceivably be of some explana- ≥ Ȗ1 but it need not be so if this inequality does not
tory significance as inertial forces in cases where hold.
the structure of constraints and opportunities 5. Elster (1991) criticizes several different ver-
or the distribution of resources is changing. But sions of the argument that action taken in con-
formity with social norms is reducible to rational
what we would in fact expect, and the decline in
action. However, his efforts to show that no version
gender differentials would, at least by analogy,
entails that such a reduction is always possible are
lend support, is that norms, in being essentially of greater philosophical than sociological interest.
epiphenomenal, would rather quickly come into One could entirely agree with Elster, yet still wish to
line with patterns of action that display a ratio- maintain that, in a particular instance of sociologi-
nal adaptation to the new circumstances that cal explanation, a reductionist view could in fact be
have come into being. upheld; or, more generally, that it is good strategy to
In sum, our model implies an explanatory start from a reductionist position and to modify it
strategy that is undoubtedly ‘reductionist’ so far only insofar as the evidence requires.
 Breen and Goldthorpe—Explaining Educational Differentials • 535

REFERENCES ———. (eds). 1996b. Can Education be Equalised?

Arum, R. and M. Hout. 1995. ‘The Early Returns: The Swedish Case in Comparative Perspective.
The Transition from School to Work in the Boulder: Westview Press.
United States.’ European University Institute, Goldthorpe, J.H. 1996. ‘Class Analysis and the
School-to-Work Conference, Florence. Reorientation of Class Theory: The Case of Per-
Boudon, R. 1974. Education, Opportunity and Social sisting Differentials in Educational Attainment.’
Inequality. New York: Wiley. British Journal of Sociology 45(3): 481–506.
Breen, R. and J.H. Goldthorpe. 1997. ‘Explaining Hout, M., A.E. Raftery and E.O. Bell. 1993. ‘Mak-
Educational Differentials. Towards a Formal Ra- ing the Grade: Educational Stratification in the
tional Action Theory.’ Rationality and Society 9: United States, 1925–1989.’ In Persistent Inequal-
275–305. ity: Changing Educational Attainment in Thir-
De Graaf, P.M. and H.G.B. Ganzeboom. 1993. teen Countries, eds Y. Shavit and H.-P. Blossfeld.
‘Family Background and Educational Attain- Boulder: Westview Press.
ment in the Netherlands for the 1891–1960 Jonsson, J.O., C. Mills and W. Müller. 1996. ‘A
Birth Cohorts.’ In Persistent Inequality: Changing Half Century of Increasing Educational Open-
Educational Attainment in Thirteen Countries, eds ness? Social Class, Gender and Educational At-
Y. Shavit and H.-P. Blossfeld. Boulder: Westview tainment in Sweden, Germany and Britain.’ In
Press. Can Education be Equalised? The Swedish Case in
Elster, J. 1991. ‘Rationality and Social Norms.’ Ar- Comparative Perspective, eds R. Erikson and J.O.
chives Européennes de Sociologie 32: 109–29. Jonsson. Boulder: Westview Press.
Erikson, R. 1996. ‘Can We Account for the Change Müller, W. and D. Haun. 1994. ‘Bildungsung-
in Inequality of Educational Opportunity?’ In leichheit im Sozialen Wandel.’ Kölner Zeitschrift
Can Education be Equalised? The Swedish Case in für Soziologie und Sozialpsychologie 46: 1–42.
Comparative Perspective, eds R. Erikson and J.O. Popper, K.R. 1972. Objective Knowledge. Oxford:
Jonsson. Boulder: Westview Press. Clarendon Press.
Erikson, R. and J.O. Jonsson. 1993. Ursprung Raftery, A.E. and M. Hout. 1990. ‘Maximally Main-
och Utbildning. Stockholm: Statens Offentliga tained Inequality: Expansion, Reform and Op-
Utredningar. portunity in Irish Education, 1921–1975.’ ISA
———. 1996a. ‘Explaining Class Inequality in Edu- Research Committee on Social Stratification and
cation: the Swedish Test Case.’ In Can Education Mobility, Madrid.
be Equalised? The Swedish Case in Comparative Shavit, Y. and H.-P. Blossfeld. (eds). 1993. Persistent
Perspective, eds R. Erikson and J.O. Jonsson. Inequality: Changing Educational Attainment in
Boulder: Westview Press. Thirteen Countries. Boulder: Westview Press.
536 • Part VI: Who Gets Ahead?

61 • Sean F. Reardon

The Widening Academic Achievement Gap

Between the Rich and the Poor

Introduction gap between high- and low-income families

The socioeconomic status of a child’s parents has has widened, has the achievement gap between
always been one of the strongest predictors of the children in high- and low-income families also
child’s academic achievement and educational at- widened?
tainment. The relationship between family socio-
economic characteristics and student achievement
is one of the most robust patterns in educational Data
scholarship, yet the causes and mechanisms of Assembling information on trends in the re-
this relationship have been the subject of consid- lationship between socioeconomic status and
erable disagreement and debate (see, for example, academic achievement requires examination of
Bowles and Gintis 1976, 2002; Brooks-Gunn multiple sources of data. I use data from nine-
and Duncan 1997; Duncan and Brooks-Gunn teen nationally representative studies, including
1997; Duncan, Brooks-Gunn, and Klebanov studies conducted by the National Center for
1994; Herrnstein and Murray 1994; Jacoby and Education Statistics (NCES), the Long-Term
Glauberman 1995; Lareau 1989, 2003). Trend and Main National Assessment of Edu-
Trends in socioeconomic achievement gaps— cational Progress (NAEP) studies, U.S. compo-
the achievement disparities between children nents of international studies, and other studies.
from high- and low-income families or between Although these studies vary in a number of ways,
children from families with high or low levels of each of them provides data on the math or read-
parental educational attainment—have received ing skills, or both, of nationally representative
[surprisingly little] attention. samples of students, together with some data on
The question [I pose] is whether and how students’ family socioeconomic characteristics,
that relationship between family socioeconomic such as family income, parental education, and
characteristics and academic achievement has parental occupation.1
changed during the last fifty years. In partic-
ular, I investigate the extent to which the ris-
ing income inequality of the last four decades Measuring Achievement Gaps
has been paralleled by a similar increase in the To compare the size of the achievement gap
income achievement gradient. As the income across studies, I report test-score differences

Sean F. Reardon. “The Widening Academic Achievement Gap Between the Rich and the Poor: New Evidence
and Possible Explanations,” in Whither Opportunity? Rising Inequality, Schools, and Children’s Life Chances (New
York: Russell Sage Foundation, 2011), pp. 91–116. Copyright © 2011 Russell Sage Foundation, 112 East 64th
Street, New York, NY 10065. Reprinted with permission.
 Reardon—The Widening Academic Achievement Gap Between the Rich and the Poor • 537

Figure 61.1. Trend in 90/10 Income Achievement Gap in Reading, by Birth Cohort (1943 to 2001 Cohorts)

1.50
Between 90th & 10th Income Percentile Families
Average Difference in Standardized Test Scores

Study
1.25 TALENT
NLS
HS&B
1.00 NLSY79
NELS
Add Health
0.75 Prospects
NLSY97
ELS
0.50 SECCYD
ECLS-K
ECLS-B
0.25 Quartic Fitted Trend
1943-2001
Fitted Trend
1974-2001
0.00

1940 1950 1960 1970 1980 1990 2000

Cohort Birth Year

Source: Author’s compilation based on data from Project Talent (Flanagan et al. n.d.); NLS, HS&B, NELS, ELS, ECLS-K,
ECLS-B (U.S. Department of Education, Center for Education Statistics 1999, 2000, 2001, 2004, 2009, 2010); Pros-
pects (U.S. Department of Education 1995); NLSY79, NLSY97 (U.S. Bureau of Labor Statistics 1980, 1999); SECCYD
(National Institute of Child Health and Human Development 2010); and Add Health (Harris 2009, reading only).
Note: See note 4 and online appendix for further details.

between groups in standard-deviation units, associations between family income and

adjusted for the estimated reliability of each achievement for measurement error in family
test. This is standard practice when comparing income.
achievement gaps measured with different tests Although each of the nineteen studies in-
(see, for example, Clotfelter, Ladd, and Vigdor cludes a measure of parental educational at-
2006; Fryer and Levitt 2004, 2006. So long as tainment, in some studies this is reported by
the true variance of achievement remains con- students, while in others it is reported by par-
stant over time, this allows valid comparisons in ents. Because reports of their parents’ education
the size of the gaps across different studies using are particularly unreliable for younger students,
different tests (see online appendix section for I include studies with student-reported paren-
technical details). tal education only if the students were in high
school themselves when reporting their parents’
educational attainment. As a measure of paren-
Measures of Socioeconomic Status tal educational attainment, I use the maximum
I rely on two key measures of socioeconomic of the mother’s and father’s attainment (or the
status: family income and parental educational attainment of the single parent in the home if
attainment. Each of the nineteen studies used both are not present).
includes information on parental educational
attainment; twelve of the studies include in-
formation on family income. Nine of the stud- Trends in Socioeconomic
ies include parent-reported family income, Status-Achievement Gradients
[and . . . ] three include student-reported in- To begin with, consider the difference in
come.2 In all studies, I adjust the estimated achievement between children from high- and
538 • Part VI: Who Gets Ahead?

Figure 61.2. Trend in 90/10 Income Achievement Gap in Math, by Birth Cohort (1943 to 2001 Cohorts)

1.50
Between 90th & 10th Income Percentile Families
Average Difference in Standardized Test Scores

Study
1.25 TALENT
NLS
HS&B
1.00 NLSY79
NELS
Add Health
0.75 Prospects
NLSY97
ELS
0.50 SECCYD
ECLS-K
ECLS-B
0.25 Quartic Fitted Trend
1943-2001
Fitted Trend
1974-2001
0.00

1940 1950 1960 1970 1980 1990 2000

Cohort Birth Year

Source: Author’s compilation based on data from Project Talent (Flanagan et al. n.d.); NLS, HS&B, NELS, ELS, ECLS-K,
ECLS-B (U.S. Department of Education, Center for Education Statistics 1999, 2000, 2001, 2004, 2009, 2010);
Prospects (U.S. Department of Education 1995); NLSY79, NLSY97 (U.S. Bureau of Labor Statistics 1980, 1999); and
SECCYD (National Institute of Child Health and Human Development 2010).
Note: See online appendix for further details.

low-income families. One way to measure this born from the 1950s through the mid-1970s,
difference is to compare the average math and and then grown steadily since the mid-1970s.
reading skills of children from families with in- There are, however, several reasons to suspect
comes at the 90th percentile of the family in- that the trend in the estimated gaps for the ear-
come distribution (about $160,000 in 2008) to liest cohorts, those born before 1970, is not as
those in families with incomes at the 10th per- accurately estimated as the later trend. For one
centile of the family income distribution (about thing, the quality of the achievement tests used
$17,500 in 2008).3 in the early studies may not have been as good as
Figures 61.1 and 61.2 present the estimated those used in the more recent studies. In addi-
90/10 income achievement gap for cohorts of tion, as I have noted, family income was reported
students born from the mid-1940s through by students rather than by a parent in three of
2001.4 These estimates are derived from the the early studies. Furthermore, [these studies]
twelve nationally representative studies available exclude [high school] dropouts, who are dispro-
that include family income as well as reading portionately low-income and low-achieving stu-
and/or math scores for school-age children. dents. Each of these factors might lead the gaps
Although the tests used are not exactly com- to be underestimated in the early cohorts relative
parable across all the studies included, both to later cohorts.
figures show a clear trend of increasing income Although the trend in achievement gaps prior
achievement gaps across cohorts born over a to 1970 is somewhat unclear, the trend from
nearly sixty-year period. The estimated income the mid-1970s to 2001 appears relatively clear.
achievement gaps among children born in 2001 Figures 61.1 and 61.2 include fitted trend lines
are roughly 75 percent larger than the estimated from 1974 to 2001 (the solid lines); these indi-
gaps among children born in the early 1940s. The cate that the income achievement gap has grown
gap appears to have grown among cohorts born in by roughly 40 to 50 percent within twenty-five
the 1940s and early 1950s, stabilized for cohorts years, a very sizable increase.
 Reardon—The Widening Academic Achievement Gap Between the Rich and the Poor • 539

Figure 61.3. Trends in Family-Income Inequality Among School-Age Children, 1967 to 2008 (Weighted
by Number of School-Age Children)

2.2
90/10 Income Ratio
50/10 Income Ratio
2.0
90/50 Income Ratio
Income Ratio (Relative to 1967)

1.8

1.6

1.4

1.2

1.0

0.8

1970 1980 1990 2000 2010

Year

Source: Author’s calculations, based on U.S. Bureau of the Census (King et al. 2010).
Note: Each line shows the trends in the ratio of household incomes at two percentiles of the income distribution. All
trends are divided by their value in 1967 in order to put the trends on a common scale.

achievement has grown substantially stronger

How Large Are These Gaps? during the last half-century. In the following
Figures 61.1 and 61.2 report income gaps in section I discuss four broad possible explana-
standard-deviation units. Although this is a met- tions for this increase: (1) income inequality
ric familiar to researchers and one that is useful has grown during the last forty years, meaning
for comparing the size of gaps across studies using that the income difference between families at
different tests, it may not be immediately obvi- the 90th and 10th percentiles of the income
ous how large these gaps are in substantive terms. distribution has grown; (2) family investment
One way to get a sense of the size of the gaps is patterns have changed differentially during the
to compare them to the amount that an average last half-century, so that high-income families
student learns during the course of a year. Data now invest relatively more time and resources in
from the NAEP indicate that the average student their children’s cognitive development than do
gains 1.2 to 1.5 standard deviations in math and lower-income families; (3) income has grown
reading between fourth and eighth grade and be- more strongly correlated with other socioeco-
tween 0.6 and 0.7 standard deviations in math nomic characteristics of families, meaning that
and reading between eighth and twelfth grade.5 high-income families increasingly have greater
Thus, a gap of 1 standard deviation is substan- socioeconomic and social resources that may
tively very large, corresponding to roughly 3 to 6 benefit their children; and (4) increasing income
years of learning in middle or high school. [ . . . ] segregation has led to greater differentiation in
school quality and schooling opportunities be-
tween the rich and the poor.
Why Has the Income Achievement
Gap Grown?
The evidence thus far indicates that the rela- Rising Income Inequality
tionship between a family’s position in the in- Income inequality in the United States has
come distribution and their children’s academic grown substantially in the last four decades and
540 • Part VI: Who Gets Ahead?

as of 2007 was at a level similar to the levels in If the increasing income achievement gap
1925 to 1940, when U.S. income inequality was is driven by increasing income inequality, we
at its twentieth-century peak (Burkhauser et al. would expect that gap to grow most sharply
2009; Piketty and Saez 2003, 2008).6 between students at the 50th and 10th percen-
If rising income inequality is responsible for tiles of the family income distribution from the
the growth in the income achievement gap, we 1960s through the 1980s (or for cohorts born
would expect to see that gap grow in a pattern in these years), and then to grow among those
similar to the growth in income inequality. To at the high end of the income distribution after
investigate this, consider the trends in measures that. Moreover, because the 50/10 ratio is larger
of family income inequality illustrated in Figure than the 90/50 ratio, we might expect the 50/10
61.3, which shows the changes in the 90/10 income achievement gap to be larger than the
family income ratio (the ratio of the family in- 90/50 income achievement gap as well.8 Figures
come of the child at the 90th percentile of the 61.4 and 61.5 display the estimated 90/50 and
family income distribution to that of the child 50/10 income achievement gaps for each of the
at the 10th percentile), the 90/50 family income studies with income data.
ratio, and the 50/10 family income ratio among Figures 61.4 and 61.5 do not exactly conform
school-age children from 1967 to 2008.7 to what we would expect if the growing income
Several key trends are evident in Figure 61.3. achievement gap were simply due to rising in-
First, the 90/10 family income ratio grew rap- come inequality among families with school-age
idly from 1967 to the early 1990s, more than children. Although the 50/10 income achieve-
doubling in twenty-five years. In 1967, the fam- ment gap in reading is generally larger than
ily income of the child at the 90th percentile of the 90/50 income achievement gap for cohorts
the family income distribution was 4.6 times born before 1990, the gaps are roughly similar
greater than that of the child at the 10th per- in size in math, and the 90/50 gap is actually
centile; in 1993 this 90/10 ratio was 9.9. After equal or larger than the 50/10 gap in the most
1993, the 90/10 ratio declined to 8.6 in 2000 recent cohorts. Moreover, the 90/50 gap appears
before climbing again to 9.9 by 2005. Second, to have grown faster than the 50/10 gap during
the growth in the ratio of the incomes in the the 1970s and 1980s, the opposite of what we
90th to those in the 10th percentiles from 1967 would predict on the basis of the rates of growth
to 1993 was driven largely by a rapid increase in of the 90/50 and 50/10 income ratios (indeed,
the 50/10 family income ratio, which grew from the 50/10 gap in reading appears to have been
2.5 in 1967 to 4.1 in 1987, a 64 percent increase basically flat through this time period, when the
in twenty years. After the late 1980s, however, 50/10 income ratio was growing most rapidly).
the 50/10 family income ratio leveled off and In sum, Figures 61.4 and 61.5 do not provide
then declined to 3.6 by 2002. Third, the ratio much support for the idea that the growing in-
between the 90th and the 50th family income come achievement gap is attributable to rising
percentiles grew steadily from the early 1970s income inequality, at least not in any simple
through 2008, increasing from 1.8 in 1974 to sense. Nor, however, do they rule out the possi-
2.5 in 2005, an increase of 36 percent. Thus, bility that rising income inequality has contrib-
from the late 1960s through the late 1980s, the uted to the rising income achievement gap.
increase in lower-tail family income inequal-
ity was largely responsible for the increase in
the ratio between the incomes of the 90th and Differential Investments in Children’s
10th percentiles. After the late 1980s, however, Cognitive Development
increasing upper-tail inequality and decreasing Families may be changing how they invest in
lower-tail inequality largely offset one another their children’s cognitive development. If so,
for the next twenty years. this may explain some of the rising income
 541

Figure 61.4. Trend in 90/50 and 50/10 Income Achievement Gap, Reading, by Birth Year (1943 to 2001
Cohorts)

1.00
Average Difference in Standardized Test Scores

Study
TALENT
NLS
0.75
HS&B
NLSY79
NELS
Add Health
0.50 Prospects
NLSY97
ELS
SECCYD
ECLS-K
0.25
ECLS-B
Fitted Trend
90/50 Gap
Fitted Trend
50/10 Gap
0.00

1940 1950 1960 1970 1980 1990 2000

Cohort Birth Year

Source: Author’s compilation based on data from Project Talent (Flanagan et al. n.d.); NLS, HS&.B, NELS, ELS, ECIS-K,
ECLS-B (U.S. Department of Education, Center for Education Statistics 1999, 2000, 2001, 2004, 2009, 2010); Pros-
pects (U.S. Department of Education 1995); NLSY79, NLSY97 (U.S. Bureau of Labor Statistics 1980, 1999); SECCYD
(National Institute of Child Health and Human Development 2010); and Add Health (Harris 2009, reading only).
Note: Solid symbols represent 90/50 income achievement gaps; hollow symbols represent 50/10 income achievement gaps.

Figure 61.5. Trends in 90/50 and 50/10 Income Achievement Gap in Math, by Birth Year (1943 to 2001
Cohorts)

1.00
Average Difference in Standardized Test Scores

Study
TALENT
NLS
0.75
HS&B
NLSY79
NELS
Add Health
0.50 Prospects
NLSY97
ELS
SECCYD
ECLS-K
0.25
ECLS-B
Fitted Trend
90/50 Gap
Fitted Trend
50/10 Gap
0.00

1940 1950 1960 1970 1980 1990 2000

Cohort Birth Year

Source: Author’s compilation based on data from Project Talent (Flanagan et al. n.d.); NLS, HS&B, NELS, ELS, ECLS-K,
ECLS-B (U.S. Department of Education, Center for Education Statistics 1999, 2000, 2001, 2004, 2009, 2010);
Prospects (U.S. Department of Education 1995); NLSY79, NLSY97 (U.S. Bureau of Labor Statistics 1980, 1999); and
SECCYD (National Institute of Child Health and Human Development 2010).
Note: Solid symbols represent 90/50 income achievement gaps; hollow symbols represent 50/10 income achievement gaps.
542 • Part VI: Who Gets Ahead?

achievement gap. Sociologists and historians of do less-educated and lower-income children

the family have argued that parents, particularly (Guryan, Hurst, and Kearney 2008; Ramey and
those in the middle class, have become increas- Ramey 2010). Moreover, the amount of time
ingly focused on children’s cognitive develop- parents spend in child-care activities (broadly
ment during the last fifty years (Lareau 1989; defined) has increased from 1965 to 2008 and
Schaub 2010; Wrigley 1989). has increased more for college-educated parents
Another factor that may contribute to par- than for less-educated parents (Bianchi 2000;
ents’ increasing focus on their children’s cog- Ramey and Ramey 2010). In addition, in a re-
nitive development is the rise of test-based cent paper using data from the Consumer Ex-
accountability systems in education. Although penditure Survey, Sabino Kornrich and Frank
some forms of standardized testing, including IQ Furstenburg (2010) find that families’ spending
tests and the SAT, have been prevalent for much on children increased substantially from 1972
of the twentieth century (Lemann 1999), stan- to 2007, particularly among high-income and
dardized achievement testing has become much college-educated families. Spending increases
more common with the rise of the accountability were particularly sharp among families with
movement following the 1983 publication of A preschool-age children. Consistent with this is
Nation at Risk (National Commission on Excel- evidence that the relationship between family
lence in Education 1983). The combination of income and preschool enrollment among three-
the increasing importance of educational success and four-year-old children grew from the late
in determining earnings (Levy and Murnane 1960s to the late 1980s (Bainbridge et al. 2005).
1992) and the increasing importance of test These patterns are broadly consistent with the
scores in defining educational success may have hypothesis that the rising income achievement
caused parents to focus more on their children’s gap is at least partly driven by the increasing
cognitive development. investment of upper-income families in their
Although both middle-class and low-income children’s cognitive development, particularly
parents may have become increasingly aware of during the preschool years, though the evidence
the intellectual development of their children, is far from conclusive on this point.
Annette Lareau (1989, 2003) argues that mid-
dle- and upper-class parents engage much more
commonly in what she calls “concerted culti- Changes in the Relationships among
vation”—the deliberate organization of child- Family Income, Family Socioeconomic
hood around intellectual and socioemotional Characteristics, and Children’s
development. If this concerted cultivation is Achievement
effective at improving children’s intellectual Another possible explanation for the rising
skills—at least, those measured by standard- income achievement gap is that high-income
ized tests—then this may contribute to the families not only have more income than
rising income achievement gap. If middle- and low-income families but also have access to a
upper-income families are increasingly likely to range of other family and social resources. On
invest in their children’s cognitive development, average, families with higher incomes tend to be
we would expect to see evidence of this in the those in which the parent(s) are highly educated.
trends in parental investment in children’s child This has long been true, though the link be-
care, education, and education-related activi- tween parental educational attainment and fam-
ties. There is, however, little available evidence ily income has grown stronger in recent decades,
with which to test this hypothesis. Studies of as the wage returns to educational attainment
parental time use show that highly educated have increased since 1979 (Levy and Murnane
and higher-income parents spend more time 1992). Because highly educated parents are
in child-care activities with their children than more able and more likely than less-educated
 Reardon—The Widening Academic Achievement Gap Between the Rich and the Poor • 543

Figure 61.6. Trend in Correlation Between Parental Education and Family Income (1943 to 2001 Cohorts)

Source: Author’s compilation based on data from Project Talent (Flanagan et al. n.d.); NLS, HS&B, NELS, ELS, ECLS-K,
ECLS-B (U.S. Department of Education, Center for Education Statistics 1999, 2000, 2001, 2004, 2009, 2010); Pros-
pects (U.S. Department of Education 1995); NLSY79, NLSY97 (U.S. Bureau of Labor Statistics 1980, 1999); SECCYD
(National Institute of Child Health and Human Development 2010); and Add Health (Harris 2009, reading only).

parents to provide resources and opportunities The trend in the correlation between family
for their children to develop cognitive and ac- income and parental education is illustrated in
ademic skills in both the preschool years and Figure 61.6, which shows a relatively unambig-
the school-age years (Lareau 1989), children of uous trend of increasing correlation between
parents with college degrees may have higher ac- parental education and family income across
ademic achievement, on average, than children cohorts.9 There are several possible explanations
of parents with lower levels of education, all else for this trend. First, as Frank Levy and Richard
being equal. Thus, the income achievement gap Murnane (1992) point out, changes in the struc-
may be partly a result of the effects of parental ture of the economy and the composition of the
educational attainment. labor force during the 1970s and 1980s, along
This argument suggests two possible expla- with declines in the real minimum wage and the
nations for the rising income achievement gap. weakening of unions, resulted in a decline in
First, the trend may result from an increase in the real wages of those with only a high-school
the correlation between parental educational degree and an increase in the wage premium
attainment and family income—which would for a college degree. These changes would be
mean that high- and low-income families are reflected in the studies of cohorts born in the
increasingly differentiated by education levels, 1950s through the 1970s because these students
leading to larger differences in children’s achieve- and their parents were surveyed in the 1970s and
ment. Second, the trend may derive from an 1980s. It is not clear, however, whether this ex-
increase in the achievement returns to parental planation can account for the continued increase
education, net of income. This would mean that in the correlation between income and educa-
children of highly educated parents benefit more tion for studies conducted after the 1980s.
from their parents’ educational attainment than A second possible reason for increasing cor-
they did in the past. relation between parental education and income
544 • Part VI: Who Gets Ahead?

is the increasing polarization of families. Sara Mc- parental education, and the association between
Lanahan (2004) argues that trends since 1960 in parental education and achievement, controlling
family structure and composition have led to an for family income. These partial associations are
increasingly polarized distribution of family con- shown in Figures 61.7 and 61.8.
texts for children—mothers with low levels of The key result evident in Figures 61.7 and
education are increasingly likely to be young, un- 61.8 is that the income coefficient grew steeply
employed, and single or divorced; mothers with for cohorts born from the 1940s to 2000. The
high levels of education are, conversely, increas- income coefficient for reading increased fourfold
ingly likely to be older, employed, and married. during this period, and it more than doubled for
As a result, the correlation of parental education math. At the same time, the parental-education
and income among families with children is likely coefficient has been generally unchanged during
to increase with time. Moreover, McLanahan ar- the six decades of cohorts in the studies. Even
gues, this polarization in family structure implies if we focus only on the cohorts born since the
a corresponding polarization in key resources mid-1970s, the income coefficient has increased
(income, parental time) available for children, substantially, more than doubling in reading and
which may have important implications for the increasing more than 50 percent in math. In this
distribution of children’s academic achievement. same time period, the coefficient on educational
Related to this argument is the fact that mar- attainment appears to have grown as well, albeit
ital homogamy (the tendency for individuals to at a slower rate.
marry those with similar levels of educational It is instructive to compare the trends in Fig-
attainment) has increased substantially since ures 61.7 and 61.8 with those in Figures 61.1
1960 (Schwartz and Mare 2005). As a result, and 61.2. Parental education accounts for a large
in two-parent families, the educational attain- proportion of the association between income
ment of the higher-educated parent is increas- and achievement in the early cohorts, but that
ingly predictive of the educational attainment proportion declines across cohorts. In reading,
of the less educated spouse. This trend, coupled for example, parental education accounts for
with the increasing disparity in single parent- roughly 60 to 80 percent of the income achieve-
hood and employment between mothers with ment gap in the studies of cohorts born in the
high and low levels of education described by 1940s, 1950s, and 1960s. But among cohorts
McLanahan, and the increasing wage premium born between 1980 and 2001, parental educa-
to education described by Levy and Murnane, tion and race explain only 40 to 60 percent of
implies that children with one highly educated the income gap. This pattern is at odds with the
parent are increasingly likely to have two highly explanation that the growing income gap is due
educated, married parents and a high family to the increasing correlation of income and pa-
income, while children with one less-educated rental education: all else being equal, we would
parent are increasingly likely to live either with expect the increasing correlation between the
a single mother or with two parents, both with two to mean that education should explain more
low levels of education and low wages. of the income gap over time, not less.
Because income and parental education are A second lesson evident in Figures 61.7 and
correlated, and increasingly so with time, as 61.8 is that the association between parental ed-
shown in Figure 61.6, I conduct a set of analyses ucation and children’s academic achievement,
to determine whether the growth in the income controlling for family income and race, remains
achievement gap is due to increases in the as- larger than the association between family in-
sociation between income and achievement or come and achievement, controlling for parental
parental education and achievement. For each education and race. That is, although the asso-
study with measures of both family income and ciation between income and achievement has
parental education, I estimate the association be- grown rapidly during the last fifty years, parental
tween income and achievement, controlling for educational attainment is still a more powerful
 545

Figure 61.7. Estimated Partial Associations Between Reading Test Scores and Both Income and Parental
Education, by Birth Cohort (1943 to 2001 Cohorts)

Study
Regression-Adjusted 90/10 Achievement Gap

1.00
TALENT
NLS
HS&B
0.80
NLSY79
NELS
Add Health
0.60 Prospects
NLSY97
ELS
0.40 SECCYD
ECLS-K
ECLS-B
0.20 Income
Coefficient
Parental
Education
Coefficient
0.00

1940 1950 1960 1970 1980 1990 2000

Cohort Birth Year

Source: Author’s compilation based on data from Project Talent (Flanagan et al. n.d.); NLS, HS&B, NELS, ELS, ECLS-K,
ECLS-B (U.S. Department of Education, Center for Education Statistics 1999, 2000, 2001, 2004, 2009, 2010; Prospects
(U.S. Department of Education 1995); NLSY79, NLSY97 (U.S. Bureau of Labor Statistics 1980,1999); and SECCYD
(National Institute of Child Health and Human Development 2010).
Note: Solid symbols represent regression-adjusted 90/10 income coefficients; hollow symbols represent regression-adjusted
parental education coefficients.

Figure 61.8. Estimated Partial Associations Between Math Test Scores and Both Income and Parental
Education, by Birth Cohort (1943 to 2001 Cohorts)

Study
Regression-Adjusted 90/10 Achievement Gap

1.00
TALENT
NLS
HS&B
0.80
NLSY79
NELS
Add Health
0.60 Prospects
NLSY97
ELS
0.40 SECCYD
ECLS-K
ECLS-B
0.20 Income
Coefficient
Parental
Education
Coefficient
0.00

1940 1950 1960 1970 1980 1990 2000

Cohort Birth Year

Source: Author’s compilation based on data from Project Talent (Flanagan et al. n.d.); NLS, HS&B, NELS, ELS, ECLS-K,
ECLS-B (U.S. Department of Education, Center for Education Statistics 1999, 2000, 2001, 2004, 2009, 2010);
Prospects (U.S. Department of Education 1995; NLSY79, NLSY97 (U.S. Bureau of Labor Statistics 1980, 1999); and
SECCYD (National Institute of Child Health and Human Development 2010).
Note: Solid symbols represent regression-adjusted 90/10 income coefficients; hollow symbols represent regression-adjusted
parental education coefficients.
546 • Part VI: Who Gets Ahead?

predictor of student achievement than is family students perform better on math tests after
income.10 moving to low-poverty schools (Schwartz 2010).
Likewise, some studies of peer effects find ev-
idence that the academic-achievement level of
Increased Segregation by Income one’s classmates may impact one’s own achieve-
A final possible explanation for the rising income ment (for recent evidence, see Lavy, Silma, and
achievement gap is the pattern of increasing in- Weinhardt 2009). Nonetheless, the evidence is
come segregation during the last forty years. far from clear if, how, and how much differences
Several recent studies have found that residential among schools in peers and school quality may
segregation by income increased from 1970 to affect achievement. As a result, there is little ev-
2000, partly as a result of rising income inequal- idence to answer the question of whether rising
ity and likely partly as a result of low-income income segregation has played a role in the in-
housing policy (Jargowsky 1996; Reardon and creasing income achievement gap.
Bischoff 2011; Watson 2009). In particular,
rising income inequality has led to the increas-
ing segregation of high-income families from Conclusion
middle- and low-income families; high-income Most of the evidence presented in this chap-
families increasingly live spatially far from the ter suggests that the achievement gap between
middle class (Reardon and Bischoff 2011). Be- children from high- and low-income families
cause residential patterns are closely linked to has grown substantially in recent decades. The
school-attendance patterns, the rise of residential income achievement gap is now considerably
income segregation has likely led to a concurrent larger than the black-white gap, a reversal of the
rise in school segregation by income, though pattern fifty years ago. At the same time, income
there is little empirical evidence on this.11 Be- inequality in the United States began to grow
cause the growth in income segregation has been sharply in the 1970s, a trend that continues
largely a result of increasing segregation of the to the present. The gap between the rich and
affluent, this might explain the pattern of the the poor has widened significantly, particularly
rising association between income and achieve- among families with children.
ment among higher-income families. Many of the other patterns in this chapter
Greater residential income segregation are not fully consistent with the simple expla-
may affect the school-quality differential be- nation that income inequality has driven these
tween high- and low-income students, because trends. First, the analyses described in the chap-
high-income parents are better able to garner ter show that the income achievement gaps do
resources for their schools. Likewise, increased not grow in the ways that would be predicted
income segregation may lead to less variance by the changes in income inequality. Although
of test scores within schools and more vari- income inequality grew sharply for families with
ance of test scores between schools, given that below-median incomes during the 1970s and
higher-income students generally have higher 1980s, the income achievement gap among chil-
scores than lower-income students.12 dren from these families was largely unchanged.
It is not clear, however, that these factors would The achievement gap did grow among children
lead to increases in the income achievement gap. from above-median-income families, but this
The evidence on the effects of school socioeco- appears to be better explained by an increase in
nomic composition is somewhat weak, though the association between income and achieve-
a new study taking advantage of quasi-random ment, not by increases in income inequality. Ev-
variation in school poverty rates experienced by idence from other studies suggests that parental
low-income students in Montgomery County, investment in their children’s cognitive devel-
Maryland, finds evidence that low-income opment has grown during the last half-century,
 Reardon—The Widening Academic Achievement Gap Between the Rich and the Poor • 547

particularly for higher-income families, a pattern Opportunity (Prospects), the National Longitudi-
that may explain the growing returns to income nal Study of Adolescent Health (Add Health), the
during this time period. Longitudinal Survey of American Youth (LSAY), the
There are a number of other possible expla- NICHD Study of Early Child Care and Youth De-
velopment (SECCYD), the Equality of Educational
nations for the evident trends in the income
Opportunity study (EEO), and Project Talent. For
achievement gap. Education policy increasingly further information about each study, see http://rus-
focuses on standardized-test scores as outcome slesage.org/duncan_murnane-online-appendix.pdf.
measures for schools; as these scores become more 2. The names of these studies are provided in full
important, families may be increasingly likely to in note 1. Although HS&B includes parent’s reported
invest in improving their children’s scores. Like- family income for a subsample of roughly 15 percent
wise, cultural perceptions of the role of parents of the full sample, the measure of family income ap-
have changed throughout the twentieth century pears highly unreliable (see online appendix 5.A2 for
to focus increasingly on early-childhood cogni- detail). I rely instead on the student-reported family
income measure for HS&B, as described in online
tive and psychological development, which may
appendix 5.A2. NLSY79 includes parent-reported
lead parents with resources to invest more in income for subjects who live with their parents; I
their young children’s development. use only the sample of sixteen- to eighteen-year-olds
In sum, the forces at work behind the rising from NLSY79 for this reason.
income achievement gap are likely complex and 3. My calculations, based on 2009 Current Popu-
interconnected. At the same time that family in- lation Survey data. See online appendix section 5.A3
come has become more predictive of children’s for details.
academic achievement, so have educational at- 4. Figures 61.1 and 61.2 display estimated 90/10
tainment and cognitive skills become more pre- income achievement gaps from all available nation-
ally representative studies that include reading- or
dictive of adults’ earnings. The combination of
math-achievement test scores for school-age children
these trends creates a feedback mechanism that
and family income. Labels indicate the modal grade
may decrease intergenerational mobility. As the in which students were tested in a given sample. For
children of the rich do better in school, and most of the longitudinal studies (HS&B, NELS,
those who do better in school are more likely to Prospects, ELS, and ECLS-K), only estimates from
become rich, we risk producing an even more the initial wave of the study are included. ECLS-B
unequal and economically polarized society. estimates come from wave 4, when children were
five years old and tested on school readiness; SEC-
NOTES CYD come from wave 5, when children were in
Online appendix available at: http://www.russell- third grade and were first administered a broad
sage.org/duncan_murnane_online_appendix.pdf. academic achievement test. The quartic fitted re-
1. The included NCES studies are the National gression line is weighted by the inverse of the sam-
Longitudinal Study (NLS), High School and Be- pling variance of each estimate. Included studies are
yond (HS&B), the National Education Longitu- Project Talent, NLS, HS&B, NLSY79, NELS, Add
dinal Study (NELS), the Education Longitudinal Health (reading only), Prospects, NLSY97, ELS,
Study (ELS), the Early Childhood Longitudinal SECCYD, ECLS-K, and ECLS-B. Family income
Study, Kindergarten Cohort (ECLS-K), and the is student-reported in Project Talent, NLS, and
Early Childhood Longitudinal Study, Birth Cohort HS&B. See online appendix for details on compu-
(ECLS-B). The included international studies are the tation of 90/10 gaps.
Third International Mathematics and Science Study 5. My calculations, based on Main NAEP math
(TIMSS), the Program of International Assessment and reading scores. See National Center for Edu-
(PISA), and the Progress in International Reading cation Statistics website, available at: http://nces.
Study (PIRLS). The additional included studies are ed.gov/nationsreportcard/naepdata/dataset.aspx (ac-
the National Longitudinal Survey of Youth: 1979 cessed March 7, 2011).
(NLSY79), the National Longitudinal Survey of 6. Figure 5.A12 in the online appendix displays
Youth: 1997 (NLSY97), Prospects: The Congressio- the trend in U.S. income inequality throughout the
nally Mandated Study of Educational Growth and last century.
548 • Part VI: Who Gets Ahead?

7. My calculations, based on Current Population Three- to Five-Year-Olds from 1968 to 2000.”

Survey, 1968–2009. See appendix section 5.A3 for Social Science Quarterly 86(3): 724–45.
details. Bianchi, Susan. 2000. “Maternal Employment and
8. We would expect this if we thought the rela- Time with Children: Dramatic Change or Sur-
tionship between achievement and log income was prising Continuity?” Demography 37(4): 401–14.
linear, which may not be the case. See online appen- Bowles, S., and H. Gintis. 1976. Schooling in Capi-
dix section 5.A6 for discussion. talist America: Educational Reform and the Contra-
9. The same trend is evident if the correlations dictions of Economic Life. New York: Basic Books.
are plotted against the year of the study rather than ———. 2002. “The Inheritance of Inequality.” Jour-
against birth year. nal of Economic Perspectives 16(3): 3–30.
10. The income coefficients displayed in Figures Brooks-Gunn, Jeanne, and Greg Duncan. 1997.
61.6 and 61.7 are roughly 20 to 40 percent the size “The Effects of Poverty on Children.” Future of
of the parental-education coefficients in the earliest Children 7(2): 55–71.
cohorts, but they are 60 to 90 percent the size of the Burkhauser, R. V., A. Feng, S. P. Jenkins, and J. Lar-
parental-education coefficients in the later cohorts. rimore. 2009. “Recent Trends in Top Income
The income coefficients here are adjusted for the es- Shares in the USA: Reconciling Estimates from
timated reliability of family income, so these differ- March CPS and IRS Tax Return Data.” NBER
ences in the magnitudes of the income and education Working Paper No. 15320. Cambridge, Mass.:
coefficients are likely not substantially biased by the National Bureau of Economic Research.
less reliable measurement of family income. Figure Clotfelter, C. T., H. F. Ladd, and J. L. Vigdor. 2006.
61.6 shows the Spearman rank-order correlation be- “The Academic Achievement Gap in Grades
tween parental educational attainment (coded as the Three to Eight.” NBER Working Paper No.
maximum level of educational attainment of both 12207. Cambridge, Mass.: National Bureau of
parents, if two are present in the home) and family Economic Research.
income. Because both income and parental education Duncan, Greg J., and Jeanne Brooks-Gunn. 1997.
are measured by ordered categories in most studies “Income Effects Across the Life-Span: Integra-
(parental education is measured in four to eight cat- tion and Interpretation.” In Consequences of
egories; income in five to fifteen categories), I com- Growing Up Poor, edited by Greg J. Duncan and
pute the rank-order correlation between income and Jeanne Brooks-Gunn. New York: Russell Sage
parental education for each of the twelve studies with Foundation.
measures of both income and parental education. Duncan, Greg J., Jeanne Brooks-Gunn, and P. K.
Correlations are disattenuated for estimated mea- Klebanov. 1994. “Economic Deprivation and
surement error in both family income and parental Early Childhood Development.” Child Develop-
educational attainment. Note that because these are ment 65(2): 296–318.
rank-order correlations, they are not directly compa- Dupriez, V., and X. Dumay. 2006. “Inequalities in
rable to standard (Pearson) correlation coefficients. School Systems: Effect of School Structure or of
11. Because of the relatively small within-school Society Structure? Comparative Education 42(2):
samples in many of the studies that include measures 243–60.
of family income, it is difficult to assess the trends in Dura-Bellat, M., and B. Suchaut. 2005. “Organisa-
school income segregation using the data available. tion and Context, Efficiency and Equity of Edu-
12. An examination (not shown) of the intraclus- cational Systems: What PISA Tells Us.” European
ter correlations of test scores from the school-based Educational Research Journal 4(3): 181–94.
studies included in this chapter provides some ev- Flanagan, John C., David V. Tiedeman, William
idence that the intracluster correlation has grown V. Clemans, and Lauress L. Wise. n.d. Project
with time, but these estimates are very noisy because Talent Public Use File, 1960–1976 [Computer
of the small sample sizes within each school in most file]. ICPSR07823-v1. Ann Arbor, Mich.: In-
of the studies. teruniversity Consortium for Political and Social
Research [distributor]. Ordering information
REFERENCES available at http://www.projecttalent.org/contact
Bainbridge, J., M. K. Meyers, S. Tanaka, and Jane (accessed June 24, 2011).
Waldfogel. 2005. “Who Gets an Early Edu- Fryer, R. G., and S. D. Levitt. 2004. “Understanding
cation? Family Income and the Enrollment of the Black-White Test Score Gap in the First Two
 Reardon—The Widening Academic Achievement Gap Between the Rich and the Poor • 549

Years of School.” Review of Economics and Statis- Lavy, V., O. Silma, and F. Weinhardt. 2009. “The
tics 86(2): 447–64. Good, the Bad, and the Average: Evidence on
———. 2006. “The Black-White Test Score Gap the Scale and Nature of Peer Effects in Schools.”
Through Third Grade.” American Law and Eco- NBER Working Paper No. 15600. Cambridge,
nomics Review 8(2): 249–81. Mass.: National Bureau of Economic Research.
Guryan, E. Hurst, and M. Kearney. 2008. “Parental Lemann, N. 1999. The Big Test: The Secret History
Education and Parental Time with Children.” of The American Meritocracy. New York: Farrar,
Journal of Economic Perspectives 22(3): 23–46. Straus and Giroux.
Harris, Kathleen Mullan. 2009. The National Longi- Levy, Frank, and Richard J. Murnane. 1992. “U.S.
tudinal Study of Adolescent Health (Add Health), Earnings Levels and Earnings Inequality: A
Waves I & II, 1994–1996; Wave III, 2001–2002; Review of Recent Trends and Proposed Expla-
Wave IV, 2007–2009 [machine-readable data file nations.”Journal of Economic Literature 30(3):
and documentation]. Chapel Hill, N.C.: Car- 1333–81.
olina Population Center, University of North Marks, G. N. 2005. “Cross-National Differences
Carolina at Chapel Hill. Ordering information and Accounting for Social Class Inequalities
available at http://www.cpc.unc.edu/projects/ in Education.” International Sociology 20(4):
addhealth/data/restricteduse (accessed June 24, 483–505.
2011). McLanahan, Sara. 2004. “Diverging Destinies: How
Herrnstein, R. J., and C. Murray. 1994. The Bell Children Are Faring Under the Second Demo-
Curve: Intelligence and Class Structure in American graphic Transition.” Demography 41(4): 607–27.
Life. New York: Free Press. Miller, Jon. 1994. Longitudinal Survey of American
Huang, M.-H., and R. M. Hauser. 2001. “Conver- Youth [dataset]. Ordering information available
gent Trends in Black-White Verbal Test-Score at: http://www.lsay.org (accessed June 24, 2011).
Differentials in the U.S.: Period and Cohort Per- Murray, C. 2007. “The Magnitude and Components
spectives.” EurAmerica 31(2): 185–230. of Change in the Black-White IQ Difference
Jacoby, R., and N. Glauberman, eds. 1995. The Bell from 1920 to 1991: A Birth Cohort Analysis of
Curve Debate: History Documents Opinions. New the Woodcock-Johnson Standardizations.” Intel-
York: Random House. ligence 35(4): 305–18.
Jargowsky, P. A. 1996. “Take the Money and Run: National Commission on Excellence in Education.
Economic Segregation in U.S. Metropolitan 1983. A Nation at Risk: The Imperative for Ed-
Areas.” American Sociological Review 61(6): ucational Reform. Washington, D.C.: National
984–98. Commission on Excellence in Education.
Kagan, J. 2002. “Empowerment and Education: National Institute of Child Health and Human De-
Civil Rights, Expert-Advocates, and Parent Poli- velopment. 2010. Study of Early Child Care and
tics in Head Start, 1964–1980.” Teachers College Youth Development [dataset]. Received 2010. Or-
Record 104(3): 516–62. dering information available at: http://www.nichd
King, Miriam, Steven Ruggles, J. Trent Alexander, .nih.gov/research/supported/seccyd/datasets.cfm
Sarah Flood, Katie Genadek, Matthew B. Schro- (accessed June 24, 2011).
eder, Brandon Trampe, and Rebecca Vick. 2010. Phillips, M., Jeanne Brooks-Gunn, Greg J. Duncan,
Integrated Public Use Microdata Series, Current P. Klebanov, and J. Crane. 1998. “Family Back-
Population Survey: Version 3.0 [Machine-readable ground, Parenting Practices, and the Black-White
database]. Minneapolis: University of Minnesota. Test Score Gap.” In The Black-White Test Score
Kornrich, Sabino, and Frank Furstenberg. 2010. Gap, edited by C. Jencks and M. Phillips. Wash-
“Investing in Children: Changes in Parental ington, D.C.: Brookings Institution Press.
Spending on Children, 1972 to 2007.” Unpub- Piketty, T., and E. Saez. 2003. “Income Inequality in
lished manuscript. the United States, 1913–1998.” Quarterly Journal
Lareau, Annette. 1989. Home Advantage: Social Class of Economics 118(1): 1–39.
and Parental Intervention in Elementary Educa- ———. 2008. “Income Inequality in the United
tion. London: Falmer Press. States, 1913–1998; Tables and Figures Updated
———. 2003. Unequal Childhoods: Class Race and to 2006.” Available at: http://www.econ.berkeley.
Family Life. Berkeley: University of California edu/~saez/TabFig2006.xls (accessed September 2,
Press. 2009).
550 • Part VI: Who Gets Ahead?

Ramey, G., and V. A. Ramey. 2010. “The Rug Rat ———. 2000. Early Childhood Longitudinal
Race.” Unpublished paper. Survey-Kindergarten (Base Year) [dataset]. Order-
Reardon, Sean F., and K. Bischoff. 2011. “Income ing information available at: http://nces.ed.gov/
Inequality and Income Segregation.” American ecls/kinderdatainformation.asp (accessed June
Journal of Sociology 116(4): 1092–153. 24, 2011) [Computer file]. 2nd ICPSR version.
Reardon, Sean F., and C. Galindo. 2009. “The Chicago, Ill.: National Opinion Research Center
Hispanic-White Achievement Gap in Math and [producer], 1980. Ann Arbor, Mich.: Interuni-
Reading in the Elementary Grades.” American versity Consortium for Political and Social Re-
Educational Research Journal 46(3): 853–91. search [distributor]. doi:10.3886/ICPSR07896.
Schaub, M. 2010. “Parenting for Cognitive De- ———. 2004. National Education Longitudinal
velopment from 1950 to 2000: The Institu- Study: Base Year Through Fourth Follow-Up,
tionalization of Mass Education and the Social 1988–2000 [Computer file], ICPSR version.
Construction of Parenting in the United States.” Washington, D.C.: U.S. Department of Ed-
Sociology of Education 83(1): 46–66. ucation, National Center for Education Sta-
Schwartz, C. R., and R. D. Mare. 2005. “Trends in tistics [producer], 2002. Ann Arbor, Mich:
Educational Assortative Marriage from 1940– Inter-university Consortium for Political and
2003.” Demography 42(4): 621–46. Social Research [distributor].
Schwartz, H. 2010. Housing Policy Is School Policy: ———. 2005. National Assessment of Educational
Economically Integrative Housing Promotes Aca- Progress-Long Term Trend [dataset]. Ordering
demic Success in Montgomery County, Maryland. information available at: http://nces.ed.gov/
New York: Century Foundation. nationsreportcard/researchcenter/datatools.asp
U.S. Bureau of Labor Statistics. 1980. National (accessed June 24, 2011).
Longitudinal Survey of Youth, 1979 [dataset]. ———. 2009. Early Childhood Longitudinal
Available at: http://www.bls.gov/nls/nlsy79.htm Survey-Birth Cohort (9 Month-Kindergarten)
(accessed June 24, 2011). [dataset]. Ordering information available at:
———. 1999. National Longitudinal Survey of http://nces.ed.gov/ecls/birthdatainformation.asp
Youth, 1997 [dataset]. Available at: http://www. (accessed June 24, 2011).
bls.gov/nls/nlsy97.htm (accessed June 24, 2011). ———. 2010. Education Longitudinal Study: Base
U.S. Department of Education. 1995. Prospects: The Year Through Second Follow-Up, 2002–2006
Congressionally Mandated Study of Educational [dataset]. Washington, D.C.: U.S. Department
Growth and Opportunity [dataset]. Ordering in- of Education, National Center for Education
formation available at: http://www.ed.gov/pubs/ Statistics [producer], 2010. Available at: http://
Prospects/index.html (accessed June 24, 2011). nces.ed.gov/surveys/els2002/orderingcds.asp (ac-
U.S. Department of Education, Center for Ed- cessed June 24, 2011).
ucation Statistics, n.d. National Assessment of Watson, T. 2009. “Inequality and the Measurement
Educational Progress-Main [dataset]. Ordering of Residential Segregation by Income.” Review of
information available at: http://nces.ed.gov/ Income and Wealth 55(3): 820–44.
nationsreportcard/researchcenter/datatools.asp Wrigley, Julia. 1989. “Do Young Children Need
(accessed June 24, 2011). Intellectual Stimulation? Experts’ Advice to Par-
———. 1999. National Longitudinal Study of the ents, 1900–1985.” History of Education Quarterly
Class of 1972 [Computer file]. ICPSR version. 29(1): 41–75.
Chicago, Ill: National Opinion Research Center Zigler, E., and S. Muenchow. 1992. Head Start: The
[producer], 1992. Ann Arbor, Mich.: Interuni- Inside Story of America’s Most Successful Educa-
versity Consortium for Political and Social Re- tional Experiment. New York: Basic Books.
search [distributor] doi:10.3886/lCPSR08085.
 Breen et al.—Nonpersistent Inequality in Educational Attainmen • 551

62 • Richard Breen, Ruud Luijkx, Walter Müller, and Reinhard Pollak

Nonpersistent Inequality in Educational Attainment

In their seminal study on the development of years of education attained and the educational
inequality in educational attainment in the 20th transitions considered. While the two analyses
century, Shavit and Blossfeld (1993) summa- address different empirical phenomena—of
rize the results under the guiding title Persistent which Shavit and Blossfeld are well aware—the
Inequality. In spite of dramatic educational results of both suggest essentially the same con-
expansion during the 20th century, of the 13 clusion, which the authors then summarize as
countries studied in their project, all but two, “stability of socio-economic inequalities of ed-
Sweden and the Netherlands, “exhibit stability ucational opportunities.” In the scientific com-
of socio-economic inequalities of educational munity, in particular in sociology and in the
opportunities. Thus, whereas the proportions of education sciences, the results have been viewed
all social classes attending all educational levels as evidence of a persistently high degree of class
have increased, the relative advantage associated inequality of educational attainment that can
with privileged origins persists in all but two of change only under rather exceptional conditions.
the thirteen societies” (p. 22). This conclusion Shavit and Blossfeld’s result echoed earlier
is based on a metanalysis of individual country findings from some single-country studies, but
studies, all of which adopt two different ap- subsequently several analyses have contested
proaches to assess socioeconomic inequalities this finding. They have shown that equalization
of educational opportunities: one is to use or- also took place in Germany (Müller and Haun
dinary least squares to regress years of education 1994; Henz and Maas 1995; Jonsson, Mills,
achieved by sons and daughters on parents’ ed- and Müller 1996), France (Vallet 2004), Italy
ucation and occupational prestige; the other is (Shavit and Westerbeek 1998), and the United
to regress, using binary logistic regression, a set States (Kuo and Hauser 1995). Rijken’s (1999)
of successive educational transitions on the same comparative analysis comes to the same conclu-
social background variables (the “Mare model”; sion. In other studies, Breen and Whelan (1993)
Mare 1980, 1981). Change or persistence in in- and Whelan and Layte (2002) confirm persistent
equalities of educational opportunities is diag- inequality for Ireland, whereas for Soviet Rus-
nosed depending on whether or not significant sia, Gerber and Hout (1995) find mixed results
variation over birth cohorts is found in the re- (declining inequality in secondary education and
gression coefficients linking social background to increasing inequality in transitions to university).

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Richard Breen, Ruud Luijkx, Walter Müller, and Reinhard Pollak, “Nonpersistent
Inequality in Educational Attainment,” American Journal of Sociology 114:5 (2009), pp. 1475–1521, published
by the University of Chicago Press. The article printed here was originally prepared by Richard Breen, Ruud
Luijkx, Walter Müller, and Reinhard Pollak for The Inequality Reader: Contemporary and Foundational Readings
in Race, Class, and Gender, Second Edition, edited by David B. Grusky, pp. 455–468. Copyright © 2011 by
Westview Press.
552 • Part VI: Who Gets Ahead?

For the postsocialist period in various countries and supportive of schoolwork than parents of
of Eastern Europe, the origin-education associ- working-class children. Different performance
ation is regularly found to be very high and is at school may also derive from different nu-
likely higher than in the socialist period (Gerber trition and health in different classes, whereas
[2000] for Russia; Iannelli [2003] for Hungary, genetic differences between individuals from
Romania, and Slovakia). different class backgrounds may play a role as
The aim of this article is to reassess the em- well as class differences in sibship sizes (see Erik-
pirical evidence concerning the conclusion of son and Jonsson 1996a). Yet, as Erikson and
Persistent Inequality using more recent data and Jonsson (1996b, p. 81) suggest, the general im-
larger samples from a selection of European provement in conditions of living should have
countries. In contrast to Shavit and Blossfeld, we made working-class children less disadvantaged
base our conclusions on analyses using ordered in terms of health and nutrition. With economic
logit models of educational attainment rather development and welfare-state protection, the
than on educational transition models. The minimum standards of living have improved and
reason is that we are interested in inequalities average family size has declined. Such changes
related to social origin in completed education, should have been more relevant for families in
which constitutes the major starting condition the less advantaged classes, such as the working
for unequal opportunities in the life course. An- classes and the small-farmer class, who have been
other reason for not using educational transition able to move out of absolute economic misery.
models is that we lack data on individuals’ com- Some decline in primary class effects should
plete educational histories. Indeed, there are no thus have occurred over the long term and par-
cross-nationally comparable large data sets that ticularly during the substantial improvement
contain complete education histories and also of general living conditions in the decades of
cover long historical periods. economic growth and welfare-state expansion
following World War II. This should have been
reinforced by changes within educational insti-
Reasons to Suppose That tutions, such as the growth in public provision
Educational Inequality Was of early child care and preschool education; the
Not Persistent development of full-day rather than part-time
Differences between students from different so- schooling; increased school support to counter-
cial classes in how they fare in the educational act performance gaps of pupils; and differences
system can, in simple terms, be seen to derive in the timing, extent, and manner of tracking, all
from differences in how they perform in the ed- of which may reduce class differences in school
ucational system (which Boudon [1974] called performance.
“primary effects”) and differences in the educa- As far as secondary effects are concerned, one
tional choices they make, even given the same factor that should have brought about a major
level of performance (“secondary effects”). In reduction is the declining costs of education. Di-
both areas, developments in the course of the rect costs, especially in secondary education, have
20th century would lead us to expect declining become smaller; school fees have been largely
class differences. abolished; the number of schools has increased,
As far as primary effects are concerned, chil- even in rural areas, so schools can be reached
dren raised in families in the more advantaged more easily; and traveling conditions have im-
classes encounter better conditions in their proved. In many countries, educational support
home environments that help them to do bet- programs for less wealthy families have been set
ter in school. They get more intellectual stimu- up, albeit of rather different kinds and levels of
lation that strengthens their cognitive abilities, generosity. Real average family income has in-
and their parents are more highly motivated creased, and that should make it easier to bear
 Breen et al.—Nonpersistent Inequality in Educational Attainmen • 553

the costs of education. While at least in the first Poland, Hungary, and the Netherlands—and
half of the last century working-class children they were originally assembled for a compara-
were urged to contribute to the family income tive analysis of social mobility in Europe (Breen
as early as possible, such pressures have declined. 2004). That project sought to bring together all
In most countries, economic growth and the re- the high-quality data sets collected between 1970
duction in family size have led to an increase in and 2000 in 11 European countries that could
disposable incomes beyond what is required for be used for the analysis of social mobility. The
basic needs. In practically all countries the length data used here are identical to those employed
of compulsory schooling has expanded, thus re- in that project except that the German data have
ducing the number of additional school years be- been augmented by six surveys. These six surveys
yond compulsory education needed to reach full contain the first three German Life History Sur-
secondary education. Countries certainly differ veys for West Germany (fielded between 1981
in the specifics of institutional reforms, and and 1989) as well as the 2000 sample for West
these probably have different implications; but Germany from the German Socio-economic
the lengthening of compulsory education should Panel and the ALLBUS Surveys for 2000 and
everywhere have contributed to a decline in the 2002. The Hungarian data are excluded from
additional costs of postcompulsory education. the final analysis, as will be described later. The
Countries also differ in their welfare-state data sets that we use are listed in Table 62.1. In
and social security arrangements and in their total we use 120 surveys collected between 1970
ability to prevent unemployment among stu- and 2002, but each country provides rather dif-
dents’ parents. In countries such as Sweden, in ferent numbers of surveys (up to a maximum of
which serious income equalization policies have 35 from the Netherlands). In Sweden, for exam-
been pursued successfully, the equalization of ple, there is a survey for every year from 1976
conditions is believed to have had an additional to 1999, whereas the analysis for Italy is based
impact on reducing the class differential in the on only two surveys and those for Ireland and
ability to bear the costs of education. The recur- Poland on only three surveys each.
rence of high levels of unemployment in many We use data on men ages 30–69 (30–59 in
countries since the 1980s, especially for the un- Great Britain, except for the years 1979–88,
skilled working class, and the increase in income when the age range is 30–49). We adopt 30 as
inequality observed in some countries in recent the lower age limit to ensure that everyone in
years (Alderson and Nielsen 2002) are probably the samples will have attained his highest level of
the most important changes that may have coun- education, and we take 69 as an upper limit in
teracted a long-term trend toward lowering the order to minimize any effects of differential mor-
impact of costs in producing class inequalities tality. We confine our analysis to men because
in educational participation, but these develop- the inclusion of both sexes, and comparisons
ments are mostly too recent to be evident in our between them, would have made a long article
data. In sum, both primary and secondary effects excessively so. We intend to analyze educational
changed in ways such that declining disparities inequality among women, and compare it with
between classes in educational attainment can the results reported here, in a further paper.
be expected; in particular, it is the children of
working-class and farm families who should have
most markedly improved their relative position. Variables
We use four variables in our analysis. Cohort (C)
defines five birth cohorts: 1908–24, 1925–34,
Data 1935–44, 1945–54, and 1955–64. Thus we have
Our data come from nine European countries— information on cohorts born throughout the first
Germany, France, Italy, Ireland, Britain, Sweden, two-thirds of the 20th century. Survey period (S)
554

Table 62.1. Sources of Data

No. of Years for Which Data

Country Tables Sources of Data Are Included
Germany 30 Zumabus 1976–77, 1979 (2), 1980, 1982
ALLBUS 1980, 1982, 1984, 1986, 1988,
1990–92, 1994, 1996,
1998, 2000, 2002
Politik in der BRD 1978, 1980
Wohlfahrtssurvey 1978
German Life History Study 1981–83 (I), 1985–88 (II),
1988–89 (III)
German Socio-economic Panel 1986, 1999, 2000
France 4 Formation-qualification professionnelle
INSEE surveys 1970, 1977, 1985, 1993
Italy 2 National Survey on Social Mobility 1985
Italian Household Longitudinal Survey 1997
Ireland 3 Survey of the Determinants of Occupational
Status and Mobility 1973
Survey of Income Distribution and Poverty 1987
Living in Ireland Survey 1994
Great Britain 15 General Household Survey 1973, 1975–76, 1979–84,
1987–92
Sweden 24 Annual Surveys of Living Conditions (ULF) 1976–99
Poland 3 Zagórski (1976) 1972
Slomczynski et al. (1989) 1988
Social Stratification in Eastern Europe after 1989 1994
Hungary 4 Social Mobility and Life History Survey 1973, 1983, 1992
Way of Life and Time Use Survey (Hungarian
Central Statistical Office) 2000
Netherlands 35 Parliamentiary Election Study 1970, 1971, 1977, 1981, 1982,
1986, 1994, 1998
Political Action Survey I 1974, 1979
Justice of Income Survey 1976
CBS Life Situation Survey 1977, 1986
National Labour Market Survey 1982
National Prestige and Mobility Survey 1982
Strategic Labour Market Survey 1985, 1988, 1990, 1992, 1994,
1996, 1998
Cultural Changes (ISSP) 1987
Justice of Income Survey 1987
Primary and Social Relationships 1987
Social and Cultural Trends 1990
Justice of Income Survey (ISJP) 1991
Family Survey I, 1992–93 1992
Households in the Netherlands pilot 1994
Households in the Netherlands 1995
Social Inequality in the Netherlands 1996
National Crime Study 1996
Social and Economic Attitudes 1998
Netherlands Family Survey II 1998
Use of Information Technology 1999
 Breen et al.—Nonpersistent Inequality in Educational Attainmen • 555

Table 62.2. CASMIN Educational Categories

Category Definition
1a ................ Inadequately completed elementary education
1b ................ Completed (compulsory) elementary education
1c ................ (Compulsory) elementary education and basic vocational qualification
2a ................ Secondary: intermediate vocational qualification or intermediate general qualification
and vocational training
2b ................ Secondary: intermediate general qualification
2c_gen.......... Full general maturity qualification
2c_voc .......... Full vocational maturity certificate or general maturity certificate and vocational
qualification
3a ................ Lower-tertiary education
3b ................ Higher-tertiary education

Table 62.3. EGP Class Categories

g
Category Definition
I .................. Higher-grade professionals, administrators, and officials; managers in large industrial
establishments; large proprietors
II .................. Lower-grade professionals, administrators, and officials; higher-grade technicians;
managers in small industrial establishments; supervisors of nonmanual employees
IIIa .............. Routine nonmanual employees, higher grade (administration and commerce)
IVa................ Small proprietors, artisans, etc. with employees
IVb .............. Small proprietors, artisans, etc. without employees
IVc .............. Farmers and smallholders; other self-employed workers in primary production
V .................. Lower-grade technicians; supervisors of manual workers
VI ................ Skilled manual workers
VIIa.............. Semi- and unskilled manual workers (not in agriculture, etc.)
VIIb ............ Agricultural and other workers in primary production
IIIb .............. Routine nonmanual employees, lower grade (sales and services)

defines the five-year interval in which the data We have only four educational categories in
were collected: 1970–74, 1975–79, 1980–84, the Hungarian data, where 2ab is missing, and
1985–89, 1990–94, 1995–99, and 2000–2004. in the Italian and the Irish data, where no dis-
Highest level of educational attainment (E) tinction has been made between 3a and 3b. The
is measured using the CASMIN educational CASMIN educational schema seeks to capture
schema (Comparative Analysis of Social Mobil- distinctions not only in the level of education
ity in Industrial Nations; see Table 62.2; Braun but also in the type, and one consequence of this
and Müller 1997). We have amalgamated cate- is that the five levels we identify cannot be con-
gories 1a, 1b, and 1c and also 2a and 2b, giving sidered to be sequentially ordered in any simple
us five educational categories: way. For example, in some countries lower ter-
1abc.—Compulsory education with or with- tiary education can be accessed directly from sec-
out elementary vocational education ondary intermediate education, whereas in most
2ab.—Secondary intermediate education, vo- countries, higher tertiary is not usually entered
cational or general after lower tertiary.
2c.—Full secondary education Class origins (O) are categorized using the
3a.—Lower tertiary education Erikson-Goldthorpe-Portocarero (EGP) class
3b.—Higher tertiary education schema (see Table 62.3; also Erikson and
556 • Part VI: Who Gets Ahead?

Table 62.4. Sample Sizes for Cohorts by Country

Cohort
Country 1908–24 1925–34 1935–44 1945–54 1955–64 Total
Germany 2,323 3,110 4,852 4,007 2,832 17,124
France 11,283 14,169 13,126 10,517 2,610 51,705
Italy 742 1,182 1,286 827 4,037
Ireland 744 1,176 1,509 1,289 863 5,581
Great Britain 6,473 9,112 24,481 20,971 5,437 66,474
Sweden 8,032 8,209 10,875 10,145 4,093 41,354
Poland 7,729 9,248 8,851 1,002 784 27,614
Hungary 5,698 6,829 7,175 5,643 2,583 27,928
Netherlands 2,385 3,507 4,736 5,612 3,511 19,751
Total 44,667 56,102 76,787 60,472 23,540 261,568

Goldthorpe 1992, chap. 2). We identify seven Table 62.4 shows the resulting sample sizes
classes: for all the countries by cohort. These vary quite
I.—Upper service considerably, and this will obviously affect our
II.—Lower service ability to detect statistically significant trends.
IIIa.—Higher-grade routine nonmanual The sample sizes for Italy and Ireland are par-
IVab.—Self-employed and small employers ticularly small, and one consequence of this is
IVc.—Farmers that we have very few observations of the oldest
V + VI.—Skilled manual workers, techni- cohort in Italy, so we omit it from our analyses.
cians, and supervisors
VIIab + IIIb.—Semi- and unskilled manual,
agricultural, and lower-grade routine non- Changes in Educational Attainment
manual workers and Class Origins
Perhaps the single most striking thing that dif-
In Britain and Poland the data allow us to ferentiates the older from the younger cohorts
identify only six class origins. In both countries in our data is the massive increase in educational
we cannot distinguish classes I and II, whereas attainment that has occurred. Figure 62.1 shows
in Britain, members of IVa are included in I + II the proportions in each cohort in each country
(see Goldthorpe and Mills 2004). Furthermore, that have attained at least upper-secondary (2c)
in Poland, we cannot split the class III, so here education, and Figure 62.2 shows the propor-
IIIb is included with IIIa rather than with VIIab. tions having attained tertiary (3a and 3b) edu-
In Ireland also we combine I and II because of cation.1 The upward trends in both are obvious
very small numbers in class I in some cohorts. and are similar across countries.
The resulting four-way table of class origins It is not only the educational distributions that
(O) by educational attainment (E) by cohort (C) have shifted, however. During the course of the
by survey period (S) is of maximum dimensions 20th century the class structures of European na-
7 × 5 × 5 × 7 = 1,225, though this number in- tions underwent major change, with a move away
cludes many structural zeros in those combina- from farming and unskilled occupations and to-
tions of cohort and survey that are not observed. ward skilled jobs and white-collar jobs. Some
Furthermore, we omitted all those observations aspects of this are shown in Figure 62.3, which
of cohort by survey in which the table of origins reports the share of the service class (I and II),
by education would have been extremely sparse. intermediate class (IIIa and IVab), the farm classes
All the cells in such a table were treated as struc- (IVc and VIIb), and working class (V + VI, VIIa
tural zeros. + IIIb) in the origins of the oldest and youngest
 Breen et al.—Nonpersistent Inequality in Educational Attainmen • 557

Figure 62.1. Proportion of Men Reaching at Least Figure 62.3. Proportions of Men in Various Class
Upper-Secondary Education, by Country and Cohort Origins for First and Last Cohort by Country (Marginal
Distributions)

Figure 62.2. Proportion of Men Reaching Tertiary

Education, by Country and Cohort
In Great Britain, class IVa is included in classes I + II; in Poland,
class IIIb is included in classes IIIa + IVab.

(–∞,∞). The logit link, which yields the ordered

logit model, sets g[Ȗj(x)] = ln [Ȗj(x)]/[1 – Ȗj(x)]
and for any two observations with covariate val-
ues x and x' we get

and so the log odds ratio of exceeding, or failing

cohorts in each country. The decline of the farm to exceed, any particular threshold is indepen-
class and growth of the service class are evident dent of the thresholds themselves and depends
everywhere, and the working class has grown or only on the betas.
remained stable everywhere except Britain. The ordered logit model has two attractive
properties: it is more parsimonious than the
multinomial logit or the Mare model, and, pro-
Analyses vided that the ordered logit is the correct model
We focus on modeling the joint distribution of for the data, the estimates of the ȕ parameters are
class origins and highest level of education using unaffected by collapsing adjacent categories of
the ordered logit model. Letting Y be the ran- Y. This invariance means that if the educational
dom variable measuring educational attainment, system makes finer distinctions than have been
we can write the probability that Y is less than a captured in the variable Y, the ȕ’s are unaffected
particular level of education, j, given background by our having measured education less precisely.
variables X, as Ȗj(x). There exists a family of sta- It also implies that, when making comparisons,
tistical models that sets g[Ȗj(x)] = tj – ȕ›x (McCul- provided that the same stochastic ordering of
lagh 1980), where g is a link function (such as categories is used in all samples, we do not re-
the logit, the inverse normal, or the complemen- quire the categories to be defined in exactly the
tary log-log) that maps the (0, 1) interval into same way. This facilitates both temporal and
558 • Part VI: Who Gets Ahead?

Table 62.5. Model Fits for Ordered Logit Models, by Country, Survey, and Cohort: Chi-Squared
Summary Measure of Change of OE Effects Across Cohorts for SCOE Data Table (Controlling for
SOE) and COE Data Table

Data from SCOE Table, Data from COE Table

Controlling for SOE (Collapsed over S)
Change No OE Change No OE
COE Change G2 COE Change G2
Country (1) (2) (3) (4) (5) (6)
Germany 557.0 642.0 84.9 134.4 234.6 100.2
(468) (492) (24) (90) (114) (24)
France 515.0 629.3 114.3 283.5 459.5 176
(312) (336) (24) (90) (114) (24)
Italy 90.1* 111.0* 20.8* 43.7* 67* 23.3*
(96) (114) (18) (48) (66) (18)
Ireland 189.2 221.1 31.8 101.8 155.3 53.4
(130) (150) (20) (50) (70) (20)
Great Britain 709.9 780.0 70.1 502.7 604.4 101.7
(275) (295) (20) (75) (95) (20)
Sweden 624.1 723.4 99.3 286.2 413.2 126.9
(450) (474) (24) (90) (114) (24)
Poland 191.5 318.8 127.3 192.5 351.3 158.8
(115) (135) (20) (50) (70) (20)
Netherlands 613.5 694.3 80.8 154.1 266.5 112.4
(516) (540) (24) (90) (114) (24)
Note: Numbers in parentheses are degrees of freedom.
* Not significant at P<.05.

cross-national comparisons. This property is not effects. This shows significant change everywhere
shared by the educational transition model or by except Italy (although the Irish data barely reach
multinomial logit models, whose estimates are significance when we control for survey period).
tied to particular categorizations of educational The deviance associated with this change is lesser
attainment. But the advantages of the ordered when we control for survey effects compared
logit come at a price: the model assumes that with when we do not, so our conclusions about
class inequalities are identical at each level of change are unaffected, though the magnitude of
education. This is sometimes called the parallel that change is a little smaller.
slopes or proportional odds assumption and is One can also see, however, that, with the .05
frequently found not to hold. We address this level of significance, the ordered logit model fails
problem in our analyses. to fit the data in almost all countries, suggest-
In Table 62.5 we report the goodness of fit ing that the parallel slopes assumption does not
of the ordered logit models in which we control hold. Elsewhere, we discuss the departures from
for survey (cols. 1–3) and in which we do not proportionality in our data and we find that they
(cols. 4–6). In all models the thresholds, tj, are are not consequential, so for the most part we
allowed to vary over cohort: in the models in base our findings on the results of the models
columns 1 and 4, class origin effects are allowed whose goodness of fit is reported in column 1
to vary over cohorts, whereas in columns 2 and 5 of Table 62.5 and whose parameter estimates are
they are held constant. The final column in each shown in Figure 62.4 (see Breen et al. 2009).
case shows the deviance comparing the model of In Figure 62.4 each line refers to a class ori-
change against the model of constancy in class gin and shows how the coefficients for that class
 Breen et al.—Nonpersistent Inequality in Educational Attainmen • 559

Figure 62.4. Ordered Logit Models for Educational Attainment in Eight Countries for
Men

Note: Class origin effects over cohorts, controlled for survey effects. Class I is classes I + II in
Ireland and Poland and classes I + II + IVa in Britain; educational levels 3a and 3b are merged in
Italy and Ireland.

evolve over cohorts, with class I always acting as observation period. The Dutch case too shows
the reference category (classes I + II in Ireland a general tendency toward a narrowing of class
and Poland and I + II + IVa in Britain) and hav- differentials; this occurs over all cohorts, for the
ing a coefficient of zero. The overall impression most part, and classes IVab, IVc, V + VI, and VII
is of a decline in class inequalities everywhere have made the greatest gains. In contrast with
(even in Italy, where the change is not statisti- these four countries, the British picture is more
cally significant), but Figure 62.4 also allows us complicated. Although all classes except III are
to see how the relative positions of the various less differentiated from class I in the youngest
classes have shifted. In Germany there was a cohort than they were in the oldest, almost all
general narrowing of class differentials in educa- the improvement for classes IVb and V + VI
tional attainment following the 1925–34 birth occurred between the 1908–24 and 1925–34
cohort and continuing until the 1945–54 co- cohorts, whereas for class VII there has been a
hort. Little change is evident between the latter more prolonged improvement and for class IVc
and the youngest cohort. In France equalization an improvement from the 1935–44 cohort on-
started later, with the 1935–44 cohort, and con- ward. In Poland there has been a steady improve-
tinued longer, persisting into the 1955–64 co- ment in the relative positions of classes IVc, V +
hort. It was slightly more differentiated by class VI, and VII until the middle cohort, at which
origins than in Germany because men from class point the position of IVc worsens, followed by
III showed no change in their position relative to a worsening of the positions of IVab and VII in
men from class I, whereas men of farm origins the youngest cohort. In Italy we find a modest
(class IVc) experienced a particularly marked im- decrease in class inequalities until the 1945–54
provement in their relative position. In Sweden cohort, after which class inequalities tended to
all classes have improved their position relative reassert themselves. Class VII is an exception
to class I, as in Germany, and the decline in in- because its position continues to improve, al-
equality largely took place throughout the whole beit slightly. Finally, in the Irish case, over the
560 • Part VI: Who Gets Ahead?

five cohorts the distance between class I and all graduates increases, young people with only an
other classes was reduced, and the positions of upper-secondary qualification will be forced to
the petty bourgeoisie (IVab) and farmers (IVc) take less attractive jobs than their counterparts
have improved noticeably. in older cohorts.
Overall it seems to be the most disadvantaged A potentially more telling objection to the
classes that experienced the greatest improve- argument that declining association implies
ment in their position over the first two-thirds of declining class inequality is that there may be
the 20th century. Whereas the position of class distinctions within the broad CASMIN educa-
II, relative to that of class I, tends to show little tional categories that are consequential for life
change and classes III and IVab show improve- chances. For example, it is commonly the case
ments in some countries and in some cohorts but that differences exist between classes in their
not in others, classes IVc, V + VI, and, particu- choice of particular subjects of study or field of
larly, VII show general and widespread improve- education (Lucas 2001; Van de Werfhorst 2001;
ment in their position vis-à-vis the other classes. Kim and Kim 2003). If these differences have
As to the more precise timing of the change, we become stronger as inequalities in level of ed-
can note that we observe a trend of declining ucation have declined, then a focus solely on
inequality in all countries for the cohorts born educational level will overestimate the extent to
between 1935 and 1954, in particular for the which inequalities have declined. Nevertheless, it
least advantaged classes. These are the cohorts is difficult to believe that inequalities stemming
that made the crucial educational transitions in from differences in field of study within a given
the first three post–World War II decades up to level of education will be as important for vari-
about 1975, during which there were significant ations in life chances as differences between the
improvements in living conditions during a pe- levels of education attained.
riod of strong economic growth in most Euro- The data that we have used are uniquely
pean countries. suited to the purpose of describing long-term
trends in educational inequality and comparing
these trends across countries. But this breadth
Conclusions of coverage in time and space comes at a cost:
Even given that the weight of evidence now we lack many items of information that would
supports the thesis of a declining association be- be needed if we were to explain the trends. As
tween class origins and educational attainment, a consequence, our article has been primarily
it may be argued that to interpret this trend as empirical, descriptive, and methodological.
demonstrating an increase in equality is mistaken But it is clear that our findings present a chal-
because education is a positional good. In this lenge to theory because much recent theorizing
case, the value of an educational qualification di- on educational differentials, such as Breen and
minishes in proportion to the number of people Goldthorpe’s (1997) relative risk aversion theory
who acquire it. But for this argument to have any and Raftery and Hout’s (1993) maximally main-
force it is not enough to show that, over time, tained inequality hypothesis, has taken as its
the value of some qualification declines; rather, it starting point the need to explain the supposed
must be demonstrated that differences between regularity exemplified by Persistent Inequality.
the returns to educational levels diminish. The Our results show that the focus needs to shift
issue is not whether the returns to a tertiary qual- not only from the explanation of stability to the
ification are less in one cohort than in an older explanation of change, but from accounts based
one, but whether, for example, the gap in returns on the assumption of widespread commonality
between a tertiary and an upper-secondary qual- to those that encompass the differences between
ification has narrowed. As far as we know, there countries in the magnitude of class inequalities
is no such evidence, and indeed, there are good in educational attainment and in the timing of
grounds for supposing that, as the number of their decline.
 Breen et al.—Nonpersistent Inequality in Educational Attainmen • 561

NOTES Soviet Period.” American Journal of Sociology,

1. At this point we omit Hungary from our 101:611–660.
comparison. Goldthorpe, John H., and Colin Mills. 2004.
“Trends in Intergenerational Class Mobility
REFERENCES in Britain in the Late Twentieth Century.” Pp.
Alderson, Arthur S., and François Nielsen. 2002. 195–224 in Social Mobility in Europe, edited by
“Globalization and the Great U-turn: Income In- Richard Breen. Oxford: Oxford University Press.
equality Trends in 16 OECD Countries.” Ameri- Henz, Ursula, and Ineke Maas. 1995. “Chancengle-
can Journal of Sociology 107:1244–99. ichheit durch die Bildungs-expansion.” Kölner
Boudon, Raymond. 1974. Education, Opportunity, Zeitschrift fur Soziologie und Sozialpsychologie
and Social Inequality: Changing Prospects in West- 47:605–33.
ern Society. New York: Wiley. Iannelli, Cristina. 2003. “Parental Education and
Braun, Michael, and Walter Müller. 1997. “Measure- Young People’s Educational and Labour Market
ment of Education in Comparative Research.” Outcomes: A Comparison across Europe.” Pp.
Comparative Social Research 16:163–201. 27–53 in School-to-Work Transitions in Europe:
Breen, Richard, ed. 2004. Social Mobility in Europe. Analyses of the EU LFS 2000 Ad Hoc Module, ed-
Oxford: Oxford University Press. ited by Irena Kogan and Walter Müller. Mann-
———. 2007. “Statistical Models of Educational heim: Mannheimer Zentrum für Europäische
Careers.” Manuscript. Yale University, CIQLE. Sozialforschung.
Breen, Richard, and John H. Goldthorpe. 1997. Jonsson, Jan O., Colin Mills, and Walter Müller.
“Explaining Educational Differentials: Towards a 1996. “Half a Century of Increasing Educational
Formal Rational Action Theory.” Rationality and Openness? Social Class, Gender and Educational
Society 9:275–305. Attainment in Sweden, Germany and Britain.”
Breen, Richard, Ruud Luijkx, Walter Müller, and Pp. 183–206 in Can Education Be Equalized? The
Reinhard Pollak. 2009. “Nonpersistent Inequal- Swedish Case in Comparative Perspective, edited
ity in Educational Attainment: Evidence from by Robert Erikson and Jan O. Jonsson. Boulder,
Eight European Countries.” American Journal of Colo.: Westview.
Sociology 114: 1475–1521. Kim, Anna, and Ki-Wan Kim. 2003. “Returns to
Breen, Richard, and Christopher T. Whelan. 1993. Tertiary Education in Germany and the UK:
“From Ascription to Achievement? Origins, Ed- Effects of Fields of Study and Gender.” Work-
ucation and Entry to the Labour Force in the Re- ing paper no. 62. Mannheimer Zentrum für Eu-
public of Ireland during the Twentieth Century.” ropäische Sozialforschung, Mannheim.
Acta Sociologica 36:3–18. Kuo, Hsiang-Hui Daphne, and Robert M. Hauser.
Erikson, Robert, and John H. Goldthorpe. 1992. 1995. “Trends in Family Effects on the Educa-
The Constant Flux: A Study of Class Mobility in tion of Black and White Brothers.” Sociology of
Industrial Societies. Oxford: Clarendon. Education 68:136–60.
Erikson, Robert, and Jan O. Jonsson, eds. 1996a. Lucas, Samuel R. 2001. “Effectively Maintained In-
“Explaining Class Inequality in Education: The equality: Education Transitions, Track Mobility,
Swedish Test Case.” Pp. 1–64 in Can Education and Social Background Effects.” American Jour-
Be Equalized? The Swedish Case in Comparative nal of Sociology 106:1642–90.
Perspective, edited by Robert Erikson and Jan O. Mare, Robert D. 1980. “Social Background and
Jonsson. Boulder, Colo.: Westview. School Continuation Decisions.” Journal of the
———. 1996b. “The Swedish Context: Educa- American Statistical Association 75:295–305.
tional Reform and Long-Term Change in Edu- ———. 1981. “Change and Stability in Educational
cational Inequality.” Pp. 65–93 in Can Education Stratification.” American Sociological Review
Be Equalized? The Swedish Case in Comparative 46:72–87.
Perspective, edited by Robert Erikson and Jan O. McCullagh, Peter. 1980. “Regression Models for Or-
Jonsson. Boulder, Colo.: Westview. dinal Data.” Journal of the Royal Statistical Society,
Gerber, Theodore P. 2000. “Educational Stratifica- ser. B, 42:109–42.
tion in Russia during the Soviet Period.” Ameri- Müller, Walter, and Dietmar Haun. 1994. “Bil-
can Journal of Sociology 101:611–60. dungsungleichheit im sozialen Wandel.” Kölner
Gerber, Theodore P, and Michael Hout. 1995. “Ed- Zeitschrift fur Soziologie und Sozialpsychologie
ucational Stratification in Russia During the 46:1–42.
562 • Part VI: Who Gets Ahead?

Müller, Walter, and Wolfgang Karle. 1993. “Social Schemat teoretyczny i warsztat badawczy. Warsaw:
Selection in Educational Systems in Europe.” Eu- Polish Academy of Sciences.
ropean Sociological Review 9:1–23. Vallet, Louis-André. 2004. “The Dynamics of In-
Raftery, Adrian E., and Michael Hout. 1993. “Maxi- equality of Educational Opportunity in France:
mally Maintained Inequality: Educational Stratifi- Change in the Association between Social Back-
cation in Ireland.” Sociology of Education 65:41–62. ground and Education in Thirteen Five-Year Birth
Rijken, Susanne. 1999. Educational Expansion Cohorts (1908–1972).” Paper prepared for the
and Status Attainment: A Cross-National and meeting of the ISA Research Committee on Social
Over-Time Comparison. Amsterdam: Thela The- Stratification and Mobility, Neuchâtel, May 6–8.
sis (ICS Dissertation). Van de Werfhorst, Herman. 2001. Field of Study and
Shavit, Yossi, and Hans-Peter Blossfeld, eds. 1993. Social Inequality: Four Types of Educational Re-
Persistent Inequality: Changing Educational At- sources in the Process of Stratification in the Nether-
tainment in Thirteen Countries. Boulder, Colo.: lands. Nijmegen: ICS Dissertation.
Westview. Whelan, Christopher T., and Richard Layte. 2002.
Shavit, Yossi, and Karin Westerbeek. 1998. “Educa- “Late Industrialisation and the Increased Merit
tional Stratification in Italy: Reforms, Expansion, Selection Hypothesis: Ireland as a Test Case.” Eu-
and Equality of Opportunity.” European Sociolog- ropean Sociological Review 18:35–50.
ical Review 14:33–47. Zagórski, Krzysztof. 1976. Zmiany struktury i ruchli-
Slomczynski, Kazimierz M., Ireneusz Bialecki, Hen- wość społeczno-zawadowa w Polsce. Warsaw: GUS.
ryk Domanski, et al. 1989. Struktura spoleczna:

63 • Michelle Jackson

Determined to Succeed

Introduction The development of mass educational

The twentieth century was a time of dramatic systems can be understood principally as a
educational transformation, marked by rapid ed- quasi-functionalist response to the demands of
ucational expansion and reform. The basic fea- changing economic and occupational structures,
tures of educational expansion are well known but also as an attempt to create greater equality
and comprise three fundamental steps: (1) the of educational opportunity. The latter objective
establishment of near-universal primary educa- remains unmet; there is much research demon-
tion at the beginning of the century, (2) the rise strating substantial inequalities in educational
of near-universal secondary education toward attainment among children of different social
the middle of the century, and (3) the ongoing classes (Shavit and Blossfeld 1993; Breen et
development of systems of mass higher educa- al. 2009). Although a trend toward a weaken-
tion toward the end of the century. ing association between social class origin and

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Michelle Jackson, “How is Inequality of Educational Opportunity Generated? The
Case for Primary and Secondary Effects,” in Determined to Succeed? Performance versus Choice in Educational
Attainment, edited by Michelle Jackson, published by Stanford University Press in 2013. The article printed
here was originally prepared by Michelle Jackson for this fourth edition of Social Stratification: Class, Race, and
Gender in Sociological Perspective, edited by David B. Grusky. Copyright © 2014 by Westview Press.
 Jackson—Determined to Succeed • 563

educational attainment exists in many European across countries with rather different educational
countries, particularly if changes over a relatively systems and institutional structures.2
long period of time are considered (Breen and
Jonsson 2005; Breen et al. 2009), inequalities
in educational attainment remain a fundamen- Data and Methods
tal feature of modern societies and are likely to The data are drawn from results presented in
linger for some time to come. The durability of Determined to Succeed? Performance versus Choice
these inequalities is particularly striking when in Educational Attainment (Jackson 2013a), in
compared to the far more substantial changes in which specialists analyze country-specific data
gender, ethnic, and racial inequalities that have on cohorts of young people to assess the level of
occurred in many countries. educational inequality and the role of primary
But why do social class inequalities in educa- and secondary effects in generating that inequal-
tional attainment exist and persist? Should these ity. I focus for the most part on cohorts born
inequalities be understood as a consequence in the early 1990s, as they provide information
of differences in academic ability and perfor- about the contemporary state of educational in-
mance between members of different social equality. For some countries, it was possible to
classes? Or should inequalities be understood as analyze data on several earlier cohorts, and I also
a consequence of differences in the educational include these cohorts in the final analysis of this
decisions made by members of different social chapter, in which I look for overarching relation-
classes, such that students from advantaged ships between primary and secondary effects and
classes choose higher levels of education more overall inequality. Data selection, variable con-
frequently than students from disadvantaged struction, and methodological approach were
classes, regardless of their academic perfor- carried out according to a common metric.
mance? These basic questions outline extreme The analyses of primary and secondary ef-
positions on how inequality is created. A cred- fects require consistent measures of social back-
ible model of educational attainment must treat ground, academic performance, and educational
inequalities as a consequence of class differences transition taking, defined as follows (more de-
in both academic performance and the choices tails can be found in Jackson 2013b):
students make, holding performance constant. • Social background. Parental class and edu-
The decomposition of educational inequality cation are both used as measures of social
into a part determined by differences in previ- background, although in some countries it
ous performance across social groups and a part is only possible to find reliable data for one
determined by the choices made by members of of these measures. I focus here on a contrast
those groups is well established in the literature, between “advantaged” and “disadvantaged”
in which performance effects are labeled “pri- social background groups. Families in which
mary” effects, and choice effects are labeled “sec- at least one parent is employed in a profes-
ondary” effects1 (Boudon 1974). The main aim sional or managerial occupation, or in which
of this chapter is to determine the importance at least one parent has degree-level educa-
of these two effects in generating inequalities in tion, are coded “advantaged.” Working-class
educational attainment. families, or families in which neither parent
In this chapter I examine the influence of has education in excess of the lower second-
primary and secondary effects on inequalities ary level, are coded “disadvantaged.”
in educational attainment in eight countries: • Academic performance. Given the behavioral
Denmark, England, France, Germany, Italy, the model that lies behind the concepts of pri-
Netherlands, Sweden, and the United States. I mary and secondary effects, it is important
ask whether it is possible to find evidence for that students who are assumed to make
the existence of primary and secondary effects educational decisions on the basis of their
564 • Part VI: Who Gets Ahead?

Figure 63.1. Role of Primary and Secondary Effects in Generating Educational Inequality at the First
Major Transition, Cohort of Students Born in Early 1990s

1 1

0.9 0.9

0.8 0.8

0.7 0.7

Probability of transition
Proportion of cases

0.6 0.6

0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1

0 0
-4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4
Academic performance

Advantaged Disadvantaged

academic performance should have access In the remaining countries, this transition
to information about that performance. occurs at the end of compulsory schooling
In most countries, therefore, performance (around age sixteen) and usually distin-
is measured using public examination re- guishes between students who continue to
sults that are known to students, teachers, postcompulsory academic education, and
and families. The performance variable is those who choose vocational education, or
standardized to a mean of 0 and a standard those who leave education. In the United
deviation of 1. States the first major transition is high
• Educational transitions. The question at school graduation.
hand is whether a student, when faced with
an educational transition, makes a positive By conforming to a common metric for analysis,
decision to go forward (measured as a 0/1 the results from the individual country analyses
variable, where 1 indicates a positive deci- can be combined and used to provide a general
sion). Although specific educational transi- assessment of the influence of primary and sec-
tions differ across countries, my focus is on ondary effects on social background inequalities
the first major transition faced by students in educational attainment.
in these countries, as this is the one expected
to be particularly important in determining
the total amount of inequality produced Decomposing Educational Inequality
within an educational system (cf. Breen et into Primary and Secondary Effects
al. 2009). In Germany and the Netherlands, The role of primary and secondary effects in
this transition is made early (ages ten to generating educational inequality is illustrated in
twelve) and distinguishes between students Figure 63.1. The figure shows the relationships
who will pursue the higher-level academic among social background, academic perfor-
tracks and those who will pursue lower-level mance, and the first major transition for cohorts
academic tracks and vocational education. of students born in the early 1990s.
 Jackson—Determined to Succeed • 565

The two sets of curves shown in Figure 63.1 background will make a positive rather than
represent the two types of effects. The primary negative choice at the first major transition as
effects are illustrated by the normal distribution compared to the corresponding chances for stu-
curves, derived by substituting the mean and dents of disadvantaged background.3 It answers
standard deviation of the performance scores the simple question: How much more likely is a
into the equation for a normal distribution, for student from an advantaged background to go
each social background group separately. These forward than one from a disadvantaged back-
curves demonstrate that social groups have dif- ground? For cohorts of students born in the
ferent distributions of performance, and that, on early 1990s, the average level of educational in-
average, students from advantaged groups have equality among the countries at this transition
higher levels of performance than those from dis- amounts to a log odds ratio of 1.8, indicating
advantaged groups. We would expect differences that students from an advantaged background
in transition rates between students from differ- are approximately six times as likely to make
ent groups to emerge simply on the basis of these the first major transition as are students from
differences in performance (i.e., the primary ef- a disadvantaged background (e1.8 = 6.05). This
fects). But the S-shaped curves demonstrate that average disguises significant variation among
secondary effects are also at work in the creation countries: the lowest level of inequality is found
of inequality. These curves show the transition in England and the United States (the log odds
propensities for students from each group con- ratio in each country is around 1.28), whereas
ditional on their level of academic performance. the highest level of inequality is found in Italy
Even at the same level of academic performance, (where the log odds ratio is 3.09).
students from an advantaged background are Using the decomposition method, I now as-
more likely to make the transition than those sess the contribution of primary and secondary
from a disadvantaged background. The curves effects to the generation of these inequalities. Av-
are derived from binary logistic regressions, in eraging across the countries, the overall log odds
which academic performance is used to predict ratio of 1.8 can be decomposed into a primary
(for each group separately) whether an individ- effect of 0.92 and a secondary effect of 0.86
ual makes a positive decision in the first major (note that total inequality = primary effect + sec-
transition. ondary effect). In other words, about half of the
This approach, in which primary and second- overall educational inequality can be attributed
ary effects are represented by normal distribu- to differences in academic performance among
tions of performance and predicted probability students of different social backgrounds, and
curves derived from logistic regressions, pro- the other half can be attributed to the choices
vides the basis for a decomposition of overall that are made by students from different back-
educational inequality proposed by Erikson and grounds with the same level of academic perfor-
colleagues (2005; see also Jackson et al. 2007; mance. Once again, the average size of primary
Kartsonaki et al. 2013). In essence, the method and secondary effects disguises variation across
decomposes the odds ratios describing inequali- countries, as demonstrated by Figure 63.2.
ties between social groups into a part due to pri- The figure shows the contribution of pri-
mary effects and a part due to secondary effects; mary effects to the overall inequality between
a percentage estimate of the relative importance the advantaged and disadvantaged groups in the
of the two effects can thus be derived. dark gray bars and the contribution of second-
ary effects in the light gray bars. Here we have
evidence that both effects are in operation in all
Cross-national Comparisons of of the countries in the sample, and importantly,
Primary and Secondary Effects that there is more to educational inequality than
The odds ratio refers in this context to the differences in performance among students from
chances that students from an advantaged different social backgrounds. For the countries
566 • Part VI: Who Gets Ahead?

Figure 63.2. Contribution of Primary and Secondary Effects to Educational Inequality (Advantaged vs.
Disadvantaged Social Background Groups), Cohort of Students Born in Early 1990s

2
Log odds

0
DEN ENG FRA GER ITA NET SWE USA

Primary Secondary

included in this analysis, the secondary effects of cohorts, and specifically to examine the strength
differences in the educational choices made, con- of the relationship between secondary effects
ditional on performance, account for between and overall inequality. The black and gray lines
13 (England) and 71 percent (Italy) of overall represent the OLS regressions of overall inequal-
inequality in the first major transition. ity on secondary effects for each transition. A ȕ
When comparing primary and secondary coefficient of 0 would indicate no relationship
effects across countries, we should also note between the size of secondary effects and the size
from Figure 63.2 that there appears to be less of overall inequality, whereas a ȕ coefficient of 1
cross-national variation in the size of primary would indicate that overall inequality increases
effects than in the size of secondary effects. by 1 for each unit increase in secondary effects,
But is this finding supported when we bring in such that differences in the overall level of in-
data from earlier birth cohorts and the transi- equality could be understood as wholly resulting
tion from high school to university? Figure 63.3 from secondary effects.
plots the relationship between secondary effects The regression lines for the two transitions
and overall inequality for each transition; each are close to parallel, indicating a similar rela-
data point represents a birth cohort in one of tionship between secondary effects and overall
the eight countries, with data points pertaining inequality at each transition point: at the first
to the first major transition marked in black and transition, the ȕ coefficient for the regression of
data points pertaining to the transition to uni- overall inequality on secondary effects is 0.98,
versity marked in grey. This figure, adapted from p = 0.000; at the transition to university, the
work by Jackson and Jonsson (2013), allows us ȕ coefficient is 0.89, p = 0.000. The intercepts
to look for general patterns across countries and of the regression lines represent the base level
 Jackson—Determined to Succeed • 567

Figure 63.3. Relationship Between Secondary Effects and Inequality at the First Major Transition and the
Transition to University (Log Odds Ratios for Advantaged vs. Disadvantaged Social Background Groups)
4

ITA70
ITA90
ITA80
3
Overall inequality (log odds)

NET90
NET70 DEN50
FRA50
NET60
2

SWE70GER90
FRA90
ITA90
SWE80 ENG50
SWE90USA90 ITA80
FRA50

ENG90 ENG70 USA90SWE80 ITA70

ENG80 SWE90
SWE70
1

DEN90
GER90 NET60
NET70
ENG50
ENG70 NET90
ENG80
ENG90
FRA90
0

0 1 2 3
Secondary effects (log odds)

First transition (ages 10-16) Transition to university (age 18)

Source: Adapted from Jackson and Jonsson (2013), Fig. 11.7.

of inequality established by the primary effects having educational regimes in which choice is
when secondary effects are 0 (at the first major heavily colored by social background. If the aim
transition, the Į is 0.96; at the transition to uni- is to reduce inequality in these countries, an ob-
versity, the Į is 0.34). In other words, the results vious way forward is to attack the institutions
of the regressions demonstrate that variation in that allow social background to enter into edu-
the size of secondary effects across countries and cational decisions.
cohorts explains variation in the level of overall
inequality. Jackson and Jonsson (2013) argue
that this finding should be taken as evidence Discussion and Policy Implications
that an “incremental reproduction” inequality The analyses in this chapter demonstrate that
regime is in place in this sample of seven Euro- both primary and secondary effects are involved
pean countries and the United States, and fur- in the generation of social background inequal-
ther that “in all countries alike there is a floor ities in educational attainment. Inequalities are
level of [inequality] established by social-origin not simply a function of differences in academic
differences in the abilities and aptitudes reflected ability or performance among social groups, but
in performance, but above that floor level there are also a product of the choices that are made
is considerable variation in the way that educa- conditional on performance. Furthermore, in-
tional choices play out to establish the overall sofar as differences are found across countries
level of [inequality]” (327). in the level of inequality, they are driven by
The implication is simple: the high-inequality cross-national differences in the size of second-
countries are distinctive mainly by virtue of ary effects. Where secondary effects are large,
568 • Part VI: Who Gets Ahead?

educational inequality tends to be large; where same level of performance have a significant im-
they are small, inequalities are smaller. In con- pact, policies aimed at changing constraints and
trast, there is much less variation in the size of incentives might hold more promise (Jackson et
primary effects across countries. This finding al. 2007).
suggests that treating inequality as a consequence
of primary and secondary effects is in fact a rather NOTES
parsimonious model of how cross-national dif- 1. Primary and secondary effects may also be
understood as restatements of basic statistical con-
ferences in inequality are generated, a model that
cepts, in that they represent the indirect and direct
identifies differences in choices made by mem-
effects of social origin on educational transitions.
bers of different social groups, conditional on The primary effects are the indirect effects of ori-
performance, as the crucial explanandum. gin on transition, operating through performance;
By considering educational inequalities as the secondary effects are the direct effects of origin
the overall consequence of two separate pro- on transition (net of all factors that operate through
cesses, performance and choice, we gain greater performance).
theoretical and empirical precision. Primary and 2. I do not discuss here the extent to which in-
secondary effects are likely to be generated by stitutional features of the countries would be ex-
rather different processes, and as a consequence pected to influence the magnitude of primary and
secondary effects, and of overall inequality, but see
the tools needed to explain the primary effects of
Jackson and Jonsson (2013) for a detailed analysis
differences in performance among social groups of this issue.
will differ from those needed to explain the sec- 3. In countries for which only one measure of
ondary effects of the differences in choices, con- social background is available, the odds ratio refers
ditional on performance (e.g., Jackson 2013b). If to inequality related to either parental class or pa-
the distinction between primary and secondary rental education. For countries where both measures
effects is ignored, sociologists are in danger of are available, the average of the inequality based on
attempting to explain educational inequality as parental class and the inequality based on parental
an unwieldy whole, thus mistaking a dual phe- education is used. There is no evidence for system-
atic differences by social background measure in the
nomenon requiring distinct and separate expla-
estimates of inequality or of primary and secondary
nations for a single composite requiring a single
effects.
explanation.
Aside from the utility of primary and sec- REFERENCES
ondary effects for explaining inequality, the Boudon, Raymond. 1974. Education, Opportunity,
concepts are highly relevant for policy. Just as and Social Inequality. New York: Wiley.
the explanatory tools will differ depending on Breen, Richard, and Jan O. Jonsson. 2005. “Inequal-
whether the focus is on primary or secondary ity of Opportunity in Comparative Perspective:
effects, the appropriate policy interventions will Recent Research on Educational Attainment and
Social Mobility.” Annual Review of Sociology 31:
also differ depending on which type of effect is
223–243.
to be addressed. If differences in performance are
Breen, Richard, Ruud Luijkx, Walter Müller, and
the main drivers of educational inequality, poli- Reinhard Pollak. 2009. “Non-Persistent Inequal-
cies aimed at reducing those differences will have ity in Educational Attainment: Evidence from
a very large impact in reducing overall inequality. Eight European Countries.” American Journal of
Early interventions, such as intensive preschool Sociology 114: 1475–1521.
education or economic support for families with Cameron, Stephen V., and James J. Heckman. 1999.
young children, would appear to have great po- “Can Tuition Combat Rising Wage Inequality?”
tential for reducing primary effects (cf. Cameron In Financing College Tuition: Government Policies
and Heckman 1999; Carneiro and Heckman and Educational Priorities, edited by Marvin H.
Kosters, 76–124. Washington, DC: AEI Press.
2003; Heckman 2006). On the other hand, if
Carneiro, Pedro, and James J. Heckman. 2003.
differences in the choices made by students at the “Human Capital Policy.” In Inequality in
 Alon—Towards a Theory of Inequality in Higher Education • 569

America, edited by James J. Heckman and Alan and Secondary Effects.” In Determined to Suc-
B. Krueger, 77–239. Cambridge, MA: MIT ceed? Performance versus Choice in Educational
Press. Attainment, edited by Michelle Jackson, 1–33.
Erikson, Robert, John H. Goldthorpe, Michelle Stanford, CA: Stanford University Press..
Jackson, Meir Yaish, and David R. Cox. 2005. Jackson, Michelle, and Jan O. Jonsson. 2013. “Why
“On Class Differentials in Educational Attain- Does Inequality of Educational Opportunity
ment.” Proceedings of the National Academy of Vary Across Countries? Primary and Secondary
Sciences 102: 9730–9733. Effects in Comparative Context.” In Determined
Heckman, James J. 2006. “Skill Formation and the to Succeed? Performance versus Choice in Educa-
Economics of Investing in Disadvantaged Chil- tional Attainment, edited by Michelle Jackson,
dren.” Science 312: 1900–1902. 306–338. Stanford, CA: Stanford University
Jackson, Michelle, Robert Erikson, John H. Press.
Goldthorpe, and Meir Yaish. 2007. “Primary and Kartsonaki, Christiana, Michelle Jackson, and David
Secondary Effects in Class Differentials in Edu- R. Cox. 2013. “Primary and Secondary Effects:
cational Attainment: The Transition to A-Level Some Methodological Issues.” In Determined to
Courses in England and Wales.” Acta Sociologica Succeed? Performance versus Choice in Educational
50: 211–229. Attainment, edited by Michelle Jackson, 34–55.
Jackson, Michelle, ed. 2013a. Determined to Succeed? Stanford, CA: Stanford University Press.
Performance versus Choice in Educational Attain- Shavit, Yossi, and Hans-Peter Blossfeld, eds. 1993.
ment. Stanford, CA: Stanford University Press. Persistent Inequality: A Comparative Study of Edu-
———. 2013b. “How Is Inequality of Educational cational Attainment in Thirteen Countries. Boul-
Opportunity Generated? The Case for Primary der, CO: Westview Press.

64 • Sigal Alon

Toward a Theory of Inequality in Higher Education

The postsecondary education system in the 2006), and existing institutions expanded. From
United States has expanded in recent decades. a modernization/industrialization perspective,
Between 1955 and 2005 total fall enrollment such growth should have narrowed class gaps
in degree-granting institutions rose from 2.6 to in college attainment by fostering high rates of
17.5 million, and the college enrollment rate of educational upward mobility (Kerr et al. 1960;
high school graduates increased from around 45 Parsons 1970; Treiman 1970). Underprivileged
to 70 percent during the same period (NCES high school graduates should disproportionately
2008). To accommodate this growth, the num- benefit from this expansion, as new avenues of
ber of institutions more than doubled during the upward mobility open to them. Some scholars
same period (reaching forty-three hundred by cast doubts on this optimistic view, predicting

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Sigal Alon, “The Evolution of Class Inequality in Higher Education: Competition,
Exclusion and Adaptation,” American Sociological Review 74:3 (2009), pp. 731–775, published by SKAGE in
2009. The article printed here was originally prepared by Sigal Alon for this fourth edition of Social Stratifi-
cation: Class, Race, and Gender in Sociological Perspective, edited by David B. Grusky. Copyright © 2014 by
Westview Press.
570 • Part VI: Who Gets Ahead?

persistent class inequality in higher education by suggesting three prototypes of evolving in-
(Jencks and Riesman 1968). Indeed, a plethora equality and exploring the causal mechanisms
of studies continues to document class-based that link class stratification to a highly stratified
gaps in college attendance. postsecondary system. I propose a refinement to
To reconcile these findings with the increase theories of social closure by unearthing the twin
in the capacity of the postsecondary system, mechanisms that the privileged use to expand
maximally maintained inequality (MMI) posits inequality: exclusion and adaptation.
that class inequalities in educational attainment
will persist until all members of high-status
groups achieve a threshold level of educational The Prototypes of Class Inequality in
attainment (Raftery and Hout 1993). Moreover, Higher Education
when high-status groups reach a saturation point Class inequality in higher education is prob-
at a certain level of education, inequality simply ably ubiquitous, given the cumulative effect
shifts upward to the next level of attainment, of inequality from kindergarten through high
thereby perpetuating relative class differences. school, but a historical review of competition in
Expanding on this theory, Lucas (2001) dis- admissions provides clues to whether and when
tinguishes between quantitative and qualitative we might discern expansion of class inequality
aspects of inequality at the high school level, in recent decades. Trends in supply of slots and
arguing that privileged groups use their status demand for postsecondary education reveal three
advantages to secure quantitatively similar but phases in the competition in admissions over the
qualitatively superior education. This in turn past several years.1 The first phase (high com-
perpetuates inequalities. Extending this view petition) started in the mid-1950s, when appli-
of effectively maintained inequality (EMI) to cations and enrollment swelled at colleges and
the postsecondary system, we presume that as universities across the United States (see Figure
a larger share of high school graduates reaches 64.1), and a record number of applicants applied
some form of higher education, class differences to schools with competitive admissions. The sec-
in access to selective college destinations become ond phase (low competition) started in the early
more prominent to preserve the status hierar- 1970s. Several social and economic factors de-
chy. Well-off youths gravitate toward the more creased the demand for higher education in the
selective schools, whereas the underprivileged midst of an ongoing expansion that continued to
are increasingly admitted only to less prestigious add new seats every year. In the mid-1970s the
institutions. college-going rate reached a low of 47 percent
These conceptions of inequality raise two of high school seniors (Figure 64.1). As a result,
fundamental issues. First is the meaning of the supply of seats in most institutions surpassed
“maintaining” inequality and whether qualitative demand, which led to less competition for the
differences replace or augment quantitative dif- available seats. Elite institutions did not literally
ferences in importance over time. If the former is experience a shortage of applicants, but they, like
true, the level of inequality will persist over time. other schools, felt the crunch. Another phase of
Yet if the privileged use their socioeconomic high competition began in the mid-1980s. As
advantages to secure both quantitatively and the rate of college attendance increased (reach-
qualitatively better outcomes, the class divide ing 62 percent by 1992 and continuing to rise
will expand. The second, greater challenge for thereafter; see Figure 64.1), the competition for
the sociological imagination is the question of slots in higher education increased dramatically,
how class inequality is able to increase, and how especially at selective institutions. The surge in
such a process is woven into the meritocratic demand allowed admissions officers at the most
ethos of postsecondary education. This study prestigious colleges to select students from a sur-
broadens the current conceptions of inequality plus of high-quality applicants (NACAC 2006).
 Alon—Towards a Theory of Inequality in Higher Education • 571

Figure 64.1. Annual College Enrollment Rate for High School Seniors, 1960 to 2006

70

65

60
Percent

55

50

45

40
1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005

Source: The National Center for Education Statistics (NCES), Digest of Education Statistics, 2007 (NCES 2008-022)
(Washington, DC: Institute of Education Sciences, US Department of Education, 2008), Table 191.
Note: Empty markers designate the cohorts of the three national surveys.

Moreover, the establishment of annual college during adjacent periods of rising and declining
rankings in the early 1980s drew attention to competition for slots in higher education.
the stratification among institutions. The evidence confirms these predictions.
I argue that this volatility of competition Using three nationally representative cohorts
in admissions delineates the magnitude of the of seniors who graduated from high school in
class divide in postsecondary education. When 1972, 1982, and 1992, respectively, I examine
the supply of postsecondary education surpasses how class-based inequality in higher education
demand, both qualitative and quantitative dif- changed between 1972 and 1992.2 The empir-
ferences should contract, and class inequality ical investigation is based on intercohort com-
regarding access to the least selective two- and parisons of the variation in college destinations
four-year destinations should decline; inequal- of students from the top and bottom socioeco-
ity should decrease or remain the same vis-à-vis nomic statuses’ quartiles.3 The results show that
four-year selective destinations. This pattern, high school graduates from low socioeconomic
dubbed effectively declining inequality (EDI), is strata are at a marked disadvantage in accessing
expected when the level of competition is low, postsecondary education in all three cohorts.
such as during the 1970s. Alternatively, when the Yet the volatility in the level of class inequal-
college squeeze intensifies, qualitative differences ity matches the competition phases that were
augment quantitative ones, and the class divide identified in the historical review: EDI during
should widen in access to all college destinations, the 1970s (the class divide narrowed substan-
especially selective institutions. We can therefore tially for all postsecondary destinations except
expect to see a pattern of effectively expanding more selective ones, for which class inequality
inequality (EEI) starting from the mid-1980s. remained unchanged) and EEI starting in the
These predictions about the evolution of class mid-1980s (a widening in the class gaps in access
inequality are tested by comparing group differ- to all postsecondary education, especially more
ences in access to and type of college attended selective schools).
572 • Part VI: Who Gets Ahead?

The key questions are how class inequality Trusheim 1988; Fischer et al. 1996; Rothstein
can grow at times of expansion and what fa- 2004; Sacks 1999), high school graduates from
cilitates privileged youths’ improved access to the top socioeconomic echelons are better sit-
the postsecondary system and their inclination uated to benefit from the shifting meritocracy.
to attend selective institutions in an ostensible A test-score barrier to admission may perpet-
meritocracy. uate class inequality, but it does not solve the
problem of hereditary transmission, that is, the
common desire to transmit privileges to one’s
A Mechanism-Based Theory for own. All students—affluent, poor, and middle
the Evolution of Class Inequality in class—face the same academic barriers, because
Higher Education exclusionary rules are universal. Parkin (1979,
Weber’s notion of social closure can help us un- 63) describes one strategy that can preserve the
derstand the mechanisms that expand class-based kinship link and the intergenerational transmis-
inequality in higher education. Social closure is a sion of status—the adaptation of the bourgeois
process whereby social collectivities seek to max- family to institutional demands:
imize rewards by restricting access to resources
and opportunities to a limited circle of eligible In systems based on aristocratic, caste, or ra-
individuals. Certain social or physical attributes cial exclusion, families of the dominant group
are used as the criteria for exclusion, based more can expect to pass on their privileged status
on selection than on hereditary transmission to their own descendants as a direct result of
(Parkin 1979). Exclusion is primarily a collective the closure rules in operation, however so-
act intended to promote class formation through cially lethargic those families might be. The
careful selection of successors (Collins 1971; bourgeois family, by contrast, cannot rest
Parkin 1979). From this perspective, high-status comfortably on the assumption of automatic
groups struggle to maintain their privileges by class succession; it must make definite social
influencing the criteria used to control access to exertions of its own or face the very real pros-
higher education (Karabel 1984, 2005; Karen pect of generational decline. In other words,
1990). Although the ability to pay for education although the typical bourgeois family will cer-
is an important exclusionary device, the main tainly be better equipped than most to cope
such device for monitoring entry into higher with a closure system on its children’s behalf,
education and determining college choice is the it must still approach the task more in the
inflated use of academic qualifications (Bowen manner of a challenge with serious risks at-
2006; McPherson and Schapiro 2006). tached than as a foregone conclusion.
Perceived by the public as a legitimate mech-
anism for sorting high school graduates accord- Accordingly, I argue that the privileged group
ing to meritocratic rules, the use of standardized not only seeks to shape the contours and the im-
test scores has intensified since the mid-1980s portance of the admissions criteria (to preserve
in the growing competition for slots, in an ef- the collective), but it also devotes considerable
fort to ease the evaluation of a growing volume effort to cultivating its offspring’s stock of ac-
of highly qualified and heterogeneous student ademic currencies to ensure succession along
pools (Alon and Tienda 2007; Lemann 1999; kinship lines. Parents of middle and high socio-
NACAC 2006).4 In this “shifting meritocracy,” economic status (SES) convey their postsecond-
institutions now rely heavily on test scores when ary expectations when their children are young,
making admissions decisions, whereas other better understand the postsecondary landscape
merit criteria, like class rank, have declined in and competitive admissions process, are aware
importance (Alon and Tienda 2007). Because of the importance of test scores in admissions
socioeconomic status is highly correlated with decisions, and invest in resources to promote
test achievement (Blau et al. 2004; Crouse and college attendance. Vigorous use of expensive
 Alon—Towards a Theory of Inequality in Higher Education • 573

test preparation tools, such as private classes The experiences of the three cohorts that are
and tutors, is a key element in the adaptation analyzed here (those who graduated from high
strategy (Buchmann, Roscigno, and Condron school in 1972, 1982, and 1992) support this
2006; Briggs 2001; NACAC 2009). These adap- argument.
tation strategies elevate privileged students’ test Changes in test scores, both their admission
achievements. In turn, this leads to class-based value and their level, are mostly responsible for
polarization of resources, placing privileged stu- the shifts in the class divide in the 1970s and
dents in a better starting position for admission 1980s. However, there is a difference between
to selective schools. the two periods. When the competition in
Adaptation (which results in class-based po- admissions was low, between 1972 and 1982,
larization in test scores) accompanies and bolsters institutions did not put much weight on test
exclusion (a greater emphasis on test scores in scores in admissions decisions, and their value
admission). These two dimensions reinforce one declined at all college destinations. By contrast,
another and amplify inequality. They also expe- the weight that selective institutions placed on
dite the pace at which the privileged cement their test scores in their admission decisions increased
superior position. In this way, class inequality in between 1982 and 1992. This greater emphasis
higher education not only persists but actually on test scores demonstrates their use as an ex-
expands, despite the system’s growing capacity. clusionary device at times of growing compe-
Figure 64.2 presents this comprehensive scheme tition in admissions, that is, to sort applicants
for the evolution of class inequality in higher and select students. However, the results also
education. When competition for admission is demonstrate that one of the dramatic changes
high, as has been the case since the 1980s, ex- in the 1970s and 1980s for the evolution of
clusion and, particularly, adaptation increase class inequality was the shift in the levels of
class-based inequality vis-à-vis all college desti- test scores. To illustrate that I ranked all high
nations. Inequality effectively expands. Alterna- school graduates on the percentile distribution
tively, when colleges struggle to fill their classes, of test scores, Figure 64.3 shows the interco-
as was the case during the 1970s, all postsecond- hort changes in the level of test scores of high
ary destinations relax their admissions barriers, school graduates from the top and bottom SES
and the system becomes more inclusive. When quartiles (this is a box-plot representation of the
exclusionary barriers are negligible, the privileged percentile distributions of test scores). It shows
do not need to cope with a closure system. that within cohorts, SES differences in test
Most sociological writing focuses on exclu- scores are considerable. In 1972 the test scores
sionary practices; there is little understanding of low-SES high school graduates were, on av-
of the role that adaptation plays in expanding erage, in the 35th percentile; high-SES seniors
inequality. Exclusion often takes ideological pre- were, on average, in the 68th percentile. This
cedence over adaptation, because of its collective class divide in test scores existed in all years. Yet
property of ensuring class reproduction (Parkin there are important temporal swings in the po-
1979). The omission of adaptation from stratifi- sition of both low- and high-SES seniors in the
cation theory, however, obscures the mechanisms test score distributions. Between 1972 and 1982
that expand class inequality in higher education; disadvantaged seniors improved their position
it is thus unclear which aspect is more important in the overall percentile test score distribution
for the polarization of the status hierarchy. Fill- (from 35 to 38 on average), especially the high
ing this lacuna, I argue that because adaptation achievers among them (those located in the
varies by social class, unlike exclusion, which is group’s top quartile). Simultaneously, all privi-
universal, adaptation is more effective in expand- leged seniors saw their advantage decline (from
ing class inequality. If this is so, adaptation is the 68th to 65th on average). In summary, during
key mechanism underlying the widening class the 1970s, a decade of declining competition
divide in US higher education. in admissions, there was a convergence in test
574 • Part VI: Who Gets Ahead?

Figure 64.2. Comprehensive Scheme for the Evolution of Class Inequality in Higher Education

 !




!

  
 
  

  


 
"#

  
 



 




  




 !





  

 

  

 
 

!


 
" #

  

 





  
 

  



 !

  

 !




  
 
 

!

 
"#







 


  



 !


  


  

scores of low- and high-SES youth. The gap in in test scores between high- and low-SES high
test scores narrowed. school graduates. Notably, among high-SES
However, parallel to the intensification of the seniors, the low achievers (those at the bottom
college squeeze in the 1980s, class-based polar- quartile) improved most in the test score distri-
ization in test scores emerged. While low-SES bution between 1982 and 1992, whereas among
seniors’ test scores declined, on average, from low-SES seniors, plummeting test scores affected
the 38th percentile in 1982 to the 32nd per- the high-achieving seniors the most. Overall, in
centile in 1992, high-SES seniors improved 1982 high-achieving, low-SES seniors surpassed
their relative standing (from the 65th to the their low-achieving, high-SES counterparts
70th percentile, on average). This shows that (60th versus 46th percentile), whereas the sit-
the privileged adapted to the changing closure uation was reversed in 1992 (49th versus 55th
rules by elevating their test scores, while their percentile, respectively). As a result, we find very
underprivileged classmates were unable to follow little overlap in the test score distributions of the
suit. This process led to the widening of the gap two groups by 1992.
 Alon—Towards a Theory of Inequality in Higher Education • 575

Figure 64.3. Change Over Time in the Percentile Distribution of Test Scores by SES Quartiles

Bottom SES Q Top SES Q

100

100
80

80
60

60
pctsat

pctsat
40

40
20

20
0

1972 1982 1992 1972 1982 1992

Bottom SES Quartile Top SES Quartile

1972 1982 1992 1972 1982 1992
Average
Average 35 38 32 68 65 70
Percentile
25% 12 14 11 50 46 55
50% 29 33 27 73 68 77
75% 53 60 49 89 87 90
Note: This plot portrays the central 50 percent of a distribution (interquartile range Q1–Q3), from the lower to the upper quartiles.
The line within the box marks the median, and the lines below and above the box cover the entire range of values. Dots represent
outliers.

In sum, starting in the mid-1980s two linked education. Privileged high school seniors bene-
developments are apparent: on the one hand, fited from the rising academic admission bar, but
selective institutions started to put greater em- even more from their ability to get ahead in the
phasis on test scores in making admissions test score distribution. For most of them, adap-
decisions; on the other hand, the gaps in test tation was more vital than exclusion in climb-
scores between youths from high and low SES ing the social hierarchy of US higher education.
widened. Did these forces, exclusion and ad- Likewise, the failure of high school graduates
aptation, contribute equally to the rising class from the bottom socioeconomic strata to keep
inequality in higher education? Further analyses pace with their privileged counterparts in the
reveal that adaptation (class-based polarization test scores race was detrimental to their college
in test scores) is more effective than exclusion prospects, especially for those with the highest
(a greater emphasis on test scores in admissions) potential. Class-based polarization in test scores
in expanding class inequality in postsecondary is ultimately the key factor in enabling high-SES
576 • Part VI: Who Gets Ahead?

groups to expand both qualitative and quantita- being attuned to changing circumstances, devote
tive advantages in postsecondary education. considerable effort to cultivating their own stock
of the currencies required for entry into lucrative
positions. Exclusion may take ideological prece-
Adaptation: The Key Mechanism for dence over adaptation, as Parkin (1979) argues.
Expanding Inequality However, as this study reveals, adaptation is a
Social class has a persistent impact on enrollment more effective mechanism for increasing class in-
and access to selective postsecondary schooling. equality in US higher education. By expanding
Yet the magnitude of class inequality is tightly the supply of high-scoring applicants, adaptation
bound to the level of competition in admissions. sparks a tightening of selection criteria, fuels ex-
When the supply of seats surpasses demand, clusion, and artificially magnifies the competi-
competition in admissions is low, admissions tiveness of the admissions process. The vicious
barriers are relaxed, and the postsecondary sys- cycle of adaptation and exclusion practices has
tem becomes more inclusive, especially at the devastating implications for the ethos and oper-
lower rungs of the ladder. Consequently, class ation of meritocracy in higher education, divert-
inequality declines. Conversely, increasing de- ing it from being the great American equalizer.
mand for a college diploma intensifies competi-
tion, which augments the value of test scores in NOTES
the admissions decisions of selective institutions 1. I gathered information from the US Census
Bureau, the US Department of Education, and the
(exclusion). Adaptation of the privileged to the
National Center for Education Statistics, as well as
changing closure rules follows, creating a po-
from the archives of several selective institutions
larization of resources and amplifying the class from the 1950s.
divide for all postsecondary destinations. Class 2. The three national data sets are the National
inequality in higher education expands. Longitudinal Study of the High School Class of 1972
There is a crucial discrepancy between exclu- (NLS72), High School and Beyond (HS&B), and
sion and adaptation that has serious ramifica- National Education Longitudinal Survey (NELS).
tions for equality of opportunity. Closure rules 3. The analyses use an inclusive measure of post-
are universal (i.e., they are the same for every secondary destinations that has five outcomes: (1)
applicant) and class blind. Thus, changes in no postsecondary education, (2) two-year open-door
colleges, (3) four-year nonselective colleges, (4)
admissions barriers uniformly shape all groups’
four-year selective colleges (median SAT 900 to
chances of admission. Rich and poor applicants 1050), and (5) four-year more selective colleges (me-
are equally affected by tightening admissions. dian SAT above 1050).
This curbs the potential of exclusion to expand 4. In addition to ranking applicants, standardized
inequality. Conversely, by assuring intergenera- test scores have played a unique role in the stratifi-
tional transmission of status along kinship lines, cation of higher education, because they are also the
adaptation is particularistic, sensitive to class dif- main criteria used to rank postsecondary institutions.
ferences, and thus unequal by default. All parents
want their offspring to succeed in life, and as part REFERENCES
Alon, Sigal, and Marta Tienda. 2007. “Diversity,
of good parenting practices they are expected to
Opportunity and the Shifting Meritocracy in
invest in their children’s education. From a soci-
Higher Education.” American Sociological Review
etal perspective, however, it is troubling that only 72 (3): 487–511.
the privileged make the behavioral adjustment Ashenfelter, Orley. 1978. “Estimating the Effect of
required to meet the rising admissions demands, Training on Earnings.” Review of Economics and
whereas others are unable to follow suit. Statistics 60 (1): 47–57.
This suggests that the concept of adaptation Ashenfelter, Orley, and David Card. 1985. “Using
is the cornerstone to building a comprehensive the Longitudinal Structure of Earnings to Esti-
theory about the evolution of inequality. Inequal- mate the Effect of Training Programs.” Review of
Economics and Statistics 67 (4): 648–660.
ity is expanding primarily because the privileged,
 Alon—Towards a Theory of Inequality in Higher Education • 577

Blau, Judith R., Stephanie Moller, and Lyle V. Jones. Long, J. Scott. 1997. Regression Models for Categor-
2004. “Why Test? Talent Loss and Enrollment ical and Limited Dependent Variables. Thousand
Loss.” Social Science Research 33: 409–434. Oaks, CA: Sage Publications.
Bowen, William G. 2006. “Extending Opportunity: Lucas, Samuel R. 2001. “Effectively Maintained In-
What Is to Be Done?” In College Access: Opportu- equality: Education Transitions, Track Mobility,
nity or Privilege?, edited by S. M. McPherson and and Social Background Effects.” American Jour-
M. O. Schapiro, 19–34. New York: The College nal of Sociology 106 (6): 1642–1690.
Board. McPherson, S. Michael, and Morton Owen Scha-
Briggs, Derek C. 2001. “The Effect of Admissions piro. 2006. “Introduction.” In College Access: Op-
Test Preparation: Evidence from NELS:88.” portunity or Privilege?, edited by S. M. McPherson
Chance 14: 10–22. and M. O. Schapiro, 3–15. New York: The Col-
Buchmann, Claudia, Vincent J. Roscigno, and lege Board.
Dennis J. Condron. 2006. “The Myth of Mer- The National Association for College Admission
itocracy? SAT Preparation, College Enrollment, Counseling (NACAC). 2006. State of College
Class and Race in the United States.” Paper pre- Admission, by David A. Hawkins and Melissa
sented at the annual meeting of the American Clinedinst, NACAC. http://chsweb.lr.k12.nj.us/
Sociological Association, Montreal, August. jmcmichael/2006%20SOCA.pdf (accessed June
Collins, Randall. 1971. “Functional and Conflict 2009).
Theories of Educational Stratification.” American ———. 2009. “Preparation for College Admission
Sociological Review 36: 1002–1019. Exams.” A NACAC Discussion Paper by Derek
Crouse, James, and Dale Trusheim. 1988. The Case C. Briggs. http://www.nacacnet.org/Publication-
against the SAT. Chicago and London: University sResources/Research/Documents/TestPrepDis-
of Chicago Press. cussionPaper.pdf (accessed June 2009).
Fischer, Claude S., Michael Hout, Martin Sanchez The National Center for Education Statistics
Jankowski, Samuel R. Lucas, Ann Swidler, and (NCES). 2008. Digest of Education Statistics,
Kim Voss. 1996. Inequality by Design: Cracking 2007. Tables 179 and 191. Washington, DC: In-
the Bell Curve Myth. Princeton, NJ: Princeton stitute of Education Sciences, US Department of
University Press. Education (NCES 2008-022).
Jencks, Christopher, and David Riesman. 1968. Parkin, Frank. 1979. Marxism and Class Theory: A
The Academic Revolution. Garden City, NY: Bourgeois Critique. London: Tavistok.
Doubleday. Parsons, Talcott. 1970. “Equality and Inequality in
Karabel, Jerome. 1984. “Status-Group Struggle, Modern Society, or Social Stratification Revis-
Organizational Interests, and the Limits of Insti- ited.” In Social Stratification: Research and Theory
tutional Autonomy: The Transformation of Har- for the 1970s, edited by E. Laumann, 13–72. In-
vard, Yale, and Princeton, 1918–1940.” Theory dianapolis, IN: Bobbs-Merrill.
and Society 13 (1): 1–40. Raftery, A. E., and M. Hout. 1993. “Maximally
———. 2005. The Chosen: The Hidden History of Maintained Inequality: Expansion, Reform and
Admission and Exclusion at Harvard, Yale, and Opportunity in Irish Education, 1921–1975.”
Princeton. Boston: Houghton Mifflin. Sociology of Education 66: 41–62.
Karen, David. 1990. “Toward a Political- Rothstein, Jesse M. 2004. “College Performance
Organizational Model of Gatekeeping: The Case Predictions and the SAT.” Journal of Econometrics
of Elite Colleges.” Sociology of Education 63: 121: 297–317.
227–240. Sacks, Peter. 1999. Standardized Minds: The High
Kerr, Clark, John T. Dunlop, Fredrick H. Harbison, Price of America’s Testing Culture and What We
and Charles A. Myers. 1960. Industrialism and Can Do to Change It. Cambridge, MA: Perseus
Industrial Man. Cambridge, MA: Harvard Uni- Books.
versity Press. Treiman, Donald. 1970. “Industrialization and
Lemann, Nicholas. 1999. The Big Test: The Secret Social Stratification.” In Social Stratification:
History of the American Meritocracy. New York: Research and Theory for the 1970s, edited by
Farrar, Straus and Giroux. Edward O. Laumann, 373–394. Indianapolis:
Bobbs-Merrill.
578 • Part VI: Who Gets Ahead?

65 • Florencia Torche

Does College Still Have Equalizing Effects?

A college degree yields substantial economic attainment is independent of their socioeco-

returns. By the early twenty-first century, col- nomic origins. In other words, a college degree
lege graduates’ earnings were about 90 percent fulfills the promise of meritocracy: it offers equal
higher than those of their high school graduate opportunity for economic success regardless of
counterparts, a premium that has increased over the advantages of birth.
the last quarter century (Autor, Katz, and Kear- This finding is not a US anomaly. Research-
ney 2008). College attainment is also related to ers have shown a weaker intergenerational as-
better health, longevity, happiness, and a host of sociation at higher levels of schooling in other
extra-economic outcomes (Ross and Mirowsky industrialized countries such as France, Sweden,
1999; Pallas 2000; Rowley and Hurtado 2003; and Germany (Vallet 2004; Breen and Jonsson
Attewell and Lavin 2007; Stevens, Armstrong, 2007; Breen and Luijkx 2007). However, the US
and Arum 2008). case is the clearest one in which the intergenera-
But college attainment is related to more tional socioeconomic association fully disappears
than economic and extra-economic well-being. among college graduates, providing “a new an-
An important finding of stratification research is swer to the old question about overcoming dis-
that the direct influence of parental resources on advantaged origins: A college degree can do it”
the economic position of adult children is much (Hout 1988, 1391).
weaker among college graduates than among
those with less schooling (Hout 1984, 1988).
That influence decreases to almost zero among A New Landscape for the
college graduates. Meritocratic Power of a
It does not of course follow that all social in- College Degree?
equality has been eliminated. Access to college is These findings come largely from the 1970s.
strongly dependent on parental resources (Ell- Many things have changed since then. College
wood and Kane 2000; Haveman and Smeeding expansion and differentiation and the increase
2006), and the socioeconomic gap in access ap- in postbaccalaureate advanced degrees define a
pears to have increased over time (Kane 2004; new educational landscape that may have altered
Belley and Lochner 2007). The finding men- mobility patterns of college graduates.
tioned previously, however, means that for those A notable change over the last quarter cen-
who receive a college degree, their socioeconomic tury has been an increase in the proportion of

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in the
following publication: Florencia Torche, “Is a College Degree Still the Great Equalizer? Intergenerational Mo-
bility Across Levels of Schooling in the United States,” American Journal of Sociology 117:3 (November 2011),
pp. 763–807, published by the University of Chicago Press. The article printed here was originally prepared by
Florencia Torche for this fourth edition of Social Stratification: Class, Race, and Gender in Sociological Perspective,
edited by David B. Grusky. Copyright © 2014 by Westview Press.
 Torche—Does College Still Have Equalizing Effects? • 579

Figure 65.1. Percentage of Adults 30–60 with a Bachelor’s Degree and an Advanced Degree by Gender
and Year, United States 1965–2005

Source: Current Population Survey, 1965–2005, March annual demographic file data. (Note: This table appears in
Torche [2011].)

adult Americans with a college degree. Accord- (Gerber and Cheung 2008). Differentiation
ing to the Current Population Survey (CPS), the matters for the intergenerational reproduction
percentage of men ages thirty to sixty with a col- of inequality if the type of college education re-
lege degree grew from 13 percent in 1965 to 30 ceived depends on socioeconomic origins and in
percent in 2005, and for women there is an even turn shapes the economic outcomes of college
more impressive increase, from 8 to 29 percent. graduates, a phenomenon called “horizontal
To date, stratification research has treated stratification.”
college graduates as a single, homogeneous cat- College differentiation involves diverse do-
egory. However, this group comprises two dis- mains, but the literature has highlighted two:
tinct levels of attainment: bachelor’s degree and institutional selectivity and field of study. Stud-
advanced degree. As Figure 65.1 shows, in 1970 ies show an association between advantaged so-
only 5 percent of adult men and 1 percent of cial origins and college selectivity (Davies and
women held a degree beyond a bachelor’s, in- Guppy 1997; Karen 2002), an association that
cluding master’s, first-professional, and doctoral appears to have increased over time (Astin and
degrees. By 2005 these numbers had reached Oseguera 2004). Evidence about field of study
11 percent for men and 10 percent for women. is limited and less conclusive. Although the as-
The substantial increase in the proportion of ad- sociation between social origins and a lucrative
vanced degree holders renders them an impor- major appears to be weak (Davies and Guppy
tant group that should be studied separately 1997), an indirect influence is likely to exist.
from those with just a bachelor’s degree. Upper-class students are more likely to major
In parallel with its expansion, postsecondary in the arts and sciences, which in turn increases
education has undergone differentiation in in- their chances of pursuing an advanced degree
stitutional characteristics and college experience (Goyette and Mullen 2006).
580 • Part VI: Who Gets Ahead?

The association between social origins and labor market (Grusky and Weeden 2008) and
college differentiation shapes inequality to the among the “overclass,” whose income largely
extent that college locations accessed by the depends on returns to capital. By using fam-
upper class yield higher economic returns. Ev- ily income, I am able to assess the economic
idence consistently suggests that graduates of well-being of those not in the labor force and
more selective institutions earn more (Brewer include occupational and extra-occupational
and Ehrenberg 1996; Thomas and Zhang 2005), sources of well-being. Furthermore, this mea-
although the “selective college” effect may be at sure accounts for family-level dynamics, such
least partially driven by academic performance as spousal selection (assortative mating), shown
and ability of recruits (Loury and Garman 1995; to play a crucial role in the transmission of ad-
Monks 2000; Dale and Krueger 2002). As for vantage across generations (Chadwick and Solon
field of study, research shows substantial variation 2002; Ermisch, Francesconi, and Siedler 2006).
in returns across fields, with business-related,
math, engineering, and health majors receiving
higher earnings, and education-related fields re- Results: Is a College Degree Still the
ceiving lower returns (Berger 1988; Grogger and Great Equalizer?
Eide 1995; Carnevale et al. 2011). Figure 65.2 presents the main findings of this
To date, virtually no study has examined analysis. It displays the level of intergenerational
differentiation at the advanced-degree level, income mobility for men with different levels of
but there is a sharp earning gradient across pro- schooling—less than high school, high school
grams—with professional degrees such as med- graduate, some college, college graduate, and
icine and law at the top, followed by doctoral advanced degree.1 The measure of mobility used
degrees, and master’s degrees a distant third (Day is the intergenerational income elasticity, which
and Newburger 2002; College Board 2005). captures the association between parents’ income
This pattern suggests that differential access by and adult children’s income. Elasticity measures
social origins to a particular type of advanced range from 0 to 1.
degree program may provide an avenue for the An elasticity of 0 represents “perfect mobil-
intergenerational reproduction of advantage. ity,” in which the incomes of parents and chil-
The findings about the “meritocratic power” dren are completely unrelated. An elasticity of
of a college degree discussed at the outset of this 1 represents “perfect immobility,” in which the
chapter referred specifically to intergenerational income advantage of parents is mirrored in their
class mobility. Recent developments in mobil- children’s generation. An elasticity of 0.4, for ex-
ity research show that measures such as class, ample, means that a 10 percent difference in two
occupational status, earnings, and family in- families’ incomes is associated with a 4 percent
come capture different dimensions of economic difference in their sons’ incomes.
well-being and suggest that mobility findings A striking U-shaped pattern of association
may be contingent on the measure used. Given across levels of schooling emerges from the anal-
these findings, this study examines various mea- ysis. The intergenerational elasticity is substan-
sures of socioeconomic advantage. I pay partic- tial for those with less than a college degree, but
ular attention to family income. Measures such it declines to a value not significantly different
as social class, occupational status, and earnings from 0 for those who graduated from college.
are based on labor market participation, but As in the past, a college degree erases the im-
total family income includes extra-occupational pact of socioeconomic origins on economic
components, such as financial assets and public success. However, the influence of social origins
and private transfers. These resources are central reemerges for those with a graduate degree. For
at either extreme of the economic distribution, graduate degree holders, a 100 percent increase
among the “underclass” poorly attached to the in parents’ income results in a more than 40
 Torche—Does College Still Have Equalizing Effects? • 581

Figure 65.2. Intergenerational Family Income Elasticity among Men Born 1957–1961 (NLSY) and
1951–1966 (PSID)
0.7
Intergenerational income elasticity

0.6

0.5

0.4

0.3

0.2

0.1

0
Less than High High School Some College Advanced
School graduate college graduate degree
Level of education
NSLY-79 PSID
Source: National Longitudinal Survey of Youth 1979 and Panel Study of Income Dynamics.

percent increase in their own incomes, an even To the extent that these attributes are rewarded
larger magnitude than for those with less than a in the labor market, lower-class college graduates
high school diploma. The U-shaped pattern is will be economically successful regardless of their
not idiosyncratic to family income, but rather humble origins. This hypothesis claims that such
emerges—although more weakly—for other in- selectivity accounts for the high level of mobility
dicators of socioeconomic advantage, including among BA holders.
class, occupational status, and earnings (Torche In contrast, the labor market meritocracy hy-
2011). pothesis suggests that college graduates tend to
The sharply different levels of intergenera- work in sectors and firms where meritocratic
tional mobility of BA holders vis-à-vis advanced selection is more prevalent and origin charac-
degree holders begs the question about mecha- teristics count for less, insofar as higher quali-
nisms. Two competing hypotheses account for fications are a powerful signal for employers,
these findings: dynamic selectivity and labor leaving little leeway for social network effects
market meritocracy. The dynamic selectivity hy- (Breen and Jonsson 2007: 1778). Highly bureau-
pothesis posits that as students advance in their cratized contexts in which college graduates are
educational careers, the association between likely to be employed operate as “great levelers”
social origins and unobserved determinants of (Baron et al. 2007) because their formal prac-
socioeconomic attainment, such as cognitive tices reduce subjectivity in personnel decisions,
ability or motivation, declines (Mare 1980). In ensuring that opportunity and rewards reflect
other words, given the substantial economic and role-specific qualifications and performance
cultural barriers that lower-class students face in (Tomaskovic-Devey 1993; Bielby 2000; Reskin
obtaining a college degree, those who “make it” 2000; Elvira and Graham 2002).
to college are positively selected on unobserved The evidence emerging from this analysis is
attributes—they are “the best and the brightest.” not consistent with the selectivity hypothesis.
582 • Part VI: Who Gets Ahead?

First, the role of selectivity in accounting for based on social origins: horizontal differentiation
intergenerational mobility at a particular educa- in postsecondary education, allocative inequal-
tional level should decrease as that level expands ity, and within-occupation differences in economic
and the relative number of credential holders rewards (Treiman and Hartmann 1981; En-
grows. In the extreme, when an educational gland 1992; Petersen and Morgan 1995; Pada-
level becomes universal, selectivity is by defini- vic and Reskin 2002). As discussed, horizontal
tion null (Raftery and Hout 1993). However, as differentiation matters for mobility if individ-
I have shown, mobility among BA holders did uals with lower-class origins attain a degree in
not decrease as this level expanded over the last low-prestige institutions and less lucrative fields
quarter century, questioning a negative effect of of study. Allocative inequality refers to differences
declining selectivity. in occupational placement after completing ed-
Second, the intergenerational association is ucation, so that lower-class individuals are con-
stronger among advanced degree holders than centrated in relatively low-paying occupations.
among BA holders. But several factors suggest Within-occupation rewards inequality emerges
that, if unobserved selectivity accounted for when individuals with disadvantaged origins re-
mobility, mobility should be higher, not lower, ceive lower economic returns than their advan-
among advanced degree holders. Research shows taged peers even if they are employed in the same
that the association between social origins and occupation. A meritocratic labor market would
enrollment in a graduate program conditional mean, then, that college graduates of lower-class
on college graduation is weaker than at earlier origins have the same chances as their upper-class
educational levels (Mare 1980) and virtually peers to access prestigious universities and lucra-
null for at least some programs, such as mas- tive fields of study and to engage in profitable
ter’s degrees and MBAs (Stolzenberg 1994; occupations, and that they receive similar earn-
Mullen, Goyette, and Soares 2003). As a re- ings conditional on their occupations.
sult, lower-class individuals who remain in the I evaluate the meritocracy hypothesis by ex-
educational system after completing a BA may amining the type of college education attained
be exceptionally selected on attributes such as and the type of occupation held by male col-
motivation and ability. In addition, attending lege graduates from different social origins. To
graduate school involves spending additional measure social origins, I divide parental income
time in the educational system, undergoing not into three groups—the poorest third, the middle
only formal training but also professional social- third, and the wealthiest third—and evaluate the
ization, as well as building social connections. educational and occupational outcomes of BA
This extended exposure may contribute to the holders coming from each third. The results are
development of networks of professional referral presented in the top left graph of Figure 65.3.
among lower-background students, providing Analysis of institutional selectivity of the college
expanded opportunities to detach themselves attended shows that BA holders with origins in
from their disadvantaged origins. Furthermore, the wealthiest income third are more likely to
advanced degrees typically provide more specific attend selective institutions: 30 percent versus
and technically sophisticated skills than those ac- 12 percent of those in the bottom income third.
quired through a BA, which could rule out the However, the differences in field of study across
use of cultural capital based on social origins or social origins are relatively minor. Upper-class
social networks as determinants of occupational BA holders are somewhat more likely to major
placement and rewards (Jackson 2007). in the social sciences and less likely to major in
If selectivity does not explain the findings, education, but these differences do not extend
does labor market meritocracy do so? To test to other majors.
this hypothesis, I focus on three processes po- The bottom left graph in the figure exam-
tentially leading to differences in opportunity ines whether BA holders from different social
 Torche—Does College Still Have Equalizing Effects? • 583

Figure 65.3. Institutional Selectivity, Field of Study, Type of Program, and Occupation among Male
Bachelor’s Degree Holders and Advanced Degree Holders

Source: Baccalaureate and Beyond 1993–2003 and National Longitudinal Survey of Youth 1979.

origins display unequal occupational outcomes. in the top income third are more likely to have
I distinguish five occupational groups: managers; attended selective institutions: 47 percent versus
professionals in computer sciences, engineering, 31 percent among lower-class graduates. But dif-
and math (CSEM); other professionals; workers ferences in social origins are not restricted to in-
in sales and administration; and craft workers stitutional selectivity. Rather, they extend to field
and operatives. Although a more finely grained of study and type of program. Advanced degree
occupational classification would be prefera- holders with origins in the top income third are
ble, the limited sample sizes constrain further also more likely to earn degrees in professional
disaggregation. fields such as business, health, and law. They
Allocation into occupational groups is re- are also much more likely to obtain lucrative
markably even across social origins. The main first-professional degrees (about 90 percent of
source of difference is that college graduates which are in medicine, dentistry, and law) and
coming from the lower income third are less MBAs, rather than MAs. The differences based
likely to hold managerial jobs: 36 percent ver- on social origins are magnified in the labor mar-
sus 44 percent of their wealthier counterparts. ket. The bottom right-hand graph shows that
This is compensated for by a higher proportion whereas about 38 percent of upper-class advanced
of lower-background individuals in craft/oper- degree holders attain a lucrative managerial occu-
atives occupations. These differences are small, pation, only 17 percent of their lower-class peers
and they pale when compared, for example, with do so. Lower-class graduates are much more
occupational gaps across gender. The same is the likely to engage in a professional occupation that
case with within-occupation earnings: the gra- is not in the CSEM field. These occupational
dient across social origins is discernible but not and educational differences are compounded by
prominent (Torche 2011). within-occupation earnings gaps. On average,
The right-hand graphs in Figure 65.3 repeat a professional with an advanced degree and or-
this exercise for advanced degree holders. As with igins in the lower income third receives earnings
BA holders, advanced degree holders with origins that are only about 60 percent of those of his
584 • Part VI: Who Gets Ahead?

upper-class counterparts—a gap substantially alternative way in which unobserved selectivity

higher than among BA holders. This pattern is may affect observed mobility. As graduate educa-
not restricted to men. Rather, it is even more pro- tion expands, the undergraduate level may serve
nounced among women (Torche 2011). an increasingly important sorting function into
graduate school, differentially allocating individ-
uals according to social background and unob-
Conclusions served characteristics. As highlighted by Goyette
The findings from this analysis are clear. The and Mullen (2006), upper-background students
intergenerational socioeconomic association is may opt to maximize their chances of attending
substantial among those without a college de- graduate school by choosing specific fields and,
gree, but it virtually disappears among those probably, specific postsecondary institutions
with only a bachelor’s degree. Intergenerational and trajectories. In contrast, lower-background
mobility among BA holders supports the hy- individuals may favor locations in the system
pothesis that labor markets for college graduates that would maximize the returns of a bachelor’s
operate on the basis of meritocratic criteria. degree, reducing the risk of downward mobility
However, a second finding from this study (Breen and Goldthorpe 1997).
questions this interpretation: a strong intergen- It is plausible, then, that upper-background
erational association reemerges among advanced students who fail to advance to graduate school
degree holders, reaching levels comparable to have miscalculated, choosing suboptimal college
those with low levels of schooling. This is an fields, institutions, or trajectories. This miscalcu-
unexpected result. Given that the human capital lation may result in downward mobility among
attained by advanced degree holders is more tech- advantaged students, whereas the optimal in-
nically specialized than that of college graduates, vestment of lower-background individuals in a
that it requires spending more time in educa- BA but no subsequent degrees would account
tional institutions undergoing socialization that for upward mobility. The overall result of this
may erase the direct influence of social origins, process would be a weak intergenerational as-
and that it is probably associated with positive sociation among those with a bachelor’s degree
unobserved selectivity, one might expect weaker only, but reduced mobility among advanced de-
rather than stronger intergenerational association gree holders. More analysis of colleges as “sort-
vis-à-vis BA holders. To explain both intergener- ing machines” could shed further light on this
ational mobility of BA holders and intergener- process. The findings from this analysis strongly
ational persistence of advanced degree holders, question the unqualified interpretation of in-
I examine horizontal differentiation in postsec- creasing meritocracy among higher levels of edu-
ondary education, and occupational inequality cation and indicate that mobility opportunity is
for these groups. I found that these factors ac- embedded in educational and labor market pro-
count for both the substantial intergenerational cesses, including, but not reduced to, horizontal
mobility among BA holders and strong intergen- educational stratification and the patterning of
erational reproduction among advanced degree occupational allocation and economic rewards
holders. Among men who attain an advanced by social origins.
degree, their social origins are strongly correlated
with the type of graduate education they obtain, NOTES
in terms of institutional selectivity, field of study, 1. The analysis presented here is restricted to
men. The reader is referred to Torche (2011) for a
and type of program attended. Social origin is
complete analysis by gender, using all measures of
also strongly correlated with the type of job at-
socioeconomic well-being.
tained and the economic rewards received.
The evidence is not consistent with unob- REFERENCES
served selectivity as the cause of high mobility Astin, Alexander, and Leticia Oseguera. 2004.
among college graduates. There is, however, an “The Declining ‘Equity’ of American Higher
 Torche—Does College Still Have Equalizing Effects? • 585

Education.” Review of Higher Education 27: Dale, Stacy, and Alan Krueger. 2002. “Estimating
321–341. the Payoff to Attending a More Selective College:
Attewell, Paul, and David Lavin. 2007. Passing the An Application of Selection on Observables and
Torch. New York: Sage. Unobservables.” Quarterly Journal of Economics
Autor, David, Lawrence Katz, and Melissa Kearney. 117: 1491–1527.
2008. “Trends in U.S. Wage Inequality: Revising Davies, Scott, and Neil Guppy. 1997. “Fields of
the Revisionists.” Review of Economics and Statis- Study, College Selectivity and Student Inequal-
tics 90 (2): 300–323. ities in Higher Education.” Social Forces 75 (4):
Baron, James, Michael Hannan, Greta Hsu, and 1417–1438.
Özgecan Koçak. 2007. “In the Company of Day, Jennifer C., and Eric Newburger. 2002. “The
Women: Gender Inequality and the Logic of Bu- Big Payoff: Educational Attainment and Syn-
reaucracy in Start-Up Firms.” Work and Occupa- thetic Estimates of Work-Life Earnings.” Current
tions 34: 35–66. Population Reports (P23-210): 1–13.
Belley, Philippe, and Lance Lochner. 2007. “The Ellwood, David, and Thomas Kane. 2000. “Who
Changing Role of Family Income and Ability in Is Getting a College Education? Family Back-
Determining Educational Achievement.” Journal ground and the Growing Gaps in Enrollment.”
of Human Capital 1: 37–89. In Securing the Future, edited by Sheldon Dan-
Berger, Mark. 1988. “Predicted Future Earnings and ziger and Jane Waldfogel, 283–324. New York:
Choice of College Major.” Industrial and Labor Russell Sage Foundation.
Relations Review 41: 418–429. Elvira, Marta, and Mary Graham. 2002. “Not Just
Bielby, William. 2000. “Minimizing Workplace a Formality: Pay System Formalization and
Gender and Racial Bias.” Contemporary Sociology Sex-Related Earnings Effects.” Organization Sci-
29: 120–129. ence 13 (6): 601–618.
Breen, Richard, and John Goldthorpe. 1997. “Ex- England, Paula. 1992. Comparable Worth: Theories
plaining Educational Differentials: Towards a and Evidence. New York: Walter de Gruyter.
Formal Rational Action Theory.” Rationality and Ermisch, John, Marco Francesconi, and Thomas
Society 9: 275–305. Siedler. 2006. “Intergenerational Mobility and
Breen, Richard, and Jan Jonsson. 2007. “Explain- Marital Sorting.” Economic Journal 116: 659–679.
ing Change in Social Fluidity: Educational Gerber, Theodore, and Sin Yi Cheung. 2008. “Hori-
Equalization and Educational Expansion in zontal Stratification in Postsecondary Education:
Twentieth-Century Sweden.” American Journal Forms, Explanations, and Implications.” Annual
of Sociology 112 (6): 1775–1810. Review of Sociology 34: 299–318.
Breen, Richard, and Ruud Luijkx. 2007. “Social Mo- Goyette, Kimberly, and Ann Mullen. 2006. “Who
bility and Education: A Comparative Analysis of Studies the Arts and Sciences? Social Background
Period and Cohort Trends in Britain and Ger- and the Choice and Consequences of Undergrad-
many.” In From Origin to Destination, edited by uate Field of Study.” Journal of Higher Education
Stefani Scherer, Reinhard Pollak, Gunnar Otte, 77 (3): 488–538.
and Markus Gangl, 102–124. New York: Campus. Grogger, Jeff, and Eric Eide. 1995. “Changes in Col-
Brewer, Dominic, and Ronald Ehrenberg. 1996. lege Skills and the Rise in the College Wage Pre-
“Does It Pay to Attend an Elite Private College? mium.” Journal of Human Resources 30: 280–310.
Evidence from the Senior High School Class of Grusky, David, and Kim Weeden. 2008. “Are There
1980.” Research in Labor Economics 15: 239–271. Social Classes? A Framework for Testing Sociolo-
Carnevale, Anthony, Stephen Rose, and Ban Cheah. gy’s Favorite Concept.” In Social Class: How Does
2011. “The College Payoff: Education, Occupa- It Work?, edited by Annette Lareau and Dalton
tions, and Lifetime Earnings.” Working paper, Conley, ch. 2. New York: Russell Sage.
The Georgetown University Center on Educa- Haveman, Robert, and Timothy Smeeding. 2006.
tion and the Workforce. “The Role of Higher Education in Social Mobil-
Chadwick, Laura, and Gary Solon. 2002. “Intergen- ity.” Future of Children 16 (2): 125–150.
erational Income Mobility among Daughters.” Hout, Michael. 1984. “Status, Autonomy, and
American Economic Review 92 (1): 335–344. Training in Occupational Mobility.” American
College Board. 2005. Education Pays Update. Trends Journal of Sociology 89 (6): 1397–1409.
in Higher Education Series. Washington, DC: ———. 1988. “More Universalism, Less Structural
College Board. Mobility: The American Occupational Structure
586 • Part VI: Who Gets Ahead?

in the 1980s.” American Sociological Review 93 Reskin, Barbara. 2000. “The Proximate Causes of
(6): 1358–1400. Employment Discrimination.” Contemporary So-
Jackson, Michelle. 2007. “How Far Merit Selection? ciology 29 (2): 319–328.
Social Stratification and the Labour Market.” Ross, Catherine, and John Mirowsky. 1999. “Re-
British Journal of Sociology 58 (3): 367–390. fining the Association between Education and
Kane, Thomas. 2004. “College-Going and Inequal- Health: The Effects of Quantity, Credential, and
ity.” In Social Inequality, edited by Kathryn M. Selectivity.” Demography 36: 445–460.
Neckerman, ch. 8. New York: Russell Sage. Rowley, Larry, and Sylvia Hurtado. 2003.
Karen, David. 2002. “Changes in Access to Higher “Non-Monetary Benefits of Undergraduate Edu-
Education in the United States: 1980–1992.” So- cation.” In The Public Research University: Serving
ciology of Education 75: 191–210. the Public Good in New Times, edited by Darrell
Loury, Linda, and David Garman. 1995. “College Lewis and James Hearn. Lanham, MD: Univer-
Selectivity and Earnings.” Journal of Labor Eco- sity Press of America.
nomics 13 (2): 289–308. Stevens, Mitchell, Elizabeth Armstrong, and Richard
Mare, Robert. 1980. “Social Background and School Arum. 2008. “Sieve, Incubator, Temple, Hub:
Continuation Decisions.” Journal of the American Empirical and Theoretical Advances in the So-
Statistical Association 75: 295–305. ciology of Higher Education.” Annual Review of
Monks, James. 2000. “The Returns to Individual Sociology 34: 127–151.
and College Characteristics: Evidence from the Stolzenberg, Ross. 1994. “Educational Continuation
National Longitudinal Survey of Youth.” Eco- by College Graduates.” American Journal of So-
nomics of Education Review 19: 279–289. ciology 99 (4): 1042–1077.
Mullen, Ann, Kimberly Goyette, and Joseph Soares. Thomas, Scott, and Liang Zhang. 2005.
2003. “Who Goes to Graduate School? Social “Post-baccalaureate Wage Growth within Four
and Academic Correlates of Educational Con- Years of Graduation: The Effects of College
tinuation after College.” Sociology of Education Quality and College Major.” Research in Higher
76 (2): 143–169. Education 46: 437–459.
Padavic, Irene, and Barbara Reskin. 2002. Women Tomaskovic-Devey, Donald. 1993. Gender and Ra-
and Men at Work. Thousand Oaks, CA: Pine cial Inequality at Work: The Sources and Conse-
Forge Press. quences of Job Segregation. Ithaca, NY: ILR Press.
Pallas, Aaron. 2000. “The Effects of Schooling on Torche, Florencia. 2011. “Is a College Degree Still
Individual Lives.” In Handbook of the Sociology the Great Equalizer? Intergenerational Mobility
of Education, edited by Maureen Hallinan, 499– across Levels of Schooling in the United States.”
525. New York: Kluwer Academic. American Journal of Sociology 117: 763–807.
Petersen, Trond, and Laurie Morgan. 1995. “Sep- Treiman, Donald, and Heidi Hartmann. 1981.
arate and Unequal: Occupation-Establishment Women, Work, and Wages: Equal Pay for Jobs of
Sex Segregation and the Gender Wage Gap.” Equal Value. Washington, DC: National Acad-
American Journal of Sociology 101 (2): 329–365. emy Press.
Raftery, Adrian, and Michael Hout. 1993. “Maxi- Vallet, Louis-Andre. 2004. “Change in Intergenera-
mally Maintained Inequality: Expansion, Reform tional Class Mobility in France from the 1970s
and Opportunity in Irish Education, 1921– to the 1990s and Its Explanation: An Analysis
1975.” Sociology of Education 66 (1): 41–62. Following the CASMIN Approach.” In Social
Mobility in Europe, edited by Richard Breen,
115–147. Oxford: Oxford University Press.
 Brand and Xie—Who Benekts Most from College? • 587

66 • Jennie E. Brand and Yu Xie

Who Benekts Most from College?

Educational expansion is one of the most ap- 1979). We call this thesis the positive selection
parent, enduring, and consequential features hypothesis.
of modern society. Concerning the significance The sociological literature usually treats the
of educational expansion in the United States two research questions posed above separately,
during the twentieth century, Fischer and Hout because of the recognition that college-going
(2006, 247) conclude that “the division between behavior is governed not only by economically
the less- and more-educated grew and emerged rational choice but also by cultural and social
as a powerful determiner of life chances and life- norms. As such, mechanisms influencing college
styles.” Social scientists have long been interested attainment may differ by social background.
in access to and the impact of higher education. For some individuals from socially advantaged
Two questions commonly posed are: What fam- backgrounds, college is a culturally expected out-
ily and individual attributes are associated with come and thus less exclusively and intentionally
the attainment of higher education? and What linked to economic gain than for those in less
are the causal effects of higher education on sub- advantaged groups, for whom a college educa-
sequent socioeconomic outcomes? tion is a novelty that may well demand economic
In the rational-behavioral model, common justification (Boudon 1974; Smith and Powell
in the economics literature, these questions 1990). In addition, as less educated workers
are intrinsically intertwined: individuals decide from disadvantaged backgrounds have partic-
whether to pursue higher education on the basis ularly bleak earnings prospects, college should
of cost-benefit analyses. People choose higher entail a significant economic return. Once we
education only if it will increase their lifetime parcel out observed covariates that help predict
earnings (Becker 1964; Mincer 1974; Willis and college education, it is possible that, because of
Rosen 1979). In other words, barring imperfect differential selection mechanisms and earnings
information, constraints on obtaining funding, prospects, individuals least likely to obtain a col-
or uncertainty, people pursue a college educa- lege education benefit most from college. We call
tion only if the economic returns outweigh the this conjecture the negative selection hypothesis.
costs. If economics is the main determinant, it To adjudicate between the positive and neg-
follows that individuals most likely to attend col- ative selection hypotheses, we conduct an em-
lege would also benefit most from it (Carneiro, pirical study analyzing data from the National
Heckman, and Vytlacil 2007; Willis and Rosen Longitudinal Survey of Youth 1979 cohort (see

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in the
following publication: Jennie E. Brand and Yu Xie, “Who Benefits Most from College? Evidence for Negative
Selection in Heterogeneous Economic Returns to Higher Education,” American Sociological Review 75 (April
2010), pp. 273–302, published by Sage Publications. The article printed here was originally prepared by Jen-
nie E. Brand and Yu Xie for this fourth edition of Social Stratification: Class, Race, and Gender in Sociological
Perspective, edited by David B. Grusky. Copyright © 2014 by Westview Press.
588 • Part VI: Who Gets Ahead?

Brand and Xie 2010 for parallel analyses of the outcomes are uncorrelated with treatment status,
Wisconsin Longitudinal Study 1957 Cohort). conditional on observed covariates; the plausibil-
First, we invoke an ignorability assumption to ity of the assumption depends on the richness of
summarize in estimated propensity scores sys- observed covariates. This approach allows us to
tematic differences in covariates between college find empirical patterns of treatment-effect het-
graduates and non–college graduates. Second, erogeneity as a function of observed covariates.
we estimate the effects of college completion on A common way of studying heterogeneous treat-
earnings by propensity score strata and examine ment effects by observed covariates is to examine
patterns of effects across these strata using a hi- the interaction between education and specific
erarchical linear model. factors that influence wages and the probability
of obtaining a college education, such as race
or gender or parents’ education. When com-
Theoretical and Methodological paring returns to college between individuals
Issues in Assessing College Returns who complete college and those who do not,
Two Sources of Selection Bias however, the most meaningful interaction is be-
As is well documented in the causal inference tween college education and the propensity to
literature, but seldom acknowledged in empirical complete college (Heckman, Urzua, and Vytlacil
sociological research, observational data contain 2006; Xie, Brand, and Jann 2012). We aggregate
two types of selection bias (Morgan and Win- heterogeneous college effects to propensity score
ship 2007; Xie, Brand, and Jann 2012). The first group-level mean effects and directly observe
type is due to heterogeneity in preexisting condi- trends in effects. Using our strategy, we focus on
tions, or attributes that are associated with both group differences in the propensity to complete
treatment condition and outcome. In the case of college and adjudicate between two potential
economic returns to higher education, attributes patterns in observed heterogeneous effects of
such as mental ability and work habits may be college completion on earnings: positive selec-
positively associated with the likelihood of ob- tion (individuals who benefit most from college
taining higher education and higher earnings. are most likely to complete college) versus nega-
The second type of selection bias is due to het- tive selection (individuals who benefit most from
erogeneity in treatment effects, that is, system- college are least likely to complete college).
atic differences in the causal effect of college on
earnings between individuals who do and do not
obtain a college education. Although it is im- Positive versus Negative Selection
possible to empirically identify individual-level In economics, human capital theory is an influ-
heterogeneity in economic returns to higher ed- ential explanation for educational acquisition
ucation, which we know exists, we can focus on (Becker 1964; Mincer 1974). The core idea of
group-level variability in the returns by aggregat- the theory is that a gradation in earnings by ed-
ing individuals according to their estimated like- ucation level reflects returns to individuals’ ratio-
lihood of completing college. Based on observed nal investment in education. If Q represents the
attributes, we ask whether individuals who are present value of the lifetime economic returns to
more likely to obtain a college education receive college, and c the cost of college, attending col-
higher or lower returns to college education than lege produces a net gain if Q > c, with the benefit
those who are less likely to do so. thus defined as U = Q–c. The association between
To systematically study heterogeneous treat- the returns to college and the decision to attend
ment effects of higher education on earnings, we college is at the core of more recent literature
adopt a simple approach using rich covariates that links variation in returns to education to
and invoking ignorability, at least provisionally. heterogeneous schooling behavior. Premised
The ignorability assumption states that potential on principles of self-selection and comparative
 Brand and Xie—Who Benekts Most from College? • 589

advantage, the thesis is that the most “college- in destination between college-educated and
worthy” individuals (those having the highest re- less-educated workers) as a function of social
turns to college) are the most likely to select into origin, this interaction pattern yields a smaller
college (Carneiro, Heckman, and Vytlacil 2007; difference by college education for individuals
Roy 1951; Willis and Rosen 1979). According of high social origin (į2) than for individuals of
to this literature, positive selection should occur. low social origin (į1). In other words, individ-
Like economists, sociologists infer that uals from relatively disadvantaged social back-
the choice to attend college can result from a grounds, or those with the lowest probability
cost-benefit analysis (Boudon 1974; Breen and of completing college, benefit the most from
Goldthorpe 1997); sociologists, however, em- completing college. This pattern results from
phasize that costs and benefits are not purely the particularly poor labor market prospects for
economic. For example, sociologists have con- workers with low levels of education combined
sidered heterogeneity in terms of both the with low levels of other forms of human, social,
financial burden and the family pressure asso- or cultural capital. This collective theoretical and
ciated with deviating from class-based cultural empirical tradition leads to our negative selec-
norms. Indeed, in contrast to the strictly eco- tion hypothesis.
nomic cost-benefit model of college attendance,
much research indicates that multiple actors and
factors influence college attendance. Beginning Behavioral Model
with the Blau-Duncan model (Blau and Duncan We specify the behavioral model for college ed-
1967), sociologists have recognized the signifi- ucation as follows: let d * represent the potential
cance of numerous family background factors likelihood that the ith person completes college,
for educational attainment, such as parents’ edu- and di the observed outcome (1 if yes; 0 other-
cation and occupation, family structure, sibship wise). We further specify that college attainment
size, occupational and educational aspirations, is determined by a weighted average of an eco-
encouragement from parents and significant oth- nomic component Ui, a noneconomic compo-
ers, social and cultural capital, and school-level nent Mi, and a residual Ji:
factors (e.g., Bourdieu 1977; Bowles and Gintis
1976; Coleman 1988; Lareau 2003; MacLeod di* = wiUi + (1–wi)Mi + Ji, (1)
1989; Morgan 2005; Sewell, Haller, and Ohlen-
dorf 1970). In summary, this literature suggests where Ji is assumed to be independent of Ui, Mi,
that individuals with a high social background and wi, with 0 ≤ wi ≤ 1. A key insight from the
are likely to go to college even in the absence of a sociological literature is that the relative weight
rational economic cost-benefit analysis, whereas wi given to the economic component may de-
those from a low social background must over- crease with the noneconomic determinant Mi
come considerable odds to attend college. (i.e., a negative correlation between the two in
Past research in social stratification provides the population). We further assume Ui to be a
a compelling theoretical and empirical basis for linear function of observed covariates (Q’X ) plus
postulating variation in the effects of education an unobserved component Ri, and Mi to be a
on earnings by social background. This research linear function of observed covariates (Q’X ). We
shows that the direct relationship between so- can rewrite Equation 1 as follows:
cial origin and destination (both measured by
occupational status) is much weaker for college di* = wiQ1’Xi + (1–wi) Q2’Xi + wiRi + Ji. (2)
graduates than for workers without college de-
grees (Hout 1988). Figure 66.1 depicts this em- The likelihood of completing college is high
pirical pattern. If we change the perspective and when di* is large. Writing out the model makes
examine returns to schooling (i.e., the difference it easier to appreciate the key difference between
590 • Part VI: Who Gets Ahead?

Figure 66.1. Hypothetical Model: Origin, Education, and Destination

Occupational
Destination College-educated
workers
Benefit of a
college degree
į2

Less-educated
workers

Benefit of a
college degree
į1

Social Origin

the economic and noneconomic factors influ- the sample. We group respondents into strata
encing college attainment. In the traditional eco- of estimated propensity scores to balance the
nomic model, wi = 1, and Ri drives the college distributions of the covariates between college
education decision, conditional on X. In most graduates and non–college graduates (p < .001).
sociological literature, familial, personal, and Second, in level 1 we estimate the treatment ef-
institutional characteristics dominate (i.e., wi is fects specific to balanced propensity score strata
much smaller than 1), so that the observed co- using ordinary least squares regression. Third, in
variates X primarily determine the decision rule, level 2 we examine the heterogeneous results by
with the self-selection component given the sec- propensity score strata and summarize the trend
ondary role or sometimes ignored (i.e., wi Ri = 0). in the variation of effects using a hierarchical
The sociological literature suggests that be- linear model. More methodological details are
cause w should be negatively correlated with the provided in Xie, Brand, and Jann (2012).
observed propensity score, the extent of misspec-
ification caused by omitting R when assuming
ignorability declines with the observed propen- Data, Measures, and Descriptive
sity score; that is, the decision to go to college Statistics
among children from high-status families is To examine heterogeneous treatment effects of
dictated less by rational choice and self-selection education on earnings, we use data from the
than it is among children from low-status fami- National Longitudinal Survey of Youth 1979
lies. When a person who is not expected to go to (NLSY). Single-cohort longitudinal surveys are
college based on observed characteristics does go advantageous in controlling for the potential
to college, there are strong factors involved, one confounding effect of cohort with experience.
of which may be the economic incentive. The NLSY is a nationally representative sam-
ple of 12,686 respondents who were fourteen
to twenty-two years old when first surveyed in
Statistical Models for 1979. These individuals were interviewed an-
Heterogeneous Treatment Effects nually through 1994 and biennially thereafter.
Our analytical strategy proceeds in three steps. We restrict our sample to respondents who were
First, we estimate binary logistic regressions pre- fourteen to seventeen years old at the time of
dicting the probability of completing college and the baseline survey in 1979 (N = 5,581), had
derive propensity scores for each individual in not graduated from high school at the time the
 Brand and Xie—Who Benekts Most from College? • 591

Armed Services Vocational Aptitude Battery a highly significant positive effect on logged
(ASVAB) tests were administered (N = 3,885), hourly wages that steadily increases over time,
had completed at least the twelfth grade by 1990 from a 20 percent advantage in men’s late twen-
(N = 3,034), and do not have any missing data ties to early thirties to a 51 percent advantage
on the set of covariates used in our analysis (N = in their late thirties to early forties. Results for
2,474). We evaluate effects of completing college NLSY employed women reveal a large, signif-
on earnings for respondents at ages twenty-nine icant effect of college completion in their late
to thirty-two (in 1994), thirty-three to thirty-six twenties to early thirties, a smaller effect in their
(in 1998), and thirty-seven to forty (in 2002), mid-thirties than in their early thirties, and then
that is, from their early- to midcareer years. a comparatively larger effect in their late thirties
To construct propensity score strata, we use to early forties. Differences in life course pat-
the following variables: race, parents’ income, terns between men and women may reflect the
mother’s and father’s education, family struc- influences of traditional gender roles in the fam-
ture, sibship size, rural/urban residence, religion, ily and corresponding intermittent labor force
ability (ASVAB), college-prep program, and attachment among women compared to men,
friends’ plans for college. We use hourly wages particularly during childbearing years.
as the outcome variable in the logarithm form
for respondents’ earnings in their late twenties
through early forties (in 1994, 1998, and 2002). Generating Propensity Score Strata
We add a small positive constant ($.50) before Our next objective is to examine the heteroge-
taking the logs. Unemployed workers are elim- neous effects of college completion by propen-
inated. Brand and Xie (2010) provide full de- sity score strata. We estimate binary logistic
tails on variable construction and descriptive regressions predicting the odds of completing
statistics. In general, college graduates, relative college by the covariates described previously,
to individuals who did not graduate from col- separately by sex, and derive estimated propen-
lege, are more likely to be white and Asian than sity scores for each individual. We then gener-
black or Hispanic and come from families with a ate balanced propensity score strata; balance is
high income, highly educated parents, an intact satisfied when within each interval of the pro-
family structure, and few siblings. High levels pensity score the average propensity score and
of secondary school academic success, cogni- the means of each covariate do not significantly
tive ability, and encouragement from teachers
and parents to attend college, as well as friends Table 66.1. Effects of College Completion on Log
who plan to attend college, are also predictive Wages under the Assumption of Homogeneity
of college education. These statistics suggest that
NLSY Men Women
many noneconomic factors figure prominently
1994 Wages .180*** .276***
in youths’ educational attainment.
(ages 29 to 32) (.047) (.051)
1998 Wages .296*** .188***
Main Analysis and Findings (ages 33 to 36) (.054) (.052)
2002 Wages .410*** .216**
College Returns under the Assumption (ages 37 to 40) (.069) (.075)
of Homogeneity
Table 66.1 shows the estimated effects of col- Note: Numbers in parentheses are standard errors.
lege completion on earnings, separately by sex, Treatment effects are conditional on the set of covariates.
through regression analyses under the homo- Estimates further condition on age at baseline, and for
women also condition on an indicator for married with
genous effect assumption, controlling for the children at age 25. All outcome variables are current
full set of covariates described previously. For hourly wages. Unemployed workers are omitted.
NLSY employed men, college completion yields *p < .05; **p < .01; ***p < .001 (two-tailed tests).
 592 • Part VI: Who Gets Ahead?

Table 66.2. Mean Covariate Values by Propensity Score Strata: NLSY Men
PROPENSITY SCORE STRATA
(.0–.1) (.1–.2) (.2–.4) (.4–.6) (.6–1.0)
Non- Non- Non- Non- Non-
Coll. Coll. Coll. Coll. Coll. Coll. Coll. Coll. Coll. Coll.
Variables Grad. Grad. Grad. Grad. Grad. Grad. Grad. Grad. Grad. Grad.
Black .37 .25 .21 .40 .23 .23 .19 .12 .15 .07
Hispanic .18 .30 .12 .08 .12 .09 .12 .09 .07 .05
Parents’ income 13381 12253 17614 18482 19324 18422 23062 21348 23469 34702
Mother’s
10.31 10.05 11.67 12.16 11.98 12.21 12.71 12.54 13.67 14.79
education
Father’s
10.17 9.95 11.79 10.72 12.08 12.53 13.33 13.97 15.11 16.30
education
Intact family .63 .55 .63 .80 .74 .67 .85 .80 .85 .91
No. siblings 3.84 4.05 3.04 3.04 2.64 2.47 2.88 2.46 2.04 2.17
Rural resident .21 .30 .26 .20 .21 .21 .19 .12 .11 .20
Avail. of college .76 .70 .80 .84 .75 .77 .73 .77 .81 .78
Jewish .00 .00 .00 .00 .01 .00 .00 .02 .04 .08
Mental ability –.14 –.01 .31 .48 .62 .57 .79 .76 .90 1.05
College track .17 .16 .32 .37 .41 .52 .57 .55 .83 .73
Friends’ plans .35 .55 .61 .52 .66 .74 .90 .85 .93 .93

differ between college graduates and non–college Points in the figures represent estimates of
graduates. We restrict the balancing algorithm stratum-specific effects of college completion
to the region of common support, that is, to re- on logged earnings. The linear plots and re-
gions of propensity scores in which both treated ported level-2 slopes in the figures are based
and control units are observed. In the case of on the hierarchical linear models (i.e., level-2
college-educated individuals, the frequency variance-weighted least squares models estimated
count increases with the propensity score, by level-1 college effects specific to propensity
whereas for non-college-educated individuals, score strata regressed on propensity stratum
the count decreases with the propensity score. To rank). More details can be found in Brand and
demonstrate the balance achieved within each Xie (2010).
stratum, we present covariate means by propen- Figure 66.2 depicts results for college ef-
sity score strata for NLSY men in Table 66.2, fects on men’s earnings at ages twenty-nine to
which also elucidates the characteristics of a typi- thirty-two (in 1994), thirty-three to thirty-six
cal individual within each stratum. For example, (in 1998), and thirty-seven to forty (in 2002).
a characteristic person in stratum 1 has parents The downward linear slopes illustrate the de-
who are high school dropouts, three siblings, clining trend in effects with propensity stratum
low ability, and friends who do not plan to go to rank at every observed time period. For example,
college, and is enrolled in a nonacademic track. for men in their late twenties to early thirties, a
By contrast, a characteristic person in stratum 5 unit change in stratum rank is associated with a
has parents with some college, one sibling, high 5 percent reduction in the treatment effect, such
ability, friends who plan to go to college, and is that the predicted effect of college completion
enrolled in an academic track. on earnings in stratum 1 is about 30 percent,
whereas the predicted effect in stratum 5 is
about 10 percent. This means, for example, that
Heterogeneous College Returns an individual with parents who are high school
Figures 66.2 and 66.3 present the main results dropouts, and who himself has low measured
of our study, for men and women respectively. ability, benefits more from completing college,
 Brand and Xie—Who Benekts Most from College? • 593

Figure 66.2. HLM of Economic Returns to College, NLSY Men

on the magnitude of an estimated 20 percent, however, we observe an oscillating return to col-

than would an individual whose parents went lege over the life course among women, possibly
to college and who himself has high measured reflecting women’s intermittent labor force at-
ability. We find similar patterns for men in their tachment during childbearing years; these fam-
mid- and late thirties and early forties. In the late ily processes could affect women differently by
thirties to early forties, the 5 percent reduction propensity score strata.
in treatment effect per stratum rank again results As there are few statistically significant level-2
in an estimated 20 percent difference between slope coefficients, our results should be taken
the lowest and highest strata, or between the as descriptive and suggestive, not definitive. In
least- and most-advantaged college goers. The Brand and Xie (2010) we revisit the ignorability
level-2 slopes thus offer support for the negative assumption and conduct auxiliary analyses that
selection hypothesis at each observed stage of the facilitate interpretation of the results.
life course. As expected, college completion is
associated with increasing economic return over
the life course, and this is true across propensity Discussion and Conclusions
score strata. Heterogeneity in response to a common treat-
The results for NLSY women, shown in Fig- ment is a norm, not an exception. Individuals
ure 66.3, are similar to those for men in suggest- differ not only in background attributes, but also
ing negative selection at each observed stage of in how they respond to a particular treatment.
the life course. For example, for women in their An important task of sociological research is to
late thirties to early forties, a unit change in stra- summarize systematic patterns in population
tum rank is associated with a 4 percent reduction variability. In this chapter we consider popu-
in the treatment effect. The predicted effect of lation heterogeneity in returns to schooling,
college completion on earnings is about 40 per- examining the effects of completing college by
cent for stratum 1 women with disadvantaged propensity score strata in a hierarchical linear
socioeconomic backgrounds, versus about 25 model. Our evidence suggests negative selection:
percent for stratum 5 women with advantaged individuals who benefit most from a college edu-
socioeconomic backgrounds. In contrast to men, cation are those who are the least likely to obtain
594 • Part VI: Who Gets Ahead?

Figure 66.3. HLM of Economic Returns to College, NLSY Women

one (i.e., individuals from the most disadvan- closely. Whatever the underlying reason, how-
taged social origins and with the lowest ability ever, we have found empirical evidence that in-
and achievement). This finding holds for both dividuals who are less likely to obtain a college
men and women and for every observed stage education benefit more from college.
over the life course. The widespread belief in the socioeconomic
A principal reason for low-propensity returns to higher education has prompted pol-
college-educated workers’ relatively large eco- icy efforts that expand educational opportu-
nomic return is that their social position is nities for all Americans. Though many policy
marked by substantial disadvantage. In the ab- makers implicitly assume homogeneity in the
sence of a college degree, low-propensity men returns to schooling, potential heterogeneity
and women have limited human, cultural, and in returns is receiving more attention as many
social capital and hence particularly limited countries are experiencing rapid expansion in
labor market prospects. By contrast, in the ab- college enrollment. This has led some to ques-
sence of a college degree, individuals from more tion the relative costs and benefits of higher
advantaged social backgrounds can still rely on education for those who were not previously re-
their superior resources and abilities. The nega- ceiving it. Yet in the presence of heterogeneous
tive selection pattern does not emerge because treatment effects, no simple summary statement
low-propensity college goers earn more wages can be made about the benefit of completing
than do high-propensity college goers; they college, either for individuals already receiving
do not. Rather, the pattern emerges because higher education or for those likely to benefit
low-propensity non–college goers earn so little. from educational expansion. The average ben-
Alternatively, we may interpret the observed efit depends on the composition at any given
pattern of negative selection as being caused by time of the group of students who complete
differential selectivity, with persons of low pro- college. One interpretation of our results is that
pensity to complete college more selective than a college education may be particularly benefi-
persons of high propensity. Our data do not cial for those in groups targeted by educational
allow us to examine this interpretation more expansion efforts—that is, individuals who are
 Brand and Xie—Who Benekts Most from College? • 595

otherwise unlikely to attend college based on Heckman, James J., Sergio Urzua, and Edward Vyt-
their observed characteristics. lacil. 2006. “Understanding Instrumental Vari-
ables in Models with Essential Heterogeneity.”
REFERENCES Review of Economics and Statistics 88: 389–432.
Becker, Gary S. 1964. Human Capital: A Theoretical Hout, Michael. 1988. “More Universalism, Less
and Empirical Analysis, with Special Reference to Structural Mobility: The American Occupational
Education. New York: Columbia University Press. Structure in the 1980s.” American Journal of So-
Blau, Peter, and Otis Dudley Duncan. 1967. The ciology 93: 1358–1400.
American Occupational Structure. New York: John Lareau, Annette. 2003. Unequal Childhoods. Prince-
Wiley & Sons. ton, NJ: Princeton University Press.
Boudon, Raymond. 1974. Education, Opportunity MacLeod, Jay. 1989. Ain’t No Makin’ It: Aspirations
and Social Inequality. New York: John Wiley & and Attainment in a Low-Income Neighborhood.
Sons. Boulder, CO: Westview Press.
Bourdieu, Pierre. 1977. “Cultural Reproduction and Mincer, Jacob. 1974. Schooling, Experience, and
Social Reproduction.” In Power and Ideology in Earnings. New York: Columbia University Press.
Education, edited by J. Karabel and A. H. Halsey, Morgan, Stephen. 2005. On the Edge of Commit-
487–510. New York: Oxford University Press. ment: Educational Attainment and Race in the
Bowles, Samuel, and Herbert Gintis. 1976. Schooling United States. Stanford, CA: Stanford University
in Capitalist America: Educational Reform and the Press.
Contradictions of Economic Life. New York: Basic Morgan, Stephen, and Christopher Winship. 2007.
Books. Counterfactuals and Causal Inference: Methods and
Brand, Jennie E., and Yu Xie. 2010. “Who Benefits Principles for Social Research. Cambridge, UK:
Most from College? Evidence for Negative Se- Cambridge University Press.
lection in Heterogeneous Economic Returns to Roy, Andrew D. 1951. “Some Thoughts on the Dis-
Higher Education.” American Sociological Review tribution of Earnings.” Oxford Economic Paper 3:
75 (2): 273–302. 135–146.
Breen, Richard, and John H. Goldthorpe. 1997. Sewell, William, Archibald O. Haller, and G. W.
“Explaining Educational Differentials: Towards a Ohlendorf. 1970. “Educational and Early Occu-
Formal Rational Action Theory.” Rationality and pational Status Attainment Process: Replication
Society 9: 275–305. and Revision.” American Sociological Review 35:
Carneiro, Pedro, James J. Heckman, and Edward 1014–1027.
Vytlacil. 2007. “Estimating Marginal and Aver- Smith, Herbert L., and Brian Powell. 1990. “Great
age Returns to Education.” Unpublished man- Expectations: Variations in Income Expectations
uscript, Department of Economics, University among College Seniors.” Sociology of Education
College London. 63: 194–207.
Coleman, James A. 1988. “Social Capital and the Willis, Robert J., and Sherwin Rosen. 1979. “Ed-
Creation of Human Capital.” American Journal ucation and Self-Selection.” Journal of Political
of Sociology 94: S95–S120. Economy 87: S7–S36.
Fischer, Claude S., and Michael Hout. 2006. Cen- Xie, Yu, Jennie Brand, and Ben Jann. 2012. “Esti-
tury of Difference: How America Changed in the mating Heterogeneous Treatment Effects with
Last One Hundred Years. New York: Russell Sage Observational Data.” Sociological Methodology
Foundation. 42: 1–34.
596 • Part VI: Who Gets Ahead?

Social Psychological Models

67 • William H. Sewell, Archibald O. Haller, and Alejandro Portes

The Educational and Early Occupational

Attainment Process

Blau and Duncan (1967:165–172) have recently single out variables indicating father’s stratifica-
presented a path model of the occupational at- tion position as the most relevant social structural
tainment process of the American adult male inputs. It is unfortunate that practical consider-
population. This basic model begins with two ations prevented the inclusion of psychological
variables describing the early stratification posi- inputs in their model, especially considering the
tion of each person; these are his father’s edu- repeated references to one such—mental abil-
cational and occupational attainment statuses. ity—in the literature on differential occupational
It then moves to two behavioral variables; these attainment (Lipset and Bendix, 1959:203–226;
are the educational level the individual has com- Sewell and Armer, 1966). More recently, this gap
pleted and the prestige level of his first job. The has been partially filled (Duncan, 1968).
dependent variable is the person’s occupational 2) Also omitted are social psychological fac-
prestige position in 1962. That the model is not tors which mediate the influence of the input
without power is attested by the fact that it ac- variables on attainment. This, too, is unfortu-
counts for about 26 percent of the variance in nate in view not only of the speculative theory
educational attainment, 33 percent of the vari- but also the concrete research in social psychol-
ance in first job, and 42 percent of the variance ogy, which suggests the importance of such
in 1962 level of occupational attainment. Vari- intervening variables as reference groups (Mer-
ous additions to the basic model are presented in ton, 1957:281–386), significant others (Gerth
the volume, but none is clearly shown to make and Mills, 1953:84–91), self-concept (Super,
much of an improvement in it. These include 1957:80–100), behavior expectations (Gross et
nativity, migration, farm origin, subgroup po- al., 1958), levels of educational and occupational
sition, marriage, and assortative mating. With- aspiration (Haller and Miller, 1963; Kuvlesky
out detracting from the excellence of the Blau and Ohlendorf, 1967; Ohlendorf et al., 1967),
and Duncan analysis, we may make several and experiences of success or failure in school
observations. (Parsons, 1959; Brookover et al., 1965).
1) Because the dependent behaviors are occu- It remains to be seen whether the addition of
pational prestige attainments—attainment levels such psychological and social psychological vari-
in a stratification system—it is appropriate to ables is worthwhile, although there are reasons

William H. Sewell, Archibald O. Haller, and Alejandro Portes. “The Educational and Early Occupational
Attainment Process,” American Sociological Review 34:1 (February 1969), pp. 82–92. Used by permission of
the American Sociological Association and the authors.
 Sewell, Haller, and Portes—The Educational and Early Occupational Attainment Process • 597

for believing that at least some of them may be. certain social structural and psychological fac-
First, an explanation of a behavior system requires tors—initial stratification position and mental
a plausible causal argument, not just a set of path ability, specifically—affect both the sets of sig-
coefficients among temporally ordered variables. nificant others’ influences bearing on the youth,
As indicated in Duncan’s (1969) recent work, and the youth’s own observations of his ability;
the introduction of social psychological mediat- (2) that the influence of significant others, and
ing variables offers this possibility, but it does possibly his estimates of his ability, affect the
not guarantee it. As it stands, the Blau-Duncan youth’s levels of educational and occupational
model fails to indicate why any connection at aspiration; (3) that the levels of aspiration affect
all would be expected between the input vari- subsequent levels of educational attainment; (4)
ables, father’s education and occupation, and the that education in turn affects levels of occupa-
three subsequent factors: respondent’s education, tional attainment. In the present analysis we
respondent’s first job, and respondent’s 1962 assume that all effects are linear; also, that the
occupation. Granting differences among social social psychological variables perform only me-
psychological positions, they all agree that one’s diating functions.
cognitions and motivations (including, among More specifically, we present theory and data
others, knowledge, self-concept, and aspirations) regarding what we believe to be a logically con-
are developed in structured situations (including sistent social psychological model. This provides
the expectations of others), and that one’s actions a plausible causal argument to link stratification
(attainments in this case) are a result of the cog- and mental ability inputs through a set of so-
nitive and motivational orientations one brings cial psychological and behavioral mechanisms
to the action situation, as well as the factors in to educational and occupational attainments.
the new situation itself. Second, if valid, a social One compelling feature of the model is that
psychological model will suggest new points at some of the inputs may be manipulated through
which the causal system may be entered in order experimental or other purposive interventions.
to change the attainment behaviors of persons, This means that parts of it can be experimen-
an issue not addressed by the Blau and Duncan tally tested in future research and that practical
volume. Variables such as the expectations of sig- policy agents can reasonably hope to use it in
nificant others offer other possibilities for ma- order to change educational and occupational
nipulating the outcomes, including educational attainments.
attainments. Third, in addition to the above
advantages, a social psychological model of edu-
cational and occupational attainment might add A Social Psychological Model
to the explanation of variance in the dependent The model treats causal relationships among
variables. eight variables. X1 is the occupational prestige
level attained by the adult person, or occupa-
tional attainment (OccAtt); X2 is the educational
The Problem level he had previously attained, or educational
The present report extends the attempts of attainment (EdAtt); X3 is the occupational pres-
the writers (Sewell and Armer, 1966; Sewell tige level to which he aspired as a youth, or level
and Orenstein, 1965; Sewell and Shah, 1967; of occupational aspiration (LOA); X4 is his level
Sewell, 1964; Haller and Sewell, 1967; Portes et of educational aspiration as a youth (LEA); X5
al., 1968; Haller, 1966; Haller and Miller, 1963; is the influence for educational achievement
Miller and Haller, 1964; Sewell et al., 1957) to exerted upon him by significant others while
apply social psychological concepts to the expla- still in high school, or significant others’ influence
nation of variation in levels of educational and (SOI); X6 is the quality of his academic perfor-
occupational attainment. We assume (1) that mance in high school (AP); X7 is the level of his
598 • Part VI: Who Gets Ahead?

Figure 67.1. Path Coefficients of Antecedents of Educational and Occupational Attainment Levels

family in the stratification system, or socioeco- variables which cannot be assigned causal prior-
nomic status (SES); and X8 is his mental ability ity in present data.
as measured while he was in high school (MA). Commencing from the left of the diagram,
Path models (Blau and Duncan, 1967:165–172; we assume, as has often been found before
Wright, 1934; Wright, 1960; Heise, 1969) re- (Sewell and Shah, 1967; Sewell et al., 1957),
quire a knowledge of the causal order among that a low positive correlation, r78, exists between
the variables. Beyond the causal arguments pre- the youth’s measured mental ability (MA) and
sented below, additional credibility is suggested his parents’ socioeconomic status (SES). This is
by the existence of a plausible temporal order the case: r78 = .21. We anticipate the existence
among variables. X7 (SES) and X8 (MA) precede of substantial effect of MA on academic perfor-
everything else. X5 (SOI) and X6 (AP) precede mance (AP). We theorize that significant oth-
both aspirations and attainments, and it can be ers’ influence (SOI) is controlled by AP, and by
assumed that for the most part X6 precedes X5. socioeconomic status, as well as by exogenous
Youthful aspirations obviously precede later edu- factors, that they exert profound effects on aspi-
cational and occupational attainments. Pre-adult ration, and that the latter in turn influences later
educational attainments precede adult occupa- attainments. A more detailed examination of the
tional attainments. theory follows.
By no means do all of the possible causal Working with partial conceptions of SOI
linkages seem defensible. The most likely ones (and using different terminology), Bordua
are indicated in Figure 67.1. In it straight solid (1960) and Sewell and Shah (1968) have shown
lines stand for causal lines that are to be theoreti- that parents’ expectations for the youths’ attain-
cally expected, dotted lines stand for possible but ments are important influences on later aspira-
theoretically debatable causal lines, and curved tions and attainment. Similarly, Cramer (1967),
lines represent unanalyzed correlations among Alexander and Campbell (1964), Campbell
 Sewell, Haller, and Portes—The Educational and Early Occupational Attainment Process • 599

and Alexander (1965), Haller and Butterworth undercurrent in the literature seems to have
(1960), and Duncan et al. (1968) have investi- held, however, that the youth’s family’s SES has
gated peer influences on aspirations and attain- a direct influence on his grades (Havighurst and
ments. Each of these sets of actors, plus some Neugarten, 1957:236–237). To our knowledge,
others, may be seen as a special case of reference this has not been adequately demonstrated, and
group influence. Building on such thinking, we in large high schools, often far removed from
have concluded that the key variable here is sig- the youth’s home and neighborhood, this may
nificant others’ influence. Significant others are well be debatable. Nevertheless, since it is at least
the specific persons from whom the individual possible that school grades (the evidences of per-
obtains his level of aspiration, either because formance) are partly determined by teachers’ de-
they serve as models or because they commu- sires to please prestigious parents or to reward
nicate to him their expectations for his behavior “middle-class” behavior, we have drawn a dotted
(Woelfel, 1967). The term “significant others” is path (p67) from SES to AP, allowing for the pos-
more appropriate than that of “reference group” sibility of such an influence.
because it eliminates the implication that collec- We hypothesize that the major effects of sig-
tivities such as one’s friends, or work groups, or nificant others’ influence (SOI) on attainment
parents are necessarily the influential agents for are mediated by its effects on levels of aspiration.
all individuals. Experimental research, beginning Thus, we have indicated a path (p35) from SOI
with Sherif ’s work (1935), has shown the im- to level of occupational aspiration (LOA) and
portance of other persons in defining one’s own another (p45) from SOI to level of educational
situation. One obtains his social behavior ten- aspiration (LEA). It is not inconsistent with this
dencies largely through the influence of others. to suspect the possibility that SOI might have a
Herriott (1963) has carried this line of thinking direct influence on later educational attainment
into the present area of research. He has shown (EdAtt); we have thus included a dotted or de-
that one’s conception of the educational behav- batable path (p25) from SOI to EdAtt. Because
ior others think appropriate to him is highly cor- we are here referring to SOI during late high
related with his level of educational aspiration. school, it must necessarily refer largely to col-
Thus, significant others’ influence is a central lege education. There is, therefore, no reason to
variable in a social psychological explanation of include such a path from SOI to occupational
educational and occupational attainment. It is attainment.
obviously important to discover the causal paths Levels of educational aspiration (LEA) and
determining SOI, as well as those by which it occupational aspiration (LOA) are known to be
exerts its effects on attainment. We hypothesize a highly correlated, since education is widely, and
substantial direct path (p57) from socioeconomic to some extent validly, considered to be a neces-
status (SES) to SOI. We also hypothesize a sub- sary condition for high occupational attainment
stantial effect of mental ability on SOI. This is (Haller and Miller, 1963:30, 39–42, 96). But
because we expect that the significant others LOA and LEA are not identical. (In these data,
with whom the youth interacts base their expec- r34.5 = rWX = .56.) We expect that LEA will have
tations for his educational and occupational at- a pronounced effect on EdAtt (p24), and that
tainments in part on his demonstrated abilities. its entire effect on level of occupational attain-
In turn, this implies that the path from mental ment will be expressed through EdAtt. On the
ability (MA) to SOI is indirect by way of aca- other hand, we do not hypothesize any effect of
demic performance (AP). Thus, we hypothesize LOA on EdAtt which is not already contained
the existence of a pronounced path from MA in its correlation with LEA. Hence, there is no
to AP (p68) and another from AP to SOI (p56). hypothetical path for LOA to EdAtt. A direct
So far we assume that one’s grades in school are effect of LOA on OccAtt (p13) is hypothesized,
based on the quality of his performance. A strong however.
600 • Part VI: Who Gets Ahead?

There are 26 possible paths, given the se-

quence laid out above. As one can see by count- Variables
ing the paths (straight lines) in Figure 67.1, we Level of occupational attainment (X1—OccAtt)
hypothesize noteworthy effects for only eight of was measured by Duncan’s (1961) socioeco-
these—ten if the dotted lines are counted. If this nomic index of occupational status.
were a rigorous theoretical model, path coeffi- Level of educational attainment (X2—EdAtt)
cients would be calculated only for these eight was operationalized with data obtained in 1964
(or ten) supposed causal connections. We believe by dividing the sample into those who have had
that because of the fact that it is not rigorous, at least some college education and those who
and at this stage of our knowledge probably can- have not had any at all.1
not be, it would be well to calculate all of the Level of occupational aspiration (X3—LOA)
possible 26 path coefficients, using the calculated was determined by assigning Duncan’s (1961)
values as rough indicators of the influences op- socioeconomic index scores to the occupation
erating in the system. If the theoretic reasoning indicated by the respondent as the one he desired
is a fair description of the reality to which it is to fill in the future.
addressed, the path coefficients for the eight (or Level of educational aspiration (X4—LEA) is
ten) predicted causal lines should be consider- a dichotomous variable corresponding to the re-
ably greater than those for the remainder where spondent’s statement in 1957 of whether or not
no causal prediction was made. Also, it is entirely he planned to attend college after graduating
possible that some unhypothesized causal lines from high school.
might turn out to be of importance. This, too, Index of significant others’ influence (X5—
argues for calculating the whole set of 26. These SOI) is a simple summated score (range: zero
data are presented in tabular form in Table 67.3. to three) of three variables: (a) The youth’s re-
port of his parents’ encouragement for college,
dichotomized according to whether or not the
Method respondent perceived direct parental encourage-
In 1957 all high school seniors in Wisconsin ment for going to college. (b) The youth’s report
responded to an extensive questionnaire con- of his teachers’ encouragement for college, di-
cerning their educational and occupational as- chotomized in a similar manner, according to
pirations and a number of potentially related whether or not direct teacher encouragement
topics. In 1964 one of the authors (Sewell) for college was perceived by the respondent. (c)
directed a follow-up in which data on later ed- Friends’ college plans, dichotomized according
ucational and occupational attainments were to the respondent’s statement that most of his
collected from an approximately one-third ran- close friends planned or did not plan to go to
dom sample of the respondents in the original college. These variables, all emphasizing edu-
survey. cation, were combined because they reflect the
This study is concerned with those 929 sub- same conceptual dimension, and that dimen-
jects for whom data are available at both times, sion is theoretically more relevant than any of its
in 1957 and 1964, and who (a) are males and (b) component parts. That the three components do
whose fathers were farmers in 1957. Zero-order in fact measure the same dimension is attested
correlations are computed on all 929 cases, using by the positive correlations among them and a
a computer program which accepts missing data. subsequent factor analysis. These correlations
All higher order coefficients are based on 739 and the correlation of each with the summated
cases for whom data on each variable were com- variable, significant others’ influence, are shown
plete. (The matrices of zero-order correlations in Table 67.1. It may be relevant to point out the
between all eight variables for those two sets of composition of this significant others’ index in
cases are practically identical.) the light of Kelley’s distinction (1952). Clearly,
 Sewell, Haller, and Portes—The Educational and Early Occupational Attainment Process • 601

Table 67.1. Zero-Order Correlations Between Indicators of Significant Others’ Influence Regarding
College

Teachers’ Friends’ Index of Significant

Influence Influence Others’ Influence
Parental Influence .37 .26 .74
Teachers’ Influence -- .32 .72
Friends’ Influence -- -- .68
Significant Others’ Influence -- -- --

the perceptions of direct parental and teacher lines to be intelligible, because path coefficients
pressures toward college conform to the classic presented in Figure 67.1 were calculated for all
case of normative reference groups. The educa- 26 possible lines implied in the causal order
tional plans of close friends, on the other hand, specified above. With the exception of the the-
may be thought of as having mixed functions. oretically dubious direct path from SES to AP,
First, close peer groups may exercise pressure to- which turned out to be p67 = .01, each of the
ward conformity, and second, friends’ plans also path coefficients for causal lines hypothesized
serve for the individual’s cognitive comparison in Figure 67.1 is larger than those not hypoth-
of himself with “people like himself.” Therefore, esized. Both sets of standardized beta (or path)
though the main character of the dimension in- coefficients are presented in Table 67.3.
dicated by this index is clearly normative, it can This table shows that the reasoning presented
be thought of as containing some elements of an in the above section, offering a social psychologi-
evaluative function as well. cal explanation for educational and occupational
Quality of academic performance (X6—AP) is attainment, cannot be too far off the mark. We
measured by a reflected arc sine transformation had hypothesized that SOI (significant others’
of each student’s rank in his high school class. influence) was of central importance. In fact, it
Socioeconomic status (X7—SES) is measured has notable direct effects on three subsequent
by a factor-weighted combination of the educa- variables, each of which bears ultimately on
tion of the respondent’s father and mother, his prestige level of occupational attainment. Both
perception of the economic status of the fam- theory and data agree that SOI has direct ef-
ily, his perception of possible parental support fects on levels of educational and occupational
should he choose to go to college and the ap- aspiration, as well as educational (i.e., college)
proximate amount of such support, and the oc- attainment. In turn, each aspiration variable
cupation of his father.2 appears to have the predicted substantial effects
Measured mental ability (X8—MA) is indexed on its respective attainment variable. Looking at
by Henmon-Nelson test scores (1942). The data its antecedents, we note theory and data again
were taken when the youths were in the junior agree that SOI is affected directly by SES and
year of high school. The scores, originally re- indirectly by measured mental ability through
corded as percentile-ranks, were treated with an the latter’s effect on the youth’s academic perfor-
arc sine transformation to approximate a normal mance. The latter variable is crucial because it
distribution.3 provides (or is correlated with) palpable evidence
that significant others can observe and, thus to
a degree, align their expectations for the youth
Results with his demonstrated ability.
The zero-order correlation coefficients among None of the unpredicted paths is very strong,
eight variables are presented in Table 67.2. A but we must recognize that there may be more
complete path diagram would involve too many operating in such a system than we were able
 602

Table 67.2. Zero-Order Correlations

X1 Occupational X2 Educational X3 Level of X4 Level of X5 Significant X6 Academic X7 Socio- X8 Measured
Attainment Attainment Occupational Educational Others’ Performance economic Mental
(Prestige Scores: (Years College) Aspiration Aspiration Influence (Grade Point) Status Ability
Duncan)

X1-Occ. Att. -- .52 .43 .38 .41 .37 .14 .33

X2-Ed. Att. -- -- .53 .61 .57 .48 .23 .40
X3-LOA -- -- -- .70 .53 .43 .15 .41
X4-LEA -- -- -- -- .59 .46 .26 .40
X5-SOI -- -- -- -- -- .49 .29 .41
X6-AP -- -- -- -- -- -- .16 .62
X7-SES -- -- -- -- -- -- -- .21
X8-MA -- -- -- -- -- -- -- --
 Sewell, Haller, and Portes—The Educational and Early Occupational Attainment Process • 603

Table 67.3. Standardized Beta Coefficients for Hypothesized and Non-Hypothesized Causal Paths

Independent Variables
Dependent X2 X3 X4 X5 X6 X7 X8
Variables EdAtt LOA LEA SOI AP SES MA
X6 AP -- -- -- -- -- (.01) .62
X5 SOI -- -- -- -- .39 .21 .13
X4 LEA -- -- -- .45 .18 .07 .08
X3 LOA -- -- -- .42 .12 –.02 .16
X2 EdAtt -- .07 .34 (.23) .17 .05 .03
X1 OccAtt .38 .19 –.10 .11 .06 .00 .04

to anticipate from previous thinking. There is a occupational attainment. Indeed, educational

pair of perhaps consequential direct paths from attainment alone accounts for 27 percent of the
academic performance to educational aspiration variance in occupational attainment (from Table
(p46 = .18) and to educational attainment (p26 67.2, r212 = .522 = .27). What we have here, then,
= .17). There are several possibilities. The data is a plausible causal system functioning primarily
might imply the existence of a mediating fac- to explain variation in educational attainment.
tor, such as one’s self-conception of his ability, This, in turn, has considerable effect on occupa-
a factor which could influence both educational tional attainment. The same set of variables adds
aspirations and attainment. They also suggest a small but useful amount to the explanation of
that not all of the effect of ability on educational occupational attainment variance beyond that
aspiration and attainments is mediated by SOI. contributed by its explanation of educational
Finally, one’s ability may exert a continuing ef- attainment.4
fect on his educational attainments quite apart
from the mediation of either significant others
or aspirations—and therefore apart from one’s Discussion and Conclusions
conception of his ability. Arguments such as Using father’s occupational prestige, the person’s
these, however, should not be pressed too far be- educational attainment, and his first job level,
cause the figures are small. Another unexpected Blau and Duncan (1967:165–172) were able to
but noteworthy path links mental ability directly account for 33 percent of the variance in occu-
to level of occupational aspiration. We offer no pational attainment of a nationwide sample of
speculation regarding it. American men. Neither our sample nor our vari-
So far we have seen that a consistent and ables are identical with theirs; so it is impossible
plausible social psychological position is at least to assess the total contribution of this study to
moderately well borne out by the analysis of the state of knowledge as reflected in their work.
lines of apparent influence of its variables when Educational attainment is strategic in both stud-
they are arranged in causal order. How well does ies and in this regard the studies are fairly com-
the total set of independent variables work in parable. The present model adds a great deal to
accounting for variance in the attainment vari- the explanation of the social psychological fac-
ables? In brief, R21.2345678 = .34 and R22.345678 = .50. tors affecting that variable. The prospects seem
Thus, the variables account for 34 percent of the good, too, that if the present model were to be
variance in level of occupational attainment and applied to a sample coming from a wider range
50 percent of the variance in level of educational of the American stratification system with greater
attainment. Obviously, variables X3 through X8 age variation, it might prove to be more pow-
are much more effective in accounting for ed- erful than it appears with our sample of young
ucational attainment than in accounting for farm-reared men. In general, the present take-off
604 • Part VI: Who Gets Ahead?

on the Blau-Duncan approach to occupational institutionalized types of achievement patterns),

attainment levels seems worthy of further testing there is also the question of the relevance of the
and elaboration. model for ascribed occupational attainment sys-
Several comments are appropriate regarding tems. Obviously we do not have data bearing on
the social psychological position and data pre- these questions but we may at least discuss them.
sented here. (1) Clearly, the variable we have Let us suppose that the same eight variables are
called significant others’ influence is an impor- measured on youth in a “sponsored” achievement
tant factor. The present evidence appears to show context. We speculate that if measured mental
that once formed its effects are far-reaching. Also, ability is the basis of selection of those who are
besides being a powerful explanatory factor, sig- to be advanced, then the direct path from men-
nificant others’ influence should be amenable to tal ability to significant others’ influence would
manipulation. It thus suggests itself as a point at increase because sponsors are significant others.
which external agents might intervene to change (This would require a more general measure of
educational and occupational attainment levels. significant others’ influence than was used here.)
This means that at least part of the system is If a variable other than mental ability or socio-
theoretically amenable to experimental testing. economic status is important to the sponsors,
The parts of the present model which are hypo- then the residual effect of unmeasured variables
thetically dependent upon this variable might be on significant others’ influence would increase.
more securely tested if such experiments can be Since one’s sponsors presumably influence one’s
worked out. Also, practical change agents might aspirations and aspirations in turn mediate at-
be able to change levels of attainment, either by tainment, the rest of the model probably would
inserting themselves or others as new significant not change much.
others or by changing the expectations existing Consider the case of ascribed attainment.
significant others have for the individual. There Here one’s parents’ position determines what
may well be a substantial pay-off from more re- one’s significant others will expect of one; mental
fined work with this variable. ability is either irrelevant or controlled by family
(2) The results seem to indicate, too, that as- position; and one’s aspirations are controlled by
pirations (a special class of attitudes) are in fact the family. The importance of higher education
performing mediational functions in transmit- may vary among basically ascribed systems: in
ting anterior factors into subsequent behaviors. one it may be unimportant, in another it may
This has been a subject of recent debate, much merely validate one’s status, or in still another it
of which has in effect held that attitudinal vari- may train ascribed elites to fulfill the key social
ables are useless epiphenomena. This was re- roles in the society. If educational attainment is
cently discussed by Fendrich (1967). important within the social system, aspirations
Such encouraging results do not, however, will mediate the influence of significant others
mitigate the need for (a) general experimental upon it, and it in turn will mediate occupational
determination of the supposed effects of atti- attainment. If not, occupational aspirations
tudes on behaviors, and (b) specific experimen- will mediate occupational attainment and ed-
tal determination of the effects of aspirations on ucational attainment will drop out of the path
attainments. model. In short, by allowing for variations in the
(3) The question may be raised as to the path coefficients, the same basic social psycho-
extent to which this system is inherently logical model might work well to describe attain-
culture-bound. One might wonder whether at- ment in stratification and mobility systems quite
tainment behavior within an institutionalized different from that of the present sample.
pattern of “sponsored” rather than “contest” (4) The linear model used here seems to be
achievement (Turner, 1960) would change the an appropriate way to operationalize social psy-
path model. Besides this (and perhaps other chological positions holding that the function of
 Sewell, Haller, and Portes—The Educational and Early Occupational Attainment Process • 605

“intervening” attitudinal variables is to mediate antecedents to occupational attainment may also

the influence of more fundamental social struc- change. In particular, mental ability may show a
tural and psychological variables on behavior. By higher path to occupational attainment.
assuming linear relations among variables and (7) Finally, although the results reported in
applying a path system to the analysis, we have this paper indicate that the proposed model has
cast the attainment problem in such a frame- considerable promise for explaining educational
work. It seems to have worked quite well. We and early occupational attainment of farm boys,
are sufficiently encouraged by this attempt to its adequacy should now be tested on popula-
recommend that a parallel tack might be made tions with a more differentiated socioeconomic
on problems in which the overt behavior vari- background. It is quite possible that in such
ables are quite different from educational and populations the effects of socioeconomic status
occupational attainment. on subsequent variables may be significantly
(5) Nonetheless, satisfactory as such a linear increased. The effects of other variables in the
model and its accompanying theory seems to be, system may also be altered when the model is
there is still the possibility that other techniques applied to less homogeneous populations.
flowing from somewhat different social psycho- The present research appears to have ex-
logical assumptions might be better. It is possible tended knowledge of the causal mechanism
that, in the action situation, enduring attitudes influencing occupational attainment. Most of
(such as educational and occupational aspira- this was accomplished by providing a consistent
tions) may function as independent forces which social psychological model which adds to our
express themselves in relevant overt behaviors to ability to explain what is surely one of its key
the degree that other personality and situational proximal antecedents, educational attainment.
variables permit. Linear models would thus be
effective to the degree that the persons modify NOTES
their aspirations to bring them in line with po- Revision of paper originally prepared for deliv-
ery at the joint sessions of the Rural Sociological
tentials for action offered by the latter variables.
Society and the American Sociological Association,
More importantly, the combined effects of aspi-
San Francisco, August 1967. The research reported
rational and facilitational variables would pro- here was supported by the University of Wisconsin
duce nonlinear accelerating curves of influence Graduate School, by the Cooperative State Research
on behavior variables. For the present types of Service, and the University’s College of Agriculture
data, this would imply that parental stratification for North Central Regional Research Committee
position, mental ability, and significant others’ NC-86, by funds to the Institute for Research on
influence not only produce aspirations, but also, Poverty at the University of Wisconsin provided by
to the extent to which these influences continue the Office of Economic Opportunity pursuant to
more or less unchanged on into early adulthood, the provisions of the Economic Opportunity Act of
1964, and by a grant from the National Institute of
they function as differential facilitators for the
Health, U. S. Public Health Service (M-6275). The
expression of aspirations in attainments. If this writers wish to thank Otis Dudley Duncan for his
is true, a nonlinear system of statistical analysis careful reading and incisive criticisms and Vimal P.
handling interaction effects would be even more Shah for help in the statistical analysis. The conclu-
powerful than the one used in this paper. sions are the full responsibility of the authors.
(6) It should be remembered that the most 1. It is important to note that the timing of the
highly educated of these young men had just follow-up was such as to allow most individuals to
begun their careers when the final data were complete their education up to the bachelor’s degree
collected. If the distance between them and the and beyond. It is unlikely that the educational at-
tainment of the sample as a whole will change much
less educated widens, the occupational attain-
in the years to come. On the other hand, while the
ment variance accounted for by the model may
span of seven years allowed those individuals who
well increase. The direct relations of some of the did not continue their education to find a stable
606 • Part VI: Who Gets Ahead?

position in the occupational structure and even im- aspirations and attainments.” American Sociologi-
prove upon it, there was not enough time for those cal Review 29 (August):568–575.
who continued their education to do the same. A few Blau, Peter M., and Otis Dudley Duncan. 1967.
of the latter were still in school; most had just begun The American Occupational Structure. New York:
their occupational careers. It is therefore possible Wiley.
that a follow-up taken five or ten years from now Bordua, David J. 1960. “Educational aspirations
would show greater differentiation in attainments as and parental stress on college.” Social Forces 38
the educated group gathers momentum and moves (March):262–269.
up in the occupational world. Brookover, Wilbur B., Jean M. LePere, Don E.
2. Naturally, father’s occupation is a constant in Hamachek, Shailer Thomas, and Edsel L. Er-
this subsample of farm-reared males. It is important ickson. 1965. Self-Concept of Ability and School
to note that the SES mean and standard deviations Achievement. East Lansing: Michigan State Uni-
for this subsample are considerably lower than for versity, Bureau of Educational Research Services.
the total sample. The low and homogeneous SES Campbell, Ernest Q., and C. Norman Alexander.
levels of this subsample may yield atypical relations 1965. “Structural effects and interpersonal re-
among the variables. lationships.” American Journal of Sociology 71
3. Our previous research (Sewell and Armer, (November):284–289.
1966; Haller and Sewell, 1967) has led us to be Cramer, M. R. 1967. “The relationship between
skeptical of claims that local ecological and school educational and occupational plans of high
class compositional factors influence aspirations and school students.” Paper presented at the meet-
attainments. Nevertheless the zero-order intercor- ing of the Southern Sociological Society, Atlanta
relations of five such variables and their correlations (unpublished).
with X1–X8 are available (although they are not pre- Duncan, Otis Dudley. 1961. “A socioeconomic
sented here). Two of these pertain to the county in index for all occupations.” Pp. 109–138 in Albert
which the youth attended high school: county level J. Reiss, Jr. (ed.), Occupations and Social Status.
of living and degree of urbanization. Three pertain to New York: Free Press.
his high school senior class: average SES of the class, ———. 1968. “Ability and achievement.” Eugenics
percentage of the class members whose fathers at- Quarterly 15 (March):1–11.
tended college, and percentage of the class members ———. 1969. “Contingencies in the construction
whose fathers had professional-level occupations. of causal models.” Edgar F. Borgatta (ed.), Socio-
Though substantially correlated with each other, the logical Methodology. San Francisco: Jossey-Bass.
variables are uncorrelated with the variables in the Duncan, Otis Dudley, Archibald O. Haller, and
above model. Alejandro Portes. 1968. “Peer influences on aspi-
4. Some readers will be interested in the path co- rations: a reinterpretation.” American Journal of
efficients as calculated only for the lines hypothesized Sociology 74 (September):119–137.
in the diagram. For this reason and because of the Fendrich, James M. 1967. “Perceived reference
diagram’s parsimony, we have calculated the values group support: racial attitudes and overt be-
for each of its eight paths (or ten, including dubious havior.” American Sociological Review 32
ones). The restricted model explains 47 and 33 per- (December):960–970.
cent of the variance in X2 and X1, respectively. Data Gerth, Hans, and C. Wright Mills. 1953. Character
not presented here show that the model reproduces and Social Structure. New York: Harcourt, Brace
the zero-order correlation matrix quite well. For this and World.
reason and because the model is an effective predic- Gross, Neal, Ward S. Mason, and Alexander W.
tor of X2 and X1, it may be considered to be fairly McEachern. 1958. Explorations in Role Analysis.
valid. Nonetheless, it seems more prudent to rest our New York: Wiley.
case on the less presumptuous data already presented Haller, Archibald O. 1966. “Occupational choices
in Table 67.3. This is why the coefficients presented of rural youth.” Journal of Cooperative Extension
in the diagram are not discussed here. 4 (Summer):93–102.
Haller, Archibald O., and Charles E. Butterworth.
REFERENCES 1960. “Peer influences on levels of occupational
Alexander, C. Norman, Jr., and Ernest Q. Campbell. and educational aspiration.” Social Forces 38
1964. “Peer influences on adolescent educational (May):289–295.
 Sewell, Haller, and Portes—The Educational and Early Occupational Attainment Process • 607

Haller, Archibald O., and Irwin W. Miller. 1963. Portes, Alejandro, Archibald O. Haller, and Wil-
The Occupational Aspiration Scale: Theory, Struc- liam H. Sewell. 1968. “Professional executive vs.
ture and Correlates. East Lansing: Michigan Agri- farming as unique occupational choices.” Rural
cultural Experiment Station Bulletin 288. Sociology 33 (June):153–159.
Haller, Archibald O., and William H. Sewell. 1967. Sewell, William H. 1964. “Community of residence
“Occupational choices of Wisconsin farm boys.” and college plans.” American Sociological Review
Rural Sociology 32 (March):37–55. 29 (February):24–38.
Havighurst, Robert J., and Bernice L. Neugarten. Sewell, William H., and J. Michael Armer. 1966.
1957. Society and Education. Boston: Allyn and “Neighborhood context and college plans.” Amer-
Bacon. ican Sociological Review 31 (April): 159–168.
Heise, David R. 1969. “Problems in path analysis Sewell, William H., Archibald O. Haller, and Mur-
and causal inference.” Edgar F. Borgatta (ed.), So- ray A. Straus. 1957. “Social status and educa-
ciological Methodology. San Francisco: Jossey-Bass. tional and occupational aspiration.” American
Henmon, V. A. C., and M. J. Nelson. 1942. The Sociological Review 22 (February):67–73.
Henmon-Nelson Test of Mental Ability. Boston: Sewell, William H., and Alan M. Orenstein.
Houghton Mifflin Company. 1965. “Community of residence and occupa-
Herriott, Robert E. 1963. “Some social determinants tional choice.” American Journal of Sociology 70
of educational aspiration.” Harvard Educational (March):551–563.
Review 33 (Spring):157–177. Sewell, William H., and Vimal P. Shah. 1967. “So-
Kelley, Harold H. 1952. “Two functions of reference cioeconomic status, intelligence, and the attain-
groups.” Pp. 410–414 in Guy E. Swanson et al. ment of higher education.” Sociology of Education
(eds.), Readings in Social Psychology. New York: 40 (Winter):1–23.
Henry Holt and Company. ———. 1968. “Social class, parental encourage-
Kuvlesky, William P., and George W. Ohlendorf. ment, and educational aspirations.” American
1967. A Bibliography of Literature on Status Pro- Journal of Sociology 73 (March):559–572.
jections of Youth: I. Occupational Aspirations and Sherif, Muzafer. 1935. “A study of some social fac-
Expectations. College Station: Texas A&M Uni- tors in perception.” Archives of Psychology Num-
versity, Department of Agricultural Economics ber 187.
and Sociology. Super, Donald E. 1957. The Psychology of Careers.
Lipset, Seymour M., and Reinhard Bendix. 1959. New York: Harper.
Social Mobility in Industrial Society. Berkeley: Turner, Ralph H. 1960. “Sponsored and contest mo-
University of California Press. bility and the school system.” American Sociolog-
Merton, Robert K. 1957. Social Theory and Social ical Review 25 (December):855–867.
Structure. New York: Free Press. Woelfel, Joseph. 1967. “A paradigm for research on
Miller, I. W., and Archibald O. Haller. 1964. significant others.” Paper presented at the Joint
“The measurement of level of occupational as- Session of the Society for the Study of Social
piration.” Personnel and Guidance Journal 42 Problems and the American Sociological Associ-
(January):448–455. ation, San Francisco (unpublished).
Ohlendorf, George W., Sherry Wages, and William Wright, Sewall. 1934. “The method of path co-
P. Kuvlesky. 1967. A Bibliography of Literature on efficients.” Annals of Mathematical Statistics 5
Status Projections of Youth: II. Educational Aspi- (September):161–215.
rations and Expectations. College Station: Texas ———. 1960. “Path coefficients and regression coef-
A & M University, Department of Agricultural ficients: alternative or complementary concept?”
Economics and Sociology. Biometrics 16 (June):189–202.
Parsons, Talcott. 1959. “The school class as a so-
cial system.” Harvard Educational Review 29
(Summer):297–318.
608 • Part VI: Who Gets Ahead?

68 • Jay MacLeod

Ain’t No Makin’ It
Leveled Aspirations in a Low-Income Neighborhood

“Any child can grow up to be president.” So anyway.” So says Freddie Piniella,1 an intelligent
maintains the dominant ideology in the United eleven-year-old boy from Clarendon Heights, a
States. This perspective characterizes Ameri- low-income housing development in a north-
can society as an open one in which barriers to eastern city. This statement, pronounced with
success are mainly personal rather than social. certitude and feeling, completely contradicts
In this meritocratic view, education ensures our achievement ideology. Freddie is pessimistic
equality of opportunity for all individuals, and about his prospects for social mobility and dis-
economic inequalities result from differences putes schooling’s capacity to “deliver the goods.”
in natural qualities and in one’s motivation and Such a view offends our sensibilities and seems a
will to work. Success is based on achievement rationalization. But Freddie has a point. What of
rather than ascription. Individuals do not in- Carnegie’s grammar school classmates, the great
herit their social status—they attain it on their bulk of whom no doubt were left behind to oc-
own. Because schooling mitigates gender, class, cupy positions in the class structure not much
and racial barriers to success, the ladder of so- different from those held by their parents? What
cial mobility is there for all to climb. A favorite about the static, nearly permanent element in
Hollywood theme, the rags-to-riches story reso- the working class, whose members consider the
nates in the psyche of the American people. We chances for mobility remote and thus despair
never tire of hearing about Andrew Carnegie, for of all hope? These people are shunned, hidden,
his experience validates much that we hold dear forgotten—and for good reason—because just
about America, the land of opportunity. Hora- as the self-made man is a testament to certain
tio Alger’s accounts of the spectacular mobility American ideals, so the very existence of an “un-
achieved by men of humble origins through their derclass” in American society is a living contra-
own unremitting efforts occupy a treasured place diction to those ideals.
in our national folklore. The American Dream is Utter hopelessness is the most striking aspect
held out as a genuine prospect for anyone with of Freddie’s outlook. Erik H. Erikson writes
the drive to achieve it. that hope is the basic ingredient of all vitality;2
“I ain’t goin’ to college. Who wants to go stripped of hope, there is little left to lose. How
to college? I’d just end up gettin’ a shitty job is it that in contemporary America a boy of

Jay MacLeod. Ain’t No Makin’ It: Leveled Aspirations in a Low-Income Neighborhood (Boulder: Westview Press,
1987), pp. 1–2, 4–5, 8, 60–63, 69–75, 78–79, 81, 137–141, 162. Copyright © 1987 by Westview Press, Inc.
Reprinted by permission of Westview Press, Inc., a member of the Perseus Books Group. The postscript to this
selection is taken from Ain’t No Makin’ It: Leveled Aspirations in a Low-Income Neighborhood (Boulder: West-
view Press, 1995), pp. 155, 169–170, 196, 240–241. Copyright © 1995 by Westview Press, Inc. Reprinted by
permission of Westview Press, Inc., a member of the Perseus Books Group.
 MacLeod—Ain’t No Makin’ It • 609

eleven can feel bereft of a future worth embrac- an analytical distinction between aspirations and
ing? This is not what the United States is sup- expectations. Both involve assessments of one’s
posed to be. The United States is the nation of desires, abilities, and the character of the oppor-
hopes and dreams and opportunity. As Ronald tunity structure. In articulating one’s aspirations,
Reagan remarked in his 1985 State of the Union an individual weighs his or her preferences more
Address, citing the accomplishments of a young heavily; expectations are tempered by perceived
Vietnamese immigrant, “Anything is possible capabilities and available opportunities. Aspira-
in America if we have the faith, the will, and tions are one’s preferences relatively unsullied by
the heart.”3 But to Freddie Piniella and many anticipated constraints; expectations take these
other Clarendon Heights young people who constraints squarely into account.4
grow up in households where their parents and
older siblings are unemployed, undereducated,
or imprisoned, Reagan’s words ring hollow. For The Hallway Hangers: Keeping a
them the American Dream, far from being a Lid on Hope
genuine prospect, is not even a dream. It is a Conventional, middle-class orientations toward
hallucination. employment are inadequate to describe the
I first met Freddie Piniella in the summer of Hallway Hangers’ approach to work. The no-
1981 when as a student at a nearby university tion of a career, a set of jobs that are connected
I worked as a counselor in a youth enrichment to one another in a logical progression, has little
program in Clarendon Heights. For ten weeks I relevance to these boys. They are hesitant when
lived a few blocks from the housing project and asked about their aspirations and expectations.
worked intensively with nine boys, aged eleven This hesitancy is not the result of indecision;
to thirteen. While engaging them in recreational rather it stems from the fact that these boys
and educational activities, I was surprised by see little choice involved in getting a job. No
the modesty of their aspirations. The world of matter how hard I pressed him, for instance,
middle-class work was entirely alien to them; Jinks refused to articulate his aspirations: “I
they spoke about employment in construction, think you’re kiddin’ yourself to have any. We’re
factories, the armed forces, or, predictably, pro- just gonna take whatever we can get.” Jinks is
fessional athletics. In an ostensibly open society, a perceptive boy, and his answer seems to be
they were a group of boys whose occupational an accurate depiction of the situation. Beg-
aspirations did not even cut across class lines. . . . gars cannot be choosers, and these boys have
The male teenage world of Clarendon nothing other than unskilled labor to offer on a
Heights is populated by two divergent peer credential-based job market.
groups. The first group, dubbed the Hallway It is difficult to gauge the aspirations of most
Hangers because of the group’s propensity for of the Hallway Hangers. Perhaps at a younger
“hanging” in a particular hallway in the project age they had dreams for their futures. At ages
[i.e., outside doorway #13], consists predomi- sixteen, seventeen, and eighteen, however, their
nantly of white boys. Their characteristics and own job experiences as well as those of fam-
attitudes stand in marked contrast to the second ily members have contributed to a deeply en-
group, which is composed almost exclusively of trenched cynicism about their futures. What is
black youths who call themselves the Brothers. perceived as the cold, hard reality of the job mar-
Surprisingly, the Brothers speak with relative ket weighs very heavily on the Hallway Hangers;
optimism about their futures, while the Hallway they believe their preferences will have almost
Hangers are despondent about their prospects no bearing on the work they actually will do.
for social mobility. . . . Their expectations are not merely tempered by
Before describing the boys’ orientation to- perceptions of the opportunity structure; even
ward work [in more detail], I would like to make their aspirations are crushed by their estimation
610 • Part VI: Who Gets Ahead?

of the job market. These generalizations may STEVE: I don’t fucking know. Working prob-
seem bold and rather extreme, but they do not ably. Have my own pad, my own house.
lack ethnographic support. Bitches, kids. Fucking fridge full of brews-
The pessimism and uncertainty with which kies. Fine wife, likes to get laid.
the Hallway Hangers view their futures emerge
clearly when the boys are asked to speculate on JINKS: Twenty years from now? Probably kicked
what their lives will be like in twenty years. back in my own apartment doing the same
shit I’m doing now—getting high. I’ll have
(all in separate interviews) a job, if I’m not in the service, if war don’t
break out, if I’m not dead. I just take one day
STONEY: Hard to say. I could be dead tomor- at a time.
row. Around here, you gotta take life day by
day. Although the Hallway Hangers expect to spend
a good portion of their waking hours on the job,
BOO-BOO: I dunno. I don’t want to think work is important to them not as an end in itself,
about it. I’ll think about it when it comes. but solely as a means to an end—money.
In probing the occupational aspirations and
FRANKIE: I don’t fucking know. Twenty years. expectations of the Hallway Hangers, I finally
I may be fucking dead. I live a day at a time. was able to elicit from them some specific hopes.
I’ll probably be in the fucking pen. Although Shorty never mentions his expec-
tations, the rest of the Hallway Hangers have
SHORTY: Twenty years? I’m gonna be in jail. responded to my prodding with some definite
answers. The range of answers as well as how
These responses are striking not only for the they change over time are as significant as the
insecurity and despondency they reveal, but particular hopes each boy expresses.
also because they do not include any mention Boo-Boo’s orientation toward work is typical
of work. It is not that work is unimportant— of the Hallway Hangers. He has held a number
for people as strapped for money as the Hall- of jobs in the past, most of them in the sum-
way Hangers are, work is crucial. Rather, these mer. During his freshman year in high school
boys are indifferent to the issue of future em- Boo-Boo worked as a security guard at school
ployment. Work is a given; they all hope to hold for $2.50 an hour in order to make restitution
jobs of one kind or another in order to support for a stolen car he damaged. Boo-Boo also has
themselves and their families. But the Hallway worked on small-scale construction projects
Hangers believe the character of work, at least through a summer youth employment program
all work in which they are likely to be involved, called Just-A-Start, at a pipe manufacturing site,
is essentially the same: boring, undifferentiated, and as a clerk in a gift shop. Boo-Boo wants to
and unrewarding. Thinking about their future be an automobile mechanic. Upon graduating
jobs is a useless activity for the Hallway Hangers. from high school, he studied auto mechanics at a
What is there to think about? technical school on a scholarship. The only black
For Steve and Jinks, although they do see student in his class, Boo-Boo was expelled early
themselves employed in twenty years, work is in his first term after racial antagonism erupted
still of tangential importance. into a fight. Boo-Boo was not altogether disap-
pointed, for he already was unhappy with what
JM: If you had to guess, what do you think you’ll he considered the program’s overly theoretical
be doing twenty years from now? orientation. (Howard London found this kind
of impatience typical of working-class students
(in separate interviews) in the community college he studied.5) Boo-Boo
 MacLeod—Ain’t No Makin’ It • 611

wanted hands-on training, but “all’s they were Jinks subsequently quit school. He had
doing was telling me about how it’s made, stuff been working twenty hours a week making
like that.” Boo-Boo currently is unemployed, but clothes-racks in a factory with his brother.
he recently had a chance for a job as a cook’s He left school with the understanding that he
helper. Although he was not hired, the event would be employed full-time, and he was mildly
is significant nevertheless because prior to the content with his situation: “I got a job. It ain’t
job interview, Boo-Boo claimed that his ambi- a good job, but other things will come along.”
tion now was to work in a restaurant. Here we Two weeks later, he was laid off. For the past
have an example of the primacy of the oppor- three months he has been unemployed, hanging
tunity structure in determining the aspirations full-time in doorway #13.
of the Hallway Hangers. One job opening in Shorty has worked construction in the past
another field was so significant that the open- and has held odd jobs such as shoveling snow.
ing prompted Boo-Boo to redefine totally his Shorty, an alcoholic, has trouble holding down a
aspirations. steady job, as he freely admits. He was enrolled
In contrast to the rest of the Hallway Hangers in school until recently. Ordered by the court to
who are already on the job market, Steve wants a detoxification center, Shorty apparently man-
to stay in school for the two years required to get aged to convince the judge that he had attended
his diploma. Yet he has a similar attitude toward enough Alcoholics Anonymous meetings in the
his future work as do the other youths. He quit meantime to satisfy the court. He has not re-
his summer job with the Just-A-Start program turned to school since, nor has he landed a job.
and has no concrete occupational aspirations. Given that Shorty is often on the run from the
As for expectations, he believes he might enlist police, he is too preoccupied with pressing ev-
in the Air Force after graduation but adds, “I eryday problems to give serious thought to his
dunno. I might just go up and see my uncle, do long-term future. It is not surprising that my
some fuckin’ construction or something.” ill-timed query about his occupational aspira-
Many of these boys expect to enter military tions met with only an impatient glare. . . .
service. Jinks and Frankie mention it as an op- The definitions of aspirations and expecta-
tion; Stoney has tried to enlist, but without suc- tions given [earlier] suggest that an assessment
cess. Although Jinks refuses to think in terms of the opportunity structure and of one’s capabil-
of aspirations, he will say what he expects to do ities impinge on one’s preferences for the future.
after he finishes school. However, the portrait of the Hallway Hangers
painted in these pages makes clear that “im-
JM: What are you gonna do when you get out? pinge” is not a strong enough word. But are the
leveled aspirations and pessimistic expectations
JINKS: Go into the service, like everybody else. of the Hallway Hangers a result of strong neg-
The navy. ative assessments of their capabilities or of the
opportunity structure?
JM: What about after that? This is not an easy question to answer.
Doubtless, both factors come into play, but in
JINKS: After that, just get a job, live around the case of the Hallway Hangers, evaluation of
here. the opportunity structure has the dominant role.
Although in a discussion of why they do not
JM: Do you have any idea what job you wanna succeed in school, the Hallway Hangers point
get? to personal inadequacy (“We’re all just fucking
burnouts”; “We never did good anyways”), they
JINKS: No. No particular job. Whatever I can look to outside forces as well. In general, they are
get. confident of their own abilities.
612 • Part VI: Who Gets Ahead?

(In a group interview) SLICK: And concentrated on studying. . . .

JM: If you’ve got five kids up the high school FRANKIE: I know I could make it. I am a
with all A’s, now are you gonna be able to businessman.
say that any of them are smarter than any
of you? JM: So all of you are sure that if you put out in
school. . . .
SLICK: (immediately) No.
FRANKIE: Yeah! If I went into business, I
JM: So how’d that happen? would, yeah. If I had the fucking money
to start out with like some of these fucking
SLICK: Because they’re smarter in some areas rich kids, I’d be a millionaire. Fucking right
just like we’re smarter in some areas. You I would be.
put them out here, right? And you put us up
where they’re living—they won’t be able to Although these comments were influenced by the
survive out here. dynamics of the group interview, they jibe with
the general sense of self-confidence the Hallway
SHORTY: But we’d be able to survive up there. Hangers radiate and indicate that they do not
have low perceptions of their own abilities.
FRANKIE: See, what it is—they’re smarter more If their assessments of their own abilities do
academically because they’re taught by teach- not account for the low aspirations of the Hall-
ers that teach academics. way Hangers, we are left, by way of explanation,
with their perceptions of the job opportunity
JM: Not even streetwise, just academically, do structure. The dominant view in the United
you think you could be up where they are? States is that American society is an open one
that values and differentially rewards individ-
FRANKIE: Yeah. uals on the basis of their merits. The Hallway
Hangers question this view, for it runs against
CHRIS: Yeah. the grain of their neighbors’ experiences, their
families’ experiences, and their own encounters
SHORTY: Yeah. with the labor market.
The Clarendon Heights community, as a
JM: When it comes down to it, you’re just as public housing development, is by definition
smart? made up of individuals who do not hold even
modestly remunerative jobs. A large majority are
FRANKIE: Yeah. on additional forms of public assistance; many
are unemployed. Like most old housing projects,
SLICK: (matter-of-factly) We could be smarter. Clarendon Heights tends to be a cloistered, in-
sular neighborhood, isolated from the surround-
FRANKIE: Definitely. ing community. Although younger residents
certainly have external points of reference, their
CHRIS: On the street, like. horizons are nevertheless very narrow. Their im-
mediate world is composed almost entirely of
FRANKIE: We’re smart, we’re smart, but we’re people who have not “made it.” To look around
just smart [inaudible]. It’s fucking, y’know, at a great variety of people—some lazy, some al-
we’re just out to make money, man. I know if coholics, some energetic, some dedicated, some
I ever went to fucking high school and college clever, some resourceful—and to realize all of
in a business course. . . . them have been unsuccessful on the job market
 MacLeod—Ain’t No Makin’ It • 613

is powerful testimony against what is billed as (in a separate interview)

an open society.
The second and much more intimate contact SLICK: That’s why I went into the army—cuz
these boys have with the job market is through there’s no jobs out here right now for people
their families, whose occupational histories only that, y’know, live out here. You have to know
can be viewed as sad and disillusioning by the somebody, right?
Hallway Hangers. These are not people who are
slothful or slow-witted; rather, they are gener- In discussing the problems of getting a job, both
ally industrious, intelligent, and very willing to Slick and Shorty are vocal.
work. With members of their families holding
low-paying, unstable jobs or unable to find work SLICK: All right, to get a job, first of all, this is
at all, the Hallway Hangers are unlikely to view a handicap, out here. If you say you’re from
the job opportunity structure as an open one. the projects or anywhere in this area, that can
The third level of experience on which the hurt you. Right off the bat: reputation.
Hallway Hangers draw is their own. These boys
are not newcomers to the job market. As we have SHORTY: Is this dude gonna rip me off, is
seen, all have held a variety of jobs. All except he. . . .
Steve are now on the job market year round, but
only Stoney has a steady job. With the excep- SLICK: Is he gonna stab me?
tions of Chris, who presently is satisfied with his
success peddling drugs, and Steve, who is still SHORTY: Will he rip me off? Is he gonna set
in school, the Hallway Hangers are actively in up the place to do a score or somethin’? I
search of decent work. Although they always tried to get a couple of my buddies jobs at
seem to be following up on some promising a place where I was working construction,
lead, they are all unemployed. Furthermore, but the guy says, “I don’t want ’em if they’re
some who were counting on prospective employ- from there. I know you; you ain’t a thief or
ment have had their hopes dashed when it fell nothing.”
through. The work they have been able to secure
typically has been in menial, dead-end jobs pay- Frankie also points out the reservations prospec-
ing minimum wage. tive employers have about hiring people who live
Thus, their personal experience on the job in Clarendon Heights. “A rich kid would have a
market and the experiences of their family mem- better chance of getting a job than me, yeah. Me,
bers and their neighbors have taught the Hallway from where I live, y’know, a high crime area, I
Hangers that the job market does not necessarily was prob’ly crime-breaking myself, which they
reward talent or effort. Neither they nor their think your nice honest rich kid from a very re-
parents, older siblings, and friends have shared spected family would never do.”
in the “spoils” of economic success. In short, the Frankie also feels that he is discriminated
Hallway Hangers are under no illusions about against because of the reputation that attaches
the openness of the job opportunity structure. to him because of his brothers’ illegal exploits.
They are conscious, albeit vaguely, of a number “Especially me, like I’ve had a few opportuni-
of class-based obstacles to economic and social ties for a job, y’know. I didn’t get it cuz of my
advancement. Slick, the most perceptive and name, because of my brothers, y’know. So I was
articulate of the Hallway Hangers, points out deprived right there, bang. Y’know they said,
particular barriers they must face. ‘No, no, no, we ain’t havin’ no Dougherty work
for us.’” In a separate discussion, Frankie again
SLICK: Out here, there’s not the opportunity to makes this point. Arguing that he would have
make money. That’s how you get into stealin’ almost no chance to be hired as a fireman, de-
and all that shit. spite ostensibly meritocratic hiring procedures,
614 • Part VI: Who Gets Ahead?

even if he scored very highly on the test, Frankie others from higher social strata do not have to
concludes, “Just cuz fuckin’ where I’m from and contend.
what my name is.” Insofar as contemporary conditions under
The Hallway Hangers’ belief that the oppor- capitalism can be conceptualized as a race by
tunity structure is not open also emerges when the many for relatively few positions of wealth
we consider their responses to the question of and prestige, the low aspirations of the Hallway
whether they have the same chance as a middle- Hangers, more than anything else, seem to be
or upper-class boy to get a good job. The Hall- a decision, conscious or unconscious, to with-
way Hangers generally respond in the negative. draw from the running. The competition, they
When pushed to explain why, Jinks and Steve reason, is not a fair one when some people have
made these responses, which are typical. an unobstructed lane. As Frankie maintains, the
Hallway Hangers face numerous barriers: “It’s a
(in separate interviews) steeplechase, man. It’s a motherfucking steeple-
chase.” The Hallway Hangers respond in a way
JINKS: Their parents got pull and shit. that suggests only a “sucker” would compete se-
riously under such conditions.
STEVE: Their fucking parents know people. Chris’s perspective seems a poignant, accurate
description of the situation in which the Hall-
Considering the boys’ employment experi- way Hangers find themselves.
ences and those of their families, it is not sur-
prising that the Hallway Hangers’ view of the CHRIS: I gotta get a job, any fucking job. Just
job market does not conform to the dominant a job. Make some decent money. If I could
belief in the openness of the opportunity struc- make a hundred bucks a week, I’d work. I
ture. They see a job market where rewards are just wanna get my mother out of the proj-
based not on meritocratic criteria, but on “who ects, that’s all. But I’m fucking up in school.
you know.” If “connections” are the keys to suc- It ain’t easy, Jay. I hang out there [in door-
cess, the Hallway Hangers know that they are way #13] ’til about one o’clock every night. I
in trouble. never want to go to school. I’d much rather
Aside from their assessment of the job oppor- hang out and get high again. It’s not that I’m
tunity structure, the Hallway Hangers are aware dumb. You gimme thirty bucks today, and I’ll
of other forces weighing on their futures. A gen- give you one hundred tomorrow. I dunno. It’s
eral feeling of despondency pervades the group. like I’m in a hole I can’t get out of. I guess I
As Slick puts it, “The younger kids have nothing could get out, but it’s hard as hell. It’s fucked
to hope for.” The Hallway Hangers often draw up.
attention to specific incidents that support their
general and vague feelings of hopelessness and of
the futility of nurturing aspirations or high ex- The Brothers: Ready at the
pectations. Tales of police brutality, of uncaring Starting Line
probation officers and callous judges, and of the Just as the pessimism and uncertainty with
“pull and hook-ups of the rich kids” all have a which the Hallway Hangers view their futures
common theme, which Chris summarizes, “We emerges when we consider what they perceive
don’t get a fair shake and shit.” Although they their lives will be like in twenty years, so do the
sometimes internalize the blame for their plight Brothers’ long-term visions serve as a valuable
(Boo-Boo: “I just screwed up”; Chris: “I guess I backdrop to our discussion of their aspirations.
just don’t have what it takes”; Frankie: “We’ve The ethos of the Brothers’ peer group is a posi-
just fucked up”), the Hallway Hangers also see, tive one; they are not resigned to a bleak future
albeit in a vague and imprecise manner, a num- but are hoping for a bright one. Nowhere does
ber of hurdles in their path to success with which this optimism surface more clearly than in the
 MacLeod—Ain’t No Makin’ It • 615

Brothers’ responses to the question of what they summers, he has worked for the city doing
will be doing in twenty years. Note the centrality maintenance work in parks and school buildings
of work in their views of the future. through a CETA-sponsored summer youth em-
ployment program. During the last year, Super’s
(all in separate interviews) occupational aspirations have fluctuated widely.
His initial desire to become a doctor was met
SUPER: I’ll have a house, a nice car, no one both- with laughter from his friends. Deterred by their
ering me. Won’t have to take no hard time mocking and by a realization of the schooling re-
from no one. Yeah, I’ll have a good job, too. quired to be a doctor, Super immediately decided
that he would rather go into business: “Maybe I
JUAN: I’ll have a regular house, y’know, with can own my own shop and shit.” This aspiration,
a yard and everything. I’ll have a steady job, however, also was ridiculed. “Yeah, right,” com-
a good job. I’ll be living the good life, the mented Mokey, “Super’ll be pimping the girls,
easy life. that kinda business.” In private, however, Super
still clings to the hope of becoming a doctor, al-
MIKE: I might have a wife, some kids. I might though he cites work in the computer field as a
be holding down a regular business job like more realistic hope. “Really, I don’t know what I
an old guy. I hope I’ll be able to do a lot of should do now. I’m kinda confused. First I said I
skiing and stuff like that when I’m old. wanna go into computers, right? Take up that or
a doctor.” The vagueness of Super’s aspirations is
CRAIG: I’ll probably be having a good job on important; once again, we get a glimpse of how
my hands, I think. Working in an office as little is known about the world of middle-class
an architect, y’know, with my own drawing work, even for somebody who clearly aspires to
board, doing my own stuff, or at least close it. Of one thing Super is certain: “I just know I
to there. wanna get a good job.”
Although Super does not distinguish between
James takes a comic look into his future with- what constitutes a good job and what does not,
out being prompted to do so. “The ones who he does allude to criteria by which the quality
work hard in school, eventually it’s gonna pay of a job can be judged. First, a good job must
off for them and everything, and they’re gonna not demand that one “work on your feet,” a
have a good job and a family and all that. Not distinction, apparently, between white and
me though! I’m gonna have myself. I’m gonna blue-collar work. Second, a good job implies at
have some money. And a different girl every day. least some authority in one’s workplace, a point
And a different car. And be like this (poses with Super makes clearly, if in a disjointed manner.
one arm around an imaginary girl and the other on “Bosses—if you don’t come on time, they yell
a steering wheel ).” at you and stuff like that. They want you to do
The Brothers do not hesitate to name their work and not sit down and relax and stuff like
occupational goals. Although some of the Broth- that, y’know. I want to try and be a boss, y’know,
ers are unsure of their occupational aspirations, tell people what to do. See, I don’t always want
none seems to feel that nurturing an aspiration people telling me what to do, y’know—the low
is a futile exercise. The Brothers have not re- rank. I wanna try to be with people in the high
signed themselves to taking whatever they can rank.” Although Super does not know what oc-
get. Rather, they articulate specific occupational cupation he would like to enter, he is certain
aspirations (although these often are subject to that he wants a job that is relatively high up in a
change and revision). vaguely defined occupational hierarchy. . . .
Like all the Brothers, Super has not had ex- The Brothers display none of the cockiness
tensive experience on the job market; he only about their own capabilities that the Hallway
has worked at summer jobs. For the past three Hangers exhibit. Instead, they attribute lack
616 • Part VI: Who Gets Ahead?

of success on the job market exclusively to per- that the Brothers apply the same criteria in judg-
sonal inadequacy. This is particularly true when ing each other’s chances to gain meaningful em-
the Brothers speculate about the future jobs the ployment. James thinks Mokey is headed for a
Hallway Hangers and their own friends will dead-end job because he is immature and un-
have. According to the Brothers, the Hallway disciplined. He also blames Juan for currently
Hangers (in Super’s words) “ain’t gonna get no- being out of work. “Juan’s outta school, and Juan
where,” not because of the harshness of the job does not have a job (said with contempt). Now
market but because they are personally lacking. that’s some kind of a senior. When I’m a senior,
The rest of the Brothers share this view. I’m gonna have a job already. I can see if you’re
gonna go to college right when you get out of
JM: Some of those guys who hang with Frankie, school; but Juan’s not doin’ nothin’. He’s just sta-
they’re actually pretty smart. They just don’t yin’ home.” Juan, in turn, thinks that Mokey and
channel that intelligence into school, it seems Super will have difficulty finding valuable work
to me. because of their attitudes. He predicts that Derek
and Craig will be successful for the same reason.
CRAIG: I call that stupid, man. That’s what they These viewpoints are consistent with the
are. dominant ideology in America; barriers to suc-
cess are seen as personal rather than social. By
JM: I dunno. attributing failure to personal inadequacy, the
Brothers exonerate the opportunity structure.
CRAIG: Lazy. Indeed, it is amazing how often they affirm the
openness of American society.
(in a separate interview)
(all in separate interviews)
SUPER: They think they’re so tough they don’t
have to do work. That don’t make sense, re- DEREK: If you put your mind to it, if you want
ally. You ain’t gonna get nowhere; all’s you to make a future for yourself, there’s no rea-
gonna do is be back in the projects like your son why you can’t. It’s a question of attitude.
mother. Depend on your mother to give you
money every week. You ain’t gonna get a good SUPER: It’s easy to do anything, as long as you
job. As you get older, you’ll think about that, set your mind to it, if you wanna do it. If you
y’know. It’ll come to your mind. “Wow, I really want to do it, if you really want to be
can’t believe, I should’ve just went to school something. If you don’t want to do it . . . you
and got my education.” ain’t gonna make it. I gotta get that through
my mind: I wanna do it. I wanna be some-
(in a separate interview) thin’. I don’t wanna be livin’ in the projects
the rest of my life.
MOKEY: They all got attitude problems. They
just don’t got their shit together. Like Steve. MOKEY: It’s not like if they’re rich they get
They have to improve themselves. picked [for a job]; it’s just mattered by the
knowledge of their mind.
In the eyes of the Brothers, the Hallway Hang-
ers have attitude problems, are incapable of con- CRAIG: If you work hard, it’ll pay off in the
sidering their long-term future, and are lazy or end.
stupid.
Because this evidence is tainted (no love is MIKE: If you work hard, really put your mind
lost between the two peer groups), it is significant to it, you can do it. You can make it.
 MacLeod—Ain’t No Makin’ It • 617

This view of the opportunity structure as an es- does not give an adequate sense of the internal
sentially open one that rewards intelligence, ef- structure of the habitus, but there is some prece-
fort, and ingenuity is shared by all the Brothers. dent in his work for incorporating other factors
Asked whether their chances of securing a remu- into constructions of the habitus; for example,
nerative job are as good as those of an upper-class he differentiates people not only by gender and
boy from a wealthy district of the city, they all class, but also by whether they come from Paris
responded affirmatively. Not a single member or not. Although Bourdieu sometimes gives the
of the Hallway Hangers, in contrast, affirms the impression of a homogeneity of habitus within
openness of American society. . . . the boundaries of social class, I understand hab-
itus to be constituted at the level of the fam-
ily and thus can include, as constitutive of the
Reproduction Theory Reconsidered habitus, factors such as ethnicity, educational
This basic finding—that two substantially dif- histories, peer associations, and demographic
ferent paths are followed within the general characteristics (e.g., geographical mobility, dura-
framework of social reproduction—is a major tion of tenancy in public housing, sibling order,
challenge to economically determinist theories. and family size) as these shape individual action.
Two groups of boys from the same social stra- Although Bourdieu never really develops the
tum who live in the same housing project and notion along these lines, he does allude to the
attend the same school nevertheless experience complexity and interplay of mediations within
the process of social reproduction in fundamen- the habitus. “The habitus acquired in the family
tally different ways. This simple fact alone calls underlies the structuring of school experiences,
into question many of the theoretical formula- and the habitus transformed by schooling, itself
tions of Bowles and Gintis.6 If, as they argue, so- diversified, in turn underlies the structuring of
cial class is the overriding determinant in social all subsequent experiences (e.g. the reception
reproduction, what accounts for the variance in and assimilation of the messages of the culture
the process between the Brothers and Hallway industry or work experiences), and so on, from
Hangers? Bowles and Gintis, in considering a restructuring to restructuring.”9 When un-
single school, maintain that social reproduction derstood along the lines I have indicated, the
takes place primarily through educational track- concept of habitus becomes flexible enough to
ing. Differential socialization through educa- accommodate the interactions among ethnicity,
tional tracking prepares working-class students family, schooling, work experiences, and peer
for working-class jobs and middle-class students associations that have been documented [here].
for middle-class jobs. But the Hallway Hangers Although we may accept the notion of hab-
and the Brothers, who are from the same social itus as a useful explanatory tool, we must reject
class background and exposed to the curricu- the inevitability of its function in Bourdieu’s
lar structure of the school in the same manner, theoretical scheme. According to Bourdieu, the
undergo the process of social reproduction in habitus functions discreetly to integrate individ-
substantially different manners. The theory of uals into a social world geared to the interests
Bowles and Gintis cannot explain this difference. of the ruling classes; habitus engenders attitudes
Bourdieu’s notion of habitus, however, can be and conduct that are compatible with the re-
used to differentiate the Hallway Hangers and production of class inequality. The outstanding
the Brothers.7 The habitus, as defined by Gir- example of this process is the development by
oux, is “the subjective dispositions which reflect working-class individuals of depressed aspira-
a class-based social grammar of taste, knowledge, tions that mirror their actual chances for social
and behavior inscribed in . . . each developing advancement.
person.”8 According to Bourdieu, the habitus The circular relationship Bourdieu posits
is primarily a function of social class. Bourdieu between objective opportunities and subjective
618 • Part VI: Who Gets Ahead?

hopes is incompatible with the findings [pre- incapable of accounting for much of what we
sented here]. The Brothers, whose objective life see. The lives of the Hallway Hangers and the
chances probably were lower originally than Brothers cannot be reduced to structural influ-
those available to the Hallway Hangers because ences or causes; although structural forces weigh
of racial barriers to success, nevertheless nurture upon the individuals involved, it is necessary,
higher aspirations than do the Hallway Hang- in the words of Willis, “to give the social agents
ers. By emphasizing structural determinants at involved some meaningful scope for viewing,
the expense of mediating factors that influence inhabiting, and constructing their own world in
subjective renderings of objective probabilities, a way which is recognizably human and not the-
Bourdieu presumes too mechanistic and sim- oretically reductive.”11 We must appreciate both
plistic a relationship between aspiration and op- the importance and the relative autonomy of the
portunity. This component of his theory fails to cultural level at which individuals, alone or in
fathom how a number of factors lie between and concert with others, wrest meaning out of the
mediate the influence of social class on individ- flux of their lives.
uals; Bourdieu cannot explain, for instance, how The possibilities open to these boys as
ethnicity intervenes in the process of aspiration lower-class teenagers are limited structurally
formation and social reproduction. from the outset. That they internalize the ob-
Thus, the theoretical formulations of Bowles jective probabilities for social advancement to
and Gintis and the deterministic elements of some degree is beyond question. The process by
Bourdieu’s theory, although elegant and intu- which this takes place, however, is influenced by
itively plausible, are incapable of accounting a whole series of intermediate factors. Because
for the processes of social reproduction as they gender is constant in the study discussed in these
have been observed and documented in Clar- pages, race is the principal variable affecting the
endon Heights. These theories give an excellent way in which these youths view their situation.
account of the hidden structural and ideologi- Ethnicity introduces new structurally deter-
cal determinants that constrain members of the mined constraints on social mobility, but it also
working class and limit the options of Clarendon serves as a mediation through which the limita-
Heights teenagers. What the Hallway Hangers tions of class are refracted and thus apprehended
and the Brothers demonstrate quite clearly, how- and understood differently by different racial
ever, is that the way in which individuals and groups. The Brothers comprehend and react
groups respond to structures of domination is to their situation in a manner entirely different
open-ended. Although there is no way to avoid from the response the Hallway Hangers make to
class-based constraints, the outcomes are not a similar situation; ethnicity introduces a new
predefined. Bowles and Gintis and Bourdieu dynamic that makes the Brothers more receptive
pay too little attention to the active, creative to the achievement ideology. Their acceptance
role of individual and group praxis. As Giroux of this ideology affects their aspirations but also
maintains, what is missing from such theories influences, in tandem with parental encourage-
“is not only the issue of resistance, but also any ment, their approach to school and the charac-
attempt to delineate the complex ways in which ter of their peer group, factors that in turn bear
working-class subjectivities are constituted.”10 upon their aspirations.
If we modify the habitus by changing the eth-
nicity variable and altering a few details of family
From Ethnography to Theory occupational and educational histories and du-
Once we descend into the world of actual human ration of tenancy in public housing, we would
lives, we must take our theoretical bearings to have the Hallway Hangers. As white lower-class
make some sense of the social landscape, but in youths, the Hallway Hangers view and inter-
doing so we invariably find that the theories are pret their situation in a different light, one that
 MacLeod—Ain’t No Makin’ It • 619

induces them to reject the achievement ideol- labor jobs. Low aspirations, in turn, make the
ogy and to develop aspirations and expectations Hallway Hangers more likely to dismiss school
quite apart from those the ideology attempts as irrelevant. Once on the job market, the Hall-
to generate. The resultant perspective, which is way Hangers’ inability to secure even mediocre
eventually reinforced by the Hallway Hangers’ jobs further dampens their occupational hopes.
contact with the job market, informs the boys’ Thus although each individual ultimately retains
approach to school and helps us understand the autonomy in the subjective interpretation of his
distinctive attributes of this peer group. Thus, situation, the leveled aspirations of the Hallway
although social class is of primary importance, Hangers are to a large degree a response to the
there are intermediate factors at work that, as limitations of social class as they are manifest in
constitutive of the habitus, shape the subjective the Hallway Hangers’ social world.
responses of the two groups of boys and produce The Brothers’ social class origins are only
quite different expectations and actions. marginally different from those of the Hallway
Having grown up in an environment where Hangers. Being black, the Brothers also must
success is not common, the Hallway Hang- cope with racially rooted barriers to success that,
ers see that the connection between effort and affirmative action measures notwithstanding,
reward is not as clearcut as the achievement structurally inhibit the probabilities for social
ideology would have them believe. Because it advancement, although to a lesser degree than
runs counter to the evidence in their lives and do shared class limitations. What appears to be
because it represents a forceful assault on their a comparable objective situation to that of the
self-esteem, the Hallway Hangers repudiate the Hallway Hangers, however, is apprehended in a
achievement ideology. Given that their parents very different manner by the Brothers.
are inclined to see the ideology in the same light, As black teenagers, the Brothers interpret
they do not counter their sons’ rejection of the their families’ occupational and educational
American Dream. records in a much different light than do the
A number of important ramifications follow Hallway Hangers. Judging by the Brothers’ con-
from the Hallway Hangers’ denial of the domi- stant affirmation of equality of opportunity, the
nant ideology: the establishment of a peer group boys believe that racial injustice has been curbed
that provides alternative means of generating in the United States in the last twenty years.
self-esteem, the rejection of school and antag- Whereas in their parents’ time the link between
onism toward teachers, and, of course, the lev- effort and reward was very tenuous for blacks,
eling of aspirations. In schematizing the role of the Brothers, in keeping with the achievement
the peer group, it is difficult not to appear tauto- ideology, see the connection today as very
logical, for the group does wield a reciprocal in- strong: “If you work hard, it’ll pay off in the
fluence on the boys: It attracts those who are apt end” (Craig). Hence, the achievement ideology
to reject school and the achievement ideology is more compatible with the Brothers’ attitudes
and those with low aspirations and then deepens than with those of the Hallway Hangers, for
these individuals’ initial proclivities and further whom it cannot succeed against overwhelming
shapes them to fit the group. But at the same contrary evidence. The ideology is not as emo-
time, the peer subculture itself, handed down tionally painful for the Brothers to accept be-
from older to younger boys, is the product of cause past racial discrimination can help account
the particular factors that structure the lives of for their families’ poverty, whereas the Hallway
white teenagers in Clarendon Heights. Hangers, if the ideology stands, are afforded no
In addition to the peer group, the curricular explanation outside of laziness and stupidity for
structure of the school solidifies the low aspira- their parents’ failures. The optimism that accep-
tions of the Hallway Hangers by channeling them tance of the achievement ideology brings for the
into programs that prepare students for manual Brothers is encouraged and reinforced by their
620 • Part VI: Who Gets Ahead?

parents. Thus, we see how in the modified hab- site that does not demean the Hallway Hangers
itus ethnicity affects the Brothers’ interpretation or drain their dignity. As workers in the under-
of their social circumstances and leads to accep- ground economy, they won’t have to take orders
tance of the achievement ideology, with all the from a boss’s arrogant son, nor will they have to
concomitant results. gossip with office colleagues and strain to cam-
ouflage their street identities. . . .
Although they have certainly fared better
Postscript: The Hallway Hangers and than the Hallway Hangers, the Brothers have
Brothers Eight Years Later themselves stumbled economically in the transi-
“Hey, Jay, what the fuck brings you back to the tion to adulthood. Even more so than the Hall-
Ponderosa?” Greeted by Steve in July 1991, I way Hangers, the Brothers have been employed
surveyed a Clarendon Heights that had changed in the service sector of the economy. They have
considerably since 1983. Steve jerked his thumb bagged groceries, stocked shelves, flipped ham-
over his shoulder at a group of African Ameri- burgers, delivered pizzas, repaired cars, serviced
can teenagers lounging in the area outside door- airplanes, cleaned buildings, moved furniture,
way #13, previously the preserve of the Hallway driven tow trucks, pumped gas, delivered auto
Hangers. “How do you like all the new niggers parts, and washed dishes. They have also worked
we got here? Motherfuckers’ve taken over, man.” as mail carriers, cooks, clerks, computer opera-
I asked Steve about Frankie, Slick, and the other tors, bank tellers, busboys, models, office pho-
Hallway Hangers. “I’m the only one holding tocopiers, laborers, soldiers, baggage handlers,
down the fort,” he answered. “Me and Jinks— security guards, and customer service agents.
he lives in the back. The rest of ’em pretty much Only Mike, as a postal service employee, holds a
cut loose, man.” unionized position. Although their experiences
In their mid-twenties, the seven Hallway on the labor market have been varied, many
Hangers should be in the labor force full time. of the Brothers have failed to move out of the
Most of them aren’t: They are unemployed or secondary labor market. Instead, like the Hall-
imprisoned, or are working sporadically either way Hangers, they have been stuck in low-wage,
for firms “under the table” or for themselves high-turnover jobs. . . .
in the drug economy. . . . The Hallway Hang- These results are depressing. The experiences
ers have been trapped in what economists call of the Hallway Hangers since 1984 show that
the secondary labor market—the subordinate opting out of the contest—neither playing the
segment of the job structure where the market game nor accepting its rules—is not a viable op-
is severely skewed against workers. Jobs in the tion. Incarceration and other less explicit social
primary labor markets provide wages that can penalties are applied by society when the contest
support families and an internal career structure, is taken on one’s own terms. There is no escape:
but the rules of the game are different in the sec- The Hallway Hangers must still generate in-
ondary labor market. Wages are lower, raises are come, build relationships, and establish house-
infrequent, training is minimal, advancement is holds. Trapped inside the game, the Hallway
rare, and turnover is high. Hangers now question their youthful resistance
When the legitimate job market fails them, to schooling and social norms. Granted the op-
the Hallway Hangers can turn to the under- portunity to do it over again, the Hallway Hang-
ground economy. Since 1984 almost all of the ers say they would have tried harder to succeed.
Hallway Hangers have at least supplemented But the Brothers have always tried, which is
their income from earnings in the burgeoning, why their experiences between 1984 and 1991
multibillion-dollar drug market. The street are as disheartening as the Hallway Hangers’. If
economy promises better money than does con- the Hangers show that opting out of the contest
ventional employment. It also provides a work is not a viable option, the Brothers show that
 Morgan—A New Social Psychological Model of Educational Attainment • 621

dutifully playing by the rules hardly guarantees 2. Erik H. Erikson, Gandhi’s Truth (New York:
success either. Conservative and liberal com- Norton, 1969), p. 154.
mentators alike often contend that if the poor 3. Ronald Reagan, “State of the Union Address to
would only apply themselves, behave responsi- Congress,” New York Times, 6 February 1985, p. 17.
4. Kenneth I. Spenner and David L. Featherman,
bly, and adopt bourgeois values, then they will
“Achievement Ambitions,” Annual Review of Sociol-
propel themselves into the middle class. The ogy 4 (1978):376–378.
Brothers followed the recipe quite closely, but 5. Howard B. London, The Culture of a Commu-
the outcomes are disappointing. They illustrate nity College (New York: Praeger, 1978).
how rigid and durable the class structure is. As- 6. Samuel Bowles and Herbert Gintis, Schooling
piration, application, and intelligence often fail in Capitalist America (New York: Basic Books, 1976).
to cut through the firm figurations of structural 7. See Pierre Bourdieu, Outline of a Theory of
inequality. Though not impenetrable, structural Practice (Cambridge: Cambridge University Press,
constraints on opportunity, embedded in both 1977).
8. Henry A. Giroux, Theory & Resistance in Ed-
schools and job markets, turn out to be much
ucation (London: Heinemann Educational Books,
more debilitating than the Brothers anticipated. 1983), p. 89.
Their dreams of comfortable suburban bliss cur- 9. Bourdieu, Outline of a Theory of Practice, p. 87.
rently are dreams deferred, and are likely to end 10. Giroux, Theory & Resistance, p. 85.
up as dreams denied. 11. Paul E. Willis, Learning to Labor (Aldershot:
Gower, 1977), p. 172.
NOTES
1. All names of neighborhoods and individuals
have been changed to protect the anonymity of the
study’s subjects.

69 • Stephen L. Morgan

A New Social Psychological Model of

Educational Attainment

The decision of whether or not to enter college is Consider two students, Max and Vinny, who
vexing for a substantial proportion of high school have been raised in similar families. Their par-
students. The alternative choices are clearly in ents have identical educational and occupational
view from early adolescence onward, but there profiles and value hard work to the same degree.
is no simple normative guide for behavior. As By the end of middle school, Max and Vinny
a result, everyday commitment to preparation have equivalent academic records, and they both
for college instruction varies across high school have access to equally strong opportunities in
students, affecting resulting enrollment decisions their respective high schools. In fact, at this point
and the potential gains from college instruction. in their lives, Max and Vinny differ substantially

This original article first appeared in Social Stratification: Class, Race, and Gender in Sociological Perspective,
Third Edition, edited by David B. Grusky, pp. 542–549. Copyright © 2008 by Westview Press.
622 • Part VI: Who Gets Ahead?

in only one way, the amount of information that transmit accurate information about a student’s
is available in their communities about the like- likely success in the educational system? Or do
lihood of graduating from college if enrolled and they systematically mislead students, so that
the typical life outcomes that college graduates some students who could succeed in college are
and college dropouts receive. discouraged while others who would likely fail
In particular, Max attends a high school in are encouraged?
which the majority of students carry on to col- The stochastic decision tree model of com-
lege after high school graduation, and about half mitment that I develop here is designed to en-
of whom then graduate from college. In contrast, able the explicit modeling of questions such as
Vinny attends a high school in which very few these, based on a belief formation framework
students per year enter college after high school that gives separable influence to the accuracy
graduation, even though they are also about and amount of information available to stu-
equally likely to graduate from college after en- dents. Before laying out the primary features of
rolling. As a result of this dearth of information, the model, I first adopt a specific conception of
Vinny sometimes wonders whether a college commitment that generates everyday behavioral
education is worth the investment of time and preparation. I then introduce a standard deci-
money. And, in fact, every once in a while when sion tree model of educational choice, which is
thinking through the attractiveness of college, he then modified for use in the stochastic decision
comes to the conclusion that a college education tree model of commitment, as justified by the
may not be worthwhile. In these moments, his literature on cognitive constraints and bounded
attention to schoolwork wanes. Max, in contrast, rationality.
remains diligent all the while. The abundant in-
formation that is available to him keeps him fo-
cused on ensuring that he is adequately prepared Commitment and Preparation
for the next phase of his educational career. The stochastic decision tree model of commit-
Now suppose that, in addition to differences ment is grounded on the specific assumption
in the amount of available information, Max’s that in the period leading up to the college entry
parents have college degrees while Vinny’s par- decision:
ents have only high school degrees. In this case,
status socialization theory (see Sewell, Haller, A1. Everyday courses of behavior are
and Portes 1969) predicts that Max will have self-regulated by commitments toward alter-
more ambitious college plans because, in gen- native future courses of behavior.
eral, students’ own expectations of future be-
havior are shaped by the expectations that their For example, the strength of a high school
parents, teachers, and peers have of them. Even student’s commitment to the future course of
if Max and Vinny receive the same amount of behavior “Go to college immediately following
encouragement from their parents, it is likely high school” is the ease with which he or she is
that Max’s teachers, who know his parents from able to envision entering and ultimately gradu-
parent-teacher conferences, are more likely to ating from a college degree program.
communicate to Max that they expect he will For modeling educational attainment, there
attend college just like his parents did. are three types of commitment toward alterna-
Stepping back from the case of Max and tive futures: purposive, normative, and imitative.
Vinny, how important are differences in the For the nonrepeatable decision “Go to college”
amount of available information, on which stu- versus “Do not go to college,” these three types
dents form their own expectations of future be- of commitment are set in response to three
havior, relative to social influence processes that forward-looking prediction rules: “I will go to
shape expectations as well? And how do teach- college if I perceive it to be in my best interest
ers and peers form their expectations? Do they to do so,” “I will go to college if my significant
 Morgan—A New Social Psychological Model of Educational Attainment • 623

others perceive it to be in my best interest to do their range of realistic college alternatives. And,
so,” and “I will go to college if I expect other because the payoff to college instruction is a
students similar to me will also go to college.” function of the ability to benefit from opportu-
These three generative dimensions are roughly nities for learning, commitment and preparation
analogous to the self-reflection, adoption, and determine final levels of well-being, above and
imitation mechanisms of status socialization beyond simply pushing students over a decision
theory (see Morgan 1998). threshold to enroll in college.
Although all three dimensions of commit-
ment are important, purposive commitment has
the most power to subsume the other two. I will A Decision Tree Model of
therefore focus development of the stochastic Educational Choice
decision tree model here on the construction Figure 69.1 presents a standard decision tree
of purposive commitment (see Morgan 2005, for choice under uncertainty. This specific tree
Chapters 5 and 6, for proposals for how to then is isomorphic with the set of equations used to
introduce reinforcing and destabilizing norma- model students’ enrollment decisions by Man-
tive and imitative sources of commitment into ski (1989). For this model, a prospective college
the model). Thus, I motivate the development student must contemplate whether it is in their
of the stochastic decision tree model by assum- best interest to choose the course of action “Go
ing that: to college” or “Do not go to college.”
Assume that students consider two possible
A2. Individuals use a decision tree to identify abstract life outcomes—a very good position in
and commit to the future course of behavior life denoted by “High” and a not very good po-
that they believe is in their best interest. sition in life denoted by “Low.” High school stu-
dents must decide which among three alternative
Having defined commitment as a cog- Figure 69.1. A Simple Decision Tree for the College Entry
nitive attachment to a future course of Decision
behavior, the course of everyday behavior
that positions an individual to realize their
commitment can then be labeled prepara-
tory commitment, or even more simply as
just preparation. I assume that:

A3. An individual’s level of preparation for

a future course of behavior is a direct func-
tion of the strength of their commitment
to that course of behavior.

Accordingly, a student with maximal

commitment toward the future course of
action “I will enroll in college” will enact
all possible behavior that prepares the stu-
dent for enrolling in college and then suc-
cessfully obtaining a college degree. While
in high school, such students will take
college preparatory classes, complete their
homework diligently, focus their attention
when taking tests, sign up early for college
entrance examinations, and investigate
624 • Part VI: Who Gets Ahead?

paths through the educational system will put The left-hand sides of Equations (1a) and
them in the best possible position to obtain the (1b) represent the expected utility of enrolling
life position High rather than Low: (1) Go to in college, and the right-hand sides represent
college and graduate with a degree, (2) Go to the expected utility of not enrolling in college.
college but fail to graduate, (3) Forego college If the expected utility of enrolling is greater than
and enter the labor market immediately. For ra- the expected utility of not enrolling, rational
tional choice models of educational attainment, choice theory assumes that a high school student
students are presumed to think of these paths as will choose to enroll in college.
alternative lotteries that are controlled by success After students choose either the upper or
parameters π, Į, ȕ, and Ȗ that are probabilities lower branch of the tree—and thus either enroll
between 0 and 1. in college or enter the labor force—they are sub-
In particular, each student’s decision tree has ject to a set of real-life lotteries. The probability
a parameter π that represents each student’s sub- of reaching High instead of Low is a function of
jective belief about the probability of completing more than simply a student’s decision. Each lot-
college if initially enrolled. Similarly, students tery has a “true” probability, denoted respectively
maintain a set of beliefs about the relative like- as π̃, Į̃, ȕ̃, and Ȗ̃. Actual outcomes are therefore
lihood of attaining High versus Low after tra- determined in large part by exogenous factors
versing each of the three possible paths through that structure these probabilities. For research-
the educational system. They expect that if they ers, claims about the relative sizes of π̃, Į̃, ȕ̃, and
complete college, they will attain High with Ȗ̃ are usually based on assumptions about institu-
probability Į and Low with probability (1–Į). If tional constraints on educational attainment and
they attempt but do not complete college, then the distribution of High versus Low outcomes
they expect that they will attain High with prob- in society. It is generally assumed that there are
ability ȕ and Low with probability (1–ȕ). And positive returns to schooling so that Į̃ ≥ ȕ̃ and
finally, if they choose to forego college, they ex- ȕ̃ > Ȗ̃. However, Manski (1989:307) argues that
pect that they will attain High with probability “analysis is trivial” unless, in the notation used
Ȗ and Low with probability (1–Ȗ). here, for at least some students ȕ̃ < Ȗ̃ and ȕ̃ < Ȗ̃. If
Rational choice theory offers the prediction that were not the case, all students would choose
that, under specific assumptions about optimi- to enter college.
zation and consistency of choice, prospective When contemplating college entry decisions,
college students will choose to go to college if: do students maintain reasonable beliefs about
the lotteries that they face? In other words, for
ʌĮ[u(High1)–u(Low1)]+(1–ʌ)ȕ[u(High2)– each individual and for a decision tree such as
u(Low2)] > Ȗ[u(High3)–u(Low3)] the one depicted in Figure 69.1, is there a close
(1a) correspondence between π, Į, ȕ, and Ȗ and π̃,
Į̃, ȕ̃, and Ȗ̃? The traditional assumption among
where u(.) is a utility function that, in its most rational choice researchers is that individuals, on
general sense, assigns subjective value to the average, have correct beliefs about π̃, Į̃, ȕ̃, and
alternative payoffs High and Low. If an indi- Ȗ̃. Accordingly, the values of π, Į, ȕ, and Ȗ that
vidual’s utility function is process-independent students rely on when making college entry deci-
so that u(High1) = u(High2) = u(High3) and sions are assumed to equal the true probabilities
u(Low1) = u(Low2) = u(Low3), then both sides that students will face in their futures.
of Equation (1a) can be divided by a common The model I develop in the next section pro-
utility difference, u(High)–u(Low), in order to vides a foundation for a more comprehensive ap-
obtain the simplified decision rule: proach. The model is based in part on rational
choice theory but also embraces the core claim
ʌĮ+(1–ʌ)ȕ > Ȗ (1b) of the Wisconsin model of Sewell, Haller, and
 Morgan—A New Social Psychological Model of Educational Attainment • 625

Portes (1969) that one’s current behavior is con- tree of this form would be justified if beliefs such
ditioned by socially structured beliefs about one’s as p or the utility evaluations of High and Low
future. depend on whether or not one enrolls in the trig-
onometry class.
Even though some very consequential
A Stochastic Decision Tree Model of short-run decisions may be approached in this
Commitment sophisticated way, I presume that students have
The action “Go to college” can be thought of as no desire to use a finely specified decision tree
a compound outcome of a series of underlying for every consequential underlying decision they
decisions, many of which must be enacted long must make, such as whether to study hard for an
before the first college tuition bill is due. More- upcoming test or whether to do their homework
over, the payoff to obtaining a college degree is on any given night. Were this the case, the large
a function not just of having enrolled in college number of preparatory commitment decisions
but of how seriously one has prepared to master that adolescents must navigate in middle school
the college curriculum before being exposed to it. and high school would render any such compre-
hensive decision trees massive and ever-changing
as each intermediate decision is considered and
Decision Tree Structure then enacted.
For the college entry decision, I assume that:

A4. The decision tree that individuals use to Belief Distributions

commit to a future course of behavior is more Instead of requiring high school students to con-
simple in structure than that which would be struct comprehensive decision trees for all inter-
required to explicitly model all underlying de- mediate decisions, I allow the complexity of the
cisions that are consequential for the college everyday decisions that constitute preparation to
entry decision. be modeled as uncertainty about the parameters
of a simple decision tree for a future course of
Equivalently, I assume that preparation can- behavior. Thus, I assume that:
not be modeled as a long series of fully conscious,
forward-looking decisions. For the numerical A5. The parameters of the decision tree that
demonstration of the framework that I offer in students use to commit to a future course of
Morgan (2005), I use the simple decision tree behavior are fundamentally stochastic.
in Figure 69.1, and hence, when referring to
the college choice example, adopt the position This assumption is the major departure from
of Manski (1989) that for enrollment decisions standard practice in rational choice modeling
potential enrollees consider four basic lotteries. of educational attainment. For the stochastic
Trees with alternative and slightly more elaborate decision tree model of commitment, students
structures are permissible, but I stipulate with recognize that the decision tree they are consult-
Assumption A4 that these must remain simple. ing is simply a rough approximation to the com-
For example, if a student were contemplating plex set of real-life lotteries to which they will
a high school course enrollment decision, the ultimately be subjected. Thus, students forecast
decision tree in Figure 69.1 could be expanded their futures as if any ostensibly true parameters
to model this decision explicitly. In this case, the such as π̃, Į̃, ȕ̃, and Ȗ̃ have distributions. Accord-
decision tree in Figure 69.1 could be specified as ingly, Assumption A5 stipulates that students
two alternative subtrees for the pressing current maintain entire belief distributions for the pa-
decision: “Enroll in trigonometry” or “Do not rameters of a decision tree, such as π, Į, ȕ, and
enroll in trigonometry.” A more comprehensive Ȗ in Figure 69.1.
626 • Part VI: Who Gets Ahead?

How are belief distributions constructed cognitive constraints. Accordingly, they cannot
and revised? I address this question in substan- numerically evaluate density functions in order
tial detail in Morgan (2005), but here I simply to quickly recover arbitrarily exact estimates of
assume for brevity that individuals store belief the means of their belief distributions. Rather,
distributions in memory based on observations individuals must rely on “fast and frugal” strate-
of individuals whom they regard as payoff mod- gies to evaluate their decision trees (see Gigeren-
els. Furthermore, I assume that individuals with zer and Selten 2001). Based on this position, I
sparse information at their disposal will have assume that:
uncertain belief distributions with large vari-
ances, and individuals with abundant informa- A6. Students evaluate their stochastic decision
tion will have precise belief distributions with trees by solving a decision rule with parameter
small variances. For example, consider two high values that are simple averages of a few ran-
school students contemplating the decision tree domly sampled candidate values drawn from
presented in Figure 69.1 when forecasting their their belief distributions.
own future decision of whether to enter college
immediately. Student A observes twenty college In the numerical simulations of the frame-
graduates and ten college dropouts. In contrast, work offered in Morgan (2005), I allow students
Student B observes only two college graduates to randomly draw only a few candidate parame-
and one college dropout. The framework I will ter values for a decision tree of the structure pre-
adopt stipulates that Student A will have more sented in Figure 69.1 and then average over these
confidence than Student B in eliminating ex- values to form a set of candidate point-value be-
treme values from his or her belief distributions liefs {π´, Į´, ȕ´, Ȗ´}. Assuming that utility eval-
about the subjective probability of graduating uations are equivalent across paths and scaled to
from college if enrolled. Student A would con- values of 1 for High and 0 for Low, an individual
sider values such as .66 much more likely than will set a dichotomous variable E equal to 1 if:
values such as .96 or .36. Student B would like-
wise consider .66 more likely than either .96 or ʌ´Į´+(1–ʌ´)ȕ´> Ȗ ´ (1c)
.36 but would be less willing than Student A
to discount the plausibility of .96 or .36. This Otherwise, E will be set equal to 0. E signifies
position is consistent with the dominant litera- an instantaneous commitment to the future af-
ture on belief distributions, which for Bayesian firmative decision “Go to college” (for example,
decision theory (see Pratt, Raiffa, and Schlaifer the choice Go in Figure 69.1).
1995) and information-integration theory in Over a set of J evaluations of the same sto-
psychology (see Davidson 1995) similarly allows chastic decision tree with stable belief distribu-
uncertainty and imprecision of beliefs to be a tions, the average level of commitment to future
function of the amount of available information. college attendance is then:

Pr(E=1)= 1– E (2)
J j=1...J 1
Decision Tree Evaluation
How do students evaluate their stochastic de-
cision trees? By principles of statistical decision Pr(E=1) is therefore equal in expectation to
theory, the traditional answer would be that they the probability that E will be evaluated as 1 for
calculate the means of their belief distributions, a given set of beliefs and prespecified total num-
declare them as their best estimates of their ber of candidate parameter draws selected for
point-value beliefs, and use them to solve a deci- decision tree evaluation. Thus, whereas the bi-
sion rule such as Equation (1a). Instead, I accept nary orientation E represents the instantaneous
the position from the bounded rationality liter- commitment that guides each preparation de-
ature that individuals are subject to substantial cision, Pr(E = 1) represents the average level of
 Morgan—A New Social Psychological Model of Educational Attainment • 627

commitment that is consistent with a set of stable educational attainment. The key innovation of
beliefs and a procedure to analyze those beliefs the framework is the specification of a simple
in the process of constructing forward-looking decision tree with fundamentally stochastic pa-
commitment. The distinction implies that for rameters. When a simple, boundedly rational
two students who at a specific point in time both decision rule is invoked to specify the way in
set E equal to 1, the student whose beliefs imply which individuals draw forecasts of future be-
a lower value for Pr(E = 1) will be less likely to havior from their stochastic decision trees, com-
set E equal to 1 when contemplating subsequent mitment and preparation can be shown to be
preparation decisions. Pr(E = 1) is therefore the functions of the accuracy and precision of beliefs
more fundamental quantity of interest. about future courses of behavior. If, as assumed
Why would students use this form of deci- by status socialization theory, current behavior
sion evaluation? I assume that students gener- is a function of beliefs about the future, then
ally wish to orient their current behavior to a everyday forecasts of future behavior operate as
long-run plan that is in their best interest, but self-fulfilling prophecies by regulating prepara-
doing so in a way that is as rigorous as stipulated tion for the course of behavior that is thought to
by statistical decision theory is too costly, espe- be in one’s best interest.
cially since students themselves must recognize The conception of educational attainment
that they cannot form comprehensive decision that I have developed here can be read as a stra-
trees to capture all consequential intermediate tegic evasion of the question that is the title of
decisions. Thus, students seek to avoid short-run Diego Gambetta’s 1987 book, Were They Pushed
mistakes in judgment by adopting a relatively or Did They Jump? Individual Decision Mecha-
frugal process of planning for their futures. nisms in Education. From the perspective devel-
The simulations and the analytic results in oped here, students wander through a series of
Chapter 4 of Morgan (2005) demonstrate that daily decisions, adjusting current behavior to
when the uncertainty of beliefs is allowed to ex- anticipation of future behavior. If belief forma-
plicitly enter a decision evaluation, preparation tion has its own inertia, is easily contaminated
for a future course of behavior such as college through social influence, and regulates current
entry is sensitive to the amount and type of behavior, then it would seem best not to conceive
uncertainty in one’s beliefs as well as the effort of students as if they are standing on the edge of
expended to analyze those beliefs. Thus, even a cliff, contemplating whether to jump. It should
in a world where the returns on investments in suffice to see students as if they are on the edge
higher education are massive, if beliefs are the of commitment to alternative futures. Although
least bit uncertain, and if students have limited less dramatic, this metaphor is sufficient to mo-
information processing and analysis capacities, tivate the new generation of theoretically guided
some high school students some of the time will empirical analysis that is now needed.
perceive it to be in their future best interest to
forego a college education. In these episodic and REFERENCES
contrarian instances, they will adjust their cur- Davidson, Andrew R. 1995. “From Attitudes to
Actions to Attitude Change: The Effects of
rent behavior accordingly, failing to enact prepa-
Amount and Accuracy of Information.” In Atti-
ration decisions that their beliefs nonetheless
tude Strength: Antecedents and Consequences, ed-
imply are genuinely in their best interest. ited by R. E. Petty and J. A. Krosnick. Mahwah,
NJ: Erlbaum.
Gambetta, Diego. 1987. Were They Pushed or Did
Conclusions They Jump? Individual Decision Mechanisms in Ed-
To construct the stochastic decision tree model ucation. Cambridge: Cambridge University Press.
of commitment, I have borrowed and then Gigerenzer, Gerd, and Reinhard Selten, eds. 2001.
extended the strongest pieces of both status Bounded Rationality: The Adaptive Toolbox. Cam-
socialization and rational choice models of bridge, MA: MIT Press.
628 • Part VI: Who Gets Ahead?

Manski, Charles F. 1989. “Schooling as Experi- Pratt, John W., Howard Raiffa, and Robert Schlaifer.
mentation: A Reappraisal of the Postsecondary 1995. Introduction to Statistical Decision Theory.
Dropout Phenomenon.” Economics of Education Cambridge, MA: MIT Press.
Review 8: 305–312. Sewell, William H., Archibald O. Haller, and Ale-
Morgan, Stephen L. 1998. “Adolescent Educational jandro Portes. 1969. “The Educational and Early
Expectations: Rationalized, Fantasized, or Both?” Occupational Attainment Process.” American So-
Rationality and Society 10: 131–162. ciological Review 34:89–92.
———. 2005. On the Edge of Commitment: Educa-
tional Attainment and Race in the United States.
Stanford, CA: Stanford University Press.
 Piore—The Dual Labor Market • 629

Labor Markets

70 • Michael J. Piore

The Dual Labor Market

Theory and Implications

The central tenet of [my] analysis is that the generally their inability to show up for work reg-
role of employment and of the disposition of ularly and on time. Secondary employers are far
manpower in perpetuating poverty can be best more tolerant of lateness and absenteeism, and
understood in terms of a dual labor market. many secondary jobs are of such short duration
One sector of that market, which I have termed that these do not matter. Work skills, which re-
elsewhere the primary market,1 offers jobs which ceive considerable emphasis in most discussions
possess several of the following traits: high of poverty and employment, do not appear a
wages, good working conditions, employment major barrier to primary employment (although,
stability and job security, equity and due process because regularity and punctuality are important
in the administration of work rules, and chances to successful learning in school and on the job,
for advancement. The secondary sector has jobs such behavioral traits tend to be highly cor-
that are decidedly less attractive, compared with related with skills).
those in the primary sector. They tend to involve Second, certain workers who possess the be-
low wages, poor working conditions, consider- havioral traits required to operate efficiently in
able variability in employment, harsh and often primary jobs are trapped in the secondary market
arbitrary discipline, and little opportunity to ad- because their superficial characteristics resemble
vance. The poor are confined to the secondary those of secondary workers. This identification
labor market. Eliminating poverty requires that occurs because employment decisions are gener-
they gain access to primary employment. ally made on the basis of a few readily (and hence
The factors that generate the dual market inexpensively) assessed traits like race, demeanor,
structure and confine the poor to the second- accent, educational attainment, test scores, and
ary sector are complex. With some injustice to the like. Such traits tend to be statistically cor-
that complexity, they may be summarized: First, related with job performance but not necessarily
the most important characteristic distinguishing (and probably not usually) causally related to it.
primary from secondary jobs appears to be the Hence, a number of candidates who are rejected
behavioral requirements they impose upon the because they possess the “wrong” traits are actu-
work force, particularly that of employment sta- ally qualified for the job. Exclusion on this basis
bility. Insofar as secondary workers are barred may be termed statistical discrimination. In addi-
from primary jobs by a real qualification, it is tion to statistical discrimination, workers are also

Michael J. Piore. “The Dual Labor Market: Theory and Implications,” in The State and the Poor, edited by Sam-
uel H. Beer and Richard E. Barringer, pp. 55–59. Copyright © 1970. Used by permission of Michael J. Piore.
630 • Part VI: Who Gets Ahead?

excluded from primary employment by discrim- work is performed. The investment necessary to
ination pure and simple. effect these changes acts to strengthen resistance
Discrimination of any kind enlarges the to antipoverty efforts.
labor force that is captive in the secondary sec- Fourth, the behavioral traits associated with
tor, and thus lowers the wages that secondary the secondary sector are reinforced by the pro-
employers must pay to fill their jobs. Such em- cess of working in secondary jobs and living
ployers thus have an economic stake in perpet- among others whose lifestyle is accommodated
uating discrimination. Since it limits the supply to that type of employment. Hence, even peo-
of labor in the primary sector and raises the ple initially forced into the secondary sector by
wages of workers who have access to jobs there, discrimination tend, over a period of time, to
primary workers also have a stake in discrimina- develop the traits predominant among second-
tion. Discrimination pure and simple is not gen- ary workers. Thus, a man who works in a world
erally of economic value to primary employers, where employment is intermittent and erratic
since it forces them to pay higher wages without tends to lose habits of regularity and punctual-
obtaining corresponding economic gains. In ity. Similarly, when reward and punishment in
statistical discrimination, however, the higher the work place are continually based upon per-
wages are compensated by the reduced cost of sonal relationships between worker and super-
screening job candidates, and here primary em- visor, workers forget how to operate within the
ployers share the interest of secondary employ- impersonal, institutional grievance procedures of
ers and primary workers in perpetuating such the primary sector. When such workers do gain
discrimination. access to primary jobs, they are frustrated by the
Third, the distinction between primary and system’s failure to respond on a personal basis
secondary jobs is not, apparently, technologically and by their own inability to make it respond on
determinate. A portion—perhaps a substantial an institutional basis.
proportion—of the work in the economy can Finally, among the poor, income sources other
be organized for either stable or unstable work- than employment, especially public assistance
ers. Work normally performed in the primary and illicit activity, tend to be more compatible
sector is sometimes shifted to the secondary with secondary than with primary employ-
sector through subcontracting, temporary help ment. The public assistance system discourages
services, recycling of new employees through full-time work and forces those on welfare either
probationary periods, and the like. Nor is the into jobs that are part-time or into jobs that pay
primary-secondary distinction necessarily asso- cash income which will not be reported to the
ciated with a given enterprise. Some enterprises, social worker or can be quickly dropped or de-
most of whose jobs constitute primary em- layed when the social worker discovers them or
ployment and are filled with stable, committed seems in danger of doing so. The relationship
workers, have subsections or departments with between social worker and client builds upon
inferior job opportunities accommodated to an the personal relationship that operates in the
unstable work force. Secondary employers gen- secondary sector, not on the institutional mech-
erally have a few primary jobs, and some have anisms that tend to operate in the primary sec-
a large number of them. Nonetheless, despite a tor. Illegitimate activity also tends to follow the
certain degree of elasticity in the distribution of intermittent work pattern prevalent in secondary
work between the primary and secondary sec- employment, and the attractions of such activity,
tions, shifts in the distribution generally involve as well as life patterns and role models it presents
changes in the techniques of production and to those not themselves involved but associating
management and in the institutional structure with people who are, foster behavioral traits an-
and procedures of the enterprises in which the tagonistic to primary employment.
 Piore—The Dual Labor Market • 631

The dual market interpretation of poverty one side; a program, for example, would then
has some central implications: the poor do par- recruit workers for training in skills that are little
ticipate in the economy; the manner of their utilized in either the secondary or the primary
participation, not the question of participation market. Alternatively, the new institutions may
as such, constitutes the manpower problem of be captured by the prevailing economic system
the poor; and their current mode of participa- and used to facilitate its operation; for example,
tion is ultimately a response to a series of pres- neighborhood employment offices may recruit
sures—economic, social, and technical—playing secondary workers for secondary jobs, and train-
upon individuals and labor market institutions. ing may be provided in primary employment
This suggests that a distinction can be drawn to workers who would have gotten it anyway
between policies that are designed to alleviate in establishments that would have financed it
the pressures which generate the dual market themselves. The central problem in the design
structure and those that attempt to attack the of direct approaches to manpower programs is to
problem directly by moving individuals from organize them in such a way that they can resist
secondary to primary employment. The latter this two-fold threat of rejection on the one hand
policies combat prevailing pressures but leave and capture on the other.
intact the forces that generate them. The thrust These conclusions follow directly from the
of [my] argument is that in concentrating upon dual market interpretation of the poverty prob-
training, counseling, and placement services for lem but they are not uniquely dependent upon
the poor, manpower policy has overemphasized it. The dual labor market is one of a class of the-
direct approaches, and that more weight should oretical constructs which views poverty in the
be placed upon policies which affect the envi- United States in terms of a dichotomy in the
ronment in which employment decisions are economic and social structures. Such a dichot-
made and the pressures which the environment omy is implicit in the concept of a “culture of
generates. Among such policies are antidiscrimi- poverty” and in the expression of public policy
nation policy, occupational licensing reform, and goals associated with poverty in terms of an in-
the structure of public assistance. come cutoff. Most such views of poverty enter-
Analysis of the dual labor market suggests a tain the idea that the dichotomy is a product of
further implication: because the “poor” do par- forces endogenous to the economy (or, more
ticipate in the economy, certain groups are in- broadly, the society as a whole). It follows that
terested in that participation and how it occurs. attempts to eliminate poverty will tend to run
Policies aimed at moving the poor out of the counter to the natural operation of the economy,
secondary market work against the interests of and that they will be resisted by existing institu-
these groups and therefore are in danger of being tions and are in danger of rejection. To say all
subverted by them. This danger is a major reason this is perhaps to say simply that if poverty were
for concentrating on indirect approaches that are easy to eliminate, it wouldn’t be around in the
not susceptible to the same kind of subversion; first place. But it does at least identify as a cer-
in fact, because such approaches alleviate the tain problem in the program design the task of
pressures generating the dual market structure, equipping the institution which works with the
they reduce the resistance to policies that move poor to withstand the rejection pressures.
directly against that structure. The dangers to What the dual labor market interpretation
which existing institutions subject programs de- implies that is not implicit in other dichoto-
signed to move the poor directly out of the sec- mous interpretations is that the poor are sepa-
ondary market are twofold. The new institutions rated from the nonpoor not only in the negative
created by these programs can be rejected by the sense of exclusion from activities and institutions
prevailing economic system and isolated off to to which the nonpoor have access, but also in
632 • Part VI: Who Gets Ahead?

the positive sense that they have economic value NOTES

where they are; that, in other words, there are 1. See Michael J. Piore, “On-The-Job Training
groups actively interested in the perpetuation of in the Dual Labor Market,” in Arnold Weber, et al.,
Public-Private Manpower Policies (Madison, Wisc.:
poverty. It is this interest that makes new institu-
Industrial Relations Research Association, 1969),
tions created to work with the poor in the labor
pp. 101–132.
market subject to threats of capture as well as of
rejection.

71 • Aage B. Sørensen and Arne L. Kalleberg

An Outline of a Theory of the Matching of

Persons to Jobs

Much recent research in sociology has focused processes for various population groups. Socio-
on labor market processes. These concerns in- logical research on attainment and mobility has
clude analysis of the processes that produce vari- not employed an explicitly stated conceptual ap-
ation in individual earnings by characteristics of paratus that informs the choice of variables and
people and their jobs; the analysis of career pat- the interpretation of parameters. Although there
terns and job mobility processes; and the analy- is a growing body of findings about the magni-
sis of employment and unemployment patterns tude of the influences of various variables on the
of various population groups. Sociologists share outcomes of labor market processes, particularly
many of these concerns with economists, and income attainment, there are few efforts by so-
there is much overlap in research topics among ciologists to identify the mechanisms that create
sociologists and economists. the influences of personal and job characteristics
Despite similarities in methodology and on income and earnings or on the other labor
research design, the research traditions in so- market outcomes.
ciology and economics have quite different intel- There is no need for sociologists to develop
lectual backgrounds. Most empirical research on a unique theory of labor market processes if
labor market processes in economics is guided by the neoclassical economic theory adequately
the dominant school of labor economics—the accounts for the findings of empirical research.
neoclassical theory of wage determination and With respect to a favorite variable of both
labor supply, with marginal productivity the- economists and sociologists—that is, educa-
ory accounting for the demand side and human tion—human capital theory does provide an
capital theory taking care of the supply side. In interpretation of results. However, the economic
contrast, sociological research on labor market theory does not provide a rationale for the so-
phenomena has its origin in research describing ciological concern for occupational attainment.
socioeconomic attainment and social mobility Job characteristics, including those presumably

Aage B. Sørensen and Arne L. Kalleberg. “An Outline of a Theory of the Matching of Persons to Jobs,” in
Sociological Perspectives on Labor Markets, edited by Ivar Berg, pp. 49–57, 65–69, 72–74. Copyright © 1981
by Academic Press. Reprinted with permission from Elsevier. All rights reserved.
 Sørensen and Kalleberg—An Outline of a Theory of the Matching of Persons to Jobs • 633

captured by the Socioeconomic Index (SEI) jobs (cf. Doeringer & Piore, 1971); or monop-
or prestige scores of occupations, play little or oly, competitive, and state economic sectors
no role in the orthodox economic theory. Still, (cf. Averitt, 1968; Bluestone, 1970; O’Connor,
occupational status accounts for a substantial 1973); or wage competition and job competi-
fraction of the explained variance in sociological tion sectors (Thurow, 1975); or internal and ex-
income attainment models. ternal markets (Doeringer & Piore, 1971; Kerr,
The amount of variance added to income 1954). These critiques all observe that jobs and
attainment models by occupation is not neces- job structures differ, contrary to the assumption
sarily a strong argument for replacing or supple- about the homogeneous nature of labor mar-
menting the economic theory. Sociologists have kets made by the economic theory. They stress
not been able to account for very much variance qualitative differences among jobs relevant for
in income attainment. Research informed by employment and earnings processes and claim
human capital theory (e.g., Mincer, 1974) has in to be able to account for the observations that
fact been able to do as well or better without in- deviate from the orthodox economic theory, as
cluding occupation. A measure of occupational well as to provide different explanations for labor
status must necessarily show some relation to market processes that also can be explained by
income, reflecting the between-occupation vari- the orthodox theory. An example of such an al-
ance in income that it captures. An observed ef- ternative explanation is Thurow’s (1975) inter-
fect of job characteristics on income or earnings pretation of the relationship between education
may be attributed to a misspecification of socio- and earnings.
logical models, both with respect to functional Most of the criticism comes from within eco-
form and omitted variables, and need not be nomics, though there are examples of research
considered a challenge to the economic theory. and conceptual elaboration by sociologists per-
There are, however, other reasons for criti- taining to the issues raised by the segmented
cally evaluating the neoclassical or orthodox eco- labor market theory (Sørensen, 1977; Spilerman,
nomic theory. The economic theory is powerful, 1977; Stolzenberg, 1975). The issues are clearly
and numerous predictions can be derived from relevant for sociological research, and more so
it regarding the earnings attainment process and since the alternatives to the neoclassical theory
other labor market processes, particularly labor provide a rationale for introducing job char-
supply. (A list of such predictions is presented acteristics sociologists are likely to continue to
by Becker, 1964.) Some of these predictions are emphasize.
borne out by empirical observations; some are The classical sociological theorists did not
not. Thurow (1975, pp. 56–70) presents a list leave labor market analysis to economists. Marx
of deviations from the theory, pertaining to such and Weber spent lifetimes analyzing the relation
issues as the relationship between wages and between economy and society, and their con-
unemployment, changes in the distribution of cerns in many ways parallel the issues raised in
earnings, and the relationship between the dis- recent controversies. Marx’s analysis of capitalist
tribution of education and the distribution of in- society is an analysis of the implications of the
come. Numerous others have identified features fundamental condition of capitalist production:
of the earnings attainment process and of labor Labor is treated as a commodity bought and sold
markets that deviate from the assumptions and freely in a market. This conception of the labor
predictions of the neoclassical theory. A review market, we shall argue in the following pages,
of these challenges to orthodox theory has been parallels the conception of the orthodox eco-
presented by Cain (1976). Particularly impor- nomic theory.
tant are those critiques that argue that labor mar- Marx treated labor in capitalist society as a
kets are segmented and that stress the differences homogeneous abstract category, and though
between either so-called primary and secondary there are occasional remarks concerning
634 • Part VI: Who Gets Ahead?

deviations from this model of labor as a com- therefore, refer to the employment relationship
modity and their relevance for class conflict (e.g., as the crucial determinant of the notion of the
Marx, 1961, Vol. 1, chap. 14), no systematic matching process and its earnings outcome.
analysis of alternative labor market structures is Employment relationships are social relation-
presented. Weber’s long analysis of the sociolog- ships created in the production of goods and
ical categories of economic action (Weber, 1947, services between an employer (or his agent) and
Pt. I, chap. 2) provides, in contrast, numerous an employee. We concentrate on employment
concepts relevant for the analysis of labor market relationships typical of capitalist production in
structures (including nonmarket relationships), which the employer appropriates the output
particularly in the sections on the social division from the production process and has complete
of labor. The concepts are highly relevant for the possession over the nonhuman means of pro-
issues raised by the challenges to orthodox eco- duction. Our analysis will focus on the conse-
nomic theory, and some of Weber’s basic con- quences for the earnings determination process
cepts will be used extensively in this chapter. and other labor market processes of variation in
The following pages provide a conceptual control over the job by the employer versus the
framework for the analysis of labor markets. employee. Two aspects of control over the job
Labor markets are arenas for the matching of may be distinguished. One is control over the
persons to jobs. The conditions that determine activities of the job, resulting in more or less au-
the earnings outcome of this matching process tonomy for the employee; the other is control
are of primary interest here, particularly the over access to the job, resulting in a more or less
identification of what determines the influence closed employment relationship. These two di-
of job and personal characteristics on earnings. mensions may vary independently. Particularly,
The purpose of this chapter is not to show the control over access to the job will be considered
neoclassical theory to be wrong, but rather to crucial, because it influences the nature of com-
identify the conditions for the emergence of the petition among employees.
matching process associated with the labor mar- The degree of control over access is a con-
ket structure assumed in the orthodox economic tinuum. At one extreme, the employee “owns”
theory. It will be argued that the conditions for the job and no one else can get access unless
the emergence of this matching process are not the current incumbent voluntarily leaves it and
present in some segments of the labor market. a vacancy is established. The length of the em-
The absence of these conditions leads to alter- ployment is then completely controlled by the
native matching processes, and a model of one employee, and the employment relationship is
important alternative matching process will be closed to outsiders. At the other extreme, the
presented. The two contrasting matching pro- employer may replace the incumbent at any
cesses will be shown to have very different impli- time. The employment contract is reestab-
cations for the earnings determination process lished in every short interval of time, and the
and for other labor market processes. employment relationship is completely open to
outsiders.
The employment relationship is established
Basic Concepts in a process assumed to involve purposive actors
The theory proposed in this chapter will rely as employers and employees where both parties
on Weber’s notion of open and closed social are attempting to maximize earnings. The earn-
relationships (Weber, 1947, p. 139) to identify ings of the employer are determined by the value
different job structures characterized by differ- of the product of the job-person combination
ent matching processes.1 The degree of closure, in relation to costs of production. The value of
in turn, is seen as determined by the bargaining production is a question of prices of products
power of employers and employees. We shall, and quantity produced. Quantity produced in
 Sørensen and Kalleberg—An Outline of a Theory of the Matching of Persons to Jobs • 635

turn reflects the performance of the employee production. This should produce different wage
and the technology used, including the techni- rates for identical labor supply because of dif-
cal division of labor adopted. For purposes of ferences in demand. However, the neoclassical
this analysis, the main variable of interest is the theory emphasizes supply differences as a source
performance of the employee and the main costs of differences in wage rates and earnings, in par-
of production of interest are the wages paid to ticular those supply differences resulting from
the employee and the costs of supervision. different skills and other individual characteris-
The performance of employees or the quan- tics related to an employee’s productive capacity.
tity of labor supplied will be taken as determined Differences in skills, according to human
by such attributes of the employees as their skills, capital theory, determine different levels of pro-
abilities, and effort. The employer’s return from ductive capacity resulting in different wage rates.
production evidently depends on his or her abil- If skills were acquired at no cost, those wage dif-
ity to obtain the highest output at the lowest ferentials would soon lead to equalizing skill ac-
costs. While numerous factors may influence the quisition. But skills are acquired at costs. These
overall level of wages, the employer’s ability to costs are partly direct in the form of tuition and
minimize costs of production depends not only living expenses and partly opportunity costs in
on the overall level of wages but also on the abil- the form of earnings foregone. No one should
ity to tie variations in wages paid to variations undertake training if the returns from this train-
in the employee’s productivity. The main argu- ing, in the form of increased earnings accumu-
ment of this chapter is that the mechanisms the lated over the working life, are not at least equal
employer can use to relate wages to performance to the costs of training.
depend on the employment relationship, partic- If only skills acquired through training are
ularly the employee’s control over access to the relevant, earnings differentials would be exactly
job, and that these different mechanisms identify off-setting the differences in training costs. How-
important differences in labor market structures ever, it is usually recognized that earnings differ-
relevant also for labor market processes other entials also capture variations in ability, where
than earnings. ability is used to refer to such characteristics as
The orthodox economic theory identifies IQ, motivation, and creativity. Ability may be
a particular set of mechanisms for relating the incorporated in the theory by recognizing that
productivity of employees to their earnings. persons with different abilities have different in-
We shall first consider these mechanisms and vestment costs and hence need different earnings
the employment relationships needed for these to induce the undertaking of training. In addi-
mechanisms to be effective. tion, some aptitudes may be innate and scarce;
these will command a rent because of their fixed
supply. Finally, some variation in earnings can
The Neoclassical Theory of Earnings be attributed to different opportunities for fi-
Determination nancing training, particularly as a result of the
In the economic theory, a wage rate is generated unequal distribution of parental wealth in com-
by a labor market as a result of the demand and bination with the unwillingness of lenders to
supply schedules of labor. Demand for labor take collateral in human capital.
varies with the derived demand for products, as The basic proposition derived from the neo-
reflected in their value. The link between wages classical theory is then that differences in earn-
and the value of products is established through ings reflect differences in the productive capacity
the concept of marginal productivity, since of persons as a result of their training, abilities,
profit-maximizing firms will be in equilibrium and training opportunities. There may be tran-
when the value of the marginal product equals sient variations in earnings as a result of differ-
the marginal cost or price of labor as a factor of ences in derived demand in combination with
636 • Part VI: Who Gets Ahead?

market imperfections, but the basic source of in- This presupposes that employees paid more than
equality in earnings is unequal endowments in their value can be replaced by others who are
productive capacities among persons. In other willing to work at the wage rate that equals mar-
words, identical persons are assumed to obtain ginal productivity, whereas employees who are
almost identical earnings, regardless of the char- paid less than their value can get access to jobs
acteristics of the jobs they are in. where the wage rate reflects their productivity.
This theory can be used to account for a Only when the employment relationship is com-
number of features of observed earnings attain- pletely open will such a clearing of the market
ment processes. Most importantly, it provides through wage rates be possible. Closed employ-
an explanation for the relation between educa- ment relationships, where new recruits can only
tion and earnings that interprets education as a get access if the incumbent leaves, insulate in-
source of marketable skills. Also, the theory pre- cumbents from competition. Employers cannot
dicts growth patterns for earnings, where earn- resolve discrepancies between productivity and
ings increase rapidly in the younger years and wage rates by threatening to replace or actually
then gradually reach a stable level, with growth replacing the current employee by someone who
after entry into the labor market explained by is more productive at the same wage rate or who
investment in on-the-job training. Empirically, is willing to work at a lower wage rate.
the theory fares well in accounting for variations It could be argued that the existence of closed
in earnings among persons, using schooling and employment relationships does not prevent the
time in the labor force (as a proxy for on-the-job employer from relating wages to performance,
training and experience) as the main indepen- even in the absence of the ability to replace an
dent variables (Mincer, 1974). employee. Most importantly, the employer can
The economic theory also emphasizes supply use promotion schemes to reward performance
in accounting for other market processes. Most and in this way obtain efficient production. This
importantly, unemployment is seen as mostly is correct. Our argument is not that closed em-
voluntary, except in certain population groups ployment relationships necessarily prevent effi-
(youngsters, blacks) where minimum wage laws cient production, but that promotion systems
make it impossible for employers to pay the mar- represent very different mechanisms for relating
ket wage. wages to performance than the use of compe-
The focus in human capital theory on the tition among employees in open employment
supply side—that is, on characteristics of per- relationships where employers make wage of-
sons—reflects the job structure assumed in the fers and employees bid for employment on the
theory—that is, one of a competitive and per- basis of their productivity. Promotions can take
fectly functioning labor market. To distinguish place only when there is a vacancy in a higher
the neoclassical theory of the earnings determi- level job and are meaningless as rewards for per-
nation process from the alternative model of the formance unless jobs at different levels provide
matching process that will be formulated later different wages, so that wages become attributes
in the chapter, we will refer to the neoclassical of jobs rather than of people. Although a firm
theory as the wage competition model (following with closed employment relationships may op-
Thurow, 1975) to emphasize the focus on com- erate efficiently because of the overall match
petition among employees for wages. . . . between job assignments and performance of
A competitive labor market that determines employees, the wages for individual employees
wage rates is one where employers make wage will reflect the jobs they hold and therefore,
offers and workers bid for employment on the not only their performance, but also the rate
basis of their productivity. The match is made at which vacancies appear, the organization of
when the value of the marginal product de- jobs, and the seniority of employees. A very dif-
manded equals the wage rate of the employee. ferent labor market structure exists from the one
 Sørensen and Kalleberg—An Outline of a Theory of the Matching of Persons to Jobs • 637

assumed in the neoclassical theory when wages equals the wage rate and can be indifferent to the
are tied to jobs and not to individual variations relationship between personal characteristics of
in performance. . . . employees and their performances. In contrast,
in vacancy competition, the employer should
be very much concerned about the relationship
Vacancy Competition between personal characteristics and produc-
When employees have control over access to the tive capacity, because once hired the employee
job, others can only get access to the job when cannot be easily dismissed. Furthermore, it is a
incumbents leave. Hence, a vacancy must exist person’s potential performance that will be of
for a person to get access to a job. We will refer concern, including the person’s ability to fulfill
to the resulting matching process as vacancy the training requirements of jobs. Previous ex-
competition. We do not wish to argue that this perience, education, and such ascriptive charac-
is the only alternative matching process to the teristics as race and sex will be used as indicators
wage competition model described by neoclas- of potential performance; the main requirements
sical economics. At least one other alternative are that the indicators chosen are visible and in
employment relationship can be identified: This the employer’s experience show some relation-
is the often met arrangement when employees ship to performance. Based on the information
are directly involved in the disposition of goods provided by these indicators, the employer will
to the market, and the “salesperson” is paid some hire the most promising candidate among those
fraction of total earnings. But such relationships available for a job. In other words, access to a
presuppose that jobs are not highly interde- vacancy will be determined by a ranking of job
pendent and that the salesperson is primarily candidates. As proposed by Thurow (1975), the
involved in the disposition, rather than in the situation may be conceived of as one where a
production, of goods.2 Vacancy competition in queue of job candidates is established for vacant
contrast is likely to emerge in closed employ- jobs. A person’s position in the labor queue will
ment relationships where jobs are interdepen- be determined, not by his or her absolute level
dent in a technical and social division of labor of productive capacity, but by the rank order in
around production. relation to other job candidates according to
In vacancy competition, as in wage compe- characteristics deemed relevant by employers.
tition, employers are assumed to be concerned As there is a queue of persons for jobs, there
about hiring the most productive employee at will be a rank order or a queue of vacant jobs,
the least cost. But because of the indeterminate where the rank order is established by the earn-
length of the employment relationship and the ings provided by vacant jobs, the career trajec-
lack of competition among employees over tories they imply, and such other characteristics
wages, it will not be possible for the employer to as status, pleasantness, and convenience. The
link marginal productivity to the wage rate. This matching process, then, is a matching of the
has important consequences for (a) the determi- queue of persons to the queue of vacant jobs.
nation of who should be hired; (b) the determi- The highest placed person in the labor queue
nation of earnings; and (c) the organization of will get the best job in the job queue. Changes in
jobs in job ladders. These consequences all fol- the supply of persons with certain characteristics
low from the employer’s attempt to secure the (say a change in the distribution of education)
highest possible return from production when and changes in the availability of jobs at different
faced with employee control over the job. levels of rewards will change the rank orderings.
In wage competition, the employer can rely As a result, whenever there is a change in the
on the wage rate as a measure of a person’s pro- labor and job queues, persons with similar char-
ductive capacity. The employer need only be con- acteristics will tend to be hired into different jobs
cerned that the value of marginal productivity and persons in similar jobs may have different
638 • Part VI: Who Gets Ahead?

personal characteristics. The organization of jobs earnings are generated by moves in mobility
into career trajectories (discussed later) will fur- regimes that are chains of vacancies in internal
ther reinforce these tendencies. labor markets. There is, in vacancy competition,
Wage rates in vacancy competition are char- no automatic correspondence between the cre-
acteristics of jobs, not of persons. Because em- ation of promotion opportunities and whatever
ployers have no effective way of enforcing a changes take place in a person’s productive ca-
translation of productivity variations into wage pacity. Employees may be promoted without a
rates other than by promotions, wages will tend preceding change in productivity, and a change
to become heavily influenced by such institu- in productive capacity need not result in a pro-
tional forces as collective bargaining and em- motion. This means that the cross-sectional
ployee desire to preserve traditional relative wage association between personal characteristics
differentials. Internally, wage differentials will and earnings will be attenuated, even though
reflect the organization of jobs into job ladders. personal characteristics are crucial for access to
The creation of job ladders in internal labor jobs. (A formal derivation of this conclusion and
markets is, as already mentioned, a way for the an empirical illustration is presented by Wise,
employer to create an incentive structure in the 1975.)
absence of open employment relationships. The In vacancy competition, variations in earn-
organization of jobs into promotion schedules ings reflect variations in job characteristics and
further acts as a screening device, inducing the organization of jobs in internal labor mar-
low-performance employees to leave on their kets. This is in contrast to the situation in the
own decision by denying or delaying promotion neoclassical model of wage competition, where
in relation to other employees. To be effective, the primary source of variation is the variation in
jobs at the same level in a promotion schedule personal characteristics that determine a person’s
should provide identical earnings, whereas jobs productive capacity.
at different levels should provide a differential Vacancy competition structures are likely to
large enough to induce employees to compete be similar to the job structures identified as pri-
for promotion opportunities. This further rein- mary jobs (e.g., Doeringer & Piore, 1971). How-
forces the tendency in vacancy competition for ever, the dualist literature has a very descriptive
earnings to become a characteristic of jobs so character, and there is also some confusion as to
that similar jobs provide similar earnings regard- whether the labor market segmentation is a seg-
less of characteristics of the incumbents. mentation of jobs or of persons (blacks, poor,
Actual promotion opportunities are cre- and women in the secondary sector, white skilled
ated when persons leave the firm or a new job workers in the primary sector). The main con-
is added, setting in motion chains of vacancies clusion derived from this literature is that there
(White, 1970). The number of job levels, the are good jobs and bad jobs.
distribution of jobs at various levels, the seniority
distribution of employees, and the demand for
products influencing the creation of new jobs Constraints on Growth in Earnings
(or the elimination of jobs), all interact to pro- The two polar models of the matching process
duce promotion schedules governing the careers suggest different constraints on a person’s ability
of employees. These promotion schedules will to increase his or her earnings. In wage compe-
under certain conditions result in career lives tition, earnings directly reflect performance and
that are similar to those predicted by human hence the skills and abilities of a person. Increases
capital theory, even though the mechanisms are in earnings then are obtained by increasing the
quite different (Sørensen, 1977). skill level of a person, and the major constraint
In wage competition, employees can change on growth in earnings will be limitations on ac-
their earnings only by changing their perfor- quiring additional human capital. In wage com-
mance. In vacancy competition, changes in petition markets, the amount of training that
 Sørensen and Kalleberg—An Outline of a Theory of the Matching of Persons to Jobs • 639

can be provided in jobs will be low, since on-the- Society, Volume 1. “A social relationship . . . will be
job training is a major cause of the emergence known as ‘open’ to those on the outside, if . . . par-
of vacancy competition (Thurow, 1975). Hence, ticipation . . . is . . . not denied to anyone who is
the major source of income inequality among inclined to participate and is actually in a position
to do so. The relationship will be known as ‘closed’
persons lies outside the labor market—that is, in
[if ] participation of certain persons is excluded, lim-
the educational and other training institutions ited or subject to conditions [Weber, 1947, p. 139].”
that produce skill differentiation. Weber argues that market relationships are open and
In vacancy competition sectors, the major gives as an example of a closed relationship the “es-
constraint on the attainment of income is access tablishment of rights to and possession of particu-
to jobs. If no job is available, a person will not lar jobs on the part of the worker [Weber, 1947, p.
be able to obtain earnings. Growth in earnings 141].” This identification of open relationships with
is produced by the utilization of opportunities market relationships (for the exchange of labor for
for mobility to better jobs, and this opportu- wages) and of closed relationships with control over
the job by the worker (and the absence of market
nity structure, not changes in skills, governs the
relationships) will be relied on heavily in this chapter.
earnings variations over time. The major source 2. A similar arrangement accounts for the appar-
of variation in earnings is then the restriction of ent contradiction of the argument presented here
access to jobs and the level of derived demand exemplified by the existence of wage competition
that determines the availability of jobs. among faculty at elite universities despite tenure.
The different constraints on growth in earn- Here the individual scholar, and not the employer
ings in wage competition and vacancy compe- (i.e., the university), disposes himself of the products
tition jobs imply that quite different policies (articles and other contributions) to a competitive
will have to be used in an attempt to increase market and obtains himself the returns from this ac-
tivity (i.e., prestige in the profession).
pretransfer earnings of poverty groups. In wage
competition sectors, policies aimed at increas-
REFERENCES
ing skill levels either through schooling or—for Averitt, R. T. 1968. The Dual Economy. New York:
those already having entered the labor market— Norton.
through various off-the-job training programs Becker, G. S. 1964. Human Capital. New York: Na-
would presumably be effective. In vacancy com- tional Bureau of Economic Research.
petition sectors such policies would be quite Bluestone, B. 1970. “The tripartite econ-
ineffective since such training would not make omy: Labor markets and the working poor.”
jobs available. Poverty and Human Resources Abstracts (5
The rather limited success of worker training March–April):15–35.
Cain, G. G. 1976. “The challenge of segmented
programs suggests that job vacancy competi-
labor market theories to orthodox theory: A
tion indeed is predominant in the U.S. econ- survey.” Journal of Economic Literature 14:
omy. More correctly, the fate of such programs 1215–1257.
suggests that it is indeed difficult to prepare Doeringer, P. B., and M. J. Piore. 1971. Internal
low-skilled workers for jobs that demand high Labor Markets and Manpower Analysis. Lexing-
skill levels, since such jobs tend to be vacancy ton, Massachusetts: Heath.
competition jobs. Kerr, C. 1954. “The Balkanization of labor markets.”
In E. W. Bakke, P. M. Hauser, G. L. Palmer, C.
NOTES A. Myers, D. Yoder, and C. Kerr (eds.), Labor
This research was supported in part by funds Mobility and Economic Opportunity. Cambridge,
granted to the Institute for Research on Poverty at Massachusetts: Technology Press of MIT.
the University of Wisconsin by the Office of Eco- Marx, K. 1961. Capital (Vol. 1–3). Moscow: Foreign
nomic Opportunity pursuant to the provisions of Language Press. (Originally published in English,
the Economic Opportunity Act of 1964. The con- 1887.)
clusions expressed herein are those of the authors. Mincer, J. 1974. Schooling, Experience and Earn-
1. The definitions are given in paragraph 10 in ings. New York: National Bureau of Economic
the section on “Basic Concepts” in Economy and Research.
640 • Part VI: Who Gets Ahead?

O’Connor, J. 1973. The Fiscal Crisis of the State. Thurow, L. C. 1975. Generating Inequality. New
New York: St. Martin’s. York: Basic.
Sørensen, A. B. 1977. “The structure of inequality Weber, M. 1947. [The Theory of Social and Economic
and the process of attainment.” American Socio- Organization] (A. M. Henderson and T. Parsons,
logical Review 42:965–978. Trans.). New York: Oxford Univ. Press.
Spilerman, S. 1977. “Careers, labor market struc- White, H. C. 1970. Chains of Opportunity: System
ture, and socioeconomic achievement.” American Models of Mobility in Organizations. Cambridge,
Journal of Sociology 83:551–593. Massachusetts: Harvard Univ. Press.
Stolzenberg, R. M. 1975. “Occupations, labor mar- Wise, D. A. 1975. “Personal attributes, job perfor-
kets and the process of wage attainment.” Ameri- mance and probability of promotion.” Economet-
can Sociological Review 40:645–665. rica 43:913–931.

72 • Arne L. Kalleberg

The Rise of Precarious Work

Employment relations have become more precar- government, business, and labor in the United
ious—more uncertain, insecure, and risky—in all States in the decades following World War II.
industrial societies over the past quarter century. The metaphor of a social contract has its roots
In the United States, the anxiety and inequal- in the philosophies of Hobbes, Locke, Rousseau
ity accompanying the expansion of precarious and Rawls. It refers to the mutual expectations
employment has not only affected how work is and responsibilities that society and individuals
experienced, but also how families and commu- have toward each other. Explicit and implicit
nities bear risks and how firms and society con- social contracts in the post-WWII period em-
duct business. Uncertainty, insecurity, and risk are phasized collective solutions to solving social
pervasive throughout the labor market, and have problems as well as long-term and fairly stable
affected both younger and older workers alike. relations between employers and their employ-
Precarious work is not new; it has existed since ees. Unfortunately, government and business
the beginning of paid employment. But globaliza- have deserted their obligations to their workers
tion, technological change, re-regulation of labor and communities over the past several decades,
markets and the removal of institutional protec- and people are now told that they are “on their
tions have shifted the balance of power away from own” to address their concerns. We need a new
workers and toward employers and made precar- social contract that will provide social insurance
ious work increasingly common across the globe. and security to individuals. What we need, in
These shifts in power relations are structural short, is to once again spread the risk around.
transformations in labor markets, not temporary
fluctuations in supply and demand associated
with the swings of business cycles. The Growth of Precarious Work
The growth in precarious employment A variety of data can be marshaled to document
has occurred just as we’ve abandoned the im- the growth of precarious work. Taken as a whole,
plicit social contract that once bound together these data strongly support the conclusion that

Arne L. Kalleberg. “The Rise of Precarious Work,” Pathways (Summer 2012), pp. 1–12. Used by permission
of the Center on Poverty and Inequality, Stanford University.
 Kalleberg—The Rise of Precarious Work • 641

Figure 72.1. Workers With Nonstandard Employment Relations, 1995–2005 (ages 16 and over, in millions)

12

10

8
MILLIONS

0
1995 1997 1999 2001 2005

Independent On-Call Temporary Help Workers Provided

Contractors Workers Agency Workers by Contract Firms

Source: Bureau of Labor Statistics, “Contingent and Alternative Employment Arrangements” press releases, 1995–2005.

there has been a transformation of employment employers to hire workers from outside the orga-
relations toward greater uncertainty and instabil- nization rather than to develop the human capi-
ity in the United States since the 1970s. tal of their employees internally. There has been
First, there has been an expansion and in- a dramatic increase since the 1950s, for instance,
stitutionalization of nonstandard employment in the proportion of managers hired from the ex-
relations, such as independent contracting and ternal market rather than being promoted from
temporary work. These forms of nonstandard within the organization. Because workers cannot
work have spread throughout the labor force in as easily climb the ladder within their firms, they
both high- and low-skill jobs. Figure 72.1 shows are likely to feel more precarious and insecure
the trends in four types of nonstandard employ- than in days past.
ment relations from 1995 to 2005. The rise was Fourth, trends in involuntary job loss sug-
greatest for independent contractors; the main gest that job stability and security have declined
increases in the others occurred before 1995, since the 1970s, especially for prime-age males
when data on nonstandard employment rela- and those in white-collar occupations. The pro-
tions began to be collected systematically. portion of males aged 35 to 54 who were perma-
Second, there has also been a general decline nently displaced from their jobs almost doubled
in job stability, with substantial reductions in the between the 1970s and 1990s.
average length of time a person spends with his Fifth, the foregoing trends explain, in part,
or her employer. This trend has been experienced the steady upward march in long-term unem-
mainly by men: The percentage of women with ployment since the 1960s. Figure 72.2 illustrates
ten or more years of employer tenure increased the growth in the percent of unemployed per-
from about 25 percent in 1983 to about 29 per- sons who have been out of work for 52 weeks or
cent in 2004, while the corresponding percent- more. This percentage has spiked dramatically
ages for men decreased from about 38 percent to since the Great Recession of the late 2000s. But
about 32 percent. even aside from this spike, the long term secular
Third, the decline in employment tenure trend has been upward.
occurred just as internal labor markets weak- Sixth, we have also seen a shifting of risk
ened, as reflected in the increasing tendency for from employers to employees, especially in
642 • Part VI: Who Gets Ahead?

Figure 72.2. The Share of the Unemployed Who Have Been Jobless For Six Months or More, 1948–2012

50

45

40

35

30
PERCENT

25

20

15

10

0
1948 1952 1956 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012

Source: Economic Policy Institute.

Notes: Data are not seasonally adjusted. Data are shown by quarter. National Bureau of Economic Research recessions
are shaded.

Figure 72.3. Trends in Defined Contribution vs. Defined Benefit Pension Plans, 1983–2007

70
PERCENT OF COVERED WORKERS

60

50

40

30

20

10

0
1983 1986 1989 1992 1995 1998 2001 2004 2007

Defined-Contribution Defined-Benefit

Note: There are no data for 1986. Data for 1986 are the midpoint between 1983 and 1989.

Source: Munnell, Alicia H., Golub-Sass, Francesca, and Muldoon, Dan. 2009. “An update on 401(k) plans: Insights from
the 2007 SCF.” Center for Retirement Research at Boston College: no 9–5; Economic Policy Institute.

relation to the employment-related benefits guaranteeing a certain level of benefits). Other

that workers have historically counted upon. types of employer-provided benefits have also
For example, Figure 72.3 shows the increase in shifted away from providing adequate security
defined contribution pension plans (in which for American workers.
employees absorb more of the risk than employ- Finally, a rising percentage of Americans say
ers) and the decline in defined benefit plans (in that they are insecure in their jobs. When the
which the employer absorbs more of the risk by 1977-2010 General Social Survey is analyzed,
 Kalleberg—The Rise of Precarious Work • 643

one finds an increasing trend in the percentage of developed institutions that equip them, more so
people responding that they both think it is very than other countries, to address the challenges
or fairly likely that they will lose their current job and consequences posed by the global division
within the next year and think that it would not of labor and the tendencies toward precarious
be at all easy to find another comparable job. Since work. These countries are well positioned to ex-
this rise takes into account changes during this ploit the labor market challenges that arise in a
period in the business cycle and underlying demo- global market.
graphics, it is consistent with the explanation that Flexicurity is win-win because it allows
there has been a sea change in employment rela- employers and labor markets to have greater
tions in the United States toward more precarious flexibility even as workers are allowed greater
employment. There is also evidence that workers protections. Flexicurity principles, however,
want more job security even as the opportunity need to be tailored to their context. The main
for such security is much reduced. Work has not contextual constraint in the United States is that
only become more precarious, but workers are employers already have considerable flexibility in
feeling the change and unhappy with it. their relations with employees. It’s accordingly
a non-starter in the American context to sug-
gest flexibility-reducing innovations. The major
A New Social Contract challenge, then, in adopting “flexicurity with
The question, then, is how do we as a society an American face” is to provide workers with
respond? A simple answer: A new social con- security in dealing with the changes that have
tract is needed to tackle the consequences of this occurred in labor markets and employment re-
growth in precarious work. The contract must lations without jeopardizing the already consid-
be sensitive to the conditions that led to the erable flexibility upon which U.S. firms count.
transformation of employment relations in the This challenge can be met. The following are
first place. Labor, product, and capital markets the three main features of a comprehensive new
are now global phenomena, and they interact social contract that meets the challenge.
to jointly intensify price competition. The ra- Economic Security: People must be assured
pidity of technological innovation both forces an adequate level of current and future income
companies to become more competitive globally if they are to be induced to make investments
and makes it relatively easy to move goods, cap- and assume risks. Consistent with the practice in
ital, and people within and across borders at an many developed industrial countries, the highest
ever-accelerating pace. Outsourcing and tempo- priority should be given to providing three types
rary work are increasingly available options for of social insurance: portable health insurance
reducing costs. Neo-liberal ideologies and pol- benefits; more generous and secure retirement
icies have encouraged a limited welfare state, benefits; and expanded unemployment benefits
weakening of unions, lowering of taxes and fees and other wage supports (including assistance
on businesses, and fiscal discipline taking prece- with acquiring new skills and relocation). These
dence over social protections. types of insurance help people navigate the in-
All countries are faced with the core problem creasingly treacherous transitions between jobs
of balancing flexibility for employers and secu- and employers and, just as importantly, they give
rity for workers. But countries have sought to them the confidence to assume risks in investing
solve this dilemma in different ways depending in human capital or exploiting entrepreneurial
on their institutional and cultural traditions. opportunities.
Flexicurity models—originating in Europe and Representation Security: The often-
now spreading to Asia and elsewhere and which unappreciated virtue of allowing for some form
involve both employers and workers in a coop- of collective representation among workers is
erative effort—suggest that labor market institu- that it forces employers to adopt longterm ra-
tions matter a good deal. Some countries have tional strategies. Put simply, employers are more
644 • Part VI: Who Gets Ahead?

likely to adopt high-road strategies if they are adopt “high road” employment systems and be
prodded by strong unions or other forms of incentivized to create more skilled jobs to take
worker organization that encourage collaborative advantage of a higher-skilled workforce. More
efforts between managers and workers and make economic security is also likely to spur entrepre-
it costly to abuse or exploit workers. The new neurial activity and make people more willing to
forms of organization won’t simply be replicates invest in their human capital. Greater represen-
of traditional industrial unions in the United tational security will help to redress the balance
States. These unions engaged in collective bar- of power between employers and labor and also
gaining with employers (either single employers spur the creation of good jobs.
or coordinated groups), and the main focus was Moreover, these three types of security are
on bread and butter issues such as earnings and likely to facilitate the acquisition of other forms
job security. The growth of precarious employ- of security. Workers with representational secu-
ment relations has reduced workers’ attachments rity are more likely to enjoy greater occupational
to their employers and increased the salience of health and safety. And workers with greater eco-
labor market intermediaries that help to create nomic and skill security should be better able to
channels for mobility between firms. It’s not get new jobs and be retrained should they lose
just a matter, then, of ramping up conventional their jobs or if the labor market does not provide
unions. The new generation of worker organiza- sufficient job opportunities.
tions must adapt to and match up with changes
in the structure of the economy and the organi-
zation of work. Implementing the New Social
Skill Reproduction Security: A new social Contract
contract must also help workers and employers A common refrain in these difficult economic
cope with the likelihood that all workers will times is that there aren’t any new ideas about
have to move among jobs relatively frequently. fixing the economy. That’s not true. Although
People need access to basic education and vo- flexicurity is hardly a new idea outside the U.S.,
cational training, thereby helping to retrain and it is new to the U.S. context and, contrary to
prepare them for good jobs. The importance conventional wisdom, it should have a special
of education and job-related training cannot resonance in the U.S. precisely because it allows
be over-emphasized. The explosive growth of us to build on our strong commitment to main-
information technology and the escalating im- taining employer flexibility.
portance of knowledge in the economy have This isn’t to suggest that a flexicurity pro-
underscored the significance of skills and ed- gram will be easy to implement. If we’re serious
ucation for obtaining and performing well in about putting the model into practice—and we
high-quality jobs and for avoiding confinement should be—it will require a complicated dance
to bad jobs. Life-long learning is becoming more between government, business, and labor. Mobi-
essential than ever, due to the need for people lizing these three partners is difficult. There are
to adjust to the technological changes that help numerous obstacles produced by (a) ideological
to create job insecurity and uncertainty. Capi- disagreements about the appropriate role of the
talizing on the skills and knowledge of Ameri- government in the economy and labor mar-
can workers also enhances the competitiveness kets, (b) a lack of trust in the government and
of American firms, which cannot compete with institutions in general, (c) the current weakness
developing countries on the basis of low labor of the labor movement, which prevents work-
costs. ers from exercising voice and labor from being
Enhancing economic representation and skill a countervailing force, (d) businesses seeking
reproduction security will result in the creation to cut costs and lobbyists trying to obtain fa-
of better jobs. Employers will be encouraged to vorable regulations for their clients, and (e) the
 Rosenfeld—Little Labor • 645

economic challenges imposed by concerns over (the Temporary Disability Benefits law). Another
budget deficits, slow growth, and stubbornly promising application of this strategy builds on
high unemployment. the Trade Adjustment Assistance (TAA) leg-
Nevertheless, a combination of strategies that islation to facilitate the transition of displaced
emanate from the “top” as well as the “bottom” workers to new jobs. The TAA originated in
is likely to be most effective, with bottom-up the early 1960s and was designed to protect
solutions perhaps especially attractive in the workers in manufacturing industries from job
short-run given the current political deadlock. displacement produced by import competition.
Although some top-down interventions will ul- The transition mechanism within TAA could be
timately be needed to create the requisite legis- replicated and expanded beyond manufacturing
lative and regulatory environment, there’s much to service industries as well as extended to cover
that can be done even now at a more local level. reasons for job displacement in addition to im-
The Hosiery Technology Center in North Caro- port competition.
lina illustrates, for example, a cooperative train- It’s altogether possible, then, that we’ll drift
ing and retraining initiative among employers’ slowly and gradually to a flexicurity-flavored
groups, local governments, and community col- labor market, not because a grand political con-
leges. There are likewise important collaborative sensus is forged, not because there’s some over-
efforts in North Carolina between community riding public outcry for it, but simply because it
colleges and firms in the biotechnology industry works well for all involved. Although it’s perhaps
to train (and retrain) workers for new jobs. unlikely that an overt flexicurity movement will
It’s also possible to develop at least some develop in the near term, the smart money will
flexicurity-flavored features in a gradual fashion be carefully watching these local experiments in
by building on existing institutions or laws. By flexicurity and tracking their spread. If it does
taking small steps, we can avoid direct confron- become a fad and even a movement in the long
tation with entrenched interests and help to run, its simple mantra will be that reforms tak-
circumvent political blockages. A version of this ing on precarious work are not just good for U.S.
strategy was used recently in New Jersey, where workers but also for U.S. competitiveness and
paid family leave was added to an existing law profits.

73 • Jake Rosenfeld

Little Labor
How Union Decline Is Changing the American Landscape

A recent Wall Street Journal editorial decrying the investment income. Such articles are typical fare
role of Big Labor in shaping the Obama admin- for a newspaper long critical of the labor move-
istration’s domestic policy expressed worry that ment’s role in American life. But what’s strange is
unions’ outsize clout would force higher taxes on the continued use of “Big Labor” as a shorthand

Jake Rosenfeld. “Little Labor: How Union Decline Is Changing the American Landscape,” Pathways (Summer
2010), pp. 3–6. Used by permission of the Center on Poverty and Inequality, Stanford University.
646 • Part VI: Who Gets Ahead?

moniker for trade unions in the contemporary standardization that govern many public-sector
United States. If organized labor remains big contracts. Organized labor, then, is disappearing
today, then back in its post–World War II peak, in the sector where historically it has had the
it was positively enormous. Fully one-third of greatest impact on people’s livelihoods.
the private sector workforce belonged to a labor But even less understood than the overall de-
union during the 1950s, and millions more re- cline in unions’ prevalence is the concomitant
sided in households reliant on a union wage. decline in unions’ activity. Academics have long
During the heyday of collective bargaining in debated whether high levels of unionization are
this country, unions helped pattern pay and a net good when it comes to global competi-
benefit packages among nonunion workers, as tiveness or overall economic performance. But
employers matched union contracts to forestall fewer dispute that unions have been a histori-
organizing drives and maintain a competitive cally positive force in bolstering the economic
workforce. Politicians, Democrats especially, prospects of union members themselves. Unions
depended on organized labor’s support during bolster workers’ clout in confrontations with
elections and consulted closely with labor leaders employers, historically winning them higher
when devising policy in office. Big Labor, then, wages, better benefits, and greater workplace
was once quite big indeed. protections than might be offered otherwise.
The only thing that remains big about labor Strikes represent unions’ most potent weapon
unions today is their problems. Figure 73.1 in confrontations with employers, and this
tracks unionization rates for private and public weapon used to be a regular feature of Amer-
sector workers between 1973 and 2009. By the ica’s industrial landscape, affecting millions of
early 1970s, organized labor had already begun workers each year. But this has changed. Figure
its decades-long decline, but still nearly a quar- 73.2 below presents two series: the first shows
ter of all private-sector employees belonged to the number of large strikes (involving 1,000 or
a labor union at this time. The late 1970s and more workers) over the last 45 years. The num-
1980s proved especially brutal for organized ber of strikes involving 1,000 or more workers
labor, with unionization rates halving during the peaked at over 400 in 1974. In 2009, there were
period. The nation’s intellectuals and journalists five. While the sheer precipitousness of this de-
covered this phenomenon extensively, linking cline is staggering, we know that strikes of such
union decline to a new post-industrial economy magnitude are often unrepresentative of more
increasingly open to global trade. Recent trends typical work stoppages. But to date, no public
have garnered less attention, yet private-sector data has been available to document strikes of
unionization rates nearly halved again between all sizes in recent decades. Because of this, I
1990 and 2009. The story for public sector filed a Freedom of Information Act (FOIA) re-
unions has been a bit brighter. Rates of orga- quest to obtain information on all strikes for the
nization among government workers increased years in which data were collected. Figure 73.2
steadily during the 1970s, settling at slightly over presents this data, and the trend mirrors what’s
one-third of all public sector workers, where they been happening with large strikes. As late as the
have remained relatively consistently up to the mid-1980s, nearly 1,000 walkouts occurred in
present. Three decades of stasis in public-sector a single year. By the dawn of the 21st century,
organization rates suggests that the earlier ex- that number had fallen to just over 200, a de-
pansion may have reached its limit. And over cline of nearly 80 percent in less than 20 years.
four-fifths of the U.S. workforce is employed What we’ve seen, then, is a rise in what might
in the private sector. Moreover, recent research be called “union dormancy,’’ whereby unions are
has demonstrated that the benefits of union no longer routinely agitating on behalf of their
membership are much smaller in the pub- membership, at least not in the traditional form
lic sector, due to the relative transparency and of the labor strike.
 Rosenfeld—Little Labor • 647

Figure 73.1. Unionization Rates by Sector, 1973–2009

25% 50%

20% 40%
PRIVATE UNIONIZATION

PUBLIC UNIONIZATION
15% 30%

10% 20%
private sector
public sector

5% 10%
1977 1981 1985 1989 1993 1997 2001 2005 2009
Data are provided by Barry T. Hirsch and David A. McPherson’s www.unionstats.com database (2010), and are based on
Current Population Survey (CPS) data. Unionization data for 1982 are unavailable; I generate 1982 estimates by averag-
ing 1981 and 1983 rates.

So what happened to Big Labor? Organized shift bargaining power away from labor unions.
labor’s penetration was especially deep in core By the early 1980s, innovative tactics adopted by
manufacturing industries. The transformation management and used against organizing drives
to a post-industrial economy hit union work- and existing unions shattered the relative detente
ers in these industries hard, as jobs became in- between business and labor that had predomi-
creasingly vulnerable to outsourcing, deskilling, nated for decades. These sophisticated strategies
and technological innovations rendering many took full advantage of existing policies governing
positions redundant. The process accelerated labor-management relations and proved incred-
throughout the 1970s and 1980s, as tradition- ibly effective at pushing back at what employers
ally protected industries like auto manufacturing felt was overreach by unions.
opened up to competition from abroad, push-
ing domestic manufacturers to search for less
labor-friendly jurisdictions. Yet private sector A New Landscape
deunionization was not limited to the manu- While the causes of organized labor’s
facturing sector; across all major industries with decades-long decline in the private sector are
some union presence, membership rates remain well known, the broad consequences are not.
lower today than in the past. This is true even in Existing research tends to focus on deunion-
those industries not threatened by cheaper labor ization’s consequences for the earnings of male,
overseas, such as transportation and retail. The blue-collar workers. But the removal of orga-
wave of deregulation that began in the Carter nized labor from much of the private sector also
administration opened up some of these sectors affects the economic assimilation of recent im-
to cutthroat competition, pressuring employ- migrants and their offspring, widens black-white
ers to shed expensive contracts and the unions wage inequality among female workers, redis-
that bargained for them. Partly as a response to tributes political power, and redefines the nature
deindustrialization and deregulation, there arose of strikes in modern America. I touch on each of
a concerted, broad-based effort by employers to these consequences below.
648 • Part VI: Who Gets Ahead?

Figure 73.2. Work Stoppages in the U.S., 1973–2009

500 1,000
large strikes
all strikes 800
400

STRIKES OF ANY SIZE

300 500
LARGE STRIKES

200 400

100 200

0 10%
1977 1981 1985 1989 1993 1997 2001 2005 2009

Data for large strikes provided by the Bureau of labor Statistics (BLS) Historical Work Stoppage Database (http://www.
bls.gov/wsp/data.htm). The BLS defines large strikes as those including 1,000 or more workers. Data for strikes of any size
provided to the author by the Federal Mediation and Conciliation Service (FMCS).

The Disappearing Economic Ladder recent decades top unions have eyed immigrant
for Hispanic Immigrants workers warily. Many assumed immigrants were
largely unorganizable, due to the precarious
Unionization has always been unevenly spread legal status of some recent arrivals, the lower
across demographic groups. The labor move- labor standards immigrants were accustomed
ment’s great upsurge between the Great De- to in their home countries, and the resulting
pression and World War II relied heavily on worry that employers would use immigrant
European immigrants and their children, with labor to undercut existing wages and benefits of
many arrivals assuming top leadership posts in native-born workers. The “Justice for Janitors”
the nation’s fastest growing unions. During the campaign in Southern California helped counter
labor movement’s peak, unions helped provide such claims and helped galvanize organizers
a firm economic foundation for these otherwise across the nation, who sought to capitalize on
disadvantaged populations, propelling millions the class-based solidarity exhibited by many His-
into the middle class. Some have argued that la- panic immigrants. And indeed, certain Hispanic
bor’s future is brightening once again, given the subgroups, including immigrants who have lived
influx of Hispanic immigration since the 1960s. in the United States for a number of years and
That is, if labor can organize recent immigrants, immigrants who are citizens, are joining unions
unions might once again reclaim a powerful po- at higher rates than native-born Whites. Figure
sition in the economic landscape. This optimism 73.3 displays the odds of joining a union over a
is driven by events like the labor movement’s one-year period for various Hispanic subgroups
success in organizing largely Hispanic janitors compared to U.S.-born Whites. Odds ratios
in Southern California, many of them recent above 1 indicate that the Hispanic subgroup is
immigrants. more likely to join a union than a White nonim-
But how is organized labor actually inter- migrant. U.S.-born Hispanics have over 40 per-
acting with this new wave of immigration? cent higher odds of joining a union compared to
Despite the historical role immigrants played U.S.-born whites, echoing the historical pattern
in building the U.S. labor movement, in more of immigrant groups and their children seeking
 Rosenfeld—Little Labor • 649

Figure 73.3. Odds of Joining a Union, 1973–2009

2
1.8
1.6
1.4
Hispanic
1.2 Hispanic
citizen
non-
1 immigrant Hispanic
immigrated
0.8 10–20 years Hispanic
0.6 prior immigrated Hispanic
5–10 years immigrated
0.4 prior less than 5
years prior
0.2
0
Odds ratios refer to the relative odds of joining a union over a one-year period where the reference category is
non-immigrant whites. Data come from matched files of the Current Population Survey (CPS), and estimates adjust for
a range of factors influencing unionization. For more on the estimation procedure, see Rosenfeld, Jake, and Meredith
Kleykamp. 2009. “Hispanics and Organized Labor in the United States, 1973 to 2007.” American Sociological Review
74:916–37.

unionized employment to assimilate upward The Declining Significance of

into the middle class. Hispanic immigrant citi- Unions for Black Females
zens and Hispanic immigrants who have lived in
the United States for many years are also joining Aside from limiting mobility for low-skilled im-
unions at higher rates than native-born Whites. migrant populations, the decline of organized
But there are limits to such trends. Despite labor exacerbates economic inequality between
the highly publicized organizing drives of the African Americans and Whites. Unionization
“Justice for Janitors” campaign, the percentage rates for African Americans have exceeded those
of Hispanic janitors in labor unions has actually of Hispanics and Whites for decades now. As
declined since 1990, as has the fraction of all jan- the labor movement began integrating its ranks,
itors who claim union membership. Unlike the African-American workers, eager to escape dis-
Southern and Eastern European migrants who criminatory treatment institutionalized in U.S.
once swelled the ranks of the union workforce, labor markets, sought out organized labor as a
recent arrivals face an economic context largely partial refuge against economic inequity. This is
hostile to trade unions. In those remaining parts especially true for females. Despite the stereotyp-
of the private sector where unions remain ac- ical image of the blue-collar male union worker,
tive, Hispanics’ and Hispanic subgroups’ rela- unionization rates for African-American females
tive unionization rates are high, but their overall rose dramatically during the 1960s and 1970s,
unionization rates are low—along with nearly with nearly one in four Black women in the pri-
everyone else’s. Thus, contemporary immigrants vate sector belonging to a union by the end of
and their offspring enter labor markets that in- the 1970s. In the heavily industrialized Midwest,
creasingly lack an established unionized pathway rates of unionization for African-American fe-
to the middle class, a pathway that past immi- males working in the private sector peaked at 40
grant populations relied upon extensively. percent. Past work by economists John Bound
650 • Part VI: Who Gets Ahead?

and Richard Freeman has found that union de- political consequences. The already high voter
cline widened wage gaps between young Black turnout rates—and education levels—among
and White males, especially in the Midwest. But government workers, union and nonunion alike,
the ramifications of deunionization for racial leave little room for unions to raise turnout in
wage inequality are actually larger for females, the public sector. Meanwhile, in the private sec-
given that differences in private sector union- tor, union status remains a significant indicator
ization rates between Black and White females of whether an individual will vote or not. How-
far exceed differences between Black and White ever, given the reduced fraction of private-sector
males. Indeed, had unionization rates remained workers in labor unions, the aggregate effect of
at their peak levels, Black–White wage differ- unionization on voting turnout is now quite
ences among private sector females would be small, and shrinking union rolls reduce the abil-
nearly 30 percent smaller than where they stand ity of unions to drive up turnout among non-
today. union citizens.
The consequences of union decline described
above largely focus on nonunion workers—those
A Political Force Diminished who in the past would have benefited from union
As unions vanish from the economic landscape, membership but who no longer will, whether
their presence in the political realm is reduced they be an immigrant employee who once would
as well. Historically, the labor movement has have been organized, a female African-American
channeled and organized the political energies of worker no longer able to rely on a union wage
the working class, helping to counter the robust, to reduce pay gaps with her white counterpart,
positive connection between civic participation or a less-educated worker lacking the training,
and socioeconomic status. Indeed, trade unions resources, and knowledge to participate in poli-
have historically stood as one of the few insti- tics. But union decline affects remaining union
tutions equalizing political participation across workers as well. Research by economists John
income and educational divides. Nowhere was DiNardo and David S. Lee suggests that the
this role more pronounced than in the private union wage benefits for newly organized manu-
sector, where voting rates run comparatively facturing firms are negligible. This may be due,
low, especially among those lacking a college in part, to the dramatic decline in strikes de-
education. This is not true in the public sec- scribed earlier. In decades past, unions often au-
tor. The combined effects of unionization and thorized a walkout during contract negotiations,
public-sector employment are not simply addi- pressuring employers to raise wages and benefits.
tive: public-sector employment bolsters political These pressure tactics worked; union members
participation, but being in a public-sector union who had participated in a strike had higher
results in only a slight increase in the propensity wages, on average, than non-striking members.
to vote. Figure 73.4 presents predicted probabili- This no longer seems true today. While we lack
ties of voting for public- and private-sector union direct measures of strikes’ impacts on an indi-
members and nonmembers. The difference in vidual striker’s pay, the Federal Mediation and
voting turnout among public sector members Conciliation Service data presented in Figure
and nonmembers is only 2.5 percentage points. 73.2 allow for comparisons between pay scales
The effect of union membership on voting in the in industries and regions in which strike activ-
private sector is nearly three times as large. ity remains relatively high and those in which
Today, the number of public sector union strikes have disappeared. I find that the positive
members equals the number of private sector wage-strike relationship has been severed; work-
union members, marking a dramatic break from ers in high-strike locales see no wage gains com-
when private sector union rolls dwarfed those of pared to workers in relatively quiescent sectors.
government employees. This shift has important Strikes now are often last-ditch attempts to hold
 Rosenfeld—Little Labor • 651

Figure 73.4. Predicted Probabilities of Voting for Union Members and Nonmembers, 1984–2006

70%

65%

60%

55%

50%
public sector union vs. private sector union vs.
public sector nonmember private sector nonmember

Probabilities generated from voter turnout models that adjust for a range of demographic, economic, and geographic
factors found to influence voting. Sample is restricted to employed citizens only, age 18 and over. Data come from the No-
vember series of the Current Population Survey (CPS). For more on the estimation procedure, see Rosenfeld, Jake, 2010.
“Economic Determinants of Voting in an Era of Union Decline.” Social Science Quarterly 91:379–96.

the line on wages and benefits, as union leaders Organized labor’s signature legislative effort
simply refrain from striking except in the most is the Employee Free Choice Act (EFCA). In
desperate situations. Thus, unions are not only its most robust form, the proposed legislation
failing to bolster the fortunes of those who once would radically recast how union elections are
would have been organized, they are also strug- held in the United States, bypassing the tra-
gling to protect the fortunes of those still in their ditional election campaign in favor of a “card
ranks. check” policy whereby a union is recognized
after over half of workers sign up in support of
collective bargaining. A compromise version of
Where from Here? the bill would retain the “secret ballot” election
It is difficult indeed to counter the self- procedure but would reduce election times,
perpetuating dynamic behind the foregoing grant organizers greater access to employees on
trends. As union ranks shrink, so too does the the worksite, and institute binding arbitration
constituency directly mobilized to press for if a contract has not been agreed upon after a
change, and with it, labor’s leverage in convinc- specified period of time. Passage of either ver-
ing lawmakers to risk the political consequences sion would shift some of the power in organizing
of business opposition. The present economic drives to labor, although it would not address
climate further dampens enthusiasm for worker the broader economic challenges labor faces,
activism, as employees cling to their positions, such as the continuing decline of manufacturing
while millions of others less fortunate scramble employment, the pressures of international com-
to find work. petition among remaining manufacturing firms,
652 • Part VI: Who Gets Ahead?

and aggressive competition in many deregulated Any policy effort to help organized labor
domestic industries. faces formidable political opposition, although
There are other institutional changes that, if we can’t rule out the possibility that the admin-
implemented, might alter the balance of power istration will creatively short-circuit the full leg-
somewhat. The Obama administration has, for islative process. For many employers, the costs
example, floated a proposal to revamp the way of unionization are substantial, and thus the
the government allocates federal contracts to benefits of continuing inaction are clear. Unions
companies. The proposal would prioritize firms often reduce flexibility in hiring and firing de-
that offer high wages while penalizing those that cisions, may slow managers’ abilities to shift re-
had committed labor violations, thereby giving sources and capital as soon as opportunities arise,
an edge to unionized companies and benefiting and substantially reduce managerial discretion
millions of nonunion workers by providing an in setting pay, all the while increasing wage and
incentive to nonunion firms to raise wages and benefit bills. Strong employer opposition has
improve treatment of workers. An estimated helped push unionization down to levels unseen
one in four workers is employed by a company since before the Great Depression. Because such
that contracts with the government, so the scale declines are self-perpetuating, at this point, it
of the regulatory change could be enormous. will take decisive legal and institutional action to
Importantly, the administration is exploring reverse or even halt the trend—action that, if not
ways to change regulations through executive taken soon, won’t have much of a constituency
order, thus avoiding difficulties in generating a behind it any longer. The simple fact: Big Labor
filibuster-proof majority in the Senate. cannot get much smaller.
 Granovetter—The Strength of Weak Ties • 653

Social Capital, Networks, and Attainment

74 • Mark S. Granovetter

The Strength of Weak Ties

Most intuitive notions of the “strength” of an with B depends (in part) on the amount A spends
interpersonal tie should be satisfied by the fol- with B and C, respectively. (If the events “A is
lowing definition: the strength of a tie is a (prob- with B” and “A is with C” were independent,
ably linear) combination of the amount of time, then the event “C is with A and B” would have
the emotional intensity, the intimacy (mutual probability equal to the product of their proba-
confiding), and the reciprocal services which bilities. For example, if A and B are together 60%
characterize the tie. Each of these is somewhat of the time, and A and C 40%, then C, A, and B
independent of the other, though the set is ob- would be together 24% of the time. Such inde-
viously highly intracorrelated. Discussion of op- pendence would be less likely after than before
erational measures of and weights attaching to B and C became acquainted.) If C and B have
each of the four elements is postponed to future no relationship, common strong ties to A will
empirical studies. It is sufficient for the present probably bring them into interaction and gener-
purpose if most of us can agree, on a rough in- ate one. Implicit here is Homans’s idea that “the
tuitive basis, whether a given tie is strong, weak, more frequently persons interact with one an-
or absent. other, the stronger their sentiments of friendship
Consider, now, any two arbitrarily selected for one another are apt to be” (1950, p. 133).
individuals—call them A and B—and the set, S The hypothesis is made plausible also by em-
= C, D, E, . . . , of all persons with ties to either pirical evidence that the stronger the tie connect-
or both of them. The hypothesis which enables ing two individuals, the more similar they are, in
us to relate dyadic ties to larger structures is: the various ways (Berscheid and Walster 1969, pp.
stronger the tie between A and B, the larger the 69–91; Bramel 1969, pp. 9–16; Brown 1965,
proportion of individuals in S to whom they pp. 71–90; Laumann 1968; Newcomb 1961,
will both be tied, that is, connected by a weak or chap. 5; Precker 1952). Thus, if strong ties con-
strong tie. This overlap in their friendship circles nect A to B and A to C, both C and B, being
is predicted to be least when their tie is absent, similar to A, are probably similar to one another,
most when it is strong, and intermediate when increasing the likelihood of a friendship once
it is weak. they have met. Applied in reverse, these two fac-
The proposed relationship results, first, from tors—time and similarity—indicate why weaker
the tendency (by definition) of stronger ties to A-B and A-C ties make a C-B tie less likely than
involve larger time commitments. If A-B and strong ones: C and B are less likely to interact
A-C ties exist, then the amount of time C spends and less likely to be compatible if they do. . . .

Mark S. Granovetter. “The Strength of Weak Ties,” American Journal of Sociology 78:6 (May 1973), pp. 1361–
1366, 1371–1373, 1378–1380. Copyright © 1973 by The University of Chicago Press. Used by permission
of the University of Chicago Press and the author.
654 • Part VI: Who Gets Ahead?

To derive implications for large networks of Figure 74.1. Forbidden Triad

relations, it is necessary to frame the basic hy-
pothesis more precisely. This can be done by C
investigating the possible triads consisting of
strong, weak, or absent ties among A, B, and any
arbitrarily chosen friend of either or both (i.e.,
some member of the set S, described above). A
thorough mathematical model would do this in
some detail, suggesting probabilities for various A B
types. This analysis becomes rather involved,
however, and it is sufficient for my purpose in any contact of A to any contact of B, and, con-
this paper to say that the triad which is most sequently, from anyone connected indirectly to
unlikely to occur, under the hypothesis stated A to anyone connected indirectly to B. Thus, in
above, is that in which A and B are strongly the study of diffusion, we can expect bridges to
linked, A has a strong tie to some friend C, but assume an important role.
the tie between C and B is absent. This triad is Now, if the stipulated triad is absent, it fol-
shown in Figure 74.1. To see the consequences lows that, except under unlikely conditions, no
of this assertion, I will exaggerate it in what fol- strong tie is a bridge. Consider the strong tie A-B:
lows by supposing that the triad shown never oc- if A has another strong tie to C, then forbidding
curs—that is, that the B-C tie is always present the triad of Figure 74.1 implies that a tie exists
(whether weak or strong), given the other two between C and B, so that the path A-C-B exists
strong ties. Whatever results are inferred from between A and B; hence, A-B is not a bridge. A
this supposition should tend to occur in the de- strong tie can be a bridge, therefore, only if nei-
gree that the triad in question tends to be absent. ther party to it has any other strong ties, unlikely
Some evidence exists for this absence. Analyz- in a social network of any size (though possible
ing 651 sociograms, Davis (1970, p. 845) found in a small group). Weak ties suffer no such re-
that in 90% of them triads consisting of two striction, though they are certainly not automat-
mutual choices and one nonchoice occurred less ically bridges. What is important, rather, is that
than the expected random number of times. If all bridges are weak ties.
we assume that mutual choice indicates a strong In large networks it probably happens only
tie, this is strong evidence in the direction of rarely, in practice, that a specific tie provides
my argument. Newcomb (1961, pp. 160–65) the only path between two points. The bridg-
reports that in triads consisting of dyads express- ing function may nevertheless be served locally.
ing mutual “high attraction,” the configuration In Figure 74.2a, for example, the tie A-B is not
of three strong ties became increasingly frequent strictly a bridge, since one can construct the
as people knew one another longer and better; path A-E-I-B (and others). Yet, A-B is the short-
the frequency of the triad pictured in Figure 74.1 est route to B for F, D, and C. This function is
is not analyzed, but it is implied that processes of clearer in Figure 74.2b. Here, A-B is, for C, D,
cognitive balance tended to eliminate it. and others, not only a local bridge to B, but, in
The significance of this triad’s absence can most real instances of diffusion, a much more
be shown by using the concept of a “bridge”; likely and efficient path. Harary et al. point out
this is a line in a network which provides the that “there may be a distance [length of path]
only path between two points (Harary, Norman, beyond which it is not feasible for u to com-
and Cartwright 1965, p. 198). Since, in general, municate with v because of costs or distortions
each person has a great many contacts, a bridge entailed in each act of transmission. If v does
between A and B provides the only route along not lie within this critical distance, then he will
which information or influence can flow from not receive messages originating with u” (1965,
 Granovetter—The Strength of Weak Ties • 655

Figure 74.2. Local Bridges

E I

F J

C G
A B

D H
(a)

C A B

(b)

(a) Degree 3; (b) Degree 13. Straight line = strong tie; dotted line = weak tie.

p. 159). I will refer to a tie as a “local bridge of broken and the changes in average path length
degree n” if n represents the shortest path be- resulting between arbitrary pairs of points (with
tween its two points (other than itself), and n > some limitation on length of path considered)
2. In Figure 74.2a, A-B is a local bridge of de- can then be computed. The contention here is
gree 3, in 74.2b, of degree 13. As with bridges that removal of the average weak tie would do
in a highway system, a local bridge in a social more “damage” to transmission probabilities
network will be more significant as a connection than would that of the average strong one.
between two sectors to the extent that it is the Intuitively speaking, this means that what-
only alternative for many people—that is, as its ever is to be diffused can reach a larger number
degree increases. A bridge in the absolute sense of people, and traverse greater social distance
is a local one of infinite degree. By the same logic (i.e., path length), when passed through weak
used above, only weak ties may be local bridges. ties rather than strong. If one tells a rumor to
Suppose, now, that we adopt Davis’s sugges- all his close friends, and they do likewise, many
tion that “in interpersonal flows of most any sort will hear the rumor a second and third time,
the probability that ‘whatever it is’ will flow from since those linked by strong ties tend to share
person i to person j is (a) directly proportional friends. If the motivation to spread the rumor is
to the number of all-positive (friendship) paths dampened a bit on each wave of retelling, then
connecting i and j; and (b) inversely proportional the rumor moving through strong ties is much
to the length of such paths” (1969, p. 549). The more likely to be limited to a few cliques than
significance of weak ties, then, would be that that going via weak ones; bridges will not be
those which are local bridges create more, and crossed. . . .
shorter, paths. Any given tie may, hypothetically, I will develop this point empirically by citing
be removed from a network; the number of paths some results from a labor-market study I have
656 • Part VI: Who Gets Ahead?

recently completed. Labor economists have long reactivate such ties. It is remarkable that people
been aware that American blue-collar workers receive crucial information from individuals
find out about new jobs more through personal whose very existence they have forgotten. . . .
contacts than by any other method. (Many stud- From the individual’s point of view, then, weak
ies are reviewed by Parnes 1954, chap. 5.) Recent ties are an important resource in making possible
studies suggest that this is also true for those in mobility opportunity. Seen from a more macro-
professional, technical, and managerial positions scopic vantage, weak ties play a role in effecting
(Shapero, Howell, and Tombaugh 1965; Brown social cohesion. When a man changes jobs, he is
1967; Granovetter 1970). My study of this ques- not only moving from one network of ties to an-
tion laid special emphasis on the nature of the tie other, but also establishing a link between these.
between the job changer and the contact person Such a link is often of the same kind which facil-
who provided the necessary information. itated his own movement. Especially within pro-
In a random sample of recent professional, fessional and technical specialties which are well
technical, and managerial job changers living defined and limited in size, this mobility sets up
in a Boston suburb, I asked those who found elaborate structures of bridging weak ties between
a new job through contacts how often they saw the more coherent clusters that constitute opera-
the contact around the time that he passed on tive networks in particular locations.
job information to them. I will use this as a
measure of tie strength. A natural a priori idea REFERENCES
is that those with whom one has strong ties are Berscheid, E., and E. Walster. 1969. Interpersonal At-
traction. Reading, Mass.: Addison-Wesley.
more motivated to help with job information.
Bramel, D. 1969. “Interpersonal Attraction, Hos-
Opposed to this greater motivation are the struc-
tility and Perception.” In Experimental Social
tural arguments I have been making: those to Psychology, edited by Judson Mills. New York:
whom we are weakly tied are more likely to move Macmillan.
in circles different from our own and will thus Brown, David. 1967. The Mobile Professors. Wash-
have access to information different from that ington, D.C.: American Council on Education.
which we receive. Brown, Roger. 1965. Social Psychology. New York:
I have used the following categories for fre- Free Press.
quency of contact: often = at least twice a week; Davis, James A. 1969. “Social Structures and Cogni-
occasionally = more than once a year but less tive Structures.” In R. P. Abelson et al., Theories
of Cognitive Consistency. Chicago: Rand McNally.
than twice a week; rarely = once a year or less.
———. 1970. “Clustering and Hierarchy in Inter-
Of those finding a job through contacts, 16.7% personal Relations.” American Sociological Review
reported that they saw their contact often at the 35 (October): 843–52.
time, 55.6% said occasionally, and 27.8% rarely Granovetter, M. S. 1970. “Changing Jobs: Channels
(N = 54). The skew is clearly to the weak end of of Mobility Information in a Suburban Commu-
the continuum, suggesting the primacy of struc- nity.” Doctoral dissertation, Harvard University.
ture over motivation. Harary, F., R. Norman, and D. Cartwright. 1965.
In many cases, the contact was someone only Structural Models. New York: Wiley.
marginally included in the current network of Homans, George. 1950. The Human Group. New
York: Harcourt, Brace & World.
contacts, such as an old college friend or a for-
Laumann, Edward. 1968. “Interlocking and Radial
mer workmate or employer, with whom sporadic
Friendship Networks: A Cross-sectional Analy-
contact had been maintained (Granovetter 1970, sis.” Mimeographed. Ann Arbor: University of
pp. 76–80). Usually such ties had not even been Michigan.
very strong when first forged. For work-related Newcomb, T. M. 1961. The Acquaintance Process.
ties, respondents almost invariably said that they New York: Holt, Rinehart & Winston.
never saw the person in a nonwork context. Parnes, Herbert. 1954. Research on Labor Mobility.
Chance meetings or mutual friends operated to New York: Social Science Research Council.
 Lin—Social Networks and Status Attainmment • 657

Precker, Joseph. 1952. “Similarity of Valuings as a Shapero, Albert, Richard Howell, and James Tom-
Factor in Selection of Peers and Near-Authority baugh. 1965. The Structure and Dynamics of the
Figures.” Journal of Abnormal and Social Psychol- Defense R & D Industry. Menlo Park, Calif.: Stan-
ogy 47, suppl. (April): 406–14. ford Research Institute.

75 • Nan Lin

Social Networks and Status Attainment

Status attainment can be understood as a pro- expansions. All subsequent theoretical revisions
cess by which individuals mobilize and invest re- and expansions must be evaluated for their con-
sources for returns in socioeconomic standings. tribution to the explanation of status attainment
These resources can be classified into two types: beyond those accounted for by the Blau-Duncan
personal resources and social resources. Personal paradigm (Kelley 1990; Smith 1990). Several
resources are possessed by the individual who lines of contributions since then, including the
can use and dispose of them with freedom and addition of sociopsychological variables (Sewell
without much concern for compensation. Social and Hauser 1975), the recasting of statuses into
resources are resources accessible through one’s classes (Wright 1979; Goldthorpe 1980), the
direct and indirect ties. The access to and use of incorporation of “structural” entities and posi-
these resources are temporary and borrowed. For tions as both contributing and attained statuses
example, a friend’s occupational or authority po- (Baron and Bielby 1980; Kalleberg 1988), and
sition, or such positions of this friend’s friends, the casting of comparative development or in-
may be ego’s social resource. The friend may use stitutions as contingent conditions (Treiman
his or her position or network to help ego to find 1970), have significantly amplified rather than
a job. These resources are “borrowed” and useful altered the original Blau-Duncan conclusion
to achieve ego’s certain goal, but they remain the concerning the relative merits of achieved versus
property of the friend or his or her friends. ascribed personal resources in status attainment.
The theoretical and empirical work for un- In the last three decades, a research tradition
derstanding and assessing the status attainment has focused on the effects on attained statuses
process can be traced to the seminal study re- of social resources. The principal proposition
ported by Blau and Duncan (1967). Their is that social resources exert an important and
major conclusion was that, even accounting for significant effect on attained statuses, beyond
both the direct and indirect effects of ascribed that accounted for by personal resources. Sys-
status (parental status), achieved status (educa- tematic investigations of this proposition have
tion and prior occupational status) remained included efforts in (1) developing theoretical ex-
the most important factor accounting for the planations and hypotheses, (2) developing mea-
ultimate attained status. The study thus set the surements for social resources, (3) conducting
theoretical baseline for further modifications and empirical studies verifying the hypotheses, and

Nan Lin. “Social Networks and Status Attainment,” Annual Review of Sociology 25 (1999), pp. 467–470, 485–
487. Copyright © 1999 by Annual Review of Sociology. Permission from Annual Reviews conveyed through
Copyright Clearance Center, Inc.
658 • Part VI: Who Gets Ahead?

(4) assessing the relative importance of social re- ties and status attainment might be indirect: The
sources as compared to personal resources in the strength of weak ties might lie in their accessing
process of status attainment. . . . social positions vertically higher in the social hi-
Contributions of social network analysis to erarchy, which had the advantage in facilitating
status attainment can be traced to the seminal the instrumental action. Second, the study im-
study conducted by Mark Granovetter (1974), plicated behavior rather than a paper-and-pencil
who interviewed 282 professional and manage- exercise, as each step in the packet-forwarding
rial men in Newton, Massachusetts. The data process required actual actions from each par-
suggested that those who used interpersonal ticipant. Thus, the study results lend behavioral
channels seemed to land more satisfactory and validity to those found in previous status attain-
better (e.g., higher income) jobs. Inferring from ment paper-pencil studies.
this empirical research, substantiated with a re- Based on these studies, a theory of social
view of job search studies, Granovetter proposed resources has emerged (Lin 1982, 1990). The
(1973) a network theory for information flow. theory begins with an image of the macro-social
The hypothesis of “the strength of weak ties” structure consisting of positions ranked accord-
was that weaker ties tend to form bridges that ing to certain normatively valued resources such
link individuals to other social circles for infor- as wealth, status, and power. This structure has
mation not likely to be available in their own a pyramidal shape in terms of accessibility and
circles, and such information should be useful control of such resources: The higher the posi-
to the individuals. tion, the fewer the occupants; and the higher the
However, Granovetter never suggested that position, the better the view it has of the struc-
access to or help from weaker rather than stron- ture (especially down below). The pyramidal
ger ties would result in better statuses of jobs structure suggests advantages for positions nearer
thus obtained (1995:148). Clues about the link- to the top, both in terms of number of occupants
age between strength of ties and attained statuses (fewer) and accessibility to positions (more). In-
came indirectly from a small world study con- dividuals within these structural constraints and
ducted in a tri-city metropolitan area in upstate opportunities take actions for expressive and in-
New York (Lin, Dayton, and Greenwald 1978). strumental purposes. For instrumental actions
The task of the participants in the study was to (attaining status in the social structure being one
forward packets containing information about prime example), the better strategy would be for
certain target persons to others they knew on a ego to reach toward contacts higher up in the
first-name basis so that the packets might even- hierarchy. These contacts would be better able
tually reach the target persons. The study found to exert influence on positions (e.g., recruiter
that successful chains (those packets successfully for a firm) whose actions may benefit ego’s inter-
forwarded to the targets) involved higher-status est. This reaching-up process may be facilitated
intermediaries until the last nodes (dipping if ego uses weaker ties, because weaker ties are
down in the hierarchy toward the locations of the more likely to reach out vertically (presumably
targets). Successful chains also implicated nodes upward) rather than horizontally relative to ego’s
that had more extensive social contacts (who position in the hierarchy.
claimed more social ties), and yet these tended
to forward the packets to someone they had not REFERENCES
seen recently (weaker ties). The small world study Baron, J. N., and W. T. Bielby. 1980. “Bringing the
Firm Back in: Stratification, Segmentation, and
thus made two contributions. First, it suggested
the Organization of Work.” American Sociological
that access to hierarchical positions might be the
Review 45:737–65.
critical factor in the process of status attainment. Blau, P. M., and O. D. Duncan. 1967. The American
Thus, the possible linkage between strength of Occupational Structure. New York: Wiley.
 Burt—Structural Holes • 659

Goldthorpe, J. H. 1980. Social Mobility and Class ———. 1990. “Social Resources and Social Mobil-
Structure in Modern Britain. New York: Oxford ity: A Structural Theory of Status Attainment.”
University Press. Pages 247–71 in Social Mobility and Social Struc-
Granovetter, M. 1973. “The Strength of Weak Ties.” ture, edited by R. L. Breiger. New York: Cam-
American Journal of Sociology 78:1360–80. bridge University Press.
———. 1974. Getting a Job. Cambridge, MA: Har- Lin, N., P. Dayton, and P. Greenwald. 1978. “Ana-
vard University Press. lyzing the Instrumental Use of Relations in the
———. 1995. Getting a Job. Rev. ed. Chicago: Uni- Context of Social Structure.” Sociological Methods
versity of Chicago Press. and Research 7:149–66.
Kalleberg, A. 1988. “Comparative Perspectives on Sewell, W. H., and R. M. Hauser. 1975. Education,
Work Structures and Inequality.” Annual Review Occupation & Earnings: Achievement in the Early
of Sociology 14:203–25. Career. New York: Academic Press.
Kelley, J. 1990. “The Failure of a Paradigm: Smith, M. R. 1990. “What Is New in New Structur-
Log-Linear Models of Social Mobility.” Pages alist Analyses of Earnings?” American Sociological
319–46, 349–57 in John H. Goldthorpe: Con- Review 55:827–41.
sensus and Controversy, edited by J. Clark, C. Treiman, D. J. 1970. “Industrialization and Social
Modgil, and S. Modgil. London: Falmer. Stratification.” Pages 207–34 in Social Stratifica-
Lin, N. 1982. “Social Resources and Instrumental tion: Research and Theory for the 1970s, edited by
Action.” Pages 131–45 in Social Structure and E. O. Laumann. Indianapolis: Bobbs-Merrill.
Network Analysis, edited by P. V. Marsden and N. Wright, E. O. 1979. Class Structure and Income De-
Lin. Beverly Hills, CA: Sage Publications. termination. New York: Academic Press.

76 • Ronald S. Burt

Structural Holes

Some people enjoy higher incomes than others. necessary to success, it is useless without the so-
Some are promoted faster. Some are leaders on cial capital of opportunities in which to apply it.
more important projects. The human capital Social capital can be distinguished in its eti-
explanation is that inequality results from dif- ology and consequences from human capital
ferences in individual ability. The usual evidence (e.g., Coleman, 1990; Bourdieu and Wacquant,
is on general populations, as is Becker’s (1975) 1992; Burt, 1992; Putnam, 1993; Lin, 1998).
pioneering analysis of income returns to edu- With respect to etiology, social capital is a qual-
cation, but the argument is widely applied by ity created between people, whereas human
senior managers to explain who gets to the top capital is a quality of individuals. Investments
of corporate America—managers who make it to that create social capital are therefore different
the top are smarter or better educated or more in fundamental ways from the investments that
experienced. But, while human capital is surely create human capital (Coleman, 1988, 1990). I

Ronald S. Burt. “The Contingent Value of Social Capital,” Administrative Science Quarterly 42 (June 1997),
pp. 339–342, 363–365. Administrative Science Quarterly by Cornell University; Johnson Graduate School of
Management (Cornell University). Copyright © 1997. Reproduced with permission of Sage Publications, Inc.,
Journals via Copyright Clearance Center.
660 • Part VI: Who Gets Ahead?

focus in this paper on consequences, a focus in there can be multiple prices because disconnec-
network analysis for many years (Breiger, 1995). tions between individuals, holes in the structure
With respect to consequences, social capital is of the market, leave some people unaware of the
the contextual complement to human capital. benefits they could offer one another. Certain
Social capital predicts that returns to intelli- people are connected to certain others, trusting
gence, education, and seniority depend in some certain others, obligated to support certain oth-
part on a person’s location in the social structure ers, dependent on exchange with certain others.
of a market or hierarchy. While human capital Assets get locked into suboptimal exchanges. An
refers to individual ability, social capital refers individual’s position in the structure of these ex-
to opportunity. Some portion of the value a changes can be an asset in its own right. That
manager adds to a firm is his or her ability to asset is social capital, in essence, a story about
coordinate other people: identifying opportu- location effects in differentiated markets. The
nities to add value within an organization and structural hole argument defines social capital in
getting the right people together to develop the terms of the information and control advantages
opportunities. Knowing who, when, and how to of being the broker in relations between people
coordinate is a function of the manager’s net- otherwise disconnected in social structure. The
work of contacts within and beyond the firm. disconnected people stand on opposite sides of a
Certain network forms deemed social capital can hole in social structure. The structural hole is an
enhance the manager’s ability to identify and de- opportunity to broker the flow of information
velop opportunities. Managers with more social between people and control the form of projects
capital get higher returns to their human capital that bring together people from opposite sides
because they are positioned to identify and de- of the hole.
velop more rewarding opportunities.

Information Benefits
The Network Structure of Social The information benefits are access, timing, and
Capital referrals. A manager’s network provides access to
Structural hole theory gives concrete meaning information well beyond what he or she could
to the concept of social capital. The theory de- process alone. It provides that information early,
scribes how social capital is a function of bro- which is an advantage to the manager acting on
kerage opportunities in a network (see Burt, the information. The network that filters infor-
1992, for detailed discussion). The structural mation coming to a manager also directs, con-
hole argument draws on several lines of network centrates, and legitimates information received by
theorizing that emerged in sociology during others about the manager. Through referrals, the
the 1970s, most notably, Granovetter (1973) manager’s interests are represented in a positive
on the strength of weak ties, Freeman (1977) light, at the right time, and in the right places.
on betweenness centrality, Cook and Emerson The structure of a network indicates the
(1978) on the power of having exclusive ex- redundancy of its information benefits. There
change partners, and Burt (1980) on the struc- are two network indicators of redundancy. The
tural autonomy created by network complexity. first is cohesion. Cohesive contacts—contacts
More generally, sociological ideas elaborated strongly connected to each other—are likely
by Simmel (1955) and Merton (1968), on the to have similar information and therefore pro-
autonomy generated by conflicting affiliations, vide redundant information benefits. Structural
are mixed in the structural hole argument with equivalence is the second indicator. Equivalent
traditional economic ideas of monopoly power contacts—contacts who link a manager to the
and oligopoly to produce network models of same third parties—have the same sources of
competitive advantage. In a perfect market, one information and therefore provide redundant
price clears the market. In an imperfect market, information benefits.
 Burt—Structural Holes • 661

Nonredundant contacts offer in- Figure 76.1. Illustrative Manager’s Networks

formation benefits that are additive
rather than redundant. Structural
holes are the gaps between nonredun- 1

dant contacts (see Burt, 1992: 25–30, 2

on how Granovetter’s weak ties gener-
alize to structural holes). The hole is a 3 JAMES
buffer, like an insulator in an electric
circuit. A structural hole between two 4

clusters in a network need not mean 5

that people in the two clusters are un-

aware of one another. It simply means
that they are so focused on their own
1
activities that they have little time to
2
attend to the activities of people in
the other cluster. A structural hole
ROBERT
indicates that the people on either 3

side of the hole circulate in different 4

flows of information. A manager who 5
spans the structural hole, by having
strong relations with contacts on both
sides of the hole, has access to both
information flows. The more holes
spanned, the richer the information
Thick lines represent a manager’s direct contacts.
benefits of the network.
Figure 76.1 provides an example.
James had a network that spanned one structural he will be informed of opportunities and im-
hole. The hole is the relatively weak connection pending disasters (access benefits). Further, since
between the cluster reached through contacts 1, Robert’s contacts are only linked through him at
2, and 3 and the cluster reached through con- the center of the network, he is the first to see
tacts 4 and 5. Robert took over James’s job and new opportunities created by needs in one group
expanded the social capital associated with the that could be served by skills in other groups
job. He preserved connection with both clusters (timing benefits). He stands at the crossroads of
in James’s network but expanded the network to social organization. He has the option of bring-
a more diverse set of contacts. Robert’s network, ing together otherwise disconnected individuals
with the addition of three new clusters of people, in the network when it would be rewarding. And
spans ten structural holes. because Robert’s contacts are more diverse, he
Information benefits in this example are en- is more likely to be a candidate for inclusion in
hanced in several ways. The volume is higher new opportunities (referral benefits). These ben-
in Robert’s network simply because he reaches efits are compounded by the fact that having a
more people indirectly. Also, the diversity of his network that yields such benefits makes Robert
contacts means that the quality of his informa- more attractive to other people as a contact in
tion benefits is higher. Each cluster of contacts their own networks.
is a single source of information because people
connected to one another tend to know the same
things at about the same time. Nonredundant Control Benefits
clusters provide Robert with a broader informa- The manager who creates a bridge between
tion screen and, therefore, greater assurance that otherwise disconnected contacts has a say in
662 • Part VI: Who Gets Ahead?

whose interests are served by the bridge. The effectively than it can be monitored bureaucrati-
disconnected contacts communicate through the cally. They move information faster, and to more
manager, giving the manager an opportunity to people, than memos. These entrepreneurial
adjust his or her image with each contact, which managers know the parameters of organization
is the structural foundation for managerial ro- problems early. They are highly mobile relative
bust action (Padgett and Ansell, 1993). Simmel to people working through a bureaucracy, eas-
and Merton introduced the sociology of peo- ily shifting network time and energy from one
ple who derive control benefits from structural solution to another. More in control of their im-
holes: The ideal type is the tertius gaudens (lit- mediate surroundings, entrepreneurial managers
erally, “the third who benefits”), a person who tailor solutions to the specific individuals being
benefits from brokering the connection between coordinated, replacing the boiler-plate solutions
others (see Burt, 1992: 30–32, for review). As of formal bureaucracy. There is also the issue of
the broker between otherwise disconnected con- costs: entrepreneurial managers offer inexpensive
tacts, a manager is an entrepreneur in the literal coordination relative to the bureaucratic alterna-
sense of the word—a person who adds value by tive. Managers with networks rich in structural
brokering the connection between others (Burt, holes operate somewhere between the force of
1992: 34–36; see also Martinelli, 1994). There corporate authority and the dexterity of markets,
is a tension here, but not the hostility of combat- building bridges between disconnected parts of
ants. It is merely uncertainty. In the swirling mix the firm where it is valuable to do so. They have
of preferences characteristic of social networks, more opportunity to add value, are expected to
where no demands have absolute authority, the do so, and are accordingly expected to enjoy
tertius negotiates for favorable terms. Structural higher returns to their human capital. The pre-
holes are the setting for tertius strategies, and diction is that in comparisons between otherwise
information is the substance. Accurate, ambig- similar people like James and Robert in Figure
uous, or distorted information is strategically 76.1, it is people like Robert who should be
moved between contacts by the tertius. The more successful.
information and control benefits reinforce one
another at any moment in time and cumulate REFERENCES
together over time. Becker, Gary. 1975. Human Capital, 2d ed. Chicago:
University of Chicago Press.
Networks rich in structural holes present
Bourdieu, Pierre, and Loïc J. D. Wacquant. 1992.
opportunities for entrepreneurial behavior. The
An Invitation to Reflexive Sociology. Chicago: Uni-
behaviors by which managers develop these op- versity of Chicago Press.
portunities are many and varied, but the oppor- Breiger, Ronald L. 1995. “Socioeconomic achieve-
tunity itself is at all times defined by a hole in the ment and social structure.” In Annual Review of
social structure around the manager. In terms of Sociology, 21: 115–136. Palo Alto, CA: Annual
the structural hole argument, networks rich in Reviews.
the entrepreneurial opportunities of structural Burt, Ronald S. 1980. “Autonomy in a social topol-
holes are entrepreneurial networks, and entre- ogy.” American Journal of Sociology, 85: 892–925.
preneurs are people skilled in building the inter- ———. 1992. Structural Holes. Cambridge, MA:
Harvard University Press.
personal bridges that span structural holes.
Coleman, James S. 1988. “Social capital in the cre-
ation of human capital.” American Journal of So-
ciology, 94: S95–S120.
Predicted Social Capital Effect
———. 1990. Foundations of Social Theory. Cam-
Managers with contact networks rich in struc- bridge, MA: Harvard University Press.
tural holes know about, have a hand in, and Cook, Karen S., and Richard M. Emerson. 1978.
exercise control over the more rewarding op- “Power, equity and commitment in exchange
portunities. They monitor information more networks.” American Sociological Review, 43:
712–739.
 Fernandez and Fernandez-Mateo—Networks, Race, and Hiring • 663

Freeman, Linton C. 1977. “A set of measures of Merton, Robert K. 1968. “Continuities in the the-
centrality based on betweenness.” Sociometry, 40: ory of reference group behavior.” In Robert K.
35–40. Merton, Social Theory and Social Structure: 335–
Granovetter, Mark S. 1973. “The strength of 440. New York: Free Press.
weak ties.” American Journal of Sociology, 78: Padgett, John F., and Christopher K. Ansell. 1993.
1360–1380. “Robust action and the rise of the Medici,
Lin, Nan. 1998. Social Resources and Social Ac- 1400–1434.” American Journal of Sociology, 98:
tion. New York: Cambridge University Press 1259–1319.
(forthcoming). Putnam, Robert D. 1993. Making Democracy Work:
Martinelli, Alberto. 1994. “Entrepreneurship and Civic Traditions in Modern Italy. Princeton, NJ:
management.” In Neil J. Smelser and Richard Princeton University Press.
Swedberg (eds.), The Handbook of Economic So- Simmel, Georg. 1955. Conflict and the Web of Group
ciology: 476–503. Princeton, NJ: Princeton Uni- Affiliations. Trans. by Kurt H. Wolff and Rein-
versity Press. hard Bendix. New York: Free Press.

77 • Roberto M. Fernandez and Isabel Fernandez-Mateo

Networks, Race, and Hiring

It is common for scholars interested in race and There is something slippery about the way
poverty to invoke a lack of access to job networks these network arguments are currently being
as a reason why minorities face difficulties in the used, however. Because “wrong network” is de-
labor market (e.g., Royster 2003; Wilson 1996). fined in terms of the eventual outcome (a net-
Previous studies on this issue, however, have work is “good” if it leads to a good outcome,
produced mixed results. Minorities have been otherwise, it is a “wrong network”), such expla-
found to be more likely to have obtained their nations run the danger of circular reasoning. To
job through networks than nonminorities (e.g., give network accounts of minority underper-
Elliott 1999). Yet, these jobs pay less than jobs formance analytical bite, we need to specify the
obtained by other means (e.g., Falcon 1995). mechanisms by which minorities are “excluded”
Rather than exclusion from white networks from productive networks or “stuck” in unpro-
(e.g., Royster 2003), the emphasis in the liter- ductive ethnic networks.
ature has shifted to minorities’ over-reliance on We argue that being “stuck” in the “wrong
ethnic networks. Thus, the imagery that emerges network” can be produced by minority un-
from these studies is that minorities are stuck in derrepresentation in any of a number of steps
the “wrong networks,” that is, those that lead to in the recruitment and hiring process. Using
low-wage jobs. unique data from one employer, we illustrate

The ideas, issues, and theories considered in this brief article are examined in greater depth in the following
publication: Roberto M. Fernandez and Isabel Fernandez-Mateo, “Networks, Race, and Hiring,” American
Sociological Review 71 (2006), pp. 42–71. The article as printed here was originally prepared by Roberto M.
Fernandez and Isabel Fernandez-Mateo for Social Stratification: Class, Race, and Gender in Sociological Perspec-
tive, Third Edition, edited by David B. Grusky, pp. 587–595. Copyright © 2008 by Westview Press.
664 • Part VI: Who Gets Ahead?

the mechanisms by which minorities can be

isolated from good job opportunities. To avoid “Wrong Networks”
circular reasoning, we form proper baselines of A key component of understanding whether
comparison using data on both networked and minorities are cut off from employment op-
nonnetworked minorities and nonminorities at portunities is to understand why they may be
each stage of the hiring pipeline and identify the underrepresented in networks that lead to good
specific points in the process where network fac- jobs. That is, in order to attribute logically the
tors could lead minorities to have less access to exclusion of minorities to the absence of net-
these desirable jobs than nonminorities. work ties to good jobs, we would need to feel
confident that minorities might plausibly have
been hired except for the lack of the contact. It
Race and Networks in the Labor is critical to define the various processes whereby
Market network factors could limit minorities’ access
A common argument in sociology is that jobs to desirable jobs. In our conceptualization, the
found through networks pay better and are of “wrong network” account is consistent with
higher status than those found through formal underrepresentation of minorities in any of a
channels (Lin 2002). Evidence on this issue, number of stages in the recruitment and hiring
however, is mixed, especially for minority process. Figure 77.1 represents a conceptual map
groups. Reingold (1999) suggests that social of the ways in which networks might affect the
networks lead to racial insularity and contribute various stages. These are separated into two sets
to the economic marginalization of minorities. of processes, referring and screening.
However, since many of these studies analyze The referral process may contribute to mi-
samples of job incumbents, they often suffer norities’ isolation either if there are no minority
from causal ambiguity (do ties to higher status employees available to refer in the pool of work-
people cause superior labor market outcomes, ers (as in Kasinitz and Rosenberg’s [1996] ac-
or is it that people with superior labor market count), or if these employees are reluctant to
outcomes gain access to high-status people?). In pass on information about good jobs (see Smith
order to avoid the causal ambiguity problem, a 2005). Even if there are potential referrers who
number of studies use samples of job seekers and are willing to refer someone, minorities could
examine the chances of obtaining employment still be cut off if these referrers were not to refer
for various search methods (e.g., Elliott 1999). minority applicants (step 1c.). This could hap-
Employer surveys (e.g., Holzer 1996) are an al- pen if job referral networks are less than perfectly
ternative way of studying this issue, by fleshing homophilous by race (see Rubineau and Fernan-
out the employer side of the hiring process. dez 2005). If all these conditions are met (steps
Neither of these approaches, however, exam- 1a.–1c.), there will be a set of networked mi-
ines actual hiring processes and their role in so- nority candidates in the applicant pool. At this
cial isolation from good jobs. Without baseline point the screening stage on the demand side of
information on the presence or absence of social the hiring interface begins.
ties for the pool of competing applicants, some The effect of screening processes on minori-
of whom are hired and others are not, it is impos- ties’ access to desirable jobs depends on the em-
sible to identify the effect of social contacts on ployer’s attitude towards referrals. If firms prefer
hiring per se. In order to address this issue, some to recruit employee referrals (Fernandez, Cas-
authors have used single-firm screening studies tilla, and Moore 2000), they will tend to hire
(e.g., Fernandez, Castilla, and Moore 2000; Fer- minority workers in proportion to their repre-
nandez and Sosa 2005), but most of this research sentation in the networked pool of applicants. In
has not analyzed the role of race in hiring due to this case, minorities would thus be cut off from
lack of appropriate data (a prominent exception good jobs only if they are underrepresented
is Petersen, Saporta, and Seidel 2000). in the networked applicant pool. However, if
 Fernandez and Fernandez-Mateo—Networks, Race, and Hiring • 665

Figure 77.1. Steps in Network Processes that Connect Candidates to Jobs

employers avoid hiring through networks (see September 1997 to November 30, 2000. We
Ullman 1966), being well represented in the pool coded data on applicants’ education, work his-
of referred applicants is bad news for minorities. tory, and other human capital characteristics and
Finally, if the employer is neutral toward hiring geocoded candidates to the addresses they listed
referrals, access to desirable jobs for minorities on the application form. Most importantly, and
will be provided in step 2 (screening) in propor- unusually, we have data on the applicants’ race.
tion to the overall applicant pool, as opposed to All applicants must turn in their form in person,
the networked applicant pool. In such case the and when they do so the receptionist at the com-
employers’ preferences would have no effect in pany (who is the same one for the whole period)
cutting off minorities from desirable jobs. records the applicant’s apparent race and gender.
Race is thus not self-identified as in most surveys
and is well suited to our purposes of understand-
Analyses ing the role of race in the screening process.
We illustrate these processes using unique data As mentioned above, in order to ensure that
on employees working at one company site and we are not misattributing the effect of networks
trace their networks of job contacts to applicants to the effect of human capital factors, we must
for desirable entry-level jobs. Minorities account show that the candidates for these jobs might
for 50 percent of employees at this site, with plausibly have been hired even without net-
Asian Americans and Hispanics being the largest work ties. In this setting, it is clear that these
groups. We collected all 2,065 paper applications entry-level jobs are within reach for people with
to the plant’s entry-level production jobs from modest education and labor force experience.
666 • Part VI: Who Gets Ahead?

The median years of education and experience We address step 1a in Figure 77.1 by studying
for the people hired into these jobs are 12 and the race and gender distribution of employees
7.9, respectively. Moreover, the local labor mar- who could have produced referral applications
ket was experiencing high rates of unemploy- during the period. For both genders, over half
ment during the period (between 6.76 and 14.1 the workers are minorities (this percentage in-
percent). Data from the local labor market (see creases to over 60 percent for entry-level em-
Table 77.1) show that the wages offered by the ployees). Relatively large percentages of Asian
firm were attractive—particularly for females Americans and Hispanics, but more modest per-
and minorities. Starting wages were $7.75/hr for centages of African Americans, are represented.
the first eight months of the study, and $8.05 Women make up 69.4 percent of entry-level
afterwards. These wages fell in the 25th and 27th workers, and they are overrepresented irrespec-
percentiles of the overall wage distribution for tive of race.
males in the area (35th and 36th percentiles for We also compare these race and gender dis-
females). For whites, these starting wages fell in tributions of workers with the composition of
the 18th and 19th percentiles of the male wage persons employed in the local area (Table 77.2).
distribution, while they were more attractive for We find that white workers are underrepre-
minority males (27th and 28th percentiles for sented in this factory compared to their pro-
African Americans; 35th and 36th for Hispanics; portion in the area. Asian Americans, however,
32nd and 35th for Asian Americans). The pat- are extremely overrepresented. While they make
tern is similar for females, although for them the up 6.1 percent of employed males in the local
wages are even more attractive than for males in labor market, they account for 26.8 percent of
every case. Furthermore, most hires receive pay male employees. Hispanics are slightly under-
raises after a 90-day probation period (to $8.50, represented (24.9 percent of employees, 35.3
$10.05, or $10.35), which makes wages even percent in the area), while the percentages of
more attractive. In fact, the top wage approaches African Americans employed at the factory are
the median of the wage distribution for white quite similar to their proportion in the area (5.9
and African American women in the area and versus 4.0 percent for males). In sum, minorities
well exceeds the median for Hispanic and Asian are definitely available to refer in this setting.
American females.

Do Minority Employees Refer?

Are Minority Employees Available to Of the 557 employees, 200 of them originated a
Refer? total of 580 applications. Asian Americans refer
In order to study the role of job-referral net- the most (50.9 percent of male Asian Americans
works in the application process, we distinguish originated at least one referral applicant). Inter-
networked candidates using data from the orig- estingly, whites show the lowest rates of produc-
inal application forms. Employee referrals made ing referrals (27.7 percent for females and 18.8
up 30.2 percent of the applications for which we percent for males). To determine whether back-
could identify recruitment source (2,556 out of ground factors might account for the observed
2,605). Using the company’s employee database, race differences in referral behavior, we estimate
we have been able to link referrers with their re- a set of negative binomial regressions (see Fer-
ferral for 83.7 percent of the “employee referral” nandez and Fernandez-Mateo 2006, Table 5).
applications. For studying who among company These models show that there are no significant
employees produces referrals, we use the firm’s gender differences in the counts of referrals orig-
personnel records. Five hundred fifty-seven inated by employees at the company. There are
workers were employed at the focal plant at any race differences, however, as minorities generate
point over the study period, and for these we more referrals than whites (with Asian Ameri-
coded their self-identified race and gender. cans producing the most). These effects remain
 Fernandez and Fernandez-Mateo—Networks, Race, and Hiring • 667

Table 77.1. Percentile Ranks of Company Offered Wages in Wage Distribution of All Non-College
Population in Local Metro Area by Racial Group

All Racial Groups Non-Hispanic White African American

Male Female Male Female Male Female
Starting Wages (N = 192)
Hired 9/97–4/98
8.3% ($7.75) 25 35 18 30 27 37
Hired 5/98–11/00
91.7% ($8.05) 27 36 19 31 28 38

Wages after 90-Day

Probation (N = 109)
15.6% ($8.05) 27 36 19 31 28 38
62.4% ($8.50) 29 40 21 34 32 41
21.1% ($10.05) 37 51 27 45 40 48
0.1% ($10.35) 38 52 28 46 41 48

Hispanic Asian American All Minorities

Male Female Male Female Male Female
Starting Wages (N = 192)
Hired 9/97–4/98
8.3% ($7.75) 35 43 32 39 33 41
Hired 5/98–11/00
91.7% ($8.05) 36 44 35 41 35 42

Wages after 90-Day

Probation (N = 109)
15.6% ($8.05) 36 44 35 41 35 42
62.4% ($8.50) 39 49 37 45 37 46
21.1% ($10.05) 49 61 43 55 46 58
0.1% ($10.35) 50 62 45 57 48 59
Source: Persons in the U.S. Census Bureau’s 5 percent 2000 Public Use Micro Sample for the local metro area who are at
least 15 years of age and have fewer than 16 years of education with positive wage and salary income in 1999. Data are
weighted to reflect the population.

even when adding the extensive controls for in- relationship between the race of the referrer
dividual background mentioned above. Thus, at and the race of the referral applicant (for both
step 1b in Figure 77.1, we have no evidence that genders). Most important, 61.8 percent of male
minorities are less likely than whites to produce and 57.2 percent of female referred applicants
referral applicants. are minorities. Clearly, by this first criterion, mi-
norities are not cut off at step 1c.
A more stringent criterion would be to assess
Do Minority Employees Refer Minority whether the referral process is reproducing the
Applicants? racial distribution of the referring population
The simplest criterion to assess if this is a point (the 200 employees identified above). We find
of disconnection for minorities is whether that whites are the most insular of the groups in
there are any minorities at all produced by the referring (76.9 percent of the referrals produced
referral process, irrespective of the race of the by white employees are white). The percentage
referrer. We analyze the race distribution of of African Americans and Asian Americans in
applicants produced by referrers of different ra- the referral pool matches the percentage of the
cial backgrounds. We find that there is a strong referring population quite closely (5.6 vs. 4.0
 668 • Part VI: Who Gets Ahead?

Table 77.2. Racial and Gender Distributions of Workers Employed During Hiring Window (September
1997–November 2000), and Persons Employed in the Metropolitan Area

All Plant Employeesa Entry Level Employees Only 2000 PUMSb

Female Male Female Male Female Male
Non-Hispanic White 44.0 41.5 38.5 28.6 54.3 50.4
African American 3.1 5.9 3.1 5.6 5.0 4.0
Hispanic 28.7 24.9 30.1 28.6 29.8 35.3
Asian American 23.3 26.8 27.6 35.7 6.5 6.1
Native American 0.9 1.0 0.7 1.6 0.9 0.7
Other, Multirace — — — — — —
Total 100.0 100.0 100.0 100.0 100.0 100.0
Total N 352 205 286 126 174838 208174
a
Each person is counted equally if they were employed at any time during the hiring window (September 1, 1997–Novem-
ber 30, 2000).
b
Persons in the 5 percent 2000 PUMS who are at least 15 years of age and less than 16 years of education with positive wage
and salary income in 1999. Data are weighted to reflect the population.

percent for African Americans; 34.6 vs. 32.5 per- with HR managers at this site suggest that the
cent for Asian Americans). Hispanics, however, employer has neither a strong preference nor dis-
constitute a third of referrers, but only 19 per- taste for referred applicants. We assembled data
cent of referrals. Hispanics, therefore, seem to on all applicants for entry-level jobs and tracked
be somehow cut off, in relative terms, from the their progress through the hiring pipeline (from
networks leading to these jobs. application, to interview, to offer, to hire).
A final criterion depends on whether mi- For females (see Figure 77.2a), there is little
norities refer minority applicants less than they evidence of a preference for networked candi-
refer white applicants. We find, however, that dates (33.9 percent of female applicants and
irrespective of which minority group one con- 35.3 percent of female hires were referrals).
siders, the percentage referring whites is lower Also, the race distribution did not change much
than the percentage referring minorities. By this across stages. Similarly for males (Figure 77.2b),
final criterion, we find no evidence that a lack the percentage of networked applicants does not
of racial homophily in minorities’ referring pat- increase across the various stages of the screen-
terns is weakening their access to this company. ing process. If anything, it decreases (networked
In sum, there is little evidence that minorities in candidates constitute 35.3 percent of the overall
this setting are cut off from the job networks that male applicant pool and 28.6 percent of hires).
lead to employment at this company due to the We also performed multivariate (probit)
behavior of the originators of referral networks analyses of who is hired versus not hired, in
(the “referring” process). order to introduce controls into the analysis. For
female applicants, both recruitment source and
race do not significantly influence the probabil-
Hiring Interface ity of hire. This does not change when human
Even if networked minorities are well represented capital and controls are added to the model
in the application pool, this does not necessarily (although some of these controls such as “years
mean that they will be similarly represented at of experience” are significant). There is thus no
subsequent stages of the screening process. This evidence of a preference for networked candi-
step depends on the employer’s racial biases and dates among females, and the race distribution
attitudes towards referrals (“screening process” of hires is not different from the distribution
in Figure 77.1). Our fieldwork and interviews of applicants. For males, the results show that
 Fernandez and Fernandez-Mateo—Networks, Race, and Hiring • 669

Figure 77.2a. Female Hiring Interface

Figure 77.2b. Male Hiring Interface

all minority racial groups are more likely to be In sum, the regression results show that there
hired than whites—irrespective of whether the is little evidence of an employer’s preference ei-
person was a referral or not. The only significant ther for, or against, candidates who were referred
pattern is that Asian American nonnetworked to the company at the hiring interface. The one
males are 8.5 percent more prevalent among exception is the case of Asian American males,
hires than among applicants, even after every- where nonnetworked candidates are more likely
thing else is controlled for. to be hired compared with networked applicants.
670 • Part VI: Who Gets Ahead?

We can only speculate whether this reflects a recruitment is inherently exclusionary (LoPresto
conscious effort to limit the number of Asian 1986). The results presented here, however, sug-
Americans—who are overrepresented in the ap- gest that this heuristic is too simple. Relying on
plication pool relative to the population in the referrals can help reproduce the distribution of
local labor market. The hiring interface is thus a the referring population. Therefore, in settings
point of disconnection for Asian American males where the current workforce is racially diverse—
(i.e., the arrow labeled “(–)” in step 2 of Figure as is the case in this setting—the referral pro-
77.1), but this employer is otherwise neutral cesses can actually help perpetuate diversity (see
with respect to networked candidates (i.e., the Rubineau and Fernandez 2005).
arrow labeled “0” in step 2 of Figure 77.1). Although the “wrong network” account does
not fit the facts in this setting, this is not to say
that in another, less diverse setting, where refer-
Summary and Conclusion rals may be preferred by screeners, the empirical
This paper makes a number of theoretical con- results would not be markedly different. While
tributions to the study of racial inequality and we can make no claims of empirical generaliz-
networks. First, it contributes to specifying the ability, however, this study has important meth-
mechanisms operating in network accounts of odological implications. Its value is apparent in
racial inequality in the labor market. Past ac- the light it sheds on mechanisms that are nor-
counts in this area often run the danger of circu- mally hidden from view. We suggest that the
lar reasoning because “wrong network” is defined fine-grained processes that we have uncovered
in terms of the eventual outcome. We argue that here need to be addressed to render the “wrong
much more analytical precision is needed to network” hypothesis testable in other settings.
specify what it means to be “stuck” in the “wrong Moreover, it is important to realize that the race
network,” as these stories are consistent with mi- and network effects that are often reported in
nority underrepresentation in any of a number analyses of highly aggregated survey data are
of steps in the recruitment and hiring process. likely conflating the effects of the multiple mech-
Indeed, this study is the first to analyze com- anisms that we have delineated. Distinguishing
prehensively the racial implications of both the among these steps should be a high priority in
referring process and the screening mechanisms future research at the intersection of networks
at the point of hire. We found that network fac- and race in the labor market.
tors operate at several stages of the recruitment
process, but we found scant evidence that these REFERENCES
factors serve to cut off minorities from employ- Elliott, James R. 1999. “Social Isolation and Labor
Market Insulation: Network and Neighborhood
ment at this company.
Effects on Less-Educated Urban Workers.” Socio-
This study also has significant implications
logical Quarterly 40: 199–216.
for policy. Since policies are often designed to Falcon, Luis M. 1995. “Social Networks and Em-
target distinct steps in the recruitment process ployment for Latinos, Blacks, and Whites.” New
(National Research Council 2004), understand- England Journal of Public Policy 11: 17–28.
ing each of these steps is crucial for crafting effec- Fernandez, Roberto M., Emilio J. Castilla, and Paul
tive policy interventions. Affirmative action, for Moore. 2000. “Social Capital at Work: Networks
example, is focused on affecting the behavior of and Employment at a Phone Center.” American
labor market screeners (Reskin 1998). Assessing Journal of Sociology 105: 1288–1356.
the effect of these policies will be very difficult Fernandez, Roberto M., and Isabel Fernandez-Mateo.
2006. “Networks, Race and Hiring.” American
without data on both hires and nonhires. Other
Sociological Review 71: 42–71.
policy prescriptions recommend that companies Fernandez, Roberto M., and M. Lourdes Sosa. 2005.
open up their recruitment practices by broad “Gendering the Job: Networks and Recruitment
advertising and use of formal recruitment sys- at a Call Center.” American Journal of Sociology
tems, on the theory that informal, referral-based 111: 859–904.
 DiMaggio and Garip—When Do Social Networks Increase Inequality? • 671

Holzer, Harry J. 1996. What Employers Want: Job Reingold, David A. 1999. “Social Networks and the
Prospects for Less-Educated Workers. New York: Employment Problem of the Urban Poor.” Urban
Russell Sage Foundation. Studies 36: 1907–32.
Kasinitz, Philip, and Jan Rosenberg. 1996. “Missing Reskin, Barbara F. 1998. The Realities of Affirmative
the Connection: Social Isolation and Employ- Action. Washington, DC: American Sociological
ment on the Brooklyn Waterfront.” Social Prob- Association.
lems 43: 180–93. Royster, Deirdre A. 2003. Race and the Invisible
Lin, Nan. 2002. Social Capital: A Theory of Social Hand: How White Networks Exclude Black Men
Structure and Action. New York: Cambridge Uni- from Blue-Collar Jobs. Berkeley: University of
versity Press. California Press.
LoPresto, Robert. 1986. “Recruitment Sources and Rubineau, Brian, and Roberto M. Fernandez. 2005.
Techniques.” Pp. 13-1 to 13-26 in Handbook of “Missing Links: Referral Processes and Job Seg-
Human Resource Administration, 2nd ed., edited regation.” MIT Sloan School of Management,
by J. J. Famularo. New York: McGraw-Hill. Massachusetts Institute of Technology, Cam-
National Research Council. 2004. Measuring Ra- bridge, MA. Unpublished manuscript.
cial Discrimination. Panel on Methods for As- Smith, Sandra. 2005. “‘Don’t Put My Name on It’:
sessing Discrimination, edited by Rebecca M. Social Capital Activation and Job-Finding Assis-
Blank, Marilyn Dabady, and Constance F. Citro. tance Among the Black Urban Poor.” American
Committee on National Statistics, Division of Journal of Sociology 111: 1–57.
Behavioral and Social Sciences and Education. Ullman, Joseph C. 1966. “Employee Referrals: A
Washington, DC: National Academies Press. Prime Tool for Recruiting Workers.” Personnel
Petersen, Trond, Ishak Saporta, and Marc-David 43: 30–35.
Seidel. 2000. “Offering a Job: Meritocracy and Wilson, William Julius. 1996. When Work Disap-
Social Networks.” American Journal of Sociology pears: The World of the New Urban Poor. New
106: 763–816. York: Alfred A. Knopf.

78 • Paul DiMaggio and Filiz Garip

When Do Social Networks Increase Inequality?

Introduction make it more difficult to overcome. This can

Sociologists ordinarily think of social networks happen when
as beneficial sources of support, stability, and 1. A good, service, or practice helps individu-
social capital. But they can have a dark side. In als get ahead;
this chapter we suggest that social networks may 2. That good, service, or practice is character-
exacerbate inequality between social groups and ized by network externalities, so that benefits

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in the
following publication: Paul DiMaggio and Filiz Garip, “How Network Externalities Can Exacerbate Intergroup
Inequality,” American Journal of Sociology 116:6 (May 2011), pp. 1887–1933, published by the University
of Chicago Press. The article printed here was originally prepared by Paul DiMaggio and Filiz Garip for this
fourth edition of Social Stratification: Class, Race, and Gender in Sociological Perspective, edited by David B.
Grusky. Copyright © 2014 by Westview Press.
672 • Part VI: Who Gets Ahead?

to adopters are higher, or risks are lower, if telephones in their homes. The more people
members of one’s social network have al- had access to phones, the more useful phones
ready adopted it; and were to everyone else. Once adoption reached
3. Social networks are segregated with respect to critical mass, telephones stopped being just busi-
some characteristic associated with adoption. ness devices and became central links to family
and friends. When a network’s value increases as
We illustrate our argument with a computa- more people join it, we say that the network pos-
tional model of Internet adoption in the United sesses “positive externalities.”1
States, supplemented by a brief account of an Tangible networks like telephones or the In-
empirical study of internal migration in Thai- ternet are not the only ones that exhibit positive
land. In each case individual choices (to use the externalities. The basic mechanism also applies
Internet or to migrate) are influenced by prior to purely social networks. Consider the prob-
choices made by members of ego’s social net- lem of socioeconomic and racial inequality in
work; and intergroup inequality (in technology advanced placement (AP) course enrollments
use or in migration rates) is amplified to the ex- in US high schools (Kao and Thompson 2003;
tent that social networks are homogeneous with Klopfenstein 2004). Students know that AP
respect to socioeconomic status. courses will help them gain admission to se-
Sociologists often view intergroup inequality lective colleges, but that they will have to work
as the aggregate result of individual behavior: hard to pass the AP exam. Whether or not they
efforts by people with varying initial advan- enroll in AP courses will be influenced not only
tages to obtain useful educations, good jobs, by their college plans or competing demands
and adequate incomes. In contrast we focus on from sports teams or after-school jobs, but also
extra-individual mechanisms (Hedstrom 2005) by their friends’ decisions. If your friends take
in which goal-directed individual actions (1) the same AP course, you can study together,
are shaped by social interactions that (2) yield collaborate on homework, and share equipment
higher-level outcomes (e.g., social inequality) like calculators or Internet connections. Having
that (a) are emergent (i.e., cannot be recovered friends in the class increases your gain (you will
simply by adding up the individual actions that learn more and enjoy socializing) and reduces
combine to produce them) and (b) vary depend- your risk of doing poorly on the AP exam.
ing upon the initial social structure (ordinarily If friendships were distributed randomly—if
depicted in terms of social networks). a student from a poor family were as likely to
know someone in the AP class as one from a
rich family is—none of this would matter for
Network Externalities, Social inequality. In real high schools, however, friend-
Homophily, and Diffusion Processes ships are homophilous: similar students hang
Today almost everyone has a cell phone or out with one another more than with those
landline. But who was the first person to buy from other groups (Quillian and Campbell
a telephone? In Palo Alto, California, it was 2003).2 Because of this, most benefits will ac-
John Parkinson (Fischer 1992, 130–34), who crue to students from privileged backgrounds,
got a telephone in 1892 to call suppliers for his whose friends disproportionately enroll in AP
hardware store. By 1893 a realtor and a butcher courses. Because poor students have only a few
had joined him. Soon a local pharmacy took a friends in AP courses, externalities help them
subscription, placing its phone in a quiet room less. Combine these influences—big incentives
where customers could pay to use it. By 1897 for advantaged students, small ones for the dis-
Palo Alto had nineteen telephone subscribers, advantaged—and gaps between rich and poor
including doctors and newspaper editors (pro- students become even larger than individual dif-
fessions reliant on quick communication) with ferences would lead one to predict.
 DiMaggio and Garip—When Do Social Networks Increase Inequality? • 673

Each of these examples has three elements: in which network effects directly enhance the
a choice (buying a telephone, enrolling in an technology’s value to the agent.
AP course); positive network externalities (your An agent-based model is an artificial social
telephone is more valuable if you can call your world that permits one to hold many things con-
friends; you get more out of AP courses if your stant while exploring a mechanism of particular
friends take them, too); and social homophily interest. One creates (or finds) a population of
(birds of a feather flock together), which makes actors, writes a computer program containing de-
positive externalities work hardest for groups cision rules that govern these actors’ choices, and
with an initial advantage. Combine these three then watches to see how these rule-governed, in-
elements—choice, externalities, and homo- dividual choices influence system-level outcomes
phily—in a diffusion process, and the results like social inequality. To ensure that the distribu-
exacerbate social inequality by reinforcing pre- tion of personal characteristics and associations
existing privilege through differential adoption among them are realistic, we base as many of the
of a useful new product or practice. parameters of our model as possible on real-world
Diffusion process is what sociologists call the data (2,257 African American and white respon-
processes of influence, imitation, or common dents to the 2002 General Social Survey), while
problem solving through which a novel practice varying two theoretically central factors—the
or product wins broad acceptance (i.e., becomes nature of network externalities and the degree of
widely adopted or diffused ). In our data analy- network homophily—as experimental conditions.
ses, we employ threshold models of diffusion. This case interested us because in the United
The threshold is the point at which a person will States, inequality in access to home Internet ser-
choose to adopt a new behavior (often a price at vice among groups based on education, income,
which one is willing to make a purchase). Thresh- and race has remained substantial (DiMaggio
old models begin with a set of actors whose indi- and Cohen 2004; DiMaggio and Garip 2011),
vidual thresholds vary as a function of personal even as the proportion of US families with Inter-
attributes and network characteristics (Coleman, net service has risen from just about one in four
Katz, and Menzel 1957; Schelling 1971; Gra- around 2000 to nearly three in four by 2012.3
novetter 1978). Our model builds on this work Given the Internet’s potential to level the infor-
by (1) exploring cases in which externalities are mation playing field, as well as the steady shift
local (in which each person’s threshold is only of commercial opportunities and public services
affected by the behavior of his or her own so- online, such persistent inequality is troublesome.
cial contacts); (2) adding status homophily as a Not everyone agrees about whether the “digital
key variable; and (3) focusing on group-specific divide” will persist, however.
adoption rates (and their implications for social Skeptics point out that intergroup inequality
inequality), rather than on rates for the popula- in adoption rates occurs whenever groups start at
tion as a whole. different baselines or reach critical mass at differ-
ent times (Leigh and Atkinson 2001). Consider
the top panel of Figure 78.1, which depicts styl-
A Case Study: The Internet— ized adoption curves for two populations, A and
Transitional Inequality or B, with differences measured at three points in
Permanent Divide? time (x, y, and z).
We develop our argument by using an The x axis represents time, and the y axis rep-
agent-based model (Macy and Willer 2002) to resents the percentage of adopters. Group B begins
predict group-specific diffusion paths for Inter- the adoption process later than and reaches takeoff
net use, varying the type of network effect and after group A. If we compare early rates (time x),
the extent of homophily. Internet diffusion is a we see no inequality. Because A reaches takeoff be-
conventional instance of new-product adoption fore B, we then see increasing inequality, reaching
674 • Part VI: Who Gets Ahead?

Figure 78.1. Stylized Diffusion Curves for 2 Pairs of Populations

a maximum at y, just before B reaches takeoff and

just as A approaches equilibrium. Trend analysis The Model
between y and z, by contrast, shows declining in- To understand these mechanisms, we created an
equality and (correctly) predicts convergence. agent-based model of Internet adoption with
The bottom panel of Figure 78.1 presents a choice interdependence in homophilous net-
similar depiction of the adoption history of two works influenced by network externalities. Our
other populations. Group C’s trajectory is iden- model produces seven artificial worlds (i.e., ex-
tical to that of Group A in the top panel. Group perimental conditions), each populated by 2,257
D gets into the game at the same point as group heterogeneous agents. The worlds vary in (1) the
B (top), but adopts more slowly and plateaus at presence and type of network externality and (2)
a lower level of penetration. As in the previous the degree of homophily. We use them to explore
case, inequality peaks at point y. But it remains how inequality in Internet use among groups
high even at point z. In this case, a pessimistic based on race (African American or white), in-
conclusion based on measurement at points x come (high, medium, and low), and education
and y (intergroup inequality is a real problem) (college graduate, some college, high school grad-
is more accurate than an optimistic prediction uate without college, and high school dropout)
based on measurement at points y and z. The varies over time as a function of the presence and
problem is this: at point x we cannot know if type of network effect (none, global, or local) and
we are in the top or the bottom panel of Figure as a function of the strength of pressures toward
78.1—that is, whether intergroup inequality will homophily (i.e., homophily bias).
increase or decrease—unless we understand the Each agent has a reservation price: an adop-
mechanisms driving diffusion. tion threshold at which it will purchase Internet
 DiMaggio and Garip—When Do Social Networks Increase Inequality? • 675

service. At the end of each period (imagine one

month), each agent compares the price of Internet What We Found
service to its reservation price and decides whether The analysis included four steps. First, we
or not to adopt. (Agents may adopt because econ- looked at the impact of the seven conditions on
omies of scale drive prices to or below their reser- overall diffusion trajectories, for the sample as
vation prices; because their network peers adopt, a whole and for groups within it, focusing on
which raises their reservation price [network ex- the rapidity of takeoff and equilibrium adoption
ternalities]; or a combination of both.) After each rates. Second, we examined how the seven con-
period, for one hundred periods, the cumulative ditions varied on inequality in adoption rates
percentage of agents with different races, income between pairs of groups defined by race, educa-
levels, and educational levels who have adopted is tion, and income. Third, we undertook logistic
reported and added to a graph. regressions to examine the net impact of race,
We focused on the impact of two variables— education, and income on individual adoption
the nature of network externalities and the de- choices throughout the process. Fourth, we
gree of homophily—on the extent of intergroup used regression analysis of seven thousand tri-
inequality in adoption throughout the diffusion als to describe how externalities and homophily
process. (For details, see appendix to DiMaggio interact to affect group-level rates and intergroup
and Garip 2011.) Two points are critical: (1) inequality.
adoption is driven by the relationship between
the cost of Internet service and each agent’s res- Overall adoption rates. Our first finding is that
ervation price (i.e., what the agent will pay) and without networks, there would be only limited
(2) this relationship varies across experimental diffusion: in the no-externalities condition,
conditions. Without externalities, reservation adoption never reaches critical mass, level-
prices remain constant, so only cost reductions ing off at about 10 percent (see Figure 78.2).
can induce adoption. With global externalities Adoption rates in other conditions rise slowly,
(when everyone’s utility is increased by each new pick up steam, and eventually level off. Under
adopter), reservation prices rise with the percent- global externalities (where rising tides lift all
age of agents who have already adopted. In this boats), adoption takes off between periods 30
condition, new adopters affect all at-risk agents and 40, rising sharply to the highest diffusion
in the same way (although susceptibility to this level of any condition, nearly 65 percent at
influence rises with income). In the five condi- equilibrium. Introducing local externalities
tions characterized by local externalities, reserva- without homophily scatters eff ects of new
tion prices are influenced by the percentage of adopters across the population, rather than ap-
the members of the agent’s personal network plying them to everyone, with the result that
who have already adopted. Each new adoption some agents are marooned in nonadopting
increases reservation prices for agents linked to networks, with adoption slower and lower at
the adopter and for no one else. Five conditions equilibrium.
with such local externalities vary in the extent of Combining local externalities with homoph-
homophily bias (Skvoretz 1990), ranging from ily induces more rapid takeoffs, but lowers pene-
no homophily bias (i.e., network partners se- tration overall. When homophily is modest (h =
lected at random from the population) to maxi- 0.25), the trajectory is similar to that for global
mum homophily (all network partners chosen externalities. Strong homophily (h = 1.0) stimu-
from a pool of those closest on a composite mea- lates early diffusion sharply, as adoption cascades
sure of similarity in education, income, and race, within homogeneous networks, inducing rapid
weighted by results in McPherson, Smith-Lovin, takeoff but producing an equilibrium adoption
and Brashears 2006). For each condition, we un- level below that of any other condition with
dertook one thousand simulation trials, report- externalities. Middling homophily levels yield
ing the mean values for each time period. middling results. The bottom line is that local
676 • Part VI: Who Gets Ahead?

Figure 78.2. Diffusion Trajectories for 7 Conditions of Network Externalities and Homophily

network externalities and homophily interact same is true for inequality by educational level
to speed up diffusion (the more homophily, the and race (DiMaggio and Garip 2011).
faster it takes off) but limit its scope (the more
homophily, the lower the equilibrium penetra- When do income, education, and race make
tion rate). the biggest difference for rates of Internet adop-
tion? We used logistic regressions to predict
Intergroup inequality. Under what condition individual-level adoption across our seven exper-
is inequality greatest? We measured inequality imental conditions. When externalities are local,
using odds ratios for each pair of groups. Figure the net effects of income, education, and race all
78.3 shows the results for income, specifically increase monotonically as homophily increases,
inequality in rates of adoption between the top providing further evidence that local externali-
and the bottom income tertiles. Differences are ties interact with homophily to exacerbate inter-
sharpest when local externalities combine with group inequality. Under global externalities, net
maximum homophily bias. Given local exter- effects of race and education are essentially zero,
nalities, increments in homophily enhance the and income effects are moderate, comparable
slope and equilibrium adoption rate for the to their impact with middling homophily bias.
highest-income group and increase that group’s (Some income effects are built into our model,
advantage at equilibrium over the lowest-income because income raises one’s reservation price.) In
tertile. The more homophily, the more rapidly concrete terms, each extra year of education in-
inequality increases, the greater it is at its peak, creases the equilibrium odds of adoption by just
and the more inequality one finds at equilib- 1 percent when externalities are global, but by
rium. When there is no homophily, or when ex- a full 7 percent when externalities are local and
ternalities are global, there is less inequality. The homophily bias is at its peak. Similarly, whites
 DiMaggio and Garip—When Do Social Networks Increase Inequality? • 677

Figure 78.3. Odds Ratios of Diffusion Rates for Highest Compared to Lowest Income Classes in 6
Conditions of Externalities and Homophily

are 11 percent more likely than blacks to adopt in Table 78.1 confirmed our interpretations of
(other things equal) under global externalities, the descriptive analyses: highest diffusion levels
but a full 50 percent more likely when local occur under global externalities, homophily with
externalities combine with maximum homoph- local externalities reduces overall adoption levels,
ily (graphs and details in DiMaggio and Garip and homophily increases intergroup inequality
2011). at equilibrium (see DiMaggio and Garip 2011
for effects on diffusion).
How do homophily and network externalities More important, these analyses help us un-
combine to exacerbate inequality? To answer this derstand why these results occur: local external-
question, we examined statistical effects of ex- ities increase inequality by significantly reducing
ternalities and homophily on diffusion and on adoption by less privileged groups, while exert-
inequality at equilibrium. Cases are seven thou- ing modest positive effects on more privileged
sand runs of the model, one thousand for each groups’ rates. Moreover, the greater the status
of seven conditions. Independent variables are distance between groups, the more homophily
dummy variables for each of six conditions. (The exacerbates inequality. Income effects are stron-
reference category, to which results from other gest on inequality between the top tertile and
conditions were compared, is local externalities bottom tertiles, with a much smaller effect on
without homophily.) Dependent variables were top-to-middle inequality and an even smaller
equilibrium adoption levels (global and for spe- (though statistically significant) impact on
cific population subgroups) and indicators of middle-to-bottom inequality. Similarly, effects of
intergroup inequality (logged odds ratios com- increased homophily bias on inequality between
paring paired subgroups). The results shown college graduates and high school dropouts
 678 • Part VI: Who Gets Ahead?

Table 78.1. Linear Regression of Logged Odds Ratios (Intergroup Inequality) on Experimental Conditions

RACE INCOME EDUCATION

BA/Less Some college/

High/Low High/Medium Medium/Low BA/High
White/Black than high Less than high
income income income school
school school

(Reference: Homophily = 0)
No network externalities -0.02 ** -0.13 ** 0.02 * -0.15 ** -0.07 ** -0.05 ** -0.13 **
General network externalities -0.05 ** 0.51 ** 0.41 ** 0.10 ** 0.02 * 0.01 0.06 **
Homophily = 0.25 0.04 ** 0.21 ** 0.15 ** 0.06 ** 0.08 ** 0.05 ** 0.06 **
Homophily = 0.5 0.09 ** 0.43 ** 0.31 ** 0.13 ** 0.16 ** 0.10 ** 0.13 **
Homophily = 0.75 0.14 ** 0.67 ** 0.46 ** 0.21 ** 0.26 ** 0.14 ** 0.22 **
Homophily = 1 0.20 ** 0.93 ** 0.62 ** 0.31 ** 0.39 ** 0.20 ** 0.33 **
Intercept 0.79 ** 3.62 ** 1.87 ** 1.76 ** 1.75 ** 1.01 ** 1.54 **
R2 0.28 0.66 0.57 0.29 0.41 0.35 0.27
N 7000 7000 7000 7000 7000 7000 7000
**p < 0.01, * p < 0.05. All independent variables are binary. Both dependent and independent variables are measured on the final
period of simulations (T = 100).

exceed the still significant effects on inequality puzzle and demonstrates the scope of behaviors
between other subgroups. to which the argument applies.
How does migration fit our model? First, it
involves a choice (to migrate) that may improve
Overview one’s well-being. Second, a young person may
The model results captured some important be more willing to migrate if siblings or friends
dimensions of Internet diffusion. Inequality in have done so. The city holds many dangers, and
Internet access has persisted as group-specific pen- having social ties there reduces risk and increases
etration rates have leveled off, dashing hopes that one’s chance of finding a good job (Garip 2008).
the digital divide would vanish on its own. The We looked at the impact of ties to previous mi-
model helps to explain why, demonstrating that grants on the likelihood of individual migration.
under almost any range of plausible conditions And we asked how social homophily moderated
(barring dramatic public policy intervention), in- the impact of network externalities at the village
tergroup adoption rates will not converge. level and created different migration patterns
across communities.4
The results were as expected. First, consistent
An Empirical Case: Internal Migration with our view of migration as a network good,
in Thailand the more migrants with whom young people
So far we have explicated our argument and were in contact, the more likely they were to mi-
used a computational model of Internet diffu- grate. Second, just as homophily depressed dif-
sion to explore its properties and implications. fusion in Internet adoption, villages with greater
DiMaggio and Garip (2011) explore a second homophily had lower rates of migration overall.
case, a study of change over twenty years in rates Third, in addition to its negative direct effect
of (mostly) short-term youth migration from on overall migration, homophily in village net-
twenty-two villages in Thailand’s Nang Rong works increased the impact on migration of social
district to nearby cities during the last quarter ties: where overall homophily was high, social
of the twentieth century. Although the villages contacts with migrants had stronger positive ef-
were similarly situated and started out with sim- fects on migration than where it was low. Finally,
ilar migration patterns, rates diverged markedly just as network effects and homophily combined
in ways that existing theories of migration can- to increase intergroup inequality in the Internet
not explain. This case complements the Inter- model, homophily increased variance in migra-
net example because it addresses an empirical tion rates among Nan Rong villages. Indeed,
 DiMaggio and Garip—When Do Social Networks Increase Inequality? • 679

most of the mysterious variation that emerged characterized by social homophily. Studies in many
among these villages over the study’s two decades societies and organizations have demonstrated per-
was concentrated among the most socially ho- vasive homophily with respect to race and ethnicity,
mogeneous villages, precisely where our argu- gender, age, educational attainment, religion, and
other factors (McPherson, Smith-Lovin, and Cook
ment would expect to find it.
2001).
3. We focus on Internet use at home because
home use ordinarily provides the greatest autonomy
Conclusion
and opportunities for learning (DiMaggio et al.
We have argued that inequality emerges from the 2004) and because it is consequential, boosting the
union of two aspects of social structure: network earnings net of technology use at work (DiMaggio
externalities, which make behaviors more pro- and Bonikowski 2008).
ductive or less risky for actors whose friends and 4. We used data on village social homogeneity
associates also pursue them, and social homoph- as a proxy for network homophily, expecting that
where everyone is more similar, there will be less
ily, which occurs when people form networks
variety in personal networks. Actual network data
with those of similar status. We have shown how collected for one year confirmed this supposition.
these processes work in technology diffusion and
have provided evidence that they operate sim- REFERENCES
ilarly in the very different case of rural-urban Coleman, James S., Elihu Katz, and Herbert Menzel.
migration in a developing country. 1957. “The Diffusion of an Innovation Among
There is still much we do not understand Physicians.” Sociometry 20: 253–270.
about the ways that networks produce cumula- DiMaggio, Paul, and Bart Bonikowski. 2008. “Make
tive advantage by enhancing the value of new Money Surfing the Web: The Impact of Internet
Use on the Earnings of U.S. Workers.” American
practices to their adopters. How widespread
Sociological Review 73: 227–252.
are they, and how large a role do they play in
DiMaggio, Paul, and Joseph Cohen. 2004. “Infor-
producing overall inequality? How might such mation Inequality and Network Externalities: A
mechanisms not only produce inequality in Comparative Study of the Diffusion of Television
rates of helpful behaviors, but also reinforce in- and the Internet.” In The Economic Sociology of
tergroup differences in such harmful practices as Capitalism, edited by Victor Nee and Richard
smoking or substance abuse? Nonetheless, our Swedberg, 227–267. Princeton, NJ: Princeton
results establish the importance of network ex- University Press.
ternalities in differentiated diffusion networks as DiMaggio, Paul, and Filiz Garip. 2011. “How
a mechanism that must be taken into account Network Externalities Can Exacerbate Social
Inequality.” American Journal of Sociology 116:
if we are to understand the structural bases of
1887–1933.
social inequality. DiMaggio, Paul, Eszter Hargittai, Coral Celeste,
and Steven Shafer. 2004. “From Unequal Access
NOTES to Differentiated Use: A Literature Review and
1. Technically, we care here about local network Agenda for Research on Digital Inequality.” In
externalities, in which each potential adopter cares Social Inequality, edited by Kathryn Neckerman,
only about whether his or her own contacts have 355–400. New York: Russell Sage Foundation.
been adopted. In some cases, we don’t care about the Fischer, Claude S. 1992. America Calling: A Social
identities of prior adopters because we gain from all History of the Telephone to 1940. Berkeley: Uni-
of them (e.g., if we want to buy or sell something on versity of California Press.
eBay or hope that lots of people watch our favorite Garip, Filiz. 2008. “Social Capital and Migration:
TV show so it gets renewed). Such global externali- How Do Similar Resources Lead to Divergent
ties don’t matter for inequality: Under those condi- Outcomes?” Demography 45 (3): 591–617.
tions, a rising tide boosts all ships. Granovetter, Mark. 1978. “Threshold Models of
2. Children aren’t the only ones whose peers tend Collective Behavior.” American Journal of Sociol-
to be similar to themselves (i.e., whose networks are ogy 83: 1420–1443.
680 • Part VI: Who Gets Ahead?

Hedstrom, Peter. 2005. Dissecting the Social: On the McPherson, Miller, Lynn Smith-Lovin, and Mat-
Principles of Analytical Sociology. New York: Cam- thew E. Brashears. 2006. “Social Isolation in
bridge University Press. America: Changes in Core Discussion Networks
Kao, Grace, and Jennifer S. Thompson. 2003. “Ra- over Two Decades.” American Sociological Review
cial and Ethnic Stratification in Educational 71: 353–375.
Achievement and Attainment.” Annual Review of McPherson, J. Miller, Lynn Smith-Lovin, and J. M.
Sociology 29: 417–442. Cook. 2001. “Birds of a Feather: Homophily in
Klopfenstein, Kristin. 2004. “Advanced Placement: Social Networks.” Annual Review of Sociology 27:
Do Minorities Have Equal Opportunity?” Eco- 415–444.
nomics of Education Review 23: 115–131. Quillian, Lincoln, and Mary E. Campbell. 2003.
Leigh, Andrew, and Robert D. Atkinson. 2001. Clear “Beyond Black and White: The Present and
Thinking on the Digital Divide. PPI Policy Report Future of Multiracial Friendship Segregation.”
(June 26). Washington, DC: Progressive Policy American Sociological Review 68: 540–566.
Institute. http://www.ndol.org/documents/digi- Schelling, Thomas. 1971. “Dynamic Models of Seg-
tal_divide.pdf (accessed March 23, 2008). regation.” Journal of Mathematical Sociology 1:
Macy, Michael W., and Robert Willer. 2002. “From 143–186.
Factors to Actors: Computational Sociology and Skvoretz, John. 1990. “Biased Net Theory: Approx-
Agent-Based Modeling.” Annual Review of Sociol- imations, Simulations, and Observations.” Social
ogy 28: 143–166. Networks 12: 217–238.
 PART VII

Race and Ethnicity

Constructing Racial Categories

79 Racial Formation in the United States 682
80 The Dynamics of Racial Fluidity and Inequality 687

Classic Modes of Incorporation

81 A Theory of Ethnic Antagonism: The Split Labor Market 696
82 The Immigrant Enclave: Theory and Empirical Examples 710
83 Assimilation Theory for an Era of Unprecedented Diversity 721

New Modes of Incorporation

84 The New Second Generation: Segmented Assimilation
and Its Variants 729
85 Why Replenishment Strengthens Racial and Ethnic Boundaries 740

Discrimination
86 Are Emily and Greg More Employable Than Lakisha and Jamal?
A Field Experiment on Labor Market Discrimination 747
87 Stereotype Threat and African-American Student Achievement 752
88 Marked: Race, Crime, and Finding Work in an Era of
Mass Incarceration 757

Are Racial and Ethnic Distinctions Declining in Signikcance?

89 The Declining Significance of Race: Blacks and Changing
American Institutions 765
The Declining Significance of Race: Revisited and Revised 777
90 How Do Latino Immigrants Fit Into the Racial Order? 780
91 Are Recent Trends in Intermarriage Consistent with
Assimilation Theory? 788
682 • Part VII: Race and Ethnicity

Constructing Racial Categories

79 • Michael Omi and Howard Winant

Racial Formation in the United States

In 1982–83, Susie Guillory Phipps unsuccess- race and assigning individuals or groups to ra-
fully sued the Louisiana Bureau of Vital Records cial categories. It shows how the racial legacies
to change her racial classification from black to of the past—slavery and bigotry—continue to
white. The descendant of an 18th-century white shape the present. It reveals both the deep in-
planter and a black slave, Phipps was designated volvement of the state in the organization and
“black” in her birth certificate in accordance interpretation of race, and the inadequacy of
with a 1970 state law which declared anyone state institutions to carry out these functions.
with at least 1/32nd “Negro blood” to be black. It demonstrates how deeply Americans both as
The Phipps case raised intriguing questions individuals and as a civilization are shaped, and
about the concept of race, its meaning in con- indeed haunted, by race.
temporary society, and its use (and abuse) in Having lived her whole life thinking that she
public policy. Assistant Attorney General Ron was white, Phipps suddenly discovers that by
Davis defended the law by pointing out that legal definition she is not. In U.S. society, such
some type of racial classification was necessary an event is indeed catastrophic. But if she is not
to comply with federal record-keeping require- white, of what race is she? The state claims that
ments and to facilitate programs for the preven- she is black, based on its rules of classification,
tion of genetic diseases. Phipps’s attorney, Brian and another state agency, the court, upholds this
Begue, argued that the assignment of racial cat- judgment. But despite these classificatory stan-
egories on birth certificates was unconstitutional dards which have imposed an either-or logic on
and that the 1/32nd designation was inaccurate. racial identity, Phipps will not in fact “change
He called on a retired Tulane University pro- color.” Unlike what would have happened
fessor who cited research indicating that most during slavery times if one’s claim to whiteness
Louisiana whites have at least 1/20th “Negro” was successfully challenged, we can assume
ancestry. that despite the outcome of her legal challenge,
In the end, Phipps lost. The court upheld Phipps will remain in most of the social rela-
the state’s right to classify and quantify racial tionships she had occupied before the trial. Her
identity.1 socialization, her familial and friendship net-
Phipps’s problematic racial identity, and her works, her cultural orientation, will not change.
effort to resolve it through state action, is in She will simply have to wrestle with her newly
many ways a parable of America’s unsolved racial acquired “hybridized” condition. She will have
dilemma. It illustrates the difficulties of defining to confront the “Other” within.

Michael Omi and Howard Winant. Racial Formations in the United States, pp. 53–61, 180–181. Copyright
© 1986 by Routledge and copyright © 1994 by Michael Omi and Howard Winant. Used by permission of
the authors.
 Omi and Winant—Racial Formation in the United States • 683

The designation of racial categories and the race. Indeed, the categories employed to differ-
determination of racial identity is no simple task. entiate among human groups along racial lines
For centuries, this question has precipitated in- reveal themselves, upon serious examination, to
tense debates and conflicts, particularly in the be at best imprecise, and at worst completely
U.S.—disputes over natural and legal rights, arbitrary.
over the distribution of resources, and indeed, If the concept of race is so nebulous, can we
over who shall live and who shall die. not dispense with it? Can we not “do without”
A crucial dimension of the Phipps case is that race, at least in the “enlightened” present? This
it illustrates the inadequacy of claims that race question has been posed often, and with greater
is a mere matter of variations in human physi- frequency in recent years. An affirmative answer
ognomy, that it is simply a matter of skin color. would of course present obvious practical dif-
But if race cannot be understood in this manner, ficulties: it is rather difficult to jettison widely
how can it be understood? We cannot fully hope held beliefs, beliefs which moreover are central
to address this topic—no less than the meaning to everyone’s identity and understanding of the
of race, its role in society, and the forces which social world. So the attempt to banish the con-
shape it—in one chapter, nor indeed in one cept as an archaism is at best counterintuitive.
book. Our goal in this chapter, however, is far But a deeper difficulty, we believe, is inherent in
from modest: we wish to offer at least the out- the very formulation of this schema, in its way
lines of a theory of race and racism. of posing race as a problem, a misconception left
over from the past, and suitable now only for the
dustbin of history.
What Is Race? A more effective starting point is the recog-
There is a continuous temptation to think of nition that despite its uncertainties and contra-
race as an essence, as something fixed, concrete, dictions, the concept of race continues to play a
and objective. And there is also an opposite fundamental role in structuring and representing
temptation: to imagine race as a mere illusion, the social world. The task for theory is to ex-
a purely ideological construct which some ideal plain this situation. It is to avoid both the uto-
non-racist social order would eliminate. It is nec- pian framework which sees race as an illusion we
essary to challenge both these positions, to dis- can somehow “get beyond,” and also the essen-
rupt and reframe the rigid and bipolar manner in tialist formulation which sees race as something
which they are posed and debated, and to tran- objective and fixed, a biological datum. Thus
scend the presumably irreconcilable relationship we should think of race as an element of social
between them. structure rather than as an irregularity within
The effort must be made to understand race it; we should see race as a dimension of human
as an unstable and “decentered” complex of so- representation rather than an illusion. These per-
cial meanings constantly being transformed by spectives inform the theoretical approach we call
political struggle. With this in mind, let us pro- racial formation.
pose a definition: race is a concept which signifies
and symbolizes social conflicts and interests by refer-
ring to different types of human bodies. Although Racial Formation
the concept of race invokes biologically based We define racial formation as the sociohistorical
human characteristics (so-called “phenotypes”), process by which racial categories are created,
selection of these particular human features for inhabited, transformed, and destroyed. Our at-
purposes of racial signification is always and nec- tempt to elaborate a theory of racial formation
essarily a social and historical process. In contrast will proceed in two steps. First, we argue that ra-
to the other major distinction of this type, that cial formation is a process of historically situated
of gender, there is no biological basis for distin- projects in which human bodies and social struc-
guishing among human groups along the lines of tures are represented and organized. Next we link
684 • Part VII: Race and Ethnicity

racial formation to the evolution of hegemony, My proposal for dealing with the racial issue
the way in which society is organized and ruled. in social welfare is to repeal every bit of leg-
Such an approach, we believe, can facilitate un- islation and reverse every court decision that
derstanding of a whole range of contemporary in any way requires, recommends, or awards
controversies and dilemmas involving race, in- differential treatment according to race, and
cluding the nature of racism, the relationship of thereby put us back onto the track that we left
race to other forms of differences, inequalities, in 1965. We may argue about the appropriate
and oppression such as sexism and nationalism, limits of government intervention in trying
and the dilemmas of racial identity today. to enforce the ideal, but at least it should be
From a racial formation perspective, race is a possible to identify the ideal: Race is not a
matter of both social structure and cultural rep- morally admissible reason for treating one
resentation. Too often, the attempt is made to person differently from another. Period.3
understand race simply or primarily in terms of
only one of these two analytical dimensions.2 For Here there is a partial but significant analysis
example, efforts to explain racial inequality as a of the meaning of race: it is not a morally valid
purely social structural phenomenon are unable basis upon which to treat people “differently
to account for the origins, patterning, and trans- from one another.” We may notice someone’s
formation of racial difference. race, but we cannot act upon that awareness. We
Conversely, many examinations of racial dif- must act in a “color-blind” fashion. This analysis
ference—understood as a matter of cultural at- of the meaning of race is immediately linked to
tributes à la ethnicity theory, or as a society-wide a specific conception of the role of race in the
signification system, à la some poststructuralist social structure: it can play no part in govern-
accounts—cannot comprehend such structural ment action, save in “the enforcement of the
phenomena as racial stratification in the labor ideal.” No state policy can legitimately require,
market or patterns of residential segregation. recommend, or award different status according
An alternative approach is to think of racial to race. This example can be classified as a par-
formation processes as occurring through a link- ticular type of racial project in the present-day
age between structure and representation. Racial U.S.—a “neoconservative” one.
projects do the ideological “work” of making Conversely, to recognize the racial dimension
these links. A racial project is simultaneously an in social structure is to interpret the meaning of
interpretation, representation, or explanation of race. Consider the following statement by the
racial dynamics, and an effort to reorganize and late Supreme Court Justice Thurgood Marshall
redistribute resources along particular racial lines. on minority “set-aside” programs:
Racial projects connect what race means in a par-
ticular discursive practice and the ways in which A profound difference separates governmental
both social structures and everyday experiences actions that themselves are racist, and govern-
are racially organized, based upon that meaning. mental actions that seek to remedy the effects
Let us consider this proposition, first in terms of prior racism or to prevent neutral govern-
of large-scale or macro-level social processes, and ment activity from perpetuating the effects of
then in terms of other dimensions of the racial such racism.4
formation process.
Here the focus is on the racial dimensions
of social structure—in this case of state activity
Racial Formation as a Macro-Level and policy. The argument is that state actions in
Social Process the past and present have treated people in very
To interpret the meaning of race is to frame it social different ways according to their race, and thus
structurally. Consider for example, this statement the government cannot retreat from its policy
by Charles Murray on welfare reform: responsibilities in this area. It cannot suddenly
 Omi and Winant—Racial Formation in the United States • 685

declare itself “color-blind” without in fact perpet- and radical, stress the incompatibility of racially
uating the same type of differential, racist treat- defined group identity with the legacy of white
ment. Thus, race continues to signify difference supremacy, and therefore advocate a social struc-
and structure inequality. Here, racialized social tural solution of separation, either complete or
structure is immediately linked to an interpre- partial. Nationalist currents represent a pro-
tation of the meaning of race. This example too found legacy of the centuries of racial absolutism
can be classified as a particular type of racial proj- that initially defined the meaning of race in the
ect in the present-day U.S.—a “liberal” one. . . . U.S. Nationalist concerns continue to influence
These two examples of contemporary racial racial debate in the form of Afrocentrism and
projects are drawn from mainstream political de- other expressions of identity politics.
bate; they may be characterized as center-right Taking the range of politically organized ra-
and center-left expressions of contemporary cial projects as a whole, we can “map” the cur-
racial politics. We can, however, expand the rent pattern of racial formation at the level of
discussion of racial formation processes far be- the public sphere, the “macro-level” in which
yond these familiar examples. In fact, we can public debate and mobilization takes place. But
identify racial projects in several other analytical important as this is, the terrain on which racial
dimensions: first, the political spectrum can be formation occurs is broader yet.
broadened to include radical projects, on both
the left and right, as well as along other polit-
ical axes. Second, analysis of racial projects can Racial Formation as Everyday
take place not only at the macro-level of racial Experience
policy-making, state activity, and collective At the micro-social level, racial projects also link
action, but also at the micro-level of everyday signification and structure, not so much as ef-
experience. . . . forts to shape policy or define large-scale mean-
ing, but as the applications of “common sense.”
To see racial projects operating at the level of
The Political Spectrum of Racial everyday life, we have only to examine the many
Formation ways in which, often unconsciously, we “notice”
We have encountered examples of a neoconser- race.
vative racial project, in which the significance of One of the first things we notice about people
race is denied, leading to a “color-blind” racial when we meet them (along with their sex) is their
politics and “hands off” policy orientation; and race. We utilize race to provide clues about who
of a “liberal” racial project, in which the signifi- a person is. This fact is made painfully obvious
cance of race is affirmed, leading to an egalitarian when we encounter someone whom we cannot
and “activist” state policy. But these by no means conveniently racially categorize—someone who
exhaust the political possibilities. Other racial is, for example, racially “mixed” or of an ethnic/
projects can be readily identified on the contem- racial group we are not familiar with. Such an
porary U.S. scene. For example, “far right” proj- encounter becomes a source of discomfort and
ects, which uphold biologistic and racist views of momentarily a crisis of racial meaning.
difference, explicitly argue for white supremacist Our ability to interpret racial meanings de-
policies. “New right” projects overtly claim to pends on preconceived notions of a racialized so-
hold “color-blind” views, but covertly manipu- cial structure. Comments such as, “Funny, you
late racial fears in order to achieve political gains. don’t look black,” betray an underlying image of
On the left, “radical democratic” projects invoke what black should be. We expect people to act
notions of racial “difference” in combination out their apparent racial identities; indeed we be-
with egalitarian politics and policy. come disoriented when they do not. The black
Further variations can also be noted. For ex- banker harassed by police while walking in casual
ample, “nationalist” projects, both conservative clothes through his own well-off neighborhood,
686 • Part VII: Race and Ethnicity

the Latino or white kid rapping in perfect Afro combination, some version, of the rules of ra-
patois, the unending faux pas committed by cial classification, and of her own racial identity,
whites who assume that the non-whites they en- often without obvious teaching or conscious in-
counter are servants or tradespeople, the belief culcation. Thus are we inserted in a comprehen-
that non-white colleagues are less qualified per- sively racialized social structure. Race becomes
sons hired to fulfill affirmative action guidelines, “common sense”—a way of comprehending,
indeed the whole gamut of racial stereotypes— explaining, and acting in the world. A vast web
that “white men can’t jump,” that Asians can’t of racial projects mediates between the discur-
dance, etc., etc.—all testify to the way a racial- sive or representational means in which race is
ized social structure shapes racial experience and identified and signified on the one hand, and the
conditions meaning. Analysis of such stereotypes institutional and organizational forms in which
reveals the always present, already active link be- it is routinized and standardized on the other.
tween our view of the social structure—its de- These projects are the heart of the racial forma-
mography, its laws, its customs, its threats—and tion process.
our conception of what race means. Under such circumstances, it is not possible
Conversely, our ongoing interpretation of our to represent race discursively without simultane-
experience in racial terms shapes our relations to ously locating it, explicitly or implicitly, in a so-
the institutions and organizations through which cial structural (and historical) context. Nor is it
we are imbedded in social structure. Thus we ex- possible to organize, maintain, or transform so-
pect differences in skin color, or other racially cial structures without simultaneously engaging,
coded characteristics, to explain social differ- once more either explicitly or implicitly, in racial
ences. Temperament, sexuality, intelligence, ath- signification. Racial formation, therefore, is a
letic ability, aesthetic preferences, and so on are kind of synthesis, an outcome, of the interaction
presumed to be fixed and discernible from the of racial projects on a society-wide level. These
palpable mark of race. Such diverse questions as projects are, of course, vastly different in scope
our confidence and trust in others (for example, and effect. They include large-scale public ac-
clerks or salespeople, media figures, neighbors), tion, state activities, and interpretations of racial
our sexual preferences and romantic images, conditions in artistic, journalistic, or academic
our tastes in music, films, dance, or sports, and fora, as well as the seemingly infinite number of
our very ways of talking, walking, eating, and racial judgments and practices we carry out at
dreaming become racially coded simply because the level of individual experience.
we live in a society where racial awareness is
so pervasive. Thus in ways too comprehensive NOTES
even to monitor consciously, and despite peri- 1. San Francisco Chronicle, 14 September 1982,
19 May 1983.
odic calls—neoconservative and otherwise—for
2. Michael Omi and Howard Winant, “On the
us to ignore race and adopt “color-blind” racial
Theoretical Status of the Concept of Race” in War-
attitudes, skin color “differences” continue to ra- ren Crichlow and Cameron McCarthy, eds., Race,
tionalize distinct treatment of racially identified Identity, and Representation in Education (New York:
individuals and groups. Routledge, 1993).
To summarize the argument: the theory of 3. Charles Murray, Losing Ground: American
racial formation suggests that society is suffused Social Policy, 1950–1980 (New York: Basic Books,
with racial projects, large and small, to which all 1984), p. 223.
are subjected. This racial “subjection” is quint- 4. Justice Thurgood Marshall, dissenting in City
essentially ideological. Everybody learns some of Richmond v. J. A. Croson Co., 488 U.S. 469 (1989).
 Saperstein and Penner—The Dynamics of Racial Fluidity and Inequality • 687

80 • Aliya Saperstein and Andrew M. Penner

The Dynamics of Racial Fluidity and Inequality

Race is generally treated as an input in the Amer- which inequality also reinforces racial distinc-
ican stratification system. People are assumed tions. Drawing on more than two decades of
to be identifiable as members of distinct racial longitudinal data from a national survey, we
populations and subject to differential treatment demonstrate that treating race as a fixed charac-
based on this presumed membership. Those be- teristic not only oversimplifies the relationship
longing to valued populations receive greater between race and social status, it also obscures
access to the resources and rewards of society, the role that the racial fluidity of individuals
on average, than do those from devalued popu- plays in stabilizing the “social invention” (ASA
lations (Massey 2007). Because membership in 2003) that is race in the United States.
these populations is assigned partly by ancestry
and/or is based on readily observable and her-
itable physical features, the hierarchy of social Fixing Race
positions can be passed on generation after Despite the sociological consensus that race is a
generation, resulting in relatively low levels of social construction that varies by context, stan-
mobility (Tilly 1998). Indeed, despite several de- dard practice in quantitative research assumes
cades of attempts to ameliorate racial inequality one’s race is set at birth and predates one’s life
in the United States, large gaps in social, phys- chances (Zuberi 2000). Racial fluidity, to the
ical, mental, and material well-being remain, extent that it is acknowledged at all, is generally
particularly between Americans of African origin believed to be limited to three circumstances: 1)
and everyone else (Fischer and Hout 2006). a small minority of present-day Americans with
We seek to extend the standard sociological widely recognized mixed ancestries, such as Lati-
account of US racial inequality by incorporat- nos, American Indians, and the children of the
ing recent research from the fields of race and post-1960s “biracial baby boom”; 2) changes in
ethnicity and social psychology. We argue that the U.S. racial hierarchy in particular historical
instead of taking an individual’s race as a given epochs, such as the early 20th century “whiten-
and examining how it affects that person’s life ing” of southern and eastern Europeans; or 3)
chances, sociologists should be asking who is places with a high degree of racial mixing, such
perceived, or identifies, as a particular race when, as Brazil.
and why? This implies a reversal of the typical However, if race is a categorical distinc-
relationship between race and stratification, in tion imposed on otherwise continuous human

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in the
following publication: Aliya Saperstein and Andrew M. Penner, “Racial Fluidity and Inequality in the United
States,” American Journal of Sociology 118:3 (2012), pp. 676–727, published by University of Chicago Press.
The article printed here was originally prepared by Aliya Saperstein and Andrew M. Penner for this fourth
edition of Social Stratification: Class, Race, and Gender in Sociological Perspective, edited by David B. Grusky.
Copyright © 2014 by Westview Press.
688 • Part VII: Race and Ethnicity

variation (Ossorio and Duster 2005), and racial and “whitening” result in an overestimate of ra-
categorization is a fundamentally social process, cial inequality.
even from a cognitive perspective (Freeman et al. However, the language of over- or under-
2011), then in many cases it is inappropriate to estimates implies that there is a correct racial
treat an individual’s race as a constant. The anal- category in which all people should be placed.
yses that follow, along with a growing literature We argue that status-driven racial change is not
documenting the fluidity and complexity of ra- a misunderstanding of the individual’s “true”
cial self-identification in the United States (e.g., race and the extent of racial inequality at a given
Harris and Sim 2002; Doyle and Kao 2007; point in time; rather, it is indicative of how ra-
Craemer 2010; Khanna and Johnson 2010), cast cial fluidity and social mobility interact to main-
further doubt on the common assumption that tain inequality over time. For example, one of
although race might be “socially constructed” in the reasons the racial hierarchy in the United
a macrolevel, sociohistorical sense, it is never- States has remained relatively stable could be
theless a fixed attribute in the life of any given that upward mobility leads to reclassification:
American. white people appear to be more successful in
part because successful people become white,
through self-identification, external classifica-
Fluidity and Inequality tion, or both. Theoretically, the relationship also
Theories of social boundary maintenance de- works in reverse. Downward mobility could be
scribe a number of ways that racial hierarchies hidden, as well: black people may appear to be
and individual racial categorizations can change more criminal, unemployable, and so forth, in
over time (see Wimmer 2008 for a review). part because criminals and the unemployed are
Sometimes racial boundaries shift to incorpo- more likely to be seen as black (or less likely to be
rate large groups of people, such as when Puerto seen as white). Either way, individual-level racial
Rico’s “white” population grew dramatically fluidity would serve to maintain existing group
between censuses (Loveman and Muniz 2007), boundaries and the hierarchy of social positions
or when southern and eastern European im- they imply.
migrants were socially “whitened” in the early
twentieth century (Jacobson 1999). Individuals
can also cross existing racial boundaries as a re- Stereotypes and Categorization
sult of changes in their personal characteristics. Research on stereotypes and cognitive racial cat-
A typical example is known as passing, such as egorization supports this perspective. Americans
when the late Anatole Broyard, a book critic for generally associate whites with positive traits,
the New York Times, severed ties with his Loui- such as “smart” and “rich,” and blacks with neg-
siana Creole family of origin in order to “pass” ative traits, such as “corrupt” and “poor” (Allen
from black to white (Gates 1997). 1996); these associations have been shown to be
For stratification scholars, such racial flu- at least partly nonconscious and active even in
idity matters when it is related to patterns of individuals who are otherwise racially egalitar-
social inequality. For most of his adult life, ian (see, e.g., Greenwald, McGhee, and Schwartz
Anatole Broyard’s income, wealth, and occupa- 1998). Indeed, research has demonstrated
tional prestige would have been counted among that negative beliefs about blacks can be acti-
those of “whites” rather than “blacks.” In Latin vated without mentioning race at all, through
America the phrase “money whitens” describes the use of racially coded words such as “inner
a related form of boundary crossing, in which: city” (Hurwitz and Peffley 2005) and “welfare”
people with Afrocentric features can be racially (Gilens 1995).
reclassified, seemingly out of courtesy, after they Social psychological research also shows
manage to climb the social ladder (Loveman and that the racial categorization process itself is
Muniz 2007). Thus, in some sense both passing influenced by stereotypes and value judgments
 Saperstein and Penner—The Dynamics of Racial Fluidity and Inequality • 689

about the individual in question. For example,

Richeson and Trawalter (2005) find that people Data
take longer to racially categorize photographs of Our data come from the 1979 cohort of the
admired blacks (e.g., Martin Luther King Jr.) National Longitudinal Survey of Youth, a repre-
and disliked whites (e.g., Jeffrey Dahmer), and sentative sample of 12,686 US men and women
they are more likely to make “mistakes” than who were fourteen to twenty-two years of age
when categorizing either disliked blacks or ad- when first surveyed in 1979. We use observa-
mired whites. These stereotypical associations tions from 1979 to 1998, the most recent year in
also can be bidirectional: not only does seeing which interviewers were asked to racially classify
black people make Americans think about crime, respondents.
but thinking about crime makes Americans pay
more attention to black people (Eberhardt et al.
2004). Thus, we expect to find that people who Measures of Race
better fit the stereotypes of a particular racial Interviewers were instructed to classify respon-
group will be more likely to be classified as such. dents as “black,” “white,” or “other” at the end
of the interview. Thus, we do not have a mea-
sure of the interviewers’ first impressions of the
Analytic Approach respondents’ race; rather, we have a classification
To test this relationship, we examine whether colored by the respondents’ answers during the
racial classifications shift over time for the same survey interviews. This is ideal for the purposes
individuals and whether such shifts are related of assessing the effects of social status on racial
to changes in social position. We focus on as- classification, because the interviewers heard a
pects of social status for which racial differences range of status information about the respon-
in outcomes are well known and linked to group dents prior to recording the respondents’ race.
stereotypes, such as incarceration, unemploy- Of the observations for which respondents have
ment, and unmarried parenthood, and we ask racial classifications in consecutive survey years,
whether changes along these status dimensions 6 percent are described by a different race than
also shape racial perceptions—by influencing the in the previous person-year, and 20 percent of
way individuals are perceived racially by others. the individuals in the sample experienced at least
We begin with descriptive comparisons using one change in how they were racially classified
sample restrictions to demonstrate that changes between 1979 and 1998.
in racial classification are related to changes in Respondents were also asked to self-identify
social position. We construct our outcomes as their “origin or descent” in 1979 and their “race
dichotomous variables to explore which charac- or races” in 2002. We distinguish these as dif-
teristics move individuals either toward or away ferent dimensions of race: racial identification,
from the traditional poles of privilege and stigma naming oneself as a member of a racial group (as
in the United States: whiteness and blackness. on a survey); and racial classification, the catego-
For example, among people who were previ- rization of others as members of particular racial
ously classified as white and had never been un- populations. Each dimension has consequences
employed, we report how differences in current for both an individual’s life chances and how re-
employment status are related to differences in searchers understand the dynamics of racial in-
the percentage of respondents who are classified equality (Saperstein 2006). The analyses below
as white in a given year. We then present pre- focus on racial classification because it is directly
dicted probabilities from a multivariate logistic linked to discrimination and stereotypical ex-
regression model examining whether respon- pectations, and because we have up to seventeen
dents are classified as white (vs. nonwhite) in a years of classification data, compared to just two
given year, controlling for how they were classi- for racial identification. Where possible, we used
fied previously. information on racial identification to conduct
690 • Part VII: Race and Ethnicity

Table 80.1. Comparison of Interviewers’ Racial Classification of Respondents Between 2 Interviews

Race in Current Year

Race in
Previous Year White Black Other Total
White 95.8 0.5 3.7 100
N 103,690 554 4,036 108,280
Black 1.3 98.3 0.5 100
N 553 42,550 199 43,302
Other 45.1 2.3 52.6 100
N 4,041 205 4,708 8,954
Total 67.5 27.0 5.6 100
N 108,284 43,309 8,943 160,536

Note: N is given in person-years. Percentages and Ns sum across the rows, but may not add up to 100 due to rounding.
Table includes all observations in which respondents had nonmissing racial classifications in two consecutive surveys.
Source: National Longitudinal Survey of Youth (NLSY) 1979.

similar analyses and confirm that our results can classified as white, black, or other by the inter-
be extended to how people describe themselves viewer in the current year. If race did not change
(see Saperstein and Penner 2012 for more details). over time, all of the observations would fall in
the three cells representing consistent classifica-
tion. Observations in the other cells indicate in-
Social Status and Other Variables consistent classification, for example, 3.7 percent
In the logistic regression analysis, our primary of respondents classified as white in the previous
independent variables are repeated measures year are currently perceived as “other.” Overall,
of the respondent’s social position, including this array suggests that the racial classification of
long-term unemployment, poverty, incarcera- Americans is more flexible than is commonly ac-
tion, and marital and parental status. We also counted for in models of racial inequality.
control for other respondent characteristics, We next demonstrate how social status influ-
including country of birth, region of residence, ences racial classification. Figure 80.1 illustrates
gender, age, and whether the respondent identi- how long-term unemployment leads to changes
fied as multiracial or of Hispanic origin, as well in respondents’ racial classification; other status
as the interviewer’s race, gender, education, and factors provide similar results. The figure in-
age. We include year fixed effects to account for cludes two comparisons, depicting the percent-
societal changes in the definition of race, along age of respondents classified as either black or
with other year-to-year changes in survey design, white in the current year. The observations are
question wording, or interviewer training. limited to people who were classified as black
or white, respectively, in the previous year. To
capture changes in employment status, we limit
Results observations to respondents who as of the previ-
We begin by establishing the extent of fluidity ous year had never been long-term unemployed.
in individuals’ racial classification in the United These sample restrictions allow us to compare
States. Table 80.1 reports year-to-year changes the current racial classifications of respondents
in racial classification in percentage terms. Each who experienced a change in their employment
row displays the racial classification of respon- status in the intervening year with the classifi-
dents in the previous year by whether they were cations of respondents who previously were
 Saperstein and Penner—The Dynamics of Racial Fluidity and Inequality • 691

Figure 80.1. Changes in Social Status and Racial Classification

100

Percent classified the same as in the previous wave

98

Employed
96 Unemployed

94

92
Classified as black Classified as white
Notes: The percentage of people who were classified as black or white in the current survey (1980–1998), by whether they
became unemployed for the first time. Restricted to those classified as black or white, respectively, in the previous survey
and who were never previously unemployed. Unemployed indicates whether the respondent was unemployed for more
than 4 months in the calendar year prior to the current survey. Error bars are ± 1SE.
Source: NLSY 1979.

classified the same way, but did not experience a a hypothetical man who was classified as white
change in social position. in the previous year and had never been incar-
Both comparisons in Figure 80.1 reveal clas- cerated, long-term unemployed, or in poverty
sification differences in the expected directions: would be classified as white again. Each sub-
A higher percentage of unemployed respondents sequent bar in the figure makes one additional
are classified as black and a lower percentage are change to his status-related characteristics, so
classified as white, even if they were classified as that the final bar illustrates the probability of
such in the previous year. Although the differ- being classified as white in the current year for
ences are relatively small—for example, 96 per- a man who is now unmarried, has been incar-
cent of continuously employed respondents were cerated, has been unemployed for more than
again classified as white the following year, com- four months, and whose annual income is below
pared to 95 percent of respondents who lost their the federal poverty line. This accumulation of
jobs in the interim—these represent year-to-year negative status results in a decreased likelihood
changes, and the differences for both black and of being classified as white, from 96 percent at
white classification are statistically significant (p baseline to 85 percent.
< .05). To the extent that status changes accu- To further explore this relationship, we also
mulate, even these small effects can play an im- estimated four types of supplemental models.2
portant role in shaping racial classification over First, we estimated multinomial logistic regres-
the life course, as we illustrate in the following sion models that allow for changes among all
figure. three racial categories available to survey inter-
Figure 80.2 depicts the cumulative effects viewers (black, white, and “other”); these mod-
of multiple status changes, drawing on the pre- els confirm that social status shapes changes
dicted probabilities from a logistic regression among all three categories (and not only changes
model estimating a person’s odds of being classi- from white to other, or other to black). Sec-
fied as white.1 We start with the probability that ond, we estimated models restricted to only
692 • Part VII: Race and Ethnicity

Figure 80.2. Cumulative Effect of Multiple Status Changes on Racial Classification

1
Probability of being classified as white

0.95

0.9

0.85

0.8
Baseline +Incarcerated +Unemployed +Divorced +Poverty

Note: The baseline category represents the predicted probability that a 29-year-old married father with a high school
education, who lives in a nonsouthern inner city, does not identify as Hispanic or multiracial, and was classified as white
in the previous year, was seen as white in 1990.
Source: NLSY 1979.

non-multiracial, non-Hispanic, non-American such changes in an individual’s race should not

Indian respondents to rule out the possibil- be thought of as permanent. Table 80.2 presents
ity that our findings were driven by the small illustrative cases of respondents who have expe-
minority of respondents for whom racial per- rienced long-term unemployment. These racial
ceptions were already believed to be fluid and classification histories underscore that though
complex. Third, we estimated models with respondents are more likely to be seen as black
person-fixed effects, which show that our results postunemployment than they were preunem-
are not fully explained by unmeasured respon- ployment, they are not now always seen as black
dent characteristics that are relatively stable over (Panel A). Similarly, people are more likely to
time, such as family ancestry or skin tone. Fi- be seen as nonwhite postunemployment than
nally, we estimated models that include racial they were before losing their jobs, but this does
classifications and status factors from each of not mean that they are never again seen as white
the five previous years to confirm that the sta- (Panel B). What seems to have changed is the
tus factors still help to explain a person’s current person’s likelihood of being seen as black or
racial classification above and beyond knowing white in any given encounter.
his or her recent history of racial classification. These findings support defining race as a flex-
Though each of our modeling strategies has its ible propensity that changes over time and across
strengths and weaknesses, we argue that in com- contexts, rather than as a characteristic that is
bination they provide consistent evidence of the ascribed at birth and fixed. Race does not sim-
reciprocal relationship between racial fluidity ply affect a person’s social status; a loss or gain
and inequality in the United States. in status can alter both how people identify and
how others perceive them. Americans who lose
their jobs are more likely to be classified and
Discussion identify as black, whereas people who get mar-
Our results clearly demonstrate that racial clas- ried are more likely to be classified and identify
sifications change over time, and they do so in as white, regardless of how they were perceived
response to changes in social position. However, or identified previously. These effects are only
 Saperstein and Penner—The Dynamics of Racial Fluidity and Inequality • 693

Table 80.2. Illustrative Racial Classification Histories Pre- and Postunemployment

Panel A. Classified as Black

% Black % Black
Person ID Preunemployment Postunemployment Preunemployment Postunemployment

343 OBO BBBBBBBBBBBBBB 33 100

9266 W.OWW BOBOWOBBOBO. 0 45
9372 OOO BBBBBBBBBBBOBB 0 93

Panel B. Classified as White

% White % White
Person ID Preunemployment Postunemployment Preunemployment Postunemployment

8857 WOOWWWW WOOOOOOWOO 71 20

9282 WWWWWWWWOW OWWOOOO 90 29
9969 WWWWWWWWWWW WWOOOO 100 33

Note: B denotes classification as black, W as white, and O as other, and a period denotes missing classification data.
Percentages reflect the number of years the respondent was classified as the given race relative to all years with nonmissing
classifications. Unemployment indicates whether the respondent was unemployed for more than 4 months in the previous
calendar year.
Source: NLSY 1979.

strengthened when one change in social position for abandoning racial categorization because it
triggers, or is accompanied by, others—such as is unnecessarily divisive and creates boundaries
when marriage increases household income or that would not otherwise exist (e.g., Connerly
unemployment follows incarceration. 2001). However, we do not believe that the idea
Of course not every change in a person’s so- that race is fluid at the individual level should be
cial position results in racial fluidity. Changes in interpreted as evidence that racial divisions are
status that are stereotype inconsistent drive racial only imagined. Race remains real because it has
fluidity, whereas stereotype consistent shifts in important consequences for people’s life chances.
status reinforce racial stability. Also, to trigger Our results show that racial divisions, like other
an observed change, the propensity to be clas- aspects of social structure, do not simply happen
sified as a particular race presumably must cross to people; racial inequality is actively (if some-
a certain threshold. People can be thought of as times unintentionally) reproduced in everyday
having different starting propensities to be clas- interactions.
sified as a given race: some will have relatively This process has implications for understand-
high baseline propensities to be racially classified ing racial discrimination. Like most research on
in a variety of ways, whereas others will have rel- racial inequality, studies typically assume that a
atively low, near-zero baseline propensities to be person’s race is fixed and self-evident. Our results
classified in more than one category. Neverthe- are an important reminder that racial percep-
less, our results suggest that a stint in prison or tions are central to the process of discrimination,
falling into poverty will increase everyone’s odds and that an act of discrimination first requires
of being seen as black and decrease everyone’s classifying an individual as belonging in a par-
odds of being seen as white in future encounters. ticular racial category.
Finally, we suggest that thinking about race
as fluid, ironically, helps explain the rigidity of
Conclusion racial hierarchy in the United States. If some
Demonstrations of the complexity and social Americans who experience an increase in sta-
construction of race are often met with calls tus are “whitened” as a result of their upward
694 • Part VII: Race and Ethnicity

mobility, and some who experience downward Freeman, Jonathan B., Andrew M. Penner, Aliya
mobility are “darkened,” these changes serve to Saperstein, Matthias Scheutz, and Nalini Am-
reinforce both racial inequality and existing ste- bady. 2011. “Looking the Part: Social Status Cues
reotypes. That is, race implies—and is defined in Shape Race Perception.” PLoS One 6 (9): e25107.
Gates, Henry Louis, Jr. 1997. “The Passing of
part by—a set of expectations by which people
Anatole Broyard.” In Thirteen Ways of Looking
are continually judged in everyday interactions at a Black Man, 180–214. New York: Random
(see also Gross 1998). When a white person is House.
someone who does what a white person is sup- Gilens, Martin. 1995. “Racial-Attitudes and Opposi-
posed to do, and a black person is someone who tion to Welfare.” Journal of Politics 57: 994–1014.
does what a black person is supposed to do, then Greenwald, Anthony G., Debbie E. McGhee, and
race is not just an input into the stratification Jordan L. K. Schwartz. 1998. “Measuring Indi-
system; it is an output, as well. vidual Differences in Implicit Cognition: The
Implicit Association Test.” Journal of Personality
NOTES and Social Psychology 74 (6): 1464–1480.
1. As noted previously, this model includes all Gross, Ariela J. 1998. “Litigating Whiteness: Trials of
measures of social position simultaneously, along Racial Determination in the Nineteenth Century
with an indicator of whether the respondent was U.S. South.” Yale Law Journal 108 (1): 109–188.
classified as white in the previous year, and controls Harris, David R., and Jeremiah Joseph Sim. 2002.
for other respondent characteristics, interviewer “Who Is Multiracial? Assessing the Complexity
characteristics, and year fixed effects. See Saperstein of Lived Race.” American Sociological Review 67:
and Penner (2012) for more details. 614–627.
2. See Saperstein and Penner (2012) for these Hurwitz, J., and M. Peffley. 2005. “Playing the Race
results. Card in the Post-Willie Horton Era—The Im-
pact of Racialized Code Words on Support for
REFERENCES Punitive Crime Policy.” Public Opinion Quarterly
Allen, B. P. 1996. “African Americans’ and Euro- 69: 99–112.
pean Americans’ Mutual Attributions: Adjective Jacobson, Matthew Frye. 1999. Whiteness of a Differ-
Generation Technique Stereotyping.” Journal of ent Color: European Immigrants and the Alchemy
Applied Social Psychology 26: 884–912. of Race. Cambridge, MA: Harvard University
American Sociological Association (ASA). 2003. The Press.
Importance of Collecting Data and Doing Social Khanna, Nikki, and Cathryn Johnson. 2010. “Pass-
Scientific Research on Race. Washington, DC: ing as Black: Racial Identity Work Among Bi-
American Sociological Association. racial Americans.” Social Psychology Quarterly 73
Connerly, Ward. 2001. “The Irrelevance of Race.” (4): 380–397.
San Francisco Chronicle, July 8, C8. Loveman, Mara, and Jeronimo O. Muniz. 2007.
Craemer, Thomas. 2010. “Ancestral Ambivalence “How Puerto Rico Became White: Boundary
and Racial Self-Classification Change.” Social Dynamics and Intercensus Racial Reclassifi-
Science Journal 47: 307–325. cation.” American Sociological Review 72 (6):
Doyle, Jamie M., and Grace Kao. 2007. “Are Ra- 915–939.
cial Identities of Multiracials Stable? Changing Massey, Douglas. 2007. Categorically Unequal: The
Self-Identification among Single and Multiple American Stratification System. New York: Russell
Race Individuals.” Social Psychology Quarterly 70 Sage Foundation.
(4): 405–423. Ossorio, Pilar, and Troy Duster. 2005. “Race and
Eberhardt, Jennifer L., Phillip Atiba Goff, Valeria J. Genetics—Controversies in Biomedical, Behav-
Purdie, and Paul G. Davies. 2004. “Seeing Black: ioral and Forensic Sciences.” American Psycholo-
Race, Crime and Visual Processing.” Journal of gist 60 (1): 115–128.
Personality and Social Psychology 87 (6): 876–893. Richeson, Jennifer A., and Sophie Trawalter. 2005.
Fischer, Claude S., and Michael Hout. 2006. Cen- “On the Categorization of Admired and Dis-
tury of Difference: How America Changed in the liked Exemplars of Admired and Disliked Racial
Last One Hundred Years. New York: Russell Sage. Groups.” Journal of Personality and Social Psychol-
ogy 89 (4): 517–530.
 Saperstein and Penner—The Dynamics of Racial Fluidity and Inequality • 695

Saperstein, Aliya. 2006. “Double-checking the Race Wimmer, Andreas. 2008. “The Making and Un-
Box: Examining Inconsistency between Survey making of Ethnic Boundaries: A Multilevel Pro-
Measures of Observed and Self-Reported Race.” cess Theory.” American Journal of Sociology 113
Social Forces 85 (1): 57–74. (4): 970–1022.
Saperstein, Aliya, and Andrew M. Penner. 2012. Zuberi, Tukufu. 2000. “Deracializing Social Statis-
“Racial Fluidity and Inequality in the United tics: Problems in the Quantification of Race.”
States.” American Journal of Sociology 113 (3): Annals of the American Academy of Political and
676–727. Social Science 568: 172–185.
Tilly, Charles. 1998. Durable Inequalities. Berkeley:
University of California Press.
696 • Part VII: Race and Ethnicity

Classic Modes of Incorporation

81 • Edna Bonacich

A Theory of Ethnic Antagonism

The Split Labor Market

Societies vary considerably in their degree of respectively. Apart from manifesting antagonism
ethnic and racial antagonism. Such territories between ethnic elements, exclusion and caste
as Brazil, Mexico, and Hawaii are generally seem to have little in common. In the one, an
acknowledged to be relatively low on this di- effort is made to prevent an ethnically differ-
mension; while South Africa, Australia, and the ent group from being part of the society. In the
United States are considered especially high. Lit- other, an ethnically different group is essential
erally hundreds of variables have been adduced to the society: it is an exploited class supporting
to account for these differences, ranging from the entire edifice. The Deep South felt it could
religions of dominant groups, to whether the not survive without its black people; the Pacific
groups who migrate are dominant or subordi- coast could not survive with its Japanese. This
nate, to degrees of difference in skin color, to an puzzle may be used as a touchstone for solving
irreducible “tradition” of ethnocentrism. While the general problem of ethnic antagonism, for to
some writers have attempted to synthesize or sys- be adequate a theory must be able to explain it.
tematize some subset of these (e.g., Lieberson, The theory presented here is, in part, a syn-
1961; Mason, 1970; Noel, 1968; Schermerhorn, thesis of some of the ideas used by Oliver Cox to
1970; van den Berghe, 1966), one is generally explain the Japanese-white conflict on the U.S.
struck by the absence of a developed theory ac- Pacific coast (Cox, 1948:408–22), and by Mar-
counting for variations in ethnic antagonism. vin Harris to analyze the difference between Bra-
One approach to this problem is to consider zil and the Deep South in rigidity of the “color
an apparent anomaly, namely that ethnic antago- line” (Harris, 1964:79–94). It stresses the role of
nism has taken two major, seemingly antithetical a certain kind of economic competition in the
forms: exclusion movements, and so-called caste development of ethnic antagonism. Economic
systems.1 An example of the former is the “white factors have, of course, not gone unnoticed,
Australia” policy; while South Africa’s color though until recently sociological literature has
bar illustrates the latter. The United States has tended to point them out briefly, then move on
shown both forms, with a racial caste system in to more “irrational” factors (even such works as
the South and exclusion of Asian and “new” im- The Economics of Discrimination, Becker, 1957).
migrants2 from the Pacific and eastern seaboards A resurgence of Marxian analysis (e.g., Blauner,

Edna Bonacich. “A Theory of Ethnic Antagonism: The Split Labor Market,” American Sociological Review (Oc-
tober 1972), pp. 547–559. Copyright © 1972 by the American Sociological Association. Used by permission
of the American Sociological Association and the author.
 Bonacich—A Theory of Ethnic Antagonism • 697

1969; Reich, 1971) has thrust economic consid- culmination of many pronouncements, actions,
erations to the fore, but I shall argue that even and enactments, and are continuously supported
this approach cannot adequately deal with the by more of the same. “Antagonism” was chosen
problem posed by exclusion movements and over terms like prejudice and discrimination be-
caste systems. In addition, both Marxist and cause it carries fewer moralistic and theoretical
non-Marxist writers assume that racial and cul- assumptions (see Schermerhorn, 1970:6–9). For
tural differences in themselves prompt the de- example, both of these terms see conflict as em-
velopment of ethnic antagonism. This theory anating primarily from one side: the dominant
challenges that assumption, suggesting that eco- group. Antagonism allows for the possibility that
nomic processes are more fundamental. conflict is mutual; i.e. a product of interaction.
No effort is made to prove the accuracy of
the following model. Such proof depends on a
lengthier exposition. Historical illustrations are The Split Labor Market
presented to support it. The central hypothesis is that ethnic antagonism
first germinates in a labor market split along eth-
nic lines. To be split, a labor market must con-
Ethnic Antagonism tain at least two groups of workers whose price of
“Ethnic” rather than “racial” antagonism was labor differs for the same work, or would differ
selected as the dependent variable because the if they did the same work. The concept “price
former is seen to subsume the latter. Both terms of labor” refers to labor’s total cost to the em-
refer to groups defined socially as sharing a com- ployer, including not only wages, but the cost
mon ancestry in which membership is therefore of recruitment, transportation, room and board,
inherited or ascribed, whether or not members education, health care (if the employer must bear
are currently physically or culturally distinctive.3 these), and the costs of labor unrest. The degree
The difference between race and ethnicity lies in of worker “freedom” does not interfere with this
the size of the locale from which a group stems, calculus; the cost of a slave can be estimated in
races generally coming from continents, and eth- the same monetary units as that of a wage earner,
nicities from national sub-sections of continents. from his purchase price, living expenses, policing
In the past the term “race” has been used to refer requirements, and so on.
to both levels, but general usage today has re- The price of a group of workers can be
versed this practice (e.g., Schermerhorn, 1970; roughly calculated in advance and comparisons
Shibutani and Kwan, 1965). Ethnicity has be- made even though two groups are not engaged
come the generic term. in the same activity at the same time. Thus in
Another reason for choosing this term is that 1841 in the colony of New South Wales, the
exclusion attempts and caste-like arrangements Legislative Council’s Committee on Immigra-
are found among national groupings within a ra- tion estimated the relative costs of recruiting
cial category. For example, in 1924 whites (Eu- three groups of laborers to become shepherds.
ropeans) attempted to exclude whites of different Table 81.1 shows their findings. The estimate of
national backgrounds from the United States by free white labor, for example, was based on what
setting up stringent immigration quotas. it would take to attract these men from compet-
The term “antagonism” is intended to en- ing activities.
compass all levels of intergroup conflict, includ-
ing ideologies and beliefs (such as racism and
prejudice), behaviors (such as discrimination, Factors Affecting the Initial Price of
lynchings, riots), and institutions (such as laws Labor
perpetuating segregation). Exclusion move- Labor markets that are split by the entrance of
ments and caste systems may be seen as the a new group develop a dynamic which may in
698 • Part VII: Race and Ethnicity

Table 81.1. Estimated Cost of Three Types of Labor to Be Shepherds in New South Wales,
1841

Free Man Prisoner Coolie

(White) (White) (Indian)
£ s. d. £ s. d. £ s. d.
Rations 16 18 0 13 14 4 9 6 4
Clothing — — — 3 3 0 1 1 8
Wages 25 0 0 — — — 6 0 0
Passage from India — — — — — — 2 0 0
Total per Annum 41 18 0 16 17 4 18 8 0
Source: From Yarwood (1968:13).

turn affect the price of labor. One must therefore available labor markets have, of course, been
distinguish initial from later price determinants. most extensively drawn upon.
The initial factors can be divided into two broad
categories: resources and motives. Workers need not accept the original wage
agreement for long after they have immigrated,
since other opportunities may exist; for instance,
1. Resources there may be ample, cheap land available for
Three types of resources are important price de- individual farming. One capitalist device for
terminants. These are: keeping wages low at least for a time is to bind
immigrants to contracts before they leave the old
a. Level of Living, or Economic Resources. The economy. The Indian indenture system, for ex-
ethnic groups forming the labor market in a ample, rested on such an arrangement (Gillion,
contact situation derive from different economic 1962:19–38).
systems, either abroad or within a conquered
territory. For members of an ethnic group to b. Information. Immigrants may be pushed into
be drawn into moving, they must at least raise signing contracts out of ignorance. They may
their wage level. In general, the poorer the econ- agree to a specific wage in their homeland not
omy of the recruits, the less the inducement knowing the prevailing wage in the new country,
needed for them to enter the new labor mar- or having been beguiled by a false account of
ket. Crushing poverty may drive them to sell life and opportunity there. Williams (1944:11),
their labor relatively cheaply. For example, Lind for example, describes some of the false promises
(1968:199) describes the effect of the living level made to draw British and Germans as workers to
on the wage scale received by immigrant work- West Indian sugar plantations before the advent
ers to Hawaii: of African slavery. Chinese labor to Australia was
similarly “obtained under ‘false and specious pre-
In every case [of labor importations] the su- tences’” (Willard, 1967:9).
perior opportunities for gaining a livelihood The possibilities for defrauding a population
have been broadcast in regions of surplus lacking access to the truth are obvious. In gen-
manpower, transportation facilities have been eral, the more people know about conditions
provided, and finally a monetary return larger obtaining in the labor market to which they are
than that already received has been offered moving, the better can they protect themselves
to the prospective laborer. The monetary against disadvantageous wage agreements.
inducement has varied considerably, chiefly
according to the plane of living of the pop- c. Political Resources. By political resources
ulation being recruited, and the cheapest I mean the benefits to a group of organizing.
 Bonacich—A Theory of Ethnic Antagonism • 699

Organization can exist at the level of labor, or home behind and whose employers need not
it can occur at higher levels, for example, in a bear the cost of housing and educating their
government that protects them. These levels are families. Even when families accompany them,
generally related in that a strong government can such workers tend to be willing to accept a lower
help organize its emigrants. There are exceptions, standard of living since it is only short term.
however: strong emigrant governments tend not Second, temporary workers avoid involve-
to extend protection to their deported convicts ment in lengthy labor disputes. Since they will
or political exiles; and some highly organized be in the labor market a short while, their main
groups, like the Jews in the United States, have concern is immediate employment. They may
not received protection from the old country. be willing to undercut wage standards if need
Governments vary in the degree to which they be to get a job and are therefore ripe candidates
protect their emigrants. Japan kept close watch for strikebreaking. Permanent workers also stand
over the fate of her nationals who migrated to to lose from lengthy conflict, but they hope for
Hawaii and the Pacific coast; and the British co- benefits to their progeny. If temporary workers
lonial government in India tried to guard against are from elsewhere, they have no such interest
abuses of the indenture system (for example, by in future business-labor relations. Altogether,
refusing to permit Natal to import Indian work- temporary workers have little reason to join the
ers for their sugar plantations until satisfactory organizations and unions of a permanent work
terms had been agreed to; cf. Ferguson-Davie, force, and tend not to do so.
1952:4–10). In contrast Mexican migrant
workers to the United States have received lit- a. Fixed or Supplementary Income Goal. Some
tle protection from their government, and Afri- temporary workers enter the market either to
can states were unable to intervene on behalf of supplement family income, or to work toward a
slaves brought to America. Often the indigenous specific purchase. The worker’s standard of living
populations of colonized territories have been does not, therefore, depend on his earnings on
politically weak following conquest. Thus Afri- the job in question, since his central source of
can nations in South Africa have been unable to employment or income lies elsewhere. Examples
protect their migrant workers in the cities. of this phenomenon are to be found throughout
In general, the weaker a group politically, the Africa:
more vulnerable it is to the use of force, hence
to an unfavorable wage bargain (or to no wage . . . the characteristic feature of the labor
bargain at all, as with slavery). The price of a market in most of Africa has always been the
labor group varies inversely with the amount of massive circulation of Africans between their
force that can be used against it, which in turn villages and paid employment outside. In
depends on its political resources. some places villagers engage in wage-earning
seasonally. More commonly today they work
for continuous though short-term periods of
2. Motives roughly one to three years, after which they
Two motives affect the price of labor, both re- return to the villages. . . . the African villager,
lated to the worker’s intention of not remain- the potential migrant into paid employment,
ing permanently in the labor force. Temporary has a relatively low, clearly-defined and rigid
workers tend to cost less than permanent work- income goal; he wants money to pay head
ers for two reasons. First, they are more willing and hut taxes, to make marriage payments
to put up with undesirable work conditions since required of prospective bridegrooms, or to
these need not be endured forever. If they are purchase some specific consumer durable
migrants, this tolerance may extend to the gen- (a bicycle, a rifle, a sewing machine, a given
eral standard of living. Often migrant temporary quantity of clothing or textiles, etc.) (Berg,
workers are males who have left the comforts of 1966:116–8).
700 • Part VII: Race and Ethnicity

Such a motive produces the “backward-sloping And so on. The lower wage level of Japanese
labor supply function” characteristic of many workers reflects both a lower standard of living,
native peoples in colonized territories. In addi- and a desire to get a foothold in the labor mar-
tion to the general depressing effects on wages ket. As Iwata (1962:27) puts it: “Their willing-
of being temporary, this motive leads to a fairly ness to accept even lower wages than laborers of
rapid turnover in personnel, making organiza- other races enabled the Japanese to secure em-
tion more difficult and hindering the develop- ployment readily.”
ment of valuable skills which could be used for Millis (1915:155) describes a basket factory
bargaining. If wages were to rise, workers would in Florin, California, where Japanese workers
reach their desired income and withdraw more had displaced white female workers because the
quickly from the market, thereby lessening their latter were unwilling to work more than ten
chances of developing the political resources nec- hours a day or on weekends. The Japanese, anx-
essary to raise their wages further. ious to return to Japan as quickly as possible,
were willing to work twelve to fourteen hours
b. Fortune Seeking. Many groups, commonly per day and on weekends, thereby saving their
called sojourners (see Siu, 1952), migrate long employers the costs of a special overtime work
distances to seek their fortune, with the ulti- force.
mate intention of improving their position in The Japanese immigrants developed political
their homeland. Such was the case with Japanese resources through a high degree of community
immigrants on the west coast and Italian immi- organization. This could be used for the conve-
grants in the east. Such workers stay longer in nience of the employer, by solving his recruit-
the labor market, and can develop political re- ment problems, seeing that work got done, and
sources. However, since they are temporary they providing workers with board and lodging. In the
have little incentive to join the organizations of case of seasonal labor, the Japanese community
the settled population. Instead they tend to cre- could provide for members during the off-season
ate competing organizations composed of people by various boarding arrangements and clubs, and
who will play a part in their future in the home- by transporting labor to areas of demand (Ichi-
land, i.e. members of the same ethnic group. hashi, 1932:172–6; Millis, 1915:44–5). These
Sojourner laborers have at least three features conveniences saved the employer money.4
which affect the price of labor: lower wages, lon- As the reader may have noted, I have omitted
ger hours, and convenience to the employer. The a factor usually considered vital in determin-
Japanese show all three. Millis (1915:45) cites ing the price of labor, i.e. differences in skills.
the U.S. Immigration Commission on the ques- I would contend, however, that this does not
tion of relative wages: in itself lead to that difference in price for the
same work which distinguishes a split labor mar-
The Japanese have usually worked for a lower ket. While a skilled worker may be able to get
wage than the members of any other race save a higher paying job, an unskilled laborer of an-
the Chinese and the Mexican. In the salmon other ethnicity may be trained to fill that job for
canneries the Chinese have been paid higher the same wage. Skills are only indirectly impor-
wages than the Japanese engaged in the same tant in that they can be used to develop political
occupations. In the lumber industry, all resources, which in turn may lead to a difference
races, including the East Indian, have been in wage level for the same work.
paid higher wages than the Japanese doing
the same kind of work. As section hands and
laborers in railway shops they have been paid Price of Labor and Ethnicity
as much or more than the Mexicans, but as a Ethnic differences need not always produce a
rule less than the white men of many races. price differential. Thus, if several ethnic groups
 Bonacich—A Theory of Ethnic Antagonism • 701

who are approximately equal in resources and/ differences in the initial price of labor only seem
or goals enter the same economic system, a split to be a factor because resources have histori-
labor market will not develop. Alternatively, cally been roughly correlated with color around
in a two-group contact situation, if one ethnic the world.5 When color and resources are not
group occupies the position of a business elite correlated in the “expected” way, then I would
and has no members in the labor force (or in a predict that price follows resources and motives
class that could easily be pushed into the labor rather than color.
force, e.g. low-capital farmers) then regardless In sum, the prejudices of business do not de-
of the other group’s price, the labor market will termine the price of labor, darker skinned or cul-
not be split. This statement is a generalization of turally different persons being paid less because
the point made by Harris (1964) that the criti- of them. Rather, business tries to pay as little as
cal difference in race relations between the deep possible for labor, regardless of ethnicity, and is
south and Brazil was that the former had a white held in check by the resources and motives of
yeomanry in direct competition with ex-slaves, labor groups. Since these often vary by ethnic-
while the Portuguese only occupied the role of a ity, it is common to find ethnically split labor
business elite (plantation owners). markets.
Conversely, a split labor force does not only
stem from ethnic differences. For example, prison
and female labor have often been cheaper than The Dynamics of Split Labor Markets
free male labor in Western societies. Prison labor In split labor markets, conflict develops between
has been cheap because prisoners lack political three key classes: business, higher paid labor, and
resources, while women often labor for supple- cheaper labor. The chief interests of these classes
mentary incomes (cf. Hutchinson, 1968:59–61; are as follows:
Heneman and Yoder, 1965:543–4).
That initial price discrepancies in labor
should ever fall along ethnic lines is a function 1. Business or Employers
of two forces. First, the original wage agreement This class aims at having as cheap and doc-
arrived at between business and new labor often ile a labor force as possible to compete effec-
takes place in the labor group’s point of origin. tively with other businesses. If labor costs are
This is more obviously a feature of immigrant too high (owing to such price determinants as
labor, but also occurs within a territory when unions), employers may turn to cheaper sources,
conquered peoples enter their conquerors’ econ- importing overseas groups or using indige-
omy. In other words, the wage agreement is nous conquered populations. In the colony of
often concluded within a national context, these Queensland in Australia, for example, it was be-
nationalities coming to comprise the ethnic el- lieved that cotton farming would be the most
ements of the new labor market. One would suitable economic enterprise:
thus expect the initial wages of co-nationals to
be similar. However, such plantations (being too large)
Second, nations or peoples that have lived could not be worked, much less cleared, by
relatively separately from one another are likely their owners; neither could the work be done
to have developed different employment mo- by European laborers because sufficient num-
tives and levels of resources (wealth, organiza- bers of these were not available—while even
tion, communication channels). In other words, had there been an adequate supply, the high
the factors that affect the price of labor are likely rates of wages would have been prohibitive.
to differ grossly between nations, even though This was a consideration which assumed vast
there may be considerable variation within each importance when it was realized that cotton
nation, and overlap between nations. Color would have to be cultivated in Queensland at
702 • Part VII: Race and Ethnicity

a considerably lower cost than in the United independent farmers. The displaced class may
States in order to compensate for the heavier move on (as occurred in many of the West Indies
freights from Queensland—the more distant when African slave labor was introduced to raise
country from England. It seemed then that sugar), but if it remains, it comes to play the role
there was no possibility of successful compe- of higher paid labor.
tition with America unless the importation The presence of cheaper labor in areas of the
of some form of cheap labor was permitted. economy where higher paid labor is not cur-
(Moles, 1968:41) rently employed is also threatening to the latter,
since the former attract older industries. The
Cheaper labor may be used to create a new in- importance of potential competition cannot be
dustry having substantially lower labor costs than over-stressed. Oftentimes writers assert the irra-
the rest of the labor market, as in Queensland. tionality of ethnic antagonism when direct eco-
Or they may be used as strikebreakers or re- nomic competition is not yet in evidence owing
placements to undercut a labor force trying to to few competitors having entered the labor mar-
improve its bargaining position with business. If ket, or to competitors having concentrated in a
cheap labor is unavailable, business may turn to few industries. Thus Daniels (1966:29) belittles
mechanization, or try to relocate firms in areas the role of trade unions in the Asiatic Exclusion
of the world where the price of labor is lower. League by describing one of the major contrib-
utors as “an organization whose members, like
most trade unionists in California, were never
2. Higher Paid Labor faced with job competition from Japanese.” It
This class is very threatened by the introduction does not take direct competition for members
of cheaper labor into the market, fearing that it of a higher priced labor group to see the possible
will either force them to leave the territory or re- threat to their well-being, and to try to prevent
duce them to its level. If the labor market is split its materializing. If they have reason to believe
ethnically, the class antagonism takes the form of many more low-priced workers are likely to fol-
ethnic antagonism. It is my contention (follow- low an initial “insignificant trickle” (as Daniels,
ing Cox, 1948:411n) that, while much rhetoric 1966:1, describes the Japanese immigration,
of ethnic antagonism concentrates on ethnicity failing to mention that it was insignificant pre-
and race, it really in large measure (though prob- cisely because a larger anticipated flow had been
ably not entirely) expresses this class conflict. thwarted, and diverted to Brazil), or if they see a
The group comprising higher paid labor large concentration of cheaper labor in a few in-
may have two components. First, it may include dustries which could easily be used to undercut
current employees demanding a greater share of them in their own, they will attempt to forestall
the profits or trying to maintain their position undercutting.
in the face of possible cuts. A second element Lest you think this fear misguided, take
is the small, independent, entrepreneur, like the note that, when business could override the in-
subsistence farmer or individual miner. The in- terests of more expensive labor, the latter have
troduction of cheaper labor into these peoples’ indeed been displaced or undercut. In British
line can undermine their position, since the Guiana the local labor force, composed mainly
employer of cheaper labor can produce at lower of African ex-slaves, called a series of strikes in
cost. The independent operator is then driven 1842 and 1847 against planters’ attempts to re-
into the labor market. The following sequence duce their wages. Plantation owners responded
occurs in many colonies: settlement by farmers by using public funds to import over 50,000
who work their own land, the introduction of cheaper East Indian indentured workers (Des-
intensive farming using cheaper labor, a rise in pres, 1969). A similar situation obtained in
land value and a consequent displacement of Mississippi, where Chinese were brought in to
 Bonacich—A Theory of Ethnic Antagonism • 703

undercut freed blacks. Loewen (1971:23) de- does not produce such antagonism, though it
scribes the thinking of white land-owners: “the too threatens people with the loss of their job.
‘Chinaman’ would not only himself supply a However, hiring practices will not necessarily
cheaper and less troublesome work force but in fall along ethnic lines, there being no advantage
addition his presence as a threatening alternative to the employer in hiring workers of one or an-
would intimidate the Negro into resuming his other ethnicity. All workingmen are on the same
former docile behavior.” Such displacement has footing, competing for scarce jobs (cf. Blalock,
occurred not only to nonwhite more expensive 1967:84–92, who uses this model of labor com-
labor, but, as the effects of slavery in the West petition). When one ethnic group is decidedly
Indies show, to whites by white capitalists. cheaper than another (i.e. when the labor market
is split) the higher paid worker faces more than
the loss of his job; he faces the possibility that
3. Cheaper Labor the wage standard in all jobs will be undermined
The employer uses this class partly to undermine by cheaper labor.
the position of more expensive labor, through
strikebreaking and undercutting. The forces
that make the cheaper group cost less permit Victory for More Expensive Labor
this to occur. In other words, either they lack If an expensive labor group is strong enough
the resources to resist an offer or use of force by (strength generally depending on the same fac-
business, or they seek a quick return to another tors that influence price), they may be able to
economic and social base. resist being displaced. Both exclusion and caste
With the possible exception of sojourners, systems represent such victories for higher paid
cheaper labor does not intentionally undermine labor.
more expensive labor; it is paradoxically its
weakness that makes it so threatening, for busi-
ness can more thoroughly control it. Cox makes 1. Exclusion
this point (1948:417–8) in analyzing why Pacific Exclusion movements generally occur when the
coast white and Asian workers could not unite in majority of a cheaper labor group resides outside
a coalition against business: a given territory but desires to enter it (often at
the request of business groups). The exclusion
. . . the first generation of Asiatic workers movement tries to prevent the physical pres-
is ordinarily very much under the control of ence of cheaper labor in the employment area,
labor contractors and employers, hence it is thereby preserving a non-split, higher priced
easier for the employer to frustrate any plans labor market.
for their organization. Clearly this cultural There are many examples of exclusion at-
bar helped antagonize white workers against tempts around the world. In Australia, for in-
the Asiatics. The latter were conceived of as stance, a group of white workers was able to
being in alliance with the employer. It would prevent capitalists from importing cheaper labor
probably have taken two or three generations from India, China, Japan, and the Pacific Islands.
before, say, the East Indian low-caste worker Attempts at importation were met with strikes,
on the Coast became sufficiently American- boycotts, petitions, and deputations (Willard,
ized to adjust easily to the policies and aims 1967:51–7). Ultimately, organized white labor
of organized labor. pressed for strong exclusion measures, and vig-
ilantly ensured their enforcement. As Yarwood
Ethnic antagonism is specifically produced (1964:151–2) puts it: “A comparison of the re-
by the competition that arises from a price dif- cords of various governments during our period
ferential. An oversupply of equal-priced labor [1896–1923] leaves no doubt as to the special
704 • Part VII: Race and Ethnicity

role of the Labour Party as the guardian of the

ports.” In other words, a white Australia policy 2. Caste
(i.e. the exclusion of Asian and Polynesian im- If cheaper labor is present in the market, and
migrants) appears to have sprung from a conflict cannot be excluded, then higher paid labor will
of interests between employers who wanted to resort to a caste arrangement, which depends
import cheap labor, and a labor force sufficiently on exclusiveness rather than exclusion. Caste is
organized to ward off such a move. essentially an aristocracy of labor (a term bor-
California’s treatment of Chinese and Jap- rowed from Lenin, e.g. 1964), in which higher
anese labor is another example of exclusion. A paid labor deals with the undercutting potential
socialist, Cameron H. King, Jr., articulates the of cheaper labor by excluding them from certain
threatened labor group’s position: types of work. The higher paid group controls
certain jobs exclusively and gets paid at one scale
Unskilled labor has felt this competition of wages, while the cheaper group is restricted to
[from the Japanese] for some time being com- another set of jobs and is paid at a lower scale.
pelled to relinquish job after job to the low The labor market split is submerged because the
standard of living it could not endure. The differentially priced workers ideally never occupy
unskilled laborers are largely unorganized and the same position.
voiceless. But as the tide rises it is reaching Ethnically distinct cheaper groups (as op-
the skilled laborers and the small merchants. posed to women, for example, who face a caste
These are neither unorganized nor voiceless, arrangement in many Western societies) may
and viewing the menace to their livelihood reside in a territory for two reasons: either they
they loudly demand protection of their mate- were indigenous or they were imported early in
rial interests. We of the Pacific Coast certainly capitalist-labor relations, when the higher paid
know that exclusion is an effective solution. group could not prevent the move. Two out-
In the seventh decade of the nineteenth cen- standing examples of labor aristocracies based on
tury the problem arose of the immigration ethnicity are South Africa, where cheaper labor
of Chinese laborers. The Republican and was primarily indigenous, and the U.S. South,
Democratic parties failed to give heed to the where they were imported as slaves.
necessities of the situation and the Working- Unlike exclusion movements, caste systems
man’s party arose and swept the state with retain the underlying reality of a price differen-
the campaign cry of “The Chinese must tial, for if a member of the subordinate group
go.” Then the two old parties woke up and were to occupy the same position as a member
have since realized that to hold the labor vote of the stronger labor group he would be paid
they must stand for Asiatic exclusion. (King, less. Hence, caste systems tend to become rigid
1908:665–6) and vigilant, developing an elaborate battery
of laws, customs, and beliefs aimed to prevent
King wrote this around the time of the Gen- undercutting. The victory has three facets. First,
tlemen’s Agreement, an arrangement of the U.S. the higher paid group tries to ensure its power
and Japanese governments to prevent further im- in relation to business by monopolizing the
migration of Japanese labor to the Pacific Coast acquisition of certain essential skills, thereby
(Bailey, 1934). The Agreement was aimed spe- ensuring the effectiveness of strike action, or
cifically at labor and not other Japanese immi- by controlling such important resources as pur-
grants, suggesting that economic and not racial chasing power. Second, it tries to prevent the
factors were at issue. immediate use of cheaper labor as undercutters
Exclusion movements clearly serve the inter- and strikebreakers by denying them access to
ests of higher paid labor. Its standards are pro- general education thereby making their training
tected, while the capitalist class is deprived of as quick replacements more difficult, or by en-
cheaper labor. suring through such devices as “influx control”
 Bonacich—A Theory of Ethnic Antagonism • 705

that the cheaper group will retain a base in their became accepted in the minds of the Euro-
traditional economies. The latter move ensures pean artisans as the means of maintaining
a backward-sloping labor supply function (cf. their own position against non-white inroads.
Berg, 1966) undesirable to business. Third, it (Doxey, 1961:23–4)
tries to weaken the cheaper group politically,
to prevent their pushing for those resources The final showdown between mine owners
that would make them useful as undercutters. and white workers occurred in the 1920s when
In other words, the solution to the devastating the owners tried to substitute cheaper non-white
potential of weak, cheap labor is, paradoxically, labor for white labor in certain semi-skilled oc-
to weaken them further, until it is no longer cupations. This move precipitated the “Rand
in business’ immediate interest to use them as Revolt,” a general strike of white workers on the
replacements. Witwatersrand, countered by the calling in of
South Africa is perhaps the most extreme troops and the declaration of martial law. The
modern example of an ethnic caste system. A result was a coalition between Afrikaner nation-
split labor market first appeared there in the min- alists (predominantly workers and small-scale
ing industry. With the discovery of diamonds in farmers being pushed off the land by larger,
1869, a white working class emerged.6 At first British-owned farms) and the English-speaking
individual whites did the searching, but, as with Labor Party (Van der Horst, 1965:117–8). The
the displacement of small farms by plantations, Revolt “showed the lengths to which white la-
they were displaced by consolidated, high-capital bour was prepared to go to defend its privileged
operations, and became employees of the latter position. From that time on, mine managements
(Doxey, 1961:18). It was this class together with have never directly challenged the colour-bar in
imported skilled miners from Cornwall (lured to the mining industry” (Van der Horst, 1965:118).
Africa by high wages) which fought the capital- The legislative history of much of South Af-
ists over the use of African labor. Africans were rica (and of the post-bellum Deep South) con-
cheaper because they came to the mines with a sists in attempts by higher priced white labor to
fixed income goal (e.g. the price of a rifle) and ward off undercutting by cheaper groups, and
did not view the mines as their main source of to entrench its exclusive control of certain jobs.7
livelihood. By contrast, European workers re- This interpretation of caste contrasts with the
mained in the mines and developed organiza- Marxist argument that the capitalist class pur-
tions to further their interests. posefully plays off one segment of the working
Clearly, it would have been to the advan- class against the other (e.g. Reich, 1971). Busi-
tage of businessmen, once they knew the skills ness, I would contend, rather than desiring to
involved, to train Africans to replace the white protect a segment of the working class supports a
miners at a fraction of the cost; but this did not liberal or laissez-faire ideology that would permit
happen. The mining companies accepted a labor all workers to compete freely in an open market.
aristocracy, not out of ethnic solidarity with the Such open competition would displace higher
white workers but: paid labor. Only under duress does business
yield to labor aristocracy, a point made in Deep
(as was to be the case throughout the later South, a book written when the depression had
history of mining) they had little or no choice caused the displacement of white tenant farmers
because of the collective strength of the white and industrial workers by blacks:
miners. . . . The pattern which was to emerge
was that of the Europeans showing every sign The economic interests of these groups [em-
of preparedness to use their collective strength ployers] would also demand that cheaper col-
to ensure their exclusive supremacy in the la- ored labor should be employed in the “white
bour market. Gradually the concept of trade collar” jobs in business offices, governmen-
unionism, and, for that matter, of socialism, tal offices, stores, and banks. In this field,
706 • Part VII: Race and Ethnicity

however, the interests of the employer group non-white workers. Consider the riots against
conflict not only with those of the lower eco- Italian strikebreakers in the coal fields of Penn-
nomic group of whites but also with those of sylvania in 1874 (Higham, 1965:47–8). In the
the more literate and aggressive middle group words of one writer: “Unions resented the ap-
of whites. A white store which employed col- parently inexhaustible cheap and relatively docile
ored clerks, for example, would be boycotted labor supply which was streaming from Europe
by both these groups. The taboo upon the obviously for the benefit of their employers”
employment of colored workers in such fields (Wittke, 1953:10).
is the result of the political and purchasing Even when no ethnic differences exist, split
power of the white middle and lower groups. labor markets may produce ethnic-like antago-
(Davis, et al., 1941:480) nism. Carey McWilliams (1945:82–3) describes
an instance:
In sum, exclusion and caste are similar re-
actions to a split labor market. They represent During the depression years, “Old Stock”—
victories for higher paid labor. The victory of ex- that is, white, Protestant, Anglo-Saxon
clusion is more complete in that cheaper labor is Americans, from Oklahoma, Arkansas, and
less available to business. For this reason I would Texas—were roundly denounced in Califor-
hypothesize that a higher paid group prefers ex- nia as “interlopers.” The same charges were
clusion to caste, even though exclusion means made against them that were made against the
they have to do the dirty work. Evidence for this Japanese: they were “dirty”; they had “enor-
comes from Australia where, in early attempts mous families”; they engaged in unfair com-
to import Asian labor, business tried to buy off petition; they threatened to “invade” the state
white labor’s opposition by offering to form and to “undermine” its institutions. During
them into a class of “mechanics” and foremen these turgid years (1930–1938) California
over the “coolies” (Yarwood, 1968:16, 42). The attempted to exclude, by various extra-legal
offer was heartily rejected in favor of exclusion. devices, those yeoman farmers just as it had
Apartheid in South Africa can be seen as an at- excluded the Chinese and Japanese. “Okies”
tempt to move from caste to the exclusion of the were “inferior” and “immoral.” There was
African work force. much family discord when Okie girl met Cal-
Most of our examples have contained a white ifornia boy, and vice versa. . . . The prejudice
capitalist class, a higher paid white labor group, against the Okies was obviously not “race”
and a cheaper, non-white labor group. Condi- prejudice; yet it functioned in much the same
tions in Europe and around the world, and not manner.
skin color, yield such models. White capitalists
would gladly dispense with and undercut their
white working-class brethren if they could, and Conclusion
have done so whenever they had the opportu- Obviously, this type of three-way conflict is not
nity. In the words of one agitator for excluding the only important factor in ethnic relations.
Chinese from the U.S. Pacific coast: “I have seen But it does help explain some puzzles, includ-
men . . . American born, who certainly would, ing, of course, the exclusion-caste anomaly. For
if I may use a strong expression, employ devils example, Philip Mason (1970:64) develops a
from Hell if the devils would work for 25 cents typology of race relations and finds that it re-
less than a white man” (cited in Daniels and Ki- lates to numerical proportions without being
tano, 1970:43). able to explain the dynamic behind this correla-
In addition, cases have occurred of white tion. Table 81.2 presents a modified version of
workers playing the role of cheap labor, and his chart. My theory can explain these relation-
facing the same kind of ethnic antagonism as ships. Paternalism arises in situations where the
 Bonacich—A Theory of Ethnic Antagonism • 707

Table 81.2. Numerical Proportion of Dominant to Subordinate Ethnic Groups

Category
Domination Paternalism Competition
South Africa Situations
Nigeria Britain
(1960) (1952) (1968)
1–4 1–2000 50–1
U. S. South Nyasaland U.S. North
(1960) (1966) (1960)
4–1 1–570 15–1
Rhodesia Tanganyika New Zealand
(1960) 1–450 13–1
1–16 1–450
Uganda 13–1
1–650
Uganda
Source: Adapted from Mason (1970:64).

cleavage between business and labor corresponds constructs like “authoritarianism,” this theory is
to an ethnic difference. A small business elite able to explain the apparent paradox.
rules a large group of workers who entered the In sum, in comparing those countries with
labor market at approximately the same price or the most ethnic antagonism with those having
strength. No split labor market existed, hence no the least, it is evident that the difference does
ethnic caste system arises. The higher proportion not lie in the fact that the former are Protestant
of the dominant ethnicity under “Domination” and the latter Catholic: Protestants are found
means that part of the dominant group must be in all three of Mason’s types, and Hawaii is a
working class. A labor element that shares eth- Protestant-dominated territory. It does not lie
nicity with people who have sufficient resources in whether the dominant or subordinate group
to become the business elite is generally likely moves: South Africa and the Deep South show
to come from a fairly wealthy country and have opposite patterns of movement. It is evident that
resources of its own. Such systems are likely to some of the most antagonistic territories have
develop split labor markets. Finally, competition been British colonies, but not all British colonies
has under it societies whose cheaper labor groups have had this attribute. The characteristic that
have not been a major threat because the indig- those British colonies and other societies high
enous population available as cheap labor has on ethnic antagonism share is that they all have
been small and/or exclusion has effectively kept a powerful white, or more generally higher paid,
business groups from importing cheap labor in working class.
large numbers.
This theory helps elucidate other observations. NOTES
One is the underlying similarity in the situation 1. I do not wish to enter the debate over the appli-
cability of the term “caste” to race relations (cf. Cox,
of blacks and women. Another is the history of
1948; Davis, et al., 1941). It is used here only for con-
political sympathy between California and the
venience and implies no particular theoretical bent.
South. And, a third is the conservatism of the 2. The term “exclusion” has not usually been ap-
American white working class, or what Daniels plied to immigrant quotas imposed on eastern and
and Kitano (1970:45) consider to be an “essential southern European immigrants; but such restrictions
paradox of American life: [that] movements for were, in effect, indistinguishable from the restric-
economic democracy have usually been violently tions placed on Japanese immigration.
opposed to a thorough-going ethnic democ- 3. This usage contrasts with that of van den Ber-
racy.” Without having to resort to psychological ghe (1967a:9–10) who reserves the term “ethnic” for
708 • Part VII: Race and Ethnicity

groups socially defined by cultural differences. In his Becker, Gary. 1957. The Economics of Discrimina-
definition, ethnicity is not necessarily inherited. I tion. Chicago: University of Chicago Press.
would contend that, while persons of mixed ancestry Berg, E. J. 1966. “Backward-sloping labor supply
may be problematic and are often assigned arbitrarily functions in dual economies—the Africa case.”
by the societies in which they reside, inheritance is Pp. 114–136 in Immanuel Wallerstein (ed.), So-
implied in the common application of the word. cial Change: The Colonial Situation. New York:
4. Sojourners often use their political resources Wiley.
and low price of labor to enter business for them- Blalock, H. M., Jr. 1967. Toward a Theory of
selves (a process which will be fully analyzed in Minority-Group Relations. New York: Wiley.
another paper). This does not remove the split in Blauner, Robert. 1969. “Internal colonialism and
the labor market, though it makes the conflict more ghetto revolt.” Social Problems 16 (Spring):
complex. 393–408.
5. It is, of course, no accident that color and Cox, Oliver C. 1948. Caste, Class and Race. New
resources have been historically related. Poverty York: Modern Reader.
among non-white nations has in part resulted from Daniels, Roger. 1966. The Politics of Prejudice.
European imperialism. Nevertheless, I would argue Gloucester, Massachusetts: Peter Smith.
that the critical factor in the development of ethnic Daniels, Roger, and Harry H. L. Kitano. 1970. Amer-
segmentation in a country is the meeting that occurs ican Racism. Englewood Cliffs: Prentice-Hall.
in the labor market of that country. The larger eco- Davis, Allison W., B. B. Gardner, and M. R. Gard-
nomic forces help determine the resources of enter- ner. 1941. Deep South. Chicago: University of
ing parties, but it is not such forces to which workers Chicago Press.
respond. Rather they react to the immediate conflicts Despres, Leo A. 1969. “Differential adaptations and
and threats in their daily lives. micro-cultural evolution in Guyana.” Southwest-
6. Such a split was not found in the early Cape ern Journal of Anthropology 25 (Spring):14–44.
Colony, where business was one ethnicity—white— Doxey, G. V. 1961. The Industrial Colour Bar in
and labor another—non-white. Actually in neither South Africa. Cape Town: Oxford University
case was the ethnic composition simple or homo- Press.
geneous; but the important fact is that, among the Ferguson-Davie, C. J. 1952. The Early History of
laborers, who included so-called Hottentots, and Indians in Natal. Johannesburg: South African
slaves from Madagascar, Mozambique, and the East Institute of Race Relations.
Indies (cf. van den Berghe, 1967b:14), no element Gillion, K. L. 1962. Fiji’s Indian Migrants. Mel-
was significantly more expensive. The early Cape is bourne: Oxford University Press.
thus structurally similar, in terms of the variables Harris, Marvin. 1964. Patterns of Race in the Ameri-
I consider important, to countries like Brazil and cas. New York: Walker.
Mexico. And it is also noted for its “softened” tone Heneman, H. G., and Dale Yoder. 1965. Labor Eco-
of race relations as reflected in such practices as nomics. Cincinnati: Southwestern.
intermarriage. Higham, John. 1965. Strangers in the Land. New
7. Ethnically based labor aristocracies are much York: Athenium.
less sensitive about cheap labor in any form than Hutchison, Emilie J. 1968. Women’s Wages. New
are systems that do not arrive at this resolution be- York: Ams Press.
cause they are protected from it. Thus, Sutherland Ichihashi, Yamato. 1932. Japanese in the United
and Cressey (1970:561–2) report that both the States. Stanford: Stanford University Press.
Deep South and South Africa continue to use var- Iwata, Masakazu. 1962. “The Japanese immigrants
ious forms of prison contract labor, in contrast to in California agriculture.” Agricultural History 36
the northern U.S. where the contract system was at- (January):25–37.
tacked by rising labor organizations as early as 1880. King, Cameron H., Jr. 1908. “Asiatic exclusion.”
International Socialist Review 8 (May):661–669.
REFERENCES Lenin, V. I. 1964. “Imperialism and the split in so-
Bailey, Thomas A. 1934. Theodore Roosevelt and the cialism.” Pp. 105–120 in Collected Works, Vol-
Japanese-American Crises. Stanford: Stanford Uni- ume 23, August 1916–March 1917. Moscow:
versity Press. Progress.
 Bonacich—A Theory of Ethnic Antagonism • 709

Lieberson, Stanley. 1961. “A societal theory of race Siu, Paul C. P. 1952. “The sojourner.” American
and ethnic relations.” American Sociological Re- Journal of Sociology 58 (July):34–44.
view 26 (December):902–910. Sutherland, Edwin H., and Donald R. Cressey.
Lind, Andrew W. 1968. An Island Community. New 1970. Criminology. Philadelphia: Lippincott.
York: Greenwood. van den Berghe, Pierre L. 1966. “Paternalistic ver-
Loewen, James W. 1971. The Mississippi Chinese. sus competitive race relations: an ideal-type ap-
Cambridge: Harvard University Press. proach.” Pp. 53–69 in Bernard E. Segal (ed.),
Mason, Philip. 1970. Patterns of Dominance. Lon- Racial and Ethnic Relations. New York: Crowell.
don: Oxford University Press. ———. 1967a. Race and Racism. New York: Wiley.
McWilliams, Carey. 1945. Prejudice: Japanese- ———. 1967b. South Africa: A Study in Conflict.
Americans. Boston: Little, Brown. Berkeley: University of California Press.
Millis, H. A. 1915. The Japanese Problem in the Van der Horst, Sheila T. 1965. “The effects of indus-
United States. New York: Macmillan. trialization on race relations in South Africa.” Pp.
Moles, I. N. 1968. “The Indian coolie labour issue.” 97–140 in Guy Hunter (ed.), Industrialization
Pp. 40–48 in A. T. Yarwood (ed.), Attitudes to and Race Relations. London: Oxford University
Non-European Immigration. Melbourne: Cassell Press.
Australia. Willard, Myra. 1967. History of the White Australia
Noel, Donald L. 1968. “A theory of the origin Policy to 1920. London: Melbourne University
of ethnic stratification.” Social Problems 16 Press.
(Fall):157–172. Williams, Eric. 1944. Capitalism and Slavery. Chapel
Reich, Michael. 1971. “The economics of racism.” Hill: University of North Carolina Press.
Pp. 107–113 in David M. Gordon (ed.), Prob- Wittke, Carl. 1953. “Immigration policy prior to
lems in Political Economy. Lexington, Massachu- World War I.” Pp. 1–10 in Benjamin M. Ziegler
setts: Heath. (ed.), Immigration: An American Dilemma. Bos-
Schermerhorn, R. A. 1970. Comparative Ethnic Re- ton: Heath.
lations. New York: Random House. Yarwood, A. T. 1964. Asian Immigration to Australia.
Shibutani, Tamotsu, and Kian M. Kwan. 1965. Eth- London: Cambridge University Press.
nic Stratification. New York: Macmillan. ——— (ed.). 1968. Attitudes to Non-European Im-
migration. Melbourne: Cassell Australia.
710 • Part VII: Race and Ethnicity

82 • Alejandro Portes and Robert D. Manning

The Immigrant Enclave

Theory and Empirical Examples

I. Introduction
society (Warner and Srole, 1945; Gordon, 1964;
The purpose of this chapter is to review existing Sowell, 1981). The focus on a “core” culture,
theories about the process of immigrant adap- the emphasis on consensus-building, and the
tation to a new society and to recapitulate the assumption of a basic patterned sequence of ad-
empirical findings that have led to an emerging aptation represent central elements of assimila-
perspective on the topic. This emerging view re- tion theory. From this perspective, the failure of
volves around the concepts of different modes of individual immigrants or entire ethnic groups to
structural incorporation and of the immigrant move up through the social hierarchies is linked
enclave as one of them. These concepts are set in either to their reluctance to shed traditional val-
explicit opposition to two previous viewpoints ues or to the resistance of the native majority to
on the adaptation process, generally identified accept them because of racial, religious, or other
as assimilation theory and the segmented labor shortcomings. Hence, successful adaptation
markets approach. depends, first of all, on the willingness of im-
The study of immigrant groups in the United migrants to relinquish a “backward” way of life
States has produced a copious historical and so- and, second, on their acquisition of characteris-
ciological literature, written mostly from the tics making them acceptable to the host society
assimilation perspective. Although the experi- (Eisenstadt, 1970). Throughout, the emphasis
ences of particular groups varied, the common is placed on the social psychological processes
theme of these writings is the unrelenting efforts of motivation, learning, and interaction and on
of immigrant minorities to surmount obstacles the cultural values and perceptions of the immi-
impeding their entry into the “mainstream” of grants themselves and those who surround them.
American society (Handlin, 1941, 1951; Wittke, The second general perspective takes issue
1952; Child, 1943; Vecoli, 1977). with this psychosocial and culturalist orientation
From this perspective, the adaptation pro- as well as with the assumption of a single basic
cess of particular immigrant groups followed a assimilation path. This alternative view begins
sequential path from initial economic hardship by noting that immigrants and their descendants
and discrimination to eventual socioeconomic do not necessarily “melt” into the mainstream
mobility arising from increasing knowledge of and that many groups seem not to want to do
American culture and acceptance by the host so, preferring instead to preserve their distinct

Alejandro Portes and Robert D. Manning. “The Immigrant Enclave: Theory and Empirical Examples,” in
Competitive Ethnic Relations, edited by Susan Olzak and Joanne Nagel, pp. 47–55, 61–68. Copyright © 1986
by Academic Press. Reprinted with permission from Elsevier. All rights of reproduction in any form reserved.
 Portes and Manning—The Immigrant Enclave • 711

ethnic identities (Greeley, 1971; Glazer and positions of advantage with members of the ma-
Moynihan, 1970). A number of writers have fo- jority (Nagel and Olzak, 1982).
cused on the resilience of these communities and A final variant focuses on the situation of
described their functions as sources of mutual contemporary immigrants to the United States.
support and collective political power (Suttles, Drawing on the dual labor market literature,
1968; Alba and Chamlin, 1983; Parenti, 1967). this approach views recent immigrants as the
Others have gone beyond descriptive accounts latest entrants into the lower tier of a segmented
and attempted to establish the causes of the per- labor market where women and other minorities
sistence of ethnicity. Without exception, these already predominate. Relative to the latter, im-
writers have identified the roots of the phenom- migrants possess the advantages of their lack of
enon in the economic sphere and, more specifi- experience in the new country, their legal vulner-
cally, in the labor-market roles that immigrants ability, and their greater initial motivation. All
have been called on to play. of these traits translate into higher productivity
Within this general perspective, several and lower labor costs for the firms that employ
specific theoretical approaches exist. The first them (Sassen-Koob, 1980). Jobs in the second-
focuses on the situation of the so-called unmelt- ary labor market are poorly paid, require few
able ethnics—blacks, Chicanos, and American skills, and offer limited mobility opportunities.
Indians—and finds the source of their plight in Hence, confinement of immigrants to this sector
a history of internal colonialism during which insures that those who do not return home are
these groups have been confined to specific areas relegated to a quasi-permanent status as disad-
and made to work under uniquely unfavorable vantaged and discriminated minorities (Piore,
conditions. In a sense, the role of colonized mi- 1975, 1979).
norities has been to bypass the free labor market, What these various structural theories have
yielding in the process distinct benefits both to in common is the view of resilient ethnic com-
direct employers of their labor and, indirectly, munities formed as the result of a consistently
to other members of the dominant racial group disadvantageous economic position and the con-
(Blauner, 1972; Geschwender, 1978). The con- sequent absence of a smooth path of assimilation.
tinuation of colonialist practices to our day These situations, ranging from slave labor to
explains, according to this view, the spatial iso- permanent confinement to the secondary labor
lation and occupational disadvantages of these market, are not altered easily. They have given
minorities (Barrera, 1980). rise, in time, either to hopeless communities of
A second approach attempts to explain the “unmeltable” ethnics or to militant minorities,
persistence of ethnic politics and ethnic mobili- conscious of a common identity and willing to
zation on the basis of the organization of subor- support a collective strategy of self-defense rather
dinate groups to combat a “cultural division of than rely on individual assimilation.
labor.” The latter confined members of specific These structural theories have provided an
minorities to a quasi-permanent situation of ex- effective critique of the excessively benign image
ploitation and social inferiority. Unlike the first of the adaptation process presented by earlier
view, this second approach does not envision writings. However, while undermining the for-
the persistence of ethnicity as a consequence of mer, the new structural perspective may have
continuing exploitation, but rather as a “reac- erred in the opposite direction. The basic hy-
tive formation” on the part of the minority to pothesis advanced in this chapter is that several
reaffirm its identity and its interests (Hechter, identifiable modes of labor-market incorpora-
1977; Despres, 1975). For this reason, ethnic tion exist and that not all of them relegate new-
mobilizations are often most common among comers to a permanent situation of exploitation
groups who have already abandoned the bottom and inferiority. Thus, while agreeing with the
of the social ladder and started to compete for basic thrust of structural theories, we propose
712 • Part VII: Race and Ethnicity

several modifications that are necessary for an Another sizable contingent of entrants whose
adequate understanding of the different types of occupational future is not easily characterized a
immigrant flows and their distinct processes of priori are political refugees. Large groups of ref-
adaptation. ugees, primarily from Communist-controlled
countries, have come to the United States, first
after the occupation of Eastern Europe by the
II. Modes of Incorporation Soviet Army, then after the advent of Fidel
In the four decades since the end of World War Castro to power in Cuba, and finally in the af-
II, immigration to the United States has experi- termath of the Vietnam War. Unlike purely “eco-
enced a vigorous surge reaching levels compara- nomic” immigrants, refugees have often received
ble only to those at the beginning of the century resettlement assistance from various governmen-
(National Research Council, 1985, chapter 2). tal agencies (Zolberg, 1983; Keely, 1981). All the
Even if one restricts attention to this movement, available evidence runs contrary to the notion
disregarding multiple other migrations elsewhere of a uniform entry of political refugees into
in the world, it is not the case that the inflow low-wage secondary occupations; on the con-
has been of a homogeneous character. Low-wage trary, there are indications of their employment
labor immigration itself has taken different in many different lines of work.
forms, including temporary contract flows, A third mode of incorporation has gained the
undocumented entries, and legal immigration. attention of a number of scholars in recent years.
More importantly, it is not the case that all im- It consists of small groups of immigrants who
migrants have been directed to the secondary are inserted or insert themselves as commercial
labor market. For example, since the promulga- intermediaries in a particular country or region.
tion of the Immigration Act of 1965, thousands These “middleman minorities” are distinct in
of professionals, technicians, and craftsmen have nationality, culture, and sometimes race from
come to the United States, availing themselves of both the superordinate and subordinate groups
the occupational preference categories of the law. to which they relate (Bonacich, 1973; Light,
This type of inflow, dubbed “brain drain” in the 1972). They can be used by dominant elites as
sending nations, encompasses today sizable con- a buffer to deflect mass frustration and also as
tingents of immigrants from such countries as an instrument to conduct commercial activities
India, South Korea, the Philippines, and Taiwan, in impoverished areas. Middlemen accept these
each an important contributor to U.S. annual risks in exchange for the opportunity to share
immigration. in the commercial and financial benefits gained
The characteristics of this type of migration through such instruments as taxation, higher re-
have been described in detail elsewhere (Portes, tail prices, and usury. Jews in feudal and early
1976, 1981). Two such traits deserve mention, modern Europe represent the classic instance of
however. First, occupationally skilled immi- a middleman minority. Other examples include
grants—including doctors, nurses, engineers, Indian merchants in East Africa, and Chinese
technicians, and craftsmen—generally enter the entrepreneurs in Southeast Asia and through-
“primary” labor market; they contribute to alle- out the Pacific Basin (Bonacich and Modell,
viate domestic shortages in specific occupations 1980, chapter 1). Contemporary examples in
and gain access, after a period of time, to the the United States include Jewish, Korean, and
mobility ladders available to native workers. Sec- other Oriental merchants in inner-city ghetto
ond, immigration of this type does not generally areas and Cubans in Puerto Rico (Kim, 1981;
give rise to spatially concentrated communities; Cobas, 1984).
instead, immigrants are dispersed throughout Primary labor immigration and middle-
many cities and regions, following different ca- man entrepreneurship represent two modes of
reer paths. incorporation that differ from the image of an
 Portes and Manning—The Immigrant Enclave • 713

homogeneous flow into low-wage employment. Jews and Japanese pursued patterns of economic
Political refugees, in turn, have followed a vari- adaptation that were quite similar both in con-
ety of paths, including both of the above as well tent and in their eventual consequences.
as insertion into an ethnic enclave economy.
The latter represents a fourth distinct mode.
Although frequently confused with middleman A. Jews in Manhattan
minorities, the emergence and structure of an The first major wave of Jewish immigration to
immigrant enclave possess distinct characteris- the United States consisted of approximately
tics. The latter have significant theoretical and 50,000 newcomers of German origin, arriving
practical implications, for they set apart groups between 1840 and 1870. These immigrants
adopting this entry mode from those following went primarily into commerce and achieved, in
alternative paths. We turn now to several exam- the course of a few decades, remarkable success.
ples of immigrant enclaves to clarify their inter- By 1900, the average income of German-Jewish
nal dynamics and causes of their emergence. immigrants surpassed that of the American pop-
ulation (Rischin, 1962). Many individuals who
started as street peddlers and small merchants
III. Immigrant Enclaves had become, by that time, heads of major indus-
Immigration to the United States before World trial, retail, and financial enterprises.
War I was, overwhelmingly, an unskilled labor The second wave of Jewish immigration ex-
movement. Impoverished peasants from south- hibited quite different characteristics. Between
ern Italy, Poland, and the eastern reaches of the 1870 and 1914, over two million Jews left the
Austro-Hungarian Empire settled in dilapidated Pale of Settlement and other Russian-dominated
and crowded areas, often immediately adjacent regions, escaping Czarist persecution. Major po-
to their points of debarcation, and took any me- groms occurred before and during this exodus
nial jobs available. From these harsh beginnings, (Dinnerstein, 1977). Thus, unlike most immi-
immigrants commenced a slow and often painful grants of the period, the migration of Russian
process of acculturation and economic mobility. and Eastern Europe Jews was politically moti-
Theirs was the saga captured by innumerable vated and their move was much more perma-
subsequent volumes written from both the as- nent. In contrast to German Jews, who were
similation and the structural perspectives. relatively well educated, the Yiddish-speaking
Two sizable immigrant groups did not follow newcomers came, for the most part, from mod-
this pattern, however. Their most apparent char- est origins and had only a rudimentary educa-
acteristic was the economic success of the first tion. Although they viewed the new Russian
generation, even in the absence of extensive ac- wave with great apprehension, German Jews
culturation. On the contrary, both groups strug- promptly realized that their future as an ethnic
gled fiercely to preserve their cultural identity minority depended on the successful integration
and internal solidarity. Their approach to ad- of the newcomers (Rischin, 1962). Charitable
aptation thus directly contradicted subsequent societies were established to provide food, shel-
assimilation predictions concerning the causal ter, and other necessities, and private schools
priority of acculturation to economic mobility. were set up to teach immigrants English, civics,
Economic success and “clannishness” also earned and the customs of the new country (Howe and
for each minority the hostility of the surround- Libo, 1979).
ing population. These two immigrant groups Aside from its size and rapidity of arrival,
did not have a language, religion, or even race turn-of-the-century Jewish immigration had
in common and they never overlapped in signif- two other distinct characteristics. First was
icant numbers in any part of the United States. its strong propensity toward commerce and
Yet, arriving at opposite ends of the continent, self-employment in general in preference to wage
714 • Part VII: Race and Ethnicity

labor: as German Jews before them, many Rus- conventional assimilation path and achieve
sian immigrants moved directly into street ped- significant economic mobility in the course of
dling and other commercial activities of the most the first generation, well ahead of complete ac-
modest sort. Second was its concentration into a culturation. Subsequent generations also pur-
single, densely populated urban area—the lower sued this path, but the resources accumulated
East Side of Manhattan. Within this area, those through early immigrant entrepreneurship were
who did not become storekeepers and peddlers dedicated primarily to further the education of
from the start found employment in factories children and their entry into the professions. It
owned by German Jews, learning the necessary was at this point that outside hostility became
rudiments for future self-employment (Sowell, most patent, as one university after another es-
1981, chapter 4). tablished quotas to prevent the onrush of Jewish
The economic activities of this population students. The last of these quotas did not come
created, in the course of two decades, a dense to an end until after World War II (Dinnerstein,
network of industrial, commercial, and finan- 1977).
cial enterprises. Close physical proximity facil- Despite these and other obstacles, the move-
itated exchanges of information and access to ment of Jews into higher education continued.
credit and raw materials. Characteristic of this Building on the economic success of the first
emerging Jewish enclave is that production and generation, subsequent ones achieved levels of
marketing of goods was not restricted to the education, occupation, and income that signifi-
ethnic community, but went well beyond it into cantly exceed the national average (Featherman,
the general economy. Jews entered the printing, 1971; Sowell, 1981, chapter 4). The original
metal, and building trades; they became increas- enclave is now only a memory, but it provided
ingly prominent in jewelry and cigar-making; in its time the necessary platform for further-
above all, the garment industry became the pri- ing the rapid social and economic mobility of
mary domain of Jewish entrepreneurship, with the minority. Jews did enter the mainstream of
hundreds of firms of all sizes engaged in the trade American society, but they did not do so start-
(Rischin, 1962; Howe and Libo, 1979). ing uniformly at the bottom, as most immigrant
The economic success of many of these ven- groups had done; instead, they translated re-
tures did not require and did not entail rapid sources made available by early ethnic entrepre-
acculturation. Immigrants learned English and neurship into rapid access to positions of social
those instrumental aspects of the new culture re- prestige and economic advantage.
quired for economic advancement. For the rest,
they preferred to remain with their own and
maintained, for the most part, close adherence B. Japanese on the West Coast
to their original religion, language, and values The specific features of Japanese immigration
(Wirth, 1956; Howe, 1976). Jewish enclave differ significantly from the movement of Eu-
capitalism depended, for its emergence and ad- ropean Jews, but their subsequent adaptation
vancement, precisely on those resources made and mobility patterns are similar. Beginning
available by a solidaristic ethnic community: in 1890 and ending with the enactment of
protected access to labor and markets, informal the Gentlemen’s Agreement of 1908, approx-
sources of credit, and business information. It imately 150,000 Japanese men immigrated to
was through these resources that upstart immi- the West Coast. They were followed primarily
grant enterprises could survive and eventually by their spouses until the Immigration Act of
compete effectively with better-established firms 1924 banned any further Asiatic immigration.
in the general economy. Although nearly 300,000 Japanese immigrants
The emergence of a Jewish enclave in East are documented in this period (Daniels, 1977),
Manhattan helped this group bypass the less than half of this total remained in the United
 Portes and Manning—The Immigrant Enclave • 715

States (Petersen, 1971). This is due, in contrast These proscriptions, which barred most of
to the case of the Jews, to the sojourner character the Japanese from the lands, accelerated their
of Japanese immigrants: the intention of many entry into urban enterprise. In 1909, Japanese
was to accumulate sufficient capital for purchas- entrepreneurs owned almost 3000 small shops in
ing farm land or settling debts in Japan. Hence several Western cities. Forty percent of Japanese
this population movement included commercial men in Los Angeles were self-employed. They
and other members of the Japanese middle class operated businesses such as dry-cleaning estab-
who, not incidentally, were explicitly sponsored lishments, fisheries, lunch counters, and produce
by their national government. stands that marketed the production of Japanese
The residential patterns of Japanese immi- farms (Light, 1972).
grants were not as concentrated as those of Jews The ability of the first-generation Issei to es-
in Manhattan, but they were geographically cape the status of stoop labor in agriculture was
clustered. Almost two-thirds of the 111,010 based on the social cohesion of their commu-
Japanese reported in the U.S. Census of 1920 nity. Rotating credit associations offered scarce
lived in California. Further, one-third of Cali- venture capital, while mutual-aid organizations
fornia’s Japanese residents lived in Los Angeles provided assistance in operating farms and urban
County in 1940, while another one-third lived businesses. Light (1972) reports that capitaliza-
in six nearby counties (Daniels, 1977). However, tions as high as $100,000 were financed through
it was not the residential segregation of Japanese ethnic credit networks. Economic success was
immigrants but rather their occupational pat- again accompanied by limited instrumental
terns that eventually mobilized the hostility of acculturation and by careful preservation of na-
the local population. tional identity and values. It was the availabil-
Japanese immigrants were initially welcomed ity of investment capital, cooperative business
and recruited as a form of cheap agricultural associations, and marketing practices (forward
labor. Their reputation as thrifty and diligent and backward economic linkages) within the
workers made them preferable to other labor ethnic enclave that enabled Japanese entrepre-
sources. Nativist hostilities crystallized, however, neurs to expand beyond its boundaries and com-
when Japanese immigrants shifted from wage pete effectively in the general economy. This is
labor to independent ownership and small-scale illustrated by the production and marketing of
farming. This action not only reduced the sup- fresh produce. In 1920, the value of Japanese
ply of laborers but it also increased competition crops was about 10% of the total for California,
for domestic growers in the fresh-produce mar- when the Japanese comprised less than 1% of
ket. In 1900, only about 40 Japanese farmers in the state’s population; many retail outlets traded
the entire United States leased or owned a total exclusively with a non-Japanese clientele (Light,
of 5000 acres of farmland. By 1909, the number 1972; Petersen, 1971).
of Japanese farmers had risen to 6000 and their During the early 1940s, the Japanese ethnic
collective holdings exceeded 210,000 acres (Pe- economy was seriously disrupted but not elim-
tersen, 1971). Faced with such “unfair” compe- inated by the property confiscations and camp
tition, California growers turned to the political internments accompanying World War II. After
means at their disposal. In 1913, the state leg- the war, economic prosperity and other factors
islature passed the first Alien Land Law, which combined to reduce local hostility toward the
restricted land ownership by foreigners. This leg- Japanese. Older Issei and many of their chil-
islation did not prove sufficient, however, and, dren returned to small business, while other
in subsequent years, the ever-accommodating second-generation Nisei, like their Jewish prede-
legislature passed a series of acts closing other cessors, pursued higher education and entered
legal loopholes to Japanese farming (Petersen, the white-collar occupations en masse. This
1971). mobility path was completed by the third or
716 • Part VII: Race and Ethnicity

Sansei generation, with 88% of their members Enclave businesses typically start small and
attending college. Other third-generation Jap- cater exclusively to an ethnic clientele. Their
anese have continued, however, the entrepre- expansion and entry into the broader market
neurial tradition of their parents (Bonacich and requires, as seen above, an effective mobilization
Modell, 1980). Like Jews before them, Japanese of community resources. The social mechanism
Americans have made use of the resources made at work here seems to be a strong sense of rec-
available by early immigrant entrepreneurship iprocity supported by collective solidarity that
to enter the mainstream of society in positions transcends the purely contractual character of
of relative advantage. The mean educational and business transactions. For example, receipt of a
occupational attainment of the group’s 600,000 loan from a rotating credit association entails the
members surpasses at present all other ethnic and duty of continuing to make contributions so that
native groups, while its average family income is others can have access to the same source of capi-
exceeded among American ethnic groups only tal. Although, in principle, it would make sense
by the Jews (Sowell, 1981).1 . . . for the individual to withdraw once his loan is
received, such action would cut him off from the
very sources of community support on which his
IV. Conclusion: A Typology of the future business success depends (Light, 1972).
Process of Incorporation Similarly, relations between enclave employ-
We can now attempt a summary description of ers and employees generally transcend a con-
the characteristics of immigrant enclaves and tractual wage bond. It is understood by both
how they differ from other paths. The emergence parties that the wage paid is inferior to the value
of an ethnic enclave economy has three prerequi- of labor contributed. This is willingly accepted
sites: first, the presence of a substantial number by many immigrant workers because the wage
of immigrants, with business experience acquired is only one form of compensation. Use of their
in the sending country; second, the availability labor represents often the key advantage making
of sources of capital; and third, the availability poorly capitalized enclave firms competitive. In
of sources of labor. The latter two conditions reciprocity, employers are expected to respond
are not too difficult to meet. The requisite labor to emergency needs of their workers and to pro-
can usually be drawn from family members and, mote their advancement through such means
more commonly, from recent arrivals. Surpris- as on-the-job training, advancement to super-
ingly perhaps, capital is not a major impedi- visory positions, and aid when they move into
ment either since the sums initially required are self-employment. These opportunities represent
usually small. When immigrants did not bring the other part of the “wage” received by enclave
them from abroad, they could be accumulated workers. The informal mobility ladders thus cre-
through individual savings or pooled resources ated are, of course, absent in the secondary labor
in the community. It is the first condition that market where there is no primary bond between
appears critical. The presence of a number of im- owners and workers or no common ethnic com-
migrants skilled in what Franklin Frazier (1949) munity to enforce the norm of reciprocity.
called the art of “buying and selling” is common Paternalistic labor relations and strong com-
[not only to the Jewish and Japanese cases re- munity solidarity are also characteristic of mid-
viewed above but also to contemporary enclave dleman minorities. Although both modes of
economies among Koreans and Cubans]. Such incorporation are similar and are thus frequently
an entrepreneurial-commercial class among early confused, there are three major structural differ-
immigrant cohorts can usually overcome other ences between them. First, immigrant enclaves
obstacles; conversely, its absence within an im- are not exclusively commercial. Unlike mid-
migrant community will confine the community dleman minorities, whose economic role is to
to wage employment even if sufficient resources mediate commercial and financial transactions
of capital and labor are available. between elites and masses, enclave firms include
 Portes and Manning—The Immigrant Enclave • 717

in addition a sizable productive sector. The latter stages. This is so for three reasons: first, the need
may comprise agriculture, light manufacturing, for proximity to the ethnic market which they
and construction enterprises; their production, initially serve; second, proximity to each other
marketed often by coethnic intermediaries, is which facilitates exchange of information, access
directed toward the general economy and not to credit, and other supportive activities; third,
exclusively to the immigrant community. proximity to ethnic labor supplies on which they
Second, relationships between enclave busi- crucially depend. Unlike the Jewish, Korean, or
nesses and established native ones are problem- Cuban cases, the Japanese enclave economy does
atic. Middleman groups tend to occupy positions partially depart from the pattern of high physical
complementary and subordinate to the local concentration. This can be attributed to the po-
owning class; they fill economic niches either litical persecution to which this group was sub-
disdained or feared by the latter. Unlike them, jected. Originally, Japanese concentration was a
enclave enterprises often enter in direct compe- rural phenomenon based on small farms linked
tition with existing domestic firms. There is no together by informal bonds and cooperative as-
evidence, for example, that domestic elites delib- sociations. Forced removal of this minority from
erately established or supported the emergence the land compelled their entry into urban busi-
of the Jewish, Japanese, Korean, or Cuban busi- nesses and their partial dispersal into multiple
ness communities as means to further their own activities.
economic interests. There is every indication, on Physical concentration of enclaves under-
the other hand, that this mode of incorporation lies their final characteristic. Once an enclave
was largely self-created by the immigrants, often economy has fully developed, it is possible for a
in opposition to powerful domestic interests. Al- newcomer to live his life entirely within the con-
though it is true that enclave entrepreneurs have fines of the community. Work, education, and
been frequently employed as subcontractors by access to health care, recreation, and a variety of
outside firms in such activities as garment and other services can be found without leaving the
construction (Bonacich, 1978), it is incorrect to bounds of the ethnic economy. This institutional
characterize this role as the exclusive or domi- completeness is what enables new immigrants to
nant one among these enterprises. move ahead economically, despite very limited
Third, the enclave is concentrated and spa- knowledge of the host culture and language.
tially identifiable. By the very nature of their Supporting empirical evidence comes from
activities, middleman minorities must often be studies showing low levels of English knowledge
dispersed among the mass of the population. among enclave minorities and the absence of a
Although the immigrants may live in certain net effect of knowledge of English on their aver-
limited areas, their businesses require proximity age income levels (Light, 1980; Portes and Bach,
to their mass clientele and a measure of physical 1985).
dispersion within it. It is true that middleman Table 82.1 summarizes this discussion by pre-
activities such as money-lending have been as- senting the different modes of incorporation and
sociated in several historical instances with cer- their principal characteristics. Two caveats are
tain streets and neighborhoods, but this is not necessary. First, this typology is not exhaustive,
a necessary or typical pattern. Street peddling since other forms of adaptation have existed and
and other forms of petty commerce require mer- will undoubtedly emerge in the future. Second,
chants to go into the areas where demand exists political refugees are not included, since this
and avoid excessive concentration of the goods entry label does not necessarily entail a unique
and services they offer. This is the typical pat- adaptation path. Instead, refugees can select
tern found today among middleman minorities or be channelled in many different directions,
in American cities (Cobas, 1984; Kim, 1981). including self-employment, access to primary
Enclave businesses, on the other hand, are labor markets, or confinement to secondary sec-
spatially concentrated, especially in their early tor occupations.
 718

Table 82.1. Typology of Modes of Incorporation

 Portes and Manning—The Immigrant Enclave • 719

Having discussed the characteristics of en- professionals, who tend to “disappear,” in a cul-
claves and middleman minorities, a final word tural sense, soon after their arrival (Stevens et al.,
must be said about the third alternative to em- 1978; Cardona and Cruz, 1980).
ployment in the lower tier of a dual labor mar- Awareness of patterned differences among
ket. As a mode of incorporation, primary sector immigrant groups in their forms of entry and
immigration also has distinct advantages, al- labor market incorporation represents a signifi-
though they are of a different order from those cant advance, in our view, from earlier undiffer-
pursued by “entrepreneurial” minorities. Disper- entiated descriptions of the adaptation process.
sal throughout the receiving country and career This typology is, however, a provisional effort.
mobility based on standard promotion criteria Just as detailed research on the condition of par-
make it imperative for immigrants in this mode ticular minorities modified or replaced earlier
to become fluent in the new language and culture broad generalizations, the propositions advanced
(Stevens, Goodman, and Mick, 1978). Without here will require revision. New groups arriving in
a supporting ethnic community, the second gen- the United States at present and a revived inter-
eration also becomes thoroughly steeped in the est in immigration should provide the required
ways of the host society. Primary sector immigra- incentive for empirical studies and theoretical
tion thus tends to lead to very rapid social and advances in the future.
cultural integration. It represents the path that
approximates most closely the predictions of as- NOTES
similation theory with regard to (1) the necessity 1. The original article from which this chapter
was drawn includes a further discussion of contem-
of acculturation for social and economic prog-
porary Korean and Cuban enclaves.—ED.
ress and (2) the subsequent rewards received by
immigrants and their descendants for shedding
REFERENCES
their ethnic identities. Alba, Richard D., and Chamlin, Mitchell B. (1983).
Clearly, however, this mode of incorporation Ethnic identification among whites. American
is open only to a minority of immigrant groups. Sociological Review 48:240–247.
In addition, acculturation of professionals and Barrera, Mario (1980). Race and class in the South-
other primary sector immigrants is qualita- west: A theory of racial inequality. Notre Dame,
tively different from that undergone by others. Indiana: Notre Dame University Press.
Regardless of their differences, immigrants in Blauner, Robert (1972). Racial oppression in America.
New York: Harper and Row.
other modes tend to learn the new language and
Bonacich, Edna (1973). A theory of middleman
culture with a heavy “local” content. Although
minorities. American Sociological Review 38
acculturation may be slow, especially in the case (October):583–594.
of enclave groups, it carries with it elements ——— (1978). U.S. capitalism and Korean im-
unique to the surrounding community—its migrant small business. Riverside, California:
language inflections, particular traditions, and Department of Sociology, University of Califor-
loyalties (Greeley, 1971; Suttles, 1968). On the nia–Riverside, mimeographed.
contrary, acculturation of primary sector immi- Bonacich, Edna, and Modell, John (1980). The eco-
grants is of a more cosmopolitan sort. Because nomic basis of ethnic solidarity: Small business in
the Japanese-American community. Berkeley: Uni-
career requirements often entail physical mo-
versity of California Press.
bility, the new language and culture are learned
Cardona, Ramiro C., and Cruz, Carmen I. (1980).
more rapidly and more generally, without strong El exodo de Colombianos. Bogota: Ediciones
attachments to a particular community. Thus, Tercer Mundo.
while minorities entering menial labor, enclave, Child, Irving L. (1943). Italian or American? The
or middleman enterprise in the United States second generation in conflict. New Haven, Con-
have eventually become identified with a certain necticut: Yale University Press.
city or region, the same is not true for immigrant Cobas, Jose (1984). Participation in the ethnic econ-
omy, ethnic solidarity and ambivalence toward the
720 • Part VII: Race and Ethnicity

host society: The case of Cuban emigres in Puerto Keely, Charles B. (1981). Global refugee policy: The
Rico. Presented at the American Sociological As- case for a development-oriented strategy. New York:
sociation Meeting, San Antonio, Texas: August. The Population Council.
Daniels, Roger (1977). The Japanese-American ex- Kim, Illsoo (1981). New urban immigrants, the Ko-
perience: 1890–1940. In Uncertain Americans (L. rean community in New York. Princeton, New
Dinnerstein and F. C. Jaher, eds.), pp. 250–267. Jersey: Princeton University Press.
New York: Oxford University Press. Light, H. Ivan (1972). Ethnic enterprise in America:
Despres, Leo (1975). Toward a theory of ethnic phe- Business and welfare among Chinese, Japanese, and
nomena. In Ethnicity and resource competition (Leo Blacks. Berkeley: University of California Press.
Despres, ed.), pp. 209–212. The Hague: Mouton. ——— (1980). Asian enterprise in America: Chi-
Dinnerstein, Leonard (1977). The East European nese, Japanese, and Koreans in small business.
Jewish migration. In Uncertain Americans (L. In Self-help in urban America (Scott Cummings,
Dinnerstein and F. C. Jaher, eds.), pp. 216–231. ed.), pp. 33–57. New York: Kennikat Press.
New York: Oxford University Press. Nagel, Joane, and Olzak, Susan (1982). Ethnic
Eisenstadt, S. N. (1970). The process of absorbing mobilization in new and old states: An exten-
new immigrants in Israel. In Integration and de- sion of the competition model. Social Problems
velopment in Israel (S. N. Eisenstadt, RivKah Bar 30:127–143.
Yosef, and Chaim Adler, eds.), pp. 341–367. Je- National Research Council (1985). Immigration
rusalem: Israel University Press. statistics: A story of neglect. Report of the Panel
Featherman, David L. (1971). The socioeconomic on Immigration Statistics. Washington, D.C.:
achievement of white religio-ethnic sub-groups: National Academy of Sciences.
Social and psychological explanations. American Parenti, Michael (1967). Ethnic politics and the per-
Sociological Review 36 (April): 207–222. sistence of ethnic identification. American Politi-
Frazier, E. Franklin (1949). The Negro in the United cal Science Review 61:717–726.
States. New York: Macmillan. Petersen, William (1971). Japanese Americans, oppres-
Geschwender, James A. (1978). Racial stratification sion and success. New York: Random House.
in America. Dubuque, Iowa: William C. Brown. Piore, Michael J. (1975). Notes for a theory of labor
Glazer, Nathan, and Moynihan, Daniel P. (1970). market stratification. In Labor market segmenta-
Beyond the melting pot: The Negroes, Puerto Ri- tion (Richard C. Edwards, Michael Reich, and
cans, Jews, Italians and Irish of New York City. David M. Gordon, eds.), pp. 125–171. Lexing-
Cambridge, Massachusetts: M.I.T. Press. ton, Massachusetts: Heath.
Gordon, Milton M. (1964). Assimilation in Amer- ——— (1979). Birds of passage, migrant labor and
ican life: The role of race, religion, and national industrial societies. New York: Cambridge Uni-
origins. New York: Oxford University Press. versity Press.
Greeley, Andrew (1971). Why can’t they be like us? Portes, Alejandro (1976). Determinants of the
America’s white ethnic groups. New York: Dutton. brain drain. International Migration Review 10
Handlin, Oscar (1941). Boston’s immigrants: A study (Winter):489–508.
of acculturation. Cambridge: Harvard University ——— (1981). Modes of structural incorporation
Press. and theories of labor immigration. In Global
——— (1951). The uprooted: The epic story of the Trends in Migration, Theory and Research on Inter-
great migrations that made the American people. national Population Movements (Mary M. Kritz,
Boston: Little, Brown. Charles B. Keely, and Silvano M. Tomasi, eds.),
Hechter, Michael (1977). Internal colonialism, the pp. 279–297. New York: Center for Migration
Celtic fringe in British national development, Studies.
1536–1966. Berkeley: University of California Portes, Alejandro, and Bach, Robert L. (1985). Latin
Press. journey, Cuban and Mexican immigrants in the
Howe, Irving (1976). World of our fathers. New York: United States. Berkeley: University of California
Harcourt, Brace, Jovanovich. Press.
Howe, Irving, and Libo, Kenneth (1979). How we Rischin, Moses (1962). The promised city, New York
lived, a documentary history of immigrant Jews in Jews 1870–1914. Cambridge, Massachusetts:
America. New York: Richard March. Harvard University Press.
 Alba and Nee—Assimilation Theory for an Era of Unprecedented Diversity • 721

Sassen-Koob, Saskia (1980). Immigrant and mi- Warner, W. Lloyd, and Srole, Leo (1945). The social
nority workers in the organization of the labor systems of American ethnic groups. New Haven:
process. Journal of Ethnic Studies (1/Spring):1–34. Yale University Press.
Sowell, Thomas (1981). Ethnic America: A history. Wirth, Louis (1956). The ghetto. Chicago: University
New York: Basic Books. of Chicago Press.
Stevens, Rosemary, Goodman, Louis W., and Mick, Wittke, Carl (1952). Refugees of revolution: The Ger-
Stephen (1978). The alien doctors, foreign medical man Forty-eighters in America. Philadelphia: Uni-
graduates in American hospitals. New York: Wiley. versity of Pennsylvania Press.
Suttles, Gerald D. (1968). The social order of the Zolberg, Aristide (1983). Contemporary transna-
slum: Ethnicity and territory in the inner city. Chi- tional migrations in historical perspective: Pat-
cago: University of Chicago Press. terns and dilemmas. In U.S. immigration and
Vecoli, Rudolph (1977). The Italian Americans. In refugee policy (Mary M. Kritz, ed.), pp. 15–51.
Uncertain Americans (L. Dinnerstein and F. C. Lexington, Massachusetts: Heath.
Jaher, eds.), pp. 201–215. New York: Oxford
University Press.

83 • Richard Alba and Victor Nee

Assimilation Theory for an Era of

Unprecedented Diversity

Introduction and Judaism of the status of charter religions, as

Assimilation remains the major path into the opposed to incorporation in the form of conver-
mainstream of American life for immigrants and sion to Protestantism, the previous mainstream
their descendants. The process has been well doc- religion (Alba 2009).
umented, and there is a strong case to be made After a long hiatus in mass immigration to
that assimilation is a predictable, large-scale out- the United States, the passage of the Immigra-
come of mass immigration to the United States. tion Act of 1965 opened legal immigration to all
Despite episodes of pressure for Americanization nationalities. The result was a swift and dramatic
from political actors, assimilation has been a change in the composition of the immigrant
bottom-up process of cultural and social change. stream. The overwhelming predominance of
The changes to immigrant groups are frequently nonwhite immigrants from Latin America, the
accompanied by shifts in the mainstream to Caribbean, and Asia has sparked a debate in the
which they are assimilating: Emblematic of the social sciences over whether assimilation might
latter is the eventual attainment by Catholicism be unattainable for many of the new immigrant

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Richard D. Alba and Victor Nee, Remaking the American Mainstream: Assimilation
and Contemporary Immigration (2005), published by Harvard University Press. The article printed here was
originally prepared by Richard Alba and Victor Nee for this fourth edition of Social Stratification: Class, Race,
and Gender in Sociological Perspective, edited by David B. Grusky. Copyright © 2014 by Westview Press.
722 • Part VII: Race and Ethnicity

minorities (Huntington 2004; cf. Glazer 1997). were the main source for populating settler so-
Portes and Zhou (1993) have argued that assim- cieties like the United States. Although immi-
ilation is being transformed into a deeply seg- grants come with their own native languages
mented process of incorporation whereby the and cultural traits, assimilation, perceived as an
children of poor nonwhite immigrants are at risk intergenerational process of endogenous cul-
of “downward assimilation” into the racialized tural and social change, gives rise to a new type
underclass of the inner city. In segmented as- of nation-state not based on common ethnicity,
similation theory, immigrant minorities may be but constructed on the institutional elements
better served by relying on their own ethnic and of common cultural beliefs, norms, rules, and
cultural resources within immigrant enclaves. organizations. Park (1930) posited a concep-
Immigration from Latin America, the Ca- tion of composite culture evolving through the
ribbean, and Asia since 1965 has now sur- interpenetration of diverse social practices and
passed the scale of the late nineteenth- and early cultural beliefs of immigrants and their children.
twentieth-century immigration of Europeans. As The more flexible and open-ended speci-
a consequence, demographic transformation and fication of assimilation of the Chicago School
growing racial and ethnic diversity are in prog- largely receded into the background in the mid-
ress in many parts of the United States. These century “straight-line” assimilation models of
dramatic demographic shifts have profound Warner and Srole (1945) and Gordon (1964).
effects on neighborhoods, public schools, and The straight-line assimilation canon was heav-
workplaces, and call attention to the need to ily influenced by structural-functionalism, the
rethink assimilation (Alba and Nee 2003; Bean paradigm that displaced the Chicago School in
and Stevens 2003; Brubaker 2001). In this chap- post–World War II American sociology. This
ter our aim is to lay out a viable conception of assimilation model conceived of society as a
assimilation and identify mechanisms that en- largely homogenous social system integrated
able it in the contemporary era of immigration around core values and norms. Consistent with
to the United States. structural-functionalism, the assimilation model
posited an abstract view of Anglo-American
middle-class culture and society as the endpoint
A Conceptual Foundation for of a one-way process of assimilation (Gordon
Assimilation Theory 1964).
The origins of assimilation theory lie in the In the early twenty-first century Alba and
last great wave of immigration. The theory de- Nee (2003) revised the concept of assimilation
veloped initially from observations by sociol- to an American society far more diverse racially
ogists at the University of Chicago about the and ethnically than during the earlier era of im-
migrants drawn to the industrializing city in migration. In place of a teleological view of an
the early twentieth century. Responding to the inevitable outcome, Alba and Nee argue that as-
transformative changes and social problems vis- similation needs to be explained, not assumed,
ible around them, the founders of the Chicago as a process of declining “ethnic distinction and
School elaborated the assimilation paradigm. its corollary cultural and social differences.” In
This framework has influenced all subsequent their view the essence of assimilation is the de-
sociological research on the incorporation of im- cline of ethno-racial origins in determining indi-
migrants and their descendants (Gordon 1964; vidual life chances and daily experiences. Defined
Portes and Rumbaut 2001; Alba and Nee 2003; in this manner, assimilation does not require the
Kasinitz et al. 2008). loss of identity or the disappearance of ethnic
Assimilation was closely linked to the project cultures. Compared to the canonical conception
of nation building in the writings of Chicago of assimilation promulgated by Gordon (1964),
School sociologists. In their view, immigrants this definition does not equate the endpoint of
 Alba and Nee—Assimilation Theory for an Era of Unprecedented Diversity • 723

assimilation with merging through intermarriage that individuals make in their everyday lives and
with white Americans, something less feasible as that shapes their actions and mental orientations
the white population declines in relative size. toward others (Barth 1969; Lamont and Molnár
According to the new definition, members of an 2002); it is typically embedded in a variety of
ethnoracial group can assimilate in large num- social and cultural differences between groups
bers even when the group maintains distinctive that give an ethnoracial boundary concrete sig-
neighborhoods and institutions. Moreover, eth- nificance (so that members of one group think,
noracial identities may experience periodic resur- “they are not like us because . . . ”). The bound-
gence as blending processes yield to segregating ary concept is useful because it does not overlook
ones (Nee, Sanders, and Sernau 1994). Alba the majority group. In research on incorporation,
and Nee argued that the new definition allows the focus typically is almost entirely on the im-
assimilation theory to jettison the tacit ethno- migrant minority and the changes it undergoes;
centrism of the past (Warner and Srole 1945; the majority group, when included, appears as a
Gordon 1964). relatively passive player, even though we know
According to Alba and Nee, the mainstream that in many situations it actively defends its
encompasses a core set of interrelated institu- privileges, greeting minorities with prejudice and
tional structures and organizations. For example, discrimination. Attention to social boundaries
the modern research university is an interrelated allows us to specify some circumstances under
set of institutional elements comprising cultural which this defense may weaken.
beliefs, social norms, and formal rules governing
faculty and student conduct. The same formal
rules of competition and cooperation apply to Mechanisms of Assimilation
all faculty and students regardless of their ethnic- Assimilation is best viewed as a process of cumu-
ity. The core institutions of the mainstream have lative causation driven by a repertoire of mech-
become cumulatively more inclusive since World anisms operating at the individual, boundary,
War II and the social movements and legislative and institutional levels. That is, no single causal
changes of the civil rights era (Dobbin 2009). mechanism explains the mode of immigrants’
The mainstream to which the immigrants and accommodation to their host society. In com-
their descendants assimilate also includes the so- bination, and cumulatively, they shape the tra-
cial and cultural settings in which the members jectories of adaptation of immigrants and their
of the dominant group (i.e., native whites) are children. These mechanisms are general to social
“at home.” Assimilation means that the entry behavior and fall broadly within two groups: the
of immigrants and their descendants into these proximate causes that operate at the individual
mainstream settings becomes unproblematic. level, and the distal, often deeper causes that are
(“Mainstream” appears here in the singular, but embedded in large structures such as the institu-
we recognize that the plural can be seen as more tional arrangements of the state, firm, and labor
appropriate because of the heterogeneity of the market.
vast US mainstream.) Although the mainstream In contrast to the canonical assimilation liter-
may be identified by the presence of whites, it ature, contemporary theory does not assume as-
is not limited to them, and mainstream settings similation is an inevitable outcome of intergroup
(even the White House!) increasingly contain contact. For one thing, contact and competition
ethnoracial diversity. can lead to collective action by the majority
The changes associated with assimilation can group aimed at segregation and exclusion of
also be analyzed in terms of social boundaries, a minorities. Furthermore, the cases of the Amish
concept exceptionally useful for addressing so- of Pennsylvania and Hasidic Jews of New York
cial stratification (Wimmer 2013; Zolberg and City show that ethnoreligious minorities can
Long 1999). A social boundary is a distinction retain separate identities and communities even
724 • Part VII: Race and Ethnicity

while participating to some degree in civil soci- States, many immigrants face choices in which
ety. Not only does assimilation occur at different the degrees of risk and benefit are difficult to
rates for different ethnic and racial groups, but gauge; it is also usually the case that these choices
within the same ethnic group there is consider- involve unforeseeable long-term consequences.
able variation in the extent of assimilation—as is When contemplating the strategies best suited to
clear, for example, in the sharp contrast between improving their lives and those of their children,
intermarried Jews and the residents of socially immigrants and the second generation weigh
encapsulated Hasidic communities. the risks and potential benefits of “ethnic” strat-
A common feature of human migration is egies, which depend on opportunities available
that it is motivated and guided by the purpo- through ethnic networks, against “mainstream”
sive action of individuals embedded in social ones, which involve the dominant society’s
networks. What we learn from the long record higher educational system and labor markets.
of human migration is the enormous capacity Often enough, there may be little choice in these
of migrants and their descendants to adapt to matters: when immigrants have little human and
new social environments (Massey 1999). In im- financial capital and/or are undocumented, they
migrant societies like the United States, Canada, will usually be limited to jobs located through
and France, contemporary immigrants and their ethnic networks and constrained to reside in
children often settle in mixed ethnic neighbor- ethnic areas (Light 1972). But other immigrants
hoods and not only establish ongoing social re- may try mixed strategies, built from ethnic and
lationships with members of their own ethnic mainstream institutional elements (Sanders and
group, but also engage in frequent social and Nee 1996), such as when second-generation
economic transactions with individuals outside young adults obtain jobs through family and
that group. The higher the level of ethnic het- ethnic networks while continuing their educa-
erogeneity, the more likely it is that individuals tion, thus leaving multiple options open.
will transact with others outside of their ethnic Individuals striving for success in American
group, and the more common are crosscutting society often do not see themselves as assimilat-
ties connecting different ethnic groups (Nee, ing. Yet unintended consequences of practical
Sanders, and Sernau 1994; Lee and Bean 2010). strategies and actions undertaken in pursuit of
Assimilation proceeds incrementally, usually familiar goals—a good education, a good job, a
as an intergenerational process, stemming both decent place to live, interesting friends and ac-
from individuals’ purposive actions and from the quaintances, economic security—often result in
unintended consequences of their workaday deci- specific forms of assimilation. For example, it is
sions to optimize on past investments in human, not uncommon for first- and second-generation
cultural, and social capital. In the case of immi- parents to raise their children speaking only
grants and their children who may not inten- English, or at least to avoid placing their chil-
tionally seek to assimilate, the cumulative effect dren in bilingual educational settings, in the
of pragmatic social action aimed at successful belief that their chances for success in school
adaptation can give rise to changes in behavior will be improved by more complete mastery
that lead to assimilation. of the English language (Alba and Nee 2003;
Kasinitz et al. 2008). This is often true of fam-
ilies that instill aspirations for professional ca-
Purposive Action reers in their children. Surprising numbers of
At the individual level, a theory of assimilation second-generation Asian children speak only
must conceptually incorporate agency stemming English at home, which suggests that their
from purposive action and self-interest and pro- families have adopted this strategy. As another
vide an account of the incentives and motivation example, the search for a more desirable place
for assimilation. In adapting to life in the United to live—with good schools, safe streets, and
 Alba and Nee—Assimilation Theory for an Era of Unprecedented Diversity • 725

opportunities for children to grow up away from expanded reach of centralized authority in mod-
the seductions of deviant models of behavior— ern societies underscores the importance of the
leads immigrant professionals and entrepre- institutional mechanisms the state commands; it
neurs to move to middle-class suburbs, because is the sovereign actor in establishing the frame-
residential amenities tend to be concentrated work for the foundational rules of the game for
there. One consequence, whether intended or competition and cooperation in a society.
not, is much greater interaction with families In the post–civil rights era, the institutional
of other backgrounds; such increased contact mechanisms monitoring and enforcing federal
tends to encourage acculturation, especially for and state rules outlawing racism have increased
the children. the cost of discrimination in significant ways
Associated with acculturation are culturally (Skrentny 2002). For example, as a result of
codified notions of appropriate behavior, which Title VII of the Civil Rights Act of 1964, firms
when learned serve as cues to others of an indi- have become more attentive in observing anti-
vidual’s level of cultural and social competence. discrimination guidelines, with increasing num-
Such competence contributes to reducing social bers offering diversity and multicultural training
distance through the signaling of behaviors that workshops for managers and employees and in-
appear familiar and trustworthy. Physical differ- stituting company rules against racial and gender
ences may persist, but their effect on social dis- discrimination (Dobbin 2009). This attentive-
tance declines as cultural competence modulates ness is attributable in part to improvements in
social behavior in ways that highlight shared the monitoring of companies’ behavior by fed-
understandings and cultural attributes. It is not eral agencies. Landmark settlements of federal
surprising, given the emphasis on social mobility lawsuits have rendered the cost of discrimination
common to immigrant families, that the second more obvious to corporations and nonprofit or-
generation strives to acquire the mainstream cul- ganizations. They have also provided the federal
ture, even if its members also struggle to hold government with useful case histories and lessons
onto their ethnic one. Institutionalized incen- in dealing with discrimination lawsuits.
tives are such that it is rational for individuals, There are also cultural sources of compliance
regardless of ethnicity, to signal that they have with the rules of the game for a very broad range
the cultural and social competence to compete of societal actors. These have developed through
and perform in schools, the workplace, and shifts since the 1960s in beliefs and values about
other mainstream institutional contexts, espe- ethnoracial difference and equality. The cultural
cially when they believe they have good chances openings created by the moral leadership of
of success there. Martin Luther King and other civil rights leaders
have expanded to a broad revision in beliefs and
norms about diversity and inclusion of racial and
Institutional Mechanisms ethnic minorities. Certainly this ideological shift
The neo-assimilation theory turns in part on the has not ended racial prejudice and racist prac-
structure of relative rewards and the rules of the tices, but it has changed their character. They are
game embedded in the institutional environ- more covert and subterranean; racism as belief
ment. Institutional mechanisms comprise some has lost much of its public legitimacy and can
of the deeper causes, insofar as they determine no longer be advocated overtly without sanction
whether the proximate causes—such as purpo- (Schuman et al. 1997).
sive action—advance blending or segregating The watershed civil rights era legislation of
processes (Nee, Sanders, and Sernau 1994). The the 1960s guided institutional change and ex-
state’s monitoring and enforcement activity ac- tended the rule of open access to political and
counts for the effectiveness of coercive isomor- economic institutions to all Americans regard-
phism of institutional mechanisms. The vastly less of race and gender. The Immigration Act
726 • Part VII: Race and Ethnicity

of 1965 was the international counterpart of is not limited to racially defined groups). The
the far-reaching civil rights laws enacted by nature of boundary crossing suggests that it is
the US Congress. Opening legal immigration a selective process, which not everyone will be
established a direct institutional mechanism willing to undertake.
connecting authorized immigrants with access When there is a blurred boundary, by con-
to the more inclusive mainstream political and trast, individuals are seen as members of groups
economic institutions. Immigrants with appro- on both sides of the boundary, either simultane-
priate visa documents could follow a sequential ously or sequentially. Under these circumstances,
process of institutionalized transition from per- assimilation may be psychically easier, insofar
manent resident status to naturalized citizenship. as the individuals undergoing it do not sense
The assimilation of Asian immigrants and their a rupture between participation in mainstream
children suggests that in a knowledge-based institutions and familiar social and cultural prac-
economy immigrants with high stocks of human tices and identities; they do not feel forced to
capital are well positioned to benefit from the choose between the mainstream and their group
inclusive economic and social institutions of of origin. Assimilation of this type involves in-
the American mainstream (Nee and Holbrow termediate, or hyphenated, stages, which allow
2012). individuals to feel they are simultaneously mem-
bers of an ethnoracial minority and of the main-
stream. This process is less individualistic, open
Boundary Mechanisms to cohorts of a minority group who then recog-
Because social boundaries are by their nature two nize similarities in their experiences.
sided, the boundary concept brings attention The nature of a boundary is not necessar-
to the powerful effect of the majority popula- ily constant, and a focus of investigation for
tion on the life chances of minority individuals assimilation theory must be the circumstances
and thus its influence on the calculations that under which a boundary changes from bright
immigrant-origin individuals make when assess- to blurred, because a blurred boundary implies
ing the possible risks and benefits of an assimila- easier access to assimilation for a larger number
tion strategy. All boundaries are not the same in of minority individuals. Another way to think
this respect. As a rough cut, we can distinguish about this variation in boundaries is the resis-
between bright and blurred boundaries (Alba tance of the majority population to assimilation.
2009; Zolberg and Long 1999). Bright bound- The majority is often invested in preserving a
aries are unambiguous distinctions, so that in- bright boundary, which helps in the defense of
dividuals know at all times which side they the systemic advantages that majority individu-
are on. Blurred boundaries allow for modes of als enjoy (Tilly 1999). In extreme cases, some
self-presentation and social representation that people may use violence to rebuff challenges to
place individuals in ambiguous zones. a boundary, as happened in the American South
In the case of a bright boundary, assimilation during the civil rights movement.
takes the form of boundary crossing by individ- Recent American history offers some per-
uals. This process is likely to be experienced as suasive cases of boundary change, such as that
something akin to a conversion, that is, a depar- involving Asian Americans, who have made
ture from one group and a discarding of signs of the transition from racial outcast status during
membership in it, linked to an attempt to enter the early twentieth century to, it appears, eth-
into another, with all of the social and psychic nic groups, whose members increasingly have
burdens a conversion process entails: increasing mixed ancestry or intermarry with white Amer-
distance from peers, feelings of disloyalty, and icans. Another theoretically critical case is the
anxieties about acceptance. The epitome of this massive post–World War II assimilation of the
process is racial “passing” (but boundary crossing so-called white ethnics: Jews, Roman Catholics,
 Alba and Nee—Assimilation Theory for an Era of Unprecedented Diversity • 727

and Orthodox Christians of Irish or southern

and eastern European ancestry. On the eve of the Conclusion
war these ethnics were confined to the margins In an ethnoracially stratified society, where soci-
of the mainstream, “here under sufferance” and etal goods are distributed very unequally across
subordinated to white Protestant leadership, as groups, assimilation remains a proven method
President Franklin Roosevelt put it bluntly to a of entry into the mainstream and of access to
confidante during a conversation in 1942. Skep- the opportunities enjoyed by the members of the
ticism about the prospects of their assimilation majority group. Assimilation in a society as di-
was prevalent, and their religions had long been verse as the contemporary United States requires
suspect to white Protestants. By 1970, however, a new conception, however, one that, unlike the
the lines of division were fading rapidly. assimilation of mid-twentieth-century Amer-
The period in which the mass assimilation ica, does not envision membership in the white
of Europeans took place was one of great eco- majority as its endpoint. The concept presented
nomic growth, involving a transformation of here acknowledges that a decline in the social
the American occupational structure and a rapid significance of ethnoracial origins does not re-
expansion of higher education to train the cadre quire individuals to become carbon copies of the
of professionals and technical workers needed dominant majority; it is compatible with some
by advanced capitalism. This economic growth persistence of ethnoracial identities and cultural
was also associated with the development of new traits. A theory of assimilation also requires an
economic sectors and occupations. In concert, account of the mechanisms that bring it about,
these conditions produced what Alba (2009) and especially how the purposive actions of in-
has described as “non-zero-sum” mobility, a sit- dividuals may contribute to their assimilation.
uation in which minorities can advance socially Assimilation need not be the overt purpose,
and economically without appearing to threaten because when individuals seek to advance in
the life chances that whites take for granted for an ethnoracially stratified society, they will gen-
themselves and their children. This state of af- erally need to enter social settings containing
fairs is conducive to greater acceptance of mobile majority-group members and be able to interact
members of minority groups by the majority. with them on a basis of parity. In other words,
In principle, non-zero-sum mobility can decisions made by immigrants and their children
also occur in periods characterized by much to better their lives often entail some degree of
less robust economic expansion. In the near assimilation. These decisions are most effective,
future it could play a role in enhancing it need hardly be said, when the discrimination
immigrant-minority mobility because of pre- of the majority is not so severe as to exclude mi-
dictable demographic changes. In particular, the norities from mainstream milieus. The demo-
exodus of the baby boomers from the active ages, graphic shift toward diversity, inevitable during
a massive demographic shake-up, could provide the next quarter century, is likely to bring about
the opening. This group of Americans, born more blurring of ethnoracial boundaries and in-
between 1946 and 1964, is disproportionately crease the relevance of assimilation theory for the
white and highly educated and occupies much of American future.
the most rewarding terrain in the labor market.
As they retire between now and the early 2030s, REFERENCES
when the youngest of them turn seventy, there Alba, Richard. 2009. Blurring the Color Line: The
New Chance for a More Integrated America. Cam-
will be fewer members of the white majority en-
bridge, MA: Harvard University Press.
tering the workforce than are leaving it. Without
Alba, Richard, and Victor Nee. 2003. Remaking the
enough whites as replacements, there could be American Mainstream: Assimilation and Contem-
more room for young Americans of immigrant porary Immigration. Cambridge, MA: Harvard
backgrounds to advance. University Press.
728 • Part VII: Race and Ethnicity

Barth, Fredrik. 1969. Ethnic Groups and Boundaries. Nee, Victor, and Hilary Holbrow. 2012. “Why Asian
Boston: Little, Brown. Americans Are Becoming Mainstream.” Working
Bean, Frank, and Gillian Stevens. 2003. America’s Papers Series, Center for the Study of Economy
Newcomers and the Dynamics of Diversity. New and Society, Cornell University, Ithaca, NY.
York: Russell Sage Foundation. Nee, Victor, Jimy Sanders, and Scott Sernau. 1994.
Brubaker, Rogers. 2001. “The Return of Assimi- “Job Transitions in an Immigrant Metropolis:
lation? Changing Perspectives on Immigration Ethnic Boundaries and Mixed Economy.” Amer-
and its Sequels in France, Germany, and the ican Sociological Review 59: 849–872.
United States.” Ethnic and Racial Studies 24 (4): Park, Robert Ezra. 1930. “Assimilation, Social.”
531–548. In Encyclopedia of the Social Sciences, edited by
Dobbin, Frank. 2009. Inventing Equal Opportunity. Edwin Seligman and Alvin Johnson, 281–283.
Princeton, NJ: Princeton University Press. New York: Macmillan.
Glazer, Nathan. 1997. We Are All Multiculturalists Portes, Alejandro, and Ruben Rumbaut. 2001. Lega-
Now. Cambridge, MA: Harvard University Press. cies: The Story of the Immigrant Second Generation.
Gordon, Milton. 1964. Assimilation in American Berkeley and Los Angeles: University of Califor-
Life: The Role of Race, Religion, and National Ori- nia Press.
gins. New York: Oxford University Press. Portes, Alejandro, and Min Zhou. 1993. “The New
Huntington, Samuel. 2004. Who Are We? The Chal- Second Generation: Segmented Assimilation and
lenges to America’s National Identity. New York: Its Variants.” Annals of the American Academy of
Simon & Schuster. Political and Social Sciences 530: 74–96.
Kasinitz, Philip, John Mollenkopf, Mary Waters, and Sanders, Jimy M., and Victor Nee. 1996. “So-
Jennifer Holdaway. 2008. Inheriting the City: The cial Capital, Human Capital, and Immigrant
Children of Immigrants Come of Age. Cambridge, Self-Employment.” American Sociological Review
MA: Harvard University Press. 61: 231–249.
Lamont, Michèle, and Virág Molnár. 2002. “The Schuman, Howard, Charlotte Steeh, Lawrence
Study of Boundaries in the Social Sciences.” An- Bobo, and Maria Krysan. 1997. Racial Attitudes
nual Review of Sociology 28: 167–195. in America: Trends and Interpretations. Cam-
Lee, Jennifer, and Frank Bean. 2010. The Diver- bridge, MA: Harvard University Press.
sity Paradox: Immigration and the Color Line in Skrentny, John. 2002. The Minority Rights Revolu-
21st Century America. New York: Russell Sage tion. Cambridge, MA: Harvard University Press.
Foundation. Tilly, Charles. 1999. Durable Inequality. Berkeley
Light, Ivan. 1972. Ethnic Enterprise in America: Busi- and Los Angeles: University of California Press.
ness and Welfare Among Chinese, Japanese, and Warner, W. Lloyd, and Leo Srole. 1945. The Social
Blacks. Berkeley and Los Angeles: University of System of American Ethnic Groups. New Haven,
California Press. CT: Yale University Press.
Massey, Douglas. 1999. “Why Does Immigration Wimmer, Andreas. 2013. Ethnic Boundary Making:
Occur? A Theoretical Synthesis.” In The Hand- Institutions, Power, Networks. New York: Oxford
book of International Migration, edited by Charles University Press.
Hirschmann, Philip Kasinitz, and Josh DeWind, Zolberg, Aristide. 2006. A Nation by Design: Immi-
34–52. New York: Russell Sage Foundation. gration Policy in the Fashioning of America. New
Massey, Douglas, Camille Charles, Garvey Lundy, York: Russell Sage Foundation.
and Mary Fischer. 2003. Source of the River: The Zolberg, Aristide, and Long Litt Woon. 1999. “Why
Social Origins of Freshmen at America’s Selective Islam Is Like Spanish: Cultural Incorporation in
Colleges and Universities. Princeton, NJ: Prince- Europe and the United States.” Politics & Society
ton University Press. 27: 5–38.
 Portes and Zhou—The New Second Generation • 729

New Modes of Incorporation

84 • Alejandro Portes and Min Zhou

The New Second Generation

Segmented Assimilation and Its Variants

My name is Herb
and I’m not poor;
I’m the Herbie that you’re looking for,
like Pepsi,
a new generation
of Haitian determination—
I’m the Herbie that you’re looking for.

A beat tapped with bare hands, a few dance Growing up in an immigrant family has al-
steps, and the Haitian kid was rapping. His song, ways been difficult, as individuals are torn by
titled “Straight Out of Haiti,” was being per- conflicting social and cultural demands while
formed at Edison High, a school that sits astride they face the challenge of entry into an unfamil-
Little Haiti and Liberty City, the largest black iar and frequently hostile world. And yet the dif-
area of Miami. The lyrics captured well the dis- ficulties are not always the same. The process of
tinct outlook of his immigrant community. The growing up American oscillates between smooth
panorama of Little Haiti contrasts sharply with acceptance and traumatic confrontation depend-
the bleak inner city. In Miami’s Little Haiti, the ing on the characteristics that immigrants and
storefronts leap out at the passersby. Bright blues, their children bring along and the social context
reds, and oranges vibrate to Haitian merengue that receives them. In this article, we explore
blaring from sidewalk speakers.1 Yet, behind some of these factors and their bearing on the
the gay Caribbean exteriors, a struggle goes on process of social adaptation of the immigrant
that will define the future of this community. second generation. We propose a conceptual
As we will see later on, it involves the second framework for understanding this process and
generation—children like Herbie—subject to illustrate it with selected ethnographic material
conflicting pressure from parents and peers and and survey data from a recent survey of children
to pervasive outside discrimination. of immigrants.

Alejandro Portes and Min Zhou. “The New Second Generation: Segmented Assimilation and Its Variants,”
The Annals of the American Academy of Political and Social Science 530 (November 1993), pp. 75–77, 81–92,
96. National American Woman Suffrage Association Collection. Copyright © 1993 by Sage Publications, Inc.
Reprinted by permission of Sage Publications, Inc., via Copyright Clearance Center.
730 • Part VII: Race and Ethnicity

Research on the new immigration—that reduced a major barrier to entry into the Amer-
which arose after the passage of the 1965 Im- ican mainstream. For this reason, the process
migration Act—has been focused almost exclu- of assimilation depended largely on individual
sively on the first generation, that is, on adult decisions to leave the immigrant culture behind
men and women coming to the United States in and embrace American ways. Such an advantage
search of work or to escape political persecution. obviously does not exist for the black, Asian, and
Little noticed until recently is the fact that the mestizo children of today’s immigrants.
foreign-born inflow has been rapidly evolving Second, the structure of economic oppor-
from single adult individuals to entire family tunities has also changed. Fifty years ago, the
groups, including infant children and those born United States was the premier industrial power
to immigrants in the United States. By 1980, 10 in the world, and its diversified industrial labor
percent of dependent children in households requirements offered to the second generation
counted by the census were second-generation the opportunity to move up gradually through
immigrants.2 In the late 1980s, another study better-paid occupations while remaining part
put the number of students in kindergarten of the working class. Such opportunities have
through twelfth grade in American schools who increasingly disappeared in recent years follow-
spoke a language other than English at home at ing a rapid process of national deindustrializa-
3 to 5 million.3 tion and global industrial restructuring. This
The great deal of research and theorizing process has left entrants to the American labor
on post-1965 immigration offers only tentative force confronting a widening gap between the
guidance on the prospects and paths of adapta- minimally paid menial jobs that immigrants
tion of the second generation because the out- commonly accept and the high-tech and profes-
look of this group can be very different from sional occupations requiring college degrees that
that of their immigrant parents. For example, it native elites occupy.6 The gradual disappearance
is generally accepted among immigration theo- of intermediate opportunities also bears directly
rists that entry-level menial jobs are performed on the race between first-generation economic
without hesitation by newly arrived immigrants progress and second-generation expectations,
but are commonly shunned by their U.S.-reared noted previously. . . .
offspring. This disjuncture gives rise to a race
between the social and economic progress of
first-generation immigrants and the material Assimilation as a Problem
conditions and career prospects that their Amer- The Haitian immigrant community of Miami
ican children grow to expect.4 is composed of some 75,000 legal and clandes-
Nor does the existing literature on tine immigrants, many of whom sold everything
second-generation adaptation, based as it is on they owned in order to buy passage to America.
the experience of descendants of pre-World War First-generation Haitians are strongly oriented
I immigrants, offer much guidance for the un- toward preserving a strong national identity,
derstanding of contemporary events. The last which they associate both with community soli-
sociological study of children of immigrants was darity and with social networks promoting indi-
Irving Child’s Italian or American? The Second vidual success.7 In trying to instill national pride
Generation in Conflict, published fifty years ago.5 and an achievement orientation in their chil-
Conditions at the time were quite different from dren, they clash, however, with the youngsters’
those confronting settled immigrant groups everyday experiences in school. Little Haiti is ad-
today. Two such differences deserve special men- jacent to Liberty City, the main black inner-city
tion. First, descendants of European immigrants area of Miami, and Haitian adolescents attend
who confronted the dilemmas of conflicting cul- predominantly inner-city schools. Native-born
tures were uniformly white. Even if of a some- youths stereotype Haitians as too docile and
what darker hue than the natives, their skin color too subservient to whites and they make fun of
 Portes and Zhou—The New Second Generation • 731

French and Creole and of the Haitians’ accent. expectation is that the foreign-born and their
As a result, second-generation Haitian children offspring will first acculturate and then seek
find themselves torn between conflicting ideas entry and acceptance among the native-born as
and values: to remain Haitian they would have a prerequisite for their social and economic ad-
to face social ostracism and continuing attacks vancement. Otherwise, they remain confined to
in school; to become American—black Ameri- the ranks of the ethnic lower and lower-middle
can in this case—they would have to forgo their classes.10 This portrayal of the requirements for
parents’ dreams of making it in America on the mobility, so deeply embedded in the national
basis of ethnic solidarity and preservation of tra- consciousness, stands contradicted today by a
ditional values.8 growing number of empirical experiences.
An adversarial stance toward the white A closer look at these experiences indicates,
mainstream is common among inner-city mi- however, that the expected consequences of as-
nority youths who, while attacking the new- similation have not entirely reversed signs, but
comers’ ways, instill in them a consciousness that the process has become segmented. In other
of American-style discrimination. A common words, the question is into what sector of Amer-
message is the devaluation of education as a ican society a particular immigrant group assim-
vehicle for advancement of all black youths, a ilates. Instead of a relatively uniform mainstream
message that directly contradicts the immigrant whose mores and prejudices dictate a common
parents’ expectations. Academically outstanding path of integration, we observe today several
Haitian American students, “Herbie” among distinct forms of adaptation. One of them rep-
them, have consciously attempted to retain their licates the time-honored portrayal of growing
ethnic identity by cloaking it in black American acculturation and parallel integration into the
cultural forms, such as rap music. Many others, white middle-class; a second leads straight in
however, have followed the path of least effort the opposite direction to permanent poverty
and become thoroughly assimilated. Assimila- and assimilation into the underclass; still a third
tion in this instance is not into mainstream cul- associates rapid economic advancement with
ture but into the values and norms of the inner deliberate preservation of the immigrant com-
city. In the process, the resources of solidarity munity’s values and tight solidarity. This pattern
and mutual support within the immigrant com- of segmented assimilation immediately raises the
munity are dissipated. question of what makes some immigrant groups
An emerging paradox in the study of today’s become susceptible to the downward route and
second generation is the peculiar forms that as- what resources allow others to avoid this course.
similation has adopted for its members. As the In the ultimate analysis, the same general process
Haitian example illustrates, adopting the out- helps explain both outcomes. We advance next
looks and cultural ways of the native-born does our hypotheses as to how this process takes place
not represent, as in the past, the first step toward and how the contrasting outcomes of assimila-
social and economic mobility but may lead to tion can be explained. This explanation is then
the exact opposite. At the other end, immigrant illustrated with recent empirical material in the
youths who remain firmly ensconced in their re- final section.
spective ethnic communities may, by virtue of
this fact, have a better chance for educational
and economic mobility through use of the ma- Vulnerability and Resources
terial and social capital that their communities Along with individual and family variables, the
make available.9 context that immigrants find upon arrival in
This situation stands the cultural blueprint their new country plays a decisive role in the
for advancement of immigrant groups in Amer- course that their offspring’s lives will follow.
ican society on its head. As presented in innu- This context includes such broad variables as
merable academic and journalistic writings, the political relations between sending and receiving
732 • Part VII: Race and Ethnicity

countries and the state of the economy in the moving ahead economically and, hence, their
latter and such specific ones as the size and struc- children have no objective reasons for embrac-
ture of preexisting coethnic communities. The ing a counter-cultural message. If successful, the
concept of modes of incorporation provides a process can effectively block parental plans for
useful theoretical tool to understand this diver- intergenerational mobility.
sity. As developed in prior publications, modes The third contextual source of vulnerability
of incorporation consist of the complex formed has to do with changes in the host economy
by the policies of the host government; the values that have led to the evaporation of occupational
and prejudices of the receiving society; and the ladders for intergenerational mobility. As noted
characteristics of the coethnic community. These previously, new immigrants may form the back-
factors can be arranged in a tree of contextual bone of what remains of labor-intensive manu-
situations, illustrated by Figure 84.1. This figure facturing in the cities as well as in their growing
provides a first approximation to our problem.11 personal services sector, but these are niches that
To explain second-generation outcomes and seldom offer channels for upward mobility. The
their segmented character, however, we need new hourglass economy, created by economic re-
to go into greater detail into the meaning of structuring, means that children of immigrants
these various modes of incorporation from the must cross a narrow bottleneck to occupations
standpoint of immigrant youths. There are three requiring advanced training if their careers are to
features of the social contexts encountered by keep pace with their U.S.-acquired aspirations.
today’s newcomers that create vulnerability to This race against a narrowing middle demands
downward assimilation. The first is color, the that immigrant parents accumulate sufficient
second is location, and the third is the absence of resources to allow their children to effect the
mobility ladders. As noted previously, the major- passage and to simultaneously prove to them the
ity of contemporary immigrants are nonwhite. viability of aspirations for upward mobility. Oth-
Although this feature may appear at first glance erwise, assimilation may not be into mainstream
as an individual characteristic, in reality it is a values and expectations but into the adversarial
trait belonging to the host society. Prejudice is stance of impoverished groups confined to the
not intrinsic to a particular skin color or racial bottom of the new economic hourglass.
type, and, indeed, many immigrants never expe- The picture is painted in such stark terms
rienced it in their native lands. It is by virtue of here for the sake of clarity, although in real-
moving into a new social environment, marked ity things have not yet become so polarized.
by different values and prejudices, that physical Middle-level occupations requiring relatively
features become redefined as a handicap. modest educational achievements have not com-
The concentration of immigrant households pletely vanished. By 1980, skilled blue-collar
in cities and particularly in central cities, as jobs—classified by the U.S. census as “precision
documented previously, gives rise to a second production, craft, and repair occupations”—had
source of vulnerability because it puts new ar- declined by 1.1 percent relative to a decade ear-
rivals in close contact with concentrations of lier but still represented 13 percent of the ex-
native-born minorities. This leads to the iden- perienced civilian labor force, or 13.6 million
tification of the condition of both groups— workers. Mostly clerical administrative support
immigrants and the native poor—as the same occupations added another 16.9 percent, or 17.5
in the eyes of the majority. More important, it million jobs. In 1980, occupations requiring a
exposes second-generation children to the ad- college degree had increased by 6 percent in
versarial subculture developed by marginalized comparison with 1970, but they still employed
native youths to cope with their own difficult less than a fifth—18.2 percent—of the Ameri-
situation.12 This process of socialization may can labor force.13 Even in the largest cities, oc-
take place even when first-generation parents are cupations requiring only a high school diploma
 Figure 84.1. Models of Incorporation: A Typology
I. Government Policy1 Receptive Indifferent Hostile

II. Societal Reception2 Prejudiced Nonprejudiced Pr N-Pr Pr N-Pr

(Pr) (N-Pr)

III. Coethnic Community3 Weak Strong W S W S W S W S W S

(W) (S)

IV. Examples4 Cambodian, Hungarians Legal Argentines, Haitian Illegal

Hmong and Other Mexicans, 1950– Boat Immigrants
Refugees, Small East 1900– People, from
1975– European 1979–83 Small
Refugee European
Groups Countries

Chinese, Cuban Koreans, Swiss, Mariel Illegal

Vietnamese Refugees, 1965– Scandinavians Cubans, Irish in
Refugees, 1960–80 in Midwest 1980 Boston,
1975– Settlements, 1975–90
1900–
Source: Adapted from Alajandro Portes and Rubén G. Rumbaut, Immigrant America: A Portrait (Berkeley: University of California Press, 1990), p. 91. Copyright ©
1990 by the Regents of the University of California.
1. Receptive Policy is defined as legal entry with resettlement assistance, indifferent as legal entry without resettlement assistance, hostile as active opposition to a group’s
entry or permanence in the country; 2. Prejudiced reception is defined as that accorded to nonphenotypically white groups; nonprejudiced is that accorded to European
and European-origin whites; 3. Weak coethnic communities are either small in numbers or composed primarily of manual workers; strong communities feature sizable
numerical concentrations and a diversified occupational structure including entrepreneurs and professionals; 4. Examples include immigrant groups arriving from the
start of the century to the present. Dates of migration are approximate. Groups reflect broadly but not perfectly the characteristics of each ideal type.
733
734 • Part VII: Race and Ethnicity

were common by the late 1980s. In New York environment where generations of Irish Amer-
City, for example, persons with 12 years or less icans had established a secure foothold. Public
of schooling held just over one half of the jobs sympathy effectively neutralized governmental
in 1987. Clerical, service, and skilled blue-collar hostility in this case, culminating in a change
jobs not requiring a college degree represented of the immigration law directly benefiting the
46 percent.14 Despite these figures, there is little newcomers.17
doubt that the trend toward occupational seg- Third, and most important, are the re-
mentation has increasingly reduced opportuni- sources made available through networks in
ties for incremental upward mobility through the coethnic community. Immigrants who join
well-paid blue-collar positions. The trend forces well-established and diversified ethnic groups
immigrants today to bridge in only one genera- have access from the start to a range of moral
tion the gap between entry-level jobs and profes- and material resources well beyond those avail-
sional positions that earlier groups took two or able through official assistance programs. Edu-
three generations to travel. cational help for second-generation youths may
Different modes of incorporation also make include not only access to college grants and
available, however, three types of resources to loans but also the existence of a private school
confront the challenges of contemporary assim- system geared to the immigrant community’s
ilation. First, certain groups, notably political values. Attendance at these private ethnic schools
refugees, are eligible for a variety of government insulates children from contact with native mi-
programs including educational loans for their nority youths, while reinforcing the authority of
children. The Cuban Loan Program, imple- parental views and plans.
mented by the Kennedy administration in con- In addition, the economic diversification of
nection with its plan to resettle Cuban refugees several immigrant communities creates niches
away from South Florida, gave many impover- of opportunity that members of the second
ished first- and second-generation Cuban youths generation can occupy, often without a need for
a chance to attend college. The high proportion an advanced education. Small-business appren-
of professionals and executives among Cuban ticeships, access to skilled building trades, and
American workers today, a figure on a par with well-paid jobs in local government bureaucra-
that for native white workers, can be traced, at cies are some of the ethnic niches documented
least in part, to the success of that program.15 in the recent literature.18 In 1987, average sales
Passage of the 1980 Refugee Act gave to subse- per firm of the smaller Chinese, East Indian, Ko-
quent groups of refugees, in particular Southeast rean, and Cuban enterprises exceeded $100,000
Asians and Eastern Europeans, access to a simi- per year and they jointly employed over 200,000
larly generous benefits package.16 workers. These figures omit medium-sized and
Second, certain foreign groups have been large ethnic firms, whose sales and work forces
exempted from the traditional prejudice en- are much larger.19 Fieldwork in these communi-
dured by most immigrants, thereby facilitating a ties indicates that up to half of recently arrived
smoother process of adaptation. Some political immigrants are employed by coethnic firms and
refugees, such as the early waves of exiles from that self-employment offers a prime avenue for
Castro’s Cuba, Hungarians and Czechs escaping mobility to second-generation youths.20 Such
the invasions of their respective countries, and community-mediated opportunities provide a
Soviet Jews escaping religious persecution, pro- solution to the race between material resources
vide examples. In other cases, it is the cultural and second-generation aspirations not available
and phenotypical affinity of newcomers to ample through competition in the open labor market.
segments of the host population that ensures a Through creation of a capitalism of their own,
welcome reception. The Irish coming to Boston some immigrant groups have thus been able
during the 1980s are a case in point. Although to circumvent outside discrimination and the
many were illegal aliens, they came into an threat of vanishing mobility ladders.
 Portes and Zhou—The New Second Generation • 735

In contrast to these favorable conditions are which the Mexican-origin population could be
those foreign minorities who either lack a com- classified. Recent Mexican immigrants were at
munity already in place or whose coethnics are one extreme. They dressed differently and un-
too poor to render assistance. The condition of stylishly. They claimed an identity as Mexican
Haitians in South Florida, cited earlier, provides and considered Mexico their permanent home.
an illustration of one of the most handicapped The most academically successful of this group
modes of incorporation encountered by con- were those most proficient in Spanish, reflecting
temporary immigrants, combining official hos- their prior levels of education in Mexico. Almost
tility and widespread social prejudice with the all were described by teachers and staff as courte-
absence of a strong receiving community.21 From ous, serious about their schoolwork, respectful,
the standpoint of second-generation outcomes, and eager to please as well as naive and unsophis-
the existence of a large but downtrodden coeth- ticated. They were commonly classified as LEP.
nic community may be even less desirable than The next category comprised Mexican-
no community at all. This is because newly ar- oriented students. They spoke Spanish at home
rived youths enter into ready contact with the and were generally classified as fluent English
reactive subculture developed by earlier gen- proficient (FEP). They had strong bicultural ties
erations. Its influence is all the more powerful with both Mexico and the United States, reflect-
because it comes from individuals of the same ing the fact that most were born in Mexico but
national origin, “people like us” who can more had lived in the United States for more than five
effectively define the proper stance and attitudes years. They were proud of their Mexican heri-
of the newcomers. To the extent that they do so, tage but saw themselves as different from the first
the first-generation model of upward mobility group, the recién llegados (recently arrived), as
through school achievement and attainment of well as from the native-born Chicanos and Cho-
professional occupations will be blocked. los, who were derided as people who had lost
their Mexican roots. Students from this group
were active in soccer and the Sociedad Bilingue
Three Examples and in celebrations of May 5th, the anniver-
Mexicans and Mexican Americans sary of the Mexican defeat of French occupying
Field High School (the name is fictitious) is lo- forces. Virtually all of the Mexican-descent stu-
cated in a small coastal community of central dents who graduated in the top 10 percent of
California whose economy has long been tied their class in 1981 were identified as members
to agricultural production and immigrant farm of this group.
labor. About 57 percent of the student popula- Chicanos were by far the largest Mexican-
tion is of Mexican descent. An intensive ethno- descent group at Field High. They were mostly
graphic study of the class of 1985 at Field High U.S.-born second- and third-generation students
began with school records that showed that the whose primary loyalty was to their in-group, seen
majority of U.S.-born Spanish-surname students as locked in conflict with white society. Chicanos
who had entered the school in 1981 had dropped referred derisively to successful Mexican students
out by their senior year. However, only 35 per- as “schoolboys” and “schoolgirls” or as “wan-
cent of the Spanish-surname students who had nabes.” According to M. G. Matute-Bianchi,
been originally classified by the school as limited
English proficient (LEP) had dropped out. The To be a Chicano meant in practice to hang
figure was even lower than the corresponding out by the science wing . . . not eating lunch
one for native white students, 40 percent. LEP in the quad where all the “gringos” and
status is commonly assigned to recently arrived “schoolboys” hang out . . . cutting classes
Mexican immigrants.22 by faking a call slip so you can be with your
Intensive ethnographic fieldwork at the friends at the 7-11 . . . sitting in the back of
school identified several distinct categories in classes and not participating . . . not carrying
736 • Part VII: Race and Ethnicity

your books to class . . . not taking the difficult rural Sikhs from the Punjab. By the early 1980s,
classes . . . doing the minimum to get by.23 second-generation Punjabi students already ac-
counted for 11 percent of the student body at
Chicanos merge imperceptibly into the last Valleyside High. Their parents were no longer
category, the Cholos, who were commonly seen only farm laborers, since about a third had be-
as “low riders” and gang members. They were come orchard owners themselves and another
also native-born Mexican Americans, easily iden- third worked in factories in the nearby San
tifiable by their deliberate manner of dress, walk, Francisco area. An ethnographic study of Val-
speech, and other cultural symbols. Chicanos leyside High School in 1980–82 revealed a very
and Cholos were generally regarded by teachers difficult process of assimilation for Punjabi Sikh
as “irresponsible,” “disrespectful,” “mistrusting,” students. According to its author, M. A. Gibson,
“sullen,” “apathetic,” and “less motivated,” and Valleyside is “redneck country,” and white resi-
their poor school performance was attributed to dents are extremely hostile to immigrants who
these traits.24 According to Matute-Bianchi, Chi- look different and speak a different language:
canos and Cholos were faced with what they saw “Punjabi teenagers are told they stink . . . told
as a forced-choice dilemma between doing well to go back to India . . . physically abused by ma-
in school or being a Chicano. To act white was jority students who spit at them, refuse to sit by
regarded as disloyalty to one’s group. them in class or in buses, throw food at them or
The situation of these last two groups exem- worse.”25
plifies losing the race between first-generation Despite these attacks and some evidence of
achievements and later generations’ expectations. discrimination by school staff, Punjabi students
Seeing their parents and grandparents confined performed better academically than majority
to humble menial jobs and increasingly aware of Anglo students. About 90 percent of the immi-
discrimination against them by the white main- grant youths completed high school, compared
stream, U.S.-born children of earlier Mexican to 70–75 percent of native whites. Punjabi boys
immigrants readily join a reactive subculture as surpassed the average grade point average, were
a means of protecting their sense of self-worth. more likely to take advanced science and math
Participation in this subculture then leads to classes, and expressed aspirations for careers in
serious barriers to their chances of upward mo- science and engineering. Girls, on the other
bility because school achievement is defined as hand, tended to enroll in business classes, but
antithetical to ethnic solidarity. Like Haitian they paid less attention to immediate career
students at Edison High, newly arrived Mexican plans, reflecting parental wishes that they should
students are at risk of being socialized into the marry first. This gender difference is indicative
same reactive stance, with the aggravating fac- of the continuing strong influence exercised
tor that it is other Mexicans, not native-born by the immigrant community over its second
strangers, who convey the message. The principal generation. According to Gibson, Punjabi par-
protection of mexicanos against this type of as- ents pressured their children against too much
similation lies in their strong identification with contact with white peers who may “dishonor”
home-country language and values, which brings the immigrants’ families, and defined “becom-
them closer to their parents’ cultural stance. ing Americanized” as forgetting one’s roots and
adopting the most disparaged traits of the ma-
jority, such as leaving home at age 18, making
Punjabi Sikhs in California decisions without parental consent, dating, and
Valleyside (a fictitious name) is a northern Cali- dancing. At the same time, parents urged chil-
fornia community where the primary economic dren to abide by school rules, ignore racist re-
activity is orchard farming. Farm laborers in this marks and avoid fights, and learn useful skills,
area come often from India; they are mainly including full proficiency in English.26
 Portes and Zhou—The New Second Generation • 737

The overall success of this strategy of se- much to do with this transformation, as it sent
lective assimilation to American society is re- the entire Cuban upper class out of the coun-
markable because Punjabi immigrants were try, followed by thousands of refugees of more
generally poor on their arrival in the United modest backgrounds. Over time, Cubans cre-
States and confronted widespread discrimina- ated a highly diversified and prosperous ethnic
tion from whites without the benefit of either community that provided resources for the ad-
governmental assistance or a well-established aptation process of its second generation. Re-
coethnic community. In terms of our typology flecting this situation are average Cuban family
of vulnerability and resources, the Punjabi Sikh incomes that, by 1989, approximated those of
second generation was very much at risk except the native-born population; the existence in
for two crucial factors. First, immigrant par- 1987 of more than 30,000 Cuban-owned small
ents did not settle in the inner city or in close businesses that formed the core of the Miami
proximity to any native-born minority whose ethnic enclave; and the parallel rise of a private
offspring could provide an alternative model of school system oriented toward the values and
adaptation to white-majority discrimination. In political outlook of this community.27 In terms
particular, the absence of a downtrodden Indian of the typology of vulnerability and resources,
American community composed of children of well-sheltered Cuban American teenagers lack
previous immigrants allowed first-generation any extensive exposure to outside discrimina-
parents to influence decisively the outlook of tion, they have little contact with youths from
their offspring, including their ways of fighting disadvantaged minorities, and the development
white prejudice. There was no equivalent of a of an enclave creates economic opportunities
Cholo-like reactive subculture to offer an alter- beyond the narrowing industrial and tourist sec-
native blueprint of the stance that “people like tors on which most other immigrant groups in
us” should take. the area depend. Across town, Haitian American
Second, Punjabi immigrants managed to teenagers face exactly the opposite set of condi-
make considerable economic progress, as attested tions, as has been shown.
by the number who had become farm owners, Among the other immigrant groups that form
while maintaining a tightly knit ethnic commu- Miami’s ethnic mosaic, two deserve mention be-
nity. The material and social capital created by cause they represent intermediate situations be-
this first-generation community compensated tween those of the Cubans and Haitians. One
for the absence of an older coethnic group and comprises Nicaraguans escaping the Sandinista
had decisive effects on second-generation out- regime during the 1980s. They were not as wel-
looks. Punjabi teenagers were shown that their comed in the United States as were the Cuban
parents’ ways paid off economically, and this exiles, nor were they able to develop a large and
fact, plus their community’s cohesiveness, en- diversified community. Yet they shared with
dowed them with a source of pride to counter- Cubans their language and culture, as well as a
act outside discrimination. Through this strategy militant anti-Communist discourse. This com-
of selective assimilation, Punjabi Sikhs appeared mon political outlook led the Cuban American
to be winning the race against the inevitable ac- community to extend its resources in support of
culturation of their children to American-style their Nicaraguan brethren, smoothing their pro-
aspirations. cess of adaptation.28 For second-generation Nic-
araguans, this means that the preexisting ethnic
community that provides a model for their own
Caribbean Youths in South Florida assimilation is not a downtrodden group but
Miami is arguably the American city that has rather one that has managed to establish a firm
been most thoroughly transformed by post- and positive presence in the city’s economy and
1960 immigration. The Cuban Revolution had politics.
738 • Part VII: Race and Ethnicity

The second group comprises West Indi- may prove the best course for immigrant minori-
ans coming from Jamaica, Trinidad, and other ties. But the extent to which this strategy is pos-
English-speaking Caribbean republics. They sible also depends on the history of each group
generally arrive in Miami as legal immigrants, and its specific profile of vulnerabilities and re-
and many bring along professional and business sources. The present analysis represents a prelim-
credentials as well as the advantage of fluency inary step toward understanding these realities.
in English. These individual advantages are dis-
counted, however, by a context of reception in NOTES
which these mostly black immigrants are put in 1. Alejandro Portes and Alex Stepick, City on the
Edge: The Transformation of Miami (Berkeley: Uni-
the same category as native-born blacks and dis-
versity of California Press, 1993), chap. 8.
criminated against accordingly. The recency of
2. Defined as native-born children with at least
West Indian migration and its small size have one foreign-born parent or children born abroad
prevented the development of a diversified ethnic who came to the United States before age 12.
community in South Florida. Hence new arrivals 3. Joan N. First and John W. Carrera, New Voices:
experience the full force of white discrimination Immigrant Students in U.S. Public Schools (Boston:
without the protection of a large coethnic group National Coalition of Advocates for Students, 1988).
and with constant exposure to the situation and 4. Michael Piore, Birds of Passage (New York:
attitudes of the inner-city population. Despite Cambridge University Press, 1979); Herbert Gans,
considerable individual resources, these disad- “Second-Generation Decline: Scenarios for the
Economic and Ethnic Futures of the Post-1965
vantages put the West Indian second generation
American Immigrants,” Ethnic and Racial Studies
at risk of bypassing white or even native black 15:173–92 (Apr. 1992).
middle-class models to assimilate into the cul- 5. Irving L. Child, Italian or American? The Sec-
ture of the underclass. . . . ond Generation in Conflict (New Haven, CT: Yale
University Press, 1943).
6. See, for example, Saskia Sassen, “Changing
Conclusion Composition and Labor Market Location of His-
Fifty years ago, the dilemma of Italian Ameri- panic Immigrants in New York City, 1960–1980,”
can youngsters studied by Irving Child consisted in Hispanics in the U.S. Economy, ed. George J. Bor-
jas and Marta Tienda (New York: Academic Press,
of assimilating into the American mainstream,
1985), pp. 299–322.
sacrificing in the process their parents’ cultural
7. See Alex Stepick, “Haitian Refugees in the
heritage in contrast to taking refuge in the ethnic U.S.” (Report no. 52, Minority Rights Group,
community from the challenges of the outside London, 1982); Alex Stepick and Alejandro Portes,
world. In the contemporary context of seg- “Flight into Despair: A Profile of Recent Haitian
mented assimilation, the options have become Refugees in South Florida,” International Migration
less clear. Children of nonwhite immigrants may Review, 20:329–50 (Summer 1986).
not even have the opportunity of gaining access 8. This account is based on fieldwork in Miami
to middle-class white society, no matter how conducted in preparation for a survey of immigrant
youths in public schools.
acculturated they become. Joining those native
9. On the issue of social capital, see James S.
circles to which they do have access may prove
Coleman, “Social Capital in the Creation of Human
a ticket to permanent subordination and disad- Capital,” American Journal of Sociology, supplement,
vantage. Remaining securely ensconced in their 94:S95–121 (1988); Alejandro Portes and Min
coethnic community, under these circumstances, Zhou, “Gaining the Upper Hand: Economic Mo-
may be not a symptom of escapism but the best bility among Immigrant and Domestic Minorities,”
strategy for capitalizing on otherwise unavailable Ethnic and Racial Studies, 15:491–522 (Oct. 1992).
material and moral resources. As the experiences On ethnic entrepreneurship, see Ivan H. Light, Eth-
of Punjabi Sikh and Cuban American students nic Enterprise in America: Business and Welfare among
Chinese, Japanese, and Blacks (Berkeley: University
suggest, a strategy of paced, selective assimilation
of California Press, 1972); Kenneth Wilson and W.
 Portes and Zhou—The New Second Generation • 739

Allen Martin, “Ethnic Enclaves: A Comparison of 18. Bailey and Waldinger, “Primary, Secondary,
the Cuban and Black Economies in Miami,” Ameri- and Enclave Labor Markets”; Min Zhou, New York’s
can Journal of Sociology, 88:135–60 (1982). Chinatown: The Socioeconomic Potential of an Urban
10. See W. Lloyd Warner and Leo Srole, The So- Enclave (Philadelphia: Temple University Press,
cial Systems of American Ethnic Groups (New Haven, 1992); Wilson and Martin, “Ethnic Enclaves”; Su-
CT: Yale University Press, 1945); Thomas Sowell, zanne Model, “The Ethnic Economy: Cubans and
Ethnic America: A History (New York: Basic Books, Chinese Reconsidered” (Manuscript, University of
1981). Massachusetts at Amherst, 1990).
11. See Alejandro Portes and Rubén G. Rum- 19. U.S. Department of Commerce, Bureau of
baut, Immigrant America: A Portrait (Berkeley: Uni- the Census, Survey of Minority-Owned Business En-
versity of California Press, 1990), chap. 3. terprises, 1987, MB-2 and MB-3 (Washington, DC:
12. See Mercer L. Sullivan, “Getting Paid”: Youth, Department of Commerce, 1991).
Crime, and Work in the Inner City (Ithaca, NY: Cor- 20. Alejandro Portes and Alex Stepick, “Unwel-
nell University Press, 1989), chaps. 1, 5. come Immigrants: The Labor Market Experiences
13. U.S. Department of Commerce, Bureau of of 1980 (Mariel) Cuban and Haitian Refugees
the Census, Census of Population and Housing, 1980: in South Florida,” American Sociological Review,
Public Use Microdata Samples A (MRDF) (Washing- 50:493–514 (Aug. 1985); Zhou, New York’s Chi-
ton, DC: Department of Commerce, 1983). natown; Luis E. Guarnizo, “One Country in Two:
14. Thomas Bailey and Roger Waldinger, “Pri- Dominican-Owned Firms in New York and the
mary, Secondary, and Enclave Labor Markets: A Dominican Republic” (Ph.D. diss. Johns Hopkins
Training System Approach,” American Sociological University, 1992); Bailey and Waldinger, “Primary,
Review, 56:432–45 (1991). Secondary, and Enclave Labor Markets.”
15. Professionals and executives represented 21. Stepick, “Haitian Refugees in the U.S.”; Jake
25.9 percent of Cuban-origin males aged 16 years C. Miller, The Plight of Haitian Refugees (New York:
and over in 1989; the figure for the total adult male Praeger, 1984).
population was 26 percent. See Jesus M. García and 22. M. G. Matute-Bianchi, “Ethnic Identities
Patricia A. Montgomery, The Hispanic Population of and Patterns of School Success and Failure among
the United States: March 1990, Current Population Mexican-Descent and Japanese-American Students
Reports, ser. P-20, no. 449 (Washington, DC: De- in a California High School,” American Journal of
partment of Commerce, 1991). Education, 95:233–55 (Nov. 1986). This study is
16. Portes and Rumbaut, Immigrant America, pp. summarized in Rubén G. Rumbaut, “Immigrant
23–25; Robert L. Bach et al., “The Economic Ad- Students in California Public Schools: A Summary
justment of Southeast Asian Refugees in the United of Current Knowledge” (Report no. 11, Center for
States,” in World Refugee Survey, 1983 (Geneva: Research on Effective Schooling for Disadvantaged
United Nations High Commission for Refugees, Children, Johns Hopkins University, Aug. 1990).
1984), pp. 51–55. 23. Matute-Bianchi, “Ethnic Identities and Pat-
17. The 1990 Immigration Act contains terns,” p. 253.
tailor-made provisions to facilitate the legalization 24. Rumbaut, “Immigrant Students,” p. 25.
of Irish immigrants. Those taking advantage of the 25. M. A. Gibson, Accommodation without Assim-
provisions are popularly dubbed “Kennedy Irish” in ilation: Sikh Immigrants in an American High School
honor of the Massachusetts Senator who coauthored (Ithaca, NY: Cornell University Press, 1989), p. 268.
the act. On the 1990 act, see Michael Fix and Jef- 26. Gibson, Accommodation without Assimilation.
frey S. Passel, “The Door Remains Open: Recent The study is summarized in Rumbaut, “Immigrant
Immigration to the United States and a Preliminary Students,” pp. 22–23.
Analysis of the Immigration Act of 1990” (Work- 27. García and Montgomery, Hispanic Popula-
ing paper, Urban Institute and RAND Corporation, tion; U.S. Department of Commerce, Bureau of the
1991). On the Irish in Boston, see Karen Tumulty, Census, Survey of Minority-Owned Business Enter-
“When Irish Eyes Are Hiding . . . ,” Los Angeles prises, MB-2.
Times, 29 Jan. 1989. 28. Portes and Stepick, City on the Edge, chap. 7.
740 • Part VII: Race and Ethnicity

85 • Tomás R. Jiménez

Why Replenishment Strengthens Racial

and Ethnic Boundaries

Introduction form after immigration ceases, what form does

It has long been argued that ethnic boundaries it take when the immigrant population is replen-
will fade and ethnic identities will become sym- ished with new waves of immigrants, such as the
bolic and optional as intermarriage increases and Mexican-origin population?
differences in socioeconomic status, residential Mexican migration to the United States is
location, and language abilities decrease (Alba distinct from that of other immigrant groups in
1990; Gans 1979; Gordon 1964; Waters 1990; many respects, but perhaps most especially in
Waters and Jiménez 2005). But the canonical ac- its long history. As Figure 85.1 shows, Mexican
count falls short by not adequately addressing the immigration continued to rise after European
role that immigration patterns play in formation of immigration declined. The diverging patterns
ethnic identity. This is probably because the sym- are particularly prominent in the period after
bolic and optional ethnic identity of white ethnics 1970, when the foreign-born Mexican popula-
today was formed against a backdrop of radically tion spiked, while the number of immigrants
reduced levels of immigration. World War I, re- from European countries continued its down-
strictive immigration laws passed in the 1920s, ward path. Later-generation Mexican Ameri-
the Great Depression of the 1930s, and World cans who descend from early waves of Mexican
War II combined to slow European immigration immigrants thus live in a US society where mi-
to a trickle. This near cessation meant that each gration from their ancestral homeland remains
generation born in the United States came of age prominent. In sharp contrast, European-origin
in an American society that was decidedly less im- immigrants are mostly absent from the ethnic
migrant in character. These US-born ethnics thus landscape that later-generation white ethnics
had less contact with individuals for whom eth- negotiate. This continuous influx of immigrants
nicity played a more central role in daily life. Yet means that the Mexican-origin population is
the literature on assimilation is silent in explain- both large and generationally diverse. Although
ing how the halt of immigration contributed to a majority of Mexicans in the United States are
the racial and ethnic identity formation of white either first- or second-generation (immigrants
ethnics. If ethnicity takes a symbolic, optional or children of immigrants), roughly three in ten

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Tomás R. Jiménez, “Mexican Immigrant Replenishment and the Continuing Sig-
nificance of Ethnicity and Race,” American Journal of Sociology 113:6 (May 2008), pp. 1527–1567, published
by the University of Chicago Press. The article printed here was originally prepared by Tomás R. Jiménez for
this fourth edition of Social Stratification: Class, Race, and Gender in Sociological Perspective, edited by David B.
Grusky. Copyright © 2014 by Westview Press.
 Jiménez—Why Replenishment Strengthens Racial and Ethnic Boundaries • 741

Figure 85.1. Number of Foreign-Born People from Mexico and Selected European Countries as a Percent
of Total US Population, 1900–2000

Source: U.S. Census.

are third- or later-generation (grandchildren and beet fields (Avila 1997). Mexican immigrant set-
beyond). tlement shifted away from Kansas in the middle
of the twentieth century, and there was a roughly
forty-year hiatus in Mexican immigration to the
Methods, Research Setting, and state. In 1980 the largest beef-packing plant in
Respondents the world opened near Garden City, attracting a
The analysis in this chapter draws on 123 new wave of Mexican migrants.
in-depth interviews with later-generation Mex- Santa Maria is located on the central coast
ican Americans and participant observation in of California; 52.3 percent of its 77,423 inhab-
Garden City, Kansas, and Santa Maria, Califor- itants are of Mexican origin. As in Garden City,
nia. I interviewed respondents whose ancestors roughly half of the Mexican-origin population is
have been in the United States since 1940 or be- foreign born. Unlike in Garden City, however,
fore, who are of Mexican descent on both their Mexican migration to Santa Maria was constant
mother’s and father’s side of the family, and who throughout the twentieth century, but especially
have lived in their respective cities for most of heavy from the early 1990s on. Agricultural
their lives. work has always attracted Mexican immigrants
Garden City is located in southwestern Kan- to Santa Maria, where they are the primary, if
sas; at the time of the interviews, it had 28,451 not only, source of agricultural labor. Variation
residents, 34.7 percent of whom were of Mex- in immigration patterns yielded some differ-
ican origin. About half of the Mexican-origin ences, but the heavy influx of Mexican immi-
population were foreign born. Between roughly grants to both cities over the last twenty years
1900 and 1930, Mexican immigrants came to suppressed any pronounced differences related to
the area to build the railroads and work the sugar the central research question.
742 • Part VII: Race and Ethnicity

it, and it’s all just hate of color, that’s when

Immigrant Replenishment and my background comes to be more impor-
Intergroup Boundaries tant to me. . . . I have a friend. . . . And he
Mexican Americans in Garden City and Santa was picking on this guy that I didn’t know.
Maria exhibit significant socioeconomic, mari- And he was Mexican and they were bagging
tal, and residential assimilation. Canonical the- on him because he was Mexican and I’m just
ories of assimilation and ethnic identity would sitting there going, “Hey. I’m Mexican too.”
thus predict that they also experience ethnicity [He said,] “No, no, no, this doesn’t concern
as a symbolic and optional aspect of their iden- you. You’re cool. This guy is not.” And I’m
tity. Instead, Mexican Americans face rigid inter- just like, “Hey back up.” And I just totally got
group boundaries resulting from the presence of in his face because I was getting mad . . . they
a large immigrant population. were calling him a wetback and just totally
Because of their large numbers and high level dissing on him because he was Mexican.
of unauthorized status, Mexican immigrants are
the primary targets of anti-immigrant antipa- Like Ryan, respondents reported that they hear
thy, or what John Higham calls nativism: “an nativist comments that rely on a language that
intense opposition to an internal minority on invokes ethnicity, which sharpens the bound-
the ground of its foreign (i.e., ‘un-American’) aries that circumscribe all people of Mexican
connections” (1955, 4). Non-Mexicans voice descent.
nativist sentiments in anti-Mexican terms, tying Respondents reported hearing similar kinds
general antipathy about changes resulting from of tirades having to do with the frequent use
immigration to Mexicans in particular. Mexi- of Spanish language. As Marcela Muñoz, a
can Americans become aware of these nativist nineteen-year-old, third- and fourth-generation
expressions through interpersonal encounters college student in Garden City, who works as
and the more public and highly visible expres- a customer service agent at a local retail store,
sions of nativism that abound in each city in the recalled:
study. Indeed, Mexican immigrants are a prism
through which nativism refracts into the lives of [W]e have a Spanish recording. And a guest
Mexican Americans. called and she was asking about American
Nearly all respondents reported wit- flags. [I said,] “No Ma’am. We’re not sched-
nessing anti-Mexican nativism on the part uled to get any more until July. We’re sorry for
of non-Mexican friends, peers, coworkers, the inconvenience.” . . . But she just opened
and strangers. Typical are the experiences of her mouth and she was like, “Oh and by the
Ryan Bradley, a sixteen-year-old, third- and way, what is up with that Mexican crap?” Like
fourth-generation high school student in that. So I of course was like, “Ma’am, over
Santa Maria, who lives in a large house in the half of our community understands Spanish.”
upper-middle-class subsection of the city. He And she started going off on me. I was like
attends a private school, where he is one of a “Ma’am, I’m Mexican American.” And she
handful of middle-class, later-generation Mexi- didn’t know what to say! She just hung up.
can Americans. Like many respondents, Ryan’s
ethnic identity becomes important to him when Here too the nativist rant couches discon-
he come up against the rigid intergroup bound- tent about an issue tied to immigration
aries that nativism animates: (Spanish-language use) as a problem seemingly
related to “Mexicans.”
If there’s a threat that’s apparent on somebody Mexican Americans also become aware of
else who is of the same descent that I am, and pervasive nativism through more visible proc-
the other person is being totally racist about lamations. Established residents use public
 Jiménez—Why Replenishment Strengthens Racial and Ethnic Boundaries • 743

forums, such as speeches, demonstrations, and I know somewhere along down before me,
the opinion section of local newspapers, to ex- somebody in my family, I’m sure, has been
press nativist fears about the ways in which in that position. And although I’m not in it,
Mexican immigrants have changed each of the and probably never will be in that position, I
study cities and the country. These expressions just think that back when my ancestors were
most often center around the increasing use of in that position and people were the same way
Spanish, a perception that immigrants take ad- towards them.
vantage of misguided multicultural policies, and
a belief that immigrants are a drain on public Although many respondents, like Mike, have
resources. It is not the frequency of these public only a vague idea about their family’s immigrant
expressions, but rather their high visibility, that history, Mexican immigrants are an en vivo rep-
accounts for their power to harden intergroup resentation of their families’ historical struggles.
boundaries. Verbal attacks on Mexican immigrants thus
Mexican Americans internalize this nativism become an affront to all people of Mexican de-
as part of their own ethnic identity. Because scent, both foreign and native born.
immigration is a salient part of historical and
present-day experience, immigration and the
struggles of assimilation are central to Mexican Immigrant Replenishment, Intergroup
American ethnic identity. Nativist expressions Boundaries, and the Continuing
directed at Mexican immigrants thus activate Significance of Race
respondents’ identity as people of Mexican de- Race matters in the lives of Mexican Americans,
scent. The comments of Mike Fernandez, a and the large and continual influx of Mexican
nineteen-year-old, third- and fourth-generation immigrants refreshes its salience and imbues it
community college student in Santa Maria, illus- with meaning. The contentious historical rela-
trate how encounters with nativism play a role in tionship between Mexicans and Anglos in the
the formation of ethnic identity. Mike describes Midwest (García 1996) and West (Almaguer
his family as “a white family who is Mexican,” 1994; Camarillo 1996 [1979]) lurks in the
because Mexican traditions play only a small role background. But it is the large influx of Mexi-
in his family life. Yet immigration as a defining can immigrants that most significantly structures
feature of ethnic identity comes to the fore when how respondents experience race. In a context
he encounters the nativist expressions that other of heavy Mexican immigration, notions of race
respondents mention: are intimately tied to ancestry, nativity, and even
legal status.
[S]omebody will say something about Mexi- Respondents with darker skin report being
cans or something like that and it’s not said mistaken for immigrants by people who assume
towards me, it’s not directed towards me. that they are Mexican born based on their skin
But at that point, I’ll feel myself discrimi- color. But race has added meaning in a context
nated against. I’ll put the discrimination on of heavy unauthorized Mexican immigration.
myself, feeling that even though they’re not The unauthorized status of more than half of
directing it towards me, I can’t help but feel all Mexican immigrants, and political attention
that it’s degrading towards me in some way, given to the US-Mexico border, only tighten the
when in fact I know it’s not meant directly relationships among race, ancestry, nativity, and
towards me—it’s a general comment. But it even legal status. Take, for example, the expe-
just kind of makes me uncomfortable. . . . rience of Pedro Ramirez, a fifty-two-year-old,
Just because they’re speaking about a Mexican third-generation high school teacher. He recalled
family or a Mexican person and I know that, the especially troubling experience of being
though my family is not in that position, that pulled over by an immigration official while
744 • Part VII: Race and Ethnicity

driving his pickup truck after doing yard work at about authentic expressions of Mexican ethnic-
a rental property he owns: ity, giving rise to rigid intragroup boundaries
that run through the Mexican-origin population.
It’s this guy with a Smokey the Bear hat and Immigrants and second-generation individuals
wrap around glasses. It’s la migra. It’s the INS, regard Mexican Americans who fail to live up to
the border patrol! So I get out [of my car] and these expectations as “inauthentic” ethnics.
the guy says “¡vete aquí!” (“come here”) I go Many people of Mexican descent, including
oh no, and I’m laughing. I come over and say, those in my sample, do not speak Spanish (Rum-
“May I help you?” He says, “Do you speak baut, Massey, and Bean 2006). But immigrants
English?” I said, “What the hell do you think from Mexico and even young second-generation
I just said?” He says, “Do you have some ID?” individuals maintain Spanish-language use as
I go, “What the hell do you want to know if a central component of identity. Interviews
I have ID for? I wasn’t going past the speed abound with reports of experiences in which
limit. Besides you’re not a cop. You’re the Bor- immigrants or second-generation individuals
der Patrol. All right, I’ll play your game.” He called into question respondents’ authenticity
said, “Do you have some ID?” So I pull out because of their inability to speak the mother
my driver’s license and show him my wallet. tongue of their immigrant ancestors. Consider
“Do you have anything else?” I said, “Yeah.” the experience of Kyle Gil, a thirty-year-old,
And I showed him my social security card. fourth-generation auto body shop owner. Some
He wanted to reach for it and I go, “You ain’t Mexican immigrants who come into Kyle’s shop
getting this. Forget that!” He goes, “You have react negatively when they realize that he does
anything else?” I go, “Sure I do.” So I pull not speak Spanish: “[T]hey’ll come in and they’ll
out my American Express card. And it’s green. look at me [and say], ‘You speak Spanish?’ [I an-
I said, “Don’t leave home without it. This is swer,] ‘No, not really.’ [They say,] ‘You dumb or
harassment!” Guilt by association: Mexican what? How come you don’t speak Spanish?’ And
needing a haircut and a shave on a Friday af- it’s like I’m not good enough for them because I
ternoon with bandana around his neck, with can’t. So you get that reverse. It’s tough.”
an old pickup truck loaded with mowers and Respondents’ inability to satisfy the gate-
edgers and stuff like that. keepers of ethnic authenticity stems from the
fact that their parents and grandparents did
Even if respondents do not have dark skin, not transmit the Spanish language across gen-
they are not entirely immune to stereotypes erations. An ideology of Americanization that
about Mexicans as foreigners. Non-Mexicans forced earlier generations of Mexican Americans
frequently tag respondents who have a Spanish to speak only English, combined with the long
surname as immigrants. Surnames often serve length of time that their families have been in
as markers of ethnicity for all groups. They may the United States, means that young respon-
signal when someone has, for example, Italian, dents are ill-equipped to use Spanish to validate
Polish, or Irish ancestry. But when immigration their ethnic roots to others who expect them to
is replenished, surnames mark both ancestry and do so.
nativity. Immigrants and young second-generation
individuals also challenge Mexican Americans
about their styles of dress, tastes, and choice
Ethnic Expectations and Intragroup of friends. In the eyes of those who enforce
Boundaries the criteria for authenticity, Mexican ethnic-
Ethnic identities are not just assigned to groups ity and “mainstream” American culture are at
and individuals; they are also asserted by group odds. Having tastes and styles perceived to be
members. The heavy influx of immigrants to devoid of Mexican overtones fails to meet the
Garden City and Santa Maria informs ideas expectations about Mexican ethnicity that many
 Jiménez—Why Replenishment Strengthens Racial and Ethnic Boundaries • 745

immigrants and young second-generation indi- to poor, laboring foreign groups that tear at the
viduals impose. American economic and social fabric. They are
These intragroup boundaries are drawn down instead seen as American ethnics who have over-
generational lines, with later-generation Mexican come the hardships of assimilation to become
Americans falling on one side and those closer to fully woven into the American mainstream
the immigrant generation on the other. The end (Alba 1990).
result is that Mexican Americans cannot sym- Race played a central role in the assimila-
bolically or optionally assert their ethnic identity tion processes of white ethnics, animating the
without being challenged. boundaries between European immigrants and
the native-born “white” population (Higham
1955). But the experience of these groups fur-
Discussion and Conclusion ther suggests the link between race and immigra-
The duration of an immigrant wave has played tion. With the cessation of large-scale European
at most a marginal role in theories of assimila- immigration, the racial markers that once served
tion and ethnic identity formation. This chapter as cues about ancestry and nativity grew weaker
demonstrates how it matters in shaping the sa- in their association with the particular groups
lience of group boundaries. from which they originated, and they “became
As the experiences of Mexican Americans in white” (Roediger 2005). This weak association
Garden City, Kansas, and Santa Maria, Califor- among race, ancestry, and nativity contributed
nia, show, the role of ethnicity as a symbolic and to the later generations’ status as “white,” free-
optional aspect of identity is in part a function ing them from the racialized foreign status with
of immigrant replenishment. Although Mexican which their immigrant and second-generation
Americans exhibit significant signs of assimila- ancestors were all too familiar.
tion, continuous waves of immigration maintain Immigrant replenishment also determines the
the rigidity of group boundaries in their lives. extent to which ethnicity is an aspect of identity
Comparing the case of white ethnics, who that can be invoked optionally. European-origin
today experience virtually no immigrant re- ethnicity has thinned in salience to such an ex-
plenishment, with that of Mexican Americans tent that white ethnics require nothing more
further illustrates the importance of immigrant from each other than the claim that their ances-
replenishment. Even after recognizing differ- tors came from a particular homeland (Waters
ences between Mexican- and European-origin 1990). Without any replenishment of immi-
experiences, a central factor differentiating grants, standards for ethnic group authenticity
how later-generation Mexican Americans and are low, and white ethnics are free to assert their
later-generation white ethnics construct their ethnic identity optionally, without challenge,
ethnic identity lies in the extent of immigrant and without running into intragroup boundar-
replenishment. Both later-generation Mexican ies. The Mexican American experience suggests
Americans and white ethnics exhibit signs of that, had European immigration continued at
assimilation, as measured by socioeconomic levels equal to those around the turn of the last
advancement, intermarriage, and residential century, white ethnics might very well face more
mobility though to a lesser degree for Mexi- stringent criteria for group authenticity. It fol-
can American. Because large-scale European lows that the duration of immigrant flows is a
immigration attenuated, so too did the accom- central factor shaping ethnic identity formation
panying forms of nativism familiar to Euro- and one for which researchers must account to
pean immigrants in the past and to Mexican more fully understand ethnic and racial change.
immigrants today. Without immigrant replen-
ishment, later-generation descendants of these REFERENECES
European immigrants negotiate an American Alba, Richard D. 1990. Ethnic Identity: The Transfor-
mation of White America. New Haven, CT: Yale
society that no longer sees them as belonging
University Press.
746 • Part VII: Race and Ethnicity

Almaguer, Tomás. 1994. Racial Fault Lines: The His- Higham, John. 1955. Strangers in the Land: Patterns
torical Origins of White Supremacy in California. of American Nativism, 1860–1925. New Bruns-
Berkeley: University of California Press. wick, NJ: Rutgers University Press.
Avila, Henry J. 1997. “Immigration and Integration: Roediger, David R. 2005. Working Toward Whiteness:
The Mexican Americans Community in Garden How America’s Immigrants Become White; The
City, Kansas, 1900–1950.” Kansas History 20: Strange Journey from Ellis Island to the Suburbs.
22–37. New York: Basic Books.
Camarillo, Albert. 1996 [1979]. Chicanos in a Rumbaut, Rubén G., Douglas S. Massey, and Frank
Changing Society: From Mexican Pueblos to Amer- D. Bean. 2006. “Linguistic Life Expectancies:
ican Barrios in Santa Barbara and Southern Cal- Immigrant Language Retention in Southern
ifornia, 1848–1930. Cambridge, MA: Harvard California.” Population and Development Review
University Press. 32: 447–460.
Gans, Herbert J. 1979. “Symbolic Ethnicity: The Waters, Mary C. 1990. Ethnic Options: Choosing
Future of Ethnic Groups and Cultures in Amer- Identities in America. Berkeley: University of
ica.” Ethnic and Racial Studies 2: 1–20. California Press.
García, Juan R. 1996. Mexicans in the Midwest, Waters, Mary C., and Tomás R. Jiménez. 2005. “As-
1900–1932. Tucson: University of Arizona Press. sessing Immigrant Assimilation: New Empirical
Gordon, Milton M. 1964. Assimilation in American and Theoretical Challenges.” Annual Review of
Life: The Role of Race, Religion, and National Ori- Sociology 31: 105–102.
gins. New York: Oxford University Press.
 Bertrand and Mullainathan—Are Emily and Greg More Employable Than Lakisha and Jamal? • 747

Discrimination

86 • Marianne Bertrand and Sendhil Mullainathan

Are Emily and Greg More Employable

Than Lakisha and Jamal?
A Field Experiment on Labor Market Discrimination

Every measure of economic success reveals sig- different to employers. So any racial difference
nificant racial inequality in the U.S. labor mar- in labor market outcomes could just as easily be
ket. Compared to Whites, African-Americans attributed to differences that are observable to
are twice as likely to be unemployed and earn employers but unobservable to researchers.
nearly 25 percent less when they are employed To circumvent this difficulty, we conduct
(Council of Economic Advisers, 1998). This a field experiment that builds on the corre-
inequality has sparked a debate as to whether spondence testing methodology that has been
employers treat members of different races dif- primarily used in the past to study minority out-
ferentially. When faced with observably similar comes in the United Kingdom. We send resumes
African-American and White applicants, do they in response to help-wanted ads in Chicago and
favor the White one? Some argue yes, citing ei- Boston newspapers and measure callback for in-
ther employer prejudice or employer perception terview for each sent resume. We experimentally
that race signals lower productivity. Others argue manipulate perception of race via the name of
that differential treatment by race is a relic of the fictitious job applicant. We randomly as-
the past, eliminated by some combination of sign very White-sounding names (such as Emily
employer enlightenment, affirmative action pro- Walsh or Greg Baker) to half the resumes and
grams and the profit-maximization motive. In very African-American-sounding names (such
fact, many in this latter camp even feel that strin- as Lakisha Washington or Jamal Jones) to the
gent enforcement of affirmative action programs other half. Because we are also interested in how
has produced an environment of reverse dis- credentials affect the racial gap in callback, we
crimination. They would argue that faced with experimentally vary the quality of the resumes
identical candidates, employers might favor the used in response to a given ad. Higher-quality
African-American one. Data limitations make applicants have on average a little more labor
it difficult to empirically test these views. Since market experience and fewer holes in their em-
researchers possess far less data than employers ployment history; they are also more likely to
do, White and African-American workers that have an e-mail address, have completed some
appear similar to researchers may look very certification degree, possess foreign language

Marianne Bertrand and Sendhil Mullainathan. “Are Emily and Greg More Employable Than Lakisha and
Jamal? A Field Experiment on Labor Market Discrimination,” The American Economic Review (September
2004), pp. 991–992, 1006–1007, 1009–1013. Reprinted by permission of The American Economic Review.
748 • Part VII: Race and Ethnicity

skills, or have been awarded some honors. In different industries does not appear correlated
practice, we typically send four resumes in re- to Census-based measures of the racial gap in
sponse to each ad: two higher-quality and two wages. The same is true for the racial gap we
lower-quality ones. We randomly assign to one measure in different occupations. In fact, we find
of the higher- and one of the lower-quality re- that the racial gaps in callback are statistically
sumes an African-American-sounding name. In indistinguishable across all the occupation and
total, we respond to over 1,300 employment ads industry categories covered in the experiment.
in the sales, administrative support, clerical, and Federal contractors, who are thought to be more
customer services job categories and send nearly severely constrained by affirmative action laws,
5,000 resumes. The ads we respond to cover do not treat the African-American resumes
a large spectrum of job quality, from cashier more preferentially; neither do larger employers
work at retail establishments and clerical work or employers who explicitly state that they are
in a mail room, to office and sales management “Equal Opportunity Employers.” In Chicago,
positions. we find a slightly smaller racial gap when em-
We find large racial differences in callback ployers are located in more African-American
rates. Applicants with White names need to send neighborhoods.1 . . .
about 10 resumes to get one callback whereas ap-
plicants with African-American names need to
send about 15 resumes. This 50-percent gap in Interpretation
callback is statistically significant. A White name Three main sets of questions arise when interpret-
yields as many more callbacks as an additional ing the results above. First, does a higher callback
eight years of experience on a resume. Since ap- rate for White applicants imply that employers
plicants’ names are randomly assigned, this gap are discriminating against African-Americans?
can only be attributed to the name manipulation. Second, does our design only isolate the effect
Race also affects the reward to having a bet- of race or is the name manipulation conveying
ter resume. Whites with higher-quality resumes some other factors than race? Third, how do
receive nearly 30-percent more callbacks than our results relate to different models of racial
Whites with lower-quality resumes. On the discrimination?
other hand, having a higher-quality resume
has a smaller effect for African-Americans.
In other words, the gap between Whites and Interpreting Callback Rates
African-Americans widens with resume qual- Our results indicate that for two identical in-
ity. While one may have expected improved dividuals engaging in an identical job search,
credentials to alleviate employers’ fear that the one with an African-American name would
African-American applicants are deficient in receive fewer interviews. Does differential treat-
some unobservable skills, this is not the case in ment within our experiment imply that employ-
our data. ers are discriminating against African-Americans
The experiment also reveals several other (whether it is rational, prejudice-based, or
aspects of the differential treatment by race. other form of discrimination)? In other words,
First, since we randomly assign applicants’ could the lower callback rate we record for
postal addresses to the resumes, we can study African-American resumes within our experiment
the effect of neighborhood of residence on the be consistent with a racially neutral review of the
likelihood of callback. We find that living in a entire pool of resumes the surveyed employers
wealthier (or more educated or Whiter) neigh- receive?
borhood increases callback rates. But, interest- In a racially neutral review process, employers
ingly, African-Americans are not helped more would rank order resumes based on their quality
than Whites by living in a “better” neighbor- and call back all applicants that are above a cer-
hood. Second, the racial gap we measure in tain threshold. Because names are randomized,
 Bertrand and Mullainathan—Are Emily and Greg More Employable Than Lakisha and Jamal? • 749

the White and African-American resumes we of White applicants (as they are trying to fill in
send should rank similarly on average. So, ir- their hiring quota for Whites) and much more
respective of the skill and racial composition of sensitive to the quality of Black applicants (when
the applicant pool, a race-blind selection rule they have so many to pick from). Thus, it is un-
would generate equal treatment of Whites and likely that the differential treatment we observe
African-Americans. So our results must imply is generated by hiring rules such as these.
that employers use race as a factor when review-
ing resumes, which matches the legal definition
of discrimination. Potential Confounds
But even rules where employers are not trying While the names we have used in this experi-
to interview as few African-American applicants ment strongly signal racial origin, they may also
as possible may generate observed differential signal some other personal trait. More specifi-
treatment in our experiment. One such hiring cally, one might be concerned that employers
rule would be employers trying to interview a are inferring social background from the per-
target level of African-American candidates. sonal name. When employers read a name like
For example, perhaps the average firm in our “Tyrone” or “Latoya,” they may assume that the
experiment aims to produce an interview pool person comes from a disadvantaged background.
that matches the population base rate. This rule In the extreme form of this social background
could produce the observed differential treat- interpretation, employers do not care at all about
ment if the average firm receives a higher pro- race but are discriminating only against the so-
portion of African-American resumes than the cial background conveyed by the names we have
population base rate because African-Americans chosen.
disproportionately apply to the jobs and indus- While plausible, we feel that some of our
tries in our sample. earlier results are hard to reconcile with this
Some of our other findings may be consis- interpretation. For example, we found that
tent with such a rule. For example, the fact that while employers value “better” addresses,
“Equal Opportunity Employers” or federal con- African-Americans are not helped more than
tractors do not appear to discriminate any less Whites by living in Whiter or more educated
may reflect the fact that such employers receive neighborhoods. If the African-American names
more applications from African-Americans. On we have chosen mainly signal negative social
the other hand, other key findings run counter background, one might have expected the esti-
to this rule. As we discuss above, we find no mated name gap to be lower for better addresses.
systematic difference in the racial gap in call- Also, if the names mainly signal social back-
back across occupational or industry catego- ground, one might have expected the name gap
ries, despite the large variation in the fraction to be higher for jobs that rely more on soft skills
of African-Americans looking for work in those or require more interpersonal interactions. We
categories. African-Americans are underrepre- found no such evidence.
sented in managerial occupations, for example. There is one final potential confound to our
If employers matched base rates in the popula- results. Perhaps what appears as a bias against
tion, the few African-Americans who apply to African-Americans is actually the result of reverse
these jobs should receive a higher callback rate discrimination. If qualified African-Americans
than Whites. Yet, we find that the racial gap in are thought to be in high demand, then employ-
managerial occupations is the same as in all the ers with average quality jobs might feel that an
other job categories. This rule also runs counter equally talented African-American would never
to our findings on returns to skill. Suppose accept an offer from them and thereby never call
firms are struggling to find White applicants her or him in for an interview. Such an argu-
but overwhelmed with African-American ones. ment might also explain why African-Americans
Then they should be less sensitive to the quality do not receive as strong a return as Whites to
750 • Part VII: Race and Ethnicity

better resumes, since higher qualification only discriminate somewhat less. However, employer
strengthens this argument. But this interpreta- discrimination models would struggle to explain
tion would suggest that among the better jobs, why African-Americans get relatively lower re-
we ought to see evidence of reverse discrimina- turns to their credentials. Indeed, the cost of in-
tion, or at least a smaller racial gap. However, we dulging the discrimination taste should increase
do not find any such evidence. The racial gap as the minority applicants’ credentials increase.
does not vary across jobs with different skill re- Statistical discrimination models are the
quirements, nor does it vary across occupation prominent alternative to the taste-based mod-
categories. Even among the better jobs in our els in the economics literature. In one class of
sample, we find that employers significantly statistical discrimination models, employers
favor applicants with White names. use (observable) race to proxy for unobservable
skills (e.g., Edmund S. Phelps, 1972; Kenneth J.
Arrow, 1973). This class of models struggles to
Relation to Existing Theories explain the credentials effect as well. Indeed, the
What do these results imply for existing models added credentials should lead to a larger update
of discrimination? Economic theories of discrim- for African-Americans and hence greater returns
ination can be classified into two main categories: to skills for that group.
taste-based and statistical discrimination models. A second class of statistical discrimination
Both sets of models can obviously “explain” our models “emphasize[s] the precision of the in-
average racial gap in callbacks. But can these formation that employers have about individ-
models explain our other findings? More specif- ual productivity” (Altonji and Blank, 1999).
ically, we discuss the relevance of these models Specifically, in these models, employers believe
with a focus on two of the facts that have been that the same observable signal is more precise
uncovered in this [chapter]: (i) the lower returns for Whites than for African-Americans (Dennis
to credentials for African-Americans; (ii) the rel- J. Aigner and Glenn G. Cain, 1977; Shelly J.
ative uniformity of the race gap across occupa- Lundberg and Richard Startz, 1983; Bradford
tions, job requirements and, to a lesser extent, Cornell and Ivo Welch, 1996). Under such
employer characteristics and industries. models, African-Americans receive lower returns
Taste-based models (Gary S. Becker, 1961) to observable skills because employers place less
differ in whose prejudiced “tastes” they empha- weight on these skills. However, how reasonable
size: customers, coworkers, or employers. Cus- is this interpretation for our experiment? First,
tomer and co-worker discrimination models it is important to note that we are using the
seem at odds with the lack of significant varia- same set of resume characteristics for both racial
tion of the racial gap by occupation and industry groups. So the lower precision of information for
categories, as the amount of customer contact African-Americans cannot be that, for example,
and the fraction of White employees vary quite an employer does not know what a high school
a lot across these categories. We do not find a degree from a very African-American neighbor-
larger racial gap among jobs that explicitly re- hood means (as in Aigner and Cain, 1977). Sec-
quire “communication skills” and jobs for which ond, many of the credentials on the resumes are
we expect either customer or coworker contacts in fact externally and easily verifiable, such as a
to be higher (retail sales for example). certification for a specific software.
Because we do not know what drives employer An alternative version of these models would
tastes, employer discrimination models could be rely on bias in the observable signal rather than
consistent with the lack of occupation and in- differential variance or noise of these signals by
dustry variation. Employer discrimination also race. Perhaps the skills of African-Americans are
matches the finding that employers located in discounted because affirmative action makes it
more African-American neighborhoods appear to easier for African-Americans to get these skills.
 Bertrand and Mullainathan—Are Emily and Greg More Employable Than Lakisha and Jamal? • 751

While this is plausible for credentials such as an implications. They suggest that training pro-
employee-of-the-month honor, it is unclear why grams alone may not be enough to alleviate the
this would apply to more verifiable and harder racial gap in labor market outcomes. For training
skills. It is equally unclear why work experience to work, some general-equilibrium force outside
would be less rewarded since our study suggests the context of our experiment would have to be
that getting a job is more, not less, difficult for at play. In fact, if African-Americans recognize
African-Americans. how employers reward their skills, they may ra-
The uniformity of the racial gap across oc- tionally be less willing than Whites to even par-
cupations is also troubling for a statistical dis- ticipate in these programs.
crimination interpretation. Numerous factors
that should affect the level of statistical discrim- NOTES
ination, such as the importance of unobservable 1. For further details on this experiment, see Ber-
trand and Mullainathan (2004).
skills, the observability of qualifications, the pre-
cision of observable skills and the ease of perfor-
REFERENCES
mance measurement, may vary quite a lot across
Aigner, Dennis J. and Cain, Glenn G. “Statistical
occupations. Theories of Discrimination in Labor Markets.”
This discussion suggests that perhaps other Industrial and Labor Relations Review, January
models may do a better job at explaining our 1977, 30(1), pp. 175–87.
findings. One simple alternative model is lexico- Altonji, Joseph G. and Blank, Rebecca M. “Race and
graphic search by employers. Employers receive Gender in the Labor Market,” in Orley Ashen-
so many resumes that they may use quick heuris- felter and David Card, eds., Handbook of labor
tics in reading these resumes. One such heuristic economics, Vol. 30. Amsterdam: North-Holland,
1999, pp. 3143–259.
could be to simply read no further when they
Arrow, Kenneth J. “The Theory of Discrimina-
see an African-American name. Thus they may
tion,” in Orley Ashenfelter and Albert Rees, eds.,
never see the skills of African-American candi- Discrimination in labor markets. Princeton, NJ:
dates and this could explain why these skills are Princeton University Press, 1973, pp. 3–33.
not rewarded. This might also to some extent Becker, Gary S. The economics of discrimination, 2nd
explain the uniformity of the race gap since the Ed. Chicago: University of Chicago Press, 1961.
screening process (i.e., looking through a large Bertrand, Marianne and Sendhil Mullainathan. “Are
set of resumes) may be quite similar across the Emily and Greg More Employable Than Lakisha
variety of jobs we study. and Jamal? A Field Experiment on Labor Market
Discrimination.” American Economic Review 94
(September 2004): 991–1013.
Conclusion Cornell, Bradford and Welch, Ivo. “Culture, Infor-
mation, and Screening Discrimination.” Journal
This [chapter] suggests that African-Americans of Political Economy, June 1996, 104(3), pp.
face differential treatment when searching for 542–71.
jobs and this may still be a factor in why they Council of Economic Advisers. Changing America:
do poorly in the labor market. Job applicants Indicators of social and economic well-being by race
with African-American names get far fewer call- and Hispanic origin. September 1998, http://w3.
backs for each resume they send out. Equally access.gpo.gov/eop/ca/pdfs/ca.pdf.
importantly, applicants with African-American Lundberg, Shelly J. and Starz, Richard. “Private Dis-
crimination and Social Intervention in Compet-
names find it hard to overcome this hurdle in
itive Labor Market.” American Economic Review,
callbacks by improving their observable skills or June 1983, 73(3), pp. 340–47.
credentials. Phelps, Edmund S. “The Statistical Theory of Rac-
Taken at face value, our results on differential ism and Sexism.” American Economic Review,
returns to skill have possibly important policy September 1972, 62(4), pp. 659–61.
752 • Part VII: Race and Ethnicity

87 • Claude Steele

Stereotype Threat and African-American

Student Achievement

Over the past four decades African-American told recently depicts some of these. The story-
college students have been more in the spotlight teller was worried about his friend, a normally
than any other American students. This is be- energetic Black student who had broken up with
cause they aren’t just college students; they are his longtime girlfriend and had since learned
a cutting edge in America’s effort to integrate it- that she, a Hispanic, was now dating a white
self in the nearly forty years since the passage of student. This hit him hard. Not long after hear-
the Civil Rights Act. These students have borne ing about his girlfriend, he sat through an hour’s
much of the burden for our national experiment discussion of The Bell Curve in his psychology
in racial integration. And to a significant degree class, during which the possible genetic inferi-
the success of the experiment will be determined ority of his race was openly considered. Then he
by their success. overheard students at lunch arguing that affirma-
Nonetheless, throughout the 1990s the na- tive action allowed in too many underqualified
tional college dropout rate for African Ameri- Blacks. By his own account, this young man had
cans has been 20 to 25 percent higher than that experienced very little of what he thought of as
for whites. Among those who finish college, racial discrimination on campus. Still, these were
the grade point average of Black students is features of his world. Could they have a bearing
two-thirds of a grade below that of whites. . . . on his academic life?
Virtually all aspects of underperformance— My colleagues and I have called such fea-
lower standardized test scores, lower college tures “stereotype threat”—the threat of being
grades, lower graduation rates—persist among viewed through the lens of a negative stereotype,
students from the African-American middle or the fear of doing something that would in-
class. This situation forces on us an uncomfort- advertently confirm that stereotype. Everyone
able recognition: that beyond class, something experiences stereotype threat. We are all mem-
racial is depressing the academic performance of bers of some group about which negative stereo-
these students. types exist, from white males and Methodists to
Some time ago two of my colleagues, Joshua women and the elderly. And in a situation where
Aronson and Steven Spencer, and I tried to see the one of those stereotypes applies—a man talking
world from the standpoint of African-American to women about pay equity, for example, or an
students, concerning ourselves less with features aging faculty member trying to remember a
of theirs that might explain their troubles than number sequence in the middle of a lecture—we
with features of the world they see. A story I was know that we may be judged by it.

Claude Steele. “Stereotype Threat and African-American Student Achievement,” in Young, Gifted, and Black:
Promoting High Achievement Among African-American Students, Theresa Perry, Claude Steele, and Asa Hilliard
III, pp. 109–116, 120–123, 172–174, 176–177. Copyright © 2003 by Theresa Perry, Asa Hilliard III, and
Claude Steele. Reprinted by permission of Beacon Press, Boston.
 Steele—Stereotype Threat and African-American Student Achievement • 753

Like the young man in the story, we can feel meant that the test—designed for graduating se-
mistrustful and apprehensive in such situations. niors—was particularly hard for them—precisely
For him, as for African-American students gen- the feature, we reasoned, that would make this
erally, negative stereotypes apply in many situ- simple testing situation different for our Black
ations, even personal ones. Why was that old participants than for our white participants,
roommate unfriendly to him? Did that young despite the fact that all the participants were of
white woman who has been so nice to him in equal ability levels measured by all available cri-
class not return his phone call because she’s teria. (The difficulty of the test guaranteed that
afraid he’ll ask her for a date? Is it because of both Black and white students would find the
his race or something else about him? He can- test frustrating. And it is in these situations that
not know the answers, but neither can his ra- members of ability-stereotyped groups are most
tional self fully dismiss the questions. Together likely to experience the extra burden of stereo-
they raise a deeper question: Will his race be a type threat. First, the experience of frustration
boundary to his experience, to his emotions, to with the test gives credibility to the limitation
his relationships? . . . alleged in the stereotype. For this reason, frus-
tration can be especially stinging and disruptive
for test-takers to whom the stereotype is relevant.
Measuring Stereotype Threat Second, it is on a demanding test that one can
Can stereotype threat be shown to affect aca- least afford to be bothered by the thoughts that
demic performance? And if so, who would be likely accompany stereotype threat.)
most affected—stronger or weaker students? A significant part of the negative stereotype
Which has a greater influence on academic suc- about African Americans concerns intellectual
cess among Black college students—the degree ability. Thus, in the stereotype threat condi-
of threat or the level of preparation with which tions of the experiments in this series, we merely
they enter college? mentioned to participants that the test was a
As we confronted these questions in the measure of verbal ability. This was enough, we
course of our research, we came in for some sur- felt, to make the negative stereotype about Af-
prises. We began with what we took to be the rican Americans’ abilities relevant to their per-
hardest question: Could something as abstract formance on the test, and thus to put them at
as stereotype threat really affect something as risk of confirming, or being seen to confirm,
irrepressible as intelligence? Ours is an individ- the negative stereotype about their abilities. If
ualistic culture; forward movement is seen to the pressure imposed by the relevance of a neg-
come from within. Against this cultural faith ative stereotype about one’s group is enough to
one needs evidence to argue that something as impair an important intellectual performance,
“sociological” as stereotype threat can repress then Black participants should perform worse
something as “individualistic” as intelligence. than whites in the “diagnostic” condition of
To acquire such evidence, Joshua Aronson this experiment but not in the “nondiagnostic”
and I (following a procedure developed with condition. As Figure 87.1 depicts, this is pre-
Steven Spencer) designed an experiment to test cisely what happened: Blacks performed a full
whether the stereotype threat that Black students standard deviation lower than whites under the
might experience when taking a difficult stan- stereotype threat of the test being “diagnostic” of
dardized test could depress their performance their intellectual ability, even though we had sta-
on the test to a statistically reliable degree. We tistically matched the two groups in ability level.
brought white and Black Stanford students into Something other than ability was involved; we
the laboratory and gave them, one at a time, a believed it was stereotype threat.
very difficult thirty-minute section of a Graduate But maybe the Black students performed
Record Exam subject test in English literature. less well than the white students because they
Most of these students were sophomores, which were less motivated, or because their skills were
754 • Part VII: Race and Ethnicity

somehow less applicable to the advanced ma- Figure 87.1. White and Black Participants’ Score
terial of this test. We needed some way to de- (Controlled for SAT) on a Difficult English Test as a
termine if it was indeed stereotype threat that Function of Characterization of the Test
depressed the Black students’ scores. We rea- 16
soned that if stereotype threat had impaired
their performance on the test, then reducing 14

this threat would allow their performance to im-

12
prove. We presented the same test as a laboratory

Mean Items Solved

task that was used to study how certain problems 10
are generally solved. We stressed that the task did
not measure a person’s level of intellectual abil- 8

ity. A simple instruction, yes, but it profoundly

6

White Participants

White Participants

Black Participants
changed the meaning of the situation. In one
stroke “spotlight anxiety,” as the psychologist 4

Participants
William Cross once called it, was turned off—
and the Black students’ performance on the test 2

Black
rose to match that of equally qualified whites
0
(see Figure 87.1). In the nonstereotype threat Diagnostic Nondiagnostic
conditions, we presented the same test as an in- Test Characterization

strument for studying problem solving that was

“nondiagnostic” of individual differences in abil- students mindful of stereotypes about their race,
ity—thus making the racial stereotype irrelevant these students should complete more fragments
to their performance. with stereotype-related words. That is just what
Aronson and I decided that what we needed happened. When Black students were told
next was direct evidence of the subjective state that the test would measure ability, they com-
we call stereotype threat. To seek this, we looked pleted the fragments with significantly more
into whether simply sitting down to take a dif- stereotype-related words than when they were
ficult test of ability was enough to make Black told that it was not a measure of ability. Whites
students mindful of their race and stereotypes made few stereotype-related completions in ei-
about it. This may seem unlikely. White students ther case. . . .
I have taught over the years have sometimes said
that they have hardly any sense of even having
a race. But Blacks have many experiences with How Stereotype Threat Affects
the majority “other group” that make their race People Differently
salient to them. Is everyone equally threatened and disrupted by a
We again brought Black and white students stereotype? One might expect, for example, that
in to take a difficult verbal test. But just be- it would affect the weakest students most. But in
fore the test began, we gave them a long list of all our research the most achievement-oriented
words, each of which had two letters missing. students, who were also the most skilled, mo-
They were told to complete the words on this tivated, and confident, were the most impaired
list as fast as they could. We knew from a pre- by stereotype threat. This fact had been under
liminary survey that twelve of the eighty words our noses all along—in our data and even in our
we had selected could be completed in such a theory. A person has to care about a domain in
way as to relate to the stereotype about Blacks’ order to be disturbed by the prospect of being
intellectual ability. The fragment “—ce,” for ex- stereotyped in it. That is the whole idea of dis-
ample, could become “race.” If simply taking a identification—protecting against stereotype
difficult test of ability was enough to make Black threat by ceasing to care about the domain in
 Steele—Stereotype Threat and African-American Student Achievement • 755

which the stereotype applies. Our earlier exper- That, however, seems to be precisely what
iments had selected Black students who identi- these students are trying to do. In some of our
fied with verbal skills and women who identified experiments we administered the test of ability
with math. But when we tested participants who by computer, so that we could see how long
identified less with these domains, what had participants spent looking at different parts of
been under our noses hit us in the face. None the test questions. Black students taking the
of them showed any effect of stereotype threat test under stereotype threat seemed to be trying
whatsoever. too hard rather than not hard enough. They re-
These weakly identified students did not per- read the questions, reread the multiple choices,
form well on the test: once they discovered its rechecked their answers, more than when they
difficulty, they stopped trying very hard and got were not under stereotype threat. The threat
a low score. But their performance did not differ made them inefficient on a test that, like most
depending on whether they felt they were at risk standardized tests, is set up so that thinking long
of being judged stereotypically. often means thinking wrong, especially on diffi-
This finding, I believe, tells us two impor- cult items like the ones we used. . . .
tant things. The first is that the poorer college In the old song about the “steel-drivin’ man,”
performance of Black students may have an- John Henry races the new steam-driven drill to
other source in addition to the one—lack of see who can dig a hole faster. When the race is
good preparation and, perhaps, of identification over, John Henry has prevailed by digging the
with school achievement—that is commonly deeper hole—only to drop dead. The social psy-
understood. This additional source—the threat chologist Sherman James uses the term “John
of being negatively stereotyped in the envi- Henryism” to describe a psychological syn-
ronment—has not been well understood. The drome that he found to be associated with hy-
distinction has important policy implications: pertension in several samples of North Carolina
different kinds of students may require different Blacks: holding too rigidly to the faith that dis-
pedagogies of improvement. crimination and disadvantage can be overcome
The second thing is poignant: what exposes with hard work and persistence. Certainly this is
students to the pressure of stereotype threat is the right attitude. But taken to extremes, it can
not weaker academic identity and skills but backfire. A deterioration of performance under
stronger academic identity and skills. They may stereotype threat by the skilled, confident Black
have long seen themselves as good students— students in our experiments may be rooted in
better than most. But led into the domain by John Henryism.
their strengths, they pay an extra tax on their This last point can be disheartening. Our re-
investment—vigilant worry that their future search, however, offers an interesting suggestion
will be compromised by society’s perception and about what can be done to overcome stereotype
treatment of their group. threat and its detrimental effects. The success
This tax has a long tradition in the Black of Black students may depend less on expecta-
community. The Jackie Robinson story is a cen- tions and motivation—things that are thought
tral narrative of Black life, literature, and jour- to drive academic performance—than on trust
nalism. Ebony Magazine has run a page for fifty that stereotypes about their group will not have
years featuring people who have broken down a limiting effect in their school world.
one or another racial barrier. Surely the academic Putting this idea to the test, Joseph Brown
vanguard among Black college students today and I asked, How can the usual detrimental ef-
knows this tradition—and knows, therefore, fect of stereotype threat on the standardized-test
that the thing to do, as my father told me, is to performance of these students be reduced? By
buckle down, pay whatever tax is required, and strengthening students’ expectations and con-
disprove the damn stereotype. fidence, or by strengthening their trust that
756 • Part VII: Race and Ethnicity

they are not at risk of being judged on the basis REFERENCES

of stereotypes? In the ensuing experiment we Aronson, J. 1997. The Effects of Conceptions of Ability
strengthened or weakened participants’ con- on Task Valuation. Unpublished manuscript, New
York University.
fidence in their verbal skills, by arranging for
———. 1999. The Effects of Conceiving Ability as
them to have either an impressive success or
Fixed or Improvable on Responses to Stereotype
an impressive failure on a test of verbal skills, Threat. Unpublished manuscript, New York
just before they took the same difficult verbal University.
test we had used in our earlier research. When Brown, J. L., and C. M. Steele. 2001. Performance
the second test was presented as a test of abil- Expectations Are Not a Necessary Mediator of Ste-
ity, the boosting or weakening of confidence in reotype Threat in African American Verbal Test
their verbal skills had no effect on performance: Performance. Unpublished manuscript, Stanford
Black participants performed less well than University.
equally skilled white participants. What does Cross, W. E., Jr. 1991. Shades of Black: Diversity in
African-American Identity. Philadelphia: Temple
this say about the commonsense idea that black
University Press.
students’ academic problems are rooted in lack Spencer, S. J., E. Iserman, P. G. Davies, and D. M.
of self-confidence? Quinn. 2001. Suppression of Doubts, Anxiety, and
What did raise the level of black students’ per- Stereotypes as a Mediator of the Effect of Stereotype
formance to that of equally qualified whites was Threat on Women’s Math Performance. Unpub-
reducing stereotype threat—in this case by ex- lished manuscript, University of Waterloo.
plicitly presenting the test as racially fair. When Spencer, S. J., C. M. Steele, and D. M. Quinn. 1999.
this was done, Blacks performed at the same “Stereotype Threat and Women’s Math Perfor-
high level as whites even if their self-confidence mance.” Journal of Experimental Social Psychology
35: 4–28.
had been weakened by a prior failure.
Steele, C. M. 1975. “Name-Calling and Compli-
These results suggest something that I think
ance.” Journal of Personality and Social Psychology
has not been made clear elsewhere: when strong 31: 361–69.
Black students sit down to take a difficult stan- ———. 1992. “Race and the Schooling of Black
dardized test, the extra apprehension they feel Americans.” Atlantic Monthly 269: 68–78.
in comparison with whites is less about their ———. 1997. “A Threat in the Air: How Stereo-
own ability than it is about having to perform types Shape Intellectual Identity and Perfor-
on a test and in a situation that may be primed mance.” American Psychologist 52: 613–29.
to treat them stereotypically. We discovered the ———. 1999. “Thin Ice: Stereotype Threat and
Black College Students.” Atlantic Monthly 248:
extent of this apprehension when we tried to
44–54.
develop procedures that would make our Black
Steele, C. M., and J. Aronson. 1995. “Stereotype
participants see the test as “race-fair.” It wasn’t Threat and the Intellectual Test Performance of
easy. African-Americans have endured so much African Americans.” Journal of Personality and So-
bad press about test scores for so long that, in cial Psychology 69: 797–811.
our experience, they are instinctively wary about Steele, C. M., S. J. Spencer, and J. Aronson. 2002.
the tests’ fairness. We were able to convince “Contending with Group Image: The Psychology
them that our test was race-fair only when we of Stereotype and Social Identity Threat.” In Ad-
implied that the research generating the test had vances in Experimental Social Psychology, vol. 34,
ed. M. Zanna. Academic Press.
been done by Blacks. When they felt trust, they
Steele, C. M., S. J. Spencer, P. G. Davies, K. Harber,
performed well regardless of whether we had
and R. E. Nisbett. 2001. African American Col-
weakened their self-confidence beforehand. And lege Achievement: A “Wise” Intervention. Unpub-
when they didn’t feel trust, no amount of bol- lished manuscript, Stanford University.
stering of self-confidence helped. Steele, S. 1990. The Content of Our Character: A New
Vision of Race in America. New York: St. Martin’s.
 Pager—Marked • 757

88 • Devah Pager

Marked
Race, Crime, and Finding Work in an Era of Mass Incarceration

While stratification researchers typically focus pairs, it becomes possible to assess the ways in
on schools, labor markets, and the family as which the effects of race and criminal record
primary institutions affecting inequality, a new interact to produce new forms of labor market
institution has emerged as central to the sorting inequalities.
and stratifying of young disadvantaged men: the
criminal justice system. With over two million
individuals currently incarcerated, and well over Mass Incarceration and the
half a million prisoners released each year, the Credentialing of Stigma
large and growing numbers of men being pro- Over the past three decades, the number of
cessed through the criminal justice system raises prison inmates has increased by more than 700
important questions about the consequences of percent, leaving the United States the country
this massive institutional intervention. Further, with the highest incarceration rate in the world
large racial disparities in incarceration lead us (Bureau of Justice Statistics, 2006; Mauer, 1999).
to question the degree to which mass incarcera- During this time, incarceration changed from
tion has particular implications for stratification a punishment reserved primarily for the most
along racial lines. heinous offenders to one extended to a much
This [chapter] focuses on the consequences greater range of crimes and a much larger seg-
of incarceration for the employment outcomes ment of the population. Recent trends in crime
of black and white men. I adopt an experimen- policy have led to the imposition of harsher sen-
tal audit approach to formally test the degree to tences for a wider range of offenses, thus casting
which race and criminal background affect sub- an ever widening net of penal intervention.
sequent employment opportunities. By using For each individual processed through the
matched pairs of individuals to apply for real criminal justice system, police records, court
entry-level jobs, it becomes possible to directly documents, and corrections databases detail
measure the extent to which contact with the dates of arrest, charges, conviction, and terms
criminal justice system—in the absence of other of incarceration. Most states make these re-
disqualifying characteristics—serves as a barrier cords publicly available, often through online
to employment among equally qualified appli- repositories, accessible to employers, landlords,
cants. Further, by varying the race of the tester creditors, and other interested parties. With

The ideas, issues, and theories considered in this brief piece are examined in greater depth in the following
publication: Devah Pager, “The Mark of a Criminal Record,” American Journal of Sociology 108 (2003), pp.
937–975, published by the University of Chicago Press. The article as printed here was originally prepared by
Devah Pager for Social Stratification: Class, Race, and Gender in Sociological Perspective, Third Edition, edited by
David B. Grusky, pp. 683–690. Copyright © 2008 by Westview Press.
758 • Part VII: Race and Ethnicity

increasing numbers of occupations, public ser- consistently rate blacks as more prone to vio-
vices, and other social goods becoming off-limits lence than any other American racial or ethnic
to ex-offenders, these records can be used as the group, with the stereotype of aggressiveness and
official basis for eligibility determination or ex- violence most frequently endorsed in ratings
clusion. The state, in this way, serves as a creden- of African Americans (Sniderman and Piazza,
tialing institution, providing official and public 1993; Smith, 1991). The stereotype of blacks as
certification of those among us who have been criminals is deeply embedded in the collective
convicted of wrongdoing. The “credential” of a consciousness of white Americans, irrespective
criminal record, like educational or professional of the perceiver’s level of prejudice or personal
credentials, constitutes a formal and enduring beliefs (Devine and Elliot, 1995; Eberhardt et
classification of social status, which can be used al., 2004; Graham and Lowery, 2004.
to regulate access and opportunity across numer- While it would be impossible to trace the
ous social, economic, and political domains. As source of contemporary racial stereotypes to any
increasing numbers of young men are marked one factor, the disproportionate growth of crim-
by their contact with the criminal justice system, inal justice intervention in the lives of young
it becomes a critical priority to understand the black men—and corresponding media coverage
costs and consequences of this now prevalent of this phenomenon, which presents an even
form of negative credential (Pager, 2007a). more skewed representation—has likely played
an important role. Experimental research shows
that exposure to news coverage of a violent inci-
Racial Stereotypes in an Era of Mass dent committed by a black perpetrator not only
Incarceration increases punitive attitudes about crime but
The expansion of the correctional population has further increases negative attitudes about blacks
been particularly consequential for blacks. Blacks generally (Gilliam and Iyengar, 2000). The more
today represent over 40 percent of current prison exposure we have to images of blacks in custody
inmates relative to just 12 percent of the U.S. or behind bars, the stronger our expectations be-
population. Over the course of a lifetime, nearly come regarding the race of assailants or the crim-
one in three young black men—and well over inal tendencies of black strangers (Cole, 1995).
half of young black high school dropouts—will The consequences of mass incarceration then
spend some time in prison. According to these may in fact extend far beyond the costs to the
estimates, young black men are more likely to individual bodies behind bars, and to the fami-
go to prison than to attend college, serve in the lies that are disrupted or the communities whose
military, or, in the case of high school dropouts, residents cycle in and out. The criminal justice
to be in the labor market (Bureau of Justice Sta- system may itself legitimate and reinforce deeply
tistics, 1997; Pettit and Western, 2004). Prison embedded racial stereotypes, contributing to the
is no longer a rare or extreme event among our persistent chasm in this society between black
nation’s most marginalized groups. Rather it has and white.
now become a normal and anticipated marker in
the transition to adulthood.
There is reason to believe that the implica- Assessing the Impact of Race and
tions of these trends extend well beyond the Criminal Background
prison walls, with assumptions of criminality In considering the labor market impacts of race
among blacks generalizing beyond those directly and criminal background, questions of causal-
engaged in crime. Blacks in this country have ity loom large. For example, while employment
long been regarded with suspicion and fear; but disparities between blacks and whites have been
unlike progressive trends in other racial atti- well established, the causes of these disparities
tudes, associations between race and crime have remain widely contested. Where racial discrimi-
changed little in recent years. Survey respondents nation may represent one explanation, a growing
 Pager—Marked • 759

number of researchers have pointed instead to characteristics. The current study included four
individual-level factors such as skill deficits and male testers, two blacks and two whites, matched
other human capital characteristics as a key into two teams—the two black testers formed
source of racial wage differentials (Neal and John- one team, and the two white testers formed a
son, 1996; Farkas and Vicknair, 1996). Likewise, second team.1 The testers were college students
assessing the impact of incarceration is not al- from Milwaukee who were matched on the basis
together straightforward. On the one hand, it’s of age, race, physical appearance, and general
not hard to imagine that a prison record would style of self-presentation. The testers were as-
carry a weighty stigma, with members of the signed fictitious resumes that reflected equiva-
general public (employers included) reluctant to lent levels of education and work experience.2
associate or work with former inmates. On the In addition, within each team, one auditor was
other hand, criminal offenders aren’t typically randomly assigned a “criminal record” for the
the image of the model employee. It’s certainly first week; the pair then rotated which member
possible that the poor employment outcomes of presented himself as the ex-offender for each suc-
ex-offenders stem rather from characteristics of cessive week of employment searches, such that
the offenders themselves, as opposed to any con- each tester served in the criminal record condi-
sequence of their criminal conviction. Poor work tion for an equal number of cases.3 By varying
habits, substance abuse problems, or deficient which member of the pair presented himself as
interpersonal skills may be sufficient to explain having a criminal record, unobserved differences
the employment disadvantages of ex-offenders, within the pairs of applicants were effectively
and yet these characteristics are difficult to cap- controlled.
ture using standard data sources. Before initiating the fieldwork, testers par-
Given the difficulties inherent to evaluating ticipated in a common training program to be-
the impact of race and imprisonment through come familiar with the details of their assumed
conventional measures, I set out to investigate profile and to ensure uniform behavior in job
this issue by constructing an experiment. I interviews. The training period lasted for one
wanted to bracket the range of personal charac- week, during which testers participated in mock
teristics associated with African Americans and interviews with one another and practice inter-
ex-offenders in order to home in on the causal views with cooperating employers. The testers
impact of race and a criminal record. The experi- were trained to respond to common inter-
mental audit methodology allowed me to do just view questions in standardized ways, and were
that. Using this approach, I pose the questions: well-rehearsed for a wide range of scenarios that
Given two equally qualified job applicants, how emerge in hiring situations. Frequent communi-
much does a criminal record affect the chances cation between myself and the testers through-
of being selected by an employer? To what ex- out each day of fieldwork allowed for regular
tent does race condition employers’ responses? supervision and troubleshooting in the event of
In answering these questions, we can begin to unexpected occurrences.
understand the ways in which race and criminal A random sample of entry-level positions
background shape and constrain important eco- (those jobs requiring no previous experience and
nomic opportunities. no education beyond high school) was drawn
each week from the Sunday classified advertise-
ment section of the Milwaukee Journal Sentinel.
The Audit Methodology The most common job titles included waitstaff,
The basic design of an employment audit in- laborers and warehouse, production/operators,
volves sending matched pairs of individuals customer service, sales, delivery drivers, and ca-
(called testers) to apply for real job openings shiers; a handful of clerical and managerial po-
in order to see whether employers respond dif- sitions were also included. I excluded from the
ferently to applicants on the basis of selected sample those occupations with legal restrictions
760 • Part VII: Race and Ethnicity

on ex-offenders, such as jobs in the health care criminal record, an arguably safer choice. Equat-
industry, work with children and the elderly, jobs ing the two applicants could in fact magnify the
requiring the handling of firearms (e.g., security impact of the criminal record, as it becomes the
guards), and jobs in the public sector. Of course, only remaining basis for selection between the
any true estimate of the collateral consequences two (Heckman, 1998). The call-back, by con-
of incarceration would also need to take account trast, does not present such complications. Typ-
of the wide range of employment fully off-limits ically employers interview multiple candidates
to individuals with prior felony convictions. for entry-level positions before selecting a hire.
Each of the audit pairs was randomly as- In fact, in a subsequent survey, employers in this
signed fifteen job openings each week. The white study reported interviewing an average of eight
pair and the black pair were assigned separate applicants for the last entry-level position filled.
sets of jobs, with the same-race testers apply- At the call-back stage, then, employers need not
ing to the same jobs. One member of the pair yet choose between the ex-offender and nonof-
applied first, with the second applying one day fender. If the applicants appear well-qualified,
later (randomly varying whether the ex-offender and if the employer does not view the criminal
was first or second). A total of 350 employers record as an automatic disqualifier, s/he can in-
were audited during the course of this study: 150 terview them both.4
by the white pair and 200 by the black pair. Ad-
ditional tests were performed by the black pair
because black testers received fewer callbacks on Hiring Outcomes by Race and
average, and there were thus fewer data points Criminal Background
with which to draw comparisons. A larger sam- Results are based on the proportion of appli-
ple size enables the calculation of more precise cations submitted by each tester which elicited
estimates of the effects under investigation (for callbacks from employers. Three main findings
a more in-depth discussion of the study design, should be noted from the audit results, presented
see Pager, 2003). in Figure 88.1. First, there is a large and signif-
This study focused only on the first stage of icant effect of a criminal record for all job seek-
the employment process, as this is the stage most ers. Among whites, for example, 34 percent of
likely to be affected by the barriers of race and those without criminal records received callbacks
criminal record. Testers visited employers, filled relative to only 17 percent of those with crimi-
out applications, and proceeded as far as they nal records. A criminal record is thus associated
could during the course of one visit. If testers with a 50 percent reduction in the likelihood of
were asked to interview on the spot, they did a callback among whites. Often testers reported
so, but they did not return to the employer for seeing employers’ levels of responsiveness change
a second visit. The primary dependent variable dramatically once they had glanced down at the
then is the proportion of applications that elic- criminal record question on the application
ited callbacks from employers. Individual voice form. Many employers seem to use the infor-
mail boxes were set up for each tester to record mation as a screening mechanism, weeding out
employer responses. applicants at the initial stage of review.
An advantage of the call-back as our key out- Second, there is some indication that the
come variable—as opposed to a job offer—is magnitude of the criminal record effect may be
that it does not require employers to narrow their even larger for blacks. While the interaction be-
selection down to a single applicant. At the job tween race and criminal record is not statistically
offer stage, if presented with an ex-offender and significant, the substantive difference is worth
an equally qualified nonoffender, even employ- noting. The ratio of callbacks for nonoffenders
ers with little concern over hiring ex-offenders relative to offenders for whites was two to one
would likely select the applicant with no (34 percent vs. 17 percent), while this same ratio
 Pager—Marked • 761

Figure 88.1. The Effects of Race and Criminal Background on Employment

for blacks is close to three to one (14 percent vs. with a troubling reality: Being black in America
5 percent).5 The estimated effect of a criminal today is just about the same as having a felony
record is thus 40 percent larger for blacks than conviction in terms of one’s chances of finding
for whites. The combination of minority status a job.6
and criminal background appears to intensify The results of this research suggest that
employers’ negative reactions, leaving few em- both race and criminal background represent
ployment prospects for black ex-offenders (200 extremely powerful barriers to job entry. The
applications resulted in only 10 callbacks). matched design allows us to separate specula-
Finally, looking at the callback rates for black tion about applicant qualifications (supply-side
and white tester pairs side by side, the funda- influences) from the preferences or biases of
mental importance of race becomes vividly employers (demand-side influences). While this
clear. Among those without criminal records, study remains silent on the many supply-side
black applicants were less than half as likely to factors that may also contribute to the employ-
receive callbacks compared to equally qualified ment difficulties of blacks and ex-offenders, it
whites (14 percent vs. 34 percent). This implies speaks loud and clear about the significance of
that young black men needed to work more than employer demand in shaping the opportunities
twice as hard (apply to twice as many jobs) to available to job seekers on the basis of their race
secure the same opportunities as whites with or criminal background. Before applicants have
identical qualifications. Even more striking, the an opportunity to demonstrate their capabilities
powerful effects of race rival even the strong in person, a large proportion are weeded out on
stigma conveyed by a criminal record. In this the basis of a single categorical distinction.
study, a white applicant with a criminal record
was just as likely, if not more, to receive a call-
back as a black applicant with no criminal his- Discussion
tory (17 percent vs. 14 percent). Despite the fact There is serious disagreement among academics,
that the white applicant revealed evidence of a policy makers, and practitioners over the extent
felony drug conviction, and despite the fact that to which contact with the criminal justice sys-
he reported having only recently returned from tem—in itself—leads to harmful consequences
a year and a half in prison, employers seemed for employment. The present study takes a strong
to view this applicant as no more risky than a stand in this debate by offering direct evidence of
young black man with no history of criminal the causal relationship between a criminal record
involvement. Racial disparities have been docu- and employment outcomes. Using matched pairs
mented in many contexts, but here, comparing and an experimentally assigned criminal record,
the two effects side by side, we are confronted this estimate is unaffected by the problems of
762 • Part VII: Race and Ethnicity

selection which plague observational data. While because blacks are so strongly associated with the
certainly there are additional ways in which in- population under correctional supervision, it be-
carceration may affect economic outcomes, this comes common to assume that any given young
study provides convincing evidence that mere black man is likely to have—or to be on his way
contact with the criminal justice system, in the to acquiring—a criminal record.
absence of any transformative or selective ef- At this point in history, it is impossible to
fects, severely limits subsequent employment tell whether the massive presence of incarcera-
opportunities. And while the audit study in- tion in today’s stratification system represents a
vestigates employment barriers to ex-offenders unique anomaly of the late twentieth century or
from a micro-perspective, the implications are part of a larger movement toward a system of
far-reaching. The finding that ex-offenders are stratification based on the official certification of
one-half to one-third as likely to be considered individual character and competence. Whether
by employers suggests that a criminal record in- this process will continue to form the basis of
deed presents a major barrier to employment. emerging social cleavages remains to be seen.
With over two million people currently behind
bars and over 12 million people with prior fel- NOTES
ony convictions, the consequences for labor mar- 1. The two teams were sent to different employ-
ers in order to minimize suspicion that might oth-
ket inequalities are potentially profound.
erwise arise from two similar applicants with similar
Second, the persistent effect of race on em-
criminal backgrounds applying within a short time
ployment opportunities is painfully clear in interval. In the present design, the criminal record
these results. Blacks are less than half as likely effect is estimated as a within-team effect while race
to receive consideration by employers relative is measured as a between-team effect. While the lat-
to their white counterparts, and black nonof- ter is less efficient, the comparison should be unbi-
fenders fall behind even whites with prior felony ased. Black and white testers used identical resumes
convictions. While this research cannot identify (something that would not have been possible had
the precise source of employers’ reluctance to they visited the same employers) and were selected
hire blacks, it is important to consider the in- and trained to present comparable profiles to em-
ployers. Likewise, employers were randomly assigned
fluence of high incarceration rates among blacks
across tester pairs, minimizing heterogeneity in em-
as one possible source of racial bias. Indeed, the ployer characteristics by team.
available evidence points to the pervasiveness of 2. Testers presented themselves as high school
images associating blacks with crime, and the graduates with steady work experience in entry-level
power of these images to strengthen negative jobs.
feelings about blacks as a group. It may be the 3. The criminal record in all cases was a drug fel-
case, then, that increasing rates of incarceration ony, possession with intent to distribute (cocaine),
among blacks, and the disproportionate coverage and 18 months of prison time. Testers presented
of criminal justice contact among blacks in the the information to employers by checking the box
“yes” in answer to the standard application question,
media, heighten negative reactions toward Af-
“Have you ever been convicted of a crime?” As addi-
rican Americans generally, irrespective of their
tional cues, testers also reported work experience in
personal involvement in crime. the correctional facility and listed their parole officer
No longer a peripheral institution, the crim- as a reference (see Pager, 2003 for a more in-depth
inal justice system has become a dominant pres- discussion of these issues).
ence in the lives of young disadvantaged men, 4. A more in-depth discussion of methodolog-
playing a key role in the sorting and stratifying ical concerns, including limits to generalizability,
of labor market opportunities. The “criminal representativeness of testers, sample restrictions, and
credential” now represents a common marker experimenter effects is presented in Pager (2007a),
appendix 4A.
among young disadvantaged men, allowing
5. While not significant in the full sample, this
for the easy identification and exclusion of
interaction becomes significant when analyzed spe-
those with a prior arrest or conviction. Further, cifically for suburban employers or among employers
 Pager—Marked • 763

with whom the testers had extended personal contact Heckman, James, and Peter Seligman. 1993. “The
(see Pager, 2007a, chap. 7). Urban Institute Audit Studies: Their Methods
6. These results do not appear specific to Mil- and Findings.” In Michael Fix and Raymond
waukee. Other audit studies have revealed effects of J. Struyk (eds.), Clear and Convincing Evidence:
similar magnitude (see Pager, 2007b for a summary); Measurement of Discrimination in America. Wash-
and a replication of the present study in New York ington, DC: Urban Institute Press.
City comes to similar conclusions (Pager and West- Mauer, Marc. 1999. Race to Incarcerate. New York:
ern, 2005). New Press.
Neal, Derek, and William Johnson. 1996. “The
REFERENCES Role of Premarket Factors in Black-White Wage
Bureau of Justice Statistics. 2002. “Recidivism of Differences.” Journal of Political Economy 104(5):
Prisoners Released in 1994,” edited by Patrick 869–895.
Langan and David Levin. Washington, DC: U.S. Pager, Devah. 2003. “The Mark of a Criminal Re-
Department of Justice. cord.” American Journal of Sociology 108(5):
———. 2006. “Prison and Jail Inmates at Midyear 937–975.
2005.” Washington, DC: U.S. Department of ———. 2007a. Marked: Race, Crime, and Finding
Justice. Work in an Era of Mass Incarceration. Chicago:
Bureau of Justice Statistics Special Report 1997. University of Chicago Press.
March. Lifetime Likelihood of Going to State or ———. 2007b. “The Use of Field Experiments for
Federal Prison, by Thomas P. Bonczar and Allen Studies of Employment Discrimination: Contri-
J. Beck. Washington, DC: U.S. Department of butions, Critiques, and Directions for the Fu-
Justice. ture.” Annals of the American Academy of Political
Cole, David. 1995. “The paradox of race and and Social Science 609: 104–133.
crime: A comment on Randall Kennedy’s ‘pol- Pager, Devah, and Bruce Western. 2005. “Discrim-
itics of distinction’.” Georgetown Law Journal ination in Low Wage Labor Markets: Evidence
83:2547–2571. from New York City.” Paper presented at Popu-
Devine, P. G., and A. J. Elliot. 1995. “Are Racial lation Association of America. Philadelphia, PA.
Stereotypes Really Fading? The Princeton Trilogy Pettit, Becky, and Bruce Western. 2004. “Mass Im-
Revisited.” Personality and Social Psychology Bulle- prisonment and the Life Course: Race and Class
tin 21(11): 1139–1150. Inequality in U.S. Incarceration.” American So-
Eberhardt, Jennifer L., Phillip Atiba Goff, Valerie J. ciological Review 69: 151–169.
Purdie, and Paul G. Davies. 2004. “Seeing black: Sampson, Robert J., and John H. Laub. 1993. Crime
Race, crime, and visual processing.” Journal of in the Making: Pathways and Turning Points
Personality & Social Psychology 87:876–893. Through Life. Cambridge, MA: Harvard Univer-
Farkas, George, and Kevin Vicknair. 1996. “Appro- sity Press.
priate tests of racial wage discrimination require Schwartz, Richard, and Jerome Skolnick. 1962.
controls for cognitive skill: Comment on Can- “Two Studies of Legal Stigma.” Social Problems,
cio, Evans, and Maume.” American Sociological Fall: 133–142.
Review 61:557–560. Smith, Tom W. 1991. “Ethnic Images. General So-
Freeman, Richard B. 1987. “The Relation of Crim- cial Survey Technical Report, 19.” Chicago: Na-
inal Activity to Black Youth Employment.” Re- tional Opinion Research Center, University of
view of Black Political Economy 16(1–2): 99–107. Chicago.
Gilliam, Franklin D., and Shanto Iyengar. 2000. Sniderman, Paul M., and Thomas Piazza. 1993. The
“Prime suspects: The influence of local television Scar of Race. Cambridge, MA: Harvard Univer-
news on the viewing public.” American Journal of sity Press.
Political Science 44:560–573. Travis, Jeremy, Amy Solomon, and Michelle Waul.
Graham, Sandra, and Brian S. Lowery. 2004. “Prim- 2001. From Prison to Home: The Dimensions and
ing unconscious racial stereotypes about ado- Consequences of Prisoner Reentry. Washington,
lescent offenders.” Law and Human Behavior DC: Urban Institute Press.
28:483–504. Uggen, Christopher. 2000. “Work as a Turning Point
Heckman, James J. 1998. “Detecting discrimi- in the Life Course of Criminals: A Duration
nation.” The Journal of Economic Perspectives Model of Age, Employment, and Recidivism.”
12:101–116. American Sociological Review 65(4): 529–546.
764 • Part VII: Race and Ethnicity

Uggen, Christopher, Jeff Manza, and Melissa Western, Bruce, and Katherine Beckett. 1999. “How
Thompson. 2006. “Citizenship and Reintegra- Unregulated Is the U.S. Labor Market? The Penal
tion: The Socioeconomic, Familial, and Civic System as a Labor Market Institution.” American
Lives of Criminal Offenders.” The Annals of the Journal of Sociology 104(4): 1030–60.
American Academy of Social and Political Science Wilson, William Julius. 1996. When Work Disap-
605:281–310. pears: The World of the New Urban Poor. New
Western, Bruce. 2002. “The Impact of Incarceration York: Vintage Books.
on Wage Mobility and Inequality.” American So-
ciological Review 67(4): 526–546.
 Wilson—The Declining Signikcance of Race • 765

Are Racial and Ethnic Distinctions

Declining in Signikcance?

89 • William Julius Wilson

The Declining Signikcance of Race

Blacks and Changing American Institutions

Race relations in America have undergone fun- In the antebellum period, and in the latter
damental changes in recent years, so much so half of the nineteenth century through the first
that now the life chances of individual blacks half of the twentieth century, the continuous
have more to do with their economic class posi- and explicit efforts of whites to construct racial
tion than with their day-to-day encounters with barriers profoundly affected the lives of black
whites. In earlier years the systematic efforts of Americans. Racial oppression was deliberate,
whites to suppress blacks were obvious to even overt, and is easily documented, ranging from
the most insensitive observer. Blacks were denied slavery to segregation, from the endeavors of the
access to valued and scarce resources through white economic elite to exploit black labor to the
various ingenious schemes of racial exploita- actions of the white masses to eliminate or neu-
tion, discrimination, and segregation, schemes tralize black competition, particularly economic
that were reinforced by elaborate ideologies of competition.1 As the nation has entered the lat-
racism. But the situation has changed. However ter half of the twentieth century, however, many
determinative such practices were for the pre- of the traditional barriers have crumbled under
vious efforts of the black population to achieve the weight of the political, social, and economic
racial equality, and however significant they were changes of the civil rights era. A new set of ob-
in the creation of poverty-stricken ghettoes and stacles has emerged from basic structural shifts
a vast underclass of black proletarians—that in the economy. These obstacles are therefore
massive population at the very bottom of the so- impersonal but may prove to be even more for-
cial class ladder plagued by poor education and midable for certain segments of the black popu-
low-paying, unstable jobs—they do not provide lation. Specifically, whereas the previous barriers
a meaningful explanation of the life chances of were usually designed to control and restrict the
black Americans today. The traditional patterns entire black population, the new barriers create
of interaction between blacks and whites, partic- hardships essentially for the black underclass;
ularly in the labor market, have been fundamen- whereas the old barriers were based explicitly
tally altered. on racial motivations derived from intergroup

William Julius Wilson. The Declining Significance of Race: Blacks and Changing American Institutions, pp. 1–9,
12–23, 183–185. Copyright © 1978 by the University of Chicago Press. Used by permission of the University
of Chicago Press and the author.
766 • Part VII: Race and Ethnicity

contact, the new barriers have racial significance industrial, and modern industrial stages of
only in their consequences, not in their origins. American race relations.
In short, whereas the old barriers bore the perva- Although this abbreviated designation of
sive features of racial oppression, the new barri- the periods of American race relations seems to
ers indicate an important and emerging form of relate racial change to fundamental economic
class subordination. changes rather directly, it bears repeating that the
It would be shortsighted to view the tradi- different stages of race relations are structured by
tional forms of racial segregation and discrim- the unique arrangements and interactions of the
ination as having essentially disappeared in economy and the polity. Although I stress the
contemporary America; the presence of blacks is economic basis of structured racial inequality in
still firmly resisted in various institutions and so- the preindustrial and industrial periods of race
cial arrangements, for example, residential areas relations, I also attempt to show how the polity
and private social clubs. However, in the eco- more or less interacted with the economy either
nomic sphere, class has become more important to reinforce patterns of racial stratification or to
than race in determining black access to privilege mediate various forms of racial conflict. More-
and power. It is clearly evident in this connection over, for the modern industrial period, I try to
that many talented and educated blacks are now show how race relations have been shaped as
entering positions of prestige and influence at a much by important economic changes as by im-
rate comparable to or, in some situations, exceed- portant political changes. Indeed, it would not
ing that of whites with equivalent qualifications. be possible to understand fully the subtle and
It is equally clear that the black underclass is in manifest changes in race relations in the mod-
a hopeless state of economic stagnation, falling ern industrial period without recognizing the
further and further behind the rest of society. . . . dual and often reciprocal influence of structural
changes in the economy and political changes
in the state. Thus, my central argument is that
Three Stages of American different systems of production and/or different
Race Relations arrangements of the polity have imposed differ-
My basic thesis is that American society has ent constraints on the way in which racial groups
experienced three major stages of black-white have interacted in the United States, constraints
contact and that each stage embodies a differ- that have structured the relations between ra-
ent form of racial stratification structured by the cial groups and that have produced dissimilar
particular arrangement of both the economy and contexts not only for the manifestation of racial
the polity. Stage one coincides with antebellum antagonisms but also for racial group access to
slavery and the early postbellum era and may rewards and privileges.
be designated the period of plantation economy In contrast to the modern industrial period
and racial-caste oppression. Stage two begins in in which fundamental economic and political
the last quarter of the nineteenth century and changes have made economic class affiliation
ends at roughly the New Deal era and may be more important than race in determining Negro
identified as the period of industrial expansion, prospects for occupational advancement, the pre-
class conflict, and racial oppression. Finally, stage industrial and industrial periods of black-white
three is associated with the modern, industrial, relations have one central feature in common,
post–World War II era, which really began to namely, overt efforts of whites to solidify eco-
crystallize during the 1960s and 1970s, and may nomic racial domination (ranging from the ma-
be characterized as the period of progressive tran- nipulation of black labor to the neutralization
sition from racial inequalities to class inequalities. or elimination of black economic competition)
For the sake of brevity I shall identify the dif- through various forms of juridical, political,
ferent periods respectively as the preindustrial, and social discrimination. Since racial problems
 Wilson—The Declining Signikcance of Race • 767

during these two periods were principally related racial lines to the extent that the capitalist class
to group struggles over economic resources, they is able to isolate the lower-priced black labor
readily lend themselves to the economic class force by not only supporting job, housing, and
theories of racial antagonisms that associate racial educational discrimination against blacks, but
antipathy with class conflict. A brief consider- also by developing or encouraging racial preju-
ation of these theories, followed by a discussion dices and ideologies of racial subjugation such
of their basic weaknesses, will help to raise a as racism. The net effect of such a policy is to
number of theoretical issues that will be useful insure a marginal working class of blacks and to
for analyzing the dynamics of racial conflict in establish a relatively more privileged position for
the preindustrial and industrial stages of Ameri- the established white labor force. Since discrim-
can race relations. However, in a later section of ination guarantees a situation where the average
this chapter I shall attempt to explain why these wage rate of the black labor force is less than the
theories are not very relevant to the modern in- average wage rate of the established white labor
dustrial stage of American race relations. force, the probability of labor solidarity against
the capitalist class is diminished.
At the same time, orthodox Marxists argue,
Economic Class Theories the members of the capitalist class benefit not
Students of race relations have paid considerable only because they have created a reserved army
attention to the economic basis of racial antag- of labor that is not united against them and the
onism in recent years, particularly to the theme appropriation of surplus from the black labor
that racial problems in historical situations are force is greater than the exploitation rate of the
related to the more general problems of eco- white labor force,7 but also because they can
nomic class conflict. A common assumption of counteract ambitious claims of the white labor
this theme is that racial conflict is merely a spe- force for higher wages either by threatening to
cial manifestation of class conflict. Accordingly, increase the average wage rate of black workers
ideologies of racism, racial prejudices, institu- or by replacing segments of the white labor force
tionalized discrimination, segregation, and other with segments of the black labor force in spe-
factors that reinforce or embody racial stratifica- cial situations such as strikes. The weaker the
tion are seen as simply part of a superstructure national labor force, the more likely it is that it
determined and shaped by the particular ar- will be replaced by lower-paid black labor espe-
rangement of the class structure.2 However, given cially during organized strikes demanding wage
this basic assumption, which continues to be the increases and improved working conditions. In
most representative and widely used economic short, orthodox Marxists argue that racial antag-
class argument,3 proponents have advanced two onism is designed to be a “mask for privilege”
major and somewhat divergent explanations of that effectively conceals the efforts of the ruling
how class conflicts actually shape and determine class to exploit subordinate minority groups and
racial relations—the orthodox Marxist theory of divide the working class.
capitalist exploitation,4 and the split labor-market In interesting contrast to the orthodox Marx-
theory of working-class antagonisms.5 ist approach, the split labor-market theory posits
The orthodox Marxist theory, which is the the view that rather than attempting to protect
most popular variant of the Marxists’ explana- a segment of the laboring class, business “sup-
tions of race,6 postulates that because the ulti- ports a liberal or laissez faire ideology that would
mate goal of the capitalist class is to maximize permit all workers to compete freely in an open
profits, efforts will be made to suppress work- market. Such open competition would displace
ers’ demands for increased wages and to weaken higher paid labor. Only under duress does busi-
their bargaining power by promoting divisions ness yield to a labor aristocracy [i.e., a privileged
within their ranks. The divisions occur along position for white workers].”8
768 • Part VII: Race and Ethnicity

The central hypothesis of the split In some cases members of the lower-paid
labor-market theory is that racial antagonism laboring class, either from within the territorial
first develops in a labor market split along racial boundaries of a given country or from another
lines. The term “antagonism” includes all aspects country, are drawn into or motivated to enter a
of intergroup conflict, from beliefs and ideolo- labor market because they feel they can improve
gies (e.g., racism), to overt behavior (e.g., dis- their standard of living. As Edna Bonacich points
crimination), to institutions (e.g., segregationist out, “the poorer the economy of the recruits, the
laws). A split labor market occurs when the price less the inducement needed for them to enter the
of labor for the same work differs for at least two new labor market.”11 In other cases, individuals
groups, or would differ if they performed the are forced into a new labor-market situation,
same work. The price of labor “refers to labor’s such as the involuntary migration of blacks into
total cost to the employer, including not only a condition of slavery in the United States. In
wages, but the cost of recruitment, transporta- this connection, the greater the employer’s con-
tion, room and board, education, health care (if trol over lower-priced labor, the more threaten-
the employer must bear these), and the cost of ing is lower-paid labor to higher-paid labor.
labor unrest.”9 However, if more expensive labor is strong
There are three distinct classes in a split enough, that is, if it possesses the power re-
labor market: (1) business or employers; (2) sources to preserve its economic interests, it can
higher-paid labor; and (3) cheaper labor. Con- prevent being replaced or undercut by cheaper
flict develops between these three classes because labor. On the one hand it can exclude lower-paid
of different interests. The main goal of business labor from a given territory. “Exclusion move-
or employers is to maintain as cheap a labor force ments clearly serve the interests of higher paid
as possible in order to compete effectively with labor. Its standards are protected, while the capi-
other businesses and to maximize economic re- talist class is deprived of cheaper labor.”12 On the
turns. Employers will often import laborers from other hand, if it is not possible for higher-paid
other areas if local labor costs are too high or if labor to rely on exclusion (cheaper labor may
there is a labor shortage. Whenever a labor short- be indigenous to the territory or may have been
age exists, higher-paid labor is in a good bargain- imported early in business-labor relations when
ing position. Accordingly, if business is able to higher-paid labor could not prevent the move)
attract cheaper labor to the market place, the in- then it will institutionalize a system of ethnic
terests of higher-paid labor are threatened. They stratification which could (1) monopolize skilled
may lose some of the privileges they enjoy, they positions, thereby ensuring the effectiveness of
may lose their bargaining power, and they may strike action; (2) prevent cheaper labor from
even lose their jobs. Moreover, the presence of developing the skills necessary to compete with
cheaper labor in a particular job market may not higher-paid labor (for example, by imposing
only represent actual competition but potential barriers to equal access to education); and (3)
competition as well. An “insignificant trickle” deny cheaper labor the political resources that
could be seen as the beginning of a major immi- would enable them to undercut higher-paid
gration. If the labor market is split along ethnic labor through, say, governmental regulations.
lines, for example, if higher-paid labor is white “In other words, the solution to the devastating
and lower-paid labor is black, class antagonisms potential of weak, cheap labor is, paradoxically,
are transformed into racial antagonisms. Thus, to weaken them further, until it is no longer in
“while much rhetoric of ethnic antagonism con- business’ immediate interest to use them as re-
centrates on ethnicity and race, it really in large placement.”13 Thus, whereas orthodox Marxist
measure (though probably not entirely) expresses arguments associate the development and insti-
this class conflict.”10 tutionalization of racial stratification with the
 Wilson—The Declining Signikcance of Race • 769

motivations and activities of the capitalist class, However, if we ignore the more categorical
the split labor-market theory traces racial strat- assertions that attribute responsibility for racial
ification directly to the powerful, higher-paid stratification to a particular class and focus se-
working class. riously on the analyses of interracial contact in
Implicit in both of these economic class the- the labor market, then I will be able to demon-
ories is a power-conflict thesis associating the strate that, depending on the historical situation,
regulation of labor or wages with properties each of the economic class theories provides ar-
(ownership of land or capital, monopolization guments that help to illuminate race relations
of skilled positions) that determine the scope during the preindustrial and industrial periods of
and degree of a group’s ability to influence be- black-white contact. By the same token, I hope
havior in the labor market. Furthermore, both to explain why these theories have little applica-
theories clearly demonstrate the need to focus tion to the third, and present, stage of modern
on the different ways and situations in which industrial race relations. My basic argument is
various segments of the dominant racial group that the meaningful application of the argu-
perceive and respond to the subordinate racial ments in each theory for any given historical pe-
group. However, as I examine the historical riod depends considerably on knowledge of the
stages of race relations in the United States, I constraints imposed by the particular systems of
find that the patterns of black/white interaction production and by the particular laws and pol-
do not consistently and sometimes do not con- icies of the state during that period, constraints
veniently conform to the propositions outlined that shape the structural relations between racial
in these explanations of racial antagonism. In and class groups and which thereby produce dif-
some cases, the orthodox Marxian explanation ferent patterns of intergroup interaction. . . .
seems more appropriate; in other instances, the
split labor-market theory seems more appropri-
ate; and in still others, neither theory can, in iso- The Influence of the System of
lation, adequately explain black-white conflict. Production
If we restrict our attention for the moment to The term “system of production” not only refers
the struggle over economic resources, then the to the technological basis of economic processes
general pattern that seems to have characterized or, in Karl Marx’s terms, the “forces of produc-
race relations in the United States during the tion,” but it also implies the “social relations of
preindustrial and industrial stages was that the production,” that is, “the interaction (for exam-
economic elite segments of the white population ple, through employment and property arrange-
have been principally responsible for those forms ment) into which men enter at a given level of
of racial inequality that entail the exploitation of the development of the forces of production.”14
labor (as in slavery), whereas whites in the lower As I previously indicated, different systems of
strata have been largely responsible for those production impose constraints on racial group
forms of imposed racial stratification that are de- interaction. In the remainder of this section I
signed to eliminate economic competition (as in should like to provide a firmer analytical basis
job segregation). Moreover, in some situations, for this distinction as it applies specifically to
the capitalist class and white workers form an the three stages of American race relations, in-
alliance to keep blacks suppressed. Accordingly, corporating in my discussion relevant theoreti-
restrictive arguments to the effect that racial cal points raised in the foregoing sections of this
stratification was the work of the capitalist class chapter.
or was due to the “victory” of higher-paid white It has repeatedly been the case that a non-
labor obscure the dynamics of complex and vari- manufacturing or plantation economy with a
able patterns of black-white interaction. simple division of labor and a small aristocracy
770 • Part VII: Race and Ethnicity

that dominates the economic and political life which were not characterized by competition
of a society has characteristically generated a pa- for scarce resources but by the exploitation of
ternalistic rather than a competitive form of race black labor.17 Social distance tended to be clearly
relations, and the antebellum South was no ex- symbolized by rituals of racial etiquette: gestures
ception.15 Paternalistic racial patterns reveal close and behavior reflecting dominance and subser-
symbiotic relationships marked by dominance vience. Consequently, any effort to impose a
and subservience, great social distance and little system of public segregation was superfluous.
physical distance, and clearly symbolized rituals Furthermore, since the social gap between the
of racial etiquette. The southern white aristoc- aristocracy and black slaves was wide and stable,
racy created a split labor market along racial lines ideologies of racism played less of a role in the
by enslaving blacks to perform tasks at a cheaper subordination of blacks than they subsequently
cost than free laborers of the dominant group. did in the more competitive systems of race re-
This preindustrial form of race relations was not lations following the Civil War. In short, the
based on the actions of dominant-group labor- relationship represented intergroup paternalism
ers, who, as we shall see, were relatively power- because it allowed for “close symbiosis and even
less to effect significant change in race relations intimacy, without any threat to status inequal-
during this period, but on the structure of the ities.”18 This was in sharp contrast to the more
relations established by the aristocracy. Let me competitive forms of race relations that accom-
briefly amplify this point. panied the development of industrial capitalism
In the southern plantation economy, public in the late nineteenth century and first few de-
power was overwhelmingly concentrated in the cades of the twentieth century (the industrial
hands of the white aristocracy. This power was period of American race relations), wherein the
not only reflected in the control of economic complex division of labor and opportunities for
resources and in the development of a juridical greater mobility not only produced interaction,
system that expressed the class interests of the competition, and labor-market conflict between
aristocracy, but also in the way the aristocracy blacks and the white working class, but also pro-
was able to impose its viewpoint on the larger so- vided the latter with superior resources (relative
ciety.16 This is not to suggest that these aspects of to those they possessed under the plantation
public power have not been disproportionately economy) to exert greater influence on the form
controlled by the economic elite in modern in- and content of racial stratification.
dustrialized Western societies; rather it indicates The importance of the system of produc-
that the hegemony of the southern ruling elite tion in understanding race relations is seen in
was much greater in degree, not in kind, than a comparison of Brazil and the southern United
in these societies. The southern elite’s hegemony States during the post-slavery periods. In the
was embodied in an economy that required little United States, the southern economy experi-
horizontal or vertical mobility. Further, because enced a fairly rapid rate of expansion during the
of the absence of those gradations of labor power late nineteenth century, thereby creating various
associated with complex divisions of labor, white middle level skilled and unskilled positions that
workers in the antebellum and early postbel- working-class whites attempted to monopolize
lum South had little opportunity to challenge for themselves. The efforts of white workers to
the control of the aristocracy. Because white la- eliminate black competition in the South gener-
borers lacked power resources in the southern ated an elaborate system of Jim Crow segregation
plantation economy, their influence on the form that was reinforced by an ideology of biological
and quality of racial stratification was minimal racism. The white working class was aided not
throughout the antebellum and early postbellum only by its numerical size, but also by its increas-
periods. Racial stratification therefore primarily ing accumulation of political resources that ac-
reflected the relationships established between companied changes in its relation to the means
blacks and the white aristocracy, relationships of production.
 Wilson—The Declining Signikcance of Race • 771

As white workers gradually translated their tends to originate outside the economic order
increasing labor power into political power, and to have little connection with labor-market
blacks experienced greater restrictions in their strife. Basic changes in the system of production
efforts to achieve a satisfactory economic, politi- have produced a segmented labor structure in
cal, and social life. In Brazil, on the other hand, which blacks are either isolated in the relatively
the large Negro and mulatto population was not non-unionized, low-paying, basically undesirable
thrust into competition with the much smaller jobs of the noncorporate sector, or occupy the
white population over access to higher-status po- higher-paying corporate and government indus-
sitions because, as Marvin Harris notes, “there try positions in which job competition is either
was little opportunity for any member of the controlled by powerful unions or is restricted to
lower class to move upward in the social hierar- the highly trained and educated, regardless of
chy.”19 No economic class group or racial group race. If there is a basis for labor-market conflict
had much to gain by instituting a rigid system in the modern industrial period, it is most prob-
of racial segregation or cultivating an ideology ably related to the affirmative action programs
of racial inferiority. Racial distinctions were in- originating from the civil rights legislation of
significant to the landed aristocracy, who con- the 1960s. However, since affirmative action
stituted a numerically small upper class in what programs are designed to improve job opportu-
was basically a sharply differentiated two-class nities for the talented and educated, their major
society originally shaped during slavery. The impact has been in the higher-paying jobs of the
mulattoes, Negroes, and poor whites were all in expanding government sector and the corporate
the same impoverished lower-ranking position. sector. The sharp increase of the more privileged
“The general economic stagnation which has blacks in these industries has been facilitated by
been characteristic of lowland Latin America the combination of affirmative action and rapid
since the abolition of slavery,” observes Marvin industry growth. Indeed despite the effectiveness
Harris, “tends to reinforce the pattern of pacific of affirmative action programs the very expan-
relationships among the various racial groups in sion of these sectors of the economy has kept
the lower ranking levels of the social hierarchy. racial friction over higher-paying corporate and
Not only were the poor whites out-numbered by government jobs to a minimum.
the mulattoes and Negroes, but there was little Unlike the occupational success achieved by
of a significant material nature to struggle over the more talented and educated blacks, those in
in view of the generally static condition of the the black underclass find themselves locked in
economy.”20 Accordingly, in Brazil, segregation, the low-paying and dead-end jobs of the non-
discrimination, and racist ideologies failed to corporate industries, jobs which are not in high
crystallize in the first several decades following demand and which therefore do not generate
the end of slavery. More recently, however, in- racial competition or strife among the national
dustrialization has pushed Brazil toward a com- black and white labor force. Many of these jobs
petitive type of race relations, particularly the go unfilled, and employers often have to turn to
southern region (for example, São Paulo) which cheap labor from Mexico and Puerto Rico. As
has experienced rapid industrialization and has Nathan Glazer has pointed out, “Expectations
blacks in economic competition with many have changed, and fewer blacks and whites today
lower-status white immigrants.21 will accept a life at menial labor with no hope for
Whereas the racial antagonism in the United advancement, as their fathers and older brothers
States during the period of industrial race rela- did and as European immigrants did.”22
tions (such as the Jim Crow segregation move- Thus in the modern industrial era neither the
ment in the South and the race riots in northern corporate nor government sectors nor the non-
cities) tended to be either directly or indirectly corporate low-wage sector provide the basis for
related to labor-market conflicts, racial antago- the kind of interracial competition and conflict
nism in the period of modern industrial relations that has traditionally plagued the labor market
772 • Part VII: Race and Ethnicity

in the United States. This, then, is the basis for since the initial state and municipal legislation of
my earlier contention that the economic class the 1940s, increasingly created changes leading
theories which associate labor-market conflicts to the erosion of traditional racial alignments; in
with racial antagonism have little application other words, instead of reinforcing racial barriers
to the present period of modern industrial race created during the preindustrial and industrial
relations. periods, the political system in recent years has
tended to promote racial equality.
Thus, in the previous periods the polity was
The Polity and American quite clearly an instrument of the white popu-
Race Relations lation in suppressing blacks. The government’s
If the patterned ways in which racial groups have racial practices varied, as I indicated above, de-
interacted historically have been shaped in major pending on which segment of the white popula-
measure by different systems of production, they tion was able to assert its class interests. However,
have also been undeniably influenced by the in the past two decades the interests of the black
changing policies and laws of the state. For an- population have been significantly reflected in
alytical purposes, it would be a mistake to treat the racial policies of the government, and this
the influences of the polity and the economy as if change is one of the clearest indications that the
they were separate and unrelated. The legal and racial balance of power has been significantly al-
political systems in the antebellum South were tered. Since the early 1940s the black population
effectively used as instruments of the slavehold- has steadily gained political resources and, with
ing elite to strengthen and legitimate the insti- the help of sympathetic white allies, has shown
tution of slavery. But as industrialization altered an increasing tendency to utilize these resources
the economic class structure in the postbellum in promoting or protecting its group interests.
South, the organizing power and political con- By the mid-twentieth century the black vote
sciousness of the white lower class increased and had proved to be a major vehicle for political
its members were able to gain enough control pressure. The black vote not only influenced the
of the political and juridical systems to legalize outcome of national elections but many congres-
a new system of racial domination (Jim Crow sional, state, and municipal elections as well. Fear
segregation) that clearly reflected their class of the Negro vote produced enactment of public
interests. accommodation and fair employment practices
In effect, throughout the preindustrial pe- laws in northern and western municipalities and
riod of race relations and the greater portion of states prior to the passage of federal civil rights
the industrial period the role of the polity was legislation in 1964. This political resurgence for
to legitimate, reinforce, and regulate patterns of black Americans increased their sense of power,
racial inequality. However, it would be unwar- raised their expectations, and provided the foun-
ranted to assume that the relationship between dation for the proliferation of demands which
the economic and political aspects of race neces- shaped the black revolt during the 1960s. But
sarily implies that the latter is simply a derivative there were other factors that helped to buttress
phenomenon based on the more fundamental Negro demands and contributed to the devel-
processes of the former. The increasing interven- oping sense of power and rising expectations,
tion, since the mid-twentieth century, of state namely, a growing, politically active black mid-
and federal government agencies in resolving dle class following World War II and the emer-
or mediating racial conflicts has convincingly gence of the newly independent African states.
demonstrated the political system’s autonomy in The growth of the black middle class was
handling contemporary racial problems. Instead concurrent with the growth of the black urban
of merely formalizing existing racial alignments population. It was in the urban areas, with their
as in previous periods, the political system has, expanding occupational opportunities, that a
 Wilson—The Declining Signikcance of Race • 773

small but significant number of blacks were employment legislation and the authorization of
able to upgrade their occupations, increase their affirmative action programs the government has
income, and improve their standard of living. helped clear the path for more privileged blacks,
The middle-class segment of an oppressed mi- who have the requisite education and training,
nority is most likely to participate in a drive for to enter the mainstream of American occupa-
social justice that is disciplined and sustained. tions. However, such government programs do
In the early phases of the civil rights movement, not confront the impersonal economic barriers
the black middle class channeled its energies confronting members of the black underclass,
through organizations such as the National who have been effectively screened out of the
Association for the Advancement of Colored corporate and government industries. And the
People, which emphasized developing political very attempts of the government to eliminate
resources and successful litigation through the traditional racial barriers through such programs
courts. These developments were paralleled by as affirmative action have had the unintentional
the attack against traditional racial alignments effect of contributing to the growing economic
in other parts of the world. The emerging newly class divisions within the black community.
independent African states led the assault. In
America, the so-called “leader of the free world,”
the manifestation of racial tension and violence Class Stratification and Changing
has been a constant source of embarrassment to Black Experiences
national government officials. This sensitivity to The problems of black Americans have always
world opinion made the national government been compounded because of their low position
more vulnerable to pressures of black protest at in both the economic order (the average eco-
the very time when blacks had the greatest pro- nomic class position of blacks as a group) and
pensity to protest. the social order (the social prestige or honor
The development of black political resources accorded individual blacks because of their as-
that made the government more sensitive to cribed racial status). It is of course true that the
Negro demands, the motivation and morale of low economic position of blacks has helped to
the growing black middle class that resulted in shape the categorical social definitions attached
the political drive for racial equality, and the to blacks as a racial group, but it is also true that
emergence of the newly independent African the more blacks become segmented in terms of
states that increased the federal government’s economic class position, the more their concerns
vulnerability to civil rights pressures all com- about the social significance of race will vary.
bined to create a new sense of power among In the preindustrial period of American race
black Americans and to raise their expectations relations there was of course very little variation
as they prepared to enter the explosive decade in the economic class position of blacks. The sys-
of the 1960s. The national government was also tem of racial-caste oppression relegated virtually
aware of this developing sense of power and re- all blacks to the bottom of the economic class
sponded to the pressures of black protest in the hierarchy. Moreover, the social definitions of
1960s with an unprecedented series of legislative racial differences were heavily influenced by the
enactments to protect black civil rights. ideology of racism and the doctrine of paternal-
The problem for blacks today, in terms of ism, both of which clearly assigned a subordinate
government practices, is no longer one of legal- status for blacks vis-à-vis whites. Occasionally, a
ized racial inequality. Rather the problem for few individual free blacks would emerge and ac-
blacks, especially the black underclass, is that cumulate some wealth or property, but they were
the government is not organized to deal with the overwhelming exception. Thus the uniformly
the new barriers imposed by structural changes low economic class position of blacks reinforced
in the economy. With the passage of equal and, in the eyes of most whites, substantiated
774 • Part VII: Race and Ethnicity

the social definitions that asserted Negroes were interests of this class were clearly reflected in the
culturally and biogenetically inferior to whites. way the race issues were defined and articulated.
The uniformly low economic class position of Thus, the concept of “freedom” quite clearly im-
blacks also removed the basis for any meaningful plied, in the early stages of the movement, the
distinction between race issues and class issues right to swim in certain swimming pools, to eat
within the black community. in certain restaurants, to attend certain movie
The development of a black middle class ac- theaters, and to have the same voting privileges
companied the change from a preindustrial to an as whites. These basic concerns were reflected in
industrial system of production. Still, despite the the 1964 Civil Rights Bill which helped to create
fact that some blacks were able to upgrade their the illusion that, when the needs of the black
occupation and increase their education and in- middle class were met, so were the needs of the
come, there were severe limits on the areas in entire black community.
which blacks could in fact advance. Throughout However, although the civil rights movement
most of the industrial period of race relations, initially failed to address the basic needs of the
the growth of the black middle class occurred members of the black lower class, it did increase
because of the expansion of institutions created their awareness of racial oppression, heighten
to serve the needs of a growing urbanized black their expectations about improving race rela-
population. The black doctor, lawyer, teacher, tions, and increase their impatience with existing
minister, businessman, mortician, excluded racial arrangements. These feelings were dra-
from the white community, was able to create matically manifested in a series of violent ghetto
a niche in the segregated black community. Al- outbursts that rocked the nation throughout the
though the income levels and life-styles of the late 1960s. These outbreaks constituted the most
black professionals were noticeably and some- massive and sustained expression of lower-class
times conspicuously different from those of the black dissatisfaction in the nation’s history. They
black masses, the two groups had one basic thing also forced the political system to recognize the
in common, a racial status contemptuously re- problems of human survival and de facto seg-
garded by most whites in society. If E. Franklin regation in the nation’s ghettoes—problems
Frazier’s analysis of the black bourgeoisie is cor- pertaining to unemployment and underemploy-
rect, the black professionals throughout the in- ment, inferior ghetto schools, and deteriorated
dustrial period of American race relations tended housing.
to react to their low position in the social order However, in the period of modern industrial
by an ostentatious display of material possessions race relations, it would be difficult indeed to
and a conspicuous effort to disassociate them- comprehend the plight of inner-city blacks by
selves from the black masses.23 exclusively focusing on racial discrimination.
Still, as long as the members of the black For in a very real sense, the current problems
middle class were stigmatized by their racial sta- of lower-class blacks are substantially related to
tus; as long as they were denied the social recog- fundamental structural changes in the economy.
nition accorded their white counterparts; more A history of discrimination and oppression cre-
concretely, as long as they remained restricted ated a huge black underclass, and the technolog-
in where they could live, work, socialize, and be ical and economic revolutions have combined to
educated, race would continue to be a far more insure it a permanent status.
salient and important issue in shaping their sense As the black middle class rides on the wave
of group position than their economic class po- of political and social changes, benefiting from
sition. Indeed, it was the black middle class that the growth of employment opportunities in the
provided the leadership and generated the mo- growing corporate and government sectors of the
mentum for the civil rights movement during economy, the black underclass falls behind the
the mid-twentieth century. The influence and larger society in every conceivable respect. The
 Wilson—The Declining Signikcance of Race • 775

economic and political systems in the United competition for jobs in the industrial period)
States have demonstrated remarkable flexibil- through various forms of political, juridical,
ity in allowing talented blacks to fill positions and social discrimination; in the modern indus-
of prestige and influence at the same time that trial period fundamental economic and politi-
these systems have shown persistent rigidity in cal changes have made economic class position
handling the problems of lower-class blacks. As a more important than race in determining black
result, for the first time in American history class chances for occupational mobility. Finally, I have
issues can meaningfully compete with race issues outlined the importance of racial norms or belief
in the way blacks develop or maintain a sense of systems, especially as they relate to the general
group position.24 problem of race and class conflict in the prein-
dustrial and industrial periods.
My argument that race relations in Amer-
Conclusion ica have moved from economic racial oppres-
The foregoing sections of this chapter present sion to a form of class subordination for the
an outline and a general analytical basis for the less-privileged blacks is not meant to suggest that
arguments that will be systematically explored racial conflicts have disappeared or have even
[elsewhere].25 I have tried to show that race rela- been substantially reduced. On the contrary,
tions in American society have been historically the basis of such conflicts have shifted from the
characterized by three major stages and that each economic sector to the sociopolitical order and
stage is represented by a unique form of racial therefore do not play as great a role in determin-
interaction which is shaped by the particular ing the life chances of individual black Ameri-
arrangement of the economy and the polity. cans as in the previous periods of overt economic
My central argument is that different systems racial oppression.
of production and/or different policies of the
state have imposed different constraints on the NOTES
way in which racial groups interact—constraints 1. See William J. Wilson, Power, Racism and Priv-
ilege: Race Relations in Theoretical and Sociohistorical
that have structured the relations between ra-
Perspectives (New York: The Free Press, 1973).
cial groups and produced dissimilar contexts
2. In Marxist terminology, the “superstructure”
not only for the manifestation of racial antago- refers to the arrangements of beliefs, norms, ideolo-
nisms but also for racial-group access to rewards gies, and noneconomic institutions.
and privileges. I emphasized in this connection 3. However, not all theorists who emphasize the
that in the preindustrial and industrial periods importance of economic class in explanations of
of American race relations the systems of pro- race relations simply relegate problems of race to the
duction primarily shaped the patterns of racial superstructure. The Marxist scholars Michael Bura-
stratification and the role of the polity was to woy and Eugene Genovese recognize the reciprocal
legitimate, reinforce, or regulate these patterns. influence between the economic class structure and
aspects of the superstructure (belief systems, politi-
In the modern industrial period, however, both
cal systems, etc.), a position which I also share and
the system of production and the polity assume which is developed more fully in subsequent sec-
major importance in creating new patterns of tions of this chapter. See Eugene D. Genovese, Roll,
race relations and in altering the context of racial Jordan, Roll: The World the Slaves Made (New York:
strife. Whereas the preindustrial and industrial Pantheon, 1974); idem, In Red and Black: Marxian
stages were principally related to group struggles Explorations in Southern and Afro-American History
over economic resources as different segments (New York: Vintage Press, 1971); and Michael Bu-
of the white population overtly sought to create rawoy, “Race, Class, and Colonialism,” Social and
and solidify economic racial domination (rang- Economic Studies 23 (1974): 521–50.
4. Oliver C. Cox, Caste, Class and Race: A Study
ing from the exploitation of black labor in the
in Social Dynamics (Garden City, New York: Dou-
preindustrial period to the elimination of black
bleday, 1948); Paul A. Baran and Paul M. Sweezy,
776 • Part VII: Race and Ethnicity

Monopoly Capital: An Essay on the American Eco- production.” However, in Marx’s writings the mode
nomic and Social Order (Harmondsworth: Penguin, of production is often discussed as equivalent only
1966); Michael Reich, “The Economics of Rac- to the “forces of production.” To avoid confusion, I
ism,” in Problems in Political Economy, ed. David have chosen the term “system of production” which
M. Gordon (Lexington, Mass.: Heath, 1971); and denotes the interrelation of the forces of production
M. Nikolinakos, “Notes on an Economic Theory of and the mode of production.
Racism,” Race: A Journal of Race and Group Relations 15. Pierre L. van den Berghe, Race and Racism: A
14 (1973): 365–81. Comparative Perspective (New York: John Wiley and
5. Edna Bonacich, “A Theory of Ethnic Antago- Sons, 1967), p. 26.
nism: The Split Labor Market,” American Sociological 16. See, for example, Genovese, Roll, Jordan, Roll.
Review 37 (October 1972): 547–59; idem, “Aboli- 17. An exception to this pattern occurred
tion, the Extension of Slavery and the Position of in the cities of the antebellum South, where
Free Blacks: A Study of Split Labor Markets in the non-slaveholding whites played a major role in the
United States,” American Journal of Sociology 81 development of urban segregation. However, since
(1975): 601–28. an overwhelming majority of the population resided
6. For examples of alternative and less orthodox in rural areas, race relations in the antebellum south-
Marxist explanations of race, see Eugene D. Geno- ern cities were hardly representative of the region.
vese, The Political Economy of Slavery: Studies in the 18. van den Berghe, Race and Racism, p. 27.
Economy and Society of the Slave South (New York: 19. Marvin Harris, Patterns of Race in the Ameri-
Pantheon, 1966); idem, The World the Slavehold- cas (New York: Walker, 1964), p. 96.
ers Made: Two Essays in Interpretation (New York: 20. Ibid., p. 96.
Pantheon, 1969); idem, In Red and Black; idem, 21. van den Berghe, Race and Racism, p. 28.
Roll, Jordan, Roll; and Burawoy, “Race, Class, and 22. Nathan Glazer, “Blacks and Ethnic Groups:
Colonialism.” The Difference, and the Political Difference It
7. “Exploitation,” in Marxian terminology, refers Makes,” in Key Issues in the Afro-American Experience,
to the difference between the wages workers receive ed. Nathan I. Huggins, Martin Kilson, and Daniel
and the value of the goods they produce. The size of M. Fox (New York: Harcourt Brace Jovanovich,
this difference, therefore, determines the degree of 1971), 2: 209.
exploitation. 23. E. Franklin Frazier, Black Bourgeoisie (New
8. Bonacich, “A Theory of Ethnic Antagonism,” York: The Free Press, 1957). See also Nathan Hare,
p. 557. Black Anglo-Saxons (New York: Collier, 1965).
9. Ibid., p. 549. 24. The theoretical implications of this develop-
10. Ibid., p. 553. ment for ethnic groups in general are discussed by
11. Ibid., p. 549. Milton Gordon under the concept “ethclass.” See
12. Ibid., p. 555. Milton M. Gordon, Assimilation in American Life
13. Ibid., p. 556. (New York: Oxford University Press, 1964).
14. Neil J. Smelser, Karl Marx on Society and So- 25. See William Julius Wilson, The Declining Sig-
cial Change (Chicago: University of Chicago Press, nificance of Race: Blacks and Changing American Insti-
1974), p. xiv. According to Smelser, Marx used tutions (Chicago and London: University of Chicago
the notions “forces of production” and “social rela- Press, 1978).
tions of production” as constituting the “mode of
 Wilson—The Declining Signikcance of Race • 777

William Julius Wilson

The Declining Signikcance of Race

Revisited and Revised

One of the basic arguments of The Declining Sig- The strength of these relationships, however,
nificance of Race is that there has been a deepen- did not increase between 1970 and 2000. “Class
ing economic schism as reflected in a widening does matter,” Freeman states. “Higher status
gap between lower-income and higher-income Blacks generally live in higher-status neighbor-
black families. In light of more recent data, not hoods and those with more Whites. But the im-
only has the family income gap between poorer portance of class has not increased since 1970.
and better-off African Americans continued to The determinants of spatial outcomes for Blacks
widen, but the situation of the bottom fifth of have been remarkably durable at the end of the
black families has deteriorated since 1975. twentieth century. . . . It appears that Blacks
In 2007, 45.6 percent of all poor blacks had will have to achieve upward mobility in other
incomes below 50 percent of the poverty line.1 domains, such as education, before achieving
Overall, poor black families fell below the pov- widespread access to higher-status and White
erty line by an average of $9,266 in 2007, a neighborhoods.”4 Reaching this goal may be
depth of poverty exceeding that of all other ra- more of a challenge for black males than for
cial/ethnic groups in the United States.2 The gap black females.
between the haves and have-nots in the African Indeed, what has also changed since I wrote
American population continues to grow. The Declining Significance of Race is that the
Research also indicates that “higher socioeco- black class structure increasingly reflects gender
nomic status Blacks have more White neighbors, differences, especially among younger blacks, as
fewer poor neighbors, and live in neighborhoods males have fallen behind females on a number
with higher housing values.”3 This fact is impor- of socioeconomic indicators: employment rates,
tant because one’s neighborhood controls access high school completion rates, and average in-
to jobs and schools, and even exposure to vio- come, with some of the sharpest discrepancies
lence. Using individual-level data from the geo- at the lower end of the income hierarchy.5 Black
coded version of the PSID for the years 1970, women have also far outpaced black men in
1980, 1990, and 2001 to correspond with the college completion in recent years. Despite the
decadal censuses, urban planner Lance Freeman fact that the gender gap in college degree attain-
found that higher socioeconomic status among ment is increasing across all racial groups, with
African Americans is generally associated with women generally exceeding men in rates of col-
greater integration and improved locational lege completion, this discrepancy is particularly
outcomes. acute among African Americans. That gap has

William Julius Wilson. “The Declining Significance of Race: Revisited & Revised,” Daedalus 140:2 (Spring
2011), pp. 63–69. Copyright © 2011 Massachusetts Institute of Technology. Reprinted by permission of MIT
Press Journals.
778 • Part VII: Race and Ethnicity

widened steadily over the past twenty-five years. of why both race-specific and race-neutral—in-
In 1979, for every 100 bachelor’s degrees earned cluding class-based—programs must be strongly
by black men, 144 were earned by black women. emphasized and pursued to combat racial
In 2006 to 2007, for every 100 bachelor’s degrees inequality.
conferred on black men, 196 were conferred Other research suggests that the white back-
on black women—nearly a two-to-one ratio. lash against racial entitlements such as affirma-
To put this gap into a larger context, for every tive action, contributed to the government’s
100 bachelor’s degrees earned by white men retreat from antidiscrimination policies during
and every 100 earned by Hispanic men, white the 1980s, a retreat that may have influenced
women earned 130 and Hispanic women earned hiring and promotion decisions in the corporate
158, respectively. The gap widens higher up on sector as well.8 It should come as no surprise that
the educational ladder. For every 100 master’s waning support for affirmative action programs
degrees and 100 doctorates earned by black men, would have an adverse effect on blacks, especially
black women earned 255 and 193, respectively. more advantaged blacks. A number of empiri-
These ratios have huge implications for the so- cal studies have revealed significant differences
cial organization of the black community. If in the family and neighborhood environments
present trends continue, future discussion of the of blacks and whites that are understated when
black class structure will have to include a gender standard measures of socioeconomic status are
component to show the increasing proportion of employed, like, for example, the question of
black women and decreasing proportion of black family background. Even when white parents
men in higher socioeconomic positions. and black parents report the same average in-
In the epilogue to the second edition of The come, white parents have substantially more as-
Declining Significance of Race, I argued that a sets than do black parents.
conclusion one could draw from my book was Whites with the same amount of schooling as
“that the sole concentration on policy programs blacks usually attend better high schools and col-
dealing with racial bias makes it difficult for leges. Furthermore, children’s test scores are af-
blacks to recognize how their fortune is inextri- fected not only by the social and economic status
cably connected with the structure and the func- of their parents but also by the social and eco-
tioning of the modern American economy.”6 In nomic status of their grandparents, meaning that
concluding the epilogue, I wrote: “Supporters of it could take several generations before adjust-
basic economic reform can only hope that in the ments in socioeconomic inequality produce their
1980s the needs and interests of the black poor full benefits. Thus, if we were to rely solely on
(as well as those of the other minority poor and the standard criteria for college admission, such
the white poor) will no longer be underrepre- as SAT scores, even many children from black
sented in serious public discussions, policies, and middle-income families would be denied ad-
programs.”7 These statements were influenced by mission in favor of middle-income whites, who
my sense that while race-specific programs like are not weighed down by the accumulation of
affirmative action had elevated and would con- disadvantages that stem from racial restrictions
tinue to improve the employment prospects of and who, therefore, tend to score higher on the
trained and highly educated blacks, they had not SAT and similar conventional tests. For all these
enhanced the employment opportunities of the reasons, the success of younger educated blacks
black poor. I felt therefore that the focus should remains heavily dependent on affirmative action
shift to more class-based, race-neutral programs. programs, whereby more flexible criteria of eval-
I no longer support this view. Recognizing that uation are used to gauge potential to succeed.
a detailed discussion of policy options would re- The policy implications are obvious.
quire far more space than that available here, I Race-specific policies like affirmative action will
would like to conclude with a brief discussion be required for the foreseeable future to ensure
 Wilson—The Declining Signikcance of Race • 779

the continued mobility of educated blacks. But inequality—some calling for race-based solu-
affirmative action programs are not designed tions, like affirmative action, others calling for
to address the problems of poor blacks, which class-based solutions, such as programs to in-
require greater emphasis on demand-side solu- crease employment in areas with the highest
tions, such as creating tight labor markets in rates of joblessness. Accordingly, if I were writing
which employers are looking for workers rather The Declining Significance of Race today, I would
than workers looking for employers. provide more balance in my policy recommen-
At the time of this writing, the nation is dations by placing much greater emphasis on the
plagued with one of the highest unemployment need to strongly and continuously embrace, as
rates since the Great Depression, affecting all ra- well as advance, both race- and class-based solu-
cial and ethnic groups in the United States. For tions to address life chances for people of color.
almost five decades, the black/white unemploy-
ment ratio was 2.0 or greater, which means that NOTES
the black unemployment rate was at least twice 1. U.S. Census Bureau. Current Population Sur-
vey. 2008 Annual Social and Economic Supplement,
that of the white unemployment rate in both
Table POV28, “Income and Deficit or Surplus of
good and bad economic times. What is unique
Families and Unrelated Individuals by Poverty Sta-
about the current economic crisis is that the un- tus: 2007.”
employment rate has surged for both blacks and 2. By comparison, poor non-Hispanic white
whites. Since December 2009, the black/white families fell below the poverty line by an average of
unemployment ratio has fallen below 2.0. The $7,957; poor Hispanic families by $8,611; and poor
ratio was 1.87 in October 2010 and 1.88 in No- Asian families by $8,959; U.S. Census Bureau, In-
vember 2010.9 come, Poverty, and Health Insurance Coverage in the
This scenario presents a dilemma for the United States: 2007, Current Population Reports
Obama administration, which has publicly ac- P60-235, Table 4, “People with Income Below Spec-
ified Ratios of Their Poverty Thresholds by Selected
knowledged the need to combat racial inequal-
Characteristics: 2007.”
ity. Given the upsurge in unemployment among 3. Lance Freeman, “Is Class Becoming a More
all racial groups, including whites, it would be Important Determinant of Neighborhood Attain-
politically prudent for the president to advance ment for African-Americans?” Urban Affairs Review
programs that address nationwide joblessness. 44 (1) (2008): 3.
However, a strong case could be made for intro- 4. Ibid., 24.
ducing programs that are designed to combat un- 5. William Julius Wilson. More Than Just Race:
employment in the highest areas of joblessness, Being Black and Poor in the Inner City (New York:
including a mix of private- and public-sector W.W. Norton, 2009).
6. Wilson, The Declining Significance of Race, 2nd
initiatives. For example, in black inner cities,
ed., 179.
where the number of very low-skilled individ-
7. Ibid., 182.
uals vastly exceeds the number of low-skill jobs, 8. Jonathan Reider. Canarsie: The Jews and Ital-
a healthy dose of public-sector job creation is ians of Brooklyn Against Liberalism (Cambridge,
needed. This approach would also apply, say, in Mass: Harvard University Press, 1985).
white and Hispanic areas that feature high rates 9. Ratio calculations based on data from the U.S.
of joblessness. Bureau of Labor Statistics, Table A-2. “Employment
The point is that a continuous struggle Status of the Civilian Population by Race, Sex, and
is needed to address the problems of racial Age,” December 3, 2010.
780 • Part VII: Race and Ethnicity

90 • Reanne Frank, Ilana Redstone Akresh, and Bo Lu

How Do Latino Immigrants Fit

into the Racial Order?

“The problem of the twentieth-century is the may not readily identify with an existing racial
problem of the color-line” (Du Bois 1903, 3). group?
A century later, scholars are revisiting W. E. B.
Du Bois’s famous proclamation and labeling
it this century’s biggest puzzle. There is in- Boundary Construction
creased uncertainty about the issue of the color In investigating the puzzle of the US color line
line and its meaning for US society (Lee and and the place of Latinos therein, we follow the
Bean 2007a, 2007b; Lewis, Krysan, and Harris tradition of Barth (1969) and, more recently,
2004). Over the past fifty years the arrival of Wimmer (2008b), who suggest that racial/eth-
unprecedented numbers of immigrants from nic boundaries are not given divisions of human
Latin America, in particular from Mexico, has populations to which all members of society as-
radically changed the racial/ethnic mix of the cribe. Rather, these divisions are conceptualized
United States and has challenged the continued as products of classificatory struggles in which
relevance of the traditional black/white model “individuals and groups struggle over who
of race relations. should be allowed to categorize, which catego-
This chapter addresses a two-part question ries are to be used, which meanings they should
about US racial boundaries and the place of imply and what consequences they should en-
the Latino immigrant population therein. First, tail” (Wimmer 2007, 11).
how do Latinos see themselves fitting into the In the language of a boundary-centered
US racial order; that is, how do Latino immi- framework, a social boundary occurs when
grants define themselves when confronted with two different schemas, one categorical and the
the US system of racial classification? Second, other behavioral, coincide (Wimmer 2008b).
on the flip side, how are Latino immigrants in- According to Wimmer (2008b), the categori-
fluenced by the existing system of racial stratifi- cal dimension of a given boundary divides the
cation in the United States? In particular, does world into social groups: “us” and “them.” For
the US color-based system of stratification affect the purposes of this chapter, we evaluate the cat-
Latino immigrants, even though they themselves egorical dimension of a possible racial boundary

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Reanne Frank, Ilana Redstone Akresh, and Bo Lu, “Latino Immigrants and the U.S.
Racial Order: How and Where Do They Fit In?” American Sociological Review 75 (June 2010), pp. 940–963,
published by Sage Publications. The article printed here was originally prepared by Reanne Frank, Ilana Red-
stone Akresh, and Bo Lu for this fourth edition of Social Stratification: Class, Race, and Gender in Sociological
Perspective, edited by David B. Grusky. Copyright © 2014 by Westview Press.
 Frank, Akresh, and Lu—How Do Latino Immigrants Fit into the Racial Order? • 781

forming around Latinos using immigrant ra- as a blurring of existing racial boundaries (Wim-
cial self-identification practices. The behavioral mer 2008b). This has led some authors to argue
dimension of social boundaries involves the that although Latinos are undoubtedly aware of
everyday action scripts that dictate how individ- the black/white color line in the United States,
uals interact with those labeled “us” and “them” they are asserting a distinct Hispanic/Latino ra-
(Wimmer 2008b). We address this aspect of cial classification by marking “some other race”
racial boundary formation through a matching (Logan 2003; Michael and Timberlake 2008)
exercise that evaluates the effect of skin-color- and rejecting the federal distinction between race
based discrimination on Latino immigrants’ an- and ethnicity (Hitlin, Brown, and Elder 2007;
nual earnings. Campbell and Rogalin 2006). According to
We investigate where Latinos fit into the ex- one Mexican American respondent interviewed
isting US racial order by first examining racial on the subject of US Census racial categories,
self-identification, one aspect of a boundary’s “I think we are big enough to be our own race,
categorical dimension. Current federal policy in especially now that we are growing” (Dowling
the United States says that individuals can be- 2004, 101).
long to one or more racial groups. Significantly, Other researchers focus less on the large
Hispanics and Latinos have been set apart as an number of other race Latinos in the US Cen-
ethnic group and are instructed to choose the sus and instead study the slightly larger number
race that best describes them (Tafoya 2005). This of Latino respondents who select white as their
forces Latino immigrants to place themselves race (Rodriguez 2000). Instead of attempting to
along a sharply drawn color line that separates modify the topography of racial boundaries by
blacks and whites and has remained relatively expanding existing options, choosing “white”
inflexible over time (Davis 1991). may reflect a process whereby the meanings
A boundary-centered framework suggests associated with particular racial boundaries
that actors may pursue several options in reac- are modified. Illustrative remarks come from a
tion to existing categorical boundaries dictated Mexican American woman living in Texas who,
by the state (Wimmer 2008a). From the immi- when asked which race she chose on the US
grant vantage point, one possible reaction to an Census form, responded: “White . . . there’s no
existing racial boundary is to reject the avail- such thing as a brown race. They call Hispanic
able choices and instead promote other nonra- people brown, right? But we are White. . . . Ig-
cial modes of classification and social practice norance is the only thing that would cause any-
(Kusow 2006; Wimmer 2008a). This process, body to check anything else but White, because
labeled “boundary contraction,” can be seen in that’s what we are. . . . There is no such thing as
the case of West Indian immigrants who have brown . . . we’ve been here too long. We’re just
made concerted efforts to resist the US racial- Americans” (Dowling 2004, 92).
ization process by stressing their cultural and The debate over who should be allowed to
national identity (Foner 1987). By asserting categorize, and which categories should be used
an alternative category not recognized in the in the formation of a social boundary, is impor-
United States, members of this group attempt tant because categories coincide with “scripts of
to position themselves in an intermediate racial action” that determine how individual mem-
category between black and white (Landale and bers of particular groups are treated by others.
Oropesa 2002). Whether a boundary can be crossed, altered, or
A second option for Latino immigrants is to redefined depends not only on those attempting
racially identify in a way that challenges existing to renegotiate it, but also on actors on the other
racial boundaries. Many researchers interpret the side who may reject the newcomers. One of the
high number of Latino respondents in the 1990 most tenacious forms of racism in the United
and 2000 Census who marked “some other race” States is discrimination by skin tone (Hochschild
782 • Part VII: Race and Ethnicity

2005). If some Latino immigrants are discrimi- we evaluate the ways in which Latinos are either
nated against on the basis of their skin color, we accepting, challenging, or expanding federally
may expect an increasingly sharper racial bound- validated racial boundaries and what this means
ary to form around them and their descendants for the future of the US color line. Skin color
(Hersch 2008). is measured, as noted previously, using the NIS
skin color scale.
We present five multinomial regression mod-
Data els predicting the log odds of either a white or a
The data come from the 2003 cohort of the New nonwhite racial self-identification as compared
Immigrant Survey (NIS), which was originally to not choosing any racial identification cate-
pilot tested with a 1996 sample cohort of im- gory. We add variable sets sequentially.
migrants. The sampling frame for the NIS 2003 Table 90.1 presents the results from the mul-
was immigrants age eighteen and older who were tinomial logistic regression modeling of racial
granted legal permanent residency between May self-identification. For each specification, two
and November 2003 (see Jasso et al., forthcom- columns present estimates of choosing white
ing). The sample includes new arrivals in the or nonwhite, respectively, as compared to not
United States and individuals who had adjusted choosing a racial category.
their visa status. Model 1 shows the relationship between skin
Among the NIS respondents, 2,729 color and racial identification. Latino immi-
self-identified as Latino. A total of 1,539 obser- grants who identify as white have significantly
vations are available for the first part of the anal- lighter skin tone than those who do not racially
ysis predicting racial identification. Most of the identify. There are no significant differences in
remaining 1,190 cases were lost because of miss- skin color between respondents who identify as
ing information on the skin color chart, primar- nonwhite and those who do not racially identify.
ily because interviews were done over the phone Model 2 tests for national origin differences
and therefore precluded an interviewer assess- in racial classification. Even after accounting for
ment of skin color. The NIS skin color scale (de- variation in racial phenotype, Cubans and South
veloped by Massey and Martin [2003]) instructs Americans are more likely to choose a white ra-
interviewers to rate a respondent’s skin color as cial category, and Dominicans have the highest
closely as possible using the shades shown in odds of not choosing a racial category. These re-
an array of ten progressively darker hands (1 is sults point to the significant influence wielded
lightest, 10 is darkest). In the second part of this by racial categorization systems prevalent in im-
chapter, where our dependent variable is annual migrants’ origin countries, which predict racial
earnings, we restrict the sample to Latinos who self-identification in the United States above and
are members of the labor force and have valid beyond individual skin tone.
skin color information (N = 954). Model 3 adds demographic controls. Cur-
rently married respondents are more likely to
identify as white or nonwhite than to not racially
Racial Self-Identification identify, although the latter effect is only signifi-
We first focus on Latino immigrants’ responses cant once the socioeconomic controls are added
to federally mandated racial identification in Model 4. This finding supports the role the
choices by predicting racial identification in mate selection process may play in racial identity
three categories: (1) white, (2) nonwhite (includ- formation.
ing those who chose Native American/Alaska Model 4 accounts for the role of socioeco-
Native, Black, Asian, and Native Hawaiian/ nomic factors in explaining differences in racial
Pacific Islander), and (3) refused to answer. By identification. Respondents with higher incomes
focusing on these patterns of refusal alongside are more likely to not report a race than to iden-
the federally mandated racial category choices, tify as white. This effect runs counter to past
 783

Table 90.1. Multinomial Regression Predicting Racial Self-Identification as White or Nonwhite Compared to None
Model  Model  Model  Model  Model 
White Nonwhite White Nonwhite White Nonwhite White Nonwhite White Nonwhite
Skin Color,
1 to 10, –.24** .01 –.21** .03 –.21** .03 –.22** .03 –.22** .06
Light–>Dark
(.05) (.07) (.05) (.07) (.05) (.07) (.05) (.07) (.05) (.07)
Cuba [Mexico] 1.45* .84 1.53* .92 1.52* .87 .88 –.23
(.60) (.78) (.60) (.79) (.61) (.79) (.65) (.87)
Dominican
–.56* –.57 –.44 –.40 –.51 –.47 –.89* –2.20**
Republic
(.28) (.47) (.29) (.48) (.29) (.49) (.41) (.61)
Central
–.11 –.21 –.03 –.13 –.03 –.12 –.13 –.30
America
(.17) (.27) (.17) (.27) (.18) (.27) (.19) (.30)
South America .55* .37 .59* .43 .59* .38 .21 –.75
(.27) (.39) (.28) (.39) (.28) (.41) (.32) (.47)
Age .01 .00 .01 .00 .00 –.00
(.01) (.01) (.01) (.01) (.01) (.01)
Married .41* .49 .42** .50* .48** .48
(.16) (.25) (.16) (.25) (.17) (.26)
Female .20 .06 .12 –.01 .14 –.01
(.15) (.23) (.16) (.24) (.16) (.25)
Earnings
–.15* –.12 –.13 –.06
(logged)
(.07) (.10) (.07) (.11)
Occ. Prestige –.00 –.00 –.00 .00
(.01) (.01) (.01) (.01)
Years of
.00 .01 .01 .04
Education
(.02) (.03) (.02) (.03)
HH < 18 –.56** –.32
(.18) (.26)
Years in United
.01 –.03
States
(.01) (.02)
English
–.47* –.60*
Proficient
(.19) (.30)
Northeast
.37 1.57**
[Southwest]
(.29) (.38)
Midwest .10 .50
(.41) (.57)
South .83* 1.19**
(.37) (.50)
West .00 –.40
(.43) (.83)
Constant 2.84** –.63 2.67** –.67+ 1.94** –1.13* 3.51** –.07 3.86** –.33
(.23) (.35) (.25) (.38) (.34) (.53) (.80) (1.16) (.83) (1.21)
Observations 1,539 1,539 1,539 1,539 1,539 1,539 1,539 1,539 1,539 1,539
Note: Standard errors in parentheses.
*p < .05; **p < .01.
784 • Part VII: Race and Ethnicity

evidence that suggests a positive relationship system based on phenotype. Here we acknowl-
between increased economic attainment and edge the multi-actor nature of racial boundary
a white racial self-classification (Tafoya 2005). formation. Not only do Latino immigrants con-
Instead, it suggests that once the effect of skin front new racial boundaries, but they may also
color is accounted for, individuals who earn be subject to effects of the existing racial bound-
more money are less likely to identify as white aries. We approach this possibility by testing
and more likely to refuse to racially identify. whether Latino immigrants experience discrim-
Model 5 adds a set of measures intended ination in annual earnings based on their racial
to evaluate whether exposure to the United phenotype. Several methodological challenges
States alters one’s racial self-identification. complicate such a test; to address them we offer
English-proficient respondents are significantly an alternative approach through propensity
more likely to not racially identify than to identify score matching with doses methodology (Lu,
as white or nonwhite. The presence of children in Hornik, and Rosenbaum 2001; Rosenbaum
a household is associated with an increased likeli- and Rubin 1983). With propensity score match-
hood of not choosing a racial category compared ing, the analysis is based on a balanced covari-
with choosing a white racial category. These find- ate distribution rather than assuming a certain
ings suggest that Latinos with more exposure to functional form of the covariates, which allows
the United States—in terms of time spent in the for a more robust inference regarding the causal
country, language facility, or exposure to schools effect than would be true with standard regres-
and other parents—are more likely to eschew fed- sion techniques. We use multivariate matching
erally mandated racial categories. with doses to conduct a more thorough test of
We find that individuals living in the South whether racial phenotype is related to earned in-
are more likely than those in the Southwest to come through the hypothesized mechanism of
choose a racial designation, either white or non- discrimination based on skin color.
white. Individuals in the Northeast have higher Matching with doses differs from the con-
odds of choosing a nonwhite racial category. ventional form of matching with treated subjects
These findings confirm that local-level interpre- and untreated controls, because all subjects are
tations of the US racial categorization scheme exposed to treatment, but the doses vary. It is
influence Latino immigrants and how they particularly useful in practical scenarios when
self-identify. Latinos in the Southwest, where a there are no clearly defined dichotomous treat-
Latino racial identity may be more salient, are ment and control groups. Instead, participants
more likely than those in the South to opt out are exposed to the treatment at numerous levels.
of traditional options. An alternative interpreta- Because we have multiple ordinal dose lev-
tion suggests that in the South, where the black/ els, we use an ordinal logit regression model
white divide has historically been more explicit, to estimate the propensity score. A propensity
Latino immigrants are more likely to identify score represents the conditional probability
racially. Residence in the Northeast, where Lati- of membership in a dose group given a vector
nos experience more external racialization as of observed covariates. Once estimated, the
black, results in a stronger identification with a propensity score is used to compare relatively
nonwhite self-classification than in the South- lighter- and darker-skinned individuals with the
west (Itzigsohn, Giorguli-Saucedo, and Vazquez same observed characteristics. We use an optimal
2005; Logan 2003). nonbipartite matching algorithm to create 477
pairs among four dose groups (Lu, Greevy, and
Xu 2008). We restrict the sample to Latinos who
Discrimination and Earned Income are members of the labor force and have valid
Next we investigate the other side of immigrants’ skin color information (N = 954).
perceptions of racial categories: whether Lati- Table 90.2 presents the results from the post-
nos are subject to effects of a racial stratification matching analysis. Comparing earned annual
 Frank, Akresh, and Lu—How Do Latino Immigrants Fit into the Racial Order? • 785

income across the matched pairs, we find an Table 90.2. Postmatching Analysis of Earned Income and
average difference of $2,435.63 between lighter- Skin Tone Among 477 Matched Pairs of Latino Immigrants
and darker-skinned individuals. This difference Group Observations Mean Income
can be interpreted to mean that, after account- Low Dose 477
$22,294.29
ing for relevant differences between any two ($20,738.48; $23,850.10)
dose groups, Latino immigrants with darker skin $19,858.66
High Dose 477
($18,329.94; $21,387.38)
earn, on average, $2,500 less per year than their
Lighter Skin–
lighter-skinned counterparts. We also explored $2,435.63
Darker Skin 954
($257.22; $4614.03)
the possibility of a dose response relationship, Contrast
but we did not find evidence that the relation-
ship was associated with dose. Instead, it appears all Latinos, however, will be defined by this
that the relative nature of racial phenotype is new boundary. Some, in particular those with
important (i.e., relatively darker-skinned indi- lighter skin, will probably be successful in their
viduals earn less than relatively lighter-skinned attempts to expand the boundary of whiteness
individuals). To the extent that this difference (Gans 1999; Lee and Bean 2007a).
represents an actual difference in earnings by Other findings, however, call into ques-
racial phenotype, net of measured premigration tion the likelihood that whiteness will expand
differences, it suggests that skin color is related again to incorporate all new Latino immigrants
to earnings via discrimination against Latino im- (Bonilla-Silva and Dietrich 2008). First, and
migrants. This finding suggests that Latino im- most important, we find evidence that Latino
migrants are not all treated equally upon arrival immigrants experience skin-color-based discrim-
in the United States; individuals with darker skin ination in the workplace. The finding that Latino
are more likely to face discrimination. immigrants, regardless of how many self-identify
as white, are not immune from penalties associ-
ated with darker skin in the United States, is a
Discussion compelling argument against the possibility that
The US racial landscape continues to undergo all newcomers will simply be accepted as white.
dramatic changes; demographers predict that Instead, we find that for Latino immigrants, skin
the Latino population will increase and the color continues to maintain its role in producing
non-Hispanic white population will become the and maintaining stratification (Montalvo and
minority in the near future. These forecasts have Codina 2001).
sparked debate over the definition of whiteness In addition, we found that respondents who
and the future of the US color line. This chapter were Dominican immigrants or who had more
focused on one dimension of these debates, ad- exposure to the United States, either by having
dressing the question of where the Latino immi- children with them in the United States, being
grant population stands in the existing US racial more fluent in English, or having higher levels
order. of economic assimilation, were more likely to
Our findings support the possibility that opt out of choosing an existing racial category
the US racial boundary system is changing. than to choose a racial category that does not fit
According to a boundary-centered approach well or is stigmatized. We interpret this finding
to racial/ethnic divisions, a social boundary is as evidence that some Latino immigrants are at-
created when two dimensions, one categorical tempting to engage in boundary change, which
and the other behavioral, coincide. Essentially, could represent a process of boundary blurring,
a social boundary forms when “ways of seeing that is, failure to choose a racial category as a way
the world correspond to ways of acting in the of emphasizing nonracial ways of belonging. Al-
world” (Wimmer 2008b, 975). Our analysis ternatively, it could also represent an attempt to
supports the possibility that a new racial bound- emphasize an alternative racial self-classification
ary is forming around Latino immigrants. Not (i.e., Latino).
786 • Part VII: Race and Ethnicity

Returning to the predictions laid out at the Bonilla-Silva, Eduardo, and David R. Dietrich.
beginning of the chapter, our findings support 2008. “The Latin Americanization of Racial
those who argue that the US racial structure is Stratification in the U.S.” In Racism in the 21st
changing. Although some Latinos will be in- Century, edited by R. E. Hall, 151–170. New
York: Springer.
cluded within the white racial boundary, our
Campbell, Mary E., and Christabel L. Rogalin.
evidence is at odds with past predictions that all 2006. “Categorical Imperatives: The Interaction
or most Latinos will be successful in their bid for of Latino and Racial Identification.” Social Sci-
whiteness (Yancey 2003). Although Latinos can ence Quarterly 87: 1030–1052.
choose their racial identification, our findings Davis, F. J. 1991. Who Is Black: One Nation’s Defini-
show that this choice is constrained by the color tion. State College: Pennsylvania State University
of their skin and skin-color-based discrimination Press.
(Golash-Boza and Darity 2008). Dowling, Julie Anne. 2004. “The Lure of Whiteness
Measuring fluctuating classification systems and the Politics of ‘Otherness’: Mexican Ameri-
can Racial Identity.” PhD diss., Department of
is by definition speculative (Bailey 2008). The
Sociology, University of Texas at Austin.
evidence presented here suggests that at present Du Bois, W. E. B. 1903. The Souls of Black Folk.
the burden of race does not fall equally on all New York: Barnes and Nobles Classics.
members of the contemporary Latino immigrant Foner, Nancy. 1987. “The Jamaicans: Race and Eth-
population (Alba 2009). Instead of a uniform ra- nicity Among Migrants in New York City.” In
cial boundary, a boundary is solidifying around New Immigrants in New York City, edited by N.
only some Latino immigrants, specifically those Foner, 195–217. New York: Columbia Univer-
with darker skin who have more experience with sity Press.
the US racial stratification system. Light-skinned Gans, Herbert J. 1999. “The Possibility of a New
Racial Hierarchy in the Twenty-first Century
Latinos and those less integrated into the United
United States.” In The Cultural Territories of Race,
States are likely to continue to test the flexibil-
edited by M. Lamont, 371–390. New York: Rus-
ity of the white racial boundary (Alba 2005). sell Sage Foundation.
Whether this strategy will prove successful, and Golash-Boza, Tanya, and William Darity. 2008.
the degree to which boundaries will become “Latino Racial Choices: The Effects of Skin Co-
stabilized and institutionalized over time, is lour and Discrimination on Latinos’ and Latinas’
necessarily left for the future. Foreign-born im- Racial Self-Identifications.” Ethnic & Racial Stud-
migrants inevitably set the stage for determining ies 31: 899–934.
how US racial boundaries will be redrawn to fit Hersch, Joni. 2008. “Profiling the New Immigrant
Worker: The Effects of Skin Color and Height.”
Latinos, but it is native-born offspring who will
Journal of Labor Economics 26: 345–386.
ultimately set the future course.
Hitlin, Steven, J. Scott Brown, and Glen H. Elder
Jr. 2007. “Measuring Latinos: Racial vs. Ethnic
REFERENCES
Classification and Self-Understandings.” Social
Alba, Richard. 2005. “Bright vs. Blurred Boundaries:
Forces 86: 587–611.
Second-Generation Assimilation and Exclusion
Hochschild, Jennifer L. 2005. “Looking Ahead: Ra-
in France, Germany, and the United States.” Eth-
cial Trends in the United States.” Daedalus (Win-
nic & Racial Studies 28: 20–49.
ter): 70–81.
———. 2009. Blurring the Color Line: The New
Itzigsohn, Jose, Silvia Giorguli-Saucedo, and Obed
Chance for a More Integrated America. Cam-
Vazquez. 2005. “Incorporation, Transnationalism,
bridge, MA: Harvard University Press.
and Gender: Immigrant Incorporation and Trans-
Bailey, Stanley R. 2008. “Unmixing for Race Mak-
national Participation as Gendered Processes.”
ing in Brazil.” American Journal of Sociology 114:
International Migration Review 39: 895–920.
577–614.
Jasso, G., D. S. Massey, M. R. Rosenzweig, and J.
Barth, Fredrik. 1969. Ethnic Groups and Boundar-
P. Smith. Forthcoming. “The U.S. New Immi-
ies: The Social Organization of Culture Difference.
grant Survey: Overview and Preliminary Results
London: George Allen & Unwin.
Based on the New-Immigrant Cohorts of 1996
 Frank, Akresh, and Lu—How Do Latino Immigrants Fit into the Racial Order? • 787

and 2003.” In Immigration Research and Statis- Abuse.” Journal of the American Statistical Associ-
tics Service Workshop on Longitudinal Surveys and ation 96: 1245–1253.
Cross-Cultural Survey Design: Workshop Proceed- Massey, Doug S., and Jennifer Martin. 2003. “The
ings, edited by B. Morgan and B. Nicholson. NIS Skin Color Scale.” Office of Population Re-
London: Crown Publishing. search, Princeton University, Princeton, NJ.
Kusow, Abdi. 2006. “Migration and Racial For- Michael, Joseph, and Jeffrey M. Timberlake. 2008.
mations Among Somali Immigrants in North “Are Latinos Becoming White? Determinants of
America.” Journal of Ethnic & Migration Studies Latinos’ Racial Self-Identification in the U.S.” In
32: 533–551. Racism in Post-Race America: New Theories, New
Landale, Nancy S., and R. S. Oropesa. 2002. “White, Directions, edited by C. A. Gallagher, 107–122.
Black or Puerto Rican? Racial Self-Identification Chapel Hill, NC: Social Forces.
among Mainland and Island Puerto Ricans.” So- Montalvo, Frank F., and G. Edward Codina. 2001.
cial Forces 81: 231–254. “Skin Color and Latinos in the United States.”
Lee, Jennifer, and Frank D. Bean. 2007a. “Re- Ethnicities 1: 321–341.
drawing the Color Line?” City & Community 6: Rodriguez, Clara E. 2000. Changing Race: Latinos,
49–62. the Census, and the History of Ethnicity in the
———. 2007b. “Reinventing the Color Line: Im- United States. New York: New York University
migration and America’s New Racial/Ethnic Di- Press.
vide.” Social Forces 86: 561–586. Rosenbaum, P. R., and D. B. Rubin. 1983. “The
Lewis, Amanda E., Maria Krysan, and Nakisha Har- Central Role of the Propensity Score in Obser-
ris. 2004. “Assessing Changes in the Meaning vational Studies for Causal Effects.” Biometrika
and Significance of Race and Ethnicity.” In The 70: 41–55.
Changing Terrain of Race and Ethnicity, edited by Tafoya, Sonya M. 2005. “Shades of Belonging: Lati-
M. Krysan and A. E. Lewis, 1–10. New York: nos and Racial Identity.” Harvard Journal of His-
Russell Sage Foundation. panic Policy 17: 58–78.
Logan, John. 2003. “How Race Counts for Hispanic Wimmer, Andreas. 2007. “How (Not) to Think
Americans.” Report, Lewis Mumford Center for about Ethnicity in Immigrant Societies.” Oxford
Comparative and Urban Regional Research, the Centre on Migration, Policy and Society Work-
University of Albany, SUNY, July 14. http:// ing Paper Series 07-44, 1–38.
mumford.albany.edu/census/BlackLatinoReport/ ———. 2008a. “Elementary Strategies of Ethnic
BlackLatino01.htm. Boundary Making.” Ethnic & Racial Studies 31
Lu, Bo, Robert Greevy, and Xinyi Xu. 2008. “Op- (6): 1025–1055.
timal Nonbipartite Matching and Its Statistical ———. 2008b. “The Making and Unmaking of
Applications.” Working Paper, Department Ethnic Boundaries: A Multilevel Process Theory.”
of Biostatistics, The Ohio State University, American Journal of Sociology 113: 970–1022.
Columbus. Yancey, George. 2003. Who Is White? Latinos, Asians,
Lu, Bo, Robert Hornik, and Paul R. Rosenbaum. and the New Black/Nonblack Divide. Boulder,
2001. “Matching with Doses in an Observa- CO: Lynne Rienner.
tional Study of a Media Campaign Against Drug
788 • Part VII: Race and Ethnicity

91 • Zhenchao Qian and Daniel T. Lichter

Are Recent Trends in Intermarriage Consistent

with Assimilation Theory?

Since the 1970s the number of racial/ethnic in- use marital prevalence data from the 1990 and
termarriages in the United States has increased 2000 Censuses to update past studies of changes
substantially. The implication is that group in intermarriage among non-Hispanic whites,
boundaries have weakened. Indeed, marriages African Americans, Asian Americans, American
between people of different racial/ethnic back- Indians, and Hispanics. We then provide addi-
grounds mean that barriers to social interaction tional analysis using marital incidence data from
and intimacy have broken down and that marital the 1980 Census and 2008 American Commu-
partners—by definition—accept each other as nity Survey.
social equals. The potential societal implications Whether racial/ethnic intermarriage con-
of intermarriage are arguably much larger. For tinued its upward trend over the past decade is
example, intermarriage commingles, to varying unclear in light of rapid demographic changes
degrees, the diverse family, friendship, and social in America (e.g., rising immigration and ed-
networks of each spouse (Romano 2003). The ucation). This chapter provides an empirical
growing population of mixed-race children from benchmark for recent trends in racial/ethnic
intermarriage also blurs racial boundaries over intermarriage and presents a conceptual frame-
successive generations (Labov and Jacobs 1998). work for interpreting intermarriage patterns in
Such trends signal improving racial/ethnic rela- light of new assimilation theory (Alba and Nee
tions, the incorporation of historically disadvan- 2003). Our principal goal and analysis focus on
taged minorities into American society, and the how rapid demographic changes in American
breakdown of persistent racial/ethnic economic society—immigration and educational attain-
and cultural distinctions (Alba and Nee 2003; ment—have influenced conventional interpre-
Bean et al. 2004). tations of changing racial/ethnic relations in
Recent increases in intermarriage should not America.
be overstated, however. In 2000 interracial mar-
riages accounted for only 6 percent of all married
couples (Simmons and O’Connell 2003). More Assimilation Theory and
important, intermarriage varies widely across ra- Intermarriage
cial/ethnic groups (Lee and Edmonston 2005; Classical assimilation theory provides a useful
Sandefur and Trudy 1986). In this chapter we conceptual framework for documenting and

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Zhenchao Qian and Daniel T. Lichter, “Social Boundaries and Marital Assimilation:
Interpreting Trends in Racial and Ethnic Intermarriage,” American Sociological Review 72 (February 2007), pp.
68–94, published by Sage Publications. The article printed here was originally prepared by Zhenchao Qian
and Daniel T. Lichter for this fourth edition of Social Stratification: Class, Race, and Gender in Sociological
Perspective, edited by David B. Grusky. Copyright © 2014 by Westview Press.
 Qian and Lichter—Are Recent Trends in Intermarriage Consistent with Assimilation Theory? • 789

interpreting long-term increases in racial/eth- intentionally seek to assimilate, [but that] the
nic intermarriage in American society (Gordon cumulative effect of pragmatic decisions aimed
1964). As defined by Park and Burgess (1969, at successful adaptation can give rise to changes
735), assimilation is “a process of interpenetra- in behavior that nevertheless lead to eventual as-
tion and fusion in which persons and groups ac- similation” (38). For example, the quest for more
quire the memories, sentiments and attitudes of education among immigrants necessarily opens
other persons and groups and, by sharing their opportunities for upward mobility and social
experience and history, are incorporated with interaction with potential white marital part-
them in a common cultural life.” This process ners, regardless of whether immigrants wish to
is arguably best exemplified in intermarriage, become “American” culturally or not. At a mini-
which is often viewed as the final step in the mum, Alba and Nee imply a continuation in the
assimilation process (Gordon 1964; Qian and upward trend of interracial marriage, albeit at a
Lichter 2001; Sassler 2005). Intermarriage with slower pace and under different circumstances
whites provides a clear signal that minority than some groups in the past experienced. They
group members have adopted cultural patterns redefine the meaning and mechanisms of assim-
of the host or majority population, such as its ilation that can occur on either side of the social
language and customs, and that they have been boundary between a minority group and the
absorbed, both economically and politically, into majority group. At the same time, recent demo-
mainstream society. graphic changes in American society have taken
Critics often question the usefulness of clas- on a momentum of their own, creating both
sical assimilation theory for explaining contem- opportunities and constraints in the marriage
porary racial/ethnic relations (Omi and Winant market. Our analyses of changing intermarriage
1994; Portes and Zhou 1992). Classical theory is incorporate the new assimilation theory in ex-
sometimes criticized as ethnocentric in its impli- amining how demographic changes, including
cations (Alba and Nee 2003); that is, it suggests immigration and educational attainment, affect
a natural progression in which immigrants jetti- intermarriage with whites differently among ra-
son maladaptive cultural repertoires and adopt cial/ethnic minorities.
a “better” way of life: the American way. This
theory also implies an inevitability of outcome
in the assimilation process. Yet distinctive phys- Demographic Sources of Change in
ical features and visibility (especially skin color) Intermarriage
make race-based social boundaries difficult to Increasing Shares of Immigrants
traverse for today’s minorities. We can no lon- Group size shapes patterns of social interaction
ger assume a simple, unilinear cultural process between groups (Blau 1977). When minority
that ultimately melds diverse racial/ethnic mi- populations become larger, opportunities for
norities and immigrants with the middle-class, intragroup contact necessarily increase, and in-
Anglo-American majority. teraction with the majority population declines.
In Remaking the American Mainstream, Alba For such minority groups (e.g., Hispanics and
and Nee (2003) reformulate classical assimi- Asian Americans), assimilation processes, in-
lation theory. They suggest that the erosion of cluding intermarriage with whites, are likely to
social distance between racial/ethnic groups, slow down. Rapid increases in the foreign-born
which culminates in intermarriage, is a two-way population have generated considerable public
process involving majority and minority pop- discourse about the cultural and economic in-
ulations. This reformulated theory takes a less corporation of immigrants and their children
ethnocentric view of the assimilation process. At into US society (Smith and Edmonston 1997).
the same time, Alba and Nee acknowledge that The concern goes beyond the immigrant pop-
America’s minorities and immigrants “may not ulation; it has become a race issue, as the racial
790 • Part VII: Race and Ethnicity

mix of immigrants has shifted over the past and socially from disadvantaged and poorly ed-
quarter century (Bean et al. 2004). Indeed, the ucated immigrants. Less educated native-born
pace of incorporation—at least with respect to minorities may be more likely than in the past
intermarriage—is likely to have decelerated with to marry their immigrant counterparts, who
increasing racial and ethnic diversity in America. comprise a growing share of the marriage pool.
In addition, recent minority immigrants are Educational attainment, however, cannot
disproportionately unskilled, have low socioeco- fully explain existing racial or ethnic differences
nomic status, and have limited English-language in intermarriage with whites. Indeed, the unique
proficiency (Bean et al. 2004; Chiswick and Sul- historical experiences and institutional arrange-
livan 1995). For new immigrants, the lack of ments of each racial/ethnic minority community
opportunities for socioeconomic mobility and may play a more fundamental role in shaping
limited social contacts with mainstream society contemporary intermarriage patterns (Nee and
magnify their social distance from whites com- Sanders 2001). African Americans and Ameri-
pared with their native-born counterparts. As a can Indians—racial minorities with the longest
result, marriages between minority immigrants history in America—are specific cases in point.
and whites are less likely to occur (Qian and African Americans are least likely of all minori-
Lichter 2001). Furthermore, the continuing in- ties to marry whites (Blackwell and Lichter
flux of immigrants has replenished the supply of 2000; Harris and Ono 2004). Although the ed-
potential partners for native-born Hispanic and ucational attainment of African Americans has
Asian American minorities (Alba and Nee 2003; increased rapidly in recent years, for various rea-
Heaton and Jacobson 2000). This suggests there sons they remain highly segregated from whites
will be less intermarriage with whites and, per- in school, neighborhoods, and the workplace.
haps, more intermarriage between natives and Racial residential segregation has declined slowly
immigrants of the same racial/ethnic group. over recent decades, but it remains much higher
for African Americans than for other minorities
(Iceland, Weinberg, and Steinmetz 2002; Neck-
Educational Upgrading erman, Carter, and Lee 1999). The legacy of past
Upward socioeconomic mobility provides mi- and current racial prejudice and discrimination
norities with new opportunities for contact with against African Americans is clearly revealed in
whites and as a result reduces social distance be- continuing physical and social isolation, which
tween minority and majority groups (Alba and has the putative effects of reinforcing racial
Nee 2003). Indeed, educational attainment, boundaries and historically low levels of inter-
a measure of socioeconomic status, is a strong marriage with whites (Omi and Winant 1994).
predictor of minority intermarriage with whites As a result, we expect that changes in educational
(Batson, Qian, and Lichter 2006; Rosenfeld attainment among African Americans will con-
2005). For Asian Americans and Hispanics, tinue to be weakly associated with interracial
intermarriage with whites increases with levels marriage with whites.
of educational attainment (Qian 1997). Strong For American Indians, educational attain-
effects of educational attainment point to the ment has played a similarly weak role in in-
obvious importance of educational composition termarriage with whites, but for very different
in shaping patterns of intermarriage between mi- reasons. To be sure, American Indians have suf-
nority and white populations. In the 1990s the fered a long and turbulent history as targets of
association between educational attainment and racial prejudice and discrimination in the United
intermarriage may have increased for Hispanics States. But unlike the situation of African Amer-
and Asian Americans, especially if highly edu- icans, American Indian–white marriages were
cated minorities separated themselves culturally tolerated (or even encouraged) for political and
 Qian and Lichter—Are Recent Trends in Intermarriage Consistent with Assimilation Theory? • 791

economic reasons (Sandefur and Trudy 1986). intermarriage is likely to have been small in
The result is that intermarriages with whites have the 1990s for African Americans and Amer-
in the past constituted a strikingly high percent- ican Indians, whose populations remain
age of all American Indian couples. Boundaries predominantly native born.
between American Indians and whites today are
weak. The racial boundary also has been blurred
by the large share of mixed-race, American Data and Methods
Indian–white individuals. The effect of educa- The data for this study come from the Inte-
tional attainment on interracial marriage is likely grated Public Use Microdata (IPUMS) 5 per-
to be weak as a result. cent samples of the 1990 and 2000 Censuses.
We distinguish among whites, African Ameri-
cans, American Indians, Asian Americans, and
A Summary of the Hypotheses Hispanics. Hispanics include individuals of any
Our fundamental goal is to update our under- race who identified themselves as being of His-
standing of contemporary patterns and recent panic/Spanish/Latino origin in the 1990 and
changes in racial/ethnic intermarriage. Our dis- 2000 Census schedules. Our sample contains
cussion and observations suggest the following only couples married at that time, which may
hypotheses: yield biased results if marital disruption rates
Hypothesis 1: The population growth of a mi- differ by marriage duration or if interracial mar-
nority population promotes in-group soli- riages are less stable than endogamous marriages
darity and enhances the likelihood of social (Jacobs and Furstenberg 1986). To reduce this
contact within the group. Rapid increases in potential bias, our analyses include only married
Hispanic and Asian American populations couples ages twenty to thirty-four at the time
through immigration during the 1990s are of each census. To reduce the number of mar-
therefore expected to have dampened rates riages contracted in other countries, we limit the
of intermarriage with whites, primarily foreign-born sample to persons who immigrated
through two processes: to the United States before age twenty. Thus,
Corollary 1a: Immigrants are much less likely mate selection patterns were shaped by marriage
to marry whites than are their native-born market conditions in the United States and are
counterparts. comparable to those of their native-born mi-
Corollary 1b: Less educated native-born mi- nority counterparts.
norities are likely to benefit more from the Log-linear models are the gold standard for
expanded marriage pools due to immigra- most statistical analyses of intermarriage patterns
tion. Thus, less educated native-born His- (Harris and Ono 2004; Qian 1997; Schwartz
panics and Asian Americans are more likely and Mare 2005). Their chief advantage is in esti-
to experience a slowdown in intermarriage mating associations between spouses’ character-
with whites than are their highly educated istics (e.g., education or race) while controlling
counterparts. for husband-wife differences in the marginal
Hypothesis 2: Because of Corollary 1b, the distributions of these characteristics. This is im-
effect of education on intermarriage with portant if the marginal distributions for spouses
whites is likely to have increased in the are different or if they have changed over time.
1990s for Hispanics and Asian Americans, We introduce two series of log-linear mod-
as the social distance between the low- and els. We first analyze the number of marriages
high-educated minority populations in- by husbands’ and wives’ race/ethnicity, nativity,
creased with the new immigrant infusion. and year. We combine race/ethnicity and nativ-
In contrast, the effect of education on ity into one variable so cross-nativity marriages
792 • Part VII: Race and Ethnicity

can be easily measured (Qian and Lichter 2001). 91.1 reveals large differences in intermarriage
The cross-tabulation has a total of 200 cells (10 among single-race, married individuals. Inter-
x 10 x 2). The basic log-linear model takes the marriage was lowest among whites (less than 5
following form: percent) and highest among American Indians
(about 60 percent) in both 1990 and 2000. To
log Fijt = G0 + G iHR + G WR
j
+ GtT + G itHRT + G WRT
jt be sure, differences in population size for each
(1) group account for part of the racial/ethnic vari-
ation in intermarriage. For example, the Asian
where Fijt is the expected number of marriages American population is much smaller in size
between husbands in race/ethnicity/nativity than the white population, which means that
i and wives in race/ethnicity/nativity j at time one Asian American-white marriage affects the
t; ȕ0 is the constant; ȕiHR (ȕjWR) denotes hus- percentage of intermarriages much more for
bands’ (wives’) race/ethnicity/nativity (i or j = Asian Americans than for whites.
non-Hispanic white, African American, Amer- Interracial marriages increased among whites
ican Indian, Asian American, and Hispanic who and African Americans, but declined during
are native born; and each of the five categories the 1990s for Hispanics, Asian Americans, and
who are foreign born). In addition to controlling American Indians. Before we conclude that this
for the marginal distributions, we also account reversal represents a clean break from the past
for the two-way interactions between race/eth- or signals a strengthening of social or cultural
nicity/nativity and year for husbands and wives, boundaries that separate these groups from
respectively (ȕitHRT, ȕjtWRT). whites, we need a better understanding of other
The second series adds educational attain- demographic shifts that may have confounded
ment. The cross-tabulation of husbands’ and statistical trends in intermarriage and perhaps
wives’ education totals 3,200 cells (10 x 10 x 4 x rendered firm conclusions ambiguous. Changes
4 x 2). The baseline model takes the form: in racial classification and the rise of cohabitation
are two sources of change in interracial marriage
, (2) (see Qian and Lichter 2007), but here we high-
light the roles of immigration and education.
The foreign born composed a larger share of
the Hispanic and Asian American populations
where Fijmnt is the expected number of marriages than of other racial groups. Among Asian Amer-
between husbands in race/ethnicity/nativity i and icans, the immigrant share increased during the
education m, and wives in race/ethnicity/nativity 1990s from 67 to 70 percent. The immigrant
j and education n at time t. In addition to the share increased more sharply for Hispanics,
parameters defined above, ȕmHE (ȕnWE) denotes hus- from 43 to 48 percent. Consequently, increasing
bands’ (wives’) educational attainment (m or n = shares of the foreign born among Asian Ameri-
no high school diploma, high school diploma, can and Hispanic populations are likely to slow
some college, and college degree and above). upward trends in intermarriage.
Our analyses also account for all two-way and For the native-born population, rising edu-
three-way interactions among race/ethnicity/ cational attainment in the 1990s was distributed
nativity, educational attainment, and year for unevenly among racial/ethnic groups. For ex-
husbands and wives, respectively (ȕimt
HRET
, ȕjnt
WRET
). ample, over two-fifths of Asian Americans, but
only a little over one-tenth of African Americans
and Hispanics, had completed a college educa-
Results tion in 2000. For the foreign-born population,
We begin by presenting evidence of changes in rising educational attainment was much less
intermarriage between 1990 and 2000. Figure apparent. Foreign-born Hispanics had lower
 Qian and Lichter—Are Recent Trends in Intermarriage Consistent with Assimilation Theory? • 793

Figure 91.1. Percentage of Intermarriages Among Married Individuals (US-Born and Immigrants
Arriving Before Age 20), Ages 20–34, 1990 and 2000

educational attainment in 2000 than in 1990, finally by African Americans. Second, intermar-
with increasing shares having a high school edu- riage with whites increased significantly among
cation or less and a declining percentage having a native-born Asian Americans and Hispanics.
college education. For Hispanics, the increase in Significantly, these multivariate results are dif-
immigration is presumably associated with some ferent from the modest declines in intermar-
erosion in economic status (or education) and riage during the 1990s reported in Figure 91.1.
perhaps with declines in rates of intermarriage Indeed, controlling for temporal changes in
with whites. marginal distributions for these two groups has
We now apply log-linear models to data on a significant impact on intermarriage patterns.
couples’ characteristics. These models control for Immigration, which has expanded the pool of
the changing marginal distributions of education marriageable partners for Asian Americans and
and other variables while isolating the indepen- Hispanics, has reduced the rate of increase in in-
dent effects of education and other variables on termarriage, but the fact that the intermarriage
each type of intermarriage. They also provide parameters increased over time suggests that “so-
evidence of whether recent declines in intermar- cial distance” between native-born Asian Amer-
riage can be attributed to changes in the share of ican and Hispanic minorities and whites would
immigrants and highly educated groups in the have narrowed in the absence of growth in the
population in the 1990s. immigrant population.
Third, foreign-born racial/ethnic minorities,
compared with their native-born counterparts,
Changes in Intermarriage have not experienced similar increases in inter-
Figure 91.2 shows the predicted number of marriage with whites. In fact, intermarriage ra-
intermarriages for one thousand endogamous tios with native-born whites declined from 27 to
marriages, controlling for marginal distribu- 22 per 1,000 for foreign-born Hispanics and in-
tions. The figure suggests several important find- creased from 28 to 31 per 1,000 for foreign-born
ings. First, intermarriage with whites was most Asians during the 1990s. Within-race marriages
likely among native-born Hispanics, followed between natives and immigrants were far more
by American Indians and Asian Americans, and common than intermarriages with whites.
794 • Part VII: Race and Ethnicity

Figure 91.2. Predicted Number of Intermarriages per 1,000 Endogamous Marriages, 1990–2000

Marriages between foreign- and native-born Americans. African American-white marriages

persons rose in the 1990s, from 160 to 240 per increased during the 1990s at every educational
1,000 among Asian Americans, and from 253 to level, but the educational gradient was com-
277 per 1,000 among Hispanics. This indicates paratively weak. Significantly, the increase was
that social distance declined between native- and greatest for those with the least education—an
foreign-born minorities during the 1990s as eth- increase of 157 percent (from 7 to 18 per 1,000)
nic identities were reinforced by the massive new among couples without a high school degree
immigration. compared to 83 percent (from 12 to 22) among
those who completed college. These data provide
little evidence that status attainment (through
Changes in Educational Assortative education alone) is a sufficient condition for in-
Mating termarriage with whites.
Table 91.1 presents the predicted number of The results reveal several additional shifts
intermarriages by educational pairings. These in intermarriage during the 1990s. For exam-
estimates reveal the educational gradient in in- ple, the educational gradient increased among
termarriage among Asian Americans and His- native-born Asian Americans and Hispanics—
panics. Among the less educated, intermarriage intermarriage with whites declined or changed
with whites is most likely for American Indians. little during the 1990s among the least educated,
Rates of interracial marriage with whites were but continued to increase among the highly edu-
similar among the least educated African Amer- cated. The boundaries between Asian Americans
icans and Asian Americans in 2000. But these and whites and between Hispanics and whites
rates were much lower for college-educated Af- are increasingly crossed at high levels of educa-
rican Americans than for college-educated Asian tional attainment. Although Hispanics are less
Table 91.1. Predicted Number of Interracial/Interethnic Marriages by Educational Combination Relative to 1,000 Endogamous Marriages

both less than high both high both some both college and one level
school school college more different
Couples 1990 2000 1990 2000 1990 2000 1990 2000 1990 2000

Native-Born White/Native-Born African American 7 18 *** 7 16 *** 10 15 *** 12 22 *** 10 20 ***

Native-Born White/Native-Born American Indian 83 62 * 80 75 66 67 85 97 * 87 87
Native-Born White/Native-Born Asian American 21 20 41 32 65 91 *** 46 62 ** 42 54 ***
Native-Born White/Native-Born Hispanic 51 47 78 85 ** 146 160 * 132 143 ** 102 112 ***

Native-Born White/Foreign-Born Black 3 3 4 4 7 9 6 8 4 7 **

Native-Born White/Foreign-Born American Indian 7 18 14 11 10 7 17 13 20 13
Native-Born White/Foreign-Born Asian 6 2 * 18 14 * 38 51 *** 34 34 27 27
Native-Born White/Foreign-Born Hispanic 12 9 * 23 18 ** 73 68 43 41 31 27 ***

Native-Born/Foreign-Born Black 44 43 107 81 * 163 170 155 139 128 122

Native-Born/Foreign-Born American Indian 59 152 92 39 57 40 124 121 83 65
Native-Born/Foreign-Born Asian 44 69 120 130 236 367 *** 132 243 *** 161 163
Native-Born/Foreign-Born Hispanic 234 222 235 263 ** 285 338 284 380 *** 272 324 ***
Note: *p < .05; **p < .01; ***p < .001 (two-tailed test of change between 1990 and 2000).
795
796 • Part VII: Race and Ethnicity

well educated than Asian Americans, intermar- Asians. Indeed, the influx of immigration from
riage rates were higher for Hispanics than for Asia and Latin America has changed the charac-
Asian Americans. teristics of the US marriage markets, as shown
The educational gradient also increased in descriptive analysis. Once this change is taken
among foreign-born Asians and Hispanics. But into account in log-linear models, we continue to
intermarriage rates declined significantly over witness an increase in intermarriage with whites
the 1990s among the least educated, while re- among Hispanics and Asian Americans. This
maining largely unchanged among those with suggests that any evidence of retreat from inter-
at least a college education. The 1990s also racial marriage is mostly a reflection of chang-
brought corresponding increases in marriage ing marriage market conditions rather than of
between native- and foreign-born Hispanics and changing mate selection choice (i.e., less willing-
Asian Americans. For Hispanics, the increase ness to marry whites, conditional on observed
was especially pronounced between the native- characteristics, such as nativity or education).
and the foreign-born with one level difference Another dominant trend is a sharp rise in mar-
in educational attainment, presumably between riages between the native- and the foreign-born
more educated foreign-born and less educated Hispanics and Asians. Clearly opportunities for
native-born Hispanics. Such findings reinforce contact between foreign- and native-born coeth-
the need for additional research on the exchange nics have increased significantly, which has ap-
of socioeconomic status for nativity in the mar- parently slowed increases in interracial marriage
riage market. with whites.
The rate of interracial marriage with whites
among African Americans was comparatively
A Coda on the Post-2000 Period low, but increased most rapidly between 1980
Our analysis of interracial marriage is based on and 2008. This indicates that the boundary be-
decennial census data from 1990 and 2000. tween African Americans and whites is eroding,
What about patterns since 2000? Unfortunately albeit slowly. Another such indication is the
the completed 2010 Census does not include effect of educational attainment on interracial
information on education and nativity, which is marriage among African Americans. Educa-
needed to replicate the preceding analysis. How- tional attainment did not play a role in 1980
ever, data from the 2008 American Community or in 1990 and 2000 (as shown previously), but
Survey can fill the void. It is the first large-scale became an important factor in 2008. In 2008,
survey since the long form of the 1980 Census when both partners had at least a college edu-
that includes information on marriage order cation the odds ratio of intermarriage between
and timing. These data, based on samples of whites and African Americans was more than
first-married couples who married in the past two times higher than when both partners had a
year for both 1980 and 2008, circumvent the high school diploma or less. Unlike in the past,
selection problems associated with analyses of racial boundaries between whites and African
all intact marriages and reflect exposure to US Americans were breaking down as more highly
marriage market conditions at the time. Be- educated men and women married interracially.
cause of space constraints, we can only provide a
brief summary of our results here (see Qian and
Lichter 2011 for details). Summary and Conclusion
Descriptive data in Qian and Lichter (2011) New assimilation theory provides a useful
show that intermarriage with whites has slowed framework for common assumptions about
among Hispanics and Asians, but the log-linear changing social boundaries and patterns of in-
analysis reveals that intermarriage with whites terracial marriage in the United States (Kalmijn
continues to increase among Hispanics and 1998; Lamont and Molnar 2002; Qian 1997).
 Qian and Lichter—Are Recent Trends in Intermarriage Consistent with Assimilation Theory? • 797

A fundamental goal of this study has been to strong educational gradient among Hispan-
reevaluate changing social boundaries among ics and Asian Americans clearly indicates that
racial/ethnic groups in the United States. Spe- their American experiences are linked to their
cifically, we have updated our understanding of educational attainment and the opportunities
marital assimilation by documenting trends and it affords them. Greater contact with whites in
patterns of recent intermarriage with whites. the workplace, school, and neighborhoods has
Our analyses provide substantial support for undoubtedly opened new opportunities to cross
our hypotheses. The 1990s generally ushered in group boundaries through marriage.
a period of slowing rates of intermarriage and For Hispanics, and to a lesser extent Asian
increasing rates of marriage between the native Americans, the 1990s brought unprecedented
and foreign born, and the educational gradient decreases in intermarriage with whites, in sharp
in intermarriage was reinforced. But our results contrast to the exceptionally large increases in
also highlight that African Americans are least intermarriage observed in prior censuses (Qian
likely to marry whites. And although the pace 1997). The retreat from intermarriage largely
of marital assimilation among African Ameri- reflects the growth in the immigrant popula-
cans proceeded more rapidly during the 1990s tion; increasing numbers of natives are marrying
and the 2000s than it had in earlier decades, their foreign-born counterparts. The substantive
the social boundaries between African Ameri- interpretation is clear: growth in size promotes
cans and whites nevertheless remain highly rigid in-group contact and interaction while reinforc-
and resistant to change. Educational attainment ing cultural and ethnic solidarity and marital
among African Americans provides no assurance endogamy. Immigration expands the marriage
of crossing the racial divide through intermar- pool of native-born counterparts and promotes
riage, although the data from 2008 suggest that endogamous marriages. Whether the 1990s rep-
race was no longer “trumping” education in the resented a short-term demographic break in the
same way it did in the past. Middle-class Afri- decades-long upward trend in marital assimila-
can Americans are less likely than Hispanics and tion or, instead, marked the beginning of a new
Asian Americans to live as children and young racial divide or accelerated cultural and racial
adults in predominantly white neighborhoods balkanization is unclear.
and attend schools with racially diverse student Corresponding increases in marriages be-
bodies (Iceland, Weinberg, and Steinmetz 2002). tween native- and foreign-born coethnics have
Racial homophily in friendship groups—even presumably been fueled by the rapid immigra-
in ostensibly integrated environments—remains tion among less educated groups. Our results
the norm in American society (Joyner and Kao imply a new kind of segmented assimilation.
2005; Moody 2001). Indeed, it is the native-born, less educated Asian
Native-born Hispanics continued to have the Americans and Hispanics who tend to share the
highest intermarriage rates with whites. Com- same pool of marriageable partners with immi-
pared with African and Asian Americans, His- grants. Educational differences in intermarriage
panics have the highest levels of intermarriage with whites have apparently exacerbated the ed-
with non-Hispanic whites at every educational ucational divide in the pace of marital assimi-
level, and race is undoubtedly a large part of lation among native-born Asian Americans and
the explanation. Almost one-half of Hispanics Hispanics. At the same time, Asian and Hispanic
consider themselves to be racially white. Asian immigrants had much lower and declining in-
Americans also have high levels of interracial termarriage rates with whites during the 1990s
marriage with whites. Although interracial (see Corollary 1a). This finding is clearly con-
marriage rates with whites are very low among sistent with the hypothesis that achieving high
less educated Asian Americans, they are excep- levels of educational attainment is an important
tionally high among well-educated Asians. The aspect of the assimilation process for Hispanics
798 • Part VII: Race and Ethnicity

and Asian Americans (see Alba and Nee 2003). Patterns of Mate Selection Among America’s Di-
Yet as we have shown, Asian American and His- verse Black Populations.” Journal of Marriage and
panic immigration has slowed the marital assim- Family 68: 658–672.
ilation process among the native born, lowered Bean, Frank D., Jennifer Lee, Jeanne Batalova, and
Mark Leach. 2004. Immigration and Fading Color
aggregate levels of intermarriage with whites,
Lines in America. New York and Washington,
and increased the socioeconomic gap in inter- DC: Russell Sage Foundation and Population
marriage with whites (see Corollary 1b). Clearly Reference Bureau.
the boundaries that separate whites from Asian Blackwell, Debra L., and Daniel T. Lichter. 2000.
Americans and Hispanics were crossed in high “Mate Selection Among Married and Cohabiting
percentages by those with exceptionally high lev- Couples.” Journal of Family Issues 21: 275–302.
els of education, but the crossing of boundaries Blau, Peter M. 1977. Inequality and Heterogeneity:
decreased among the less educated. A Primitive Theory of Social Structure. New York:
The marital assimilation story is different Free Press.
Chiswick, Barry R., and Teresa A. Sullivan. 1995.
for American Indians. The racial boundary be-
“The New Immigrants.” In State of the Union:
tween American Indians and whites appears to America in the 1990s, vol. II, Social Trends, edited
have undergone a remarkable shift over time. For by R. Farley, 211–270. New York: Russell Sage
American Indians, the long history of substan- Foundation.
tial intermarriage with whites has produced large Gordon, Milton M. 1964. Assimilation in American
numbers of mixed-race persons and has rendered Life. New York: Oxford University Press.
them phenotypically indistinguishable in many Harris, David R., and Hiromi Ono. 2004. “How
cases from native-born, European-origin whites. Many Interracial Marriages Would There Be If
Racial boundaries have blurred (Snipp 2002). All Groups Were of Equal Size in All Places? A
New Look at National Estimates of Interracial
Our results reveal a surprisingly weak effect of ed-
Marriage.” Social Science Research 34: 236–251.
ucational attainment on intermarriage. The least
Heaton, Tim B., and Cardell K. Jacobson. 2000.
educated American Indians are more likely than “Intergroup Marriage: An Examination of Op-
other less-educated minorities to marry whites. portunities.” Sociological Inquiry 70: 34–41.
These results are consistent with the views of Alba Iceland, John, Daniel H. Weinberg, and Erika
and Nee (2003), who argue that social boundar- Steinmetz. 2002. Racial and Ethnic Residential
ies that separate whites from nonwhites can shift Segregation in the United States: 1980–2000.
over time or across generations in ways that re- Washington, DC: US Government Printing
define minority “outsiders” as part of the white Office.
Jacobs, Jerry A., and Frank F. Furstenberg Jr. 1986.
majority population. Obviously one factor that
“Changing Places: Conjugal Careers and Wom-
prevents the elimination of this group boundary
en’s Marital Mobility.” Social Forces 64: 714–732.
altogether is that single-race American Indians Jacobs, Jerry A., and Teresa G. Labov. 2002. “Gen-
disproportionately live on highly segregated and der Differentials in Intermarriage Among Sixteen
geographically isolated reservations that limit op- Race and Ethnic Groups.” Sociological Forum 17:
portunities for interracial marriage with whites, 621–646.
impede upward socioeconomic mobility, and re- Joyner, Kara, and Grace Kao. 2005. “Interracial Re-
inforce traditional cultural repertoires. lationships and the Transition to Adulthood.”
American Sociological Review 70: 563–581.
REFERENCES Kalmijn, Matthijs. 1998. “Intermarriage and Ho-
Alba, Richard D., and Victor Nee. 2003. Remaking mogamy: Causes, Patterns, Trends.” Annual Re-
the American Mainstream: Assimilation and Con- view of Sociology 24: 395–421.
temporary Immigration. Cambridge, MA: Har- Labov, Teresa G., and Jerry A. Jacobs. 1998. “Pre-
vard University Press. serving Multiple Ancestry: Intermarriage and
Batson, Christie D., Zhenchao Qian, and Daniel Mixed Births in Hawaii.” Journal of Comparative
T. Lichter. 2006. “Interracial and Intraracial Family Studies 29: 481–502.
 Qian and Lichter—Are Recent Trends in Intermarriage Consistent with Assimilation Theory? • 799

Lamont, Michele, and Virag Molnar. 2002. “The Qian, Zhenchao, and Daniel T. Lichter. 2001.
Study of Boundaries in the Social Sciences.” An- “Measuring Marital Assimilation: Intermarriage
nual Review of Sociology 28: 167–195. Among Natives and Immigrants.” Social Science
Lee, Sharon M., and Barry Edmonston. 2005. New Research 30: 289–312.
Marriages, New Families: U.S. Racial and His- ———. 2007. “Social Boundaries and Marital As-
panic Intermarriage. Vol. 60, no. 2. Washington, similation: Interpreting Trends in Racial and Eth-
DC: Population Reference Bureau. nic Intermarriage.” American Sociological Review
Lichter, Daniel T., and Zhenchao Qian. 2004. Mar- 72: 68–94.
riage and Family in a Multiracial Society. New ———. 2011. “Changing Patterns of Interracial
York and Washington, DC: Russell Sage Foun- Marriage in a Multicultural Society.” Journal of
dation and Population Reference Bureau. Marriage and Family 73: 1065–1084.
Massey, Douglas S. 1995. “The New Immigration Romano, Renee C. 2003. Race Mixing: Black-White
and Ethnicity in the United States.” Population Marriage in Postwar America. Cambridge, MA:
and Development Review 21: 631–652. Harvard University Press.
Moody, James. 2001. “Race, School Integration, and Rosenfeld, Michael J. 2005. “A Critique of Exchange
Friendship Segregation in America.” American Theory in Mate Selection.” American Journal of
Journal of Sociology 107: 679–716. Sociology 110: 1284–1325.
Neckerman, Kathryn M., Prudence Carter, and Jen- Sandefur, Gary D., and McKinnell Trudy. 1986.
nifer Lee. 1999. “Segmented Assimilation and “American Indian Intermarriage.” Social Science
Minority Cultures of Mobility.” Ethnic and Ra- Research 15: 347–371.
cial Studies 22: 945–965. Sassler, Sharon. 2005. “Gender and Ethnic Differ-
Nee, Victor, and Jimy Sanders. 2001. “Understand- ences in Marital Assimilation in the Early Twen-
ing the Diversity of Immigrant Incorporation: tieth Century.” International Migration Review
A Forms-of-Capital Model.” Ethnic and Racial 39: 608–636.
Studies 24: 386–411. Schwartz, Christine R., and Robert D. Mare. 2005.
Omi, Michael, and Howard Winant. 1994. Racial “Trends in Educational Assortative Marriage
Formation in the United States: From the 1960s to from 1940 to 2003.” Demography 42: 621–646.
the 1990s. New York: Routledge. Simmons, Tavia, and Martin O’Connell. 2003.
Park, Robert E., and Ernest W. Burgess. [1921] Married-Couple and Unmarried-Partner House-
1969. Introduction to the Science of Sociology. Chi- holds: 2000. Washington, DC: U.S. Census
cago: University of Chicago Press. Bureau.
Pattillo, Mary. 2005. “Black Middle-Class Neighbor- Smith, James P., and Barry Edmonston. 1997. The
hoods.” Annual Review of Sociology 31: 305–329. New Americans: Economic, Demographic, and Fis-
Portes, Alejandro, and Min Zhou. 1992. “Gaining cal Effects of Immigration. Washington, DC: Na-
the Upper Hand: Economic Mobility Among tional Academy Press.
Immigrant and Domestic Minorities.” Ethnic and Snipp, C. Mattew. 2002. “American Indians: Clues
Racial Studies 15: 491–522. to the Future of Other Racial Groups.” In The
Qian, Zhenchao. 1997. “Breaking the Racial Bar- New Race Question: How the Census Counts Mul-
riers: Variations in Interracial Marriage Between tiracial Individuals, edited by J. Perlmann and
1980 and 1990.” Demography 34: 478–500. M. C. Waters, 189–214. New York: Russell Sage
———. 2004. “Options: Racial/Ethnic Identifica- Foundation.
tion of Children of Intermarried Couples.” Social
Science Quarterly 85: 746–766.
 PART VIII

Gender Inequality

Labor Force Participation

92 The Time Bind: When Work Becomes Home and Home
Becomes Work 803
93 The Opt-Out Revolution 807
94 Opting Out? 811

Discrimination
95 Orchestrating Impartiality: The Impact of “Blind” Auditions
on Female Musicians 820
96 Getting a Job: Is There a Motherhood Penalty? 831
97 Do Openly Gay Men Experience Employment Discrimination? 842
98 Rethinking Employment Discrimination and Its Remedies 849
99 Discrimination: Conscious or Nonconscious? 858

Sex Segregation
100 The Structure and Process of Sex Segregation 865
101 Revolving Doors: Sex Segregation and Women’s Careers 876
102 Labor Markets as Queues: A Structural Approach to Changing
Occupational Sex Composition 881
103 Glass Ceilings in Corporate Law Firms 890
104 Egalitarianism and Gender Inequality 902

Gender Gap in Wages

105 The Within-Gender Wage Gap 912
106 Devaluation and the Pay of Comparable Male and
Female Occupations 919
107 Why Do Female Occupations Pay Less? 924
108 The Sources of the Gender Pay Gap 929
 How Gender Intersects
109 Why Race, Class, and Gender Matter 942
110 Double Jeopardy 943

A Stalling Out?
111 The Gender Revolution: Uneven and Stalled 955
112 The Anti-Feminist Backlash and Recent Trends in
Gender Attitudes 965
113 The Persistence of Gender Inequality 973
 Hochschild—The Time Bind • 803

Labor Force Participation

92 • Arlie Russell Hochschild

The Time Bind

When Work Becomes Home and Home Becomes Work

The Waving Window parents. It’s called “the waving window.” Timmy
It is 6:45 a.m. on a fine June day in the midwest- dashes from the breakfast table, climbs up the
ern town of Spotted Deer. At a childcare center in wooden steps, and waits.
the basement of the Baptist church, Diane Caselli, His dad, an engineer, briskly strides past the
a childcare worker in blue jeans and loose shirt, first window toward his red Volvo parked down
methodically turns over small upended chairs the street. He stops for a moment in front of the
that rest on a Lilliputian breakfast table. She sets waving window, tilts his head, eyebrows lifted
out small bowls, spoons, napkins, and a pitcher of clownishly, then walks on without a backward
milk around a commanding box of Cheerios. The glance. Timmy returns to his cereal, sighs, and
room is cheerful, clean, half-asleep. Diane moves declares excitedly, “My Dad sawed me wave!” . . .
slowly past neatly shelved puzzles and toys, a hat At 7:40 a.m., four-year-old Cassie sidles in,
rack hung with floppy, donated dress-up hats her hair half-combed, a blanket in one hand, a
and droopy pocketbooks, a tub filled with bits of fudge bar in the other. “I’m late,” her mother ex-
colored paper. Paintings of swerving trains and plains to Diane. “Cassie wanted the fudge bar so
tipsy houses are taped to the wall. bad, I gave it to her,” she adds apologetically—
At seven, a tall, awkward-looking man peers though Diane has said nothing. Gwen Bell is a
hesitantly into the room, then ventures a few sturdy young woman, with short-cropped dark
steps forward looking for Diane. His son Timmy hair. Lightly made up and minimally adorned
tromps in behind him. Diane walks over, takes with gold stud earrings, she is neatly dressed in
Timmy’s hand, and leads him to the breakfast khaki slacks and jacket. Some Amerco mothers
table, where she seats him and helps him pour don business suits as soldiers don armor while a
cereal and milk into his bowl. Timmy’s dad, few wear floral dresses suggesting festivity and
meanwhile, hurries toward the door. leisure. But Cassie’s mother is dressed in a neu-
One wall of the room has four large windows tral way, as if she were just getting the job of
that overlook a sidewalk. In front of the second self-presentation done.
window is a set of small wooden steps children “Pleeese, can’t you take me with you?” Cassie
climb to wave good-bye to their departing pleads.

Arlie Russell Hochschild. Adapted from “The Waving Window” in The Time Bind: When Work Becomes Home and
Home Becomes Work, pp. 3–5, 11–14. Copyright © 1997 by Arlie Russell Hochschild. Reprinted by permission
of Henry Holt and Company, LLC. And from “Time in the Balance,” The Nation, Vol. 264, No. 20, May 26,
1994. Copyright © 1994 by Arlie Russell Hochschild. Reprinted by permission of Georges Borchardt, Inc. on
behalf of the author.
804 • Part VIII: Gender Inequality

“You know I can’t take you to work,” Gwen and children to yell. Marie ushers him away to
replies in a tone that suggests she’s heard this re- dictate a note to his mother explaining that it
quest before. Cassie’s shoulders droop in defeat. hurts his feelings when she doesn’t wave.
She’s given it a try, but now she’s resigned to her Cassie still stands at the front door holding
mother’s imminent departure, and she’s agreed, her fudge bar like a flag, the emblem of a truce in
it seems, not to make too much fuss about it. a battle over time. Every now and then, she licks
Aware of her mother’s unease about her long day one of its drippy sides while Diane, uncertain
at childcare, however, she’s struck a hard bargain. about what to do, looks on disapprovingly. The
Every so often she gets a morning fudge bar. This cereal eaters watch from their table, fascinated
is their deal, and Cassie keeps her mother to it. and envious. Gwen Bell turns to leave, waving
As Gwen Bell later explained to me, she contin- goodbye to Cassie, car keys in hand. By our prior
ually feels that she owes Cassie more time than arrangement, I am her shadow for this day, and
she actually gives her. She has a time-debt to her I follow her out the door and into the world of
daughter. If many busy parents settle such debts Amerco.
on evenings or weekends when their children ea- Arriving at her office, as always, exactly at
gerly “collect” promised time, Cassie insists on a 7:50 a.m., Gwen finds on her desk a cup of cof-
morning down payment, a fudge bar that makes fee in her personal mug, milk no sugar (exactly
her mother uneasy but saves her the trouble and as she likes it), prepared by a coworker who has
embarrassment of a tantrum. Like other parents managed to get in ahead of her. The remaining
at the center, Gwen sometimes finds herself in- half of a birthday cake has been left on a table in
dulging her child with treats or softened rules in the hall outside her office by a dieting coworker
exchange for missed time together. Diane speaks who wants someone else to eat it before she does.
quietly to Cassie, trying to persuade her to stop Gwen prepares materials for her first meeting
sulking and join the others. (having e-mailed messages to the other partic-
The center works on “child time.” Its ipants from home the previous night), which
rhythms are child-paced, flexible, mainly slow. will inaugurate her official 8 a.m. to 5:45 p.m.
Teachers patiently oversee the laborious task of workday. As she does so, she nibbles at a sliver of
tying a shoelace, a prolonged sit on the potty, the the cake while she proofreads a memo that must
scrambled telling of a tall tale. In this and other be Xeroxed and handed out at a second meeting
ways it is an excellent childcare center, one of a scheduled for 9 a.m.
dozen islands of child time I was to discover in As the assistant to the head of the Public
my three summers of field research at Amerco, Relations Office, Gwen has to handle Amerco’s
a Fortune 500 company headquartered in Spot- responses to any press reports that may appear
ted Deer.1 Scattered throughout the town, such about the company. This time the impending
islands—a playground, a pediatrician’s wait- media attention is positive, and her first meet-
ing room, the back of a family van—stand out ing has been called to discuss how to make the
against the faster paced, more bureaucratically most of it. As the members of the publicity team
segmented blocks of adult work time. . . . straggle into her office and exchange friendly
At the Spotted Deer Childcare Center, the greetings, she sighs. She’s ready.
group of young breakfasters gradually expands, Gwen loves her job at Amerco, and she is very
early arrivals watching the entertainment pro- good at it. Whatever the daily pressures, she also
vided by yet more newcomers. Sally enters suck- feels remarkably at home there. Her boss, a man
ing her thumb. Billy’s mother carries him in even who she says reminds her of the best aspects of
though he’s already five. Jonathan’s mother for- her father, helps her deal with work problems
gets to wave, and soon after, Jonathan kicks the and strongly supports her desire to rise in the
breakfast table from below, causing milk to spill company. In many ways, her “Amerco dad” has
 Hochschild—The Time Bind • 805

been better to her than her own father, who, home and returns earlier than his wife. He has
when she was small, abruptly walked out on her eaten an ample lunch, knowing that they usu-
mother and her. She feels lucky to have such a ally have a late, light dinner, but this evening
caring boss, and working for him has reinforced he’s hungrier than he means to be. He plays with
her desire to give her all to work—insofar as she Cassie while Gwen makes dinner.
can. She and her husband need her salary, she To protect the dinner “hour”—8:00 to
tells me, since his job, though more lucrative 8:30—Gwen checks that the phone machine
than hers, is less steady. is on, and we hear a series of abbreviated rings
Gradually, over the last three years Gwen’s several times during dinner. John says grace, and
workday has grown longer. She used to work we all hold hands. It is time, it seems, to let go
a straight eight-hour day. Now it is regularly and relax. Right after dinner, though, it’s Cas-
eight and a half to nine hours, not counting sie’s bath time. Cassie both loves and protests her
the work that often spills over into life at home. nightly bath. Gwen has come to expect Cassie to
Gwen is not happy about this. She feels Cassie’s dawdle as she undresses or searches for a favorite
ten-hour day at Spotted Deer is too long, but bath toy, trying to make a minivacation of it.
at the same time she is not putting energy into Gwen lets Cassie linger, scans through her phone
curbing her expanding workday. What she does messages, and sets them aside.
do is complain about it, joke about it, com- At 9 p.m., the bath over, Gwen and Cassie
pare stories with friends at work. Hers are not have “quality time” or “QT,” as John affection-
the boastful “war stories” of the older men at ately calls it. This they see as their small castle of
Amerco who proudly proclaim their ten-hour time protected from the demands of the outside
workdays and biweekly company travel sched- world. Half an hour later, at 9:30 p.m., Gwen
ules. Rather, Gwen’s stories are more like situa- tucks Cassie into bed.
tion comedies: stories about forgetting to shop Later, as Gwen and John show me around
and coming home to a refrigerator containing their home, John points out in passing an expen-
little more than wilted lettuce and a jar of ol- sive electric saw and drill set he bought two years
ives, stories told in a spirit of hopeless amuse- earlier with the thought of building a tree house
ment. Gwen is reasonably well informed about for Cassie, a bigger hutch for her rabbit Max,
Amerco’s flextime and reduced-hours policies, and a guest room for visiting friends. “I have
which are available to white-collar employees the tools,” John confides. “I just don’t have the
like her. But she has not talked with her boss time to use them.” Once, those tools must have
about cutting back her hours, nor have her jok- represented the promise of future projects. Now
ing coworkers, and her boss hasn’t raised the they seemed to be there in place of the projects.
possibility himself. There is just so much to get Along with the tools, perhaps John has tried
done at the office. to purchase the illusion of leisure they seemed
At 5:45 p.m. on the dot, Gwen arrives back to imply. Later, as I interviewed other working
at Spotted Deer. Cassie is waiting eagerly by the parents at Amerco, I discovered similar items
door, her coat over her arm, a crumpled picture banished to attics and garages. Timmy’s father
she has drawn in her hand. Gwen gives Cassie a had bought a boat he hadn’t sailed all year. Jarod
long, affectionate hug. By the time Gwen and and Tylor’s parents had a camper truck they had
Cassie roll into the driveway of their two-story hardly driven. Others had cameras, skis, guitars,
white frame house, surrounded by a border of encyclopedia sets, even the equipment to harvest
unruly shrubs, it is 6:25 p.m. John Bell is already maple syrup, all bought with wages that took
there, having shopped, taken the messages off time to earn.
the phone machine, set the table, and heated the John’s tools seemed to hold out the promise
oven. This is one of two days a week he leaves of another self, a self he would be “if only I had
806 • Part VIII: Gender Inequality

time.” His tools had become for him what Cas- It’s a little less clear than it once was who’s sup-
sie’s fudge bar was for her—a magical substitute posed to take care of whom, how much and for
for time, a talisman. how long. So we’ve grown more anxious. Given
There were, in a sense, two Bell households: our tradition of individualism, many of us feel
the rushed family they actually were and the re- alone in this anxiety, and insofar as work is our
laxed family they imagined they could be if only rock, we cling to it.
they had time. Gwen and Bill complained that In this context, the idea of cutting back the
they were in a time bind: they wanted more time workday is an idea that seems to have died, gone
for life at home than they had. It wasn’t that they to heaven and become an angel of an idea. We
wanted more small segments of “quality time” dream about it, but it’s something we’d never re-
added into their over-busy days. They wanted ally expect to do. No matter how a movement
a quality life, and Gwen, at least, worked for a for work-family balance is structured, in the
family-friendly company whose policies seemed long run, no such balance will ever take hold
to hold out hope for just that. So what was pre- if the social conditions that would make it pos-
venting them from getting it? sible—men who are willing to share parenting
and housework, supportive communities, and
policy-makers and elected officials who are pre-
Time in the Balance pared to demand family-friendly reforms—re-
We’ve gotten ourselves into a time bind. Feeling main out of reach. . . .
that we are always late, having no free time, try- As my study of Amerco has shown, however,
ing to adapt as best we can to the confines of our even when the jobs of working parents are secure,
time prisons—these are all symptoms of what pay a sufficient wage and provide family-friendly
has become a national way of life. There are sev- programs, many working parents are still reluc-
eral reasons for this. Over the past two decades, tant to spend more time at home. American fa-
global competition and inflation have lowered thers spend less time at home than mothers do,
the buying power of the male wage. In response, expand their work hours when children are small
many women have gone to work in order to and, if Amerco is any indication, are reluctant
maintain the family income. But the legacy of to take paternity leaves. We know from previous
patriarchy has given cultural shape to this eco- research that many men have found a haven at
nomic story. As women have joined men at work. This isn’t news. The news is that growing
work, they have absorbed the views of an older, numbers of working women are leery of spending
male-oriented work world—its views of “an hon- more time at home as well. They feel guilty and
est day’s work”—at a much faster rate than men stressed out by long hours at work, but they feel
have absorbed their share of domestic work and ambivalent about cutting back on those hours.
culture. One reason women have changed more Women fear losing their place at work; hav-
than men is that the world of “male” work seems ing such a place has become a source of secu-
more honorable and valuable than the “female” rity, pride and a powerful sense of being valued.
world of home and children. As a survey conducted by Bright Horizons (a
There is another factor too—we are increas- Boston-based company that runs on-site day-
ingly anxious about our “culture of care.” Where care centers in twenty-three states) indicates,
do we turn when we are down and out? This is a women are just as likely to feel appreciated as
question we ask, understandably, even when we men at the workplace; as likely as men to feel
are up and in. With recent welfare “reform,” the underappreciated at home; and even more likely
government is cutting off aid to women and the than men to have friends at work. To cut back
poor. With the growth of the benefit-less contin- on work hours means risking loosening ties to a
gency labor force, many corporations are doing world that, tension-filled as it is, offers insurance
the same thing to middle-class men. Meanwhile, against even greater tension and uncertainty at
modern families have grown more ambiguous: home. For a substantial number of time-bound
 Belkin—The Opt-Out Revolution • 807

working parents, the stripped-down home and adjust to or resist the time bind—and all of its
the community-denuded neighborhood are sim- attendant consequences—more or less on their
ply losing out to the pull of the workplace. own. It is unsettling because while children
Many women are thus joining men in a flight remain precious to their parents, the “market
from the “inner city” of home to the “suburbs” of value” of the world in which they are growing up
the workplace. In doing so, they have absorbed has declined drastically. One need not compare
the views of an older, male-oriented work world their childhoods with a perfect childhood in a
about what a “real career” and “full commitment mythical past to conclude that our society needs
to the job” really mean. Women now make up to face up to a serious problem.
nearly half the labor force. The vast majority of
them need and want to be there. There is defi- NOTES
nitely no going back. But women have entered 1. To protect the privacy of the people in this
chapter, I have given the company a fictional name
the workplace on “male” terms. It would be less
and declined to specify what its workers produce or
problematic for women to adopt a male model
where they live. I’ve also altered the names, occu-
of work—to finally enjoy privileges formerly re- pations, and defining details of the personal lives
served for men—if the male model of work were of individuals. I have, however, tried to document
one of balance. But it is not. as accurately as possible their experiences of life at
All this is unsettling news, in part because home and at work and to capture the essence of the
the children of working parents are being left to culture that infuses both worlds.

93 • Lisa Belkin

The Opt-Out Revolution

The scene in this cozy Atlanta living room expected to do. They received law degrees from
would—at first glance—warm an early feminist’s Harvard and Columbia. They chose husbands
heart. Gathered by the fireplace one recent eve- who could keep up with them, not simply sup-
ning, sipping wine and nibbling cheese, are the port them. They waited to have children because
members of a book club, each of them a bene- work was too exciting. They put on power suits
ficiary of all that feminists of 30-odd years ago and marched off to take on the world.
held dear. Yes, if an early feminist could peer into this
The eight women in the room have each scene, she would feel triumphant about the fu-
earned a degree from Princeton, which was ture. Until, of course, any one of these polished
a citadel of everything male until the first and purposeful women opened her mouth.
co-educated class entered in 1969. And after “I don’t want to be on the fast track leading
Princeton, the women of this book club went to a partnership at a prestigious law firm,” says
on to do other things that women once were not Katherine Brokaw, who left that track in order to

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Lisa Belkin, “The Opt-Out Revolution,” The New York Times Magazine (October
26, 2003, Section 6), pp. 42, 44–46, 85–86. The article printed here was originally prepared by Lisa Belkin
for Social Stratification: Class, Race, and Gender in Sociological Perspective, Third Edition, edited by David B.
Grusky, pp. 735–738. Copyright © 2008 by Westview Press.
808 • Part VIII: Gender Inequality

stay home with her three children. “Some people than it is about the obstacles faced by mothers.
define that as success. I don’t.” As Joan C. Williams, director of the Program on
“I don’t want to be famous; I don’t want to WorkLife Law at American University, wrote in
conquer the world; I don’t want that kind of the Harvard Women’s Law Journal last spring,
life,” says Sarah McArthur Amsbary, who was a “Many women never get near” that glass ceiling,
theater artist and teacher and earned her master’s because “they are stopped long before by the ma-
degree in English, then stepped out of the work ternal wall.”
force when her daughter was born. “Maternity Look, for example, at the Stanford class of
provides an escape hatch that paternity does not. ’81. Fifty-seven percent of mothers in that class
Having a baby provides a graceful and conve- spent at least a year at home caring for their
nient exit.” infant children in the first decade after gradua-
Wander into any Starbucks in any Starbucks tion. One out of four have stayed home three or
kind of neighborhood in the hours after the com- more years. Look at Harvard Business School.
muters are gone. See all those mothers drink- A survey of women from the classes of 1981,
ing coffee and watching over toddlers at play? 1985 and 1991 found that only 38 percent were
If you look past the Lycra gym clothes and the working full time. Look at professional women
Internet-access cell phones, the scene could be in surveys across the board. Between one-quarter
the ’50s, but for the fact that the coffee is more and one-third are out of the work force, depend-
expensive and the mothers have M.B.A.s. . . . ing on the study and the profession. Look at
Arguably, the barriers of 40 years ago are the United States Census, which shows that the
down. Fifty percent of the undergraduate class number of children being cared for by stay-at-
of 2003 at Yale was female; this year’s graduat- home moms has increased nearly 13 percent in
ing class at Berkeley Law School was 63 percent less than a decade. At the same time, the per-
women; Harvard was 46 percent; Columbia was centage of new mothers who go back to work fell
51. Nearly 47 percent of medical students are from 59 percent in 1998 to 55 percent in 2000.
women, as are 50 percent of undergraduate busi- Look, too, at the mothers who have not left
ness majors (though, interestingly, about 30 per- completely but have scaled down or redefined
cent of M.B.A. candidates). They are recruited their roles in the crucial career-building years (25
by top firms in all fields. They start strong out to 44). Two-thirds of those mothers work fewer
of the gate. than 40 hours a week—in other words, part
And then, suddenly, they stop. Despite all time. Only 5 percent work 50 or more hours
those women graduating from law school, they weekly. Women leave the workplace to strike out
comprise only 16 percent of partners in law on their own at equally telling rates; the num-
firms. Although men and women enter corpo- ber of businesses owned or co-owned by women
rate training programs in equal numbers, just jumped 11 percent since 1997, nearly twice the
16 percent of corporate officers are women, and rate of businesses in general.
only eight companies in the Fortune 500 have Look at how all these numbers compare with
female C.E.O.s. Of 435 members of the House those of men. Of white men with M.B.A.s, 95
of Representatives, 62 are women; there are 14 percent are working full time, but for white
women in the 100-member Senate. women with M.B.A.s, that number drops to
Measured against the way things once were, 67 percent. (Interestingly, the numbers for
this is certainly progress. But measured against African-American women are closer to those for
the way things were expected to be, this is a white men than to those for white women.)
revolution stalled. During the ’90s, the talk was And look at the women of this Atlanta book
about the glass ceiling, about women who were club. A roomful of Princeton women each
turned away at the threshold of power simply be- trained as well as any man. Of the 10 members,
cause they were women. The talk of this new de- half are not working at all; one is in business
cade is less about the obstacles faced by women with her husband; one works part time; two
 Belkin—The Opt-Out Revolution • 809

freelance; and the only one with a full-time job chosen to leave their high-powered jobs, most
has no children. voluntarily, for lives that are less intense and
Social scientists—most of them women— more fulfilling.
have made a specialty in recent years of studying It’s why President Bush’s adviser Karen
why all this is so. Joan Williams (“Unbending Hughes left the White House, saying her family
Gender”), Sylvia Ann Hewlett (“Creating a was homesick and wanted to go back to Austin.
Life”), Arlie Russell Hochschild (who coined It’s why Brenda C. Barnes, who was the presi-
the phrase “the second shift”) and Felice N. dent and C.E.O. of Pepsi-Cola North America,
Schwartz (who made popular the phrase “the left that job to move back to Illinois with her
mommy track”), to name just a few, have done family. And it’s why Wendy Chamberlin, who
important work about how the workplace has was ambassador to Pakistan, resigned, because
failed women. security concerns meant she never saw her two
But to talk to the women of the book club— young daughters.
or to the women of a San Francisco mothers’ Why don’t women run the world?
group with whom I also spent time, or the doz- Maybe it’s because they don’t want to.
ens of other women I interviewed, or the count- Attitudes cluster in place and time. This is
less women I have come to know during the four particularly true of a college campus, where
years I have reported on the intersection of life one-quarter of the student population turns
and work—is to sense that something more is over every year. Undergraduates tend to think
happening here. It’s not just that the workplace that the school they find is the one that always
has failed women. It is also that women are re- was, with no knowledge of the worldview of
jecting the workplace. . . . those even a few short years before. Looked at
As these women look up at the “top,” they that way, the women of the Atlanta book club
are increasingly deciding that they don’t want are a panoramic snapshot of change.
to do what it takes to get there. Women today Sally Sears, the oldest of the group, entered
have the equal right to make the same bargain Princeton in the fall of 1971. Women had been
that men have made for centuries—to take time fully admitted two years earlier, and the school
from their family in pursuit of success. Instead, was still very much a boys club. Sears had gone
women are redefining success. And in doing so, to a small public school in Alabama and entered
they are redefining work. college “very conscious of being a representative
Time was when a woman’s definition of success of women and a representative of the South.” As
was said to be her apple-pie recipe. Or her hus- she describes it the air was electric with femi-
band’s promotion. Or her well-turned-out chil- nism. “Margaret Mead came to talk one night,
dren. Next, being successful required becoming and I was stunned by how penetrating her ques-
a man. Remember those awful padded-shoulder tions were about what it was like to be the first
suits and floppy ties? Success was about the male women,” she says. “I thought, my God, she’s
definition of money and power. thinking of us as Samoans.”
There is nothing wrong with money or Upon graduation in 1975, Sears felt both en-
power. But they come at a high price. And lately titled and obligated to make good. “The clear
when women talk about success they use words message was, ‘You’ve been given the key to a
like satisfaction, balance and sanity. kingdom that used to be denied to people like
That’s why a recent survey by the research you,’” she says. “It never occurred to me that my
firm Catalyst found that 26 percent of women choices would be proscribed. I could have any-
at the cusp of the most senior levels of manage- thing I wanted.”
ment don’t want the promotion. And it’s why What she wanted, at first, was to be “a con-
Fortune magazine found that of the 108 women firmed single person, childless, a world traveler.”
who have appeared on its list of the top 50 most She spent a couple of years running The Childers-
powerful women over the years, at least 20 have burg Star, a small Alabama newspaper owned by
810 • Part VIII: Gender Inequality

her family, and then, in 1978, she took a job college, she says, she gave no thought to melding
on the air at a television station in Birmingham. life and work, but now, “I think about it almost
That led to a job in Memphis, followed by a constantly.”
yearlong trip around the world, then another And what she has concluded, after all this
TV job in Dallas. By 1984 she was on the air in thinking, is that the exodus of professional
Atlanta, where she became a local celebrity and women from the workplace isn’t really about
where she met Richard Belcher, a fellow reporter motherhood at all. It is really about work.
and now a local anchor. They were married in “There’s a misconception that it’s mostly a pull
1988, when Sears was 35. toward motherhood and her precious baby that
Three years later, their son, Will, was born. drives a woman to quit her job, or apparently,
Soldiers of feminism take only the shortest of her entire career,” she says. “Not that the pre-
maternity leaves, and as soon as Sears recovered cious baby doesn’t magnetize many of us. Mine
from her C-section she was back at work. The certainly did. As often as not, though, a woman
O.J. Simpson trial was the first real test of what would have loved to maintain some version of
she calls “work plus love plus a child,” because a career, but that job wasn’t cutting it anymore.
both she and her husband were sent out to Cali- Among women I know, quitting is driven as
fornia for the duration. “I got my mom and dad much from the job-dissatisfaction side as from
to bring Will out, and we all camped out at the the pull-to-motherhood side.”
New Otani Hotel for a few weeks,” she says. “I She compares all this to a romance gone sour.
was determined not to blink.” . . . “Timing one’s quitting to coincide with a baby
Sears took nine years to quit. And she did so is like timing a breakup to coincide with gradu-
with great regret. “I would have hung in there, ation,” she says. “It’s just a whole lot easier than
except the days kept getting longer and longer,” breaking up in the middle of senior year.”
she explains. “My five-day 50-hour week was That is the gift biology gives women, she
becoming a 60-hour week.” As news reports says. It provides pauses, in the form of pregnancy
could be transmitted farther and farther from and childbirth, that men do not have. And as
the “mother ship,” she found herself an hour the workplace becomes more stressful and
or two from home when the nightly news was all-consuming, the exit door is more attractive.
done. “Will was growing up, and I was driving “Women get to look around every few years and
home from a fire,” she says. “I knew there would say, ‘Is this still what I want to be doing?’” she
always be wrecks and fires, but there wouldn’t says. “Maybe they have higher standards for job
always be his childhood.” satisfaction because there is always the option
First she tried to reduce her schedule. “The of being their child’s primary caregiver. When
station would not give me a part-time con- a man gets that dissatisfied with his job, he has
tract,” she says. “They said it was all or noth- to stick it out.”
ing.” So in August 2000, she walked away from This, I would argue, is why the workplace
her six-figure income and became a homeroom needs women. Not just because they are 50 per-
mom at her son’s school. cent of the talent pool, but for the very fact that
“It was wrenching for me to leave Channel they are more willing to leave than men. That,
2,” she says. “I miss being the lioness in the in turn, makes employers work harder to keep
newsroom—to walk through and have the in- them. It is why the accounting firm Deloitte &
terns say, ‘There she goes.’ It kills me that I’m Touche has more than doubled the number of
not contributing to my 401(k) anymore. I do employees on flexible work schedules over the
feel somehow that I let the cause down.” . . . past decade and more than quintupled the num-
Sarah McArthur Amsbary of the Atlanta ber of female partners and directors (to 567,
group leads a much-examined life. Back in from 97) in the same period. It is why I.B.M.
 Percheski—Opting Out? • 811

employees can request up to 156 weeks of to act like men, men are being freed to act like
job-protected family time off. It is why Hamot women. Because women are willing to leave,
Medical Center in Erie, Pa., hired a husband and men are more willing to leave, too—the num-
wife to fill one neonatology job, with a shared ber of married men who are full-time caregivers
salary and shared health insurance, then let them to their children has increased 18 percent. Be-
decide who stays home and who comes to the cause women are willing to leave, 46 percent of
hospital on any given day. It is why, everywhere the employees taking parental leave at Ernst &
you look, workers are doing their work in untra- Young last year were men.
ditional ways. Looked at that way, this is not the failure of a
Women started this conversation about life revolution, but the start of a new one. It is about
and work—a conversation that is slowly com- a door opened but a crack by women that could
ing to include men. Sanity, balance and a new usher in a new environment for us all.
definition of success, it seems, just might be Why don’t women run the world?
contagious. And instead of women being forced “In a way,” Amsbary says, “we really do.”

94 • Christine Percheski

Opting Out?

Women’s participation in paid employment has as “male” and free of personal responsibilities
increased substantially in the United States over (Acker 1990). Many working women with chil-
the past fifty years, particularly among married dren consequently experience significant dif-
women and mothers of young children. This ficulties because of the competing demands of
occurred without a substantial reorganization work and family life. This has prompted ques-
of work or family life. Although men perform tions about the sustainability of high employ-
more housework and child care than in the past, ment rates for women. In this chapter I consider
they still spend on average far less time engaged whether recent cohorts of women in professional
in these activities than women do, leaving the and managerial occupations are increasing or
gendered nature of family life and childrearing maintaining high employment rates or are “opt-
essentially intact. The nature of paid work also ing out” of professional employment to stay at
remains fundamentally unchanged. Despite the home with children.
expansion of maternity leave policies and flexible Whether US women will maintain high em-
work schedules, the organization of most work- ployment levels has sparked interest outside of
places is still predicated on a concept of workers academia. Mainly relying on anecdotal evidence,

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Christine Percheski, “Opting Out? Cohort Differences in Professional Women’s
Employment Rates from 1960 to 2005,” American Sociological Review 73 (June 2008), pp. 497–517, published
by Sage Publications. The article printed here was originally prepared by Christine Percheski for this fourth
edition of Social Stratification: Class, Race, and Gender in Sociological Perspective, edited by David B. Grusky.
Copyright © 2014 by Westview Press. Used with permission of the author.
812 • Part VIII: Gender Inequality

many media outlets in the United States have led to a decline in women’s employment rates
run stories either predicting or claiming an ex- across cohorts, including a conservative cultural
odus of women from professional work (Wil- shift, increased time demands of professional
liams, Manvell, and Bornstein, 2006). One of occupations, more marriage opportunities
these articles made a particularly controversial for educated women, and a shift toward more
claim: a New York Times Magazine article by mother-intensive parenting practices.
Belkin (2003) describes an “opt-out revolution” Cohort change can be thought of as hav-
among highly educated professional women and ing two sources: changes in the composition
asserts that women’s voluntary employment exits of the population and changes in the behavior
to accommodate childrearing account for per- of subgroups of the population. For example,
sistent gender inequalities in employment. an increase across cohorts in the percentage of
Media attention to “opting out” centers on professional women who are mothers—a com-
highly educated women in professional and positional change—could decrease employment
managerial occupations, and perhaps for good rates. Employment rates could similarly de-
reason. The expansion of women’s educational crease if the percentage of mothers stayed the
opportunities has yielded more women qualified same while the employment rates of mothers
for historically male-dominated jobs that require decreased—a behavioral change. The changes
advanced schooling. Because professional and outlined previously may have impacted the com-
managerial occupations confer prestige, social positional characteristics of professional women,
influence, and economic rewards, women’s suc- their behaviors, or both.
cess in these fields may be particularly important The evidence on cohort change among
for gender equality. women in professional and managerial occupa-
This chapter describes cohort changes in tions is limited; in 2008 I found just one study
employment patterns of college-educated pro- of cohort employment patterns for this group
fessional and managerial women in the United using nationally representative data. Whitting-
States from 1960 to 2010 and evaluates whether ton, Averett, and Anderson (2000) examine
recent cohorts are increasingly “opting out” of postpartum employment patterns among mar-
paid employment. ried mothers using data from the Panel Study of
Income Dynamics (PSID) from 1968 to 1992.
They do not find significant cohort differences
Theory and Previous Research in models controlling for period effects, but their
Widespread social change—such as the expan- sample size is small, and their model specifica-
sion of women’s employment opportunities—is tion may be inadequate.
experienced differently by birth cohorts who Two other studies are particularly relevant
mature under different historical circumstances to this research question. Goldin’s (2006) anal-
and are at different stages in the life cycle when ysis of women born around 1958 who attended
change occurs (Ryder 1965). Many societal selective colleges finds relatively small employ-
changes over the past fifty years may have con- ment differences between mothers and childless
tributed to employment differences across birth women. In addition, less than half of the moth-
cohorts among professional women. Changes ers in the sample were out of the labor force for
that may have increased employment rates in- more than six months at a time. Another cohort
clude the expansion of employment opportuni- study considers the fertility and employment
ties and legal protections for women, women’s patterns of college-educated women born from
increased control over their fertility owing to 1960 to 1979 between ages twenty-two and
technological and legal changes, and an increase twenty-seven. Vere (2007) finds modest de-
in the earnings premiums for higher education creases in cumulative annual employment hours
and professional work. Other changes may have for single-year cohorts born after 1974 and
 Percheski—Opting Out? • 813

increases in fertility among cohorts born after Occupational characteristics. The historical gen-
1966, but this analysis cannot show whether der composition or sex-type of an occupation
employment decreases were concentrated among captures a broad array of factors that may in-
women with children. fluence employment rates, including prestige,
working conditions, and salary. In the category
of historically male professions, I include doc-
Data and Methods tors and physicians, lawyers and judges, engi-
Data neers, accountants, college professors and deans,
I use the Integrated Public Use Microdata Se- clergy, and dentists. In the category of histori-
ries (IPUMS) Census data for the 1960 through cally female professions are nurses, social work-
2000 Censuses and American Community Sur- ers, teachers, and librarians.
vey (ACS) data for 2005 and 2010 (Ruggles et
al. 2004). (The original analysis published in the Family characteristics. In this analysis I consider
American Sociological Review in 2008 included whether a woman has children and the age of
ACS data from 2005; I have used data from her youngest child. I classify a woman who re-
2010 instead of 2005 for most of this updated ports her youngest child’s age as five or younger
analysis.) My sample includes college-educated as “having young children” and a woman whose
women ages twenty-five to fifty-four in profes- youngest child is between the ages of six and
sional or managerial occupations (hereafter re- eighteen as “having older children.” Women who
ferred to as “professional women”) in the 1960 do not report any of their own children in the
General, 1970 Form 2 Metro, 1980 1% Metro household are classified as “without children.”
B, 1990 1% Metro, 2000 1% Census, and 2005 This operational definition is likely to slightly
and 2010 ACS IPUMS samples. For some anal- underestimate maternal status.
yses I consider only women in the prime child-
bearing years (ages twenty-five to thirty-nine).
Analysis Plan
Professionals and managers. I define professional To examine changes in professional women’s
women as those with a college degree (or four employment over time, I use a cohort analysis
years of college) and an occupation classified as approach. First, I show how age-specific rates
professional or managerial by the Census Bureau. of LFP, FTYR, and working long hours have
Specifically, I use the IPUMS code occ1950. changed by cohort. This part of the analysis is
similar to Sayer, Cohen, and Casper’s (2004) co-
Birth cohorts. I divide the sample into seven hort analysis; my analysis differs primarily in the
ten-year cohorts, starting with the birth year population examined and the number of cohorts
1906 and ending with the year 1975. I use the considered. Second, I track changes in the com-
cohort parameters defined by Sayer, Cohen, and position of professional women across cohorts
Casper (2004) to facilitate comparisons with and in employment rates among subgroups, de-
their research on women’s employment. The fined by family characteristics and occupation
final sample, which includes observations from sex-type. To calculate employment rates for the
1960 to 2010, contains 370,713 observations. main childbearing years, I compute single-year,
age-specific rates and then average these
Measures of employment. I examine three mea- single-year rates together to compute an average
sures of employment by birth cohorts: labor rate for twenty-five- to thirty-nine-year-olds. For
force participation (LFP), full-time, year-round data from 1990 and later years, I use weights as
employment (FTYR), and working more than suggested by the US Census Bureau. Finally, I
fifty hours per week (hereafter referred to as describe the characteristics of women who were
“working long hours”). out of the labor force in the previous year.
814 • Part VIII: Gender Inequality

Figure 94.1. Labor Force Participation Rates for Professional Women by Age and Cohort

full-time, year-round (FTYR) has greatly in-

Results creased across cohorts. Age-specific rates (not
Age-Specific Employment Rates shown) increased across every cohort. For exam-
Labor force participation rates. Figure 94.1 shows ple, the percentage of women ages thirty-five to
the trends in labor force participation (LFP) rates thirty-nine who were working FTYR increased
for professional women by age and birth cohort. from 42.7 percent of the Early Baby Boom co-
Each line in the figure shows the trend in LFP hort, to 52.1 percent of the Late Baby Boom
rates for one five-year age group. The horizontal cohort, to 64.0 percent of women in the Gen-
axis shows birth cohorts. By following each line eration X cohort.
from left to right, one can see the trend of in-
creasing age-specific rates across cohorts. Steep Percentage working long hours. The percent-
increases across cohorts are most pronounced age of professional women working more than
for the youngest age groups, with more mod- fifty hours a week increased from less than 10
est increases among older women. Many of the percent in each age group for the oldest three
age-specific trend lines plateau or slightly dip cohorts (born before 1935) to over 15 percent
across the youngest cohorts. For example, the for most ages in the youngest two cohorts (born
LFP rate for thirty- to thirty-four-year-olds de- after 1956). For this measure, there is a greater
creased from 87.1 for the Late Baby Boom co- percentage of women ages thirty to thirty-four
hort (1956 to 1965) to 85.0 for Generation X working long hours in the Generation X cohort
(1966 to 1975). In contrast, a higher percentage (18.0 percent) than in the Late Baby Boom
of Generation X women in their late thirties and (13.9 percent), but similar percentages of these
early forties are in the labor force (91.5 percent cohorts work long hours across other age groups.
and 92.6 percent, respectively) than similarly
aged women from the Late Baby Boom cohort
(86.0 percent and 89.7 percent, respectively). Compositional Changes Across
Cohorts
Full-time, year-round employment rates. The To better understand what drives cohort change,
percentage of professional women working I examine changes in the composition of cohorts
Table 94.1. Cohort Compositional Changes: Educational and Occupational Characteristics of Women Ages 30 to 34 for the Total Female
Population and for Professional Women
Percent College-Educated Percent with Percent in Historically Percent in Historically
Professionals and Managers Advanced Degrees Male Professionsa Female Professionsb
Total Total Total Total
Female Female Female Female
Birth Cohort Population Professionals Population Professionals Population Professionals Population Professionals
Baby Boom Parents 4.5 .8 12.7 .2 6.4 3.0 62.9
100
1926 to 1935 (4.3–4.7) (.69–.82) (11.5–13.9) (.17–.24) (5.5–7.3) (2.9–3.2) (61.1–64.7)
World War II 8.0 1.3 14.2 .5 7.7 5.6 66.6
100
1936 to 1945 (7.7–8.2) (1.2–1.4) (13.2–15.2) (.46–.58) (6.9–8.4) (5.4–5.8) (65.2–67.9)
Early Baby Boom 13.1 5.0 30.0 1.3 10.2 7.7 56.8
100
1946 to 1955 (12.9–13.3) (4.9–5.2) (29.2–30.8) (1.2–1.4) (9.7–10.8) (7.5–7.8) (55.9–57.7)
Late Baby Boom 15.4 6.1 31.3 2.5 15.4 5.7 39.7
100
1956 to 1965 (15.2–15.6) (5.9–6.2) (30.6–32.0) (2.4–2.6) (14.8–15.9) (5.6–5.9) (39.0–40.5)
Generation X 21.2 9.8 37.7 3.7 17.0 7.1 34.9
100
1966 to 1975 (21.1–21.4) (9.7–10.0) (37.2–38.1) (3.6–3.8) (16.6–17.4) (7.0–7.3) (34.4–35.3)
Source: 1960 to 2000 Census and 2005 ACS microdata from IPUMS.
Note: 95 percent confidence intervals are in parentheses.
a
Historically male professions are accountants, clergy, dentists, doctors, engineers, lawyers, and college professors and deans.
b
Historically female professions are librarians, nurses, social workers, and teachers.
815
816 • Part VIII: Gender Inequality

and in the behavior of subgroups within and occupations contributed to rising rates of FTYR
across cohorts. Table 94.1 shows that the per- and working long hours, but did not affect LFP
centage of the total female population who are rates.
college educated and in professional and mana-
gerial occupations has increased across cohorts, Employment rates by maternal status. As Table
indicating decreased selectivity, which would 94.2 shows, LFP rates among women with
predict falling employment rates. At the same young children increased greatly across cohorts,
time, more women in younger cohorts have ad- climbing from 33.5 percent for the Baby Boom
vanced degrees, suggesting rising employment Parents cohort (1926 to 1935) to 76.2 percent
rates, because highly educated women have for the Late Baby Boom and 78.4 percent for
higher employment rates than do less educated Generation X (1956 to 1975). The percentage of
women. Table 94.1 also shows that the share of women who have older children and are in the
professional women in historically male occupa- labor force has also increased across cohorts, and
tions has increased across cohorts. Would this in the youngest cohorts, women with children
be expected to result in higher or lower employ- over age five have LFP rates similar to those of
ment levels? On the one hand, there is evidence women without children. Not only are mothers
of continuing discrimination against women in in younger cohorts more likely to be working,
some of these professions (Blair-Loy 2001; Kay but they are working more hours. In the early
and Hagan 1995; Roth 2003). On the other cohorts, FTYR employment among women
hand, these professions confer higher salaries and with young children was quite rare; less than 10
more prestige, raising the opportunity costs of percent of these mothers worked full-time. For
not working in these fields. the Late Baby Boom and Generation X cohorts
The percentage of professional women ages (1956 to 1975), FTYR employment among
thirty to thirty-four who have children has de- mothers of young children is much more com-
creased across cohorts, reflecting the rising mean mon and has become the norm among mothers
age of US mothers at their first birth. However, of older kids.
the majority of professional women between the
ages of thirty-five and thirty-nine have children, Despite the overall trend of increasing em-
and this has not varied much across cohorts. ployment rates, this analysis shows some ev-
idence of a leveling off of employment rates
across the youngest cohorts. For all three sub-
Employment Rates of Population groups defined by maternal status, there were no
Subgroups statistically significant differences in LFP rates
Employment rates by occupation. Differences in between Generation X and the Late Baby Boom
labor force participation (LFP) rates by occu- cohort. Notably, however, there was a substan-
pational sex-type are small or nonexistent across tial increase in the FTYR rates for women with
the cohorts studied. In contrast, rates of FTYR young children (from 31.6 to 43.8) and for
employment vary more across sex-type of oc- women with older children (51.9 to 63.0). In
cupation. Women work FTYR at much higher addition, the percentage of women with young
rates in historically male unclassified occupations children working more than fifty hours per week
than in female-typed occupations. Working increased substantially across the youngest two
long hours varies even more across occupational cohorts (from 7.6 to 10.5).
types. Among Late Baby Boomers, 24 percent in
male-typed occupations work long hours, com-
pared to 17 percent of those in unclassified occu- Percentage Opting Out
pations and 12 percent of those in female-typed Table 94.3 shows the percentage of professional
occupations. The increase across cohorts in the women between the ages of twenty-five and
percentage of professional women in male-typed thirty-nine who were not employed or enrolled
Table 94.2. Employment Rates of Professional Women Ages 25 to 39 by Motherhood Status and Age of Youngest Child
Full-Time, Year-Round
Cohort Labor Force Participation Rates Employment Rates Percentage Working 50+ hrs./week

With With With With With With

Without Young Older Without Young Older Without Young Older
Children Childrena Childrenb Children Children Children Children Children Children
Baby Boom
Parents 88.7 33.5 77.2 33.2 5.6 17.9 7.2 1.3 2.1
1926 to 1935
World War II
92.3 47.2 78.5 36.9 9.3 19.8 8.5 2.1 6.1
1936 to 1945
Early Baby Boom
96.0 68.6 91.9 54.0 19.8 40.0 15.3 4.1 7.5
1946 to 1955
Late Baby Boom
95.9 76.2 92.9 67.6 31.6 51.9 23.8 7.6 11.5
1956 to 1965
Generation X
95.2 78.4 93.4 70.8 43.8 63.0 24.1 10.5 15.5
1966 to 1975
Source: 1960 to 2000 Census and 2010 ACS microdata from IPUMS.
Notes: Respondents enrolled in school are excluded from this analysis. Standard errors are the following: less than 2.5 for all LFP rates except for
Baby Boom Parents with older children (3.0) and World War II with older children (4.3); less than 2.0 for full-time, year-round rates except for Baby
Boom Parents with older children (2.2), Early Baby Boom with older children (2.2), and Generation X with young children (2.1); and less than 1.5
for all estimates of working more than 50 hrs./week.
a
Children 5 years old or younger.
b
Children 6 to 18 years old.
817
818 • Part VIII: Gender Inequality

Table 94.3. Percentage of Professional Women Ages 25 to 39 Not Enrolled in School and Not
Working in the Previous Year
Percent of All Percent of Those Percent of Those with
Professional Women in Male Professions Advanced Degrees
Baby Boom Parents 30.6 29.8 17.2
1926 to 1935 (25.7–35.5) (23.7–35.8) (13.9–20.4)
World War II 20.7 16.3 9.8
1936 to 1945 (15.2–26.2) (12.2–20.5) (7.4–12.2)
Early Baby Boom 8.0 6.6 5.1
1946 to 1955 (6.4–9.5) (5.2–8.0) (4.0–6.1)
Late Baby Boom 6.1 5.3 4.4
1956 to 1965 (4.9–7.3) (3.7–6.9) (3.4–5.3)
Generation X 3.8 3.5 2.7
1966 to 1975 (2.1–5.5) (1.9–5.1) (1.5–3.9)
Source: 1960 to 2000 Census and 2010 ACS microdata from IPUMS.
Note: 95 percent confidence intervals reported in parentheses.

in school in the previous year. The first column degree or work in a historically male profession is
shows that the percentage decreased, from 30.6 increasing across cohorts.
for the Baby Boom Parents (1926 to 1935) to
3.8 for Generation X (1966 to 1975). The sec-
ond and third columns show that the percentage Conclusions
among those with advanced degrees and in his- I find little evidence that recent cohorts of pro-
torically male professions also decreased across fessional women are opting out of paid work at
cohorts. Notably, this finding of a continued higher rates than preceding cohorts did. Indeed,
decrease in the percentage “opting out” between the FTYR employment rate of professional
the Late Baby Boom and Generation X cohorts women with young children in Generation X
differs from my conclusions published in 2008 is higher than that of any previous cohort. In
based on the 2005 data; using the 2005 data, addition, women in younger cohorts—even
I found no statistically significant difference in those with young children—are more likely to
the percentage opting out between the Late Baby be working long hours than women in older co-
Boom and Generation X. Regardless of whether horts. Notably, however, labor force participa-
I use 2005 or 2010 data, my conclusion is the tion rates have held relatively constant—instead
same: there is no evidence of a new opt-out of continuing to rise—over the last two cohorts
phenomenon. of professional women.
Three characteristics of women who did no If there is no opt-out phenomenon, why have
paid work in the previous year are particularly media stories suggesting otherwise garnered so
striking. First, opting-out seems to be concen- much attention? I suggest three possible reasons.
trated among women with high-earning hus- First, the opt-out journalistic account may ring
bands. Second, the percentage of nonworking partially true for working mothers in profes-
professional women who have children is de- sional occupations. Having children is associated
creasing. Although the vast majority of these with substantially lower rates of employment,
women have children at home, a greater per- and although professional women are working
centage in the younger cohorts do not. Finally, more now than ever before, a large percentage
whereas the percentage of women with advanced are not working full-time, year-round.
degrees and in historically male occupations who Second, many people expected greater prog-
are not working is decreasing (see Table 94.3), ress toward gender equality in employment
the percentage opting out who hold an advanced would have been achieved by now. Though it is
 Percheski—Opting Out? • 819

more feasible for women in younger cohorts to Executives.” Research in the Sociology of Work 10:
simultaneously pursue a career and raise children 51–83.
than it was for women in older cohorts, profes- Goldin, Claudia. 2006. “The Quiet Revolution That
sional women with children still work fewer Transformed Women’s Employment, Education,
and Family.” American Economic Review 96 (2):
hours than men or women without children.
1–21.
Finally, there are more professional women Kay, Fiona M., and John Hagan. 1995. “The Per-
now than ever before, so the average person is sistent Glass Ceiling: Gendered Inequalities in
more likely to know a professional woman who the Earnings of Lawyers.” The British Journal of
has left the labor force. In addition, although Sociology 46: 279–310.
the percentage of women with advanced degrees Roth, Louise Marie. 2003. “Selling Women Short: A
who are not working is declining across cohorts, Research Note on Gender Differences in Com-
the percentage of nonworking women who have pensation on Wall Street.” Social Forces 82 (2):
an advanced degree is growing, because the en- 783–802.
Ruggles, Stephen, Matthew Sobek, Trent Alexander,
tire population is becoming more educated.
Catherine Fitch, Ronald Goeken, Patricia Kelly
“Opting out” does not describe the trend Hall, Miriam King, and Charles Ronnander.
characterizing the employment patterns of recent 2004. Integrated Public Use Microdata Series: Ver-
cohorts of professional women. By most mea- sion 3.0 [MRDF]. Minneapolis: Minnesota Pop-
sures, professional women—including mothers ulation Center [producer and distributor].
with young children—are working more than Ryder, Norman. 1965. “The Cohort as a Concept in
ever. By narrowly focusing on motherhood as the Study of Social Change.” American Sociologi-
the primary source of women’s employment dis- cal Review 30 (6): 843–861.
advantages, the new opting-out rhetoric pins the Sayer, Liana C., Philip N. Cohen, and Lynne M.
Casper. 2004. The American People Census 2000:
explanation for gender inequality on individual
Women, Men, and Work. New York: Russell Sage
choices. This diverts attention from structural
Foundation.
and institutional factors that may depress wom- US Bureau of the Census. 1950. Alphabetic Index
en’s employment levels. We need to ask new of Occupations and Industries: 1950. Washington,
questions about women’s employment—includ- DC: US Government Printing Office.
ing issues of work-family conflict—but also tran- Vere, James. 2007. “‘Having It All’ No Longer: Fer-
scend them if we are to better understand gender tility, Female Labor Supply and the New Life
inequality in the workplace. Choices of Generation X.” Demography 44 (4):
821–828.
REFERENCES Whittington, Leslie, Susan Averett, and Donna An-
Acker, Joan. 1990. “Hierarchies, Jobs, Bodies: A derson. 2000. “Choosing Children over Career?
Theory of Gendered Organizations.” Gender and Changes in the Postpartum Labor Force Behavior
Society 4 (2): 139–158. of Professional Women.” Population Research and
Belkin, Lisa. 2003. “The Opt-Out Revolution.” Policy Review 19: 339–355.
New York Times Magazine, October 26, 42–47, Williams, Joan C., Jessica Manvell, and Stephanie
58, 85–86. Bornstein. 2006. “‘Opt-Out’ or Pushed Out?
Blair-Loy, Mary. 2001. “It’s Not Just What You How the Press Covers Work/Family Conflict.”
Know, It’s Who You Know: Technical Knowl- In WorkLifeLaw Report. San Francisco, CA: UC
edge, Rainmaking, and Gender Among Finance Hastings College of the Law.
820 • Part VIII: Gender Inequality

Discrimination

95 • Claudia Goldin and Cecilia Rouse

Orchestrating Impartiality
The Impact of “Blind” Auditions on Female Musicians

Sex-biased hiring has been alleged for many oc- more than 100 applicants where fewer than 20
cupations but is extremely difficult to prove. The would have been considered before. Audition
empirical literature on discrimination, deriving committees were restructured to consist of mem-
from the seminal contributions of Gary Becker bers of the orchestra, not just the conductor and
(1971) and Kenneth Arrow (1973), has focused section principal. The audition procedure be-
mainly on disparities in earnings between groups came democratized at a time when many other
(e.g., males and females), given differences in institutions in America did as well.
observable productivity-altering characteristics. But democratization did not guarantee im-
With the exception of various audit studies (e.g., partiality, because favorites could still be identi-
Genevieve Kenney and Douglas A. Wissoker, fied by sight and through resumes. Another set
1994; David Neumark et al., 1996) and others, of procedures was adopted to ensure, or at least
few researchers have been able to address directly give the impression of, impartiality. These pro-
the issue of bias in hiring practices. A change in cedures involve hiding the identity of the player
the way symphony orchestras recruit musicians from the jury. Although they take several forms,
provides an unusual way to test for sex-biased we use the terms “blind” and “screen” to describe
hiring. the group. The question we pose is whether the
Until recently, the great symphony orchestras hiring process became more impartial through
in the United States consisted of members who the use of blind auditions. Because we are able
were largely handpicked by the music director. to identify sex, but no other characteristics, for
Although virtually all had auditioned for the po- a large sample, we focus on the impact of the
sition, most of the contenders would have been screen on the employment of women.
the (male) students of a select group of teachers. Screens were not adopted by all orchestras
In an attempt to overcome this seeming bias in at once. Among the major orchestras, one still
the hiring of musicians, most major U.S. orches- does not have any blind round to their audi-
tras changed their audition policies in the 1970s tion procedure (Cleveland) and one adopted
and 1980s, making them more open and rou- the screen in 1952 for the preliminary round
tinized. Openings became widely advertised in (Boston Symphony Orchestra), decades before
the union papers, and many positions attracted the others. Most other orchestras shifted to

Claudia Goldin and Cecilia Rouse. “Orchestrating Impartiality: The Impact of ‘Blind’ Auditions on Female
Musicians,” American Economic Review 90 (2000), pp. 715–726, 734–738, 740–741. Reprinted by permission
of the American Economic Review.
 Goldin and Rouse—Orchestrating Impartiality • 821

blind preliminaries from the early 1970s to the being the lowest (generally at zero) for decades.
late 1980s. The variation in screen adoption at Thus the increase of women in the nation’s finest
various rounds in the audition process allows us orchestras has been extraordinary. The increase
to assess its use as a treatment. is even more remarkable because, as we discuss
The change in audition procedures with the below, turnover in these orchestras is exceedingly
adoption of the screen allows us to test whether low. The proportion of new players who were
bias exists in its absence. In both our study and women must have been, and indeed was, exceed-
studies using audits, the issue is whether sex (or ingly high.
race or ethnicity), apart from objective criteria Similar trends can be discerned for four
(e.g., the sound of a musical performance, the other orchestras—the Los Angeles Symphony
content of a resume), is considered in the hiring Orchestra (LA), the San Francisco Philharmonic
process. Why sex might make a difference is an- (SF), the Detroit Symphony Orchestra, and the
other matter. . . . Pittsburgh Symphony Orchestra (PSO)—given
in Figure 95.1B.2 The upward trend in the pro-
portion female is also obvious in Figure 95.1B,
Sex Composition of Orchestras although initial levels are higher than in Figure
Symphony orchestras consist of about 100 mu- 95.1A. There is somewhat more choppiness to
sicians and, although the number has varied be- the graph, particularly during the 1940s. Al-
tween 90 and 105, it is rarely lower or higher. though we have tried to eliminate all substi-
The positions, moreover, are nearly identical tute, temporary, and guest musicians, especially
between orchestras and over time. As opposed during World War II and the Korean War, this
to firms, symphony orchestras do not vary much was not always possible.
in size and have virtually identical numbers and The only way to increase the proportion
types of jobs. Thus we can easily look at the pro- women is to hire more female musicians and
portion of women in an orchestra without being turnover during most periods was low. The
concerned about changes in the composition of number of new hires is graphed in Figure 95.2
occupations and the number of workers. An in- for five orchestras. Because “new hires” is a vol-
crease in the number of women from, say, 1 to atile construct, we use a centered five-year mov-
10 cannot arise because the number of harpists ing average. In most years after the late 1950s,
(a female-dominated instrument) has greatly ex- the top-ranked orchestras in the group (Chicago
panded. It must be because the proportion fe- and NYPhil) hired about four musicians a year,
male within many groups has increased. whereas the other three hired about six. Prior
Among the five highest-ranked orchestras to 1960 the numbers are extremely high for LA
in the nation (known as the “Big Five”)—the and the PSO, because, it has been claimed, their
Boston Symphony Orchestra (BSO), the Chi- music directors exercised their power to termi-
cago Symphony Orchestra, the Cleveland Sym- nate, at will, the employment of musicians. Also
phony Orchestra, the New York Philharmonic of interest is that the number of new hires trends
(NYPhil), and the Philadelphia Orchestra— down, even excluding years prior to 1960. The
none contained more than 12 percent women important points to take from Figure 95.2 are
until about 1980.1 As can be seen in Figure that the number of new hires was small after
95.1A, each of the five lines (giving the propor- 1960 and that it declined over time.
tion female) greatly increases after some point. The proportion female among the new hires
For the NYPhil, the line steeply ascends in the must have been sizable to increase the proportion
early 1970s. For the BSO, the turning point ap- female in the orchestras. Figure 95.3 shows the
pears to be a bit earlier. The percentage female trend in the share of women among new hires
in the NYPhil is currently 35 percent, the high- for four of the “Big Five” (Figure 95.3A) and
est among all 11 orchestras in our sample after four other orchestras (Figure 95.3B).3 In both
Figure 95.1. Proportion Female in Nine Orchestras, 1940 to 1990s. A:
The “Big Five”; B: Four Others
A 0.4
BSO
Philadelphia
Chicago
Proportion Female in Orchestra 0.3
NYPhil
Cleveland

0.2

0.1

0.0

1940 1950 1960 1970 1980 1990

B 0.4
SF
PSO
Proportion Female in Orchestra

LA
0.3
Detroit

0.2

0.1

0.0

1940 1950 1960 1970 1980 1990

Source: Roster sample. See text.

Figure 95.2. Number of New Hires in Five Orchestras, 1950 to 1990s

24
20 NYPhil
Chicago
18
LA
16 SF
Number of New Hires

PSO
14
12
10
8
6
4
2
0

1950 1960 1970 1980 1990

Source: Roster sample. See text.
 Goldin and Rouse—Orchestrating Impartiality • 823

Figure 95.3. Female Share of New Hires in Eight Orchestras, 1950 to 1990s.
A: Four of the “Big Five”; B: Four Others
A
0.6 NYPhil
BSO
0.5 Chicago
Female Share of New Hires

Cleveland

0.4

0.3

0.2

0.1

0.0

1950 1960 1970 1980 1990

B LA
0.6 SF
Detroit
0.5 PSO
Female Share of New Hires

0.4

0.3

0.2

0.1

0.0

1950 1960 1970 1980 1990

Source: Roster sample. See text.
Note: A five-year centered moving average is used. New hires are musicians who were not
with the orchestra the previous year, who remain for at least one additional year and who
were not substitute musicians in the current year.

groups the female share of new hires rose over 1980s, a time when the labor force participation
time, at a somewhat steeper rate for the more of women increased generally and when their
prestigious orchestras. Since the early 1980s the participation in various professions greatly ex-
share female among new hires has been about panded. The question, therefore, is whether the
35 percent for the BSO and Chicago, and about screen mattered in a direct manner or whether
50 percent for the NYPhil, whereas before 1970 the increase was the result of a host of other fac-
less than 10 percent of new hires were women. tors, including the appearance of impartiality or
Even though the fraction of new hires who are an increased pool of female contestants coming
female rises at somewhat different times across out of music schools. Because the majority of
the orchestras, there is a discernible increase for new hires are in their late twenties and early thir-
the group as a whole from the late 1970s to early ties, the question is whether the most selective
824 • Part VIII: Gender Inequality

Figure 95.4. Proportion Female of Julliard Graduates, Total and by Section, 1947
to 1995
Proportion Female of Juilliard Graduates 0.7

All Sections
0.6 Strings
Winds
0.5 Brass

0.4

0.3

0.2

0.1

0.0

1940 1950 1960 1970 1980 1990 2000

Source: Julliard Music School files.

music schools were producing considerably more and some continue to have, stated policies not
female students in the early 1970s. We currently to hire women. The Vienna Philharmonic has
have information by instrument for only the only recently admitted its first female member
Juilliard School of Music. With the exception of (a harpist). Female musicians, it can be convinc-
the brass section, the data, given in Figure 95.4, ingly argued, have historically faced considerable
do not reveal any sharp breaks in the fraction discrimination. Thus a blind hiring procedure,
of all graduates who are female. Thus, it is not such as a screen that conceals the identity of the
immediately obvious that an expansion in the musician auditioning, could eliminate the possi-
supply of qualified female musicians explains bility of discrimination and increase the number
the marked increase in female symphony orches- of women in orchestras.
tra members; it could, therefore, be because of
changes in the hiring procedures of orchestras.
But why would changes in audition proce- Orchestral Auditions
dures alter the sex mix of those hired? Many of To understand the impact of the democratiza-
the most renowned conductors have, at one time tion of the audition procedure and the screen,
or another, asserted that female musicians are we must first explain how orchestra auditions
not the equal of male musicians. Claims abound are now conducted. After determining that an
in the world of music that “women have smaller audition must be held to fill an opening, the or-
techniques than men,” “are more temperamen- chestra advertises that it will hold an audition.
tal and more likely to demand special attention Each audition attracts musicians from across the
or treatment,” and that “the more women [in country and, often, from around the world. Mu-
an orchestra], the poorer the sound.”4 Zubin sicians interested in auditioning are required to
Mehta, conductor of the Los Angeles Symphony submit a resume and often a tape of compulsory
from 1964 to 1978 and of the New York Phil- music (recorded according to specific guidelines)
harmonic from 1978 to 1990, is credited with to be judged by members of the orchestra. In
saying, “I just don’t think women should be in some orchestras this prescreening is dispositive;
an orchestra.”5 Many European orchestras had, in others the musician has the right to audition
 Goldin and Rouse—Orchestrating Impartiality • 825

live in the preliminary round, even if the audi- authoritarianism. The music director can ignore
tion committee rejects the candidate on the basis the audition committee’s advice, but does so at
of the tape. All candidates are given, in advance, greater peril. Once an applicant is chosen to
most of the music they are expected to perform be a member of an orchestra, lifetime tenure is
at the live audition. awarded after a brief probationary period. The
Live auditions today generally consist of basis for termination is limited and rarely used.
three rounds: preliminary, semifinal, and final. The positions we are analyzing are choice jobs in
But there is considerable variation. Although all the musical world. In 1995 the minimum starting
orchestras now have a preliminary round, some base salary for musicians at the BSO was $1,400
have two final rounds and in many there was per week (for a 52-week year), not including
no semifinal round until the 1980s. The prelim- recording contracts, soloist fees, overtime and
inary is generally considered a screening round extra service payments, bonuses, and per diem
to eliminate unqualified candidates. As a result, payments for tours and Tanglewood. . . .
the committee is free to advance as many, or as The audition procedures of the 11 orchestras
few, as they wish. Candidates advanced from the in the roster sample are summarized in Table
semifinal round are generally considered “accept- 95.1.6 Although audition procedures are now
able for hire” by the audition committee (which part of union contracts, that was not the case in
does not include the music director, a.k.a. con- the more distant past and the procedures were
ductor, until the finals). Again, this means that not apparently recorded in any surviving docu-
the committee can advance as many as it wishes. ments. We gathered information on these pro-
The final round generally results in a hire, but cedures from various sources, including union
sometimes does not. contracts, interviews with personnel managers,
In blind auditions (or audition rounds) a archival documents on auditions, and a mail
screen is used to hide the identity of the player survey we conducted of orchestral musicians
from the committee. The screens we have seen are concerning the procedures employed during the
either large pieces of heavy (but sound-porous) audition that won them their current position.
cloth, sometimes suspended from the ceiling An obvious question to ask is whether the
of the symphony hall, or look like large room adoption of the screen is endogenous. Of par-
dividers. Some orchestras also roll out a carpet ticular concern is that more meritocratic orches-
leading to center stage to muffle footsteps that tras adopted blind auditions earlier, producing
could betray the sex of the candidate. Each can- the spurious result that the screen increased the
didate for a blind audition is given a number, likelihood that women were hired.7 We estimate
and the jury rates the candidate’s performance a probit model of screen adoption by year, con-
next to the number on a sheet of paper. Only ditional on an orchestra’s not previously having
the personnel manager knows the mapping from adopted the screen (an orchestra exits the exer-
number to name and from name to other per- cise once it adopts the screen). Two time-varying
sonal information. The names of the candidates covariates are included to assess commonly held
are not revealed to the juries until after the last notions about screen adoption: the proportion
blind round. female (lagged) in the orchestra, and a measure
Almost all preliminary rounds are now blind. of tenure (lagged) of then-current orchestra
The semifinal round, added as the number of members. Tenure is included because personnel
applicants grew, may be blind. Finals are rarely managers maintain the screen was advocated
blind and almost always involve the attendance more by younger players.
and input of the music director. Although the As the proportion female in an orchestra in-
music director still wields considerable power, creases, so does the likelihood of screen adop-
the self-governance that swept orchestras in the tion in the preliminary round, as can be seen in
1970s has served to contain the conductor’s columns (1) and (2) in Table 95.2, although the
826

Table 95.1. Orchestra Audition Procedure Summary Table

y
Orchestra Preliminaries Semifinals Finals
A Blind since 1973 Blind (varies) since 1973 Not blind
B Blind since at least 1967 Use of screen varies Blind 1967–1969; since
winter 1994
C Blind since at least 1979 Not blind: 1991–present Not blind
(definitely after 1972) Blind: 1984–1987
D Blind since 1986 Blind since 1986; varies 1st part blind since 1993;
until 1993 2nd part not blind
E Use of screen varies until 1981 Use of screen varies Not blind
F Blind since at least 1972 Blind since at least 1972 Blind since at least 1972
G Blind since 1986 Use of screen varies Not blind
H Blind since 1970 Not blind Not blind
I Blind since 1979 Blind since 1979 Blind since fall 1983
J Blind since 1952 Blind since 1952 Not blind
K Not blind Not blind Not blind
Notes: The eleven orchestras (A through K) are those in the roster sample described in the text. A subset of eight form
the audition sample (also described in the text). All orchestras in the sample are major big-city U.S. symphony orches-
tras and include the “Big Five.”
Sources: Orchestra union contracts (from orchestra personnel managers and libraries), personal conversations with
orchestra personnel managers, and our mail survey of current orchestra members who were hired during the probable
period of screen adoption.

Table 95.2. Estimated Probit Models for the Use of a Screen

Preliminaries blind Finals blind

(1) (2) (3)
(Proportion female)t–1 2.744 3.120 0.490
(3.265) (3.271) (1.163)
[0.006] [0.004] [0.011]
(Proportion of orchestra personnel –26.46 –28.13 –9.467
with <6 years tenure)t–1 (7.314) (8.459) (2.787)
[–0.058] [–0.039] [–0.207]
“Big Five” orchestra 0.367
(0.452)
[0.001]
pseudo R 2 0.178 0.193 0.050
Number of observations 294 294 434
Notes: The dependent variable is 1 if the orchestra adopts a screen, 0 otherwise. Huber standard errors (with orches-
tra random effects) are in parentheses. All specifications include a constant. Changes in probabilities are in brackets.
“Proportion female” refers to the entire orchestra. “Tenure” refers to years of employment in the current orchestra. “Big
Five” includes Boston, Chicago, Cleveland, New York Philharmonic, and Philadelphia. The data begin in 1947 and an
orchestra exits the exercise once it adopts the screen. The unit of observation is an orchestra-year.
Source: Eleven-orchestra roster sample. See text.
 Goldin and Rouse—Orchestrating Impartiality • 827

effects are very small and far from statistically recorded whether the individual was advanced to
significant. We estimate a similar effect when we the next round of the current audition.
assess the role of female presence on the adop- We rely on the first name of the musicians
tion of blind finals [see column (3)]. The impact to determine sex. For most names establishing
of current tenure, measured by the proportion sex was straightforward. Sexing the Japanese
with less than six years with the orchestra, is— and Korean names was equally straightforward,
contrary to general belief—negative and the at least for our Japanese and Korean consul-
results do not change controlling for whether tants. For more difficult cases, we checked the
the orchestra is one of the “Big Five.” In all, it names in three baby books (Connie Lockhard
appears that orchestra sex composition had lit- Ellefson, 1990; Alfred J. Kolatch, 1990; Bruce
tle influence on screen adoption, although the Lansky, 1995). If the name was listed as male-
stability of the personnel may have increased its or female-only, we considered the sex known.
likelihood. The gender-neutral names (e.g., Chris, Leslie,
and Pat) and some Chinese names (for which
sex is indeterminate in the absence of Chinese
The Role of Blind Auditions on the characters) remained ambiguous. Using these
Audition and Hiring Process procedures, we were able to determine the sex of
A. Data and Methods 96 percent of our audition sample.
Audition Records. We use the actual audition In constructing our analysis sample, we ex-
records of eight major symphony orchestras ob- clude incomplete auditions, those in which there
tained from orchestra personnel managers and were no women (or only women) competing,
the orchestra archives. The records are highly rounds from which no one was advanced, and
confidential and occasionally contain remarks the second final round, if one exists, for which
(including those of the conductor) about mu- the candidates played with the orchestra. In ad-
sicians currently with the orchestra. To preserve dition, we generally consider each round of the
the full confidentiality of the records, we have audition separately. These sample restrictions
not revealed the names of the orchestras in our exclude 294 rounds (199 contained no women)
sample. and 1,539 individuals. Our final analysis sam-
Although availability differs, taken together ple has 7,065 individuals and 588 audition
we obtained information on auditions dating rounds (from 309 separate auditions) resulting
from the late 1950s through 1995. Typically, in 14,121 person-rounds and an average of 2.0
the records are lists of the names of individuals rounds per musician. . . .
who attended the auditions, with notation near
the names of those advanced to the next round. Roster Data. Our second source of informa-
For the preliminary round, this would indicate tion comes from the final results of the audition
advancement to either the semifinal or final process, the orchestra personnel rosters. We
round. Another list would contain the names of collected these data from the personnel page
the semifinalists or finalists with an indication of concert programs, one each year for eleven
of who won the audition. From these records, major symphony orchestras. These records are
we recorded the instrument and position (e.g., in the public domain and thus we have used the
section, principal, substitute) for which the orchestra names in the graphs containing those
audition was held. We also know whether the data alone. As opposed to the auditionees, we
individual had an “automatic” placement in a were able to confirm the sex of the players with
semifinal or final round. Automatic placement the orchestra personnel managers and archivists.
occurs when a musician is already known to We considered a musician to be new to the or-
be above some quality cutoff and is invited to chestra in question if he or she had not previ-
compete in a semifinal or final round. We also ously been a regular member of that orchestra
828 • Part VIII: Gender Inequality

Table 95.3. Linear Probability Estimates of the Effect of Blind Auditions on the Likelihood of Being
Hired with Individual Fixed Effects

Without semifinals With semifinals All

(1) (2) (3) (4) (5) (6)
Completely blind –0.024 0.047 0.001 0.006 0.001 0.005
audition (0.028) (0.041) (0.009) (0.011) (0.008) (0.009)
Completely blind 0.051 0.036 0.001 –0.004 0.011 0.006
audition female (0.046) (0.048) (0.016) (0.016) (0.013) (0.013)
Year effects? No Yes No Yes No Yes
Other covariates? No Yes No Yes No Yes
R2 0.855 0.868 0.692 0.707 0.678 0.691
Number of
observations 4,108 4,108 5,883 5,883 9,991 9,991
Notes: The unit of observation is person-round. The dependent variable is 1 if the individual is advanced (or hired)
from the final round and 0 if not. Standard errors are in parentheses. All specifications include individual fixed effects,
whether the sex is missing, and an interaction for sex being missing and a completely blind audition. “Other covariates”
are the size of the audition, the proportion female at the audition, the number of individuals advanced (hired), whether
a “Big Five” orchestra, the number of previous auditions, and whether the individual had an automatic semifinal or
final.
Source: Eight-orchestra audition sample. See text.

(i.e., we did not count returning members as The impact of completely blind auditions on
new). We excluded, when possible, temporary the likelihood of a woman’s being hired is given
and substitute musicians, as well as harpists and in Table 95.3, for which all results include indi-
pianists. Our final sample for 1970 to 1996 has vidual fixed effects.9 The impact of the screen is
1,128 new orchestra members. . . . positive and large in magnitude, but only when
there is no semifinal round. Women are about 5
percentage points more likely to be hired than
B. The Effect of the Screen on the are men in a completely blind audition, although
Hiring of Women the effect is not statistically significant. The effect
Using the Audition Sample. We turn now to the is nil, however, when there is a semifinal round.
effect of the screen on the actual hire and esti- The impact for all rounds [columns (5) and (6)]
mate the likelihood an individual is hired out of is about 1 percentage point, although the stan-
the initial audition pool.8 Whereas the use of the dard errors are large and thus the effect is not
screen for each audition round was, more or less, statistically significant. Given that the probabil-
an unambiguous concept, that for the entire pro- ity of winning an audition is less than 3 percent,
cess is not and we must define a blind audition. we would need more data than we currently have
The definition we have chosen is that a blind to estimate a statistically significant effect, and
audition contains all rounds that use the screen. even a 1-percentage-point increase is large, as we
In using this definition, we compare auditions later demonstrate.
that are completely blind with those that do not
use the screen at all or use it for the early rounds Using the Roster Data. The roster data afford us
only. We divide the sample into auditions that another way to evaluate the effect of the screen
have a semifinal round and those that do not, on the sex composition of orchestras. Using the
because the earlier analysis suggested they might rosters we know the sex of new hires each year for
differ (Goldin and Rouse 2000). 11 orchestras, and we also have information (see
 Goldin and Rouse—Orchestrating Impartiality • 829

Table 95.4. Probit Estimates of the Effect of Blind Auditions on the Sex of the New Members, 1970
to 1996
Only blind preliminaries
and/or semifinals vs.
Any blind auditions completely blind auditions
(1) (2)
Any blind auditions 0.238
(0.183)
[0.075]
Only blind preliminaries and/or 0.232
semifinals (0.184)
[0.074]
Completely blind auditions 0.361
(0.438)
[0.127]
Section:
Woodwinds –0.187 –0.188
(0.114) (0.114)
[–0.058] [–0.058]
Brass –1.239 –1.237
(0.157) (0.157)
[–0.284] [–0.284]
Percussion –1.162 –1.164
(0.305) (0.305)
[–0.235] [–0.235]
p-value of test: only blind preliminaries 0.756
and/or semifinals = completely blind
pseudo R2 0.106 0.106
Number of observations 1,128 1,128
Notes: The dependent variable is 1 if the individual is female and 0 if male. Standard errors are in parentheses. All
specifications include orchestra fixed effects and orchestra-specific time trends. Changes in probabilities are in brack-
ets; see text for an explanation of how they are calculated. New members are those who enter the orchestra for the
first time. Returning members are not considered new. The omitted section is strings.
Source: Eleven-orchestra roster sample. See text.

Table 95.1) on the year the screen was adopted 95.4, the effects of the screen on the hiring of
by each orchestra. We treat the orchestra posi- women from 1970 to 1996 using a probit model.
tion as the unit of observation and ask whether The screen is first defined to include any blind
the screen affects the sex of the individual who auditions [column (1)]. In column (2) we esti-
fills the position. We model the likelihood that a mate separate effects for orchestras using blind
female is hired in a particular year as a function preliminary (and semifinal) rounds but not blind
of whether the orchestra’s audition procedure finals and those with completely blind auditions.
involved a screen, again relying on the variation To interpret the probit coefficient, we first
over time within a particular orchestra. Thus, in predict a base probability, under the assumption
all specifications, we include orchestra fixed ef- that each orchestra does not use a screen. We
fects and an orchestra-specific time trend. then predict a new probability assuming the or-
The roster data extend further back in time chestra uses a screen. The mean difference in the
than do the audition data and could conceivably probabilities is given in brackets.
begin with the orchestra’s founding, although The coefficient on blind in column (1) is pos-
there is no obvious reason to include many years itive, although not significant at any usual level
when none used the screen. We report, in Table of confidence. The estimates in column (2) are
830 • Part VIII: Gender Inequality

positive and equally large in magnitude to those 5. Seltzer (1989), p. 215.

in column (1). Further, these estimates show 6. We identify the orchestras by letter, rather than
that the existence of any blind round makes a by name, to preserve confidentiality of the audition
difference and that a completely blind process sample.
7. Note, however, it is unlikely that the orchestras
has a somewhat larger effect (albeit with a large
that sought to hire more women chose to adopt the
standard error). According to the point estimates screen earlier since the best way to increase the num-
in column (1) of Table 95.4, blind auditions in- ber of women in the orchestra is to have not-blind
crease the likelihood a female will be hired by auditions (so that one could be sure to hire more
7.5 percentage points. The magnitude of the women).
effect must be judged relative to the overall av- 8. The original article (Goldin and Rouse 2000)
erage and, for the period under consideration, includes much additional analysis examining the ef-
it was about 30 percent. Thus blind auditions fect of the screen on the likelihood of being advanced
increased the likelihood a female would be hired through each stage of the audition process.—EDS
9. There are four auditions in which the com-
by 25 percent.10 . . .
mittee could not choose between two players and
therefore asked each to play with the orchestra. We
consider both to be winners. The results are not
Conclusion
sensitive to this classification. For this analysis we
The question is whether hard evidence can sup- exclude auditions with no women, all women, or no
port an impact of discrimination on hiring. Our winner; these exclusions do not change the results.
analysis of the audition and roster data indicates 10. In Table 95.3 we are identified off of individ-
that it can, although we mention various caveats uals who competed in auditions that were completely
before we summarize the reasons. Even though blind and those that were not completely blind (that
is, any one round could not be blind). The unit of
our sample size is large, we identify the coeffi-
observation is the person-round and there are 92 ful-
cients of interest from a much smaller sample.
filling this criterion for auditions without a semifinal
Some of our coefficients of interest, therefore, do [columns (1) and (2)]; on average these persons com-
not pass standard tests of statistical significance peted in 3.6 auditions in this sample. There are 625
and there is, in addition, one persistent result person-rounds fulfilling this criterion that included a
that goes in the opposite direction. The weight semifinal [columns (3) and (4)] and on average these
of the evidence, however, is what we find most persons competed in 3.5 auditions in this sample.
persuasive and what we have emphasized. . . . Finally, there are 911 person-rounds fulfilling this
As in research in economics and other fields criterion across all auditions [columns (5) and (6)]
and on average these persons competed in 3.5 au-
on double-blind refereeing (see, e.g., Blank,
ditions in this sample. The sample off of which we
1991), the impact of a blind procedure is toward
are identified is larger for all auditions than for the
impartiality and the costs to the journal (here to sum of the other two because some individuals au-
the orchestra) are relatively small. We conclude ditioned both with and without a semifinal round.
that the adoption of the screen and blind au-
ditions served to help female musicians in their REFERENCES
quest for orchestral positions. Arrow, Kenneth. “The Theory of Discrimination,”
in Orley Ashenfelter and Albert Rees, eds., Dis-
NOTES crimination in labor markets. Princeton, NJ:
1. The data referred to, and used in Figures 95.1 Princeton University Press, 1973, pp. 3–33.
to 95.3, are from orchestral rosters, described in Becker, Gary. The economics of discrimination, 2nd
more detail below. Ed. Chicago: University of Chicago Press, 1971
2. Our roster sample also includes the Metropol- [orig. pub. 1957].
itan Opera Orchestra and the St. Louis Symphony. Blank, Rebecca. “The Effects of Double-Blind versus
3. A centered five-year moving average is also Single-Blind Refereeing: Experimental Evidence
used for this variable. from the American Economic Review.” American
4. Seltzer (1989), p. 215. Economic Review, December 1991, 81(5), pp.
1041–67.
 Correll, Benard, and Paik—Getting a Job • 831

Ellefson, Connie Lockhard. The melting pot book Kolatch, Alfred J. The Jonathan David dictionary of
of baby names, 2nd Ed. Cincinnati, OH: Better first names. New York: Perigee Books, 1990.
Way Books, 1990. Lansky, Bruce. 35,000+ baby names. New York:
Goldin, Claudia and Cecilia Rouse. 2000. “Orches- Meadowbrook Press, 1995.
trating Impartiality: The Impact of ‘Blind’ Audi- Neumark, David (with Bank, Roy and Van Nort,
tions on Female Musicians.” American Economic Kyle D.). “Sex Discrimination in Restaurant Hir-
Review 90 (September): 715–741. ing: An Audit Study.” Quarterly Journal of Eco-
Kenney, Genevieve and Wissoker, Douglas A. “An nomics, August 1996, 111(3), pp. 915–41.
Analysis of the Correlates of Discrimination Seltzer, George. Music matters: The performer and the
Facing Young Hispanic Job-Seekers.” American American Federation of Musicians. Metuchen, NJ:
Economic Review, June 1994, 84(3), pp. 674–83. Scarecrow Press, 1989.

96 • Shelley J. Correll, Stephen Benard, and In Paik

Getting a Job
Is There a Motherhood Penalty?

Mothers experience disadvantages in the work- are judged as less committed to their jobs, less
place in addition to those commonly associated dependable, less authoritative, but warmer, more
with gender. Recent studies show employed emotional, and more irrational than otherwise
mothers in the United States suffer a 5 percent equal women managers (Halpert, Wilson, and
per-child wage penalty on average after con- Hickman 1993; Corse 1990). While the pat-
trolling for the usual human capital and occu- tern is clear, the underlying mechanism remains
pational factors that affect wages (Budig and opaque. Why would being a parent lead to dis-
England 2001; Anderson, Binder, and Krause advantages in the workplace for women? And
2003). The pay gap between mothers and non- why might similar disadvantages not occur for
mothers under age 35 is larger than the pay gap men?
between men and women (Crittenden 2001), This paper presents a laboratory experiment
and employed mothers now account for most of and an audit study. The laboratory experiment
the “gender gap” in wages (Glass 2004). evaluates the hypothesis that the “motherhood
The disadvantages are not limited to pay. De- penalty” occurs because cultural understandings
scribing a consultant as a mother leads evaluators of motherhood lead evaluators to, perhaps un-
to rate her as less competent than when she is consciously, expect mothers to be less compe-
described as not having children (Cuddy, Fiske, tent and less committed to their jobs (Blair-Loy
and Glick 2004), and visibly pregnant managers 2003; Ridgeway and Correll 2004). As a result,

The ideas, issues, and theories considered in this brief piece are examined in greater depth in the following
publication: Shelley J. Correll, Stephen Benard, and In Paik, “Getting a Job: Is There a Motherhood Pen-
alty?” American Journal of Sociology 112 (March 2007), pp. 1297–1338, published by the University of Chicago
Press. The article as printed here was originally prepared for Social Stratification: Class, Race, and Gender in
Sociological Perspective, Third Edition, edited by David B. Grusky, pp. 759–770. Copyright © 2008 by West-
view Press.
832 • Part VIII: Gender Inequality

we argue, employers will discriminate against because employers discriminate against mothers
mothers when making hiring, promotion, and (or some combination of the two processes).
salary decisions. We do not expect fathers to
experience similar workplace disadvantages
since being a good father is not seen in our cul- Productivity and Discrimination
ture as incompatible with being a good worker To distinguish between discrimination and pro-
(Townsend 2002). By having participants rate ductivity explanations, ideally one would com-
job applicants, we expect that applicants pre- pare the outcomes of employed mothers and
sented as women with children will be viewed nonmothers who have equal levels of workplace
as less competent and less committed to work, productivity. If differences in pay or promotion
will need to present evidence that they are more rates were found between equally productive
qualified for the job, will be rated as less pro- mothers and nonmothers, this would suggest
motable, and will be offered lower starting sala- that discrimination factors were at work. How-
ries compared with otherwise similar applicants ever, the datasets analyzed in the previous stud-
presented as women without children. While the ies lack direct measures of worker productivity.
laboratory experiment isolates and examines the One likely reason for this is that it is inherently
mechanism of discrimination, the audit study problematic to specify what makes someone a
provides external validity by evaluating whether productive employee. This difficulty leads to
actual employers discriminate against mothers. another: unexplained gaps in wages between
employed mothers and nonmothers can always
be attributed to unmeasured productivity differ-
Wage Penalty for Motherhood ences between the two groups.
Explanations for the motherhood wage penalty To address these problems, we experimentally
generally can be classified as worker explana- held constant the workplace performances and
tions, which seek to identify differences in the other relevant characteristics of a pair of fictitious
traits, skills, and behaviors between mothers and job applicants and varied only their parental
nonmothers, and discrimination explanations, status. By holding constant workplace-relevant
which rely on the differential preference for or characteristics, differences between the ratings of
treatment of mothers and nonmothers. Empiri- mothers and nonmothers cannot be attributed
cal evaluations of these explanations have largely to productivity or skill differences. While this
focused on the former. design cannot rule out the possibility that pro-
Budig and England (2001) find that inter- ductivity differences account for part of the wage
ruptions from work, working part-time, and de- penalty found in previous studies, the laboratory
creased seniority/experience collectively explain study will isolate a potential status-based dis-
no more than about one-third of the mother- crimination mechanism by evaluating whether
hood penalty. In addition, “mother-friendly” job being a parent disadvantages mothers in the
characteristics (i.e., differences in the type of jobs workplace even when no productivity differ-
chosen) explain very little of the penalty. Simi- ences exist between them and women without
larly, Anderson, Binder, and Krause (2003) find children.
that human capital and occupational and house-
hold resource variables (e.g., number of adults in
household) collectively account for 24 percent of Performance Expectations
the total penalty for one child and 44 percent for and Evaluations of Workplace
women with two or more children. As Budig and Competence
England (2001) conclude, the remaining wage Status Characteristics Theory
gap likely arises either because mothers are some- The laboratory study evaluates the theoretical
how less productive at work than nonmothers or claim that motherhood is a “status characteristic”
 Correll, Benard, and Paik—Getting a Job • 833

that, when salient, results in biased evaluations of commitment because contradictory schemas
of competence and commitment, a stricter stan- govern conceptions of “family devotion” and
dard for evaluating the workplace performances “work devotion” (Blair-Loy 2003: 5). A cultural
of mothers, and a bias against mothers in hiring, norm that mothers should always be on call for
promotion, and salary decisions. A status char- their children coexists in tension with another
acteristic is a categorical distinction among peo- widely held normative belief in our society that
ple, such as race or occupational status, that has the “ideal worker” be unencumbered by com-
attached to it widely held cultural beliefs associ- peting demands and “always there” for his or her
ating greater status worthiness and competence employer (Acker 1990; Hays 1996; Williams
with one category of the distinction over others 2001; Blair-Loy 2003). The tension between
(Berger et al. 1977). these two roles occurs at the level of normative
Theory and empirical research suggest that cultural assumptions, and not necessarily at the
ability standards are stricter for those with lower level of mothers’ own commitment to work. In-
performance expectations (Foschi 1989). The deed, Bielby and Bielby (1984) found no differ-
logic behind this prediction is that good perfor- ences in the workplace commitment of mothers
mances are inconsistent with expectations for and nonmothers. Instead it is the perceived ten-
lower status actors; therefore when they perform sion between these two roles that leads us to
well, their performances are critically scrutinized suggest that motherhood is a devalued status in
and judged by a stricter standard compared with workplace settings.
higher status actors. Thus, performances of low Therefore, we predict that mothers will be
status actors—even when “objectively” equal to rated as less competent, less committed, less
that of their high status counterparts—are less suitable for hire and promotion, and deserving
likely to be judged as demonstrating task ability of lower starting salaries compared with other-
or competence (for a comparison of status dis- wise equal women who are not mothers. In ad-
crimination and economic theories of statistical dition, we expect mothers will be judged by a
discrimination, see Correll and Benard 2006). harsher standard. Since being a good father is
not seen as culturally incompatible with being
an ideal worker (Townsend 2002), we do not ex-
Motherhood as a Status pect that fathers will experience lower workplace
Characteristic evaluations.
To understand how motherhood might function
as a devalued status characteristic in workplace
settings, it is helpful to broaden the conventional The Laboratory Experiment
usage of “performance expectations.” While re- Paid undergraduate volunteers (84 men and
searchers typically focus on the anticipated rel- 108 women) rated a pair of equally qualified,
ative competence of group members, cultural same-gender (either male or female), same-race
beliefs about the relative effort that social groups (either African American or white) fictitious job
exert in task situations can also be the basis for applicants, presented as real, who differed on
forming differentiated performance expecta- parental status. Since there were very few signif-
tions. While it is logically difficult to understand icant effects of participant gender or applicant
why taking on the motherhood role should affect race, we do not discuss these results here.
a person’s underlying competence, there is con-
siderable evidence that contemporary cultural
beliefs assume that employed mothers are less The Use of Undergraduates
committed to work than nonmothers and, con- The laboratory setting ensures sufficient control
sequently, put less effort into it (Ridgeway and over factors that would interfere with tests of our
Correll 2004). Motherhood affects perceptions hypotheses, such as other people in the room to
834 • Part VIII: Gender Inequality

prime other status characteristics, telephones, Parental status was manipulated on the resume
or other distractions, and it allows us to collect and on the human resources memo. The resume
detailed measures to fully test our argument. By described one applicant as an officer in an ele-
necessity we rely on a sample of undergraduates. mentary school Parent-Teacher Association, and
The theory presented here implies that to the ex- the accompanying memo included the follow-
tent that employers share the belief that mothers ing phrase: “Mother/Father to Tom and Emily.
are less committed to or competent in workplace Married to John/Karen.” The nonparent was
settings, they too will subtly discriminate against presented as married, but with no mention of
mothers. Qualitative and quantitative research children.
provides some evidence that employers share
this belief (Blair-Loy 2003; Crittenden 2001;
Kennelly 1999; Cleveland and Berman 1987; Dependent Measures
Cleveland 1991; Olian and Schwab 1988). The There are eight dependent measures: two that
audit study, described below, will provide more measure competence and commitment, two that
direct evidence regarding the extent to which measure the ability standard participants used to
employers discriminate against mothers. judge the applicants, and four that serve as our
key evaluation measures. The competence mea-
sure is a weighted average of participants’ ratings
Procedure of the applicants on seven-point scales ranging
Participants read a description of a company that from “not at all” to “extremely” capable, efficient,
was purportedly hiring for a mid-level market- skilled, intelligent, independent, self-confident,
ing position and examined application materials aggressive, and organized (alpha = .85). The
for two equally qualified applicants who differed commitment measure comes from a single-item
on parental status. To increase their task orien- question that asked participants how commit-
tation, participants were told that their input ted they thought the applicant would be relative
would impact actual hiring decisions. They then to other employees in similar positions at the
inspected each applicant’s file, containing three company.
items: a short memo, a “fact sheet,” and a re- There are two ability standard items. Partici-
sume. The memos contained notes purportedly pants were asked: (1) what percentile the appli-
from a company human resources staff member cant would need to score on an exam diagnostic
who conducted a screening interview with the of management ability, and (2) “how many days
applicant. The “fact sheet” summarized relevant could this applicant be late or leave early per
information about the potential employee (e.g., month before you would no longer recommend
college GPA) not presented on the resume. The him/her for management track?” We predict that
fact sheets and the resumes established that the mothers will be required to score in a higher per-
candidates were equally productive in their past centile than nonmothers before being considered
jobs and had equivalent skills and backgrounds. hirable and will be allowed fewer days of being
Prior to the actual experiment, pretesting of the late or leaving early.
two versions of the materials confirmed that There are four evaluation measures. Partic-
the resumes were perceived to be of equivalent ipants were asked: (1) to recommend a salary
quality. for each applicant, (2) to estimate the likeli-
hood that an applicant would be subsequently
promoted if hired, (3) to judge whether the ap-
Experimental Manipulations plicant, if hired, should be recommended for a
The race and gender of applicants were ma- management-training course designed for those
nipulated by altering first names on the appli- with strong advancement potential, and (4) to
cant files (Bertrand and Mullainathan 2003). decide if they would recommend each applicant
 Correll, Benard, and Paik—Getting a Job • 835

Table 96.1. Means or Proportions of Status, Standards and Evaluation Variables by Gender and Parental
Status of Applicant (Standard Deviation in Parentheses)
pp ( p )
Female Applicants Female Applicants Male Applicants Male Applicants
Mothers Non-mothers Fathers Non-fathers
Competence 5.19 (0.73)** 5.75 (0.58) 5.51 (0.68) 5.44 (0.66)
Commitment 67.0 (19.1)** 79.2 (15.2) 78.5** (16.3) 74.2 (18.6)
Days allowed late 3.16 (1.98)** 3.73 (2.01) 3.69** (2.55) 3.16 (1.85)
Percent score 72.4 (27.5)** 67.9 (27.7) 67.3 (32.7) 67.1 (33.0)
required on exam
Salary recommended $137,000** $148,000 $150,000** $144,000
(21,000) (25,000) (23,000) (20,700)
Proportion recommend .691++ .862 .936+ .851
for management
Likelihood of 2.74 (0.65)** 3.42 (0.54) 3.30* (0.62) 3.11 (0.70)
promotion
Proportion recommend .468++ .840 .734+ .617
for hire
*
p<.1, test for difference in means between parent and non-parents
**
p<.05, test for difference in means between parent and non-parents
+
z<.1, test for difference in proportion between parents and non-parents
++
z<.05, test for difference in proportion between parents and non-parents
Notes: 94 participants rated female applicants and 94 rated male applicants. For this table, the data for male and female sub-
jects are pooled, as are the data by race of applicant.

for hire. We predict that mothers will be offered Mothers were also rated significantly less pro-
lower starting salaries, will be rated as less pro- motable and were less likely to be recommended
motable, will be less likely to be recommended for management. Finally, while participants rec-
for management, and will be less likely to be rec- ommended 84 percent of female nonmothers for
ommended for hire than nonmothers. hire, they recommended a significantly lower 47
percent of mothers.
Fathers were not disadvantaged, and in fact
Laboratory Experiment Results were advantaged on some of these measures.
As predicted, mothers were judged significantly Relative to nonfathers, fathers were rated signifi-
less competent and committed than women cantly more committed to their jobs, allowed to
without children (see Table 96.1, left side) and be late to work significantly more times, and of-
were held to harsher performance and punctu- fered significantly higher salaries.
ality standards. They were allowed significantly
fewer times of being late to work and needed
a significantly higher score on the management Multivariate Analysis
exam than nonmothers before being considered We now turn to multivariate models to evaluate
hirable. Similarly, the evaluation measures show the motherhood penalty hypothesis by estimat-
significant and substantial penalties for moth- ing the effects of gender of applicant, parental
erhood. The recommended starting salary for status, and the interaction of gender of appli-
mothers was $11,000 (7.4 percent) less than cant with parental status on each of the eight
that offered nonmothers, a significant difference. dependent variables. We refer to the interaction
836 • Part VIII: Gender Inequality

Table 96.2. Estimated Regression Coefficients for the Effects of Gender, Parental Status and Race on
Applicant’s Perceived Competence and Commitment (Robust Standard Errors in Parentheses, Clustered
by Participant ID)
yp p
Independent Variables Competence Commitment
Parent 0.089 5.15 ***
(0.088) (1.73)
Female applicant 0.376 *** 5.68 **
(0.104) (2.51)
African American -0.038 -2.01
(0.090) (2.27)
Female participant 0.060 -2.61
(0.094) (2.26)
Motherhood interaction^ -0.750 *** -17.3 ***
(0.132) (2.32)
Intercept 5.42 *** 75.8 ***
(0.100) (2.55)
^
Parent * Female applicant
*
p<.1
**
p<.05
***
p<.001
Note: N=188 participants

term (gender of applicant x parental status) as significant, negative motherhood penalty inter-
the “motherhood penalty interaction.” Appli- action indicates that being a parent lowers the
cant race and participant gender are included in competence ratings for women, but not men (see
all models, and standard errors are clustered by left column, Table 96.2). Participants also per-
participant ID to take into account the nonin- ceived mothers as less committed than other ap-
dependence of observations that results from plicants: the motherhood penalty interaction is
asking participants to rate applicants in pairs. significant and negative in the model predicting
Linear regression models are used for the con- commitment ratings (see right column, Table
tinuous dependent variables. Logistic regression 96.2). The positive and significant main effect
models are estimated for the binary evaluation for parental status implies that fathers are actu-
variables (recommend for management and rec- ally rated as more committed than nonfathers.
ommend for hire). Ordered logistic regression, Consistent with the status-based discrimina-
with the proportional odds specification, is used tion argument, mothers were held to a stricter
for the ordered categorical evaluation variable, performance standard (see Table 96.3). The
likelihood of promotion. Parental status, gender motherhood interaction is significant and pos-
of applicant, gender of participant, and race of itive in the model predicting the required test
applicant are dummy variables, with parents, fe- score, while the main effects of gender of appli-
males, and African Americans coded as 1. cant and parental status are insignificant, show-
The estimated regression coefficients are pre- ing that participants require mothers (but not
sented in Tables 96.2–4. For all eight dependent fathers) to score higher on a test of management
variables, the motherhood penalty interaction is ability than other applicants before considering
significant and in the predicted direction. This them for a job. Mothers are also held to a higher
result shows strong support for the main predic- standard of punctuality, being allowed fewer
tion that parental status negatively impacts rat- days of being late.
ings for female, but not male, applicants. In Table 96.4, the motherhood penalty inter-
Confirming our prediction, mothers were action is significant and negative across all four
viewed as less competent than nonmothers. The models, indicating that mothers, relative to other
 837

Table 96.3. Estimated Regression Coefficients for the Effects of Gender, Parental Status and Race on Ability
Standard Variables (Robust Standard Errors in Parentheses, Clustered by Participant ID)

Independent Variables Days allowed late Test score required (%)

Parent 0.515 *** 1.03
(0.137) (0.968)
Female applicant 0.572 ** 1.25
(0.294) (4.52)
African American -0.361 -4.06
(0.294) (4.38)
Female participant 0.234 -9.44 **
(0.289) (4.30)
Motherhood interaction^ -1.10 *** 3.56 ***
(0.213) (1.21)
Intercept 3.22 *** 73.7 ***
(0.322) (4.27)
^
Parent * Female applicant
*
p<.1
**
p<.05
***
p<.001
Note: N=188 participants

Table 96.4. Estimated Regression Coefficients for the Effects of Gender, Parental Status and Race on Eval-
uation Variables (Robust Standard Errors in Parentheses, Clustered by Participant ID)
p yp p
Promotion Mgmt Hire? Recommended
likelihood training? (binary logistic salary in
(ordered logistic (binary logistic estimates) thousands of dollars
Independent Variables estimates) estimates) (linear estimates)
Parent 1.03 * 0.605 * 0.570 4.47 ***
(0.545) (0.321) (0.366) (1.84)
Female applicant 0.256 1.009 *** 1.21 *** 2.56
(0.425) (0.319) (0.365) (3.18)
African American 0.309 -0.211 -0.163 -6.80 **
(0.299) (0.218) (0.197) (2.94)
Female participant 0.496 * 0.526 ** 0.606 *** 0.691
(0.298) (0.226) (0.199) (2.82)
Motherhood interaction^ -2.14 *** -2.72 *** -2.38 *** -15.9 ***
(0.651) (0.426) (0.548) (2.42)
Intercept ^^ 4.56 *** 0.210 148
(0.601) (0.266) (2.55)
^
Parent * Female applicant
^^
Since ordered logistic regression produces multiple intercepts, we do not present them here.
*
p<.1
**
p<.05
***
p<.001
Note: N=188 participants
838 • Part VIII: Gender Inequality

Table 96.5. Estimated Regression Coefficients for the Mediation of Competence and Commitment on the
Impact of Parental Status on Workplace Evaluations (Robust Standard Errors in Parentheses, Clustered by
Participant ID)

Promotion Mgmt Hire? Recommended

likelihood training? (binary logistic salary in
(ordered logistic (binary logistic estimates) thousands of dollars
Independent Variables estimates) estimates) (linear estimates)
Competence 0.628 ** 1.263 *** 1.21 *** 7.00 ***
(0.295) (0.281) (0.258) (1.99)
Commitment 0.237 *** 0.206 ** 0.308 *** 1.08
(0.095) (0.099) (0.081) (0.762)
Parent 0.901 * 0.508 0.433 3.23 *
(0.558) (0.340) (0.426) (1.78)
Female applicant -0.140 0.661 ** 0.755 * -0.817
(0.426) (0.332) (0.410) (3.31)
African American 0.374 -0.154 -0.092 -6.30 ***
(0.319) (0.237) (0.244) (2.86)
Female participant 0.557 * 0.606 *** 0.755 *** 0.512
(0.316) (0.236) (0.254) (2.81)
Motherhood interaction^ -1.34 ** -1.89 *** -1.39 ** -8.52 ***
(0.646) (0.437) (0.606) (2.66)
Intercept ^^ 3.64 *** -2.09 *** 140
(0.947) (0.702) (6.37)
Percent reduction of 37.4 % 30.5 % 41.6 % 46.4 %
motherhood penalty
^
Parent * Female applicant
^^
Since ordered logistic regression produces multiple intercepts, we do not present them here.
*
p<.1
**
p<.05
***
p<.001
Note: N=188 participants

applicants, are believed to deserve lower salaries salary of approximately $148,000. Fathers were
and to be less suitable for hiring, promoting, and offered a significantly higher salary of approx-
training for management. In the model predict- imately $152,000. Women without children
ing likelihood of promotion, the main effect of were offered approximately $151,000, whereas
parental status is marginally significant and pos- mothers were recommended a significantly lower
itive, while the motherhood penalty interaction salary of about $139,000, or about 7.9 percent
is significant and negative, indicating that the less than otherwise equal childless women.
negative effect of parental status on perceptions While the motherhood penalty interaction is
of promotability accrues only to women. significant and its sign is in the predicted direc-
Consistent with previous literature on the tion for each model, to complete our argument,
motherhood wage penalty, mothers are offered we need to give evidence that motherhood dis-
lower starting salaries than other types of appli- advantages job applicants because it is a status
cants, as indicated by the significant, negative characteristic. If the theory is correct, then eval-
coefficient for the motherhood interaction term. uations of competence and commitment should
Childless men were recommended an average mediate the motherhood penalty.
 Correll, Benard, and Paik—Getting a Job • 839

Table 96.6. Proportions of Applicants Receiving Callbacks by Gender and Parental Status

Callbacks / Total Jobs Proportion Called Back

Mothers 10 / 320 .0313
Childless women 21 / 320 .0656++
Fathers 16 / 318 .0503
Childless men 9 / 318 .0283
++
z<.05 Test for difference in proportions between parents and non-parents.
Notes: Mothers and childless women applied to the same 320 jobs; fathers and childless men applied to the same 318 jobs.

When the competence and commitment rat- assigned to either the male or female condition.
ings were added as independent variables to the We manipulated parental status on the resume
models (see Table 96.5), the negative effect of and on the cover letter. We did not manipulate
motherhood status on workplace evaluations was race in this study.
significantly reduced, by a magnitude of 31–46 We monitor whether gender and parental
percent. Thus, mothers are rated as less hirable, status impact the odds that an employer will
less suitable for promotion and management call back an applicant. Based on the 5–8 percent
training, and deserving of lower salaries in part callback rate found in an audit study of race in
because they are believed to be less competent hiring (Bertrand and Mullainathan 2003), and
and less committed to paid work. Having estab- to ensure that we had sufficient statistical power
lished support for the causal mechanism with to evaluate the effect of parental status, we sub-
the laboratory data, we turn to the audit study mitted 1,276 resumes and cover letters to 638
to assess whether actual employers discriminate employers.
against mothers.

Results
The Audit Study The results suggest that real employers do dis-
The audit methodology combines experimental criminate against mothers (see Table 96.6).
design with real-life settings. As in laboratory Childless women received 2.1 times as many
experiments, audit studies isolate a characteris- callbacks as equally qualified mothers. This
tic (e.g., race or gender) and test for discrimi- finding is similar to the laboratory experiment
natory behavior. Distinct from most laboratory (see Table 96.1) in which childless women were
studies, audit study participants are the people recommended for hire 1.8 times more frequently
who make important decisions about actual than mothers. In the laboratory study, fathers
applicants, such as employers conducting new were recommended for hire at a slightly higher
employee searches. While laboratory experi- rate, although the difference was only marginally
ments permit closer investigation of social and significant; in the audit study, fathers were called
cognitive processes, audit studies provide greater back at a higher rate, although the difference was
generalizability of results. not significant.
Resumes and cover letters from a pair of ficti- We now consider a multivariate model for the
tious, equally qualified, same-gender applicants effects of parental status, applicant gender, and
were sent to employers advertising for marketing the interaction of parental status and applicant
and business job openings in a large, northeast- gender on the odds that an applicant receives a
ern city newspaper over an 18-month period of callback from an employer. Table 96.7 shows the
time. The same-sex pair contained one parent motherhood penalty interaction is significant
and one nonparent. Job openings were randomly and negative, while the main effect for parental
840 • Part VIII: Gender Inequality

status is insignificant, and the main effect for Table 96.7. Estimated Binary Logistic Regression
the female applicant variable is significant and Coefficients for the Effects of Parental Status
positive. The significant negative motherhood and Gender on the Odds of Receiving a Callback
(Robust Standard Errors in Parentheses, Clustered
penalty interaction term indicates that being a
by Job)
parent lowers the odds that a woman, but not a
man, will receive a callback from employers. In Independent Variables Callback?
sum, the audit data show that, compared with Parent 0.598
their equally qualified childless counterparts, (.433)
mothers are disadvantaged when actual employ- Female applicant 0.887**
ers make hiring decisions. (.407)
Motherhood interaction^ -1.38**
(.590)
Strengths and Limitations Intercept -3.54***
While the audit study evaluates whether actual (.338)
employers discriminate against mothers in the ^
Parent * Female applicant
*
hiring process, it does not give us insight into the p<.1
**
mechanism underlying discrimination, because p<.05
***
p<.001
it was not possible to collect employers’ rankings Notes: Mothers and childless women applied to the same
of commitment, competence, performance stan- 320 jobs; fathers and childless men applied to the same
dards, and other relevant variables. These limits 318 jobs, for a total of 1276 applications for 638 jobs.
mean that while the audit study establishes that
actual employers discriminate against mothers, it explains some of the disadvantages mothers
cannot establish why. experience in the paid labor market. Second, it
By considering the results of these two com- shows that real employers discriminate against
panion studies simultaneously, however, we mothers. The results of this study have impli-
find support for the status-based discrimination cations for understanding some of the enduring
mechanism using the laboratory data and see real patterns of gender inequality in paid work. Stud-
world implications of the argument with data ies have documented the motherhood penalty in
generated from the audit study. Further, these at least 15 countries (Harkness and Waldfogel
results are consistent with qualitative work show- 1999; Misra, Budig, and Moller 2005) and
ing that employers discriminate against mothers shown its stability over time (Avellar and Smock
(Blair-Loy 2003; Crittenden 2001; Kennelly 2003). This study offers a partial explanation for
1999) and with survey research that consistently the mechanism behind a widespread, durable
finds a wage penalty for motherhood (Budig and phenomenon with implications for a broad seg-
England 2001; Anderson, Binder, and Krause ment of the population.
2003). Thus, across a wide range of methodolog- More generally, a gender gap in wages has
ical approaches—each of which has its unique persisted despite the vast movement of women
strengths and weaknesses—we find evidence that into paid labor in the United States since the
mothers experience disadvantages in workplace early 1970s, and employed mothers account
settings and that discrimination plays a role in for most of this gap (Glass 2004). This study
producing these disadvantages. suggests that cultural beliefs about the tension
between the motherhood and “ideal worker”
roles may play a part in reproducing this pat-
Summary and Conclusions tern of inequality. A second enduring pattern
This project makes two main contributions. of gender inequality is the so-called “glass ceil-
First, it isolates and experimentally evaluates a ing,” a metaphor for the barriers that restrict
status-based discrimination mechanism that women’s movement up the career ladder to the
 Correll, Benard, and Paik—Getting a Job • 841

highest positions in organizations and firms. To than Lakisha and Jamal? A field experiment on
the extent that employers view mothers as less labor market discrimination.” National Bureau
committed to their jobs and less “promotable,” of Economic Research Working Paper Series, Work-
the glass ceiling women face could be, in part, a ing Paper no. 9873. Retrieved September 2003
(http://www.nber.org/papers/w9873).
motherhood ceiling.
Bielby, Denise D., and William T. Bielby. 1984.
Writing for the National Center for Policy “Work commitment, sex-role attitudes, and
Analysis, Denise Venable (2002) reports that women’s employment.” American Sociological Re-
among people ages 27 to 33 who have never had view 49: 234–247.
children, women’s earnings approach 98 percent Blair-Loy, Mary. 2003. Competing Devotions: Ca-
of men’s. She concludes, “When women behave reer and Family Among Women Executives. Cam-
in the workplace as men do, the wage gap be- bridge, MA: Harvard University Press.
tween them is small.” Claims of unequal pay, Budig, Michelle, and Paula England. 2001. “The
she continues, “almost always involve comparing wage penalty for motherhood.” American Socio-
logical Review 66: 204–225.
apples and oranges.” However, since most em-
Cleveland, Jeanette N. 1991. “Using hypothetical and
ployed men and employed women have children actual applicants in assessing person-organization
at some point in their lives, the most illustrative fit: A methodological note.” Journal of Applied
“within fruit” comparison is not the compar- Social Psychology 21: 1004–1011.
ison of childless men to childless women, but Cleveland, Jeanette N., and Andrew H. Berman.
the comparison of men with children to women 1987. “Age perceptions of jobs: Agreement be-
with children. As the two studies reported here tween samples of students and managers.” Psy-
show, when women “behave as men do,” by giv- chological Reports 61: 565–566.
ing evidence of being a parent, they are discrim- Correll, Shelley J., and Stephen Benard. 2006. “Bi-
ased estimators? Comparing status and statistical
inated against, while their male counterparts
theories of gender discrimination.” Pp. 89–116
are often advantaged. Far from being an “apples
in Social Psychology of the Workplace (Advances
to oranges” comparison, the male and female in Group Process Volume 23), edited by Shane R.
applicants who were evaluated in these studies Thye and Edward J. Lawler. New York: Elsevier.
were exactly equal by experimental design. That Corse, Sara J. 1990. “Pregnant managers and their
parental status disadvantaged only female appli- subordinates: The effects of gender expectations
cants is strong evidence of discrimination. on hierarchical relationships.” Journal of Applied
Behavioral Science 26: 25–48.
REFERENCES Crittenden, Ann. 2001. The Price of Motherhood:
Acker, Joan. 1990. “Hierarchies, jobs, and bodies: Why the Most Important Job in the World Is Still
a theory of gendered organizations.” Gender & the Least Valued. New York: Metropolitan Books.
Society 4: 139–158. Cuddy, Amy J. C., Susan T. Fiske, and Peter Glick.
Anderson, Deborah J., Melissa Binder, and Kate 2004. “When professionals become mothers,
Krause. 2003. “The motherhood wage penalty warmth doesn’t cut the ice.” Journal of Social Is-
revisited: Experience, heterogeneity, work effort sues 60: 701–718.
and work-schedule flexibility.” Industrial and Foschi, Martha. 1989. “Status characteristics, stan-
Labor Relations Review 56: 273–294. dards and attributions.” Pp. 58–72 in Sociological
Avellar, Sarah, and Pamela Smock. 2003. “Has Theories in Progress: New Formulations, edited by
the price of motherhood declined over time? Joseph Berger, Morris Zelditch, Jr., and Bo An-
A cross-cohort comparison of the motherhood derson. Boston: Houghton Mifflin.
wage penalty.” Journal of Marriage and the Fam- Glass, Jennifer. 2004. “Blessing or curse?
ily 65: 597–607. Work-family policies and mother’s wage growth
Berger, Joseph, Hamit Fisek, Robert Norman, and over time.” Work and Occupations 31: 367–394.
Morris Zelditch, Jr. 1977. Status Characteristics Halpert, Jane A., Midge L. Wilson, and Julia Hick-
and Social Interaction. New York: Elsevier. man. 1993. “Pregnancy as a source of bias in per-
Bertrand, Marianne, and Sendhil Mullainathan. formance appraisals.” Journal of Organizational
2003. “Are Emily and Greg more employable Behavior 14: 649–663.
842 • Part VIII: Gender Inequality

Harkness, Susan, and Jane Waldfogel. 1999. “The Organizational Behavior and Human Decision
family gap in pay: Evidence from seven indus- Processes 41: 180–195.
trialised countries.” CASE paper 29: Centre for Ridgeway, Cecilia, and Shelley J. Correll. 2004.
Analysis of Social Exclusion. “Motherhood as a status characteristic.” Journal
Hays, Sharon. 1996. The Cultural Contradictions of of Social Issues 60: 683–700.
Motherhood. New Haven, CT: Yale University Townsend, Nicholas W. 2002. The Package Deal:
Press. Marriage, Work and Fatherhood in Men’s Lives.
Kennelly, Ivy. 1999. “That single mother element: Philadelphia, PA: Temple University Press.
How white employers typify black women.” Gen- Venable, Denise. 2002. “The wage gap myth.” Na-
der & Society 13: 168–192. tional Center for Policy Analysis, April 12, 2002,
Misra, Joya, Michelle Budig, and Stephanie Moller. Brief Analysis no. 392. Retrieved December 12,
2005. “Employment, wages, and poverty: Recon- 2004 (http://www.ncpa.org/pub/ba/ba392/).
ciliation policies and gender equity.” Paper pre- Waldfogel, Jane. 1997. “The effect of children on
sented at the Annual Meeting of the American women’s wages.” American Sociological Review 62:
Sociological Association, Philadelphia, PA. 209–217.
Olian, Judy D., and Donald P. Schwab. 1988. Williams, Joan. 2001. Unbending Gender: Why Work
“The impact of applicant gender compared and Family Conflict and What to Do About It. Ox-
to qualifications on hiring recommendations: ford: Oxford University Press.
A meta-analysis of experimental studies.”

97 • András Tilcsik

Do Openly Gay Men Experience Employment

Discrimination?

In recent years the rights of and legal protec- sexual orientation in employment contexts. Al-
tions for lesbian, gay, bisexual, and transgen- though scholars have produced a considerable
der (LGBT) people have been at the center amount of research relevant to this debate, most
of heated debates in the United States. In the of the literature has focused on wage inequality
absence of a federal law specifically protecting and has produced little direct evidence about
LGBT employees and job seekers, one debate the difficulties that LGBT people may face in
has focused on sexual orientation discrimination obtaining a job. This is a significant omission,
in employment; that is, the behaviors and prac- because hiring discrimination is an important
tices—both deliberate and unconscious—that inequality-generating mechanism with poten-
disadvantage individuals of a particular sexual tially powerful effects on a job seeker’s access to
orientation compared to individuals of another a broad range of opportunities.

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: András Tilcsik, “Pride and Prejudice: Employment Discrimination Against Openly
Gay Men in the United States,” American Journal of Sociology 117:2 (2011), pp. 586–626, published by the
University of Chicago Press. The article printed here was originally prepared by András Tilcsik for this fourth
edition of Social Stratification: Class, Race, and Gender in Sociological Perspective, edited by David B. Grusky.
Copyright © 2014 by Westview Press.
 Tilcsik—Do Openly Gay Men Experience Employment Discrimination? • 843

I begin to address this lacuna by directly ex- perceptions, rather than the actual incidence, of
amining hiring discrimination against openly discrimination. A third approach focused on the
gay men. Limiting the scope of this study to one number of employment discrimination com-
LGBT group—gay men—was advantageous, plaints that LGBT employees filed in states that
because the precise nature of prejudice based outlaw sexual orientation discrimination (e.g.,
on sexual orientation might vary among LGBT Rubenstein 2002). However, like self-reports,
groups. By focusing on a single group—and complaint rates do not necessarily represent the
leaving future research to explore discrimination actual incidence of discrimination.
against other LGBT groups—it was possible to Seeking more direct evidence about sexual
delve more deeply into the nature of discrimina- orientation discrimination in the United States,
tion against that one group, gay men. To do so, some researchers have adopted an experimental
I conducted an audit study. I responded with a approach. Crow, Fok, and Hartman (1998), for
pair of fictitious but ostensibly real resumes to example, asked full-time employees in a southern
1,769 postings of white-collar, entry-level jobs city to select six out of eight fictitious applicants
in seven states, randomly assigning a signal of for an accounting position. Similarly, Horvath
sexual orientation to each resume. The findings and Ryan (2003) instructed undergraduates to
from this study provide evidence about the ex- rate resumes for which sexual orientation and
tent of discrimination as well as the factors that gender were experimentally manipulated. Taking
affect the likelihood of its occurring, including a somewhat different approach, Hebl and col-
the presence of antidiscrimination laws and the leagues (2002) conducted a field experiment in
extent to which employers value stereotypically which male and female confederates applied for
male heterosexual personality traits. retail jobs in a mall. For each store, the confed-
erates were randomly assigned to wear a baseball
hat with the words “Gay and Proud” or “Texan
Prior Research and the Audit and Proud.” Although these experiments were an
Approach important first step toward directly measuring
Much of the extant literature on sexual orienta- hiring discrimination, they had significant lim-
tion discrimination in the United States has fo- itations. First, all three experiments were limited
cused on compensation. Controlling for human to a single context: a single city, university, or
capital, these studies have found that gay men mall area. Thus, it is unclear how accurately the
earn 10–32 percent less than heterosexual men results reflect broader patterns of discrimina-
(Badgett et al. 2007). Although these studies tion. Second, in two of these studies the decision
generated important evidence, they focused on makers knew that they were participating in an
earnings rather than on the hiring decision and experiment and that their choices had no con-
were not designed to provide direct evidence sequences for real hiring outcomes (Crow et al.
about discrimination. Indeed, regression anal- 1998; Horvath and Ryan 2003). Whether these
yses that define discrimination as unexplained decision makers would make the same hiring
income differences may lead to biased estimates choices in a real employment context, faced with
of discrimination if differences in employee real incentives and constraints, remains unclear.
productivity or preferences are observed in- To overcome these limitations, I conducted a
completely. Another line of research examined large-scale correspondence audit, sending pairs
employee self-reports and found that many of fictitious resumes in response to real job post-
LGBT individuals report experiencing some ings and assessing the effect of a randomly as-
form of discrimination in the workplace (e.g., signed signal of sexual orientation on whether a
Croteau 1996). The generalizability of these resume elicited an invitation for a job interview.
studies, however, is limited, because they rely This methodology offers important advantages.
on convenience samples and capture subjective By experimentally controlling for human capital
844 • Part VIII: Gender Inequality

factors that might be confounded with minority In addition to geographic differences, we

status, audit studies provide more direct evidence might also observe important variation in the
about the causal impact of discrimination than level of discrimination across jobs if stereotypes
do wage regressions (Bertrand and Mullainathan of gay men play a role in hiring decisions. Ste-
2004; Pager 2007). By gathering such evidence reotypes are socially shared sets of implicit or
in a real employment context from real employ- explicit beliefs about the typical characteristics
ers, audit studies are also more generalizable than of members of a social group. Because stereo-
studies with undergraduate participants and ex- typed judgments simplify and justify social real-
periments in which participants know that their ity, stereotyping has potentially powerful effects
choices will not affect real hiring outcomes (Cor- on how people perceive and treat one another.
rell et al. 2007). To date, however, few studies However, stereotyping has received relatively
have taken an audit approach to sexual orien- little empirical attention in the audit literature.
tation discrimination (Adam 1981; Hebl et al. One notable exception is a correspondence test
2002; Weichselbaumer 2003; Drydakis 2009), in Austria that compared the callback rates of
and there has been no large-scale audit of sexual three fictitious job applicants (a man, a feminine
orientation discrimination in the United States. woman, and a masculine woman) for job post-
ings that emphasized different personality traits
(e.g., “powerful,” “dynamic,” “friendly”; We-
Discrimination, Regional Variation, ichselbaumer 2004). This research design made
and Stereotypes it possible to assess whether employers looking
The likelihood of discrimination against gay men for job candidates with stereotypically masculine
probably varies across geographic areas and jobs. traits favored men over women and masculine
First, differences in the level of tolerance toward women over feminine women.
gay men and in antidiscrimination laws may lead I adopt a similar approach. If stereotyped
to geographic variation in the incidence of dis- judgments influence callback decisions, employ-
crimination. Public opinion polls indicate con- ers should be more likely to engage in discrim-
siderable regional variation in attitudes toward ination if they value and emphasize attributes
gay men. Whereas almost half of Americans in that gay men are stereotypically perceived to
the Northeast and the West have a favorable view lack. What might these attributes be? Research
of gay men, only slightly more than a third of suggests that gay men are often perceived to ex-
respondents express similar views in the Mid- hibit behaviors associated with “feminine” char-
west, and even fewer do so in the South (Pew acteristics; for example, they are commonly seen
Research Center 2003). The adoption of state as sensitive, emotional, gentle, affectionate, and
laws that prohibit sexual orientation discrimi- passive. Consequently, gay men are often seen as
nation follows a similar geographic pattern. At lacking “toughness” and “masculinity” (Madon
the time of this study, twenty states and the Dis- 1997). Even a quick perusal of job postings re-
trict of Columbia prohibited sexual orientation veals that it is not uncommon for employers to
discrimination in the private sector, but most of emphasize personality characteristics that are
these states were in the Northeast and the West. considered typical of heterosexual men, such
The geographic distribution of counties and cit- as decisiveness, assertiveness, and aggressiveness
ies that ban sexual orientation discrimination in (Bem 1974; Gorman 2005). Searches in online
private employment was roughly similar, with job databases return hundreds of postings in
relatively few antidiscrimination laws in south- which employers seek, for example, “an aggres-
ern cities and counties. Notably, however, such sive, motivated self-starter,” “an assertive associ-
laws have been passed in some major cities in ate,” or “a decisive, results-oriented leader.” This
the South and the Midwest, including Atlanta, emphasis on stereotypically male heterosexual
Austin, Chicago, Dallas, and Detroit. characteristics may in turn be associated with a
 Tilcsik—Do Openly Gay Men Experience Employment Discrimination? • 845

higher likelihood of discrimination against gay irrelevant to a job application. Thus, instead of
men. If stereotypes of gay men—as feminine, being just a member of a gay and lesbian cam-
passive, gentle, or lacking “toughness”—play a pus organization, the applicant served as the
significant role in callback decisions, employers elected treasurer for several semesters, managing
who characterize their ideal job candidate with the organization’s financial operations. Accord-
stereotypically male heterosexual traits should be ingly, rather than focusing on the organization’s
particularly likely to engage in discrimination. nature or goals, this resume item explicitly em-
phasized the applicant’s managerial and finan-
cial skills. Thus the applicant’s participation in
Methods this organization could be seen as a meaningful,
Signaling Sexual Orientation valuable experience with potentially important
An important challenge in resume-based audit transferable skills. In other words, omitting this
studies is to signal the characteristic of interest experience from the resume would have meant
without introducing a confounding factor into concealing relevant and important human cap-
the analysis. In the present case, for example, ital. Given the heavy emphasis on the specific
employers may perceive openly gay applicants financial and organizational activities associated
as tactless or lacking business savvy because they with the treasurer position on the resume, it
list an irrelevant experience on their resumes, would be difficult to dismiss this experience as a
apparently “trumpeting” their sexual orientation social activity or a way of simply trumpeting the
(Weichselbaumer 2003). In addition, perceiving applicant’s sexual orientation.
such applicants as radical or liberal, employers Third, I used a control organization to ensure
may discriminate against them for their per- that any observed differences in callbacks could
ceived political views and activism, rather than be attributed to antigay discrimination rather
for their sexual orientation (Badgett et al. 2007). than other factors. An important consideration
Moreover, if the “control organization” that is as- was that participation in a gay organization
signed to the resume of the ostensibly heterosex- might be associated with progressive, liberal, or
ual applicant is not carefully chosen, it may lead leftist political views. Thus, if I had used an apo-
to differences in the applicants’ perceived level of litical control organization (or no control orga-
human capital, making it difficult to assess the nization at all), observed differences in callbacks
extent of discrimination. might have been attributable to discrimination
I took several steps to address these issues.1 based on either sexual orientation or political
First, the fictitious job seekers in this study affiliation, and it would have been impossible
were graduating college seniors applying for to determine the net effect of sexual orienta-
entry-level jobs. For this population of appli- tion. Accordingly, to determine whether there is
cants, listing volunteer experiences in political, a “gay penalty” above and beyond the possible
cultural, ethnic, religious, or other identity-based effect of political discrimination, I chose a con-
campus organizations in a resume is a common trol organization that is associated with leftist or
practice, especially if the experience involves an progressive views. At first glance, a campus chap-
elected position with important responsibilities. ter of college Democrats might seem suitable
Thus, in the case of college seniors, listing in- for this purpose. However, because Democratic
volvement in political or identity-based groups campus groups are typically larger than gay and
is less likely to be perceived as unprofessional lesbian student groups, leadership experiences
or unusual than it would be if experienced job in a Democratic organization might seem more
seekers did so. valuable than similar experiences in a gay organi-
Second, to signal homosexual orientation, zation. To avoid this problem, the control group
I chose an experience in a gay community or- was a small left-wing campus organization (the
ganization that could not be easily dismissed as “Progressive and Socialist Alliance”).
846 • Part VIII: Gender Inequality

cities and counties that banned sexual orienta-

Resumes, Randomization, and Sample tion discrimination.
Over six months in 2005, I sent fictitious re-
sumes, via e-mail, in response to advertisements
for full-time, entry-level positions on three re- Findings
cruitment Web sites targeted to college seniors The submission of resumes led to a total of 331
and recent graduates. I sent two resumes in interview invitations, an overall callback rate of
response to each job posting, with one day or 9.35 percent. Table 97.1 tabulates callback rates
less in between. Before sending out each resume by sexual orientation. Whereas heterosexual ap-
pair, I randomly assigned the gay signal to one plicants had an 11.5 percent chance of being
of the resumes and the control signal to the invited for an interview, equally qualified gay
other. Thus, even though the resumes differed applicants had only a 7.2 percent chance of re-
in order to avoid raising suspicion, there was no ceiving a positive response. This is a difference
systematic relationship between their quality and of 4.3 percentage points, or about 40 percent.
sexual orientation. Consequently, any significant This gap is statistically significant (P < .001)
difference in callback rates could be attributed to and implies that a heterosexual job seeker had
the experimental manipulation of the resumes. to apply to fewer than nine different jobs to ob-
Importantly, I varied only the name of the treat- tain a positive response, whereas a gay applicant
ment organization and the control organization. needed to reply to almost fourteen ads to achieve
As a result, the activities of the applicants in the same result.
their respective groups were not systematically The size of the callback gap, however, varied
related to sexual orientation. I thus effectively substantially across states. On the one hand, in
controlled for any differences in the applicants’ the southern and midwestern states (Texas, Flor-
achievements in the treatment organization and ida, and Ohio) there was a substantial difference
the control organization. For each application, in the callback rates for gay and heterosexual
I recorded whether it led to an invitation to a applicants. In Texas and Ohio, for example, the
first-round job interview (either an in-person or size of the callback gap (8.3 and 8.6 percentage
a telephone interview). points, respectively) was substantially larger than
I submitted a total of 3,538 resumes, respond- in the overall sample (4.1 percentage points). By
ing to 1,769 job postings by private employers. contrast, there was no statistically significant
The sample included jobs in five occupations and callback gap in any of the western and north-
seven states. The five occupations were managers eastern states (California, Nevada, Pennsylvania,
(including management trainees), business and and New York). Similarly, there was variation in
financial analysts, sales representatives, customer the callback gap across legal environments. In
service representatives, and administrative assis- the case of employers subject to a relevant an-
tants. The sampled states included four states in tidiscrimination law at the city, county, or state
the Northeast and the West (New York, Pennsyl- levels, the callback gap was less than 3 percentage
vania, California, Nevada) and three states in the points; in the case of employers not subject to
Midwest and the South (Ohio, Florida, Texas). such regulation, the gap was as large as 6 per-
These states—and the counties and cities in centage points. In both cases, however, the call-
them—offered an intriguing mosaic of legal en- back gap was statistically significant.
vironments. California, Nevada, and New York In addition, as expected the callback gap was
prohibited sexual orientation discrimination in particularly large among employers who em-
private employment, but the other four states phasized the importance of stereotypically male
had no such legislation. At the same time, with heterosexual traits, particularly assertiveness, ag-
the exception of Nevada, each state had some gressiveness, and decisiveness. Notably, however,
there was a statistically significant callback gap
 Tilcsik—Do Openly Gay Men Experience Employment Discrimination? • 847

Table 97.1. Callback Rates by Sexual Orientation

Sample (n job ads) % Callback Difference
Not
Gay Gay Ratio (P-value)
Total Sample (n = 1,769) 11.5 7.2 1.59 4.3 (.000)
California (n = 337) 11.0 9.2 1.20 1.8 (.443)
Nevada (n = 131) 12.2 6.1 2.00 6.1 (.087)
New York (n = 236) 10.2 11.4 0.89 -1.2 (.656)
Pennsylvania (n = 201) 12.9 9.4 1.37 3.5 (.268)
Ohio (n = 219) 14.1 5.5 2.56 8.6 (.002)
Florida (n = 347) 9.5 5.5 2.11 4.0 (.044)
Texas (n = 298) 12.0 3.7 3.24 8.3 (.000)
Employers subject to a city, county,
or state law that prohibits sexual
orientation discrimination:
Yes (n = 983) 11.6 8.7 1.33 2.9 (.037)
No (n = 786) 11.3 5.3 2.13 6.0 (.000)
Job postings that require
stereotypically male heterosexual
traits:
Yes (n = 361) 13.0 3.9 3.33 9.1 (.000)
No (n = 1,408) 11.1 8.1 1.37 3.00 (.007)
*Assertiveness, aggressiveness, or decisiveness

even within the sample of employers who did not at the initial point of contact, employers more
specifically require such traits. Thus, even when readily disqualify openly gay applicants than
these traits were not emphasized, gay applicants equally qualified heterosexual applicants. These
suffered significant callback discrimination. findings are consistent with less direct indicators
Using logistic regression models, I found that of discrimination against LGBT people; taken
these basic patterns were robust to the inclusion together, these lines of evidence suggest that sex-
of control variables, including occupation and ual orientation discrimination is a prominent
industry dummies, employer size, and the local feature of many American labor markets.
unemployment rate, as well as the degree of ur- At the same time, the results indicate dra-
banness, political conservatism, and level of ed- matic geographic variation in the level of dis-
ucation characteristic of the area where the job crimination. What explains this variation? A key
was located. Under a variety of conditions, these finding is that employers in states and counties
models consistently indicated discrimination with a relevant antidiscrimination law were sig-
overall, as well as variation in the level of dis- nificantly less likely to engage in discrimination.
crimination by region and by the traits required This result, however, does not necessarily imply
in the job postings. Moreover, the effect of geo- that lower levels of discrimination were due en-
graphic differences and the influence of required tirely to antidiscrimination laws. Because public
personality traits were significant even net of opinion about gay people might affect both the
each other. level of discrimination and the probability that
an antidiscrimination policy will be adopted,
the effect of laws on discrimination may be con-
Conclusions founded with the effect of attitudes. Data limita-
This study shows that openly gay men encounter tions prevented me from fully untangling these
significant barriers in the hiring process because, factors in the present study, but the mechanisms
848 • Part VIII: Gender Inequality

behind the observed geographic variation are an Findings.” Journal of Vocational Behavior 48:
important topic for future research. 195–209.
Moreover, this study has identified stereotyp- Crow, Stephen M., Lillian Y. Fok, and Sandra J.
ing as a potentially important mechanism un- Hartman. 1998. “Who Is at Greatest Risk of
Work-Related Discrimination—Women, Blacks,
derlying hiring discrimination against gay men.
or Homosexuals?” Employee Responsibilities and
Employers who sought applicants with stereo- Rights Journal 11: 15–26.
typically male heterosexual traits were much Drydakis, Nick. 2009. “Sexual Orientation Discrim-
more likely to discriminate against gay appli- ination in the Labour Market.” Labour Economics
cants than employers who did not emphasize the 16: 364–372.
importance of such traits. This finding suggests Gorman, Elizabeth. 2005. “Gender Stereotypes,
that employers’ implicit or explicit stereotypes Same-Gender Preferences, and Organizational
of gay men are inconsistent with the image of Variation in the Hiring of Women: Evidence
an assertive, aggressive, and decisive employee. from Law Firms.” American Sociological Review
70: 702–728.
It seems, therefore, that the discrimination doc-
Hebl, Michelle R., Jessica Bigazzi Foster, Laura M.
umented in this study is partly rooted in specific Mannix, and John Dovidio. 2002. “Formal and
stereotypes and cannot be completely reduced to Interpersonal Discrimination: A Field Study of
a general antipathy against gay employees. Bias Toward Homosexual Applicants.” Personality
and Social Psychology Bulletin 28: 815–825.
NOTES Horvath, Michael, and Ann Marie Ryan. 2003.
1. For further details on this experiment, the “Antecedents and Potential Moderators of the
analyses, and robustness checks, see Tilcsik (2011). Relationship Between Attitudes and Hiring Dis-
crimination on the Basis of Sexual Orientation.”
REFERENCES Sex Roles 48: 115–130.
Adam, Barry. 1981. “Stigma and Employability: Dis- Madon, Stephanie. 1997. “What Do People Believe
crimination by Sex and Sexual Orientation in the About Gay Males? A Study of Stereotype Con-
Ontario Legal Profession.” Canadian Review of tent and Strength.” Sex Roles 37: 663–685.
Sociology and Anthropology 18: 216–221. Pager, Devah. 2007. “The Use of Field Experiments
Badgett, M. V. Lee, Holning Lau, Brad Sears, and for Studies of Employment Discrimination:
Deborah Ho. 2007. “Bias in the Workplace: Contributions, Critiques, and Directions for the
Consistent Evidence of Sexual Orientation and Future.” Annals of the American Academy of Polit-
Gender Identity Discrimination.” Report, Wil- ical and Social Science 609: 104–133.
liams Institute, University of California School Pew Research Center. 2003. “Religious Beliefs Un-
of Law. derpin Opposition to Homosexuality.” http://
Bem, Sandra. 1974. “The Measurement of Psycho- www.people-press.org/report/197/religiou
logical Androgyny.” Journal of Consulting and s-beliefs-underpin-opposition-to-homosexuality.
Clinical Psychology 42: 155–162. Rubenstein, William B. 2002. “Do Gay Rights Laws
Bertrand, Marianne, and Sendhil Mullainathan. Matter? An Empirical Assessment.” Southern Cal-
2004. “Are Emily and Greg More Employable ifornia Law Review 75: 65–119.
Than Lakisha and Jamal? A Field Experiment on Tilcsik, András. 2011. “Pride and Prejudice: Em-
Labor Market Discrimination.” American Eco- ployment Discrimination Against Openly Gay
nomic Review 94: 991–1013. Men in the United States.” American Journal of
Correll, Shelley J., Stephen Benard, and In Paik. Sociology 117 (2): 586–626.
2007. “Getting a Job: Is There a Motherhood Weichselbaumer, Doris. 2003. “Sexual Orientation
Penalty?” American Journal of Sociology 112: Discrimination in Hiring.” Labour Economics 10:
1297–1338. 629–642.
Croteau, James M. 1996. “Research on the Work ———. 2004. “Is It Sex or Personality? The Impact
Experiences of Lesbian, Gay, and Bisexual Peo- of Sex Stereotypes on Discrimination in Appli-
ple: An Integrative Review of Methodology and cant Selection.” Eastern Economic Journal 30:
159–186.
 Reskin—Rethinking Employment Discrimination and Its Remedies • 849

98 • Barbara F. Reskin

Rethinking Employment Discrimination

and Its Remedies

The paradigmatic definition of discrimination Economists have theorized that employers

is the differential treatment of persons because may also intentionally discriminate against in-
of status characteristics that are functionally dividuals on the rational rather than emotional
irrelevant to the outcome in question (Merton ground that they belong to a race or sex thought
1972; see also Allport 1954, 51). Implicit in this to be less productive or more costly to employ,
definition is the proposition that individuals dis- or that they will otherwise adversely affect
criminate because of their negative feelings (an- business because of customers’ or coworkers’
tipathy, distaste, fear) toward or negative beliefs prejudices (Phelps 1972; Bergmann and Dar-
(social stereotypes) about members of a status ity 1981). Economists term this intentional,
group (Allport 1954, 14; Becker 1957). Media non-animus-based discrimination “statistical dis-
accounts of whites’ hostility toward blacks prior crimination.” The Equal Employment Oppor-
to and during the civil rights movement lent face tunity Commission (EEOC) and the Supreme
validity to an antipathy-based theory of discrim- Court (Los Angeles Department of Water and Power
ination. Although the success of the civil rights v. Manhart 1978; 435 U.S. 702)—the federal
movement depended partly on its ability to ex- agencies charged with enforcing and interpreting
pose the antagonism that many whites harbored Title VII—have interpreted the law as prohib-
toward blacks, it simultaneously reinforced the iting intentional, non-animus-based discrimi-
theory that discrimination resulted from the nation, including statistical (stereotype-based)
deliberate efforts of individual whites to harm discrimination. The law prohibits employers
blacks because of their hostility to them. from deliberately using race and sex in personnel
In emphasizing differential treatment based decisions, regardless of their motive (except to
on or because of an irrelevant characteristic, social redress egregious discrimination of the past; see
scientists construed discrimination as intentional Reskin 1998). Thus, the major U.S. antidiscrim-
behavior. So too did the major antidiscrimina- ination law bans individuals—whether acting on
tion law, Title VII of the 1964 Civil Rights Act. their own behalf or as agents of their employer—
Although the statute did not explicitly define from intentionally discriminating regardless of
discrimination, it conceptualized it as an in- whether their motives derive from negative sen-
tended and consciously motivated act (Stryker timents toward a group or from concerns about
2001, 18). Thus, the law operationalized illegal the economic ramifications of employing mem-
employment discrimination as intentional. bers of that group.

Barbara F. Reskin. “Rethinking Employment Discrimination and Its Remedies,” in The New Economic So-
ciology: Developments in an Emerging Field, edited by Mauro F. Guillén, Randall Collins, Paula England, and
Marshall Meyer (2002), pp. 218–230, 232–233, 237, 239–244. Copyright © 2002 Russell Sage Foundation,
112 East 64th Street, New York, NY 10065. Reprinted with permission.
850 • Part VIII: Gender Inequality

There is ample evidence that individu- automatically categorize the people we encoun-
als continue to engage in intentional acts that ter into ingroups and outgroups, into “we” and
conform to the paradigmatic definition of dis- “they” (Rothbart and Lewis 1994; Brewer and
crimination (Kirschenman and Neckerman Brown 1998). In 1908 Georg Simmel wrote that
1991; Kasinitz and Rosenberg 1996; Browne we necessarily think of others in terms of general
and Kennelly 1999; U.S. Equal Employment categories (1971 [1908], 10–12). Almost half a
Opportunity Commission 2001a–g). However, century later Gordon Allport (1954, 19) echoed
both the paradigmatic theory of intentional this insight, asserting that “the human mind
disparate treatment and the theory of statistical must think with the aid of categories.” Recent
discrimination ignore another important source research supports these claims: we understand
of employment discrimination: automatic, non- the social world in terms of categorical distinc-
conscious cognitive processes.1 In this chapter, tions and readily categorize others into ingroups
I review research that implies that employment and outgroups, on the basis of both highly visi-
decisions are routinely biased as a result of nor- ble as well as arbitrary and trivial characteristics
mal cognitive processes. If nonconscious cogni- (Brewer 1997, 200; Brewer and Brown 1998,
tive processes—unless checked by employment 566; Fiske 1998, 364, 375).
structures—regularly contribute to discrimina- A critical concomitant of categorization is
tion, our theoretical approach to discrimination the generalization of similarity and difference.
needs to be modified. Having categorized others on trait A, we infer
In the years since the animus theory of dis- both that ingroup members resemble us and that
crimination gained paradigmatic status, social outgroup members differ from us on traits B, C,
cognition researchers have established that au- D, and so on. As a result, we systematically un-
tomatic cognitive processes give rise to cognitive derestimate within-group differences and exag-
errors whose results are likely to be discrimina- gerate between-group differences on a variety of
tory. Largely outside individual control, these characteristics (Brown 1995, 78). Categorization
processes are marked by a cognitive efficiency also affects how and how quickly we encode and
that makes them generally adaptive. Because recall information about others, thereby (unbe-
of their automaticity, neither intention nor an- knownst to us) distorting our perceptions and
tipathy is a necessary condition for discrimina- shaping our preferences (Perdue et al. 1990).
tory outcomes. The role of automatic cognitive It is important to note that we automatically
processes in discrimination means that we also prefer ingroup members to outgroup members.
cannot continue to view discrimination as anom- We are more comfortable with them, feel more
alous acts by a few biased individuals within an obligated and loyal to them, impute to them
otherwise fair system of employment opportu- positive attributes, trust them, remember their
nities (Black 1989). Instead, we must recognize positive traits while forgetting their negative
that discriminatory outcomes—many of them ones, are predisposed to cooperate with them,
involving micro acts of discrimination—are per- and favor them when distributing rewards (Per-
vasive, permitting members of favored groups to due et al. 1990; Baron and Pfeffer 1994; Roth-
accumulate advantages while members of disfa- bart and Lewis 1994, 369; Brown 1995; Brewer
vored groups accumulate disadvantages. 1997, 201; Brewer and Brown 1998, 567; Fiske
1998, 361).
Although we tend to distrust and deperson-
Social Cognitive Processes and alize outgroup members and see them as com-
Employment Discrimination petitors (Brewer 1997, 201), laboratory research
Categorization indicates that we apparently do not automat-
The core cognitive process that links gender and ically see them in negative ways (Perdue et al.
race to workplace outcomes is categorization. We 1990, 482) or discriminate against them when
 Reskin—Rethinking Employment Discrimination and Its Remedies • 851

allocating negative outcomes (Brewer and Brown

1998, 559). Together, this body of research sug- Stereotyping
gests that automatic categorization gives rise to Categorization leads to a second automatic cog-
discrimination primarily through pervasive in- nitive process with potentially discriminatory
group favoritism rather than pervasive outgroup effects: stereotyping. Stereotyping is an inferen-
antipathy. tial process in which we attribute traits that we
Automatic categorization is not inevitably habitually associate with a group to individuals
group-serving. We categorize others into in- who belong to that group. Although conscious
groups and outgroups largely independently stereotypes can culminate in intentional statis-
of our conscious feelings toward the groups tical discrimination (see, for example, Wilson
or our desire to protect our own status (Fiske 1996, ch. 5; Moss and Tilly 1996, 264), here I
1998, 364). Nonetheless, ingroup preferences address the unintentional discriminatory effects
reinforce dominant groups’ privileges. For ex- of nonconscious stereotypes. Almost invariably,
ample, group status affects whether people infer noticing that a “target” belongs to a stereotyped
“illusory correlations” between positive traits and group automatically brings to mind characteris-
group membership. Members of high-status in- tics that are stereotypically linked to that group,
groups show more ingroup favoritism than do even when we consciously reject those stereo-
members of low-status ingroups (Brewer and types as false (Bodenhausen, Macrae, and Garst
Brown 1998, 570). Also, ingroup attachment is 1998, 316). In fact, we automatically pursue,
especially strong among people who think that prefer, and remember “evidence” that supports
group membership affects their opportunities or our stereotypes (including untrue “evidence”)
risks, both among competing groups and among and ignore, discount, and forget facts that chal-
ingroups in a numerical minority (Weber 1994). lenge them. Stereotypes have been described
Group membership is often based on as- as “hypotheses in search of evidence” (Brown
cribed characteristics. Ingroup favoritism cannot 1995) because we are predisposed toward be-
operate unless we can readily distinguish who is havioral interpretations that conform to our
“us” and who is “them” (Brewer 1997, 205). stereotypes, blind to stereotype-inconsistent in-
Given its cultural salience and socially exagger- terpretations, and cognitively better able to find
ated visibility, we habitually use sex in assigning stereotype-consistent than-inconsistent evidence
others to ingroups or outgroups (Brewer and (Fiske 1998, 367). This selectivity means that
Brown 1998). Because sex and race are strongly our nonconscious stereotypes include a confir-
correlated with control over workplace oppor- matory bias (Brown 1995).
tunities, with white men monopolizing these Given our cognitive capacity to override ev-
roles, ingroup preference favors whites, men, and idence with stereotypes, it stands to reason that
especially white men. Thus, ingroup favoritism we invoke stereotypes when we lack complete
produces status-group discrimination. information about others. For example, given
In sum, concomitant to our automatic sort- the stereotype of blacks as lazy (Bobo 2001),
ing of people into categories are automatic biases an employer who is choosing between a black
in our feelings toward others based on whether applicant and a white applicant and has limited
we see them as “us” or “them.” The effects of information about the black is likely to assume
such biases are discriminatory, although they automatically that the black applicant is lazy. Ste-
are neither intended nor even known to us as reotypes also distort our recollections of others.
we perpetrate them (Fiske 1998, 362). In addi- We more readily remember stereotypical than
tion, they apparently operate more through our nonstereotypical descriptions—an attractive
automatic inclination to favor others like our- flight attendant, for example, is more memora-
selves than to deprive or punish those who differ ble than an unattractive flight attendant (Brown
(Brewer and Brown 1998; DiTomaso 2000). 1995, 96). As discussed later, stereotypes also
852 • Part VIII: Gender Inequality

bias the attributions we automatically construct stereotypic statements about women were more
for others’ behavior; these attribution errors in likely to ask a female job applicant sexist ques-
turn distort our assessments of others and our tions than were nonprimed men, and they took
expectations of their future behavior. longer to recognize nonsexist words (Fiske et al.
Automatic stereotypes often prevail even 1999). These findings imply that an idle remark
when we are consciously motivated not to ste- can activate sex or race stereotypes that in turn
reotype. For instance, experimental subjects affect how decisionmakers assess job or pro-
asked to judge the heights of women and men motion candidates or other workers (Heilman
from photographs routinely underestimated 1995). Importantly, by priming race or sex ste-
women’s height relative to same-height men, al- reotypes, diversity training may actually foster
though they had been instructed to judge each distorted impressions and differential treatment
photograph as an individual and they knew that based on sex or race.
accurate judgments could yield a substantial pay- Status affects the propensity to automatic
off (Brown 1995, 92). Subjects in another study stereotyping. Subordinates are less likely to ste-
who were instructed to suppress their stereotypes reotype members of dominant groups than the
were able to do so, but in a subsequent task they reverse, presumably because they need to make
expressed stronger stereotypes than members of accurate assessments of the people who control
a control group who had not been told to avoid their work environment and rewards (Fiske et
stereotyping (Bodenhausen et al. 1998, 326). al. 1999, 241). In addition, dominant-group
Automatic stereotyping is tenacious because members appear to have greater confidence in
it is cognitively efficient (Fiske 1998, 366; Fiske, their stereotyped assessments than do members
Lin, and Neuberg 1999, 237, 244). We can pro- of subordinate groups, probably both because
cess information that supports our stereotypes their dominant-group status gives them confi-
more quickly than inconsistent information. dence and because being wrong is less likely to
Time pressure, information overload, mental have negative consequences.
busyness, and pressure to make a decision—con- In sum, social cognition research suggests that
ditions that characterize the work lives of many unconscious sex and race stereotyping can lead
decisionmakers—increase our likelihood of ste- to discrimination by distorting our impressions
reotyping and exacerbate the effects of stereotypes of individuals based on group membership. This
on judgment and memory, presumably because occurs, unless checked, when group-based ste-
these situations require cognitive efficiency reotypes automatically are applied to individuals,
(Bodenhausen et al. 1998: 319; Fiske 1998, replace missing information about individuals,
389; Goodwin, Operario, and Fiske 1998, 694). distort what we recall of others’ behavior, bias
For example, subjects in a sentence-completion our attributions of others’ success and failures,
task could refrain from making sexist statements and influence our predictions of others’ future
when they had ample time; given time limits, behavior.
however, their statements were more sexist than
those of members of a control group (Bodenhau-
sen et al. 1998, 326). Attribution Error
Any attribute that is stereotypically linked to How we expect others to perform affects the
a status characteristic can activate (or “prime”) meaning we assign to their behavior. When
group stereotypes (Heilman 1995; Fiske 1998, people’s performance conforms to our expec-
366). For example, research subjects who were tations, we attribute it to their stable, internal
subliminally exposed to the words “black” and traits (for example, ability); when it contradicts
“white” were more likely to link negative traits our expectations, we attribute it to transient,
to blacks and positive traits to whites than those external causes (for example, task difficulty or
who were not subliminally exposed (Dovidio luck). Both ingroup versus outgroup member-
et al. 1997), and men who were primed with ship and sex and race stereotypes influence our
 Reskin—Rethinking Employment Discrimination and Its Remedies • 853

expectations about others’ performance. We individuals. Their efficiency frees up cognitive

expect members of socially preferred groups to resources for other purposes, making these pro-
succeed and members of devalued groups to fail. pensities cognitively adaptive to individuals,
Thus, we tend to chalk up successes by ingroup especially those who are juggling multiple de-
members and white men and failures by out- mands (Brown 1995, 95). Given what seems to
group members, minorities, and white women be the near-impossibility of suppressing these
to their stable predispositions, while we attribute cognitive propensities, the resulting biases are
failures by ingroup members and white men and pervasive and predictable. Because they link
successes by outgroup members, minorities, and employment outcomes to functionally irrelevant
white women to external (and hence, unstable) group membership, they are also discriminatory.
factors (Crocker, Major, and Steele 1998, 539). In addition, although many of the daily dis-
For example, we credit a man’s talents for his criminatory events stemming from cognitive bi-
success in a customarily male job, while we at- ases such as excluding someone from a project
tribute a woman’s success to situational factors, or denying them credit for a success may seem
such as help from others. Importantly, we do not trivial (and certainly not legally actionable), over
see successes or failures that we have attributed time members of preferred groups accumulate
to external causes as predicting future success advantages and members of disparaged groups
(Swim and Sanna 1996; Brewer and Brown accumulate disadvantages. Eventually group dif-
1998, 560). ferences emerge that appear to justify unequal
Moreover, our brains tend to encode behavior career outcomes (Krieger 1995). Thus, unless
that conforms to our expectations in terms of ab- employers implement structures to check the bi-
stract traits (intelligence, honesty, laziness) and asing effects of these microlevel processes, their
encode unexpected behavior in concrete terms long-term consequences create or exacerbate
(his car would not start; she took the class twice). macrolevel disparities across race and sex groups
Thus, we attribute an ingroup member losing in their economic and social fates.
his temper to provocation and the same behavior In view of the automaticity of these cognitive
by an outgroup member to personality. Attribu- processes, the proximate cause of employment
tions in abstract terms are more likely than ones discrimination becomes whether employers’
in concrete terms to affect our global evaluation personnel practices prevent or permit these cog-
of others as well as our predictions of their fu- nitive propensities from linking workers’ status
ture performance, and they are more resistant to characteristics to their employment outcomes
challenge by counterevidence. As a result, these (Bielby 2000; Reskin 2000). In the next section,
cognitive propensities give ingroup members I discuss employment structures and practices
and persons from socially valued groups the ben- that can reduce their discriminatory impact.
efit of the doubt, while preventing new data on
outgroup members and persons in socially sub-
ordinate status groups from dispelling negative Organizational Practices, Cognitive
attributions. Biases, and Discriminatory
Once again, while these propensities are Outcomes
universal, perceptions by members of majority Work organizations cannot stamp out our pro-
groups are more prone to attribution error than pensities to categorize others automatically or
those by minority-group members (Brown 1995, to treat ingroup members more favorably than
101–2). outgroup members. As we have seen, attempts
to repress stereotyping can backfire. But employ-
ment structures can curb the biasing effects of
Micro-Macro Links these cognitive processes, thus inhibiting the dis-
Automatic categorization, ingroup preference, crimination that would otherwise result (Fiske
and stereotyping are cognitively efficient for 1998, 375). The impact of audition practices on
854 • Part VIII: Gender Inequality

women’s share of positions in major symphony triggered stereotypes about Asians, while sup-
orchestras illustrates my point. With the adop- pressing stereotypes about women (Bodenhau-
tion in the 1970s and 1980s of blind auditions sen et al. 1998, 317). In general, recategorization
that physically screened the musician from the exploits people’s readiness to categorize others on
auditors, women’s odds of being selected rose the most trivial criteria by encouraging the cate-
sharply (Goldin and Rouse 2000). Screens that gorization of workers on characteristics that are
concealed the sex of the candidates circumvented independent of their sex and race (Brewer and
intentional discrimination and obviated the Brown 1998, 583). Employers can encourage re-
discriminatory consequences of nonconscious categorization based on work-related functions,
sex stereotyping and ingroup favoritism. Social for example. Function-based categories such as
cognition and organizational research points teams, projects, and divisions are cognitively
to employment practices or structures that can available and exploitable by employers through
arrest—or activate—the biasing effects of cog- work-group culture and structured competition.
nitive processes. Based on that scholarship, this Firms can also emphasize recategorization to a
section proposes organizational practices that “superordinate category” such as an officer of the
mitigate the biasing effects of automatic cogni- law or a UPS employee (Brewer 1997, 202). . . .
tive processes.

Organizational Practices and

Organizational Practices and Stereotyping
Categorization Although the propensity to automatic stereo-
The automaticity of ingroup preference probably typing is universal, whether we stereotype oth-
plays an important role in producing disparate ers varies, and whether our stereotypes give rise
employment outcomes for different race or sex to biased cognitions depends on several factors,
groups. One way to minimize its discriminatory such as the type and quality of information that
effects is to dissociate ingroup membership from is available to decisionmakers about “targets”;
ascribed statuses. As Marilynn Brewer (1997, whether the decisionmaking process highlights
205) observes, a precursor of ingroup favoritism status-group membership and hence primes
is the identification of who is “us” and who is stereotyping; decisionmakers’ awareness of po-
“them.” Employers can take advantage of this tential biases and their motivation to perceive
fact by discouraging ingroups based on race or others accurately; and whether decisionmakers
sex through decategorization and recategorization. have the cognitive resources to consciously min-
Decategorization conveys individuating in- imize stereotype-based distortions (Bodenhausen
formation about workers to decisionmakers et al. 1998, 330).
(which in turn reduces stereotyping). Recatego- Making available accurate individuating in-
rization recognizes that people have many char- formation about persons from negatively ste-
acteristics that compete to serve as the basis of reotyped groups can prevent stereotyping. (But
categorization. According to Galen Bodenhau- feeling informed in the absence of objective in-
sen and her colleagues (1998, 316), “the stereo- formation may promote stereotyping; see Fiske
types associated with the category that ‘wins’ the 1998, 387.) When we lack information that
competition” are activated, while we suppress the we need to make employment decisions, our
“losing” stereotype. Consider, for example, an ex- descriptive stereotypes automatically provide
periment in which exposure to a photograph of it. Because stereotypes about socially devalued
a Chinese woman activated subjects’ stereotypes groups tend to be negative and those about so-
about women, while inhibiting their stereotypes cially favored groups to be positive, the effect
about Asians, but exposure to a picture of the is usually discriminatory. In addition, because
same woman eating noodles with chopsticks multiple bases for categorizing a person compete
 Reskin—Rethinking Employment Discrimination and Its Remedies • 855

for activation, providing relevant information (Is remedies misconstrue discrimination as based in
the applicant a college graduate? An experienced antipathy toward a group. Despite the applica-
manager?) about members of stereotyped groups bility of the paradigmatic construction of dis-
can prevent their being stereotyped by sex, race, crimination in many instances, that construction
and other ascribed characteristics (Bodenhausen ignores an even more important reason individ-
et al. 1998, 316). Thus, employment practices uals’ sex and race are routinely and illegitimately
that increase the amount of unambiguous, rel- linked to employment rewards: automatic non-
evant information available to decisionmakers conscious cognitive processes that distort our
may suppress stereotyping (Heilman 1995, 12). perceptions and treatment of others.
By compiling and disseminating to decisionmak- Categorization, ingroup preference, stereo-
ers standardized information for all candidates typing, attribution error—all part of the nor-
on all the criteria relevant to a personnel deci- mal information processing that occurs largely
sion, organizations can reduce stereotyping. . . . outside our conscious control—bias our per-
ceptions, evaluations, and treatment of people
because of their sex, color, accent, and other dis-
Decisionmakers’ Discretion cernible characteristics that signal membership
Discretion is an important perquisite for man- in more or less valued groups. They engender
agers. It not only permits autonomy but allows biases because we automatically categorize others
decisionmakers to control the amount of effort into ingroups and outgroups—often based on
they must expend. Although organizational these same characteristics. Moreover, we prefer
members tend to assume that managers’ deci- ingroup members to outgroup members, evalu-
sions are rational (Baron 1984, 56), discretion ate them more favorably, treat as factual impres-
invites unconscious and conscious biases to in- sions that are based on stereotypes, and make
fluence their decisions, thus spawning discrim- erroneous inferences about the successes or fail-
inatory treatment. Employment decisions that ures of ingroup and outgroup members. These
are based on unstructured observations are es- cognitive biases occur during interpersonal rela-
pecially vulnerable to cognitive biases (Fiske et tions, and interpersonal relations are an intrin-
al. 1991). The key to preventing such automatic sic part of most jobs. Thus, ordinary, everyday
biases from giving rise to unintended micro acts cross-sex and cross-race interactions elicit auto-
of discrimination lies in curbing decisionmak- matic ingroup favoritism, stereotyping, and at-
ers’ discretion by requiring specific procedures tribution biases. These cognitive distortions lead
for distributing opportunities. . . . on a daily basis to micro acts of discrimination:
omitting a qualified but female inside candidate
from the short list for a top position because no
Conclusion one thinks of her; not recognizing the contribu-
In keeping with the paradigmatic conception of tion that a Japanese woman makes to a project;
discrimination as conscious invidious treatment assuming that women do not want challeng-
motivated by antipathy, Congress outlawed in- ing assignments; not inviting a black man to a
tentional discrimination in Title VII of the 1964 meeting of his specialty group; being confused
Civil Rights Act. Although the enforcement of about which black worker made a suggestion;
Title VII has reduced employment discrimi- giving a white male a tip on a job opening (Cose
nation (Burstein 1989; Burstein and Edwards 1993; Fiske 1998, 371–72; Catalyst 1999; Bar-
1994), decades after the enactment of Title VII rett 1999; DiTomaso 2000; Reskin and McBrier
the careers of thousands of Americans continue 2000, 224; Sturm 2001, 113, 136).
to be stunted by employment discrimination At the individual level, these “countless small
(Blumrosen et al. 1998; Reskin 2001). Discrim- acts by a changing cast of characters . . . incre-
ination persists partly because antidiscrimination mentally and consistently limit the employment
856 • Part VIII: Gender Inequality

prospects of one group of workers compared with Project Report 3. Newark, N.J.: Rutgers Univer-
those of another” (Nelson and Bridges 1999, sity School of Law.
241–43, emphasis added; see also Feagin 2000, Bobo, Lawrence D. 2001. “Racial Attitudes and Re-
139; Sturm 2001, 113). As a result, workers who lations at the Close of the Twentieth Century.” In
America Becoming: Racial Trends and Their Conse-
regularly benefit from ingroup preference accu-
quences, vol. 1, edited by Neil J. Smelser, William
mulate advantages, and members of disparaged J. Wilson, and Faith N. Mitchell. Washington:
groups who suffer the negative discriminatory National Academy Press.
effects of cognitive biases accumulate disad- Bodenhausen, Galen V., C. Neil Macrae, and Jen-
vantages. Over time, career disparities between nifer Garst. 1998. “Stereotypes in Thought and
sex- and race-based groups also accumulate. In Deed: Social Cognitive Origins of Intergroup
the aggregate, the unchecked consequences of Discrimination.” In Intergroup Cognition and
automatic cognitive biases help to preserve job Intergroup Behaviors, edited by Constantine Se-
segregation, promotion differences, and earnings dikides, John Schopler, and Chester A. Insko.
Mahwah, N.J.: Erlbaum.
disparities by race and sex. That they result from
Brewer, Marilynn B. 1997. “The Social Psychology
nonconscious cognitive processes rather than of Intergroup Relations: Can Research Inform
conscious antipathy does not mitigate their im- Practice?” Journal of Social Issues 53(1): 197–211.
pact or reduce the need for more effective policy Brewer, Marilynn B., and Rupert J. Brown. 1998.
remedies. The results are the same. “Intergroup Relations.” In Handbook of Social
Psychology, edited by Daniel T. Gilbert, Susan
NOTES T. Fiske, and Gardner Lindzey. New York:
1. According to Shiffrin and Schneider (1977), McGraw-Hill.
automatic cognitive processes are those that are not Brown, Rupert. 1995. Prejudice. Oxford: Blackwell.
effortful, intentional, or consciously controlled. Browne, Irene, and Ivy Kennelly. 1999. “Stereotypes
and Realities: Images of Black Women in the
REFERENCES Labor Market.” In Latinas and African American
Allport, Gordon W. 1954. The Nature of Prejudice. Women at Work, edited by Irene Browne. New
Reading, Mass.: Addison-Wesley. York: Russell Sage Foundation.
Baron, James N. 1984. “Organizational Perspectives Burstein, Paul. 1989. “Attacking Sex Discrimination
on Stratification.” Annual Review of Sociology 10: in the Labor Market: A Study in Law and Poli-
37–69. tics.” Social Forces 67: 641–65.
Baron, James N., and Jeffrey Pfeffer. 1994. “The So- Burstein, Paul, and Mark E. Edwards. 1994. “The
cial Psychology of Organizations and Inequality.” Impact of Employment Discrimination Litiga-
Social Psychology Quarterly 57: 190–209. tion on Racial Disparity in Earnings: Evidence
Barrett, Paul M. 1999. The Good Black: A True Story and Unresolved Issues.” Law and Society Review
of Race in America. New York: Dutton. 28: 79–111.
Becker, Gary. 1957. The Economics of Discrimina- Catalyst. 1999. “Women of Color Report a ‘Con-
tion. Chicago: University of Chicago Press. crete Ceiling’ Barring Their Advancement in
Bergmann, Barbara R., and William Darity Jr. Corporate America.” Catalyst Press Room, July 13.
1981. “Social Relations, Productivity, and Em- Available at: www.catalystwomen.org/press/media
ployer Discrimination.” Monthly Labor Review kit/release071399woc.html (May 20, 2001).
104(April): 47–49. Cose, Ellis. 1993. Rage of the Privileged Class. New
Bielby, William T. 2000. “Minimizing Workplace York: HarperCollins.
Gender and Racial Bias.” Contemporary Sociology Crocker, Jennifer, Brenda Major, and Claude Steele.
29(1): 120–29. 1998. “Social Stigma.” In Handbook of Social
Black, Donald. 1989. Sociological Justice. New York: Psychology, edited by Daniel T. Gilbert, Susan
Oxford University Press. T. Fiske, and Gardner Lindzey. New York:
Blumrosen, Alfred W., Mark Bendick, J. J. Miller, McGraw-Hill.
and Ruth G. Blumrosen. 1998. “Employment DiTomaso, Nancy. 2000. “Why Antidiscrimina-
Discrimination Against Women in Washing- tion Policies Are Not Enough: The Legacies and
ton State, 1997.” Employment Discrimination Consequences of Affirmative Inclusion—for
 Reskin—Rethinking Employment Discrimination and Its Remedies • 857

Whites.” Rutgers University, Newark. Unpub- Moss, Philip, and Chris Tilly. 1996. “Soft Skills
lished manuscript. and Race: An Investigation of Black Men’s Em-
Dovidio, John F., Kerry Kawakami, Craig Johnson, ployment Problems.” Work and Occupations 23:
Brenda Johnson, and Adaiah Howard. 1997. 252–76.
“On the Nature of Prejudice: Automatic and Nelson, Robert L., and William S. Bridges. 1999.
Controlled Processes.” Journal of Experimental Legalizing Gender Inequality. New York: Oxford
Social Psychology 33: 510–40. University Press.
Feagin, Joe R. 2000. Racist America: Roots, Current Perdue, Charles, John F. Dovidio, Michael B. Gut-
Realities, and Future Reparations. New York: man, and Richard B. Tyler. 1990. “Us and Them:
Routledge. Social Categorization and the Process of Inter-
Fiske, Susan T. 1998. “Stereotyping, Prejudice, and group Bias.” Journal of Personality and Social Psy-
Discrimination.” In Handbook of Social Psychol- chology 59: 475–86.
ogy, edited by Daniel T. Gilbert, Susan T. Fiske, Phelps, Edmund S. 1972. “The Statistical Theory of
and Gardner Lindzey. New York: McGraw-Hill. Racism and Sexism.” American Economic Review
Fiske, Susan T., Donald N. Bersoff, Eugene Borgida, 62: 659–61.
Kay Deaux, and Madeline E. Heilman. 1991. Reskin, Barbara F. 1998. The Realities of Affirmative
“Social Science Research on Trial: Use of Sex Ste- Action. Washington, D.C.: American Sociologi-
reotyping Research in Price Waterhouse v. Hop- cal Association.
kins.” American Psychologist 46: 1049–60. ———. 2000. “The Proximate Causes of Discrim-
Fiske, Susan T., Monica Lin, and Steven L. Neuberg. ination: Research Agenda for the Twenty-First
1999. “The Continuum Model: Ten Years Later.” Century.” Contemporary Sociology 29: 319–29.
In Dual Process Theories in Social Psychology, ed- ———. 2001. “Employment Discrimination and
ited by Shelly Chaiken and Yaacov Trope. New Its Remedies.” In Sourcebook on Labor Market
York: Guilford Press. Research: Evolving Structures and Processes, ed-
Goldin, Claudia, and Cecilia Rouse. 2000. “Orches- ited by Ivar Berg and Arne Kalleberg. New York:
trating Impartiality: The Impact of ‘Blind’ Audi- Plenum.
tions on Female Musicians.” American Economic Reskin, Barbara F., and Debra B. McBrier. 2000.
Review 90: 715–41. “Why Not Ascription?: Organizations’ Employ-
Goodwin, Stephanie A., Don Operario, and Susan ment of Male and Female Managers.” American
T. Fiske. 1998. “Situational Power and Interper- Sociological Review 65: 210–33.
sonal Dominance Facilitate Bias and Inequality.” Rothbart, Myron, and Susan Lewis. 1994. “Cogni-
Journal of Social Issues 54: 677–98. tive Processes and Intergroup Relations: A His-
Heilman, Madeline E. 1995. “Sex Stereotypes and torical Perspective.” In Social Cognition: Impact
Their Effects in the Workplace: What We Know on Social Psychology, edited by Patricia G. Devine
and What We Don’t Know.” Journal of Social Is- and David L. Hamilton. San Diego: Academic
sues 10: 3–26. Press.
Kasinitz, Philip S., and Jan Rosenberg. 1996. “Miss- Simmel, Georg. 1971 [1908]. On Individuality and
ing the Connection: Social Isolation and Em- Social Forms: Selected Writings. Edited and with
ployment on the Brooklyn Waterfront.” Social an introduction by Donald N. Levine. Chicago:
Problems 43: 180–96. University of Chicago Press.
Kirschenman, Joleen, and Kathryn M. Neckerman. Stryker, Robin. 2001. “Disparate Impact and the
1991. “‘We’d Love to Hire Them But . . .’: The Quota Debates: Law, Labor Market Sociology,
Meaning of Race for Employers.” In The Urban and Equal Employment Policies.” Sociological
Underclass, edited by Christopher Jencks and Quarterly 42(1): 13–46.
Paul Peterson. Washington, D.C.: Brookings Sturm, Susan. 2001. “Second Generation Employ-
Institution. ment Discrimination: A Structural Approach.”
Krieger, Laura H. 1995. “The Contents of Our Cat- Columbia Law Review 101(3): 101–207.
egories: A Cognitive Bias Approach to Discrim- Swim, Janet K., and Lawrence J. Sanna. 1996.
ination and Equal Employment Opportunity.” “He’s Skilled, She’s Lucky: A Meta-analysis of
Stanford Law Review 47: 1161–1248. Observers’ Attributions for Women’s and Men’s
Merton, Robert K. 1972. “Insiders and Outsiders.” Successes and Failures.” Personality and Social Psy-
American Journal of Sociology 78: 9–47. chology Bulletin 22: 507–19.
858 • Part VIII: Gender Inequality

U.S. Equal Employment Opportunity Commis- ———. 2001e. “EEOC Settles Racial Harassment
sion. 2001a. “EEOC and Private Plaintiffs Settle Suit Against Georgia-Pacific Corporation.”
Harassment Suit for $485,000 Against Chicken EEOC Press Releases (April 3). Available at: www.
Processing Plant.” EEOC Press Releases (April 10). eeoc.gov/press/4-3-01.html (June 20, 2001).
Available at: www.eeoc.gov/press/4-10-01.html ———. 2001f. “EEOC Sues Two Indiana Em-
(June 20, 2001). ployers for Race Harassment.” EEOC Press Re-
———. 2001b. “EEOC Responds to Final Report leases (May 22). Available at: www.eeoc.gov/
of Mitsubishi Consent Decree Monitors.” EEOC press/5-22-01.html (June 20, 2001).
Press Releases (May 23). Available at: www.eeoc. ———. 2001g. “Joe’s Stone Crab Liable for In-
gov/press/5-23-01.html (June 20, 2001). tentional Discrimination Court Rules in Sex
———. 2001c. “EEOC Seeks to Join Nationwide Bias Suit Brought by EEOC.” EEOC Press Re-
Sex Discrimination Suit Against Rent-A-Cen- leases (March 28). Available at: www.eeoc.gov/
ter.” EEOC Press Releases (March 12). Available press/3-28-01.html (June 20, 2001).
at: www.eeoc.gov/press/3-2-01.html (June 20, Weber, Joseph G. 1994. “The Nature of Ethno-
2001). centric Attribution Bias: Ingroup Protection or
———. 2001d. “EEOC Settles Bias Suit for $2.6 Enhancement?” Journal of Experimental Social
Million Against TWA.” EEOC Press Releases (May Psychology 20: 177–94.
24). Available at: www.eeoc.gov/press/5-24-01. Wilson, William J. 1996. When Work Disappears.
html (June 20, 2001). Chicago: University of Chicago Press.

99 • Trond Petersen

Discrimination
Conscious or Nonconscious?

Introduction inequality and that they hence need to be reg-

The struggle for equality between men and ulated through legislation and other means.
women in the workplace resulted in one of the Underlying the presumption was an implicit
most significant achievements of the twentieth and often explicit purposive-action framework.
century and probably the single most lasting leg- Employers were seen as consciously treating men
acy of the 1960s. It changed in major ways how and women differently, either due to animus or
organizations operate and how men and women to cost considerations.
live their lives. A remarkable and perhaps unexpected chal-
The movement was prefaced on the idea lenge to such reasoning has arisen with great
that employers were a major culprit for gender strength over the last decade. It originated in

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth
in the following publication: Trond Petersen, “Motive and Cognition,” in Understanding Choice: Explaining
Behavior, edited by Elster et al. (2006), pp. 225–248, published by Unipub Forlag/Oslo Academic Press. The
article printed here was originally prepared by Trond Petersen for Social Stratification: Class, Race, and Gender
in Sociological Perspective, Third Edition, edited by David B. Grusky, pp. 780–785. Copyright © 2008 by
Westview Press.
 Petersen—Discrimination • 859

cognitive psychology, recently bolstered by neu- One motive arises from animus, where an em-
roscience, and has spread to sociology and parts ployer dislikes or harbors a grudge against
of legal and economic scholarship. It has fur- women and/or ethnic minorities. The other mo-
thermore entered the consciousness of journal- tive arises from statistical considerations, where
ists and human resource managers in the United women on average are viewed to be less produc-
States and Europe. It calls for abandoning the tive, referred to as statistical discrimination.
standard purposive-action framework for under- Discrimination, from animus or statistical
standing female disadvantage in the workplace. considerations, is rational in the sense that (a) it
The claim behind the challenge is this. Dis- reflects preference, (b) it is intended, and (c) the
crimination is no longer intended and conscious. employer is conscious about (a) and (b). Action
It results today primarily from cognitive pro- is conscious and directed toward achieving goals
cesses that are nonconscious. The mechanisms that the agent values.
behind discrimination are hardwired by biology In case of animus discrimination, someone
into our brains and are not accessible to human who acts on it maximizes his preferences. An em-
decision makers. ployer is willing to lose money in order to avoid
Reskin (2000, p. 327), a sociologist, writes hiring women and/or minorities. The preference
about the sources of discrimination: “I have ar- itself, the animus against women or minorities, is
gued that much of it results from nonconscious widely seen to be irrational, and certainly despi-
cognitive processes.” Valian (1998, pp. 2, 22), cable. But once in place, acting upon it, as such,
a psychologist, refers to the sources as “a set of is not irrational. Economists refer to this as taste
implicit, or nonconscious hypotheses” or as “the discrimination (Becker 1971).
unexpressed and nonconscious nature of gender Discrimination from statistical consider-
schemas and their subterranean mode of action.” ations has the same structure. An employer may
This view extends beyond researchers. In a harbor no animus against groups, but treating
significant U.S. media story, one commentator them differently may maximize preferences for
stated in a front-page newspaper article: “Every- higher profits. The underlying reason is that
body has nonconscious gender bias. It shows up groups may differ in how productive they are.
in every study” (Rimer 2005, p. A15). For example, average productivity may be higher
In this chapter I discuss the challenge posed for men in the construction industry and higher
by this literature. I first outline the standard for women in electronics assembly. It may, how-
view stemming from a purposive-action frame- ever, be costly to assess individual productivity
work where agents are viewed as being conscious prior to hire or once hired. Hiring primarily
about the motivations for their actions, then from the more productive group may then save
outline the challenge from cognitive psychology, on hiring costs and increase productivity and
followed by an evaluation, and ending with the profits.
implications for jurisprudence, our understand- Legislation, judicial practice, and much pub-
ing of gender inequality, and for its solutions. lic policy were premised on the idea that dis-
The focus is on gender inequality in the work- crimination is intentional, and hence conscious,
place, such as hiring, wages, and promotions. I and also rational in the sense that the agent is
abstract from the separate and different issues trying to satisfy his preferences, from either ani-
that arise in race inequality, where in addition mus or desire for increased profits. Legislation, at
to discrimination, disadvantages stemming from a minimum, made discriminatory choices cost-
class play a large role. lier. For those acting on animus, it would be-
come riskier to satisfy these. For those engaging
in statistical discrimination, the law would erode
Standard View: Conscious Action the expected profit advantage. Legislation hence
In the standard view there are two main motives shifts the trade-offs, and rational agents should
of discriminatory behaviors from employers. be expected to adjust accordingly.
860 • Part VIII: Gender Inequality

While the standard view of conscious action A person with no conscious animus or stereo-
undoubtedly applies to disparate treatment dis- type, with no conscious intent to discriminate,
crimination, it needs to be modified for analyz- nevertheless, due to entirely automatic noncon-
ing disparate impact discrimination. The latter scious cognitive processes, may end up making
refers to the situation where two groups are biased evaluations, may discriminate, and, to top
treated equally within the given set of organiza- it off, will be totally unaware that the behavior
tional rules and procedures, but where the pro- was discriminatory.
cedures advantage one group over another and at This change in theoretical orientation from
the same time are irrelevant for job performance. conscious to nonconscious processes was in
An example in hiring would be a physical height part prompted by the realization among many
requirement unrelated to job performance, researchers that conscious prejudice and overt
which would favor men over women. The hir- discrimination no longer are central sources of
ing agent or institution may be unaware that disadvantage for women and minorities. There
the procedures are discriminatory, and there is are, however, still substantial inequalities be-
thus no discriminatory motive, conscious or tween groups. The purported central cause is
nonconscious. now nonconscious cognitive biases.

The Challenge: Nonconscious Evaluation

Processes The claim that discrimination primarily is non-
What is then the challenge from cognitive psy- conscious rests on the view that most discrimina-
chology? It proceeds from exactly the opposite tion arises from bias, giving less role to statistical
assumptions of the purposive-action framework. discrimination. Before discussing this claim, the
As an explanation of discrimination it claims evidence for it, and its implications, it is useful
that there is (a) no necessary bias against women, to assess the presumption that most discrimina-
(b) no necessary intent to treat the sexes differ- tion arises from nonconscious bias. I have no
ently, and (c) discriminatory agents are driven by empirical information to evaluate this claim, nor
nonconscious preferences or biases. do I think any other researcher has. But I think
The reasons for discrimination are now cog- it needs one important qualification.
nitive structures unknown to the agent. These I would like to suggest that in the instances
structures are part and parcel of how we think where discrimination has occurred, it is pru-
and function as human beings. They involve dent to reserve judgment on the prevalence of
processes of categorization (e.g., by sex), the nonconscious versus conscious practices. To the
use of simplified schemas, and errors in infor- extent that it arises from bias, I think most dis-
mation processing and causal attribution. Such crimination is nonconscious. As argued by cog-
nonconscious cognitive processes can lead to bi- nitive psychologists, one would have to search
ases against women and minorities, without the long to find many card-carrying sexists proudly
agent ever being aware of it. The biases could thinking that “I will do my part for the patriar-
reflect broader societal stereotypes, or may just chical project to keep women from getting hired
reflect hardwiring of the brain that compels hu- and promoted.”
mans to make categorizations. But some employers may still consciously act
The forceful claim from cognitive psychol- upon correct, or what they perceive to be correct,
ogy is that the nonconscious schemas we use, to- statistical information. This is entirely conscious.
gether with the nonconscious biases, can crowd And it is entirely rational. It may minimize hir-
out well-meaning conscious preferences and in- ing costs and maximize profits. It is still illegal:
tentions. They may result in a total disconnec- A woman should not be treated as a representa-
tion between conscious intent and actual action. tive member of her sex; qualifications should be
 Petersen—Discrimination • 861

evaluated separately for each person. This could often while carefully monitoring their potential
well be the most prevalent source of employment adverse impacts on women and minorities. This
discrimination. will be the case in most large organizations. Parts
There are also instances where employers act of employment processes will clearly involve
on incorrect statistical information, though be- on-the-spot decisions, such as how to greet in-
lieving it to be correct, and then engage in what terviewees, and may hence be less deliberate. But
has been called error discrimination (England much happens both before and after that may
1992, chap. 3). Though based on unfounded counterbalance split-second biases. And deci-
stereotypes, positive or negative, such action sions are not only made with care, in midsized
would not be nonconscious. The stereotypes will to large organizations they are typically made by
in all likelihood be eradicated as better informa- multiple people, including equity officers, and
tion becomes available. create an extensive paper trail of job resumes, re-
Exactly what evidence then exists for the ports from meetings, and written justifications.
claim of nonconscious discrimination? Most of The decisions are far removed from those made
the evidence comes from laboratory experiments in the laboratory.
on undergraduate students at U.S. and European The difference between the two decision-
universities and now increasingly from com- making situations is hence stark: mostly spot
puter experiments conducted on large samples and split-second in experiments with minimal
(Dasgupta 2004). consequences, versus deliberate and slow in the
The main procedure currently used for de- workplace with almost invariably significant
tecting nonconscious bias is the Implicit Associa- consequences.
tion Test (see Greenwald, McGhee, and Schwartz The question is not one of truth: cogni-
1998). Subjects are asked to compare words at a tive science has almost certainly delivered real
brisk pace while sitting in front of a computer findings. The question is one of applicability,
screen. Since the interval between receiving a whether the findings extend to decisions in the
stimulus and making a response is short, often workplace, and moreover one of balance—what
a third or less of a second, the claim is that the is the relative weight of nonconscious processes
response is beyond the subject’s conscious con- in creating differential treatment and outcomes.
trol and hence must be caused by nonconscious
processes. The tests document widespread biases.
But do these tests and the conclusions from Implication 1: Jurisprudence
them identify the central problems women face Implications for legal practice and for how or-
in the workplace today? ganizations construct their procedures to avoid
One need be careful with respect to drawing litigation may vary according to whether dis-
wide-ranging conclusions from laboratory and crimination is conscious, and often rational,
computer experiments. Besides the question of versus nonconscious and probably irrational. In
whether the results reflect animus versus stereo- the latter case there is no discriminatory intent,
types or statistical generalizations (Arkes and which may lead to difficulties in the courts.
Tetlock 2004), there is a broader issue that needs White and Krieger (2001, p. 510) argue that
addressing. How undergraduates, but also older proving nonconscious discrimination should rely
people, react in experimental settings may be on the same types of evidence as proving covert
entirely different from how decisions are made but conscious discrimination; cases where the
by experienced people in real-life settings where employer hides his discriminatory motives. The
consequences may be far-reaching economi- idea is that in such cases it should be sufficient
cally, socially (workplace culture), and legally. to show that a disparate outcome by sex or race
Employment decisions are often made with cannot be justified. One infers from the outcome
considerable deliberation, not on the spot, and that the employer acted from bias, in spite of no
862 • Part VIII: Gender Inequality

direct evidence of it. The same idea can be used major problem, then penalties and enforcement
for nonconscious discrimination. In either case, need to be strengthened. This will shift incen-
the courts would in principle not have to rule tives. There will also be shaming of violators,
on whether the motive was conscious, but not an added cost. For statistical discrimination this
revealed, or nonconscious. should be sufficient. For nonconscious discrimi-
nation, the suggested remedies would depend on
the kinds of evidence for bias that are acceptable.
Implication 2: Inequality If the nonconscious motive must be shown, there
For understanding how inequality is created, and is a difficulty. But if all that is required is proof
for public policy, the prevalence and importance of a disparate outcome that appears unjustified,
of a mechanism matters. We need to assess which strengthened enforcement and penalties should
are the central forces in generating inequality: again suffice. Many employers are aware that
discrimination, conscious and nonconscious, ed- nonconscious bias may exist, understand the dif-
ucational choices, family adaptations, and more. ficulties of proving or disproving this, and may
Krieger (1995, p. 1164) argues with respect take precautions to guard against accusations
to nonconscious discrimination that “these sub- and litigation. A purposive-action framework
tle forms of bias, I suggest, represent today’s serves well to analyze such adaptive behavior.
most prevalent type of discrimination.” And Although only a small portion of equality issues
concerning its scope, White and Krieger (2001, can be brought to the courts, legislation and
p. 509) state: “that ‘nonconscious’ discrimina- legal practice can set important frameworks for
tion frequently occurs is well documented.” how employers will act, on the assumption that
But how can these processes, which are non- the overwhelming majority of employers seek to
conscious and often hardly noticeable, add up comply with the law.
to a major gender gap in employment? Valian Scholars taking their view from cognitive
(1998, p. 18) stresses that “a succession of small science offer several kinds of advice for fighting
events, such as not getting a good assignment, discrimination arising from nonconscious biases.
result in large discrepancies in advancement and Some scholars see the need for broader cul-
achievement.” They are like accrued interest, tural changes. Jolls and Sunstein (2006) argue
which in any one year makes little difference, for the role of government information cam-
but over time adds up. paigns aided by the law. These may sensitize de-
There is no question that subtle and noncon- cision makers to nonconscious biases and induce
scious mechanisms may operate. The question is them to safeguard against their operation. In a
whether they are the central causes of the major similar vein, Krieger (1995, pp. 1245–1247) sees
differences we still observe between men and the biases as part of a broad cultural problem,
women. I would like to suggest that other fac- and concludes that to alter the cognitive sche-
tors are more important and overpower the small mas we use thus requires a broad-based cultural
causes, namely the educational choices and fam- solution.
ily adaptations men and women make. These are Other scholars focus on workplace processes.
observable, not at all subtle, and undoubtedly Bielby (2000) argues that organizations need
have major impacts. to be more careful in their hiring and promo-
tion procedures, and that procedures should be
made sensitive to the operation of nonconscious
Implication 3: Solutions biases. This can minimize their impact. Krieger
What are the implications for public policy and (1995, p. 1216) also sees a role for policies at
organizational remedies? the workplace level, but focuses more on chang-
A purposive-action framework would argue ing the psychology of decision makers than on
that to the extent that discrimination still is a procedures.
 Petersen—Discrimination • 863

These proposals can be justified without re- major cultural change in most western societies.
course to a theory of nonconscious biases. For The framework for debate and decision making
example, broad-based cultural solutions can is entirely different in year 2000 than in 1960.
change the nature of conscious preferences. They Our conscious processes are hardly recogniz-
may root out animus against specific groups. able. But perhaps our nonconscious categories
This has already occurred. There used to be have changed as well? If they can be influenced
widespread conscious animus. There no longer at all by broader cultural changes, as White and
is. What is left, according to cognitive psycholo- Krieger (1995, p. 510) argue, then there must
gists, is widespread nonconscious bias. The puz- have been significant responses to the very vis-
zle is that the successful campaign to eradicate ible material and cultural transformations that
conscious bias may have failed in also changing have occurred. Krieger (1995, p. 1202, n. 179)
nonconscious cognitive categories. thinks this has yet to happen. I would think that
no change seems unlikely, and major change ap-
pears likely.
Conclusions There is simply reason to think that pro-
Summary longed attention to a particular problem, in this
Legislation and theorizing on gender and race case gender inequality, may have consequences
inequality originally proceeded on the assump- not only for our procedures in organizations, as
tion that employers consciously and rationally they have had, not only for the content of our
discriminated against women and minorities. A conscious thought, as they also have had due to
challenge to this view has emerged from cogni- major cultural changes, but also for the content
tive psychology. The claim is that discrimination of our nonconscious cognitive categories.
today no longer is conscious, but arises rather
from nonconscious schemas that underlie all de- REFERENCES
cision making. This change in emphasis reflects in Arkes, Hal R., and Philip Tetlock. 2004. “Attri-
butions of Implicit Prejudice, or ‘Would Jesse
part the realization that conscious prejudice and
Jackson “Fail” the Implicit Association Test?’”
overt discrimination no longer are central forces
Psychological Inquiry 15(4): 257–278.
behind disadvantage. In part it reflects a broader Becker, Gary S. 1971. The Economics of Discrimina-
movement in the social sciences that draws on tion. Chicago: University of Chicago Press.
cognitive psychology, neuroscience, and biology. Bielby, William T. 2000. “Minimizing Workplace
For differential treatment discrimination that Gender and Racial Bias.” Contemporary Sociology
has its genesis in animus or unfounded stereo- 29(1): 120–129.
types, I think cognitive psychology is on target: Dasgupta, Nilanjana. 2004. “Implicit Ingroup Fa-
It is mostly nonconscious. I just doubt that non- voritism, Outgroup Favoritism, and Their Be-
conscious discrimination is a major source of havioral Manifestations.” Social Justice Research
17(2): 143–168.
female disadvantage in the workplace. Statistical
England, Paula. 1992. Comparable Worth: Theo-
discrimination may still be important, and it is ries and Evidence. Hawthorne, NY: Aldine de
entirely conscious and economically rational, Gruyter.
though illegal. Greenwald, Anthony G., Debbie E. McGhee, and
Jordan L. K. Schwartz. 1998. “Measuring Indi-
vidual Differences in Implicit Cognition: The
Cognition Extended Implicit Association Test.” Journal of Personality
There is no question that our conscious cogni- and Social Psychology 74(6): 1464–1480.
tive structures have changed dramatically during Jolls, Christine, and Cass R. Sunstein. 2006. “The
Law of Implicit Bias.” California Law Review
the last 40 years. The focus on equality issues in
94(4): 969–996.
legislation, in public policy, and in how organi-
Krieger, Linda Hamilton. 1995. “The Content of
zations operate no doubt has brought along a Our Categories: A Cognitive Bias Approach to
864 • Part VIII: Gender Inequality

Discrimination and Equal Employment Oppor- Rimer, Sandra. 2005. “For Women in Sciences, the
tunity.” Stanford Law Review 47(6): 1161–1248. Pace of Progress in Academia Is Slow.” New York
Petersen, Trond. 2006. “Motive and cognition: Con- Times, April 15, 2005, pp. A1, A15.
scious and unconscious processes in employment Valian, Virginia. 1998. Why So Slow? The Advance-
discrimination.” Chap. 14 (pp. 225–248) in Jon ment of Women. Cambridge, MA: MIT Press.
Elster, Olav Gjelsvik, Aanund Hylland, and Karl White, Rebecca Hanner, and Linda Hamilton
Moene (Eds.), Understanding Choice, Explaining Krieger. 2001. “Whose Motive Matters?: Dis-
Behavior. Oslo, Norway: Unipub Forlag/Oslo crimination in Multi-Actor Employment De-
Academic Press. cision Making.” Louisiana Law Review 61(3):
Reskin, Barbara F. 2000. “The Proximate Causes of 495–538.
Employment Discrimination.” Contemporary So-
ciology 29(2): 319–328.
 Bielby—The Structure and Process of Sex Segregation • 865

Sex Segregation

100 • William T. Bielby

The Structure and Process of Sex Segregation

My work seeks to understand the persistence of summarize how my research with James Baron
occupational segregation by sex and the sex gap and Denise Bielby challenges neoclassical expla-
in earnings. Neoclassical economists, of course, nations and then raise some issues concerning
have an elegant, coherent theory of the sources of the promise and limitations of recent interdisci-
both occupational segregation by sex and earn- plinary approaches to the study of job segrega-
ings differentials: they come from the rational, tion by sex and other forms of discrimination.
utility-maximizing behavior of men and women
within households and labor markets.
Sociologists view the situation differently. In- I. Neoclassical Views
deed, depending on whom you ask, a sociologist In a sense, it all starts with the household divi-
might offer any of a number of explanations for sion of labor. Women prefer, get stuck with, or
sex differences in labor-market outcomes. Social have a comparative advantage caring for children
psychologists would stress sex-role socialization. and the home. Because of that, they invest less
Feminist scholars would emphasize the “patriar- in market human capital. For example, intermit-
chal” interests of male employers and workers. tent labor-force participation leads to less work
Marxists would tell a story about capitalists experience and thus to lower productivity and
creating divisions in the work force in order to flatter age-earnings profiles. Women rationally
control labor and boost profits. Organizational choose occupations that are easy to leave and re-
theorists might talk about the unintended con- enter, where skills do not atrophy, and so on.
sequences of bureaucratic rules and procedures. These choices lead to occupational segregation
Not surprisingly, sociologists from different and wage differentials (see Polachek and Kao
camps often talk past one another. [1991]; Polachek [1979]).
However, sociologists are increasingly taking Becker [1985] has offered a clever twist on
the work of neoclassical economists seriously. this explanation. People do not simply allocate
The language of “human capital,” “statistical dis- time between household and market activities.
crimination,” and the like is increasingly part of They also allocate effort. Since women allo-
the sociological discourse on gender and work. cate more effort to household activities, they
However, sociologists do not accept the neoclas- have less effort to allocate to work outside the
sical account uncritically. In this chapter, I first home. So, even when men and women have

William T. Bielby. “The Structure and Process of Sex Segregation,” in New Approaches to Economic and Social
Analyses of Discrimination, edited by Richard R. Cornwall and Phanindra V. Wunnava, pp. 97–112. Copyright
© 1991 by Praeger Publishers/Greenwood Publishing Group, Inc. Reproduced with permission of Greenwood
Publishing Group, Inc. Permission conveyed through Copyright Clearance Center, Inc.
866 • Part VIII: Gender Inequality

the same amount of work experience, training, allocation of work effort (Bielby and Bielby
etc., women are less productive. An additional [1988]). Here I summarize the findings rele-
hour of women’s work is less effort-intensive vant to the issue of job segregation and earnings
than one of men’s work. As a result, women disparities.
are worth less to employers and accumulate less Contrary to the clear implication from Beck-
market human capital with each additional hour er’s model, women do not exert less effort than
of work. This leads to an explanation for both men in the workplace in order to conserve ef-
occupational segregation by sex and earnings fort for household activities. Compared to men
differentials within occupations. Women choose with similar household responsibilities, market
less effort-intensive jobs, and they are less pro- human capital, earnings, promotion opportu-
ductive than men working on the same jobs. nities, and job responsibilities, women allocate
There is, of course, another view of occupa- substantially more effort to work activities; the
tional segregation that admits to a certain kind net sex difference is nearly one-half a standard
of “discrimination,” but only a rational kind of deviation on the scale we use. To the extent that
“statistical discrimination.” In a world filled with women do allocate effort away from the work-
uncertainty, employers cannot always know the place in order to meet family demands, these
relevant traits of job applicants. Suppose that trade-offs bring their work effort back to the
there are important traits that are difficult to level of the typical male with no such family
measure, but we know that, on average, women responsibilities.
exhibit more (or less) of them than men. One How could this be? We believe that the neo-
example often cited is the likelihood of job classical model of the allocation of work effort
turnover. Suppose also that there are two kinds is seriously flawed in several ways. First, it may
of jobs—those for which it is costly to replace be inappropriate to assume that the “stock” of
a worker who quits and those for which it is effort to be allocated by an individual is fixed,
not. Not knowing actual quit propensities, ra- analogous to a fixed stock of physical capital.
tional, profit-maximizing employers will reserve For women to work just as hard as men, if not
the former jobs for men and fill the latter with harder, despite their greater household responsi-
women, even if there is considerable overlap in bilities, women must be able to draw on a reserve
the underlying distributions of quit propensities of energy that is either not available to the typ-
of men and women (Phelps [1972]; Aigner and ical male, or more realistically, that men choose
Cain [1977]). not to draw upon. Stephen Marks, a sociologist
My research in collaboration with Denise who has written on this topic [1977: 927], cites
Bielby (Bielby and Bielby [1988]) and James physiological evidence in support of his claim
Baron (Bielby and Baron [1986]) suggests that that “individuals have abundant and perpetually
despite the elegance of the neoclassical accounts, renewing [energy] resources.” He suggests that
the world just does not work that way. Men and individuals generate the energy needed to partic-
women employees do not act in a way that is ipate in activities to which they are committed
consistent with neoclassical accounts, nor do and that they often feel more “energetic” after
employers. engaging in them. Our findings imply that as
women add work roles to their family roles, they
generate the energy necessary to fulfill their com-
II. Empirical Challenges to mitments to the two sets of activities.
Neoclassical Views: The Allocation A second factor ignored by human capital
of Effort models is what social psychologists call “entitle-
Denise Bielby and I, using the 1973 and 1977 ment norms.” The allocation to effort at work is
Quality of Employment Surveys, attempted to influenced by internalized norms of a “fair day’s
operationalize a human-capital model of the work,” which may differ by sex. Determinations
 Bielby—The Structure and Process of Sex Segregation • 867

of fairness are based on perceptions of contri- and the denominator of the “equity ratio” of
butions relative to rewards. Experiments show reward to effort for men’s work. According to
that men and women differ in how they invoke social psychologists, it is precisely when external
equity considerations in allocating effort and bases of social comparison are unavailable that
rewards. On average, women pay themselves individuals rely on internal, same-sex norms
less than men performing the same task, and (Berger, Zelditch, Anderson, and Cohen [1972];
women tend to undervalue their efforts relative Berger, Rosenholtz, and Zelditch [1980]; Austin
to men (Lenney [1977]; Callahan-Levy and [1977]). Major, McFarlin, and Gagnon found
Messe [1979]; Major, McFarlin, and Gagnon that when external comparison information was
[1984]). It appears that, on average, women available, neither sex relied on same-sex norms
have lower internal standards of “personal enti- to make judgments of fair pay for a given task.
tlement,” and, in the absence of salient external Thus equity research suggests why employers
comparison standards, they make fairness judg- might have a stake in a sex-segregated workforce:
ments based on application of same-sex norms as long as women lack information about the re-
about appropriate rewards (Berger et al. [1972]; ward structure for men, they will be willing to
Crosby [1982]). work for less pay. (Moreover, husbands of work-
In a laboratory experiment designed spe- ing women benefit by avoiding responsibility for
cifically to examine sex differences in effort household activities, despite the fact that their
and standards of personal entitlement, Major, wives expend as much or more energy as they
McFarlin, and Gagnon [1984] asked men and do at work.)
women to do as much work as they thought was Fourth, another aspect of our analysis calls
fair for a fixed amount of money. They collected into question the extent to which decisions
objective measures of the accuracy and efficiency about work and family are made in a deliber-
of performance on the task as well as informa- ately self-conscious, rational manner. Bielby
tion on each subject’s perceptions of his or her and Bielby [1988] found that men’s work effort
level of performance. They discovered that, is greatly influenced by family demands, while
on average, “women worked longer, did more the effects of “human-capital” investments such
work, completed more correct work, and were as schooling and labor-market experience are
more efficient than men” (1409). At the same relatively weak. In contrast, compared to men,
time, men and women did not differ in their women’s work effort is less strongly influenced
self-evaluations of their performance, despite by family demands, but is more strongly affected
women’s superior objective performance. Nor by investments in market human capital. For
did men and women differ in their reports of example, a man with child-care responsibilities
satisfaction from the task. In short, laboratory reduces his work effort, but a woman in the same
research on sex differences in internal standards situation does not. At the same time, a man with
of personal entitlement suggests that, all else meager investments in “market human capi-
constant, women can be expected to allocate tal”—education and labor-market experience—
more effort than men to work activities. does not decrease his work effort, but a woman
Third, structural features of workplaces with comparable investments does decrease
might reinforce sex differences in entitlement her work effort. This suggests that individuals
norms. For example, research on sex segrega- take sex-specific culturally prescribed roles for
tion in the workplace suggests why same-sex granted. Women do not adjust their effort allo-
norms about rewards for performance persist. cations in response to household responsibilities,
Most men and women work in sex-segregated and men do not adjust their effort allocations in
jobs (Bielby and Baron [1984], [1986]; Reskin response to workplace investments.
and Hartmann [1986]). Consequently, many It appears that individuals do rationally re-
women lack information on both the numerator flect upon their efforts when they cannot fall
868 • Part VIII: Gender Inequality

back upon culturally prescribed roles. Men 2, the opposite is true. In firm 3, however, the
confronted with household responsibilities re- first job is staffed exclusively by men, and the
duce their work efforts, as do women with mea- second and third jobs exclusively by women. We
ger investments in market human capital. For can imagine other patterns besides the one de-
men, some level of work effort may be culturally picted in Figure 100.1 for an occupation that is
prescribed and “taken for granted” regardless 40 percent female. For example, in a world with-
of investments in education and training. For out statistical discrimination, we might find the
women, a certain level of work effort may be sex composition of specific jobs normally dis-
culturally prescribed and “taken for granted” re- tributed around 40 percent female. In any case,
gardless of the level of household responsibilities. according to neoclassical models of the labor
Moreover, the level of effort allocated to work by market, knowledge of differences between firms
a woman with heavy household responsibilities and jobs in work requirements and differences
is at least as large as that allocated by men with between the sexes in job traits should explain
no such responsibilities at home. how men and women in mixed occupations get
In sum, Becker’s human-capital model of distributed across firms and jobs.
the allocation of work effort provides a plau- Baron and I assembled data like this for sev-
sible explanation for both sex segregation and eral hundred California firms studied by the
wage differentials by sex. However, our research U.S. Employment Service in the late 1960s and
and other social psychological studies call into 1970s. That is, we looked at how firms hiring
question some basic premises of this model. The from a sex-mixed labor pool staffed their jobs as
assumption of a fixed stock of effort may be un- a function of firm and job characteristics. Neo-
realistic, the social psychological processes may classical models suggested the following hypoth-
be more complex than what is typically assumed eses: scale, bureaucratization, and job-specific
about “economic man,” and segregation of men’s training are related to turnover costs, so if
and women’s jobs may reinforce sex differences women are more likely to quit, they should be
in social psychological processes. underrepresented in firms and jobs where these
costs are high. If employers statistically discrim-
inate based on turnover costs, women should be
III. Empirical Challenges to excluded altogether from jobs where these costs
Neoclassical Views: Noneconomic are high. The remaining job traits we looked at
Bases of Statistical Discrimination are job-relevant skills. Women have been shown
James Baron and I set out to analyze how men to perform better, on average, on traits such as
and women in the same general line of work are verbal skills, motor coordination, finger dexter-
sorted into different firms and into different jobs ity, interpersonal skills, and stamina under re-
within firms (Bielby and Baron [1984], [1986], petitive, monotonous working conditions. Men
[1987]; Baron and Bielby [1985]). Figure 100.1 have been shown to perform better, on average,
shows the issue schematically. Consider an oc- on numerical skills, spatial skills, and upper
cupation like “electronics assembler” that is 40 body strength. Thus the distribution of men and
percent female. Suppose that three firms in an women across jobs should reflect differences in
industry employ electronics assemblers and that jobs in the extent to which they require these
each of these firms uses assemblers in three of its sex-linked traits.
job titles. In the hypothetical example, firm 1 Our most startling finding was that Cali-
relies disproportionately on male workers, firm fornia firms in the late 1960s and 1970s rarely
2 relies disproportionately on female workers, employed men and women in the same job ti-
and firm 3 relies on a balanced mix of male and tles. Even when drawing from a sex-mixed labor
female workers. In firm 1, each of the jobs is pool for a work role like “electronics assembler,”
staffed disproportionately by men, and in firm firms reserved some job titles exclusively for men
 Bielby—The Structure and Process of Sex Segregation • 869

Figure 100.1. Hypothetical Distribution of Workers in a Mixed Occupation

Across Firms and Jobs

Job 1 - (90/10)
Firm 1 Job 2 - (70/30)
(80/20) Job 3 - (80/20)

Mixed Job 1 - (10/90)

Occupation Firm 2 Job 2 - (30/70)
(60/40*) (20/80) Job 3 - (20/80)

Job 1 - (100/0)
Firm 3 Job 2 - (0/100)
(50/50) Job 3 - (0/100)

*Ratio of Male Workers to Female Workers

and others exclusively for women. That is, firms are assigned jobs defined as “detail” work, and
either employed male workers exclusively or fe- men are not.
male workers exclusively, or, if they employed According to neoclassical economists, em-
both, men and women were completely segre- ployers are sacrificing profits in order to sustain
gated by job title, as in firm 3 in Figure 100.1. the type of sex segregation we observed. Our
In addition, we found that the relationships measure of physical demands was whether a job
between firm and job traits and the sex com- required lifting more than twenty-five pounds.
position of work were exactly as hypothesized. In most of the firms we studied, no woman
“Male” traits were positively related to the like- would be assigned to these jobs, regardless of
lihood of excluding women from a job and the that woman’s own physical strength. Yet it is
effects of “female” traits were in the opposite simple and inexpensive to devise a test to learn
direction. Our research indicates that employ- whether a female (or male) job applicant is ca-
ers exclude women from some jobs and men pable of lifting twenty-five pounds. Similarly, it
from others (in the same general line of work) should be easy to devise a simple test that would
based on their perceptions of group differences allow an employer to observe whether a male (or
between men and women in job-related traits. female) job applicant could successfully com-
That is, employers statistically discriminate. plete a task requiring finger dexterity. Given that
However, our findings did not support the they were drawing from sex-mixed labor pools,
neoclassical version of statistical discrimination. why did employers persist in sustaining almost
Statistical discrimination is efficient only when complete job segregation by sex? We had access
it is costly to measure the actual job-relevant to narrative material about the employment
traits of individual workers. The factors that practices in these California firms, and analyses
made the greatest difference in whether men of these qualitative materials suggested several
or women were employed were not the subtle important factors.
and difficult-to-measure factors like turnover
costs. They were physical demands and finger Policies. In many of the firms studied by the
dexterity. Men are assigned the jobs perceived to Employment Service, it was a matter of com-
involve heavy work and women—in the same oc- pany policy not to employ women in certain
cupation—are assigned the “light” work. Women jobs and to favor employing women in others.
870 • Part VIII: Gender Inequality

Jobs closed to women were typically perceived Cultural Stereotypes. Cognitive psychologists
to be physically demanding and those favoring have demonstrated that we all use stereotypes,
women usually involved “detail” work. Curi- or cognitive schema, as a cognitive shorthand
ously, while we often came across policies spec- we invoke to achieve economy in perception
ifying that women were preferred for certain (Ashmore and Del Boca [1981]). Of course, the
jobs, we rarely encountered official prohibitions particular stereotypes we hold about sex roles are
against men being employed in these jobs. Yet in widely shared and continually reinforced in our
practice, men rarely were employed in predom- culture. Despite what is assumed by neoclassical
inantly female jobs unless they were supervisors models of discrimination, the profit motive is
and in that case, they would almost always have not necessarily strong enough to overcome the
distinct job titles. cognitive processes that lead to the use of ste-
There are at least two reasons for the lack reotypes. Thus it may be that employers make
of any explicit prohibition against using men hiring decisions based on widely shared misper-
in specific jobs, despite the nearly perfect seg- ceptions of differences in the average traits of men
regation in practice. First, if female labor is less and women. To complicate matters further, sta-
expensive, employers have no incentive to em- tistical discrimination can lead to “self-fulfilling
ploy men in the same job categories as women, prophecies,” or what social psychologists call “ex-
even if the company has no policy against doing pectancy confirmation sequences” (Darley and
so. Second, prohibitions against women work- Fazio [1980]). That is, employers expect certain
ing in certain jobs often originated in protective behaviors from women and assign them to rou-
legislation (even if these regulations no longer tine tasks and dead-end jobs. Women respond
applied), and thus complying with government by exhibiting the very behaviors employers ex-
regulations was often the original rationale for pect, reinforcing the stereotype. Moreover, social
having a policy on women’s work. No similar psychological research suggests that employers
regulations mandated policies on men’s work. are more likely to attend to and retain informa-
tion that confirms stereotypes, ignoring behavior
Organizational Inertia. Once a sex-based divi- by women that does not fit their expectations.
sion of labor is established, it becomes “taken for Given the kind of data available to Baron and
granted,” sustained over the years unless some me, we had no direct evidence of cultural ste-
deliberate effort is taken to undo it. This was reotypes held by specific employers or managers
especially evident in the exclusion of women about men’s work and women’s work. However,
from jobs perceived to be physically demand- such stereotypes were invoked quite often in
ing. California laws prohibited the employment policies regarding the employment of women
of women in jobs requiring heavy lifting until in “heavy” work and in “detail” work. Indeed,
these laws were struck down in 1971. Yet our it was not uncommon to encounter a policy
data showed that the exclusion of women from prohibiting employment of women in certain
so-called physically demanding jobs was just as jobs because of heavy lifting requirements and
widespread through the late 1970s—when the to then find in detailed job analyses by the Em-
practice was illegal—as it was during the period ployment Service that these jobs required no
before 1971. We also saw this inertia in data heavy physical labor. In other words, the work is
from firms analyzed by the Employment Ser- often labelled as “heavy work” because it is done
vice at two points in time, typically six or seven by men.
years apart. The sex-based division of labor rarely
changed over that six- or seven-year period, even Interests. Finally, the extreme sex segregation
when total employment fluctuated substantially, we discovered within mixed occupations could
unless there was some deliberate effort taken to not have been sustained without someone’s
undo sex segregation. interests being served. It is not difficult to see
 Bielby—The Structure and Process of Sex Segregation • 871

how employers might have an interest in sex drawn on both institutional approaches to the
segregation. For example, our research on the structure of work and social psychological theo-
allocation of effort suggests that segregation ries of entitlement norms. Of course, these areas
sustains sex-specific norms of entitlement. Male are no longer beyond the purview of neoclassical
employees who view integration as a threat to economics. For example, Williamson ([1975],
their wages may resist the assignment of women [1985]) explains variation in organizational
to the job titles they hold. However, not at all structures with the concept of economizing on
clear from existing research are the circumstances transaction costs, and “efficiency wage” theorists
under which male workers have the capacity to invoke entitlement norms to explain interindus-
act on these interests, given employers’ economic try wage differentials (Krueger and Summers
incentive to substitute cheap female labor for ex- [1987]; Akerlof and Yellen [1988]).
pensive male labor. Although my research with It is customary to applaud cross-disciplinary
Baron on statistical discrimination did not get at efforts, and the blurring of the boundaries be-
the interests issue directly, other work we have tween sociology and neoclassical economics may
done suggests a slight tendency for greater seg- eventually lead to a comprehensive explanation
regation in settings where male workers have of job segregation by sex that is consistent with
substantial market power (Bielby and Baron the empirical evidence on the subject. By incor-
[1984]). porating “sociological factors” (e.g., institutional
barriers, entitlement norms), neoclassical mod-
els may be able to more faithfully represent the
IV. Conclusion: Reconciling factors that slow the rate at which market forces
Alternative Perspectives on the erode discrimination in pay and job assignment.
Structure and Process of Sex Moving in that direction requires continued
Segregation commitment to the neoclassical assumption that
We can identify six more or less distinct kinds were it not for various “drags” on the system,
of explanations for a single phenomenon: sex market forces would inevitably discount ascrip-
segregation in the workplace. These explana- tive factors like race and gender in valuing oth-
tions are (1) neoclassical, based on models of erwise comparable labor inputs.
statistical discrimination and investments in Adapting the neoclassical model in this way
human capital; (2) social psychological, em- has appeal due to the power of the model’s for-
phasizing socialization and internalized norms; mal analytical properties. However, the perva-
(3) institutional, emphasizing the intended and siveness and persistence of job segregation by sex
unintended consequences and inertia of orga- seems too far at odds with the imagery of even
nizational arrangements; (4) cultural, empha- the most extreme adaptations of the model (for
sizing taken-for-granted notions of men’s work a similar view, see Darity [1991]). For example,
and women’s work, often shared by both men it is difficult to imagine what kinds of “trans-
and women; (5) political, stressing the different action costs” could explain the nearly universal
interests of male employers and employees with job segregation of men and women within the
respect to maintaining the status quo; and (6) same occupation that Baron and I found across a
patriarchal, emphasizing the common interests wide variety of organizational settings. That such
of male workers and employers in maintaining a segregation would persist is especially problem-
sex-based division of labor. atic from a neoclassical perspective, since the
The research my colleagues and I have pur- rationale was typically based on stereotypical
sued has sought to test neoclassical explanations perceptions of men’s and women’s job aptitudes,
of gender differences in the workplace, not to often with little basis in the actual duties men
offer fully developed alternatives. In posing and women performed on the job. Indeed, on
tentative accounting for our findings, we have more than one occasion a labor economist has
872 • Part VIII: Gender Inequality

suggested to me that the kind of uniform sex The first concerns the circumstances under
segregation within occupations that Baron and which women were demobilized from factory
I detected simply cannot exist in the real world! jobs in the automobile industry after World War
Radical and Marxist economists who study II. In the most definitive study to date on the
discrimination have always been more open than topic, Milkman [1987] documented that the
their more orthodox colleagues to incorporating women who held these jobs performed at least
concepts and findings of sociologists. Further- as capably as the men they replaced. She also
more, the kind of persistent job segregation by showed that many of the women mobilized by
sex detected in my research with Baron is fully the automobile industry during the war wanted
consistent with their models. Radical and Marx- to keep their jobs. Moreover, at the war’s end,
ist models are built upon an imagery of con- the automobile manufacturers could no longer
flict and struggle over control of the workplace rely on “cost-plus” government contracts and
among contending groups defined by race, class, therefore should have been more sensitive to
and gender. In attempting to account for the labor costs after the war than during the war.
persistence of the racial gap in unemployment, Furthermore, Milkman showed that the unions
Shulman [1991] moves this line of theorizing that represented the interests of white male
forward by addressing the economic and social workers had virtually no power to influence
costs of dismantling discrimination. According management policy regarding hiring preferences
to this perspective, job segregation and other and job assignments. In short, management ap-
forms of discrimination are likely to persist in peared to have both the incentive and the ability
the absence of any exogenous shock or direct to rely disproportionately on women during the
intervention that changes the social relations tremendous expansion in employment during
among contending groups. the years following the war.
However, while radical and Marxist ap- Such behavior by employers would be con-
proaches appear more consistent with the find- sistent with both neoclassical and neo-Marxist
ings of our research, I am not yet prepared to approaches to the labor market. Instead, Milk-
accept their models. My reservation is based on man showed that management acted on a strong
the absence of gender as a cultural or ideological preference for young, male veterans and went to
phenomenon in their explanations of segrega- great lengths to purge women from all but the
tion and other forms of discrimination. Instead, traditionally female job classifications. Perhaps
their models are driven by the logic of systemic management feared that female substitution in
properties like concentration and monopoliza- the automobile factories would destabilize labor
tion and the self-interested, rational action of so- relations and were taking into account one of
cial groups (e.g., employers, white male workers) the costs of dismantling discrimination empha-
intent on preserving their privileged positions sized in Shulman [1991]. However, there is little
in the division of labor (Darity and Williams evidence that the very decisive actions taken by
[1985]). Based on my research and my reading management to demobilize women were based
of the work of other sociologists, I have come on an informed view of the costs of possible resis-
to believe that gender is more than a dimension tance by male workers. Instead, for reasons that
along which the labor force is segmented and are difficult to explain from any perspective em-
that gender ideologies are a strong, semiautono- phasizing economic self-interest (or class inter-
mous force shaping segregation and other mani- est), employers concluded that women ought not
festations of socioeconomic inequality. I have yet to hold these jobs once men returned from war.
to see this notion of gender and ideology incor- My second example comes from research I
porated into formal models of discrimination, have done in collaboration with Denise Bielby
but I can illustrate the empirical reality of the (Bielby and Bielby [1991]) empirically testing
issue with two examples. Jacob Mincer’s [1978] neoclassical model of
 Bielby—The Structure and Process of Sex Segregation • 873

Figure 100.2. Reluctance to Relocate Due to Family Considerations, by Sex, Spouse’s Earnings, and
Gender-Role Beliefs

1.0

Traditional females
Probability of Reporting Reluctance to Relocate

.8

Nontraditional females

.6

.4

Nontraditional males

.2

Traditional Males

0
0 20 40 60
Spouse's Earnings, $1000s

family migration decisions among dual-earner contingent upon both gender and gender-role
couples. According to the model, couples maxi- beliefs. As Figure 100.2 illustrates, traditional
mize family well-being. When a husband or wife males—those who believed in the primacy of the
is confronted with a job advancement requiring husband’s role as provider and who disapproved
geographic relocation, the couple migrates when of working mothers—were not influenced at all
that person’s gain is greater than the cost of dis- by their wives’ earnings. In contrast, traditional
rupting his or her spouse’s career. Otherwise, females were extremely sensitive to their hus-
there is no net gain to the family, and family bands’ earnings. In effect, such women expressed
well-being is maximized by forgoing the job op- reluctance to relocate for job advancement un-
portunity at the new location. less their husbands earned very little. Holding
Presumably, the risk to joint family earnings nontraditional gender-role beliefs had opposite
due to a relocation increases with the level of effects on men and women. Compared to tradi-
earnings of the spouse whose career will be dis- tional women, nontraditional women were more
rupted. If so, then the neoclassical model predicts inclined to pursue job advancement at a new
that reluctance to relocate for a new job should location, while the opposite was true for non-
increase directly with the level of the spouse’s traditional men compared to traditional men.
earnings. Accordingly, it was not surprising In other words, our results showed that among
when data from the 1977 Quality of Employ- those holding traditional beliefs, men were likely
ment Survey showed spouse’s earnings to be the to put their private career interests ahead of the
strongest predictor of a reluctance to relocate for interest of overall family economic well-being
a much better job in a new location. What was (contrary to the assumption of the neoclassical
surprising (from the neoclassical perspective) was model), while women behaved altruistically to-
that the net effect of spouse’s earnings was highly ward the family, maximizing family well-being
874 • Part VIII: Gender Inequality

instead of personal well-being (consistent with REFERENCES

Mincer’s assumption). However, the relationship Aigner, Dennis J. and Glen C. Cain. 1977. “Statis-
was sharply mediated by gender-role beliefs, and tical Theories of Discrimination in Labor Mar-
kets.” Industrial and Labor Relations Review 30, 2
the sex differences in orientation toward self
(January): 175–187.
and family were substantially attenuated among
Akerlof, George A. and Janet L. Yellen. 1988. “Fair-
those holding nontraditional beliefs. ness and Unemployment.” American Economic
Thus our research suggests that traditional Review 78, 2 (May): 44–49.
gender-role ideologies sustain women’s disadvan- Ashmore, Richard D. and Frances K. Del Boca.
tages in both workplace and family dynamics. 1981. “Conceptual Approaches to Stereotypes
Traditional notions of women’s role provide a and Stereotyping.” Pp. 1–35 in D. L. Hamilton
rationale for job segregation and statistical dis- (ed.), Cognitive Procession Stereotyping and Inter-
crimination in the workplace, and those same group Behavior. Hillsdale, N.J.: Erlbaum.
beliefs make it more difficult for women than Austin, William. 1977. “Equity Theory and Social
Comparison Processes.” Pp. 279–305 in J. Suls
men to pursue job opportunities (which in
and R. Miller (eds.), Social Comparison Theory:
turn can reinforce stereotypes and provide a ra- Theoretical and Empirical Perspectives. Washing-
tionale for statistical discrimination). In short, ton, D.C.: Hemisphere Publishing.
gender-role ideologies appear to directly shape Baron, James N. and William T. Bielby. 1985. “Or-
the extent to which husbands and wives define ganizational Barriers to Gender Equality: Sex
and act upon their interests. Segregation of Jobs and Opportunities.” Pp.
In sum, in an important sense, the economic 233–251 in A. S. Rossi (ed.), Gender and the Life
theories of segregation and discrimination— Course. New York: Aldine.
neoclassical supply-side, demand-side, radical, Berger, Joseph, Susan J. Rosenholtz, and Morris
Zelditch. 1980. “Status Organizing Processes.”
and neo-Marxist—are neutral with respect to
Annual Review of Sociology 6: 479–508.
gender (and race). In these theories, whichever
Berger, Joseph, Morris Zelditch, Jr., Bo Anderson,
group happens to have fewer resources or less and Bernard P. Cohen. 1972. “Structural Aspects
predictable traits or arrived last when either cap- of Distributive Justice: A Status Value Formula-
ital or “good jobs” were being allocated happens tion.” Pp. 119–146 in J. Berger, M. Zelditch, and
to end up disadvantaged. The ideological con- B. Anderson (eds.), Sociological Theories in Prog-
tent of cultural idioms of gender and race have ress, vol. 2. Boston: Houghton Mifflin.
no causal impact in any of these theories. Curi- Bielby, William T. and James N. Baron. 1984. “A
ously, this is true even when these theories are Woman’s Place Is with Other Women: Sex Seg-
regation Within Organizations.” Pp. 27–55 in B.
borrowed by sociologists, who, in recent years,
F. Reskin (ed.), Sex Segregation in the Workplace:
have been captivated by theories that purport to
Trends, Explanations, Remedies. Washington,
explain social behavior as the outcome of “ratio- D.C.: National Academy Press.
nal action” (Hechter [1983]; Coleman [1990]). ———. 1986. “Men and Women at Work: Sex Seg-
Yet the examples I have cited suggest that culture regation and Statistical Discrimination.” Ameri-
and ideology might have considerable explana- can Journal of Sociology 91, 4 (January): 759–799.
tory power in more fully accounting for both ———. 1987. “Undoing Discrimination: Compa-
persistence and change in job segregation and rable Worth and Job Integration.” In Christine
other forms of discrimination. At least at a de- Bose and Glenna Spitze (eds.), Ingredients for
Women’s Employment Policy. Albany: State Uni-
scriptive level, sociocultural factors are impossi-
versity of New York Press.
ble to ignore in any close empirical investigation
Bielby, William T. and Denise D. Bielby. 1988. “She
of these phenomena. However, it is quite clear Works Hard for the Money: Household Respon-
that we have a long way to go before such factors sibilities and the Allocation of Work Effort.”
are fully incorporated into our theoretical and American Journal of Sociology 93, 5 (March):
analytical models of discrimination. 1031–1059.
 Bielby—The Structure and Process of Sex Segregation • 875

———. 1991. “I Will Follow Him: Family Ties, the Nature of Gender Differences in Personal
Gender-Role Beliefs, and Reluctance to Relocate Entitlement.” Journal of Personality and Social
for a Better Job.” Forthcoming, American Journal Psychology 47, 6 (December): 1399–1412.
of Sociology 94. Marks, Stephen R. 1977. “Multiple Roles and Role
Becker, Gary S. 1985. “Human Capital, Effort, and Strain: Some Notes on Human Energy, Time,
the Sexual Division of Labor.” Journal of Labor and Commitment.” American Sociological Review
Economics 3, 1, pt. 2 (January): 33–58. 42, 6.
Callahan-Levy, Charlene M. and Lawrence A. Messe. Milkman, Ruth. 1987. Gender at Work: The Dynam-
1979. “Sex Differences in the Allocation of Pay.” ics of Job Segregation by Sex During World War II.
Journal of Personality and Social Psychology 37, 3 Urbana: University of Illinois Press.
(March): 433–446. Mincer, Jacob. 1978. “Family Migration Decisions.”
Coleman, James S. 1990. Foundations of Social The- Journal of Political Economy 86, no. 5 (October):
ory. Cambridge, Mass.: Harvard University Press. 749–773.
Crosby, F. 1982. Relative Deprivation and Working Phelps, E. S. 1972. “The Statistical Theory of Rac-
Women. New York: Oxford University Press. ism and Sexism.” American Economic Review 62,
Darity, William A., Jr. 1991. “Efficiency Wage The- 4 (September): 659–661.
ory: Critical Reflections on the Neokeynesian Polachek, Solomon W. 1979. “Occupational Segre-
Theory of Unemployment and Discrimination.” gation Among Women: Theory, Evidence, and
Pp. 39–53 in R.R. Cornwall and Phanindra V. a Prognosis.” Pp. 137–157 in C. B. Lloyd, E. S.
Wunnava (eds.), New Approaches to Economic and Andrews, and C. L. Gilroy (eds.), Women in the
Social Analyses of Discrimination. New York: Prae- Labor Market. New York: Columbia University
ger Publishers. Press.
Darity, William A. and Rhonda A. Williams. 1985. Polachek, Solomon W. and Charng Kao. 1991.
“Peddlars Forever? Culture, Competition, and “Lifetime Work Expectations and Estimates of
Discrimination.” American Economic Review: Pa- Gender Discrimination.” Pp. 199–238 in R.R.
pers and Proceedings 75, 2 (May): 256–261. Cornwall and P.V. Wunnava (eds.), New Ap-
Darley, J. M. and R. H. Fazio. 1980. “Expectancy proaches to Economic and Social Analyses of Dis-
Confirmation Sequences.” American Psychologist crimination. New York: Praeger Publishers.
35, 10 (October): 867–881. Reskin, Barbara F. and Heidi I. Hartmann (eds.)
Hechter, Michael. 1983. The Microfoundations of 1986. Women’s Work, Men’s Work: Sex Segregation
Macrosociology. Philadelphia: Temple University on the Job. Washington, D.C.: National Academy
Press. Press.
Krueger, Alan B. and Lawrence H. Summers. 1987. Shulman, Steven. 1991. “Why Is the Black Un-
“Reflections on the Inter-Industry Wage Struc- employment Rate Always Twice as High as the
ture.” Ch. 2, pp. 17–47, in K. Lang and J. S. White Unemployment Rate?” Pp. 5–37 in R.R.
Leonard (eds.), Unemployment and the Structure Cornwall and P.V. Wunnava (eds.), New Ap-
of Labor Markets. New York: Basil Blackwell. proaches to Economic and Social Analyses of Dis-
Lenney, Ellen. 1977. “Women’s Self-Confidence in crimination. New York: Praeger Publishers.
Achievement Settings.” Psychological Bulletin 84, Williamson, Oliver E. 1975. Markets and Hierar-
1 (January): 1–13. chies. New York: Free Press.
Major, Brenda, Dean B. McFarlin, and Diana Gag- ———. 1985. The Economic Institutions of Capital-
non. 1984. “Overworked and Underpaid: On ism. New York: Free Press.
876 • Part VIII: Gender Inequality

101 • Jerry A. Jacobs

Revolving Doors
Sex Segregation and Women’s Careers

This chapter is an effort to understand why heightened interest in working women. Legisla-
women do “women’s work.” In 1985, over tive policies such as affirmative action, as well as
two-thirds of the women in the U.S. civilian the entry of feminist scholars into academia, have
labor force worked in occupations that were 70 generated research on the disadvantaged position
percent or more female.1 The study of women’s of working women. Most recently, calls for equal
work has attracted scholars from many disci- pay for work of comparable value have brought
plines, from psychology (Gutek, 1985) to history public and research attention to the concentra-
(Kessler-Harris, 1982) to anthropology (Sanday, tion of women in low-paid, female-dominated
1981) to economics (Bergmann, 1986) to sociol- occupations (Remick, 1984).
ogy (Reskin, 1984). A flurry of recent papers has This chapter investigates the way women’s ca-
addressed such topics as trends in occupational reers intersect with the sexual division of labor in
segregation (Bianchi and Rytina, 1986; Jacobs, the workplace. We follow women’s (and men’s)
1986), sex segregation among teenagers in the life histories in order to understand the gender
workplace (Greenberger and Steinberg, 1983), tracking system. We ask whether there is a deci-
and sex segregation within voluntary organi- sive moment during women’s lives when they are
zations (McPherson and Smith-Lovin, 1986), channeled into female-dominated occupations,
between industries (Tienda, Smith, and Ortiz, or whether the process is more gradual. Is sex
1987), across cities (Abrahamson and Sigelman, segregation principally due to a single life-course
1987), and within firms (Bielby and Baron, event, such as meeting with a guidance coun-
1986). (See Reskin and Hartmann, 1985, for selor in high school or becoming a parent, or is
discussion of sex segregation in the workplace.) it the result of a cumulative process extending
This interest has undoubtedly been spurred throughout much of the women’s (and men’s)
by the growth in women’s labor force participa- lives?
tion (Goldin, 1983; Smith and Ward, 1984). As the title of this chapter suggests, we
Over 70 percent of women between the ages of find surprising rates of mobility for women
20 and 44 worked for pay in 1985, including among male-dominated, sex-neutral, and
nearly half of those with children under one year female-dominated occupations (see Jacobs,
old. The overall rate of women’s labor force par- 1989). This pattern of change is also evident
ticipation has grown from 33.9 percent in 1950 when the dynamics of sex segregation of career
to 54.5 percent in 1985 (U.S. Bureau of the aspirations and college majors are analyzed. In-
Census, 1986). The women’s movement has also dividual mobility is common, yet change in the

Jerry A. Jacobs. Revolving Doors: Sex Segregation and Women’s Careers, pp. 1–4, 8–10, 50–53, 190. Copyright
© 1989 by the Board of Trustees of the Leland Stanford Jr. University. All rights reserved. Used with the per-
mission of the publishers, Stanford University Press, www.sup.org.
 Jacobs—Revolving Doors • 877

structure of sex segregation is quite slow. This the flows in both directions. If there were less
paradox is the central puzzle of this chapter. attrition of women from male-dominated occu-
The mobility described here and elsewhere pations, there would be more progress toward in-
(Jacobs, 1989) does not imply the existence of tegrating men and women in the workplace. . . .
equal opportunity for women. Quite the con-
trary. The rates of mobility we document are
roughly proportionate to the distribution of Sex Segregation and Social Control
women across occupations; they do not imply To maintain any sexual division of labor, a soci-
that women have the same chances as men. For ety has to develop mechanisms for slotting men
example, the probability of a woman starting into male-dominated positions and women into
her career in a male-dominated occupation is female-dominated positions. Once positions
approximately one in three for recent cohorts or tasks are defined as masculine or feminine,
of women entering the labor force. Our data societies must enforce these definitions. Sociol-
indicate that, for a woman who changes her ogists refer to mechanisms by which societies
occupation, the probability of moving into a bring about these results as mechanisms of social
male-dominated occupation is close to one in control.
three, no matter whether her previous occupa- The earliest stage of social control is social-
tion was in a female-dominated, sex-neutral, ization, during which society imbues the young
or male-dominated field (Jacobs, 1989, chap- with its prevailing values and beliefs. Central to
ter 7). This pattern suggests that women in our interest is vocational socialization, the de-
female-dominated occupations who are able velopment of gender identity that bears on the
to change occupations stand a good chance of pursuit of careers. While the reproduction of
moving into a male-dominated occupation. Yet social values may seem a simple matter in pre-
by the same token, women in male-dominated industrial societies, socialization is problematic
fields who change occupations also stand a in our own society. We argue that social control
good chance of moving to a sex-neutral or continues well beyond early socialization, involv-
female-dominated occupation. Indeed, one of ing the educational system, the decisions and be-
the more disturbing findings documented here havior of employees and employers in the labor
(also, see Jacobs, 1989, chapter 7) is a startling market, and the influences of family and friends
rate of attrition of women in male-dominated throughout life. We raise questions about the
occupations, even as these occupations appear to efficacy of various gender-tracking mechanisms
be opening their doors to them. that are often assumed to be responsible for the
The revolving-door metaphor does not imply maintenance of sex segregation.
that women are getting nowhere. Rather, it sug- The central thesis of this chapter is that the
gests that gross mobility far exceeds the overall maintenance of sex segregation depends on a
net change in opportunities for women. In recent lifelong system of social control. While sex-role
years, for every 100 women in male-dominated socialization is important, since it instills values
occupations who were employed in two consecu- and goals, it is inadequate by itself to maintain the
tive years, 90 remained in a male-dominated oc- system of sex segregation. Vocational aspirations
cupation, while 10 left for either a sex-neutral or have short life expectancies, and young women
female-dominated occupation. At the same time, frequently aspire to female-dominated occupa-
11 entered a male-dominated occupation from tions at one point in time and male-dominated
one of these other occupation groups. Thus, the jobs at another (see Jacobs, 1989, chapter 5).
revolving door sends 10 out for every 11 it lets Further, the relationship between the sex type
in.2 There has been a slow net accumulation of of young women’s vocational goals and the sex
women in male-dominated occupations, but type of the jobs they obtain is weak (see Jacobs,
the net change is small compared to the size of 1989, chapter 5). These patterns suggest that
878 • Part VIII: Gender Inequality

the influence of sex-role socialization is limited. macro level despite its instability or permeability
Thus, sex-role socialization can be viewed as a at the individual or micro level. Our resolution
system of social control for the early years, nec- of this paradox is that sex roles are subject to
essary but not sufficient to account for the per- continual, systematic, but imperfect control
sistence of occupational sex segregation. throughout life. Individuals move, but the sys-
For those men and women who attend col- tem remains segregated, owing to the cumulative
lege, social control continues during higher ed- force of social pressures. . . .
ucation. Many consider the educational system
to be the realm of universalism, the paramount
social institution in our society for promoting Cumulative Disadvantage Versus
opportunities for all. Indeed, the declines in sex Revolving Doors
segregation in higher education have surpassed Two competing versions of the lifelong social
those in the labor force. Yet, as shown in Jacobs control perspective must be distinguished. The
(1989, chapter 6), important gender differences more familiar is what we label the “cumulative
persist in America’s colleges. Men and women disadvantage” view, which we contrast with the
pursue different majors, and informal social con- “revolving door” perspective. The cumulative
trol plays a prominent role on campus. disadvantage model holds that occupational
The field of study in college is an important segregation by sex is the result of the accumu-
but not decisive component of gender tracking. lation of obstacles women face in their pursuit
Segregation remains fluid during the college years of careers. Thus, socialization is compounded by
as many students move among male-dominated, educational inequality and labor market discrim-
sex-neutral, and female-dominated majors. Fur- ination, the cumulative force of which leaves
ther, the link between a person’s major in college women in a segregated set of occupations.
and subsequent occupational pursuits is far from This view is clearly implicit in Berryman’s
direct for many fields. Thus, the sex segregation discussion of women in science and engineering
of college majors resembles in both structure and (1983). Women face obstacles beginning with
process the segregation of aspirations and the ambivalent socialization about the importance of
segregation of occupations. Like sex-role social- careers and the stereotype that science is an un-
ization, sex segregation in the educational system feminine pursuit. Women are underrepresented
is a necessary but not sufficient cause of gender in science and engineering professions because of
inequality in the labor force. inadequate preparation in high school, attrition
Sex segregation in the labor force is the from science and engineering fields in college,
most durable of any context we consider, yet and minority status and hostility on the job for
sex-role assignments are not fixed when women those women who pursue such careers.
and men enter the labor force. Many women If the obstacles women confront in pursuing
move among male-dominated, sex-neutral, science and engineering careers are characteristic
and female-dominated occupations. The entry of male-dominated occupations, this model im-
of women into male-dominated fields often plies a number of specific predictions concern-
does not immediately follow schooling, but ing sex segregation. The first is that segregation
instead may follow a period of employment in would increase over women’s lifetimes as women
a female-dominated field or a spell as a home- continually leave male-dominated pursuits. One
maker. The patterns of stability and change in would expect to find more segregation as a co-
sex segregation are strikingly similar in aspira- hort ages, as the disadvantages women face ac-
tions, college majors, and careers, at both the cumulate. Second, mobility observed between
aggregate and the individual level. male-dominated and female-dominated occupa-
The problem, then, is accounting for the tions would take the form of attrition of women
persistence of sex segregation at the societal or from male-dominated pursuits. Delayed entry
 Jacobs—Revolving Doors • 879

into male-dominated occupations would be dif- barriers to women’s opportunities are permanent
ficult; at the very least, delayed entries of women and irreversible.
into male-dominated occupations should be less Thus we arrive at a lifelong social control per-
frequent than exits. Finally, if declines in sex spective that is not characterized by cumulative
segregation were to occur, this change would disadvantage. In recognition of the dramatic lev-
primarily take the form of changes between els of mobility we find, we refer to this view as
cohorts. The best one could expect for those the “revolving door” model of sex segregation.
already in the pipeline is to reduce the rate of The revolving door perspective can account for
attrition; real change must await a new cohort extensive sex-type mobility because it recognizes
that is exposed to different socialization, broader a variety of stages in the career development pro-
opportunities in school, and less hostility on the cess, and it recognizes a host of pressures women
job. One would not expect increased integration face. Most women will face one or more barriers
among those already in the labor force, since the to the pursuit of a career at some point; they are
cumulative effects of these social control pro- likely to overcome some of these and not others.
cesses are essentially irreversible. Thus there are reasons to expect substantial flows
While the cumulative disadvantage model of women into and out of male-dominated oc-
is attractive because of its recognition of the cupations, while overall the system changes only
pervasive nature of social control, three strik- gradually.
ing findings suggest that while social control is This perspective also accounts for the pat-
continuous, its effects do not increase over the terns of change in sex segregation we are be-
life course. First, occupational segregation by sex ginning to see. If social control is exercised
does not increase as men and women grow older throughout life, social change is likely to be ex-
(see Jacobs 1989, chapter 2). Second, sex-type perienced by women of different ages. Change
mobility involves substantial rates of midcareer should affect those already in the labor force as
entry into, as well as exit from, male-dominated well as those entering it and those in the edu-
occupations (see Jacobs 1989, chapter 7). Third, cational system. Theories that emphasize labor
overall change in segregation is observed across market entry as the decisive time for segrega-
different age groups, and is not simply the result tion cannot account for decreasing segregation
of cohort replacement (see Jacobs 1989, chapter among those already in the labor force. As
2). In other words, where the cumulative disad- shown elsewhere (Jacobs 1989, chapter 2), the
vantage view holds that change must come about declines in labor market segregation by sex in
principally because a new cohort has different the 1970s occurred throughout the age structure
experiences, the evidence suggests that a general and were not simply restricted to labor market
weakening of the system of controls broadens entrants.
opportunities for those already in the labor force. There is a striking similarity in the dynam-
There is as much change in young women’s as- ics of sex segregation in the development of
pirations over time as there is across new cohorts aspirations, in the segregation of college ma-
of women in the process of forming their aspi- jors, and in the sex segregation of the occupa-
rations. There is as much change for people al- tional structure. All three are characterized by
ready in college as there is among those entering change occurring among different age groups,
college. In recent years, there has been a nota- and all three have striking levels of mobility
ble decline in segregation among those already between male-dominated, sex-neutral, and
in the labor market (see Jacobs 1989, chapter female-dominated fields. The similarities of
2). The insight of the cumulative disadvantage these patterns in these diverse contexts and
view is that it recognizes the range of constraints across different stages of life convince us of
imposed on women throughout their lives. Its the utility of the social control/revolving door
limitation is that it assumes that the effects of perspective.
880 • Part VIII: Gender Inequality

NOTES ———. 1989. Revolving Doors: Sex Segregation and

1. Calculated from data in Employment and Earn- Women’s Careers. Stanford, CA: Stanford Univer-
ings (U.S. Bureau of Labor Statistics, 1986). sity Press.
2. These figures reflect the analysis of Current Kessler-Harris, Alice. 1982. Out to Work: A History
Population Survey data discussed in Jacobs (1989, of Wage-Earning Women in the United States. New
chapter 7). York: Oxford University Press.
McPherson, J. Miller, and Lynn Smith-Lovin. 1986.
REFERENCES “Sex Segregation in Voluntary Associations,”
Abrahamson, Mark, and Lee Sigelman. 1987. American Sociological Review, 51(1): 61–79.
“Occupational Sex Segregation in Metropoli- Remick, Helen, ed. 1984. Comparable Worth and
tan Areas,” American Sociological Review, 52(5): Wage Discrimination: Technical Possibilities and
588–97. Political Realities. Philadelphia: Temple Univer-
Bergmann, Barbara. 1986. The Economic Emergence sity Press.
of Women. New York: Basic Books. Reskin, Barbara, ed. 1984. Sex Segregation in the
Berryman, Sue R. 1983. Who Will Do Science? New Workplace: Trends, Explanations, Remedies. Wash-
York: Rockefeller Foundation. ington, D.C.: National Academy of Sciences.
Bianchi, Suzanne M., and Nancy Rytina. 1986. Reskin, Barbara F., and Heidi I. Hartmann. 1985.
“The Decline in Occupational Sex Segregation Women’s Work, Men’s Work: Sex Segregation on the
During the 1970s: Census and CPS Compari- Job. Washington, D.C.: National Academy of
sons,” Demography, 23(1): 79–86. Sciences.
Bielby, William T., and James N. Baron. 1986. “Men Sanday, Peggy R. 1981. Female Power and Male
and Women at Work: Sex Segregation and Statis- Dominance: On the Origins of Sexual Inequality.
tical Discrimination,” American Journal of Sociol- Cambridge: Cambridge University Press.
ogy, 91(4): 759–99. Smith, James P., and Michael P. Ward. 1984. Wom-
Goldin, Claudia. 1983. “The Changing Economic en’s Wages and Work in the Twentieth Century.
Role of Women: A Quantitative Approach,” Jour- Santa Monica, Calif.: Rand Corporation.
nal of Interdisciplinary History, 13(4): 707–33. Tienda, Marta, Shelley A. Smith, and Vilma Ortiz.
Greenberger, Ellen, and Laurence D. Steinberg. 1987. “Industrial Restructuring, Gender Segre-
1983. “Sex Differences in Early Labor Force Ex- gation, and Sex Differences in Earnings,” Ameri-
perience: Harbinger of Things to Come,” Social can Sociological Review, 52(2): 195–210.
Forces, 62(2): 467–86. U.S. Bureau of the Census. 1986. “Women in the
Gutek, Barbara. 1985. Sex and the Workplace. San American Economy,” Current Population Reports,
Francisco: Jossey-Bass. P-23, no. 146. Washington, D.C.: U.S. Govern-
Jacobs, Jerry A. 1986. “Trends in Contact Between ment Printing Office.
Men and Women at Work, 1971–1981,” Sociol- U.S. Bureau of Labor Statistics. 1986. Employment
ogy and Social Research, 70(3): 202–6. and Earnings. Washington, D.C.: U.S. Govern-
ment Printing Office.
 Reskin—Labor Markets as Queues • 881

102 • Barbara F. Reskin

Labor Markets as Queues

A Structural Approach to Changing Occupational Sex Composition

The historic persistence of sex segregation (Gross then shows how the determinants of occupa-
1968; Beller 1984) made women’s dramatic tional feminization conformed to queuing pro-
gains during the 1970s in such diverse male cesses. Finally, it argues that the queue model as I
occupations as pharmacy, bank management, have developed it offers a structural approach to
bartending, and typesetting noteworthy. Why understanding change in job composition.
did women disproportionately enter these and
a few other male occupations when they made
only modest progress in most and lost ground Queuing: An Overview
in a few? To answer this question, Patricia Roos Simply stated, a queuing perspective views labor
and I conducted a two-part study: multivariate markets as composed of labor queues and job
analyses of the changing sex composition in all queues that reflect, respectively, employers’ rank-
detailed census occupations and in-depth case ing of possible workers and workers’ ranking of
studies of 14 occupations in which women’s rep- jobs.2 The idea of the labor queue originated
resentation increased at least twice as much as in Lester Thurow’s (1969) work on race differ-
it had in the labor force as a whole (see Table ences in unemployment and was elaborated by
102.1). This chapter draws on those case studies: Hodge (1973) and Lieberson (1980). Rotella’s
bank managers (Bird 1990), bartenders (Detman (1977) analysis of the feminization of clerical
1990), systems analysts (Donato 1990a), public work, Lieberson’s (1980) study of racial and
relations specialists (Donato 1990b), pharma- ethnic inequality, and Strober’s (1984) theory of
cists (Phipps 1990a), insurance adjusters/exam- sex segregation implicitly invoked job as well as
iners (Phipps 1990b), typesetters/compositors labor queues. The queuing perspective presented
(Roos 1990), insurance salespersons (Thomas here views occupational composition as result-
1990), real estate salespersons (Thomas and ing from the simultaneous operation of job and
Reskin 1990), bakers (Steiger and Reskin 1990), labor queues (also see Reskin and Roos 1990).
book editors (Reskin and Roos 1990), print and Employers hire workers from as high in the labor
broadcast reporters, and accountants.1 Although queue as possible, and workers accept the best
the case studies revealed that widely different jobs available to them, so the most desirable jobs
factors precipitated these occupations’ feminiza- go to the most preferred workers, less attractive
tion, the process of change can be encompassed jobs go to workers lower in the labor queue, and
in a single perspective—queuing. This chapter the most lowly workers end up jobless or in jobs
begins by outlining the queuing approach. It others have rejected.

Barbara F. Reskin. “Labor Markets as Queues: A Structural Approach to Changing Occupational Sex Composi-
tion,” in Micro-Macro Linkages in Sociology, edited by Joan Huber, pp. 170–186, 188–192. Copyright © 1991.
Reproduced with permission of Sage Publications, Inc., Books via Copyright Clearance Center.
882 • Part VIII: Gender Inequality

Table 102.1. Percent Females in Feminizing Occupations, 1970, 1980, 1988

Occupation 1970 1980 1988 1970–1988 Increase

% % % %
Financial managers 19.4 31.4 42.4 23.0
Accountants/auditors 24.6 38.1 49.6 25.0
Operation/systems analysts 11.1 27.7 39.6 28.5
Pharmacists 12.1 24.0 31.9 19.8
Public relations specialists 26.6 48.8 59.1 32.5
Editors and reporters 41.6 49.3 n.a. n.a.
Insurance sales 12.9 25.4 29.7 16.8
Real estate sales 31.2 45.2 48.5 17.3
Insurance adjusters/examiners 29.6 60.2 72.2 42.6
Bartenders 21.2 44.3 49.6 28.4
Bakers 25.4 40.7 47.8 22.4
Bus Drivers 28.3 45.8 48.5 20.2
Typesetters/compositors 16.8 55.7 73.9 57.1
Labor force 38.0 42.6 45.0 7.0

Three structural properties characterize both example, in a society with relatively few workers
job and labor queues: (a) the ordering of their in the preferred group (A2) and few desirable
elements, (b) whether or not their elements over- jobs (B2), preferred groups will monopolize bet-
lap, and (c) their shape. By elements, I mean the ter or worse jobs than persons from their group
constituent units—groups of potential workers normally command. In consequence, when good
or jobs. Thus the ordering of elements simply in- jobs sharply outnumber highly ranked workers
dicates in what order workers rank possible jobs as in panels A2 and B1, employers must fill
and employers rank prospective workers. I pass the better jobs with workers from lower in the
over how workers order jobs in the labor queue labor queue than they usually hire. In contrast,
as something both sociologists and workers un- when the job queue is bottom heavy, only the
derstand. In ordering the labor queue, employers highest-ranked workers get desirable jobs, and
consider potential productivity and cost.3 How- workers moderately high in the labor queue are
ever, the indirect effect of workers’ sex on em- forced to settle for less. As Lieberson (1980, p.
ployers’ estimations of the former and its direct 297) has shown, both the absolute and the rela-
influence on how employers rank workers result tive sizes of each group in the labor queue affect
in labor queues that encompass and often act as lower-ranked workers’ chances of getting desir-
gender queues. able jobs. The larger a subordinate group relative
The absolute and relative numbers of ele- to the size of the preferred group, the more costly
ments in a queue determine its shape. Thus the it is for employers to deny its members good jobs
number of prospective workers in each subgroup (Hodge 1973).
in a labor market sets the shape of the labor Whether or not elements overlap denotes the
queue, and the number of jobs at each level in strength of rankers’ preferences for one element
the job queue fixes its shape. Panels A and B of over another. Among some employers, group
Figure 102.1 show how the shape of labor and membership is the overriding consideration in
job queues can vary while their order remains ordering the labor queue, and rankers favor per-
constant. This variation influences each group’s sons from the group they prefer, regardless of
probable access to occupations of varying desir- qualifications. For other labor queues, employ-
ability. (Put differently, it shows how a queue’s ers are indifferent to group membership except
shape influences each occupation’s chance of to break ties between otherwise equally qualified
recruiting workers from particular groups.) For prospects. Figure 102.2 illustrates variation in
 Reskin—Labor Markets as Queues • 883

Figure 102.1. Illustration of Variation in the Shape of Labor and Job Queues
Panel A. Hypothetical labor queues ordered by race for predominantly white and predominantly black
labor markets, respectively

A1: A2: Whites

Whites

Blacks

Blacks

Panel B. Hypothetical job queues ordered by nonmanual-manual work for predominantly nonmanual
and predominantly manual occupational structures, respectively

B1: B2: Nonmanual

Nonmanual

Manual

Manual

the intensity of raters’ preferences with respect rank, occasionally eschewing jobs in occupa-
to workers’ race in three hypothetical labor tions usually reserved for the preferred group
queues. The space between the races in panel A and appropriating more desirable jobs within
depicts the ranking for employers who invari- lower-ranked occupations that usually go to less
ably hire the lowest-ranked white worker over preferred groups. Strong preferences mean that
the best black worker; the overlapping groups workers invariably prefer one type of job (e.g.,
in panel B depict the situation in which raters’ professional or nonmanual work) over another.
aversion to blacks is slight. Panel C illustrates an Changes in any of these structural features
intermediate situation in which employers prefer of queues—employers’ rankings of workers or
white workers over equally qualified blacks but workers’ ranking of jobs, the intensity of either
will set aside racial biases to hire very talented groups’ preferences, and the relative distributions
blacks over mediocre whites. Where preferences of workers or jobs—can transform occupations’
are weak, substantial job-level integration exists. composition.
In contrast, when preferences are overriding, all
members of the favored group precede all mem-
bers of the disfavored group in the labor queue, Queuing Models as Structural
and the labor market is totally segmented. Approaches to Occupational
Workers’ job preferences also vary in inten- Composition
sity. Some workers may categorically prefer any Queuing emphasizes how employers rank groups
job in a high-ranked occupation to working in of potential workers as well as how workers rank
a lower occupation (expressed, for example, in jobs. Because rankings simply boil down to
rejecting manual jobs in favor of nonmanual preferences, is queuing not simply a fancy way
work). Alternatively, they may be more attuned to say that workers’ and employers’ preferences
to specific job characteristics than their overall affect their labor market decisions? What does
884 • Part VIII: Gender Inequality

Figure 102.2. Illustration of Variation in the Intensity of Raters’ Preferences with Respect to Race
Panel A. Racial group membership is an overriding consideration to rankers. Employers hire applicants as
qualified as possible, but hire unqualified whites before hiring highly qualified blacks.

Blacks Blacks Blacks Whites Whites Whites

Low Moderate High Low Moderate High

Level of Qualification

Panel B. Racial group membership is a minor consideration to rankers. Employers hire the most qualified
applicants, but within levels of qualification, give white applicants an edge over equally qualified blacks.

Blacks Whites Blacks Whites Blacks Whites

Low Low Moderate Moderate High High

Level of Qualification

Panel C. Racial group membership is an intermediate consideration to rankers. Employers will prefer a
more qualified white over a less qualified black, but will hire very qualified blacks over unqualified whites.

Blacks Blacks Whites Blacks Whites Whites

Low Moderate Low High Moderate High

Level of Qualification

the queuing perspective add to the neoclassical A queuing approach to changing occupa-
economic approach to occupational segregation? tional sex composition differs in several ways.
The neoclassical economic approach assumes First, queuing emphasizes the collective nature
(a) that workers’ labor market decisions represent of sex segregation that results from socially struc-
attempts to maximize their lifetime earnings, tured rankings by groups in conflict. In contrast to
(b) that differences in the sexes’ distributions the neoclassical perspective, queuing highlights
across jobs stem primarily from women’s family the roles played by power and conflict between
responsibilities, (c) that employers act to maxi- groups with contradictory interests in shaping
mize revenues, and (d) that, in the short run, dis- occupations’ composition. As I show below, the
criminatory “tastes” sometimes divert employers queuing perspective elucidates the role of group
from their economic best interests, but, in the power in determining occupational access.
long run, competitive disadvantages that accrue Second, it takes seriously the effects of non-
from indulging economically irrational tastes economic factors on workers’ rankings of occu-
lead discrimination to decline over time (Becker pations and employers’ rankings of prospective
1957). Thus traditional economic analysis treats workers. It reminds us that working conditions,
sex segregation as the aggregate outcome of the autonomy, social standing, career opportunities,
mostly rational choices of individual workers and sex composition influence how workers
and employers. appraise jobs and that employers’ prejudices,
 Reskin—Labor Markets as Queues • 885

stereotypes, and desire to preserve their own and dominance by legitimating unequal wages, insu-
other men’s advantages influence how they rank lating men from female competition, supporting
workers. men’s dominance in other public realms and the
Third, the queuing perspective assumes that home, and shoring up myths of essential sex dif-
the sexes rank most occupations similarly, so it ferences and male superiority (Reskin 1988). As
predicts women’s influx into accessible male oc- Cockburn (1988, p. 41) put it, “Behind occupa-
cupations in view of their superior rewards. In tional segregation is gender differentiation, and
explaining occupational feminization, it directs behind that is male power.”
our attention to changes in employers’ need for Other considerations inclining employers to
women and to declining barriers to women’s ac- preserve good jobs for men include (a) the abil-
cess to traditionally male jobs. In other words, ity of male workers—and sometimes custom-
the queuing perspective redirects us from char- ers—to penalize employers for hiring women
acteristics of female workers to structural prop- for “men’s” jobs (Bergmann and Darity 1981;
erties of labor markets that in turn are shaped by Strober 1988); (b) organizational inertia that
the preferences of employers and male workers. makes following standard practices the course
of least resistance; (c) custom—in other words,
jobs’ sex labels—combined with the uncertain-
Power and Coalitions in Ordering ties attendant on radically altering employment
the Labor Queue practices (Cohn 1985; Strom 1987; Figart and
In examining how conflict affects occupations’ Bergmann 1989, p. 36); (d) gender solidarity
composition, we must bear in mind three points: with male workers;4 and (e) forestalling later
First, three groups with competing interests have threats to their own jobs. As long as employ-
a stake in which sex predominates in an occu- ers can operate profitably, preserving male jobs
pation—employers (who are overwhelmingly can be a fairly cheap amenity, especially if their
male), male workers, and female workers. Sec- competitors also practice male preference (Stol-
ond, the greater social and economic power of zenberg 1982; Strober 1988). Thus, in reserving
employers and of men give them a substantial male-dominated jobs for men, employers at once
advantage in the struggle over what sex heads the appease male workers, who could make hiring
queue for desirable occupations. Third, the out- women costly; simplify organizational decisions;
come of the struggle over what sex dominates an and shore up the gender hierarchy from which
occupation—occupations’ sex compositions— they benefit.
represents a de facto coalition between compet-
ing groups. Male Workers. Male workers’ disadvantaged
economic position vis-à-vis employers and their
Employers. Given their orientation toward advantaged sex status divide their interests. They
minimizing costs and maximizing labor docility, stand to gain in the long run from acting col-
employers should prefer female workers for all lectively with all workers, but, in the short run,
jobs. That so few have done so reminds us that men are vulnerable to job and wage competition
other considerations affect male employers’ labor from women who must settle for lower wages.
decisions. Foremost among these is the personal Equal job opportunities for women can also
stake of both male employers and male workers cost men domestic benefits they derive from sex
in excluding women from male-labeled jobs to inequality in the workplace and can threaten
preserve sex differentiation and hence male priv- the gender hierarchy from which working men
ilege in all spheres. Because sex differentiation benefit (Hartmann 1976; Reskin 1988). Thus,
both legitimates and facilitates unequal treat- for the most part, male workers do better in the
ment, physically segregating the sexes at work short run by monopolizing male jobs than by
and assigning them different tasks preserves male welcoming female coworkers with whom they
886 • Part VIII: Gender Inequality

could collectively fight for better pay. Only adjusters and examiners. Heightened competi-
when large numbers of women already have tion for customers following deregulation forced
their foot in the door, or when labor shortages banks to provide more customer services while
or outside agencies have imposed integration, are trying to increase productivity and keep the lid
male workers freed to pursue their class interests on wages. In typesetting and composing, the
collectively with female workers (Hirsch 1986; International Typographical Union’s frequent
Milkman 1987; Cobble forthcoming). Under recourse to strikes spurred newspaper publish-
such conditions, male unions have tried to orga- ers to develop a printing technology that the
nize women and advocated equal pay. union could not control, and then they hired
women for the newly clericalized typesetting
Coalitions Between Employers and Male Work- and composing jobs. It is important, however,
ers. Male workers’ monopoly of most occupa- that, in occupations that became less attractive
tions stems from the de facto coalition between to men for reasons not tied to employers’ actions
male employers and male workers. Although I (book editing, real estate sales, retail pharmacy,
use the term coalition figuratively, employers’ and bartending), employers did not counter by
power to hire workers and assign them jobs improving male jobs. Employers’ economic in-
means that ordinarily male workers can resist terests prevailed, and they replaced men with
women’s entry only with their male bosses’ ac- cheaper women.
tive contrivance (Baker 1964; Hartmann 1976; In sum, the case studies suggested that em-
Cockburn 1988). The result of this coalition be- ployers put women ahead of men in labor queues
tween male workers and employers is also seen only when labor costs became a pressing concern
in feminizing occupations in women’s confine- or when action by women or government regu-
ment to the least desirable jobs, thereby provid- lators raised the cost of male preference.5 Faced
ing male workers a refuge from integration in with economic exigencies, employers’ willingness
still-male jobs (Cockburn 1988). For example, to donate profits to support white male privilege
when outside pressure forced a large publishing went by the wayside, and they struck a bargain
firm to promote women to jobs as editorial as- with women workers.
sistants, it created a new, higher rank into which
it promoted male former editorial assistants (Os- Coalitions Between Male Employers and Female
terman 1979). Workers. Why women should enter into de
Despite the forces that preserve sex segre- facto coalitions with male employers is no mys-
gation, it broke down in the occupations we tery. Disadvantaged relative to both male work-
studied when male employers or male workers ers and male employers, women stand to gain
withdrew from their coalition. Employers never from cooperating with either group. Employers
violated a coalition by directly substituting are more likely to accept a coalition with women
women for men, but they set aside customary when their numbers are large. Cobble’s (forth-
practices, their own sex biases, and concern coming) comparison of women’s efforts early in
with the reactions of male workers when their this century to preserve the right to serve drinks
firms’ survival depended on cutting labor costs with their efforts to tend bar illustrates the im-
or averting unionization. In the occupations we portance of numbers. Women waitresses’ moder-
studied, increased competition and the concom- ate representation within union locals gave them
itant need to cut costs as well as vulnerability the clout to block proposed restrictions against
to work disruptions by unions precipitated this their serving alcoholic drinks. In contrast, wom-
action. For example, industry-wide demand en’s small numbers doomed their efforts to tend
for increased profits during the high-inflation bar until they could invoke the Civil Rights
1970s prompted insurance companies to cut Act. In the 1970s, women capitalized on their
costs by standardizing the work of insurance numbers in the book publishing, newspaper,
 Reskin—Labor Markets as Queues • 887

broadcasting, and banking industries to challenge high-turnover occupations like book editing,
their exclusion from male jobs or their confine- bartending, and residential real estate sales had
ment in low-status posts. Large numbers mean no stake in what happened in the long run. High
that women can better spot their concentration turnover also means workers were too poorly or-
in the least desirable jobs and more effectively ganized to resist integration. Finally, in reorga-
protest it. Thus women’s increased share in labor nizing work, employers had an effective tool in
queues for male occupations contributed to their phasing out men. By changing the work process
gains both by increasing the odds that employers to incorporate female-labeled tasks, employers
would get down to women in the hiring queue could disguise the fact that they were filling what
and by making them more powerful adversaries. had been men’s jobs with women (Davies 1975,
p. 282; Rotella 1977, pp. 162, 165) and recast
Resistance by Male Workers. Male workers’ stake them as women’s work. For instance, after trans-
in monopolizing the best jobs encourages them forming the work of adjusters, industry trade
to oppose employers’ efforts to integrate women. journals construed the ideal insurance adjuster
To this end, they may try to reduce women’s pro- as a woman.7
ductivity by denying them training and infor- However, men did effectively resist women’s
mation or withhold their own labor (Doeringer employment in the most desirable specialties
and Piore 1971; Bergmann and Darity 1981).6 within desegregating occupations, such as pro-
Whether men win the battle depends on their duction baking, commercial real estate, and
power base. For example, early in the Industrial industrial pharmacy—partly by informal pres-
Revolution, before strong unions existed, em- sures that discouraged women’s entry. I suspect,
ployers could easily feminize jobs (Kessler-Harris however, that women’s continued exclusion de-
1982, p. 266; Cohn 1985; Milkman 1987; pended more on employer discrimination in job
Cockburn 1988). However, once unions were assignment and promotion practices than on
well established, they could preserve their do- male workers’ actions.
mains. They did so in bartending, baking, and
typesetting and composing as well as many other
occupations throughout most of the century. Conclusions
Obviously, men did not ultimately prevent In summary, the queuing model proposed here
women’s entry in the occupations we studied. In holds that (a) employers rank prospective work-
most of those occupations, by abandoning jobs, ers in terms of their potential productivity as
male workers gave employers the go-ahead to well as their personal characteristics, but that
hire women. Sometimes men do not see women they are also influenced by current employees
as a threat. For instance, although men fought and others who can impose costs for hiring or
women’s integration into clerical jobs in the Brit- failing to hire women; (b) shortages—whether
ish railroads (Cohn 1985), they did not resist from job growth or from a job’s inability to at-
women’s movement in similar jobs in the United tract enough customary workers—prompt em-
States because integration occurred slowly, and ployers to hire workers from lower in the labor
Pullman reserved the best jobs for men (Hirsch queue; (c) male workers affect women’s access to
1986, p. 33). In rapidly growing occupations, jobs through their ability to preempt jobs, their
men did not resist women’s entry because growth power to enforce their monopoly over desirable
ensured that good jobs were plentiful, and jobs, and their ability, in abandoning jobs, to be-
within-occupation segregation assured men’s ac- stow them on workers lower in the labor queue;
cess to the best jobs. and (d) women’s search for better jobs leads
Thus the case studies of accounting and them to individually and collectively challenge
systems analysis revealed no systematic ef- their exclusion from men’s jobs and to move into
forts to exclude women. Male incumbents in male lines of work that become open. The case
888 • Part VIII: Gender Inequality

study data support these predictions. I caution on which this chapter is based appear in our book,
readers that our selection of case studies of oc- Job Queues, Gender Queues: Explaining Women’s In-
cupations in which women’s gains were excep- roads into Male Occupations (1990).
tionally large limits the generalizability of these 1. I conducted the last two unpublished studies.
Unless otherwise noted, assertions about these oc-
findings. However, most occupations included
cupations are based on the case studies whose full
specialties or work settings that had feminized at citations are in the references. For a description of
varying rates, so the case studies provide varia- the case study method, see Reskin and Roos (1990).
tion on the dependent variable. Nonetheless, the 2. This analysis takes for granted the existence
bases for women’s movement into occupations of multiple labor markets, each composed of a labor
not marked by rapid feminization may have dif- queue of potential workers and a job queue of avail-
fered. Patricia Roos and I addressed this question able jobs.
in statistical analyses of all 503 detailed occupa- 3. Incumbent workers can affect the ordering of
tions (Reskin and Roos 1985). the labor queue, for example, through seniority rules
that restrict employers’ hiring prerogatives.
The case studies support two empirical gener-
4. This should be especially common among
alizations, both of which are consistent with the bosses who share a similar background with male
operation of queuing processes. First, in most workers, previously held the jobs in question, and
of the occupations we studied, after they had currently work with or remain friendly with their
become less attractive to men, employers hired former coworkers. Two examples are blue-collar su-
disproportionate numbers of women. Second, pervisors or sales managers.
within these nominally desegregating occupa- 5. Unless it is backed by threats from regula-
tions, women tended to be relegated to female tory agencies, pressure from women rarely suffices
enclaves, while men retained the most desirable to persuade employers to assign women to male
jobs, partly because it elicits opposition from male
jobs.8 These findings confirm Bielby and Baron’s
workers.
(1986) contention that greater balance in the sex
6. Employers can rarely forestall resistance
composition of occupations does not necessarily by feminizing entire work teams. Seniority rules
imply decreased segregation of jobs and also es- may prevent their replacing men, they may doubt
tablish the fallacy of inferring declines in sex seg- whether enough qualified women are available, and
regation from occupational-level data. Yet social they may be unable to train women because male
analysts cannot ignore such data, because policy unions control training or because training is too
makers and the media use them to assess change, long and expensive.
and, as W. I. Thomas said, situations men define 7. Kessler-Harris (1986) and Strober and Arnold
(1987) also recounted how banks exploited sex ste-
as real are real in their consequences. Exagger-
reotypes to justify hiring women for banking jobs as
ated conclusions about women’s progress in male
well as to exclude them from the same jobs.
occupations support claims that governmen- 8. For further evidence on this point, see Chapter
tal agencies no longer need to intervene in the 3 in Reskin and Roos (1990).
workplace to ensure equal treatment. Thus it is
incumbent on researchers to determine whether REFERENCES
or not the trends that superficial comparisons Baker, Ross. 1964. “Entry of Women into Federal
imply are genuine. Job World—At a Price.” Smithsonian 8:82–91.
Becker, Gary. 1957. The Economics of Discrimina-
NOTES tion. Chicago: University of Chicago Press.
I am grateful to Ross Boylan, Lowell L. Hargens, Beller, Andrea. 1984. “Trends in Occupational Seg-
Heidi Hartmann, and Patricia A. Roos for their regation by Sex, 1960–1981.” Pp. 11–26 in Sex
comments on earlier drafts of this chapter and to Segregation in the Workplace: Trends, Explanations,
James N. Baron and Ronnie Steinberg for helpful Remedies, edited by B. Reskin. Washington, DC:
discussions of the issues it addresses. The larger study National Academy Press.
that gave rise to the ideas in this chapter is a collabo- Bergmann, Barbara R., and William Darity.
rative project with Patricia A. Roos. The case studies 1981. “Social Relations, Productivity, and
 Reskin—Labor Markets as Queues • 889

Employer Discrimination.” Monthly Labor Re- Paper prepared for the U.S. Department of
view 104:47–49. Labor, Commission on Workforce Quality and
Bielby, William T., and James N. Baron. 1986. “Men Labor Market Efficiency, American University.
and Women at Work: Sex Segregation and Statis- Gross, Edward. 1968. “Plus ça Change. . . . The
tical Discrimination.” American Journal of Sociol- Sexual Segregation of Occupations over Time.”
ogy 91:759–99. Social Problems 16:198–208.
Bird, Chloe. 1990. “High Finance, Small Change: Hartmann, Heidi I. 1976. “Capitalism, Patriarchy,
Women’s Increased Representation in Bank Man- and Segregation by Sex.” Signs 1: 137–70.
agement.” Pp. 145–66 in B. F. Reskin and P. A. Hirsch, Susan E. 1986. “Rethinking the Sexual Di-
Roos, Job Queues, Gender Queues: Explaining vision of Labor: Pullman Repair Shops, 1900–
Women’s Inroads into Male Occupations, Philadel- 1969.” Radical History Review 3:26–48.
phia: Temple University Press. Hodge, Robert W. 1973. “Toward a Theory of Ra-
Cobble, Dorothy Sue. Forthcoming. “‘Drawing the cial Differences in Employment.” Social Forces
Line’: The Construction of a Gendered Work- 52:16–31.
force in the Food Service Industry.” In Work En- Kessler-Harris, Alice. 1982. Out to Work. New York:
gendered, edited by Ava Baron. Ithaca: Cornell Oxford University Press.
University Press. ———. 1986. “Women’s History Goes to Trial:
Cockburn, Cynthia. 1988. “The Gendering of Jobs: EEOC vs. Sears, Roebuck and Co.” Signs
Workplace Relations and the Reproduction of 1:767–79.
Sex Segregation.” Pp. 29–42 in Gender Segrega- Lieberson, Stanley. 1980. A Piece of the Pie. Berkeley:
tion at Work, edited by S. Walby. Milton Keynes, University of California Press.
England; New York: Open University Press. Milkman, Ruth. 1987. Gender at Work. Urbana:
Cohn, Samuel. 1985. The Process of Occupational University of Illinois Press.
Sex-Typing: The Feminization of Clerical Labor in Osterman, Paul. 1979. “Sex Discrimination in Pro-
Great Britain. Philadelphia: Temple University fessional Employment: A Case Study.” Industrial
Press. and Labor Relations Review 32: 451–64.
Davies, Margery W. 1975. “Women’s Place Is at the Phipps, Polly A. 1990a. “Occupational and Indus-
Typewriter: The Feminization of the Clerical trial Change: Prescription for Feminization.”
Labor Force.” Pp. 279–96 in Labor Market Seg- Pp. 111–28 in B. F. Reskin and P. A. Roos, Job
mentation, edited by R. Edwards, M. Reich, and Queues, Gender Queues: Explaining Women’s In-
D. Gordon. Lexington, MA: D. C. Heath. roads into Male Occupations. Philadelphia: Tem-
Detman, Linda. 1990. “Women Behind Bars: The ple University Press.
Feminization of Bartending.” Pp. 241–56 in B. ———. 1990b. “Occupational Resegregation: A
F. Reskin and P. A. Roos, Job Queues, Gender Case Study of Insurance Adjusters and Examin-
Queues: Explaining Women’s Inroads into Male Oc- ers.” Pp. 224–40 in B. F. Reskin and P. A. Roos,
cupations. Philadelphia: Temple University Press. Job Queues, Gender Queues: Explaining Women’s
Doeringer, Peter B., and Michael J. Piore. 1971. Inroads into Male Occupations. Philadelphia:
Internal Labor Markets and Manpower Analysis. Temple University Press.
Lexington, MA: D. C. Heath. Reskin, Barbara F. 1988. “Bringing the Men Back
Donato, Katharine M. 1990a. “Programming for In: Sex Differentiation and the Devaluation of
Change? The Growing Demand for Women Sys- Women’s Work.” Gender & Society 2:58–81.
tems Analysts.” Pp. 167–82 in B. F. Reskin and P. Reskin, Barbara F., and Patricia A. Roos. 1985.
A. Roos, Job Queues, Gender Queues: Explaining “Collaborative Research on the Determinants
Women’s Inroads into Male Occupations. Philadel- of Changes in Occupations’ Sex Composition
phia: Temple University Press. Between 1970 and 1980.” National Science
———. 1990b. “Keepers of the Corporate Image: Foundation Grant no. 85-NSF-SES-85-12452.
Women in Public Relations.” Pp. 129–44 in B. Manuscript.
F. Reskin and P. A. Roos, Job Queues, Gender ———. 1990. Job Queues, Gender Queues: Explain-
Queues: Explaining Women’s Inroads into Male Oc- ing Women’s Inroads into Male Occupations. Phil-
cupations. Philadelphia: Temple University Press. adelphia: Temple University Press.
Figart, Deborah M., and Barbara Bergmann. 1989. Roos, Patricia A. 1990. “Hot Metal to Electronic
“Facilitating Women’s Occupational Integration.” Composition: Gender, Technology, and Social
890 • Part VIII: Gender Inequality

Change.” Pp. 275–98 in B. F. Reskin and P. A. Strober, Myra, and Carolyn Arnold. 1987. “The
Roos, Job Queues, Gender Queues: Explaining Dynamics of Occupational Segregation Among
Women’s Inroads into Male Occupations. Philadel- Bank Tellers.” Pp. 107–48 in Gender in the Work-
phia: Temple University Press. place, edited by C. Brown and J. Pechman. Wash-
Rotella, Elyce. 1977. From Home to Office: U.S. ington, DC: Brookings Institution.
Women at Work, 1870–1930. Ann Arbor: UMI Strom, Sharon Hartman. 1987. “‘Machines Instead
Research Press. of Clerks’: Technology and the Feminization of
Steiger, Thomas, and Barbara F. Reskin. 1990. “Bak- Bookkeeping, 1910–1950.” Pp. 63–97 in Com-
ing and Baking off: Deskilling and the Chang- puter Chips and Paper Clips. Vol. 2, edited by H.
ing Sex Make Up of Bakers.” Pp. 257–74 in B. Hartmann. Washington, DC: National Academy
F. Reskin and P. A. Roos, Job Queues, Gender Press.
Queues: Explaining Women’s Inroads into Male Oc- Thomas, Barbara J. 1990. “Women’s Gains in In-
cupations. Philadelphia: Temple University Press. surance Sales: Increased Supply, Uncertain De-
Stolzenberg, Ross. 1982. “Industrial Profits and the mand.” Pp. 183–204 in B. F. Reskin and P. A.
Propensity to Employ Women Workers.” Paper Roos, Job Queues, Gender Queues: Explaining
presented at the Workshop on Job Segregation by Women’s Inroads into Male Occupations. Philadel-
Sex, Committee on Women’s Employment and phia: Temple University Press.
Related Social Issues, National Research Council, Thomas, Barbara J., and Barbara F. Reskin. 1990.
Washington, DC. “A Woman’s Place Is Selling Homes: Women’s
Strober, Myra. 1984. “Toward a General Theory of Movement into Real Estate Sales.” Pp. 205–24 in
Occupational Sex Segregation.” Pp. 144–56 in B. F. Reskin and P. A. Roos, Job Queues, Gender
Sex Segregation in the Workplace: Trends, Explana- Queues: Explaining Women’s Inroads into Male Oc-
tions, Remedies, edited by B. Reskin. Washington, cupations. Philadelphia: Temple University Press.
DC: National Academy Press. Thurow, Lester. 1969. Poverty and Discrimination.
———. 1988. “The Processes of Occupational Seg- Washington, DC: Brookings Institution.
regation: Relative Attractiveness and Patriarchy.” ———. 1972. “Education and Economic Equality.”
Paper presented at the meetings of the American The Public Interest 28:66–81.
Educational Research Association, New Orleans.

103 • Elizabeth H. Gorman and Julie A. Kmec

Glass Ceilings in Corporate Law Firms

Women remain underrepresented at the top women encounter increasing obstacles relative
levels of business organizations, despite a slow to men as they move up the organizational
increase over time in their presence there (Cres- ladder. This proposition, which we label the
well 2006; Padavic and Reskin 2002, 97–107). increasing-disadvantage model, is a core element
Recent research has examined the claim that of the popular “glass ceiling” metaphor (Baxter

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Elizabeth H. Gorman and Julie A. Kmec, “Hierarchical Rank and Women’s Or-
ganizational Mobility: Glass Ceilings in Corporate Law Firms,” American Journal of Sociology 114:5 (March
2009), pp. 1428–1474, published by the University of Chicago Press. The article printed here was originally
prepared by Elizabeth H. Gorman and Julie A. Kmec for this fourth edition of Social Stratification: Class, Race,
and Gender in Sociological Perspective, edited by David B. Grusky. Copyright © 2014 by Westview Press.
 Gorman and Kmec—Glass Ceilings in Corporate Law Firms • 891

and Wright 2000; Federal Glass Ceiling Com-

mission 1995a, 1995b). Despite continued Why Expect an Increasing Female
widespread public acceptance of the glass ceil- Disadvantage in Mobility?
ing idea and some findings that are consistent If women’s mobility chances decline relative to
with it (Cohen, Broschak, and Haveman 1998; men’s as they move up the organizational hi-
Paulin and Mellor 1996), most research to date erarchy, their increasing disadvantage must be
has failed to support the increasing-disadvantage attributable to an increasing decision-maker
model (see Bruderl, Preisendorfer, and Ziegler preference for men over women. Why would
1993; Dencker 2008; Petersen and Saporta employers have such a preference, and why
2004; Spilerman and Petersen 1999; Lewis would it increase at higher organizational levels?
1986; Powell and Butterfield 1994; DiPrete and
Soule 1988; Baxter and Wright 2000). Instead,
women’s scarcity in top organizational ranks may Decision Makers’ Gender
simply be the result of the cumulative effect of a Preferences: Biasing Processes
constant—or even decreasing—disadvantage at Three processes can bias organizational deci-
successive hierarchical levels (Baxter and Wright sion makers’ conscious or unconscious gender
2000). preferences.
Theoretical and empirical problems make
it difficult to draw firm conclusions from these Gender as an indicator of competence. Employ-
studies, however. Researchers have neglected to ers want competent employees, but competence
theorize why we should expect women’s and men’s cannot be observed directly. It must be inferred
relative mobility prospects to vary by organiza- from other evidence, such as past academic and
tional level and have failed to identify scope con- job performance. Evaluators also base inferences
ditions under which an increasing-disadvantage of competence on gender, race, and social class
pattern might occur. In addition, most research (Berger, Rosenholtz, and Zelditch 1980; Ridge-
has focused exclusively on promotion, neglect- way 1997). An extensive body of laboratory re-
ing the issue of external hiring into upper-level search demonstrates that men are presumed to
positions. Empirically, previous research often be more competent than women (e.g., Dovidio
used inadequate organizational data, with most et al. 1988; Lucas 2003; Wagner, Ford, and Ford
prior studies analyzing the personnel records of 1986). Moreover, evidence of strong past perfor-
single organizations over specific spans of time. mance does not erase the impact of gender. Even
Researchers have also paid little attention to if a man and a woman have performed equally
defining and measuring the sex composition of well, inferences based on gender will still raise
relevant candidate pools, which may have pro- the ability expectation for the man and lower it
duced biased results. for the woman (Pugh and Wahrman 1983; Wag-
In this study we develop a theoretical basis ner, Ford, and Ford 1986).
for explaining why women’s relative mobility
prospects decline at higher hierarchical levels Masculine role-incumbent schemas and gender
and outline scope conditions defining settings stereotypes. In addition to looking for compe-
in which the increasing-disadvantage pattern is tence, decision makers charged with filling an
likely to be observed. We also empirically demon- organizational position usually have in mind a
strate the presence of an increasing-disadvantage more detailed mental model, or schema, of the
pattern in internal promotion and the absence desired incumbent and seek candidates who
of such a pattern in external hiring in the are congruent with it (Heilman 1983; Perry,
professional ranks of corporate law firms, a Davis-Blake, and Kulik 1994). The attributes
setting that permits us to study multiple orga- of role-incumbent schemas can include both sex
nizations with highly comparable organizational and stereotypically gendered abilities and skills.
structures. At the same time, because people’s impressions
892 • Part VIII: Gender Inequality

Table 103.1. Effects of Characteristics of High-Ranking Positions on Mechanisms of Gender Preference

Mechanisms

Masculine role-incumbent schemas as

heuristic to assess role suitability

Reliance on gender Role requires

as an indicator of Role is sex- stereotypically
Characteristic general competence labeled as male masculine skills In-group favoritism
Accentuates the Makes a role Makes a role seem to Strengthens the
importance of seem more require stereotypically preference for allocating
Status and
competence, thus appropriate for masculine skills, position to in-group
authority
increasing the impact men than for such as leadership, members, because
of this mechanism women decisiveness, etc. rewards are greater
Makes it more difficult
Makes a role seem to
to draw inferences Intensifies the need
require stereotypically
Work of ability from past for trust, and people
— masculine skills,
uncertainty performance, so are more comfortable
such as leadership,
gender is weighted trusting similar others
decisiveness, etc.
more heavily
Makes a role Makes a role seem to
Male Means that in-group
seem more require stereotypically
predominance favoritism is directed
--- appropriate for masculine skills,
among predominantly toward
men than for such as leadership,
incumbents men
women decisiveness, etc.

of others are influenced by applicable stereotypes

(see Fiske, Lin, and Neuberg 1999; Hilton and The Intensification of Gender Bias in
von Hippel 1996), employers are likely to per- High-Ranking Positions
ceive male candidates as possessing stereotypi- Higher-ranking positions typically have higher
cally masculine abilities and female candidates social status and greater authority. Work pro-
as lacking them. Thus, when the attributes of a cesses at higher organizational levels also tend
role-incumbent schema include either the male to involve more uncertainty; they encompass a
sex category or stereotypically masculine quali- wider and less predictable variety of issues, re-
ties, male candidates will seem better suited to quire decision making based on more ambigu-
the role than female candidates (Eagly and Karau ous information, and depend more heavily on
2002; Gorman 2005). cooperation from peers and subordinates (Kanter
1977; Thomas and Gabarro 1999). Finally, the
In-group favoritism. People tend to rate mem- predominance of men has been especially per-
bers of their “in-group” more positively than sistent in high-ranking positions. These char-
“out-group” members, especially on dimensions acteristics should affect the basic mechanisms
such as trustworthiness and cooperativeness, and outlined previously in multiple ways (see Table
are likely to favor in-group members when allo- 103.1).
cating rewards (Brewer and Brown 1998; Fiske Higher status and authority are likely to
et al. 2002). Consistent with these tendencies, intensify the impact of gender as an indicator
male decision makers are more likely than fe- of competence in selection decisions. When a
male decision makers to prefer male candidates role incumbent can more substantially affect
for hiring and promotion (Beckman and Phillips the organization’s collective performance, as
2005; Gorman 2005, 2006). high-ranking jobholders generally can, decision
 Gorman and Kmec—Glass Ceilings in Corporate Law Firms • 893

makers are likely to pay closer attention to all operate in these selection decisions, disadvantag-
indicators of competence—including gender— ing women. As a result, women are likely to de-
and will be less willing to take risks by selecting velop skills and resources more slowly than men
a candidate whose attributes raise doubts. The do. Employees build skills and networks through
greater unpredictability involved in upper-level on-the-job experience, as they are given increas-
positions should also diminish decision mak- ingly challenging assignments and promotions,
ers’ relative reliance on candidates’ performance and through mentoring and sponsorship by se-
records—because the same problems may not nior employees. Over time, gender differences
recur—and thus increase their reliance on gen- in the allocation of developmental assignments
der as evidence of competence. and mentoring can create a real gap between
Role-incumbent schemas for high-level jobs men and women who began their careers with
typically have a masculine slant, because cultural similar capabilities. Because of these cumulative
beliefs associate status and authority with men effects, objective sex differences in skills and re-
(Lucas 2003; Ridgeway and Correll 2004). Such sources are likely to be greater among candidates
role-incumbent schemas also include stereotyp- for high-level positions than among those for
ically masculine “soft skills,” such as leadership low-level positions.
ability and decisiveness (Dodge, Gilroy, and
Fenzel 1995; Martell et al. 1998). More work
uncertainty may also intensify perceptions that Scope Conditions
such a job requires stereotypically masculine The increasing-disadvantage model is subject
traits. Greater historical male predominance also to three scope conditions. First, the pattern
shapes role-incumbent schemas by labeling the should be more evident in large organizations
role as male (Cohen, Broschak, and Haveman than in small ones, because differences between
1998; Konrad and Pfeffer 1991). Finally, deliber- high- and low-level positions in authority, work
ate in-group favoritism may operate with greater uncertainty, and gender composition are likely
force in high-status positions, because men have to be greater there. Second, the model is most
a greater incentive to preserve their control over applicable to comparisons between higher and
those positions (Reskin 1988). Unintentional lower ranks within identifiable job ladders.
in-group favoritism may also increase, because Large organizations often cluster jobs that share
work uncertainty enhances the value of smooth similar functions and technology into recog-
communication and shared cultural understand- nizable ladders, with successive levels linked by
ing (Kanter 1977). increasing skill gradients and clear promotion
patterns (Althauser 1989; Baron, Davis-Blake,
and Bielby 1986). Unmistakable promotion
The Cumulative Effects of Past ceilings often distinguish lower-tier manual,
Selection Decisions clerical, or technical jobs from upper-tier profes-
In organizations with many levels, employees sional and managerial ones (DiPrete and Soule
typically enter a promotion tournament in- 1988; Spilerman and Petersen 1999). Third, the
volving several rounds of formal selection pro- increasing-disadvantage pattern is most likely to
cesses as they climb the hierarchy (Rosenbaum occur within relatively sex-integrated job lad-
1979). Less formally, decision makers engage ders. Where job ladders are severely segregated
in similar processes of evaluating and selecting by sex, the very idea of a female mobility disad-
employees for important developmental assign- vantage is not really meaningful; the disadvan-
ments, membership on key project teams, and tage lies at the point of entry to separate male
mentor-protégé relationships. All three of the bi- and female career paths. All three scope condi-
asing processes discussed previously are likely to tions are likely to be met in the professional and
894 • Part VIII: Gender Inequality

managerial ranks of large corporations or profes- disadvantage for women in partner hiring than
sional service firms. in entry-level hiring.
We drew our sample of 796 law firm offices
from the 1996 edition of the National Direc-
External Hiring into High-Level tory of Legal Employers (the NALP Directory),
Positions prepared by the National Association for Law
We expect the processes discussed thus far to Placement. Within the 796 establishments, a
apply to external hiring into high-level posi- total of sixty-nine hundred selection decisions
tions, as well as to internal promotion. Never- took place between 1995 and 1996. We gathered
theless, women’s disadvantage in mobility into data on the number and gender of new first- and
high-ranking jobs should be less in the case of second-year associate hires and newly promoted
external hiring than in internal promotion—if and hired partners by comparing the names of
candidates typically hold comparable positions lawyers listed in the 1995 and 1996 editions of
in organizations of similar size and promi- the Martindale-Hubbell Law Directory, which
nence. Such candidates have already been vetted lists almost every law office and lawyer in the
through selection processes in those organiza- United States.
tions and have received a seal of approval. For
women, this positive signal should offset some or
all of the negative impact of their sex on decision Descriptive Analyses
makers’ expectations about their competence An increasing-disadvantage pattern manifests
(see Heilman, Martell, and Simon 1988). itself in a decline in the ratio of the probability
of women’s mobility, relative to men’s, at higher
hierarchical levels—as long as this decline is not
Data and Methods due to women’s disproportionate self-selection
We focus here on large law firms that serve corpo- into alternative career paths (Baxter and Wright
rate clients. Such “corporate law firms” are orga- 2000). Accordingly, we begin with simple com-
nized as partnerships (or the equivalent). Below parisons of women’s and men’s mobility rates
the rank of partner are salaried “associates”: ju- and the ratio of these rates at the entry and
nior lawyers employed on a probationary basis partnership levels. We calculate male and female
during a “partnership track” period before being mobility rates for each type of selection decision
considered for promotion to partner. Although for each establishment by taking the ratio of the
most new partners are promoted from within, number of selected men or women to the num-
law firms may also hire partners externally. ber of men or women in the relevant candidate
Corporate law firms meet our scope conditions pool. These rates describe the mobility chances
because they are relatively large and have clear of the average lawyer in the relevant candidate
professional job ladders that are sex-integrated pool in a given establishment.
on the lower rungs. As a research setting, they To rule out differential self-selection as an al-
offer additional advantages. Because almost all ternative explanation, it is necessary to identify
corporate law firms have the same basic struc- the relevant candidate pool with some care. Ide-
ture, job titles and ranks are similar across or- ally, the candidate pool for selection into a given
ganizations. This setting minimizes the problem position should include both those who were ac-
of unobserved heterogeneity, because almost all tually considered for the position and those who
new lawyers have the same amount of education were previously eliminated from contention by
and little or no prior work experience. We expect the employer, either formally or informally. In-
that women’s mobility disadvantage, relative to formal elimination by the employer includes
men, is greater in promotion to partnership than ostensibly voluntary withdrawal in anticipation
in entry-level hiring. We do not expect a greater of formal rejection. In contrast, individuals who
 Gorman and Kmec—Glass Ceilings in Corporate Law Firms • 895

choose other options without employer pressure Wolf 2003). Departures in the fourth and later
do select themselves out of the competition and years of the partnership track are more likely to
should not be included in the candidate pool. reflect associates’ discouragement about partner-
Because data bearing on individuals’ entrances ship prospects (Noonan and Corcoran 2004).
into and exits from the competition for a partic- Law firms generally begin serious evaluation
ular position are usually difficult to obtain, or if of associates’ partnership prospects around the
obtainable are difficult to interpret, it is almost fourth year of the partnership track, and asso-
always necessary to estimate the size and gender ciates who perceive their promotion chances as
composition of the candidate pool. slim often leave their firms in the fourth, fifth,
For hiring decisions, the key difficulty lies in or later years (Davis 2000). The turning point at
knowing who has chosen to enter the candidate about the fourth year of the partnership track is
pool. Appropriately defined, the candidate pool also reflected in gendered attrition patterns. Dif-
for a position should include both individuals ferences between men’s and women’s departure
who formally applied or expressed interest and rates in the first three years are slight, but be-
those who would have done so, had they not ginning in the fourth year, women’s attrition in-
been prevented or discouraged by the firm. Yet creasingly exceeds men’s (Noonan and Corcoran
it is rarely possible to know much about the in- 2004). Following this reasoning, we identify the
dividuals who formally applied, let alone about cohort of associates considered for promotion
those who were discouraged from applying. One to partner in the fall of 1995 and estimate the
approach to this problem is to assume that the numbers of women and men in that cohort at
characteristics of the candidate pool reflect those the beginning of that cohort’s fourth year, taking
of the broader pool from which candidates are account of the length of each office’s partnership
drawn (see Cohen, Broschak, and Haveman track and the average male and female attrition
1998). This is reasonable, because organizations rates for each year for the focal cohort (calculated
often shape the demographic composition of from a separate NALP survey).
their applicants through their choice of recruit-
ing methods (Reskin and McBrier 2000), and
because women and men are not usually dis- Multivariate Analyses
proportionately attracted to different employers We estimate multilevel logistic regression models
(Konrad et al. 2000). For entry-level hiring de- to predict the probability that a selection deci-
cisions, we use the numbers of male and female sion results in the hiring of a woman. We treat
students in the law schools where a given law selection decisions as the unit of analysis and
office conducted on-campus recruiting (named estimate the probability that decision makers
in the office’s NALP Directory listing). For exter- choose a candidate with a specific characteris-
nal partner hires, we use the numbers of female tic, such as sex, to fill the vacancy (e.g., Cohen,
and male partners in the other sample law firm Broschak, and Haveman 1998; Phillips and So-
offices in the same metropolitan area or state. rensen 2003). Because of the nested structure
In the case of promotions to partnership, the of the data, we estimate two-level hierarchical
chief difficulty lies in interpreting the behav- logistic models using grand mean centering for
ior of associates who depart before the end of continuous variables and uncentered binary vari-
the partnership track. Most departures during ables (see Raudenbush and Bryk 2002).
the first three years of the partnership track are The dependent variable in our multivar-
probably truly voluntary, motivated by a desire iate models is a binary indicator, coded “1” if
to pursue other careers. Firms generally do not a woman was selected for the position and “0” if
encourage associates to leave before the end of a man was selected. The principal independent
the third year, because they begin to be profit- variables are two binary indicators, coded “1” if
able for firms at about that point (Ginsburg and the selection decision was a partner promotion or
896 • Part VIII: Gender Inequality

Figure 103.1. Gender Composition of Law Students and Lawyers Across Law Firm Establishments,
1994–1995

a lateral partner hire, respectively, and “0” if it (see Epstein 1993). Other control variables in-
was an entry-level hire. clude the starting salary offered to entry-level
We control for the percentage of candidates in associates, adjusted for local labor market varia-
the relevant pool who are female, which we cal- tion, and indicators of whether or not the office
culate using the numbers of women and men exhibited a strong commitment to charitable “pro
discussed in the preceding section. We also in- bono” work, the presence or absence of a two-tier
clude the number of vacancies for the relevant partnership (including both equity-holding and
type of selection decision (entry-level hire, part- non-equity-holding partners) in an establish-
ner promotion, partner hire) and the number ment, and geographical region. Organizational
of female departures per vacancy. Three dichoto- size is tapped by both establishment size and
mous variables represent minimum client-billable whether the establishment is part of a national
hours requirements of up to 1,800 hours per year, or international firm. Finally, we include the per-
1,801 to 1,900 hours per year, and more than centage of women among existing partners in the
1,900 hours per year, respectively (coded “1” if fall of 1994.
yes, “0” if no; the reference category is no min-
imum billable hours requirement). Two other
binary variables measure whether an establish- Results
ment permits part-time work policy or offers paid In our sample women constitute 43 percent of
family-related leave. We also control for the pro- the pool of students at the law schools where the
portion of each establishment’s lawyers working average law firm office recruited new employees
in each of three broad areas that are sometimes (see Figure 103.1). In the average establishment,
perceived as gender-typed: litigation (mascu- women’s representation drops slightly to 39
line); business transactions law (masculine); and percent among newly hired entry-level associ-
people-oriented fields, such as family law, trusts ates and stays at 39 percent among all associ-
and estates law, and employment law (feminine) ates. Among newly promoted partners, women’s
 Gorman and Kmec—Glass Ceilings in Corporate Law Firms • 897

Table 103.2. Women’s and Men’s Mobility Rates in Law Firm Establishments, 1994–1995

25th 75th
Mean S.D. Min. Percentile Median Percentile Max. N

Entry-level
hiring
Women .002 .002 0 .000 .001 .003 .027 715
Men .002 .002 0 .001 .001 .003 .040 715

Partnership
promotion
Women .334 .397 0 0 0 .691 1 457
Men .584 .342 0 .300 .593 .984 1 457
Lateral partner
hiring
Women .009 .056 0 0 0 .002 1 475
Men .005 .011 0 .000 .001 .004 .107 475
Note: Rates represent the ratio of the number of women or men hired or promoted in 1994–1995 to the number of men or
women, respectively, in the appropriate candidate pool as described in the text.

presence declines noticeably to 25 percent, and it to male promotion rates is at its mean, the fe-
then falls further, to only 14 percent of the entire male rate is only half the male promotion rate.
partnership. Finally, in an office where the ratio of female to
Table 103.2 compares women’s and men’s male external partner hiring rates is at its mean,
mobility rates across offices in the sample. the female rate is approximately three-quarters
Entry-level hiring rates for both women and men of the male rate. The lower panel examines ra-
are quite low, which is to be expected given the tios of mobility rates in the smaller subsamples
large pools of potential candidates. However, it of establishments where both the men’s and
is clear that women’s and men’s entry-level hiring women’s selection rates were greater than zero.
rates are approximately equal in the average law Among establishments that hired both men and
firm office. In contrast, the average promotion women into entry-level positions, when the ratio
rate for women is only a little more than half of of female to male entry-level hiring rates is at
the average promotion rate for men. Moreover, is mean, the women’s rate actually exceeds the
more than half of the establishments in the sam- men’s rate by almost one-third. Among estab-
ple promoted no women at all, although more lishments that promoted both women and men
than three-quarters promoted at least one man. to partnership, the average female-to-male ratio
The average partner hiring rate for women is ac- of promotion rates still slightly favors women,
tually higher than that for men, but this average but it is lower than the entry-level ratio. Finally,
is pulled up by a few establishments with espe- in the average establishment that hired both
cially high female hiring rates (more than half of women and men as partners, the ratio is sub-
the sample hired no female partners at all). stantially higher than the entry-level hiring ratio.
Ratios of women’s and men’s mobility rates These results show an increasing-disadvantage
in the same establishment are presented in Table pattern across the sample of law firm offices with
103.3. The upper panel reports ratios based on respect to promotion: the ratio of the probability
the subsamples of offices where at least one man of women’s mobility, relative to men’s, declines
was selected. The mean ratio of female to male as the hierarchical level increases. With respect
entry-level hiring rates is .997, indicating that to external partner hiring, we observe a weak
the two rates are approximately equal in the av- increasing-disadvantage pattern across the sam-
erage office. In contrast, when the ratio of female ple if establishments that hired no women are
898 • Part VIII: Gender Inequality

Table 103.3. Ratios of Women’s and Men’s Mobility Rates in the Same Establishment

Percent
25th 75th Where
Mean Min. Percentile Median Percentile Max. Ratio < 1 N
Based on subsamples
where men’s rates > ,
Ratio of women’s to
men’s rates of:
Hiring into entry-level
.997 0 .185 .701 1.328 10.357 61 648
positions
Promotion into partner
.495 0 0 0 .997 5.125 75 414
positions
Hiring into partner
.775 0 0 0 0 14.413 82 428
positions

Based on subsamples
where both men’s
and women’s rates > ,
Ratio of women’s to
men’s rates of:
Hiring into entry-level
1.318 .164 .617 1.034 1.499 10.357 49 490
positions
Promotion into partner
1.115 .241 .677 1.000 1.335 5.125 44 184
positions
Hiring into partner
3.684 .368 1.751 3.018 5.230 14.413 12 90
positions

included; this pattern is reversed if we examine increases as they move higher in organizational
only establishments that hired both men and hierarchies. Fundamental processes—reliance
women. on gender as a proxy for competence, use of
Turning to our multilevel logistic regression sex-labeled roles and gender stereotypes as
models (see Table 103.4), we find that the prob- heuristics to assess candidate suitability, and
ability that a woman will be selected is consider- in-group favoritism—lead decision makers to
ably less in the case of partner promotions than prefer men over women in selection decisions
in the case of entry-level hires (Model 1). This at all levels. The high status, work uncertainty,
negative effect remains significant when the sex and traditional male domination of upper-level
composition of the candidate pool is controlled positions only intensify these gender biases.
(Model 2) and other organizational characteris- In addition, the cumulative effects of differen-
tics are added (Model 3), although its magnitude tial developmental opportunities can produce
diminishes somewhat. In contrast, although the real gender differences in skills among candi-
coefficient for external partner hires is negative in dates for higher-level positions. The resulting
the model without controls (Model 1), includ- increasing-disadvantage pattern is most likely
ing the sex composition of the candidate pool to occur in large organizations, on identifiable
causes the coefficient to become significantly pos- job ladders, and where entry-level positions are
itive (Model 2). This effect remains positive and relatively sex integrated. However, it should not
significant when establishment-level controls are occur in external upper-level hiring if external
added (Model 3). female candidates already hold similar positions
in comparable organizations. Empirical analysis
supports our ideas, revealing in the setting of
Discussion and Conclusion corporate law firms an increasing-disadvantage
There are theoretical reasons to expect that pattern in internal promotion and the absence
women’s disadvantage in organizational mobility of such a pattern in external upper-level hiring.
 899

Table 103.4. Multilevel Logistic Models of the Probability of Selecting a Woman

Variables Model 1 Model 2 Model 3

Odds Odds Odds
Coeff. Coeff. Coeff.
Ratio Ratio Ratio
-.410*** -.843*** -.950***
Intercept
(.034) (.056) (.166)
Selection decision type
(reference: entry-level
hire)
-.662*** -.238** -.245**
Partner promotion .52 .79 .78
(.072) (.090) (.095)
-1.389*** .527* .500*
Partner hire .25 1.69 1.65
(.090) (.214) (.220)
Decision-level controls
.065*** .064***
Percentage female in candidate pool 1.07
(.007) (.007) 1.07
-.002 .004
Number of vacancies 1.00 1.00
(.003) (.005)
Number of female departures .036 .039
1.04 1.04
per vacancy (.055) (.057)
Establishment-level
controls
Extent of practice in:
.071
Business law 1.07
(.202)
.073
Litigation 1.08
(.231)
-.268
People-oriented law .77
(.346)
Part-time availability:
(reference: no part-time)
.267*
Available case by case 1.31
(.137)
.249†
Available to all 1.28
(.136)
-.043
Family leave policy .96
(.064)
Minimum billable hours
requirement:
(reference: no minimum)
.164†
< 1,800 hours 1.19
(.085)
.040
≥ 1,800 hours but < 1,900 hours 1.04
(.079)
-.097
≥ 1,900 hours .91
(.106)
Starting salary less city mean -.016†
.98
(in thousands $US) (.008)
.126*
Pro bono emphasis 1.13
(.063)
900 • Part VIII: Gender Inequality

Within the scope conditions we identify, we Psychology, edited by D. T. Gilbert, S. T. Fiske,

expect that the same patterns hold in organi- and G. Lindzey, 554–594. New York: Oxford
zations with more complex structures, such as University Press.
business corporations. The substantive processes Bruderl, J., P. Preisendorfer, and R. Ziegler. 1993.
“Upward Mobility in Organizations: The Effects
we describe are not unique to any organizational
of Hierarchy and Opportunity Structure.” Euro-
form or industry. Our study also makes a meth- pean Sociological Review 9: 173–188.
odological contribution to research on inequality Cohen, Lisa E., Joseph P. Broschak, and Heather A.
in hiring and promotion by demonstrating the Haveman. 1998. “And Then There Were More?
importance of accounting for the sex composi- The Effect of Organizational Sex Composition
tion of the relevant candidate pool to avoid bi- on the Hiring and Promotion of Managers.”
ased estimates of the gender effect. American Sociological Review 63: 711–727.
Our theory and findings are at odds with Creswell, Julie. 2006. “How Suite It Isn’t: A Dearth
the view of some scholars that discrimination of Female Bosses.” New York Times, December
17, B1.
is constant across organizational levels and that
Davis, Wendy. 2000. “Flight Leads to New Look at
variations in women’s mobility disadvantage are Pyramid.” New York Law Journal, June 22, 1.
due primarily to variations in decision makers’ Dencker, John. 2008. “Corporate Restructuring
opportunities to discriminate, which are said and Sex Differences in Managerial Promotion.”
to be greatest at the entry level. In contrast, we American Sociological Review 73: 455–476.
contend that bias is more severe in upper-level DiPrete, Thomas A., and Whitman T. Soule. 1988.
selection decisions. Our findings have clear pol- “Gender and Promotion in Segmented Job Lad-
icy implications: efforts to recruit women into der Systems.” American Sociological Review 53:
the lower rungs of job ladders will not, by them- 26–40.
Dodge, Katherine, Faith Gilroy, and L. Mickey
selves, eliminate women’s underrepresentation at
Fenzel. 1995. “Requisite Management Charac-
the top. To shatter the glass ceiling, organizations
teristics Revisited: Two Decades Later.” Journal
must adopt practices focused on eliminating the of Social Behavior and Personality 10: 253–264.
disadvantages women experience as they attempt Dovidio, John F., Clifford E. Brown, Karen Helt-
to climb upward after the point of hire. mann, Steve L. Ellyson, and Caroline F. Keating.
1988. “Power Displays between Women and
REFERENCES Men in Discussions of Gender-Linked Tasks: A
Althauser, Robert P. 1989. “Internal Labor Markets.” Multichannel Study.” Journal of Personality and
Annual Review of Sociology 15: 143–161. Social Psychology 55: 580–587.
Baron, James N., Alison Davis-Blake, and William Eagly, Alice, and Steven Karau. 2002. “Role Congru-
T. Bielby. 1986. “The Structure of Opportunity: ity Theory of Prejudice Toward Female Leaders.”
How Promotion Ladders Vary within and among Psychological Review 109: 573–598.
Organizations.” Administrative Science Quarterly Ely, Robin J. 1995. “The Power in Demography:
31: 248–273. Women’s Social Constructions of Gender Iden-
Baxter, Janeen, and Erik Olin Wright. 2000. “The tity at Work.” Academy of Management Journal
Glass Ceiling Hypothesis: A Comparative Study 38: 589–634.
of the United States, Sweden, and Australia.” Epstein, Cynthia Fuchs. 1993. Women in Law. Ur-
Gender and Society 14: 275–294. bana: University of Illinois Press.
Beckman, Christine, and Damon Phillips. 2005. Federal Glass Ceiling Commission. 1995a. Good for
“Interorganizational Determinants of Promo- Business: Making Full Use of the Nation’s Human
tion: Client Leadership and the Attainment of Capital; The Environmental Scan: A Fact-Finding
Women Attorneys.” American Sociological Review Report of the Federal Glass Ceiling Commission.
70: 678–701. Washington, DC: US Government Printing
Berger, Joseph, Susan Rosenholtz, and Morris Zeld- Office.
itch. 1980. “Status Organizing Processes.” An- ———. 1995b. A Solid Investment: Making Full Use
nual Review of Sociology 6: 479–508. of the Nation’s Human Capital: Recommendations
Brewer, Marilynn, and Rupert Brown. 1998. “In- of the Federal Glass Ceiling Commission. Washing-
tergroup Relations.” In Handbook of Social ton, DC: US Government Printing Office.
 Gorman and Kmec—Glass Ceilings in Corporate Law Firms • 901

Fiske, Susan T., Amy J. C. Cuddy, Peter Glick, and Executive Suite: Much Ado about Something.”
Jun Xu. 2002. “A Model of (Often Mixed) Ste- Journal of Social Behavior and Personality 13:
reotype Content: Competence and Warmth 127–138.
Respectively Follow from Perceived Status and Martindale-Hubbell Inc. 1995. The Martindale-
Competition.” Journal of Personality and Social Hubbell Law Directory. New York: Martindale-
Psychology 82: 878–902. Hubbell, Inc.
Fiske, Susan T., Monica Lin, and S. L. Neuberg. ———. 1996. The Martindale-Hubbell Law Direc-
1999. “The Continuum Model: Ten Years Later.” tory. New York: Martindale-Hubbell, Inc.
In Dual-Process Theories in Social Psychology, ed- NALP Foundation for Research and Education.
ited by S. Chaiken and Y. Trope, 231–254. New 1998. Keeping the Keepers: Strategies for Associate
York: Guilford Press. Retention in Times of Attrition. Washington, DC:
Ginsburg, Tom, and Jeffrey Wolf. 2003. “The Mar- NALP Foundation for Law Career Research and
ket for Elite Law Firm Associates.” Florida State Education.
University Law Review 31: 909–963. National Association for Law Placement. 1996. Na-
Gorman, Elizabeth. 2005. “Gender Stereotypes, tional Directory of Legal Employers. Chicago: Har-
Same-Gender Preferences, and Organizational court Brace Legal & Professional Publications.
Variation in the Hiring of Women: Evidence Noonan, Mary, and Mary Corcoran. 2004. “The
from Law Firms.” American Sociological Review Mommy Track and Partnership: Temporary
70: 702–728. Delay or Dead End?” Annals of the American
———. 2006. “Work Uncertainty and the Promo- Academy of Political and Social Science 596:
tion of Professional Women: The Case of Law 130–150.
Firm Partnership.” Social Forces 85: 864–890. Padavic, Irene, and Barbara Reskin. 2002. Women
Heilman, Madeline E. 1983. “Sex Bias in Work and Men at Work. Thousand Oaks, CA: Pine
Settings: The Lack of Fit Model.” In Research in Forge Press.
Organizational Behavior, edited by B. M. Staw Paulin, Elizabeth, and Jennifer Mellor. 1996. “Gen-
and L. L. Cummings, 5:269–298. Greenwich, der, Race, and Promotions within a Private-Sector
CT: JAI Press. Firm.” Industrial Relations 35: 276–295.
Heilman, Madeline E., Richard F. Martell, and Perry, Elissa L., Allison Davis-Blake, and Carol
Michael C. Simon. 1988. “The Vagaries of Sex Kulik. 1994. “Explaining Gender-Based Selec-
Bias: Conditions Regulating the Undervaluation, tion Decisions: A Synthesis of Contextual and
Equivaluation, and Overvaluation of Female Job Cognitive Approaches.” Academy of Management
Applicants.” Organizational Behavior and Human Review 19: 786–820.
Decision Processes 41: 98–110. Petersen, Trond, and Ishak Saporta. 2004. “The Op-
Hilton, J. L., and W. von Hippel. 1996. “Stereo- portunity Structure for Discrimination.” Ameri-
types.” Annual Review of Psychology 47: 237–271. can Journal of Sociology 109: 852–901.
Kanter, Rosabeth Moss. 1977. Men and Women of the Phillips, Damon, and Jesper Sorensen. 2003. “Com-
Corporation. New York: Basic Books. petitive Position and Promotion Rates: Com-
Konrad, A. M., J. E. Ritchie, P. Lieb, and E. Corri- mercial Television Station Top Management,
gall. 2000. “Sex Differences and Similarities in 1953–1988.” Social Forces 81: 819–842.
Job Attribute Preferences: A Meta-analysis.” Psy- Powell, Gary N., and D. Anthony Butterfield. 1994.
chological Bulletin 126: 593–641. “Investigating the ‘Glass Ceiling’ Phenomenon:
Konrad, Alison, and Jeffrey Pfeffer. 1991. “Under- An Empirical Study of Actual Promotions to Top
standing the Hiring of Women and Minorities.” Management.” Academy of Management Journal
Sociology of Education 70: 151–173. 37: 68–88.
Lewis, Gregory. 1986. “Gender and Promotions: Pugh, Meredith D., and Ralph Wahrman. 1983.
Promotion Chances of White Men and Women “Neutralizing Sexism in Mixed-Sex Groups: Do
in Federal White-Collar Employment.” Journal of Women Have to Be Better Than Men?” American
Human Resources 21: 406–419. Journal of Sociology 88: 746–762.
Lucas, Jeffrey. 2003. “Status Processes and the Insti- Raudenbush, Stephen, and Anthony Bryk. 2002.
tutionalization of Women as Leaders.” American Hierarchical Linear Models: Applications and
Sociological Review 68: 464–480. Data Analysis Methods. Thousand Oaks, CA:
Martell, Richard F., C. Parker, C. G. Emrich, and Sage Publications.
M. S. Crawford. 1998. “Sex Stereotyping in the
902 • Part VIII: Gender Inequality

Reskin, Barbara F. 1988. “Bringing the Men Back Rosenbaum, James. 1979. “Tournament Mobility:
In: Sex Differentiation and the Devaluation of Career Patterns in a Corporation.” Administrative
Women’s Work.” Gender and Society 2: 58–81. Science Quarterly 24: 220–241.
Reskin, Barbara F., and Debra B. McBrier. 2000. Spilerman, Seymour, and Trond Petersen. 1999.
“Why Not Ascription? Organizations’ Employ- “Organizational Structure, Determinants of Pro-
ment of Male and Female Managers.” American motion, and Gender Differences in Attainment.”
Sociological Review 65: 210–233. Social Science Research 28: 203–227.
Ridgeway, Cecilia. 1997. “Interaction and the Con- Spurr, Stephen J. 1990. “Sex Discrimination in the
servation of Gender Inequality: Considering Legal Profession: A Study of Promotion.” Indus-
Employment.” American Sociological Review 62: trial and Labor Relations Review 43: 406–417.
218–235. Thomas, David A., and John J. Gabarro. 1999.
Ridgeway, Cecilia, and Shelley Correll. 2004. “Un- Breaking Through: The Making of Minority Ex-
packing the Gender System: A Theoretical Per- ecutives in Corporate America. Boston: Harvard
spective on Gender Beliefs and Social Relations.” Business School Press.
Gender & Society 18: 510–531. Wagner, David, Rebecca S. Ford, and Thomas W.
Ford. 1986. “Can Gender Inequalities Be Re-
duced?” American Sociological Review 51: 47–61.

104 • Maria Charles and David B. Grusky

Egalitarianism and Gender Inequality

The rise of egalitarian values and associated the forces for egalitarianism have been resisted,
egalitarian institutional reforms is a distinctive sometimes violently (as with recurrences of eu-
feature of modernity and postmodernity. This genics and fascism) and sometimes in quieter
development, which dates at least to the En- but still profound ways (as in the persistence of
lightenment, intensified throughout the twenti- residential segregation).
eth century as formal legal rights were extended This chapter is about one of those quieter
to previously excluded groups (e.g., women), anti-egalitarian forces. Although it is often taken
wide-reaching institutional reforms were imple- for granted, it is arguably striking that women
mented to equalize life chances (e.g., bureau- and men continue to work in very different oc-
cratic personnel policies), and anti-egalitarian cupations, with women typically crowding into
doctrines (e.g., racism) were challenged. These a relatively small number of historically female
processes of equalization, dramatic though they occupations (e.g., teacher, secretary, nurse). If
are, obviously do not exhaust the story of mo- one sought, for example, to undo all sex segrega-
dernity and postmodernity. As is well known, tion, a full 52 percent of the employed women
this story is replete with counterpoints at which in the United States would have to be shifted

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in the
following publication: Maria Charles and David B. Grusky, Occupational Ghettos: The Worldwide Segregation
of Women and Men, published by Stanford University Press in 2004. The article printed here was originally
prepared by Maria Charles and David B. Grusky for The Inequality Reader: Contemporary and Foundational
Readings in Race, Class, and Gender, Second Edition, edited by David B. Grusky, pp. 389–404. Copyright ©
2011 by Westview Press.
 Charles and Grusky—Egalitarianism and Gender Inequality • 903

out of their current occupational categories There is, to be sure, clear and substantial ev-
(Jacobs 2003). This extreme sex segregation is idence of desegregation in many sectors of the
typical of what prevails throughout the advanced occupational structure, yet any careful observer
industrial world. Because sex segregation is so of this process has to be struck by its uneven-
extreme, and because it colors the life chances ness and by the persistence, in particular, of
and life experiences of so many women and men, many female occupational ghettos (e.g., secre-
we characterize the contemporary occupational tary, nursery school teacher). Moreover, when
structure as “hypersegregated” (see Massey and male-dominated occupations embark on what
Denton 1993 for a related usage). appears to be integrative trajectories, the influx
Why is the occupational structure so resistant of women often continues well past the point
to egalitarian forces? It could be argued that fun- of gender parity and ultimately creates a new
damental institutional change is inevitably pro- female-dominated ghetto (e.g., Reskin and Roos
longed and that full integration will ultimately 1990). These results suggest that sex segregation,
be achieved through ongoing reform efforts far from being a holdover, is actively advanced
(Jackson 1998). Although we cannot rule out by dynamics that are part and parcel of modern
the possibility of full integration in the distant industrialism.
future, we would stress that this outcome is by We therefore present the following puzzle:
no means inevitable under prevailing policies, Why has sex segregation proven resistant to egal-
practices, and commitments. That is, rather itarian pressures even as other forms of gender
than viewing sex segregation as a residual that inequality have given way? Although we return
is destined to wither away under contempo- to this puzzle repeatedly, it is important to recog-
rary egalitarian pressures, it is best regarded as nize that it is but one part of a larger complex of
an organic feature of modern economies that is perplexing findings that scholars of segregation
ideologically consistent with egalitarianism, at have recently reported. The following two puz-
least as the latter is understood and practiced zles might, in particular, be cited: (a) the com-
today. In this sense, there is a deep structure to mon view that male power and privilege allows
sex segregation that makes it a viable long-term men to dominate the best occupations is partly
feature of modern economies, even as pressures inconsistent with the typical pattern of contem-
for equalization mount in other domains of the porary sex segregation; and (b) the highest levels
social stratification system. of segregation are often found in socially and
It is instructive in this regard to ask whether culturally progressive countries (such as Swe-
countries that have committed most explicitly den) rather than in their more traditional coun-
to family-friendly policies, antidiscrimination terparts (such as Japan). Taken together, these
legislation, and other forms of egalitarianism findings seem to suggest a topsy-turvy world in
have made substantial headway in reducing sex which males do not straightforwardly dominate
segregation. If they have, such progress suggests the best jobs, family-friendly policies have a per-
that conventional egalitarian commitments, at verse segregating effect, and contemporary gen-
least when carried out to their logical extremes, der regimes continue to have a highly segregated
can serve to more quickly root out residual “1950s feel” even in the twenty-first century.
segregative processes. The available evidence is We argue here that these empirical puzzles
largely disappointing on this front. For exam- have emerged in the literature because stratifi-
ple, Sweden is well known for its egalitarian and cation scholars tend to treat sex segregation in
family-friendly policies, yet it remains deeply sex unidimensional terms and accordingly fail to
segregated to the present day. Conversely, coun- appreciate that a complex amalgam of processes
tries that are commonly viewed as bastions of underlies gender inequality and renders some
conservative gender practice, such as Japan and forms of segregation more entrenched than oth-
Italy, are not any more segregated than less con- ers. Although the tendency to represent segrega-
servative countries. tion unidimensionally is widespread, it emerges
904 • Part VIII: Gender Inequality

especially clearly in classical theorizing about the best occupations will be dominated by men,
long-term trends in inequality. For example, either because women have disproportionate
structural-functional theorists typically treat all domestic responsibilities that reduce their incen-
forms of “ascription,” including sex segregation, tive to invest in demanding careers (e.g., Becker
as a generic residue destined to wither away either 1991), or because employers practice discrimi-
because discriminatory practices are inefficient nation through “glass ceiling” personnel policies
or because bureaucratic forms of social organi- and other forms of male-biased queuing in the
zation have diffused widely and served to un- labor market (e.g., Reskin and Roos 1990).
dermine discriminatory practices (e.g., Parsons This conventional account falls short because
1970). In similar fashion, neoinstitutionalists it fails to appreciate the distinction between ver-
assume that egalitarian practices and organiza- tical and horizontal forms of segregation and,
tional forms will gradually diffuse and generate in particular, the interaction between these two
across-the-board reductions in segregation, al- forms (see Blackburn, Jarman, and Brooks 2000;
though the main impetus for such diffusion is Semyonov and Jones 1999). The model in Figure
not so much the intrinsic efficiency of univer- 104.1, which underlies all our analysis, builds
salistic practices as the characteristically mod- explicitly on this distinction: the “nonmanual
ern commitment to cultural stories about their slope parameter” governs the extent to which
efficiency (e.g., Meyer 2001). Finally, some early men dominate the most desirable classes in the
feminist scholars (e.g., Huber 1988) conceptu- nonmanual sector; the “manual slope parame-
alized occupational sex segregation as one of the ter” governs the extent to which men dominate
main outcomes of “patriarchal” forms of social the most desirable classes in the manual sector;
organization, again implying that the fate of sex and the “horizontal gap parameter” governs the
segregation is simply a function of the larger fate extent to which men are disproportionately al-
of patriarchal social relations that “create solidar- located into the manual sector rather than the
ity and interdependence among men and enable nonmanual one.
them to dominate women” (Hartmann 1981, The horizontal axis of this figure arrays the
p. 14). This approach typically treats both pa- nine major occupational categories defined by
triarchy and inequality in monolithic terms and the International Labour Office (ILO) on an
thus draws scholars into weaving stories about approximate socioeconomic scale ranging from
the extent of segregation rather than its many high (professional) to low (agricultural). Follow-
dimensions and their different responsiveness to ing convention, the first five categories in this
egalitarian forces. list may be characterized as nonmanual, and the
We argue, then, that various puzzles have second four may be characterized as manual. The
emerged in the literature because conventional vertical axis of this figure, labeled “female repre-
theories and methods blind us to the multidi- sentation,” indicates the extent to which women
mensional structure of segregation. By advanc- or men are overrepresented in each of these nine
ing a two-dimensional conceptualization of sex categories. In the interior of the figure, the slopes
segregation and a matching methodological ap- of the two lines reveal the strength of vertical
proach, we seek to solve the puzzles and build an segregation, with a steep positive slope indicating
alternative understanding of the development of that men are much advantaged in the competi-
segregation regimes. tion for desirable occupations, a moderate pos-
itive slope indicating that men are only weakly
advantaged in this competition, and a negative
A New Multidimensional Model slope (which is logically possible but empirically
It is useful to begin by asking whether the un- unlikely) indicating that women are advantaged.
derlying structure of sex segregation is consistent The extent of horizontal segregation (the “hor-
with unidimensional accounts of segregation. In izontal gap parameter”) is given by the vertical
many such accounts, it is simply presumed that distance between the manual and nonmanual
 Charles and Grusky—Egalitarianism and Gender Inequality • 905

Figure 104.1. Anatomy of a Hypothetical Sex-Segregation Regime

Female Representation

Nonmanual
slope parameter

Horizontal gap Manual slope

parameter parameter
Professional

Manager

Associate
Professional

Clerical

Ser vice/
Sales

Craft

Operative

Agriculture

Laborer
Note: In our models, the major occupational categories are coded in terms of a socioeconomic scale,
thus allowing the intercategory distances to vary freely. For the purpose of simplifying the presentation,
we have assumed here that intercategory distances are the same.

lines.1 The foregoing three-parameter specifica- segregation are fundamentally two-dimensional.

tion, which serves to summarize the aggregate Under the parameterization of Figure 104.1, it is
structure of segregation, allows us to resolve assumed that (a) the cultural tenet of male pri-
long-standing empirical puzzles in the field that macy undergirds vertical segregation, and (b) the
arose because most scholars have defaulted to a complementary cultural tenet of gender essen-
unidimensional view. tialism undergirds horizontal segregation. The
The two vertical slope parameters in Figure tenet of male primacy represents men as more
104.1 are partly consistent with a unidimen- status-worthy than women and better suited for
sional queuing formulation (whereby men se- positions of authority and domination, and the
cure better occupations than women), but our tenet of gender essentialism represents women
specification may be understood as a revision of as more competent than men in personal ser-
this formulation because queuing theory does vice, nurturance, and social interaction. In the
not allow for a “horizontal gap” expressing the modern context, these two cultural tenets tend
tendency of women to be disproportionately al- to coexist with one another, thus giving segrega-
located to the nonmanual sector (even though tion systems a hybrid character. We review these
nonmanual occupations are, on average, more two tenets below.
desirable than manual ones). Moreover, we allow
the vertical principle to be stronger in some re-
gions of the labor market (i.e., the manual sec- Essentialism
tor) than in others, and we also allow the vertical Why do women crowd into the nonmanual sec-
principle to be stronger at the aggregate level tor and men crowd into the manual sector? In
than at the level of detailed occupations (which addressing this question, one has to be struck
are not represented in Figure 104.1). Although by the strong correspondence between (a) the
queuing theory thus motivates some aspects of traits that are regarded as distinctively male or
our parameterization, it cannot provide a com- female (gender essentialism) and (b) the task
plete account of segregation (see Reskin and requirements of manual and nonmanual labor.
Roos 1990; Strober 1984). Although prevailing characterizations of male
The deficiencies of queuing theory and and female traits are complex and multifac-
other unidimensional formulations arise in eted, a core feature of such characterizations is
large part because the cultural underpinnings of that women are presumed to excel in personal
906 • Part VIII: Gender Inequality

service, nurturance, and interpersonal interac- there persist deeply rooted and widely shared
tion, and men are presumed to excel in inter- cultural beliefs that men are better suited than
action with things (rather than people) and in women for all forms of labor outside the fam-
strenuous or physical labor. These stereotypes ily, and that men are, in particular, better suited
about natural male and female characteristics than women for labor involving the exercise of
are disseminated and perpetuated through pop- authority and power (Deaux and Kite 1987;
ular culture and media, through social interac- Ridgeway 1997). We argue, then, that vertical
tion in which significant others (parents, peers, segregation is maintained and reproduced in part
teachers) implicitly or explicitly support such because it is consistent with the cultural value of
interpretations, and through micro-level cogni- male primacy. The main proximate mechanisms
tive processes in which individuals pursue and by which beliefs in male primacy are translated
remember evidence consistent with their preex- into vertical segregation are (a) the recognition
isting stereotypes and ignore, discount, or for- among men that, because they are regarded as
get evidence that undermines those stereotypes primary breadwinners, they should make sub-
(Fiske 1998; Reskin 2000). The main claim we stantial investments in human capital (i.e.,
wish to advance is that horizontal segregation is supply-side sources), and (b) the recognition
maintained and reproduced in large part because among employers that, because men are regarded
nonmanual occupations embody characteristics as primary breadwinners, their commitment to
(such as service orientation) regarded as proto- the labor force will be greater and hence there is
typically female, while manual occupations em- a greater payoff to investing in them rather than
body characteristics (such as strenuousness and in women (who may exhibit intermittency).
physicality) regarded as prototypically male (see, Moreover, employers reward men with better
e.g., Crompton 2001; Lorber 1993; Milkman jobs not just because they assume that men
and Townsley 1994; Tilly 1998). This linkage have a greater commitment to the labor force,
is converted into durable horizontal segregation but also because they regard men as intrinsically
because (a) employers internalize these essen- more competent. Because such assumptions
tialist presumptions and allocate occupations about intrinsic competency are internalized by
to men and women in accord with them (i.e., everyone, male workers will also come to believe
essentialist discrimination), and (b) workers that they are more competent at high-status tasks
internalize the same essentialist presumptions than females and hence more likely to succeed
and aspire to occupations that satisfy them (i.e., in those tasks, thereby motivating them to make
essentialist preferences). It follows that hori- the requisite investments in human capital. The
zontal segregation has both demand-side and premise of male primacy, like that of gender es-
supply-side sources. sentialism, is therefore expressed in supply-side
as well as demand-side processes.

Male Primacy
Why are men disproportionately allocated to The Dynamics of Gender Segregation
the best-paid and most desirable occupations in This two-dimensional understanding of seg-
both the nonmanual and manual sectors? In ac- regation casts light on the processes by which
counting for such vertical segregation, we again the spread of egalitarian commitments will af-
understand the main forces at work as being fect the structure of gender inequality. Within
cultural in form, but now the relevant cultural the cultural domain, the diffusion of egalitari-
principle is the long-standing belief that men anism is an extremely important development,
are more status-worthy than women and accord- one that will likely continue apace unless some
ingly better suited for positions of high pay or unforeseen catastrophic event has a recalibrat-
authority. Despite the rise of universalistic ideals, ing effect. Although the future of egalitarianism
 Charles and Grusky—Egalitarianism and Gender Inequality • 907

Table 104.1. Data and Sample Characteristics

Country Census Year Sample Size

Belgium 1991 3,418,512
France 1990 900,255
West Germany 1993 128,912
Italy 1991 21,071,282
Portugal 1991 4,037,130
Sweden 1990 4,059,813
Switzerland 1990 3,076,445
United Kingdom 1991 2,405,091
United States 1990 1,152,885
Japan 1990 12,220,974

appears bright, one should consider the limits identities will not be scrutinized or challenged
of the particular version of egalitarianism that to the extent that they would be under more
has taken hold and that continues to diffuse. radical egalitarian commitments. On the de-
Among the many competing egalitarian vi- mand side, the liberal egalitarian commitment
sions, it is clear that “liberal egalitarian” strands delegitimates all forms of pure discrimination,
remain dominant, implying that our collective but it does not as directly challenge statistical
commitment to gender equality mainly takes discrimination that rests on essentialist pre-
the form of developing procedural guarantees sumptions. In a world in which women have
of equal opportunity. This commitment to lib- disproportionately “invested” in nurturance and
eral egalitarianism is quite compatible with the service, essentialist presumptions about gender
essentialist presumption that men and women differences in capabilities have ample room to
have fundamentally different tastes, skills, and flourish, and employers may well reason that
abilities (see Charles and Grusky 2004). That is, gender provides a good signal of capabilities
the liberal egalitarian vision of women and men in nurturing and service. The main argument
as autonomous agents entitled to equal rights, we would make, then, is that liberal variants of
opportunities, and treatment allows for the per- egalitarianism serve principally to undermine the
sistence of fundamentally gendered outlooks and presumption of male primacy rather than gen-
identities. For a liberal egalitarian, it is enough to der essentialism; and, consequently, horizontal
defend the right of women to fairly compete for forms of segregation may prove to be quite resis-
any occupation to which they aspire, without in tant to egalitarianism.
any way questioning how those aspirations were
formed or why they may differ from the aspira-
tions of men. It follows that liberal egalitarians Is the Argument Supported?
embrace an “equal but different” conceptualiza- This argument can be addressed with the inter-
tion of gender and social justice. national data archive described in Table 104.1.
Insofar as this version of egalitarianism con- We apply a cross-nationally harmonized clas-
tinues to diffuse, the push toward complete sification of sixty-four detailed occupational
equality may be slowed. This suppressive effect categories to census segregation data from ten
occurs through proximate mechanisms on both industrial market economies. The data are or-
the supply and demand sides. On the supply ganized in the form of a three-dimensional,
side, we cannot expect liberal egalitarians to pay 1,280-cell matrix with sixty-four occupations,
much attention to individual aspirations and two sexes, and ten countries. For most of our
self-assessments, meaning that the persistence analyses, we examine segregation principally in
of gender differences in these outlooks and terms of nine major classes, each of which is an
908 • Part VIII: Gender Inequality

aggregation of a subset of the sixty-four detailed percent of the total association at the aggregate
occupations in the full array. We report results level can be explained in vertical terms. More-
based on “self-weighted” data in which the actual over, our estimate of İ under this model is .008,
sample size in each country is preserved. Because meaning that female representation increases by
large-sample countries (such as Italy and Japan) a factor of 1.008 for each unit increase in so-
have much leverage on our estimates, we have cioeconomic status. This estimate, which indi-
reestimated many of our models after standard- cates that women are (slightly) overrepresented
izing sample sizes to an arbitrary constant (N in high-status occupations, is of course incon-
= 10,000), but such standardized results have sistent with simple queuing perspectives (see
proved to be much the same as self-weighted Blackburn, Brooks, and Jarman 2001; Charles
results and therefore will not be reported here. and Grusky 1995; Roos 1985). Although such
We have measured horizontal inequality at a result is counterintuitive, it is consistent with
the aggregate level by distinguishing the five the long-standing argument that socioeconomic
nonmanual categories (managerial, professional, scores overstate the desirability of routine non-
associate professional, clerical, service/sales) manual occupations and hence create the (mis-
from the four manual categories (agriculture, leading) appearance of female advantage (e.g.,
craft, operative, laborer). We measure vertical Acker 1980; England 1979). We have no quar-
inequality with the internationally standardized rels with this long-standing account, but it may
socioeconomic index (SEI) published by Ganze- not be complete. The queuing model fails, as we
boom and Treiman (1996). For the purposes of see it, not only because socioeconomic scales are
our analyses here, it is useful to apply an aggre- flawed or because the vertical dynamic is weak
gate variant of the SEI index (V1), a variant that but also because this dynamic is obscured in the
assigns average SEI values to each of the nine absence of controls for horizontal segregation.
major categories. The structure of segregation is in this sense fun-
The key question that arises is whether the damentally multidimensional.
between-category component may be explained We can test this argument by examining
in vertical terms. In its purest form, a queuing whether the vertical coefficient reverses sign and
model implies that men are disproportionately strengthens in the context of a multidimensional
allocated to the most desirable major occupa- model. When vertical and horizontal effects
tions, thus suggesting the following specification: are simultaneously fit, the following model is
generated:
mijk=ĮkȕikȖjkeİ(Z V1 ),
i j

mijk=ĮkȕikȖjkeİ(Z V1 )+Ȧ(Z H ),
i j i j

where i, j, and k index gender, occupation, and

country respectively, Įk is the grand mean in the where Hj is the horizontal term (i.e., Hj = 1 if j
kth country, ȕik is the country-specific marginal is a manual occupation and Hj = 0 otherwise), Ȧ
effect for the ith gender, Ȗjk is the country-specific is the effect of horizontal status on female repre-
marginal effect for the jth occupation, İ refers sentation, and the remaining terms are defined
to the effect of socioeconomic status on female as before. The explained association under this
representation (at the aggregate level), Zi is an specification increases dramatically from 2.5 to
indicator variable for gender (i.e., Z1 = 0 and Z2 80.4 percent, and the vertical coefficient further
= 1), and V1j refers to the aggregate version of assumes the expected negative sign and becomes
our socioeconomic scale. We have identified this quite strong (–.050). The horizontal coefficient
model by imposing standard constraints on the is likewise very strong: The estimate of Ȧ, – 1.96,
parameters. implies that female representation is 7.10 times
The test statistic for this model, L2 = greater in a nonmanual occupation than a cor-
16,081,116 with 629 df, implies that only 2.5 responding manual occupation of the same
 Charles and Grusky—Egalitarianism and Gender Inequality • 909

socioeconomic status. This effect, which is equiv- arise because egalitarianism has drawn women
alent to that associated with a downward shift into the labor force at precisely the time when
in status of nearly forty points, is surely strong the nonmanual sector is expanding. The “Swed-
enough to suggest that scholars of gender strati- ish puzzle” is solved, therefore, by recognizing
fication should move beyond their long-standing that the logic of egalitarian policy is not incon-
focus on vertical inequality and begin attending sistent with the persistence and even the growth
to horizontal forms of stratification. of horizontal forms of segregation (see Charles
We might also allow the strength of the verti- and Grusky 2004 for details).
cal effect to differ across sectors. This model may
be represented as follows:
Conclusion
mijk=ĮkȕikȖjke ȥ(ZiNjV1j)+İ(ZiMjV1j)+Ȧ(ZiHj)
, The preceding discussion suggests that the fu-
ture of gender inequality rests on a struggle be-
where Nj is an indicator variable for nonman- tween egalitarian and essentialist forces that is
ual occupations (i.e., Nj = 1 if j is a nonmanual not quite as one-sided as modernization theorists
occupation and Nj = 0 otherwise), Mj is an indi- have sometimes claimed (see, e.g., Jackson 1998;
cator variable for manual occupations (i.e., Mj = Parsons 1970). If gender segregation is especially
Hj), and ȥ and İ express the strength of vertical durable, it is partly because it has a deep essen-
segregation within the nonmanual and manual tialist undergirding. The “first revolution” in
sectors respectively.2 With this specification, the gender inequality, which has generated impor-
explained association increases modestly (from tant reductions in segregation over the past thirty
80.4 to 85.3 percent), and the vertical effect is years, has been driven in large part by declines
revealed to be slightly weaker in the nonmanual in vertical inequality (see Weeden 2004). Will
sector (–.040) than in the manual sector (–.047). there be a “second revolution” that leads to an
The modern segregation regime thus takes on analogous decline in horizontal segregation? In
the three-parameter form represented in Figure answering this question, conventional function-
104.1: The horizontal gap parameter captures alist and neoinstitutionalist accounts fall short
the dramatic overrepresentation of women in the because they fail to recognize that horizontal seg-
nonmanual sector, and the two slope parameters regation proceeds from an essentialist ideology
capture the tendency for men to dominate the that can persist—even thrive—in the context of
best occupations within the manual and non- liberal egalitarian norms of equal opportunity.
manual sectors. In the contemporary context, men and
We can now turn to the question of why women are presumed to have rather different
countries with a comparatively deep commit- tastes and aptitudes, and liberal egalitarianism
ment to egalitarianism, such as France, West works merely to ensure that such differences,
Germany, Sweden, the United Kingdom, and however they might be generated, can then be
the United States, have remained quite sex seg- pursued or expressed in a fair (gender-neutral)
regated. If our model is now reestimated within contest. The assumption that men and women
each country in our sample, we find that the have fundamentally different tastes and capac-
average value of the horizontal segregation pa- ities is reinforced in various social settings, not
rameter for these five “egalitarian” countries is just in families (with their gender-specific so-
–3.20, whereas the corresponding average for cialization practices) and work organizations
the five less egalitarian countries in our sample (with their discriminatory hiring practices) but
is –2.07. There is accordingly no evidence that in other institutional contexts as well. By way
egalitarianism reduces the horizontal variant of of (trivial) example, consider the practice among
segregation. To the contrary, it would appear American fast-food restaurants of providing
to increase such segregation, an effect that may gender-specific toys to children, a practice of
910 • Part VIII: Gender Inequality

interest only because it is widely diffused and those forms of inequality that can be represented
evidently unobjectionable to all but a small mi- as consistent with it.
nority of “gender progressives.” If these same
restaurants distributed toys on the basis of racial NOTES
or class standing, the practice would be deemed 1. If the two vertical segregation lines are paral-
lel, the distance between them is of course constant.
absurd at best and racist or classist at worst. This
However, insofar as the slopes of these lines are al-
example suggests that, at least in the United
lowed to differ, the size of the horizontal parameter
States, it is less legitimate to interpret racial or will depend on the implicit zero point of the vertical
class-based inequalities in essentialist terms than scale.
to interpret gender segregation and inequality in 2. Under this specification, the two vertical pa-
these terms. rameters no longer generate parallel lines, meaning
In the long run, it is of course possible that that the estimated size of the horizontal parameter
a yet deeper form of egalitarianism will emerge depends on the implied zero point of the socio-
and delegitimate (a) the tendency of males and economic scale. We have fixed the zero point of
females to develop different tastes, aspirations, this scale at 37 (which is the midpoint between the
score for the lowest nonmanual category [40] and
and market capacities, and (b) the tendency of
the highest manual category [34]). The two vertical
employers to make judgments about productiv- segregation lines are furthest from one another at this
ity through essentialist lenses. There are indeed point (assuming that the nonmanual slope is weaker
many signs that just such a form of egalitarian- than the manual one).
ism is developing. Most notably, conventional
sociological understandings of the roles of social- REFERENCES
ization, social exchange, and power differentials Acker, Joan R. 1980. “Women and Stratification: A
in generating preferences have diffused widely Review of Recent Literature.” Contemporary So-
in contemporary industrial societies, suggesting ciology 9: 25–39.
Becker, Gary S. 1991. A Treatise on the Family. Cam-
that preferences and choices formerly regarded as
bridge, MA: Harvard University Press.
sacrosanct are increasingly treated as outcomes of
Blackburn, Robert M., Jennifer Jarman, and Bradley
unequal and unfair social processes. This deeper Brooks. 2000. “The Puzzle of Gender Segrega-
form of egalitarianism is reflected in attempts by tion and Inequality: A Cross-National Analysis.”
some parents to minimize gender bias in the so- European Sociological Review 16: 119–135.
cialization of their children, at least in the early ———. 2001. “Occupational Stratification: The
years of childrearing before the unremitting in- Vertical Dimension of Occupational Segre-
fluence of societywide essentialism typically un- gation.” Work, Employment, and Society 15:
dermines their efforts. It is surely plausible that 5111–5138.
Charles, Maria, and David B. Grusky. 1995. “Mod-
this deeper form of egalitarianism will ultimately
els for Describing the Underlying Structure of
take hold (see Ramirez 1987, p. 270).
Sex Segregation.” American Journal of Sociology
For our part, we would merely stress that 100: 931–971.
prevailing forms of egalitarianism do not fully ———. 2004. Occupational Ghettos: The Worldwide
delegitimate essentialist processes and that a Segregation of Women and Men. Stanford, CA:
true “second revolution,” one that establishes Stanford University Press.
this new and broader definition of equality, will Crompton, Rosemary. 2001. “The Gendered Re-
therefore be needed to eliminate essentialist seg- structuring of the Middle Classes.” In Janeen
regation. There should be no illusions about how Baxter and Mark Western, eds., Reconfigurations
formidable the remaining barriers are. Far from of Class and Gender, pp. 29–54. Stanford, CA:
Stanford University Press.
being some “inevitable destiny” (Jackson 1998,
Deaux, Kay, and Mary Kite. 1987. “Thinking About
p. 271), the second revolution will face many Gender.” In B. Hess and M. Ferree, eds., Analyz-
obstacles, not the least of which is an entrenched ing Gender: A Handbook of Social Science Research,
tradition of classical liberalism that celebrates in- pp. 92–117. Newbury Park, CA: Sage.
dividual choice and thus supports and sustains
 Charles and Grusky—Egalitarianism and Gender Inequality • 911

England, Paula. 1979. “Women and Occupational Parsons, Talcott. 1970. “Equality and Inequality in
Prestige: A Case of Vacuous Sex Equality.” Signs Modern Society, or Social Stratification Revis-
5: 252–265. ited.” In Edward O. Laumann, ed., Social Strat-
Fiske, Susan T. 1998. “Stereotyping, Prejudice, and ification: Research and Theory for the 1970s, pp.
Discrimination.” In D. T. Gilbert, S.T. Fiske, and 14–72. Indianapolis: Bobbs-Merrill.
G. Lindzey, eds., Handbook of Social Psychology, Ramirez, Francisco O. 1987. “Global Changes,
pp. 357–411. New York: McGraw-Hill. World Myths, and the Demise of Cultural Gen-
Ganzeboom, Harry B. G., and Donald J. Treiman. der: Implications for the United States.” In Terry
1996. “Internationally Comparable Measures of Boswell and Albert Bergesen, eds., America’s
Occupational Status for the 1988 International Changing Role in the World System, pp. 257–73.
Standard Classification of Occupations.” Social New York: Praeger.
Science Research 25: 201–239. Reskin, Barbara F. 2000. “The Proximate Causes of
Hartmann, Heidi I. 1981. “The Unhappy Marriage Employment Discrimination.” Contemporary So-
of Marxism and Feminism: Towards a More Pro- ciology 29: 319–328.
gressive Union.” In Lydia Sargent, ed., Women Reskin, Barbara F., and Patricia A. Roos. 1990. Job
and Revolution: A Discussion of the Unhappy Mar- Queues, Gender Queues: Explaining Women’s In-
riage of Marxism and Feminism, pp. 1–41. Bos- roads into Male Occupations. Philadelphia: Tem-
ton: South End Press. ple University Press.
Huber, Joan. 1988. “A Theory of Family, Economy, Ridgeway, Cecilia L. 1997. “Interaction and the
and Gender.” Journal of Family Issues 9: 9–26. Conservation of Gender Inequality: Considering
Jackson, Robert Max. 1998. Destined for Equality: Employment.” American Sociological Review 62:
The Inevitable Rise of Women’s Status. Cambridge, 218–235.
MA: Harvard University Press. Roos, Patricia A. 1985. Gender and Work: A Com-
Jacobs, Jerry. 2003. “Detours on the Road to Equal- parative Analysis of Industrial Societies. New York:
ity: Women, Work, and Higher Education.” Con- State University of New York Press.
texts 2: 32–41. Semyonov, Moshe, and Frank L. Jones. 1999.
Lorber, Judith. 1993. Paradoxes of Gender. New “Dimensions of Gender Occupational Dif-
Haven, CT: Yale University Press. ferentiation in Segregation and Inequality: A
Massey, Douglas S., and Nancy A. Denton. 1993. Cross-National Analysis.” Social Indicators Re-
American Apartheid: Segregation and the Making search 46: 225–247.
of the Underclass. Cambridge, MA: Harvard Uni- Strober, Myra H. 1984. “Toward a General Theory
versity Press. of Occupational Sex Segregation: The Case of
Meyer, John W. 2001. “The Evolution of Modern Public School Teaching.” In Barbara F. Reskin,
Stratification Systems.” In David B. Grusky, ed., ed., Sex Segregation in the Workplace. Washington,
Social Stratification: Class, Race, and Gender in So- DC: National Academy Press.
ciological Perspective, 2d ed., pp. 881–90. Boul- Tilly, Charles. 1998. Durable Inequality. Berkeley:
der, CO: Westview Press. University of California Press.
Milkman, Ruth, and Eleanor Townsley. 1994. “Gen- Weeden, Kim A. 2004. “Profiles of Change: Sex Seg-
der and the Economy.” In Neil J. Smelser and regation in the United States.” In Maria Charles
Richard Swedberg, eds., The Handbook of Eco- and David B. Grusky, Occupational Ghettos: The
nomic Sociology, pp. 600–619. Princeton, NJ: Worldwide Segregation of Women and Men. Stan-
Princeton University Press. ford, CA: Stanford University Press.
912 • Part VIII: Gender Inequality

Gender Gap in Wages

105 • Trond Petersen and Laurie A. Morgan

The Within-Gender Wage Gap

There are three types of discrimination that discrimination and more a matter of allocative
can produce wage differences between men and and valuative processes. That is, the segregation
women. (1) Women are differentially allocated of women into lower paying occupations, estab-
to occupations and establishments that pay lower lishments, or both, and lower pay in occupations
wages. This may involve discrimination partly held primarily by women are more important
through differential access to occupations and than pay differences within the same job in
establishments—that is, the matching process explaining the gender wage gap. Treiman and
at the point of hire—and partly through subse- Hartmann (1981, 92–93) write, “Although the
quent promotions. We call this process allocative committee recognizes that instances of unequal
discrimination. (2) Occupations held primarily pay for the same work have not been entirely
by women are paid lower wages than those held eliminated, we believe that they are probably not
primarily by men, although skill requirements now the major source of differences in earnings.”
and other wage relevant factors are the same. This conjecture is drawn primarily from a
This process, which we call valuative discrimi- large literature that focuses on pay differences
nation, is addressed by comparable worth initia- across and within occupations. One pattern of
tives. (3) Women receive lower wages than men findings is that the wage gap between men and
within a given occupation and establishment. women becomes smaller as occupational controls
We call this process within-job wage discrimi- become finer, suggesting that a large proportion
nation. Allocative and valuative discrimination of the wage gap is explained by occupational dis-
involve the segregation of men and women into tribution (Treiman and Hartmann 1981; Res-
different occupations, establishments, or both, kin and Roos 1990). For example, Treiman and
and may occur without within-job wage discrim- Hartmann (1981, 33–39) explained 10 to 20
ination. Thus, although it may be the case that percent of the raw gap using 222 occupational
where men and women share the same jobs they categories and 35 to 40 percent using 479 cate-
receive the same pay, they simply do not share gories. These studies usually draw on data from
the same jobs all that often. One conjecture cur- the Census or national probability samples that
rently accepted by many researchers is that wage allow no analysis of practices in specific estab-
differences are less an issue of within-job wage lishments. Additional evidence suggests that,

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in the
following publication: Trond Petersen and Laurie A. Morgan, “Separate and Unequal: Occupation-Establishment
Sex Segregation and the Gender Wage Gap,” American Journal of Sociology 101:2 (September 1995), pp. 329–
365, published by the University of Chicago Press. The article printed here was originally prepared by Trond
Petersen and Laurie A. Morgan for Social Stratification: Class, Race, and Gender in Sociological Perspective, Second
Edition, edited by David B. Grusky, pp. 734–742. Copyright © 2001 by Westview Press.
 Petersen and Morgan—The Within-Gender Wage Gap • 913

within occupations, the distribution of women of observed patterns, neither from the demand
across firms or establishments also accounts for side—that is, discriminatory behavior by em-
some portion of the wage gap. For example, Blau ployers—or the supply side—that is, behav-
(1977) found that in 11 clerical occupations, dif- iors by employees and prospective employees
ferences in men’s and women’s wages were larger (see England 1992, chap. 2). Nevertheless, our
between than within establishments. results have implications for the kinds of the-
Yet the prevailing conjecture remains a con- oretical issues that are most in need of being ad-
jecture. It has not been shown that men and dressed and for the type of data that need to be
women receive equal pay within given occupa- collected and analyzed.
tions in given establishments. What has been
shown is that sex segregation is extensive and
pervasive (Bielby and Baron 1984; Petersen and Data
Morgan 1995), but not the extent to which sex We use two large-scale data sets. The first data
segregation accounts for the wage gap or that, set comes from 16 Industry Wage Surveys (IWS)
when sex segregation is absent, the sexes receive conducted by the U.S. Bureau of Labor Statis-
more or less equal treatment. To confirm such tics (BLS) in the period 1974 to 1983 (see, e.g.,
a claim, one needs data on wages of men and U.S. Department of Labor 1976), correspond-
women in the same detailed occupational group ing to industry codes at three and more digits as
or position within the same establishment. Such defined in the Standard Industrial Classification
data are not widely available except for isolated Manual (see U.S. Executive Office of the Presi-
establishments. dent 1987). Eleven industries were surveyed in
This [chapter] reports a large-scale empirical 1974 to 1978, whereas 5 were surveyed in 1980
investigation of wage differences between men to 1983. The populations for the surveys and the
and women within the same detailed occupa- sampling from the populations are described in
tional position within the same establishment. several U.S. Department of Labor publications
We use establishment-level data from a wide (see Petersen and Morgan 1995, table 1). Of the
variety of industries. In each establishment, 16, 11 are manufacturing industries, and 5 are
individual-level wage data for a large array of service industries.
detailed occupational groups were collected, In each industry, the BLS drew a sample of
providing more accurate wage as well as occu- several hundred establishments, often cover-
pational data than probably any other surveys ing a large proportion of the establishments in
available, except in some case studies of single the industry. For each establishment, informa-
establishments (e.g., Hartmann 1987). We tion was obtained from establishment records
focus first on production and clerical employ- both on establishment characteristics and on a
ees in 16 U.S. industries in the 1974 to 1983 large number of the production and/or clerical
period, primarily 1974 to 1978, analyzing data workers in the establishment. Within each in-
on 787,577 employees, 705 industry-specific dustry, only a selection of occupations were sur-
occupations, 6,057 establishments, and 71,214 veyed—on average 42 occupations per industry.
occupation-establishment pairs. Second, we The occupations were selected by the BLS in
focus on seven professional and three adminis- order to provide a wide representation of pro-
trative occupations across a broad range of indus- duction or clerical occupations in an industry.
tries in 1981, analyzing data on about 740,000 The individual-level data provide information
employees distributed across 2,162 establish- on each individual in the relevant occupation
ments and 16,433 occupation-establishment and establishment. Excluded from the data col-
pairs. lection were professional and managerial em-
We make no attempt to settle the important ployees. Because these occupations may have
conceptual issues that go along with the em- exhibited wider variations in wages, even at the
pirical patterns we address, namely the sources occupation-establishment level, there may have
914 • Part VIII: Gender Inequality

been less occupation-establishment level vari- industry are given by w f and w m, and the raw
ation in wages here than in samples including proportional wage between women and men
professionals and managers. is wr,r = w f / w m. The average wages for women
For each employee surveyed, information was and men in occupation o are wo,f and wo,m, and
obtained on sex, occupation (an industry-specific the proportional wage is wo,r = w o,f / w o,m. The
code), method of wage payment (incentive- or average wages for women and men in establish-
time-rated), and hourly earnings. No informa- ment e are we,f and we,m, and the proportional
tion was collected on race, age, experience, or wage is we,r = we,f / we,m. The average wages for
education. The occupational classification is women and men in occupation-establishment
unusually detailed, corresponding in many cases pair oe are woe,f and woe,m, and the proportional
to nine digits in the Dictionary of Occupational wage is woe,r =woe,f / woe,m. These computations can
Titles (see U.S. Department of Labor 1977). In only be done for units that are integrated by sex.
other cases, the titles are specific to the BLS data, We multiply each proportional wage by 100 in
based on industry-specific codes, but are usually order to get the average female wage as percent
as detailed as the nine-digit titles in the Dictio- of average male wage at the relevant level. This
nary of Occupational Titles. percentage is referred to as the relative wage in
Wage data are straight-time hourly wages what follows.
in 13 industries and full-time weekly earnings We make three computations with rel-
in the other three, excluding premium pay for ative wages. Each computation answers
overtime and work on weekends, holidays, and the following question: suppose sex segre-
late shifts. Thus, we do not conflate pay earned gation—by occupation, establishment, or
on regular hours with pay earned on overtime occupation-establishment—were abolished,
and irregular hours, making the wage data less what then would the remaining relative wage be?
prone to bias than virtually any other study used These computations are defined in the appendix.
for assessing wage discrimination. Men work At the occupation level, we report the aver-
more overtime hours than women, due either to age of the relative wage across all sex-integrated
preference for or better access to overtime hours, occupations. We do the same at the establish-
and overtime hours are usually paid at a higher ment level. At the occupation-establishment
rate. Non-production bonuses, such as year-end level, we likewise report the average of the rel-
bonuses, are also excluded, whereas incentive pay ative wage (woe,r × 100) across all sex-integrated
is included. occupation-establishment units (oe). The female
The second data set we use is the National wage gap at each of these levels is then given
Survey of Professional, Administrative, Techni- as 100 minus the average of the relative wage
cal, and Clerical (PATC) employees in 1981, at each level. This equals the percent by which
also conducted by the BLS (U.S. Department of women are paid less than men.
Labor 1981). The sampling and data collection The remaining wage gap at the occupation-
design for this survey is similar to that in the establishment level can reasonably be inter-
IWS data. We use the data on weekly pay for preted as an estimate of the upper bound on
full-time employees in seven professional and the amount of within-job wage discrimination,
three administrative occupations. the gap one would observe in the absence of
occupation-establishment sex segregation. It is an
upper bound because the remaining gap at that
Methods level could have been caused by other factors,
We report all statistics separately for each of such as differences in experience and human cap-
the 16 industries in the IWS data.1 The raw ital between men and women in a given category.
(average) wages, either hourly or full-time The difference between the raw wage gap and
weekly earnings, for women and men in an the wage gap at the occupation-establishment
 Petersen and Morgan—The Within-Gender Wage Gap • 915

Table 105.1. Women’s Wages Relative to Men’s With and Without Occupation, Establishment, and
Occupation-Establishment Controls

Women’s Wages as a Percent of Raw Wage

Percentage of Men’s Gap Explained by:
%F Raw Occ Est Occ-Est Occ Est Occ-Est
Industry 1 2 3 4 5 6 7 8
Men’s and Boys’ Shirts 92.5 84.0 92.7 83.1 98.0 54.5 –5.4 87.5
Hospitals 84.8 105.3 95.8 103.7 101.0 178.7 30.7 81.3
Banking* 82.8 64.7 97.3 65.5 98.1 92.4 2.3 94.6
Life Insurance* 75.9 61.5 98.0 61.4 98.9 94.8 –0.3 97.1
Wool Textiles 55.4 87.3 93.8 91.7 98.6 51.3 34.8 89.0
Cotton Man-Made 54.6 89.2 96.5 90.5 99.7 67.5 11.8 97.2
Fiber Textiles
Miscellaneous Plastics 51.9 77.5 87.4 80.6 97.8 44.1 14.0 90.2
Products
Hotels and Motels 51.5 87.0 91.9 89.1 99.7 37.9 16.4 97.7
Computer and Data 44.5 67.5 98.4 81.4 98.9 95.1 42.8 96.6
Processing*
Wood Household 35.9 85.4 91.3 90.4 95.7 40.3 34.2 70.5
Furniture
Textile Dyeing and 19.8 82.7 92.2 90.2 99.0 55.0 43.4 94.2
Finishing
Machinery 15.5 74.3 88.8 89.0 99.6 56.4 57.2 98.4
Nonferrous Foundries 9.4 81.3 84.1 84.4 97.5 15.1 16.7 86.7
Paints and Varnishes 5.6 85.1 92.6 91.8 96.6 50.2 44.8 77.1
Industrial Chemicals 2.5 83.2 93.2 90.1 97.1 59.6 41.2 82.8
Fabricated Structural Steel 0.6 80.7 86.0 81.5 96.9 27.4 4.1 83.9
Average Across 42.7 81.0 92.5 85.3 98.3 63.8 24.3 89.1
Industries**
Notes: Column 1, %F, gives the percent female in the industry. Column 2 gives the raw relative wage, from equa-
tion (1) in the Appendix. Columns 3–5 give the average of the within occupation, within establishment, and within
occupation-establishment relative wages, computed from equations (2)–(4). Columns 6–8 give the percent of the raw
wage gap, defined as 100 minus the number in column 2, that can be attributed to occupation, establishment, and
occupation-establishment segregation, computed from equations (5)–(7).
*
This is based on weekly wages for full-time employees.
**
Each column gives the unweighted average across industries of the percentages in the respective column.

level can thus be interpreted as being attributable computed from equations (1) to (4) in the ap-
to occupation-establishment sex segregation. pendix. Column 1 shows the percentage female
This part of course need not be caused by dis- in each industry. Column 2 shows that the av-
crimination alone. erage of the industry-specific raw relative wages
between men and women was 81 percent. Thus
the average of the industry-specific female wage
Results gaps was 19 percent, where the wage gap is the
Table 105.1 gives first in column 2 the raw percentage by which women earned less than
relative wage and then in columns 3 to 5 the men, computed as 100 minus the relative wage.
relative wages controlling for occupation, for es- Column 3 shows that the female wage gap is
tablishment, and for occupation-establishment, reduced to an average, across the 16 industries, of
916 • Part VIII: Gender Inequality

8 percent when we control for occupation, using by occupation-establishment segregation, and

equation (2). Column 4 shows that the female among seven of those more than 90 percent is
wage gap is reduced, across the 16 industries, to explained, whereas in two industries only 70
an average of 15 percent when we control for percent and 77 percent of the gap is explained.
establishment, using equation (3). Column 5 Again, we have the striking result that the
shows that the female wage gap is reduced to interaction of occupation and establishment
about 0 to 4 percent when we control for the segregation not only explains dramatically more
occupation-establishment pair, using equation of the wage gap than either establishment or
(4). The average of the industry-specific female occupational segregation alone, it also explains
wage gaps is 1.7 percent, after controlling for the most of the wage differences between men and
occupation-establishment pair. women. Had men and women been equally dis-
These results are striking. Controlling for oc- tributed on occupation-establishment pairs, on
cupation or establishment alone reduces the wage the average 89 percent of the wage gap would
gap somewhat, but not drastically. However, have disappeared, assuming that other forms of
controlling for the occupation-establishment discrimination would not have emerged.
pair reduces the wage gap to a point of virtually One point requires discussion here. The fe-
no difference between men and women. The male wage gap for the workers in our sample
remaining female wage gap is on average 1.7 is about 20 percent. This is substantially less
percent across the 16 industries, even without than the about 40 percent gap for all full-time
controlling for any individual-level characteris- workers during the period of the surveys (e.g.,
tics such as education, age, seniority, and race. Goldin 1990, 61). There are several reasons for
Occupation-establishment segregation accounts this discrepancy. One is that prior studies look at
better for wage differences between men and median or average annual earnings for full-time
women than any other set of variables studied in employees. It is probable that men on the average
the literature on wage differences.2 work more hours per year than women and that
Columns 6 to 8 give a decomposition of the they also work more overtime hours. Therefore,
raw wage gap, calculated as 100 minus the rel- differences between men and women are likely
ative wage from column 2, into the percentage overstated. Another reason for this difference is
that is due to occupation, to establishment, and that the occupations covered here comprise a
to occupation-establishment segregation, when relatively narrow range of the overall spectrum.
each dimension is considered alone. These re- In particular, professional and managerial oc-
sults are based on equations (5) to (7) in the cupations are excluded from the sample. The
appendix. Column 6 shows that occupational excluded occupations are expected to be on av-
sex segregation alone accounts for about 64 erage higher paid than those included and to be
percent of the wage gap, somewhat more than occupied primarily by men. As a result, we have,
other studies have shown (e.g., Treiman and when computing the raw wage gap of 19 percent,
Hartmann 1981; Reskin and Roos 1990). Col- some degree of de facto control for occupation.
umn 7 shows that establishment sex segregation The latter results were based on data mostly
alone accounts for about 24 percent of the wage for blue-collar, clerical, and some technical em-
gap. This is a new result. To date, occupational ployees. It might be suspected that the wage
segregation has been the primary focus of gen- gap among professional, administrative, and
der wage gap studies. Column 8 shows that managerial employees is greater. However, when
occupation-establishment sex segregation alone we turn our attention to analysis of wage dif-
accounts for a large percentage of the wage gap, ferences among these employees using PATC
ranging from a low of 70.5 percent to a high data, the results are similar to those in the IWS
of 98.4 percent. In 14 of 16 industries, more data. The wage gap in the PATC data is just 3.1
than 80 percent of the wage gap is explained percent among men and women in the same
 Petersen and Morgan—The Within-Gender Wage Gap • 917

position (occupation and rank) and the same valuative processes should be given the most
establishment. attention, and within-job wage discrimination,
which is covered by the Equal Pay Act of 1963
and which has been the implicit or explicit focus
Conclusion and Discussion of much discussion and research, should receive
The findings of this [chapter] are simple to less. Future research on differential wage attain-
summarize. First, wage differences within given ment between men and women should be re-
occupation-establishment pairs were relatively focused as follows. The emphasis should be less
small: on average 1.7 percent among blue-collar on within-job wage discrimination and more on
and clerical as well as some technical employ- three prior processes: (1) the entry of employ-
ees, and on average 3.1 percent in seven pro- ees into occupations and establishments, that
fessional and three administrative occupations, is, the differential access of men and women to
ranging from 1 percent in the lower to 5 per- positions during the initial hiring or matching
cent in the higher ranks in an occupation. Thus, process, an allocative mechanism that is not easy
occupation-establishment segregation, not to research; (2) career advancement within es-
within-job wage discrimination, was the driving tablishments, that is, the differential rates of pro-
force for observed wage differences.3 Second, es- motion for men and women, also an allocative
tablishment segregation was important for wage mechanism, but one which is more easily studied
differences between men and women, although (Spilerman 1986); and (3) how jobs occupied
not as important as occupational segregation, a primarily by women tend to be paid less than
result we could compute only for the blue-collar, those occupied by men—the comparable worth
clerical, and some technical employees in the issue, or what we refer to as valuative discrimi-
IWS data. nation, also a line of research well under way (see
The first finding is important. It shows that England 1992).
occupation-establishment segregation accounted Two issues arise in the identification of allo-
for more of the gender wage gap than any other cative processes as responsible for the gender wage
variable or set of variables currently used in gap. The first is whether segregation patterns are
studying the gender wage gap. Occupational due primarily to discrimination or to differences
segregation alone tends to account for about 40 in productive capacities. The meaning of our
percent of the wage gap, and human capital and results for theory and policy depends on which
other variables also account for about 40 per- mechanism operated here. This cannot be settled
cent (e.g., Treiman and Hartmann 1981, chap. with our data and is obviously a task for future
2). But no set of variables, either individual or research. It requires information about produc-
structural, accounts for as much as 89 percent, tive capacities and would require an analysis of
as occupation-establishment segregation did here the matching between these and particular jobs.
in the case of blue-collar, clerical, and some tech- The second issue concerns the role of
nical employees in the IWS data. supply-side sources of differential attainment
This first finding establishes the conjecture between men and women. Of particular interest
already made in the literature, but not yet doc- are the constraints put on women’s career attain-
umented: wage differences are to a larger extent ment from family obligations and household
generated by occupation-establishment segrega- choices. Although these constraints may result
tion than by within-job wage discrimination. most proximally from a traditional household di-
Along with the first, the second finding shows vision of labor, not directly from discriminatory
the need to study establishment as well as occu- behaviors by employers, household decisions
pational segregation. are made in light of labor market opportuni-
The implications of the findings are straight- ties, so the two are interdependent. The role of
forward. In terms of policy, allocative and supply-side behaviors and their interrelationship
918 • Part VIII: Gender Inequality

with employer behaviors in generating the ob- The percentage of the raw wage gap due to
served occupation-establishment segregation is occupation-establishment segregation alone is
in need of research. given by

w(oe,r) – w(r,r)
%w(oe,r) = 100. (7)
Appendix 100 – w(r,r)
The raw relative wage between women and men
is given as NOTES
The chapter is a version of Petersen and Morgan
—
w (1995) that was excerpted and partially rewritten for
w(r,r) = f 100. (1)
—
w this volume by Petersen and Morgan with the help
m
of David Grusky.
1. Unfortunately, we cannot compare wages
The relative wage controlling for occupation across industries, because the industry data come
obtains as from different years and hence reflect inflation as
well as general wage increases.
w(o,r) = 1 [ No(I)
o=1
(wo,r )] 100. (2) 2. Although Groshen (1991) and Tomaskovic-
No(I) Devey (1993) report similar results on wage differ-
Here, No(I ) is the number of occupations in ences, neither author reported the wage gap at the
which both men and women are present. occupation-establishment level.
3. Similar results are shown in Petersen et al.
The establishment-level relative wage obtains
(1997) for Norway in 1984 and 1990.
as
REFERENCES
w(e,r) = 1 [ Ne(I)
e=1
(we,r )] 100. (3) Bielby, William T., and James N. Baron. 1984. “A
Ne(I) Woman’s Place Is with Other Women: Sex Seg-
Here, Ne(I ) is the number of establishments regation Within Organizations.” Pages 27–55 in
where both men and women are present. Sex Segregation in the Workplace: Trends, Expla-
The occupation-establishment-level relative nations, Remedies, edited by Barbara F. Reskin.
wage obtains as Washington, D.C.: National Academy Press.
Blau, Francine D. 1977. Equal Pay in the Office. Lex-
ington, MA: Lexington Books.
w(oe,r) = 1 [ Noe(I)
(woe,r )] 100. (4)
Noe(I) oe=1 England, Paula. 1992. Comparable Worth: Theo-
ries and Evidence. Hawthorne, NY: Aldine de
Here, Noe(I) is the number of occupation- Gruyter.
establishment pairs in which both men and Goldin, Claudia. 1990. Understanding the Gender
women are present. Gap: An Economic History of American Women.
The raw wage gap obtains as 100 minus the New York: Oxford University Press.
number in (1). The percentage of the raw wage Groshen, Erica L. 1991. “The Structure of the Fe-
gap due to occupational segregation alone is male/Male Wage Differential: Is It Who You Are,
given by What You Do, or Where You Work?” Journal of
Human Resources 26(3):457–72.
w(o,r) – w(r,r) Hartmann, Heidi I. 1987. “Internal Labor Markets
%w(o,r) = 100. (5) and Gender: A Case Study of Promotion.” Pages
100 – w(r,r)
59–92 in Gender in the Workplace, edited by Clair
The percentage of the raw wage gap due to Brown and Joseph Pechman. Washington, D.C.:
establishment segregation alone is given by Brookings Institution.
Petersen, Trond, and Laurie A. Morgan. 1995. “Sep-
w(e,r) – w(r,r)
%w(e,r) = 100. (6) arate and Unequal: Occupation-Establishment
100 – w(r,r) Sex Segregation and the Gender Wage Gap.”
American Journal of Sociology 101(2): 329–61.
 England—Devaluation and the Pay of Comparable Male and Female Occupations • 919

Petersen, Trond, Vemund Snartland, Lars-Erik Jobs of Equal Value. Washington, D.C.: National
Becken, and Karen Modesta Olsen. 1997. Academy Press.
“Within-Job Wage Discrimination and the Gen- U.S. Department of Labor. 1976. Industry Wage
der Wage Gap, The Case of Norway.” European Survey: Miscellaneous Plastics, September 1974.
Sociological Review 13(2):199–215. Bureau of Labor Statistics, Bulletin 1914. Wash-
Reskin, Barbara F., and Patricia Roos (Eds.). 1990. ington, D.C.: U.S. Government Printing Office.
Job Queues, Gender Queues: Explaining Women’s ———. 1977. Dictionary of Occupational Titles. 4th
Inroads into Male Occupations. Philadelphia: ed. Bureau of Employment Security. Washing-
Temple University Press. ton, D.C.: U.S. Government Printing Office.
Spilerman, Seymour. 1986. “Organizational Rules ———. 1981. National Survey of Professional, Ad-
and the Features of Work Careers.” Research in ministrative, Technical, and Clerical Pay, March
Social Stratification and Mobility 5:41–102. 1981. Bureau of Labor Statistics, Bulletin 2108.
Tomaskovic-Devey, Donald. 1993. Gender and Race Washington, D.C.: U.S. Government Printing
Inequality at Work: The Sources and Consequences Office.
of Job Segregation. Ithaca, NY: ILR Press. U.S. Executive Office of the President. 1987. Stan-
Treiman, Donald J., and Heidi I. Hartmann (Eds.). dard Industrial Classification Manual. 1987. Of-
1981. Women, Work, and Wages: Equal Pay for fice of Management and Budget. Washington,
D.C.: U.S. Government Printing Office.

106 • Paula England

Devaluation and the Pay of Comparable Male and

Female Occupations

The terms comparable worth and pay equity refer to some jobs because they are filled largely with
to a form of sex discrimination that went vir- women has come to be called the devaluation
tually unrecognized until the 1980s. Evidence perspective. Petersen and Saporta (2004) have
abounds that jobs filled mostly by women have referred to this type of discrimination as “valua-
pay levels lower than they would be if the jobs tive discrimination.”
were filled mostly by men. This is seen as sex At first glance, the issue of comparable worth
discrimination by advocates of the principle of sounds like the more familiar issue of “equal pay
comparable worth, which posits that jobs that for equal work,” which refers to men and women
are equally demanding and onerous and of equal in the same job with the same seniority, perform-
value to society should be paid equally. The the- ing the same work equally well, but being paid
ory positing that employers assign lower wages differently. Petersen and Saporta (2004) have

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publications: Paula England, Comparable Worth: Theories and Evidence, published by Aldine Press
in 1992; and Paula England, Joan H. Hermsen, David A. Cotter, “The Devaluation of Women’s Work: A
Comment on Tam,” American Journal of Sociology 105 (2000), pp. 1741–1751, published by the University of
Chicago Press. The article printed here was originally prepared by Paula England for inclusion in The Inequality
Reader: Contemporary and Foundational Readings in Race, Class, and Gender, First Edition, edited by David B.
Grusky, pp. 352–356. Copyright © 2006 by Westview Press.
920 • Part VIII: Gender Inequality

called this “within job discrimination,” and it San Jose, California, discovered that secretaries
has been illegal in the United States since the were generally earning less than workers in male
1963 Equal Pay Act and Title VII of the 1964 jobs that required no more than an eighth grade
Civil Rights Act. However, comparable worth is education, including, for example, men who
distinct because it refers to comparisons between washed cars for the city. Eventually, women in
the pay in different jobs that entail at least some San Jose succeeded in getting the city to do a job
distinct tasks. Comparable worth advocates evaluation study. The results showed, to choose
compare jobs that are largely male with those some examples, that nurses earned $9,120 less
that are largely female (which I will sometimes, annually than fire truck mechanics and that
for brevity, refer to as “male” and “female” jobs) legal secretaries made $7,288 less annually than
and, in some cases, claim that the difference be- equipment mechanics. In 1985, the California
tween the pay of the two jobs results from gen- School Employees Association complained that
der bias in wage setting rather than from other school librarians and teaching assistants (fe-
job characteristics. Needless to say, this approach male jobs) were paid less than custodians and
creates a thorny issue in distinguishing when two groundskeepers (male jobs). In recent years, the
distinct jobs are comparable in the sense that we city of Philadelphia, Pennsylvania, was paying
would expect them to pay the same in the ab- practical nurses (mostly women) less than gar-
sence of sex discrimination. deners (mostly men). These are not atypical ex-
The wage discrimination at issue in a com- amples. In addition, one is hard-pressed to come
parable worth scenario is also distinct from dis- up with a single example of a male job paying
crimination in hiring, initial job placement, and less than a female job that reasonable people
promotion, all of which Petersen and Saporta would find comparable in skill, effort, and diffi-
(2004) refer to as “allocative discrimination.” culty of working conditions. Nor are these dif-
Such discrimination is a violation of Title VII ferences in pay a result of men averaging more
of the Civil Rights Act of 1964. Hiring discrim- years of seniority than women, since in the above
ination against women seeking to enter tradi- comparisons (from England 1992: 2), a constant
tionally male jobs is one (although not the only) level of experience was assumed.
reason for occupational sex segregation. There U.S. courts have ruled that, under most con-
are supply-side processes at work as well, such ditions, this type of valuative discrimination and
as those that lead women and men to major in lack of comparable worth is not a violation of
different fields in college. Without segregation of U.S. law (Nelson and Bridges 1999; England
jobs, female jobs could not be given a discrim- 1992: Chapter 5). Studies on data from a single
inatory pay level. But engaging in discrimina- employer like those reviewed here are not seen as
tion on the basis of sex in setting the pay levels evidence of illegal discrimination in wage setting.
assigned to male and female jobs is analytically At present there is no legal remedy for this kind
distinct from engaging in hiring discrimination of discrimination unless there is a “smoking-gun”
on the basis of sex. Moreover, valuative discrim- statement by employers that the reason they paid
ination may exist whether or not the segregation one job less than another was simply because it
came from allocative discrimination or from was filled with women. Statistical evidence of
supply-side forces. consistently biased pay levels is not enough.
Some examples may help the reader to visual- However, occasionally, some of the job evalu-
ize the sorts of comparisons at issue in compara- ation studies have led private and city or state
ble worth. In 1975, nurses in Denver, Colorado, government entities to revise their pay scales to
sued the city claiming that their jobs paid less equalize salaries in jobs deemed comparable.
than male jobs such as tree trimmer and sign Economists propose two explanations for the
painter. It would be hard to argue that the latter lower pay of occupations with a high percent of
two jobs require as much skill or are as demand- female workers. The first is the theory of com-
ing as nursing. Women workers for the city of pensating differentials, which posits that the full
 England—Devaluation and the Pay of Comparable Male and Female Occupations • 921

pay of a job consists of both pecuniary (wage) discrimination discussed above), leading to a
and nonpecuniary compensation (the utility supply of applicants for traditionally female jobs
experienced from doing the work itself ). Jobs larger than it would be in the absence of hiring
with more comfortable, less hazardous working discrimination as women denied entrance to
conditions can be filled with workers easily at male jobs crowd the female jobs. The excess sup-
lower wages, ceteris paribus. If women care more ply lowers wages in female jobs. Although this is
about nonpecuniary rewards (such as avoiding plausible, it is very difficult to test directly.
physical danger or having mother-friendly work Sociologists have proposed the devaluation
conditions such as flexible hours or on-site child thesis, which posits that gender bias leads to
care centers) than men, and men focus more on valuing (and thus paying) female jobs less than
maximizing earnings, then women will trade off comparably skilled male jobs (England et al.
earnings for amenities by choosing safer, more 2000; Steinberg 2001). According to these so-
mother-friendly jobs. Most tests have failed to ciologists, it doesn’t matter if segregation arises
find that greater nonpecuniary amenities explain from allocative discrimination (in hiring and job
much of the lower pay of women’s jobs (En- placement) or not; employers engage in valua-
gland 1992; Glass 1990; Glass and Camarigg tive discrimination once the segregation has oc-
1992; Jacobs and Steinberg 1990; Kilbourne et curred. Of course, the same evidence offered for
al. 1994). The idea that women might choose this view would also be consistent with crowd-
lower-paying jobs for other compensations ing. However, sociologists believe it is unlikely
seems on first glance consistent with the find- that hiring is the only place where bias occurs;
ing that mothers earn less than nonmothers, it occurs in wage-setting between jobs as well.
even after controlling for part-time work status, Evidence that the sex composition of an oc-
experience, and seniority (Budig and England cupation or job affects its wage level supports
2001; Lundberg and Rose 2000; Waldfogel the devaluation theory. Such effects of sex com-
1997). But neither Glass (1990) nor Glass and position, after statistically controlling for the
Camarigg (1992) found women’s jobs to have other factors already discussed, have led some
more mother-friendly characteristics than men’s researchers to conclude that employers set lower
jobs. Similarly, Budig and England (2001) could wages (relative to job demands) when jobs are
not find any job characteristics (except part-time filled largely by women. One type of study takes
status) that reduced the motherhood wage pen- the U.S. Census Bureau’s detailed occupational
alty. Nonmothers were about as concentrated in categories as units of analysis. The measures in-
female jobs as mothers. Thus, oddly enough, in clude the percent female of those in the occu-
the United States, segregation by sex does not pation, average or median male wage, median
seem to stem from women choosing or being female wage, average education of incumbents,
consigned to more mother-friendly jobs. The and a myriad of measures of skill demands from
“motherhood penalty” explains only some of the the Dictionary of Occupational Titles. These
sex gap in pay, and this portion is largely unre- studies examine the effect of occupational per-
lated to the effect of occupational sex segregation cent female on average male and female wages
on the sex gap in pay. Thus, although the general (separately for men and women) after adding
theory of compensating differentials may have all the educational, skill, and working condition
some merit, it cannot explain most of the lower demands as control variables whose effects are
pay of female jobs. adjusted out statistically. These studies find that
A second explanation offered by economists both men and women earn less when in a more
for the lower pay in female jobs is crowding. female occupation (England 1992; England et
Bergmann (1986) argues that women’s jobs pay al. 2001). That is, in a regression predicting me-
less because they are crowded. According to this dian wage, percent female of the occupation is
view, women seeking to enter male occupations significant and negative. (See Filer 1989 for a
face sex discrimination in hiring (the allocative contrary finding.)
922 • Part VIII: Gender Inequality

Other studies use individuals as units of consultants make about the wage levels for var-
analysis, and find that individuals in more fe- ious jobs. In part, this results from a cognitive
male occupations earn less, after similar statis- distortion in underestimating the relative con-
tical controls to those described above (England tribution to profits or other organization goals
et al. 2000; see Tam 1997, 2000 for a contrary of the work done in female jobs. In part, this
view). Other studies take person/years as units bias on the part of organizational decision mak-
of analysis, and percent female in occupations ers results from generalizing to women’s jobs
as contextual variables. The studies with per- the relatively low status accorded women as
son/years as units employ person-fixed effects to persons by our culture. Later, bureaucratic iner-
remove omitted variable bias; that is, they use tia takes over, setting the wage consequences of
persons as their own control and generally show these cultural biases in stone; the relative wage
that the same person earns more when moving levels of jobs within an organization often re-
to a more male-dominated occupation or earns main quite constant for decades. Relative wage
less when moving to a more female-dominated scales get institutionalized. If any attempt is
occupation (Kilbourne et al. 1994; MacPherson made to change relative wage levels in a way
and Hirsch 1995). favorable to female occupations, individual or
Other studies use jobs from a single employer collective action by male workers may increase
as cases and show that female jobs pay less than the wages in male-dominated jobs over those of
male jobs when a job evaluation (of skill de- female-dominated jobs again. Market processes
mands and working conditions) finds these jobs then reproduce all these effects. That is, if most
comparable (reviewed in England 1992). employers have at some point inscribed these bi-
All of these studies suggest that jobs pay ei- ases into their wage structures, this affects the
ther men or women less if they are filled mostly “going wage” in predominantly female and male
by women. The low pay of the largely female occupations that other new organizations must
jobs cannot be entirely explained by how much pay to hire workers of typical quality into the
skill the job requires or how onerous the working occupation. These market forces dictate that
conditions are, because measures of these factors even an organization not beset by these cogni-
are controlled in the regression analyses that find tive and cultural biases will be forced to pay a
a significant negative effect of the percent female higher wage to recruit in male occupations than
in the occupation or job. These studies convince comparable female occupations. Although they
many sociologists that employers’ evaluation of could choose to pay more than they need to in
jobs is gendered—and that devaluation infects female occupations—more than the market or
the wage levels of female jobs. Sociologists posit average wage—few profit-minded employers will
that this devaluation is perpetuated through do so. In these ways, gender bias is perpetuated
cultural, institutional, and market mechanisms. over time. Absent significant collective action or
Although the evidence of devaluation is quite new legislation that requires comparable worth
strong, there is little direct evidence of the mech- in pay scales, the devaluing social and inertial
anisms through which it occurs. forces described above may keep the pay penalty
I close with a somewhat speculative descrip- for working in female jobs in force indefinitely.
tion of the process by which the devaluation of
female jobs has probably occurred. At certain REFERENCES
points in the life of organizations, particularly Bergmann, Barbara. 1986. The Economic Emergence
of Women. New York: Basic Books.
when they are first formed, or undergo signifi-
Budig, Michelle J., and Paula England. 2001. “The
cant restructuring, the relative wages of various
Wage Penalty for Motherhood.” American Socio-
jobs are somewhat up for grabs, although not logical Review 66: 204–225.
unaffected by market forces. At these points, England, Paula. 1992. Comparable Worth: Theories
gender bias infects the decisions managers or and Evidence. New York: Aldine.
 England—Devaluation and the Pay of Comparable Male and Female Occupations • 923

England, Paula, Jennifer Thompson, and Carolyn the Wages of White Women and Men.” American
Aman. 2001. “The Sex Gap in Pay and Compa- Journal of Sociology 100: 689–719.
rable Worth: An Update.” Pp. 551–556 in Ivar Lundberg, Shelley, and Elaina Rose. 2000. “Par-
Berg and Arne Kalleberg, eds., Sourcebook on enthood and the Earnings of Married Men and
Labor Markets: Evolving Structures and Processes. Women.” Labour Economics 7:689–710.
New York: Plenum. MacPherson, D. A., and B. T. Hirsch. 1995. “Wages
England, Paula, Joan M. Hermsen, and David A. and Gender Composition: Why Do Women’s
Cotter. 2000. “The Devaluation of Women’s Jobs Pay Less?” Journal of Labor Economics 13:
Work: A Comment on Tam.” American Journal 426–471.
of Sociology 105: 1741–1751. Nelson, Robert, and William Bridges. 1999. Legal-
Filer, Randall K. 1989. “Occupational Segregation, izing Gender Inequality. New York: Cambridge
Compensating Differentials and Comparable University Press.
Worth.” Pp. 153–170 in Michael, Robert T., Petersen, Trond, and Ishak Saporta. 2004. “The Op-
Heidi I. Hartmann, and Brigid O. Farrell, eds., portunity Structure for Discrimination.” Ameri-
Pay Equity: Empirical Inquiries. Washington, DC: can Journal of Sociology 109: 852–901.
National Academy Press. Steinberg, Ronnie. J. 2001. “Comparable Worth
Glass, Jennifer. 1990. “The Impact of Occupational in Gender Studies.” Pp. 2293–2397 in N. J.
Segregation on Working Conditions.” Social Smelser and P. B. Baltes, eds., International Ency-
Forces 68: 779–796. clopedia of the Social and Behavioral Sciences, vol.
Glass, Jennifer, and Valerie Camarigg. 1992. “Gen- 4. London: Elsevier.
der, Parenthood, and Job-Family Compatibility.” Tam, Tony. 1997. “Sex Segregation and Occupa-
American Journal of Sociology 98: 131–151. tional Gender Inequality in the United States:
Jacobs, Jerry, and Ronnie Steinberg. 1990. “Com- Devaluation or Specialized Training?” American
pensating Differentials and the Male-Female Journal of Sociology 102: 1652–1692.
Wage Gap: Evidence from the New York State ———. 2000. “Occupational Wage Inequality and
Comparable Worth Study.” Social Forces 69: Devaluation: A Cautionary Tale of Measure-
439–468. ment Error.” American Journal of Sociology 105:
Kilbourne, Barbara S., Paula England, George Farkas, 1752–1760.
Kurt Beron, and Dorothea Weir. 1994. “Returns Waldfogel, Jane. 1997. “The Effect of Children on
to Skill, Compensating Differentials, and Gender Women’s Wages.” American Sociological Review
Bias: Effects of Occupational Characteristics on 62: 209–217.
924 • Part VIII: Gender Inequality

107 • Tony Tam

Why Do Female Occupations Pay Less?

Background the laboratory and by surveys of prestige ratings

For the past two decades, scholars of gender of occupations. Thus the typical comparable
inequality in the labor market have sought to worth policy in the United States seeks to equal-
understand why occupations with more female ize wages for occupations (within a firm) that
workers tend to pay less. This question addresses differ mainly in sex composition.
a robust finding about occupational sex compo-
sition and wages: studies based on different sam-
ples, units of analysis, and statistical techniques Human Capital and Specialization
almost always find significant wage effects of oc- The main competing interpretations in the liter-
cupational sex composition even after account- ature on occupational sex composition effects are
ing for a wide range of personal attributes and (a) the devaluation hypothesis that female work
occupational characteristics. The evidence of is deemed less important and thereby culturally
lower wages in “female occupations” (i.e., occu- devalued (England et al. 1988), and (b) the spe-
pations that are disproportionately female) may cialized human capital hypothesis that female
reflect subtle gender discrimination. Instead of occupations require, on average, less substantial
wage discrimination against female workers, gen- investments in specialized skills (Tam 1997). The
der discrimination may take the form of wage devaluation hypothesis simply refers to a general
discrimination against female work. cultural devaluation of women’s labor—a special
The finding of persistent wage effects of case of discrimination against women. In a fe-
occupational sex composition is socially impor- male occupation, all workers in the occupation
tant because it offers crucial support for com- are subject to the devaluation effect. By contrast,
parable worth policies. The comparable worth the specialized human capital hypothesis claims
movement has long identified occupational wage that the apparent wage penalties against female
inequality as a major source of gender inequity occupations arise mainly from occupational
in the labor market. The movement claims that differences in specialized training. In essence,
there is widespread valuative discrimination the same worker is expected to receive different
against female work, as suggested by research in wages for occupations associated with different

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Tony Tam, “Sex Segregation and Occupational Gender Inequality in the United
States: Devaluation of Specialized Training?” American Journal of Sociology 102:6 (May 1997), pp. 1652–1692,
published by the University of Chicago Press. The article printed here was originally prepared by Tony Tam
for Social Stratification: Class, Race, and Gender in Sociological Perspective, Second Edition, edited by David B.
Grusky, pp. 776–780. Copyright © 2001 by Westview Press.
 Tam—Why Do Female Occupations Pay Less? • 925

training requirements for specialized human

capital, which may be occupation specific, in- Weak Old Tests and Strong New
dustry specific, or firm specific. Evidence
Human capital may be general (e.g., reading The empirical tests necessary for adjudicating
skills and health) or specialized (e.g., knowledge between the two hypotheses are surprisingly dif-
of computer science). Although any investment ficult to achieve, as is the measurement of the
in human capital is costly, an investor in special- extent and degree of discrimination in general
ized human capital faces an additional opportu- (Heckman 1998). There are many traps that
nity cost because specialized skills by definition would invalidate a test or misguide the interpre-
have restricted applicability. For instance, a med- tation of a finding.1 Recently I have reexamined
ical doctor cannot expect to recoup the training the evidence for the devaluation hypothesis and
cost of attending medical school by becoming showed that there are strong reasons to question
a high school teacher, even if s/he may one day its validity (Tam 1997).
find teaching in high school the most personally The first reason is a crucial bias in the ev-
gratifying occupation. But a specialized skill also idential strategy. The establishment of general
tends to earn higher returns than does a general empirical support for the devaluation hypothesis
skill that requires the same amount of training essentially follows a “residual” approach because
time. The lower pay attached, on average, to fe- of the difficulty in developing any direct measure
male occupations may thus arise because such of devaluation in survey data. That is, the empir-
occupations require less substantial investments ical support for the hypothesis crucially hinges
in specialized human capital. The sources of sex on the existence of substantial and unexplained
segregation by specialized training requirement (i.e., residual) wage penalties against female oc-
are outside the purview of this paper, but one cupations. The more comprehensive are the the-
might hypothesize that (a) employers for occu- oretically relevant control variables and the larger
pations requiring specialized human capital may are the residual effects among workers with
avoid hiring women, (b) specialized occupations comparable characteristics, the stronger is the
discourage potential female recruits because the evidence in favor of the devaluation hypothesis.
work environment is often hostile (by virtue of Virtually all quantitative tests of the devaluation
employer or coworker behavior), and (c) women hypothesis adopt this strategy. Nevertheless, this
have relatively stronger incentive than men to evidential strategy is biased in favor of the de-
make training investments that can pay off in valuation hypothesis because of the potential
many lines of work. for omitted-variable bias and, most important,
The devaluation hypothesis of course does measurement errors in correlated control vari-
not deny the existence of market forces in wage ables (Tam 2000). Past tests and evidence have
determination. The key point of contention is been produced and interpreted without taking
whether gendered cultural bias has sufficiently any account of measurement error and thus fail
distorted the market’s wagesetting process to gen- to provide critical support for devaluation.
erate substantial, unexplained sex composition The second reason for skepticism is new
effects. The specialized human capital hypothesis counter-evidence that simultaneously contra-
claims that it does not and that the apparent sex dicts the devaluation hypothesis and confirms
composition effects can be explained by taking the specialized human capital hypothesis (Tam
account of occupational differences in special- 1997, 2000). This evidence uses the May 1988
ized human capital. This hypothesis is a plau- Current Population Survey (CPS) as the main
sible application of human capital theory but, source of data. Appended to the CPS are two
unfortunately, has never been properly tested in constructed variables: the sex composition
the literature. (PWOM) of three-digit census occupations, and
 926 • Part VIII: Gender Inequality

Table 107.1. Description of Variables

Variable Description Mean SD

Wage Hourly wage is the ratio of weekly earnings and usual weekly 9.679 5.497
hours.
SMSA 1 = residing in a standard metropolitan statistical area, 0 = .725 .446
otherwise.
South 1 = residing in the South, 0 = otherwise. .303 .459
West 1 = residing in the West, 0 = otherwise. .181 .385
Midwest 1 = residing in the Midwest, 0 = otherwise. .272 .445
Veteran 1 = ever in the armed forces, 0 = otherwise. .160 .366
Union 1 = union member or under union contract, 0 = otherwise. .187 .390
School Years of schooling completed. 14.028 2.632
Experience Potential work experiences, estimated by age – school – 5.1 17.702 12.177
Tenure Years with current employer. 6.779 7.721
Full-time 1 = usually work at least 35 hours a week, 0 = otherwise. .829 .376
PWOM Women as a proportion of workers in respondent’s occupation. .465 .324
SVP Years of specific vocational training for an occupation.2 1.977 1.611

Note: The data are for whites in the expanded May 1988 CPS. Most of the valid Ns are 23,019 or slightly less. Valid Ns for experi-
ence and tenure are 19,666.
1
For consistency, negative values of experience are recoded “0.”
2
From England and Kilbourne’s (1988) match of SVP (DOT, 4th ed.) with three-digit 1980 census occupations.

occupation-specific training time (SVP, mea- control for industry effects, respectively. This
sured in years) derived from the fourth edition of finding therefore contradicts the devaluation
the Dictionary of Occupational Titles (DOT). hypothesis at any reasonable level of statistical
Table 107.1 describes the key variables used in significance.
the analysis. The devaluation hypothesis implies that
Table 107.2 presents the core results that women’s work is devalued for all incumbents
show the extent to which occupation-specific (i.e., men and women alike) and via the same
(SVP) and/or industry-specific (INDUSTRY) cultural process. Columns 5–8 of table 107.2 ex-
training can explain PWOM effects. The first amine this implication. After controlling for spe-
major finding pertains to white female workers. cialized human capital (SVP and INDUSTRY),
The gross PWOM effect (–.255, not shown in the PWOM effect among white men becomes
the table) is substantially reduced (–.158) after practically zero (.002), as was the case for white
controlling for a set of basic variables. This women (–.007). Equally significant, the model
PWOM effect is still negative and substantial, with INDUSTRY alone (column 6) or SVP
apparently confirming a wage disadvantage alone (column 7) fails to fully account for the
against female occupations. However, after add- PWOM effect (–.094 and –.141, respectively)
ing occupational differences in two measures of among white men. These results therefore suggest
specialized human capital, column 4 of Table that the sex composition effects among men and
107.2 shows that there is virtually no occupa- women are attributable to gender-specific occu-
tional sex composition effect left to be explained. pational differences in specialized human capital.
Comparing the coefficients of PWOM across These two results strongly suggest that the
columns 1–4 of Table 107.2 reveals that SVP is PWOM effects are mediated by specialized
the key explanatory variable. After controlling human capital differences and that the sources
for SVP, the PWOM effects are dramatically of the wage effects of sex composition are gender
reduced to .011 and –.007 without and with specific. Both findings contradict the devaluation
 Tam—Why Do Female Occupations Pay Less? • 927

Table 107.2. Regressions of Log Hourly Wages on Demographic, Human Capital, and Industry Variables

White Women White Men

(1) (2) (3) (4) (5) (6) (7) (8)

PWOM −.158 −.187 .011# −.007# −.224 −.094 −.141 .002#

(.016) (.016) (.017) (.018) (.018) (.019) (.018) (.019)
SVP .088 .078 .065 .069
(.004) (.004) (.003) (.003)
INDUSTRY Yes Yes Yes Yes
(56.792) (50.145) (51.883) (54.429)
Adjusted R2 .403 .470 .440 .496 .452 .503 .482 .532

Note: Variables included in all models are South, West, Midwest, SMSA, veteran, union, school, experience and its
square, tenure and its square, and full-time. INDUSTRY denotes the control for 22 industries: construction,
education, financial/insurance/real estate, wholesale, retail trade, medical (except hospital), hospital, agriculture,
mining, manufacturing of durable goods, manufacturing of nondurable goods, postal, other transportation, other
utilities, private household service, business/repair, personal services (except private household), entertainment/
recreation, welfare/religious, other professional services, forestry/fisheries, and public administration/armed forces.
Nos. in parentheses are standard errors, except that F-statistics are reported for industry effects. N = 9,269 for (1) and
(2), 9,265 for (3) and (4), 10,383 for (5) and (6), 10,372 for (7) and (8). Data are for whites in the May 1988 CPS.

# Not significant at the .05 level; other presented coefficients are significant at the .05 level.

hypothesis. Moreover, both findings are remark- conducive to further opportunities for special-
ably robust across racial groups, as shown in Tam ized training, and only certain jobs will generate
(1997, p. 1674). There is absolutely no wage returns to the specialized skills of a worker. Not
penalty against female occupations: not among surprisingly, jobs are often the focus of career
women or men, not among whites or blacks.2 investment, just as human capital is. The capi-
These striking results are hard to recon- tal value of a job is high if it provides the neces-
cile with a devaluation story but are consistent sary gateway to the accumulation of specialized
with the specialized human capital hypothesis. human capital or the unique promotion path to
Occupational sex composition is just a handy other jobs that can make the most use of the skills
proxy for unobserved specialized human capital of a worker. The “positional capital” of a worker is
investment. In a related article (Tam 2000), I made up of the worker’s portfolio of ever-occupied
have further demonstrated that even seemingly jobs that can provide long-term returns in addi-
supportive findings for the devaluation hypoth- tion to income (Tam 1998). To the extent that
esis can be accurately predicted (i.e., numeri- male and female occupations offer different po-
cally reproduced) under the specialized human sitional capital, any discriminatory job allocation
capital hypothesis by assuming a modest level of by sex could have profound long-term impacts
measurement errors. In sum, the accumulated on women’s labor market outcomes even in the
evidence to date is remarkably strong for the absence of devaluation of female occupations.
specialized human capital hypothesis.
NOTES
This chapter draws on analyses and argu-
Positional Capital and Allocative ments that are presented in more detail elsewhere
Inequality (Tam 1997, 2000). The research presented here
was supported by Academia Sinica through the
Jobs are highly differentiated, interconnected, Organization-Centered Society Project and the In-
and form a social structure. Some jobs are more stitute of European and American Studies.
928 • Part VIII: Gender Inequality

1. For instance, it is poorly understood that mar- England, Paula, and Barbara S. Kilbourne. 1988.
ket prices are not determined by the preferences of Occupational Measures from the Dictionaries of
the average or majority of buyers and sellers, but set Occupational Titles for 1980 Census Detailed Oc-
at the margin by the final set of exchanges that align cupations (MRDF). ICPSR no. 8942. Ann Arbor,
supply and demand. The result is that discriminatory Mich.: Inter-University Consortium for Political
intentions, attitudes, and behavior can be pervasive and Social Research.
while market wages remain competitive (nondis- Heckman, James J. 1998. “Detecting Discrim-
criminatory) under a wide variety of scenarios. ination.” Journal of Economic Perspectives 12
2. As Tam (1997) has further demonstrated, these (2):101–116.
results are reinforced by the fact that there is a large Tam, Tony. 1997. “Sex Segregation and Occupa-
wage advantage, instead of disadvantage, for female tional Gender Inequality in the United States:
occupations among women working part-time. Also, Devaluation or Specialized Training?” American
see Tam (1997) for a discussion of various fallback Journal of Sociology 102:1652–1692.
defenses of the devaluation hypothesis, all of which ———. 1998. “Getting Ahead in the Labor Market:
can be shown to be untenable. The Positional Capital Approach.” Paper pre-
sented at the 14th World Congress of Sociology,
REFERENCES Montreal, Canada, July 26–August 1.
England, Paula, George Farkas, Barbara S. Kil- ———. 2000. “Occupational Wage Inequality
bourne, and Thomas Dou. 1988. “Explaining and Devaluation: A Cautionary Tale of Meas-
Occupational Sex Segregation and Wages: Find- urement Error.” American Journal of Sociology
ings from a Model with Fixed Effects.” American 105:1741–1760.
Sociological Review 53:544–558.
 Blau—The Sources of the Gender Pay Gap • 929

108 • Francine Blau

The Sources of the Gender Pay Gap

The sources of the gender pay gap can usefully occurred over the 1980s, and I will also consider
be divided into two broad sets of causes directly what we know about why the pace of change
related to gender: gender differences in qualifi- slowed in the 1990s.
cations and labor market discrimination against
women. Based on the available evidence, my
own view is that both of these factors play a role The Sources of the Gender
in producing a gender pay gap. And, changes in Wage Gap
each dimension—that is, in the extent of gender There are many possible concepts that can mea-
differences in qualifications and in the extent sure pay. When economists consider pay, they
of labor market discrimination—have played a generally focus on wages, or pay per hour. This
role in the decrease in the gender pay gap we differentiates the pay measure from how many
have observed over time. How women fare in hours an individual worked. This is always im-
the labor market may also be affected by broader portant but may be especially so for gender com-
economy-wide or labor market-wide forces, such parisons since women tend to work fewer hours
as the relative demand for workers in various oc- than men. Thus, my focus here is on wages and
cupations and industries. While such factors can the gender wage gap.
be quite important, I give them less emphasis
here, given my primary focus on the role of labor
market discrimination. Gender Differences in Qualifications
According to published government statistics, One of the major gender differences in qualifi-
in 2010, women fulltime workers earned about cations, and the one that has received the most
81 percent of what men earned, still a substantial attention from economists, is the gender differ-
gender gap in pay. Nonetheless, this represents a ence in the amount and type of human capital
large increase over the 64 percent of men’s pay investments. Jacob Mincer and Solomon Po-
that women full-time workers earned in 1980.1 lachek have done especially important work in
In this essay, I first describe the potential sources highlighting the role of labor market experience
of the gender pay gap that economists have iden- in explaining the gender wage gap.2 Given the
tified. I then consider the empirical evidence on traditional division of labor by gender in the
the role that each has played in producing the family, in which men have primary responsibility
gender pay gap and in explaining the substan- for earning the income and women have primary
tial increase in women’s relative pay that has oc- responsibility for homemaking and child care,
curred in recent decades. The sharpest increase women tend to accumulate less labor market

Francine Blau. “The Sources of the Gender Pay Gap,” in The New Guilded Age: The Critical Inequality Debates
of Our Time, edited by David Grusky and Tamar Kricheli-Katz (Stanford: Stanford University Press, 2012),
pp. 189–208, 271–275. Copyright © 2012 by the Board of Trustrees of the Leland Stanford Jr. University. All
rights reserved. Used with the permission of Stanford University Press, www.sup.org.
930 • Part VIII: Gender Inequality

experience than men. Further, because women women still exists and is perhaps all the more
anticipate shorter and more discontinuous work difficult to combat because the remaining dis-
lives, they have lower incentives to invest in crimination is likely to be less overt and more
market-oriented formal education and on-the- subtle than in the past.5
job training. Their resulting smaller human cap- A useful place to begin the consideration of
ital investments lower their earnings relative to labor market discrimination is with a definition:
those of men. The traditional division of labor labor market discrimination exists when there
may also disadvantage women if the longer are wage or occupational differences between
hours women spend on housework decrease the men and women that are not accounted for by
effort they put into their market jobs compared productivity differences.6
to men, controlling for hours worked, and hence It is important to emphasize that, in defining
also reduce their productivity and wages.3 The labor market discrimination, it is not a question
traditional division of labor could potentially of whether men and women in the labor market
lower the income of women relative to men in are equally well qualified, on average, but rather
a variety of other ways by impacting women’s whether the wage gap that we observe between
hours worked, commitment to their jobs, will- them exceeds what could be expected based on
ingness to undertake job-related travel, etc. any gender differences in qualifications. Such
To the extent that women choose occupa- “unexplained” gender differences in wages are
tions for which on-the-job training is less im- a departure from what would be expected from
portant, gender differences in occupations are the operation of a purely competitive labor mar-
also expected. Women may especially avoid ket with no discrimination. In such a compet-
jobs requiring large investments in skills that are itive labor market, employers would, in their
unique to a particular enterprise, because the re- hiring, promotion, training, and wage decisions,
turns to such investments are reaped only as long care about nothing other than the qualifications
as one remains with that employer. At the same of workers.
time, employers may be reluctant to hire women Why might labor market discrimination
for such jobs because the firm bears some of the exist? There are a number of interesting models
costs of such firm-specific training, and fears not developed by economists that analyze the reasons
getting a full return on that investment.4 for and the consequences of such discrimination.
I do not have space to go into them in detail,
but I would like to give a flavor of such models
Labor Market Discrimination to suggest some of the reasons that labor market
I have no doubt that human capital is indeed discrimination might exist. Economic modeling
important in explaining the gender wage gap, in this area was spearheaded by the pathbreaking
and that changes in gender differences in human work of Gary Becker, who received a Nobel Prize
capital have played an important role in explain- in economics, in part for his theory of discrim-
ing trends over time in the gap. (I review some ination. In the Becker model, discrimination is
evidence on this later.) However, in my view: due to what he terms “tastes for discrimination”7
(i) Labor market discrimination against women and what Ronald G. Ehrenberg and Robert S.
has also played an important role in causing the Smith have expressed in more everyday lan-
gender wage gap. (ii) Decreases in discrimina- guage as “personal prejudices” against a partic-
tion against women have likely played a role and ular group;8 in this case, it would be prejudices
contributed to the decline in the gender wage against women.
gap, including over the 1980s when the largest Becker postulated that discrimination was
declines occurred, (iii) Even though there have rooted in the desire to maintain social distance
been important and significant decreases in the from the group. He originally developed his
extent of discrimination, discrimination against model to explain race differences in labor market
 Blau—The Sources of the Gender Pay Gap • 931

outcomes, but these ideas are readily applicable One answer to this question is that discrim-
to women (as well as to other groups potentially ination may simply result from a lack of such
facing discrimination). It might seem odd to al- competitive pressures in the economy. Becker
lege that males in the labor force would be prej- hypothesized that, on average, employer dis-
udiced against women since men and women crimination would be less severe in competitive
often live together in the same families. But I than in monopolistic industries, and there is
think the issue here is one of socially appropri- some empirical support for this prediction. It
ate roles. A male manager might be delighted to has also been suggested that monopsony power
have a female secretary but object to having to by employers in the labor market plays a role
deal with a software engineer who is a woman. in producing and perpetuating the gender pay
The members of a construction crew might be differential.
perfectly happy to have a woman bring them One way in which firms may gain monop-
their food during the lunch break, but not if she sony power over workers is due to workers’ im-
stood next to them doing wiring or plumbing. perfect information about job opportunities and
Becker postulated that discrimination could the resulting necessity for workers to engage in
be due to such prejudices on the part of employ- job search. In a competitive labor market with
ers, in which case employers engage in hiring perfect information, even a slightly higher wage
or wage discrimination because of their own at another firm will induce workers to move to
disutility of employing women. But it could also that better opportunity. However, when infor-
be coworkers who have such prejudices, as my mation is imperfect, workers must search among
construction example implies. And the Becker employers for a good job match, thus incurring
model postulates that it could also be customers “search costs.” Since search is costly, workers will
or clients who have such prejudices. In the case be less mobile across firms than they would be
of coworker or customer discrimination, em- if information was perfectly and costlessly avail-
ployers discriminate against women due to the able, and it will take larger wage premiums at
higher costs or lower revenues associated with other firms to bid them away. Thus, the presence
hiring them. of search costs can give employers a degree of
One problem that economists have identi- monopsony power over workers. A model de-
fied with the model of employer discrimination veloped by Dan A. Black shows how such mon-
based on personal prejudices is that discrimina- opsony power could result in the persistence of
tion is not costless to the employer who forgoes discrimination over time.9 If we assume that
the opportunity to hire more of the lower-priced some employers discriminate against women and
female labor and less of the higher-priced male are not willing to hire them, we see that women
labor. Therefore, less discriminatory firms should will have higher search costs than men. As a
have lower costs of production. Such a compet- consequence, employers will exploit this greater
itive advantage would enable them to expand monopsony power over women and offer them
and drive the more discriminatory firms out of lower wages than men. Thus, when information
business in the long run. As the less discrimina- is imperfect and there are search costs, it is more
tory firms expand, the demand for female labor credible that employer discrimination can persist
would be increased and the male-female wage in the long run.10
gap would be reduced. If there were enough en- An effort to understand the persistence of
tirely nondiscriminatory firms to absorb all the discrimination in the face of competitive forces
women workers, the discriminatory wage gap was also a motivation behind the development
would be eliminated. Hence, the question is how of models of statistical discrimination. Such
discrimination, which represents a departure models assume a world of uncertainty and im-
from profit-maximizing behavior, can withstand perfect information for employers; thus differ-
the impact of competitive pressures. ences between groups in the expected value of
932 • Part VIII: Gender Inequality

productivity are expected to result in differences perceptions are correct. In that case, the concept
in the treatment of members of each group. So, of statistical discrimination provides a plausible
for example, employers may believe that women reason for the existence and persistence of wage
are more likely to quit their jobs or to put their and occupational differences between men and
families ahead of their work responsibilities than women with similar measured characteristics;
men with similar observed characteristics (edu- this is in itself a useful contribution even if it
cation, experience, etc.). As a consequence, they does not shed light on the persistence of eco-
may pay women less, exclude them from jobs nomic discrimination in the long run. But if
requiring substantial firm-specific training, or the employer views are correct, is it appropri-
deny them promotions.11 They will discriminate ate to consider this a form of “discrimination”
against all women because they cannot easily dis- in any sense? From a normative perspective, the
tinguish the more from the less career-oriented answer may be yes, to the extent that basing
women. Since the real or perceived average gen- employment decisions on a characteristic like
der differences that underlie statistical discrimi- sex—a characteristic that the individual cannot
nation against women in the labor market tend change—could be viewed as inequitable. Indeed,
to stem from the traditional division of labor the practice of judging an individual on the basis
in the family, this constitutes another route by of group characteristics rather than upon his or
which traditional gender roles within the family her own merits seems the very essence of ste-
adversely affect women’s labor market outcomes. reotyping of discrimination. Such behavior is
It has been argued that such statistical dis- certainly not legal under the antidiscrimination
crimination, making decisions on the basis of the laws and regulations, for example.
average characteristics of the group, is consistent Now consider the situation where employer
with profit maximization and can thus persist in perceptions are incorrect. If statistical discrimi-
the face of competitive forces. However, Dennis nation is accompanied by feedback effects, this
J. Aigner and Glen G. Cain contend that such may indeed be a credible source of persistent
models are no more convincing in explaining the discriminatory wage differences between men
persistence of discrimination than models based and women. For example, if employers incor-
on employers’ tastes for discrimination.12 To the rectly expect that women are more likely to quit
extent that employers’ views are accurate, the their jobs, they may respond by giving women
lower expected productivity of women will re- less firm-specific training or assigning them to
duce their wages, but women as a group will be dead-end jobs. Faced with few incentives to re-
paid their expected productivity. This does not main on the job, women may respond by exhib-
constitute labor market discrimination as econ- iting exactly the higher turnover that employers
omists define it, that is, pay differences that are expect.
not accounted for by productivity differences. Although the foregoing considerations help
Presumably they would make a similar argu- to shed light on the persistence of employer
ment regarding hiring and promotion decisions. discrimination in the long run, it should be
Moreover, they note that when employer beliefs acknowledged that there is probably no sim-
regarding average differences are false or erro- ple answer to the question of why competitive
neous, economic discrimination clearly exists, forces have not entirely eliminated discrimina-
but in their view discrimination based on such tion against women and other groups. This has
misperceptions is even less likely to persist in the led some economists to doubt that labor market
long run than discrimination based on tastes. discrimination is responsible, in whole or part,
However, such models may have greater use- for gender inequality in economic rewards. Yet
fulness in explaining the persistence of gender it is important to recognize that the phenom-
differentials in wage than Aigner and Cain sug- enon that we seek to understand is intrinsi-
gest. Consider first the situation where employer cally complex. Further, the various models of
 Blau—The Sources of the Gender Pay Gap • 933

discrimination, each emphasizing different mo- not going so far as to suggest that the division of
tivations and different sources of this behavior, labor in the family is solely due to discrimination
need not be viewed as alternatives. Rather, each in the labor market, but I am arguing that the
may serve to illuminate different aspects of this traditional division of labor in the family is rein-
complex reality. forced by gender inequities in the labor market.13
Another example I would give of feedback
effects relates to occupational choice. The first
Feedback Effects two women to serve on the U.S. Supreme Court,
So far I have for the most part considered gender Sandra Day O’Connor and Ruth Bader Gins-
differences in human capital and discrimination burg, graduated from law school in the 1950s.
as if they are two completely distinct sources of They both graduated at or near the top of their
the wage gap. However, the case is more compli- class. However, neither was able to obtain a
cated than that. These two sources of the wage clerkship with a justice of the Supreme Court or
gap can reinforce each other and this is impor- with a distinguished judge at another level of the
tant because it may make it difficult to separate judiciary, and neither was able to obtain a job at
the impact of these two components when we a major law firm. They both pursued stopgap ap-
look at the empirical evidence. On the one hand, proaches for a while and eventually did succeed.
the gender division of labor in the family may But the women of that time were likely aware of
result in women investing less in human capi- these sorts of difficulties, and hence many might
tal, thereby lowering their wages and adversely not have gone to law school at least partly for
affecting their occupations in the labor market. that reason. Then when we analyze the sources
The statistical discrimination theory fits in well of their wage gap, we would attribute some of it
here too, with employers treating all women as to women not having the same advanced train-
if they were pursuing a traditional division of ing as men. Yet some of the causation may go in
labor in the family; this generates discrimination the opposite direction: anticipating labor market
against women as a group in the labor market. discrimination, women may be less likely to in-
On the other hand, there is also an effect in vest in their human capital than men—particu-
the opposing direction. This is sometimes called larly training to enter predominately male fields.
a feedback effect, where gender differences in
labor market outcomes caused by discrimina-
tion feed back onto the gender division of labor Wage Structure
in the family and reinforce it. Thus, it has been Both the human capital and discrimination
pointed out that even a small amount of dis- explanations for the gender pay gap might be
crimination at the beginning of the career can considered “gender-specific” in nature—that is,
accumulate to have a large impact on the gender focused on gender differences in either quali-
wage gap as this initial discrimination influences fications or treatment. Analyses of gender dif-
the division of labor in the family. For example, ferentials have traditionally tended to emphasize
if couples feel that one partner needs to drop these types of gender-specific factors. However,
out of the labor force for a while when the chil- important work by Chinhui Juhn, Kevin M.
dren are small or that one partner should be the Murphy, and Brooks Pierce on trends in the race
secondary worker, that is, give priority to the pay gap over time in the United States has high-
other partner’s career, then it makes sense for the lighted an additional factor, overall wage struc-
low-wage partner to be the one to drop out or to ture, which is also relevant to analyzing gender
be the secondary worker. Thus, if labor market differences in wages.14 Wage structure is the array
discrimination lowers the wages of women rela- of prices determined for labor market skills and
tive to their husbands, this works to reinforce the the rewards to employment in particular sectors.
traditional division of labor in the family. I am This insight has been applied by Lawrence M.
934 • Part VIII: Gender Inequality

Kahn and me to analyze international differences

and changes over time in the gender wage gap.15 Empirical Evidence of
The human capital model suggests that Discrimination
men and women tend to have different levels Evidence from Statistical Analyses
of labor market qualifications (especially work Now let us turn to empirical estimates of dis-
experience) and to be employed in different oc- crimination. How do we generally measure dis-
cupations and perhaps in different industries. crimination and how good are such estimates?
Discrimination models too suggest that women The typical approach to analyzing the sources of
may be segregated into different sectors of the the gender wage gap is to estimate the statistical
labor market. This implies that the returns to relationship between wages and qualifications
skills and the size of the wage premiums that or productivity-related characteristics for men
workers receive for employment in high-wage and women. The gender wage gap can then be
occupations and industries potentially play an statistically decomposed into two components:
important role in determining the gender wage one that is due to gender differences in measured
gap. All else equal, the larger the returns to skills characteristics and the other that is unexplained
and the larger the rewards received by individu- and potentially due to discrimination. My read-
als in predominantly male occupations and in- ing of such empirical studies is that they provide
dustries, the larger will be the gender wage gap. evidence consistent with both human capital
While gender-specific factors and wage struc- differences and labor market discrimination in
ture each potentially play a distinct role in af- explaining the gender wage gap and actually give
fecting the gender wage gap, they are likely to a bit of a quantitative edge to discrimination
interact. For example, since the 1970s, the labor rather than human capital.
market returns to skills such as education, spe- Accepting that traditional statistical estimates
cialized training, and experience have risen in the of discrimination are imperfect, it is nonetheless
United States and in many other countries, likely of interest to see what can be learned from them.
due in part to technological change, including After discussing results from one study that is
computerization. To the extent that the prices typical of this kind of work, based on research
of skills for which women have a relative deficit that I did with Lawrence Kahn,17 I will then con-
have risen, such changes in wage structure will sider information from alternative approaches,
raise the gender wage gap. However, technologi- which tend to support the central conclusion
cal change itself is not likely to have gender neu- from the statistical analyses: there is evidence
tral effects on labor demand, given occupational of discrimination against women in the labor
and industrial segregation patterns by gender. So, market.
for example, it is likely that computerization has The study that Kahn and I did is based on
reduced the demand for blue-collar production data from the Panel Study of Income Dynamics
labor and therefore lowered the relative demand (PSID), a large, nationally representative data
for sectors where men are disproportionately set that includes information on actual labor
represented.16 market experience, which, as suggested by my
Thus, changes in wage structure and in the discussion of the human capital explanation, is
favorableness of demand and supply for women a crucial variable in analyzing the gender wage
workers are factors affecting trends over time in gap. The sample is restricted to full-time work-
the gender wage gap. To the extent possible, their ers in order to focus on women and men whose
effects, along with changes in the relative qual- commitment to employment is as similar as pos-
ifications of men and women, must be netted sible. Figure 108.1 shows three measures of the
out in order to infer changes in the extent of female-male wage ratio for each of three years:
discrimination against women. 1979, 1989, and 1998. The “unadjusted” ratio
 Blau—The Sources of the Gender Pay Gap • 935

Figure 108.1. Unadjusted and Adjusted Female-Male Wage Ratios, Full-Time Workers, 1979–1998

Source: Constructed from data presented in Francine D. Blau and Lawrence M. Kahn, “The US Gender Pay Gap in the
1990s: Slowing Convergence,” Industrial and Labor Relations Review 60, no. 1 (2006).

is based on the observed mean wages for each control for the human capital variables. Further
group. The ratio “adjusted for human capital adjusting for occupation, industry, and union
only” controls for gender differences in actual membership, in addition to human capital,
labor market experience, education in years, and raises the gender ratio to 81.6 percent. However,
whether the individual has a college diploma or I would again point out there is still a sizeable
an advanced degree. The ratio “adjusted for all remaining gap—18.4 percent, despite fairly de-
variables” additionally controls for gender dif- tailed controls for industry and occupation.
ferences in occupation (nineteen categories), in- Figure 108.1 indicates that the unadjusted
dustry (twenty-six categories), and unionization. wage ratio increased over time, particularly over
The “unexplained gap” from each of the latter the 1980s when it rose by over 11 percentage
two estimates provides suggestive information points to 74.5 percent, with a smaller further rise
about the extent of labor market discrimination. of 5 percentage points to 79.7 percent in 1998.18
To establish a baseline and illustrate the First, I shall consider how the determinants of
findings of this type of analysis, let us begin by the gender wage gap differed between the most
looking at the ratios for 1979. We see that the recent year, 1998, and 1979 and then look at
unadjusted wage ratio is 63.2 percent, mean- the question of the determinants of the changes
ing that the wages of women full-time workers in the gender wage gap over the 1979 to 1998
were about 63 percent of male full-time workers’ period, focusing particularly on the reasons for
wages in that year. Controlling for the human and significance of the slowing convergence in
capital variables only, the gender wage ratio the gender gap over the 1990s compared to the
rose to 70.8 percent. This shows that gender 1980s.
differences in human capital were indeed quite A striking difference between 1979 and 1998
important. However, women still earned 29.2 is that, in the latter year, adjusting for gender
(= 100 – 70.8) percent less than men when we differences in the human capital variables does
936 • Part VIII: Gender Inequality

not increase the gender ratio by very much: the How do we explain the decrease in the gen-
unadjusted ratio of 79.7 percent increases only der gap between 1979 and 1998, and why was
slightly, to 81.9 percent, when we control for convergence slower in the 1990s than the 1980s?
human capital. Kahn and my detailed analy- Kahn and I found that convergence in men’s and
ses indicate that one reason for the smaller role women’s human capital was an important factor
played by measured human capital in 1998 is in the narrowing of the gender wage gap, and
that women narrowed the gap in prior full-time that such improvements had a comparable effect
experience from 6.6 years in 1979 to 3.5 years in both the 1980s and the 1990s.20 In the 1980s,
in 1998.19 The remaining male advantage in ex- women’s human capital upgrading consisted
perience was largely offset by a female advantage entirely of rising experience. In the 1990s, de-
in education. Although men had an edge in the creases in the gender gap in experience contrib-
incidence of college degrees in the earlier years, uted considerably less to wage convergence, but
by 1998 the incidence of college degrees was rising relative educational attainment of women
slightly (1.2 percentage points) higher among played a much larger role.
women than men in our sample of full-time Occupational upgrading of women also con-
workers. This in turn reflects faster increases tributed to women’s relative wage gains in both
in female than in male educational attainment decades, as women moved out of clerical and ser-
in the population as a whole, such that, among vice jobs and into professional and managerial
younger cohorts, women are now more likely to employments, and as men moved out of (or lost)
be college graduates than men. Another factor craft and operator jobs, particularly in the 1980s.
contributing to the female advantage in educa- Deunionization, or a decline in the share of the
tional attainment in our sample is the positive labor force that is unionized, reduced the gender
selection on education of women into employ- wage gap in both decades as well, because men,
ment in general and into full-time employment who are more likely than women to be union-
in particular. In contrast, when we add controls ized, lost union jobs at a faster pace than women.
for industry, occupation, and unionism, the gen- The impact of these factors—occupational gains
der ratio increases substantially to 91.1 percent. for women and deunionization—was greater
This indicates that gender differences in these in the 1980s than in the 1990s. Thus, slowing
variables, particularly industry and occupation, convergence in sectoral location in the 1990s is
remained a substantial source of gender differ- part of the explanation for the slowdown in wage
ences in wages in the late 1990s. convergence.
In terms of the unexplained gap, we find that However, the main reason for the slower con-
it fell over the 1979-to-1998 period as a whole. vergence in the gender gap in the 1990s than in
The unexplained gap fell from 29.2 percent in the 1980s (at least in an accounting sense) is that
1979 to 18.8 percent in 1998, adjusting for the “unexplained gap” decreased by much less in
human capital variables only, and from 18.4 the 1980s than in the 1990s. This may readily
percent to 8.9 percent, when all variables are be seen in Figure 108.1, which shows that the
controlled for. This may correspond to a reduc- adjusted wage ratio increased substantially over
tion in discrimination over this period, though, the 1980s, but remained roughly constant over
as we have seen, the unexplained gap can reflect the 1990s.
other factors, and we shall note some of them in What is the significance of this change? Per-
greater detail ahead. In both 1979 and 1998, the haps a reasonable starting point is to inquire
contribution of the unexplained gap to the gen- about the reasons for a decline in the unex-
der differential was far more important that the plained gender wage gap in general. Such a shift
human capital variables, suggesting at least the may reflect a decrease in labor market discrim-
possibility that discrimination was a more im- ination against women, but also an upgrading
portant factor than differences in human capital of women’s unmeasured labor market skills, or
in accounting for the gender wage gap. a shift in labor market demand favoring women
 Blau—The Sources of the Gender Pay Gap • 937

over men. This implies that slower convergence then one might expect to find less wage conver-
in the unexplained gap in the 1990s than in gence with men at the top of the distribution
the 1980s could be due to a smaller decline in than at other points. We find some evidence that
discrimination against women in the 1990s; a this is the case.
slower improvement in women’s unmeasured
qualifications relative to men’s in the 1990s; or
less favorable demand shifts for women in the Other Evidence of Discrimination
1990s. Kahn and I present evidence that each Given the problems with traditional statistical
of these factors appear to have played a role in studies of discrimination, some economists have
explaining the observed trends, although it is not taken different approaches to the question. First,
possible to specifically apportion a share to each let me consider two studies that have applied
factor.21 This suggests that the entire difference traditional statistical techniques to especially
between the two decades is not necessarily due homogeneous groups and employed extensive
to trends in the extent of discrimination against controls for qualifications, thus minimizing the
women, but that this factor did play a role. effect of gender differences in unmeasured pro-
It might at first appear unlikely that labor ductivity characteristics. Mary C. Noonan, Mary
market discrimination against women decreased E. Corcoran, and Paul Courant studied two co-
by more in the 1980s than in the 1990s because horts of graduates of the University of Michigan
civil rights legislation passed in 1991 made the Law School, five and fifteen years after gradu-
legal environment more favorable toward an- ation; the first graduated between 1972 and
tidiscrimination lawsuits in the 1990s than it 1978 and the second between 1979 and 1985.
was in the 1980s, by making it more profitable The results for the two cohorts were quite sim-
for private law firms operating on a contingency ilar. The gap in pay between women and men
fee basis to represent defendants in job bias law- was relatively small at the outset of their careers,
suits. This contributed to a more rapid growth but fifteen years later, women graduates earned
in such lawsuits over the 1990s.22 Nonetheless, if only about 60 percent as much as men. Some of
women’s labor force commitment changed more this difference reflected choices that workers had
dramatically in the 1980s, and we find some ev- made, including the propensity of women law-
idence consistent with this, it is possible that yers to work shorter hours. But, even controlling
employers’ perceptions of women’s labor force for current hours worked, as well as an extensive
commitment also changed more dramatically list of worker qualifications and other covari-
in the 1980s. If so, one of the possible bases ates, including grades while in law school, and
for statistical discrimination against women detailed work history data, such as years prac-
may have eroded faster in the 1980s than in the ticed law, months of part-time work, and type
1990s. and size of employer, a male advantage of 11
An additional scenario whereby discrimina- percent remained.23 In a similar vein, Catherine
tion could have narrowed more slowly in the J. Weinberger examined wage differences among
1990s is related to the glass ceiling hypothesis. recent college graduates in 1985. Her controls
The so-called glass ceiling problem refers to the included narrowly defined college major, college
explicit or, more likely, subtle barriers that in- grade point average, and specific educational
hibit women’s progress at the highest echelons. institution attended. She found an unexplained
If there is indeed such a problem, it may have pay gap of 10 to 15 percent between men and
had a greater negative impact on women in the women.24 While one must be cautious in inter-
1990s than in the 1980s, as women’s 1980s gains preting the unexplained gaps identified in these
placed more of them into the higher-level posi- two studies as entirely due to discrimination, for
tions where glass ceiling barriers might hinder the reasons we discussed previously, these find-
their further upward progression. If women are ings are consistent with discrimination against
indeed increasingly constrained by glass ceilings, highly educated women.
938 • Part VIII: Gender Inequality

A second set of studies used an experimen- added profits fostered by regulation primarily
tal approach. David M. Neumark analyzed the with men. It was thus men who lost the most in
results of a hiring “audit” in which male and the shift to deregulation. And, Black and Eliza-
female pseudo-job seekers were given similar beth Brainerd find that increasing vulnerability
resumes and sent to apply for jobs waiting on to international trade reduced estimated gender
tables at the same set of Philadelphia restau- wage discrimination in concentrated industries,
rants.25 In high-priced restaurants, a female ap- again as predicted by Becker’s model.31
plicant’s probability of getting an interview was Finally, additional evidence on discrimina-
40 percentage points lower than a male’s and her tion comes from court cases. A number of em-
probability of getting an offer was 50 percent- ployment practices that explicitly discriminated
age points lower. A second study examined the against women used to be quite prevalent, in-
impact of the adoption of “blind” auditions by cluding marriage bars restricting the employ-
symphony orchestras in which a screen is used ment of married women32 and the intentional
to conceal the identity of the candidate.26 The segregation of men and women into separate job
screen substantially increased the probability categories with associated separate and lower pay
that a woman would advance out of preliminary scales for women (e.g., Bowe v. Colgate-Palmolive
rounds and be the winner in the final round. The Co., 416 F.2d 711 [7th Cir. 1969]; IUE v. West-
switch to blind auditions was found to explain inghouse Electric Co., 631 F.2d 1094 [3rd Cir.
25 percent of the increase in the share of musi- 1980]). While many such overt practices have
cians in the top five symphony orchestras in the disappeared, court cases suggest that employ-
United States who were women, from less than ment practices still exist that produce discrim-
5 percent of all musicians in 1970 to 25 percent inatory outcomes for women.
in 1996).27 One high-profile case is the $54 million set-
Third, several studies have examined predic- tlement of a sex discrimination lawsuit against
tions of Becker’s discrimination model.28 As we Morgan Stanley in 2004, in which plaintiffs
have seen, Becker and others have pointed out claimed that the firm underpaid and did not
that competitive forces should reduce or elim- promote women. Allegations of sexist practices
inate discrimination in the long run because reportedly included claims that Morgan Stanley
the least discriminatory firms, which hire more withheld raises and desirable assignments from
lower-priced female labor, would have lower women who took maternity leave, and that it
costs of production and should drive the more condoned a hostile workplace where men made
discriminatory firms out of business. For this sexist comments and organized trips to top-
reason, Becker suggested that discrimination less bars and strip clubs.33 Another example of
would be more severe in firms or sectors that a major case was the $31 million settlement
are shielded to some extent from competitive in 2002 of sex bias charges against American
pressures. Consistent with this reasoning, Judith Express Financial Advisors, where it was also
K. Hellerstein, David Neumark, and Kenneth claimed that female employees were underpaid
Troske found that, among plants with high lev- and given fewer job opportunities. According
els of product market power, those employing to the plaintiffs, the company steered the most
relatively more women were more profitable.29 profitable accounts to male financial advisers
In a similar vein, Sandra E. Black and Philip E. and, corporate-wide, men were given preferen-
Strahan report that, with the deregulation of the tial treatment in training, mentoring, and pro-
banking industry beginning in the mid-1970s, motion. It was alleged that this was “the product
the gender pay gap in banking declined as men’s of a stereotype—pervasive both inside Amex
wages fell by considerably more than women’s and throughout the industry—that women do
(12 percent vs. 3 percent).30 This suggests that not have what it takes to succeed in the financial
during the period of regulation, banks shared the planning business and that only young males
 Blau—The Sources of the Gender Pay Gap • 939

have the temperament and the ability to achieve one to doubt that such segregation was entirely
aggressive sales.”34 As another example, in 2000, voluntary.
the U.S. Information Agency agreed to pay
$508 million to settle a case in which the Voice
of America rejected women who applied for Conclusion
high-paying positions in the communications The past thirty years have been a period of enor-
field. A lawyer representing the plaintiffs said mous change in the status and achievements of
that the women were told things like, “These women in the labor market. I have focused here
jobs are only for men,” or “We’re looking for a on one specific measure of outcomes, the gen-
male voice.”35 der wage gap. There is no doubt that the relative
A number of sex discrimination suits have earnings of women have increased dramatically
been filed against grocery chains as well. In during this period, with particularly sharp gains
1994, Lucky Stores, a major chain, agreed to in the 1980s. This development is extremely im-
a settlement of $107 million after the judge portant, not only because of the improvement it
found that “sex discrimination was the standard indicates in women’s labor market outcomes, but
operating procedure at Lucky with respect to also because higher wages for women encourage
placement, promotion, movement to full-time greater investment in human capital and higher
positions, and the allocation of additional hours” levels of commitment to the labor market, pav-
(Stender v. Lucky Stores, Inc., 803 F. Supp. 259; ing the way for further increases in their wages
[N.D. Cal. 1992]). Similar lawsuits against in the future.
several other grocery chains have also ended in While the gains are important and dramatic,
settlements, including Publix Super Markets they are not an invitation to complacency.
Inc. of Florida, which, in 1997, agreed to pay Women still earn less than their male coun-
$81.5 million to settle a sex discrimination law- terparts with similar measured characteristics.
suit that accused the chain of keeping women This “unexplained gap,” which is potentially
in dead-end, low-wage jobs. Although women due at least in part to labor market discrimina-
made up half the company’s workforce, less tion against women, has been diminished but
than 5 percent of its 535 store managers were by no means eliminated. Moreover, while the
women.36 decline in the unexplained gap was particularly
Some insight into the underrepresentation pronounced during the 1980s, there was no ev-
of women in higher-level positions found in idence of a further decline in the 1990s. To the
the cases cited previously is provided by a recent extent that this reflects slower decreases in the
study of eight years of data from an unidentified extent of labor market discrimination during
regional grocery chain on gender differences in the 1990s, it means that we must continue to be
job titles and wage rates.37 The authors find a concerned with inequitable treatment of women
pattern of gender differences in initial job as- in the labor market. . . . This means that fur-
signment and upward mobility within the firm ther change in the gender wage gap in the fu-
that “generally penalized women, even when ture also requires that we address the issue of
the analysis accounted for individuals’ charac- work-family conflict and continue to seek ways
teristics.”38 While one might dispute the reason that allow both women and men to successfully
for these differences, the authors found that job combine challenging careers with their family
segregation of women and men was dramatically responsibilities.
lower in the period after the company lost a dis-
crimination suit (1984) and reached a settlement NOTES
(1986) in which it initiated affirmative action 1. U.S. Department of Labor, Bureau of Labor
Statistics, “2011 Employment & Earnings Online,”
policies. This implies that it was possible to find
available at http://www.bls.gov/opub/ee/2011/
women interested in higher-level jobs, leading
cps/annual.htm; and Francine D. Blau, Marianne
940 • Part VIII: Gender Inequality

A. Ferber, and Anne E. Winkler, The Economics Black-White Wage Convergence,” in Workers and
of Women, Men, and Work, 6th ed. (Upper Saddle Their Wages, ed. Marvin Kosters (Washington, DC:
River, NJ: Pearson/Prentice-Hall, 2010), 142. AEI Press, 1991).
2. Jacob Mincer and Solomon Polachek, “Fam- 15. Francine D. Blau and Lawrence M. Kahn,
ily Investments in Human Capital: Earnings of “Wage Structure and Gender Earnings Differentials:
Women,” Journal of Political Economy 82 (March/ An International Comparison,” Economica 63 (1996,
April 1974, pt. 2). Supp.); “Gender Differences in Pay,” Journal of Eco-
3. Gary S. Becker, “Human Capital, Effort, and nomic Perspectives 14 (Fall 2000); and “The US Gen-
the Sexual Division of Labor,” Journal of Labor Eco- der Pay Gap in the 1990s: Slowing Convergence,”
nomics 3 (January 1985 Supp.). Industrial and Labor Relations Review 60 (October
4. Francine D. Blau and Lawrence M. Kahn, 2006).
“Race and Sex Differences in Quits by Young Work- 16. E.g., Bruce Weinberg, “Computer Use and
ers,” Industrial & Labor Relations Review 34 (July the Demand for Female Workers,” Industrial and
1981); Michael Ransom and Ronald L. Oaxaca, “In- Labor Relations Review 53 (January 2000).
trafirm Mobility and Sex Differences in Pay,” Indus- 17. Francine D. Blau and Lawrence M. Kahn,
trial and Labor Relations Review 58 (January 2005). “The US Gender Pay Gap in the 1990s”; Joseph G.
5. See Barbara F. Reskin, “The Proximate Causes Altonji and Rebecca M. Blank, “Race and Gender in
of Employment Discrimination,” Contemporary So- the Labor Market,” in Handbook of Labor Economics,
ciology 29 (March 2000); and Marianne Bertrand, ed. Orley Ashenfelter and David Card (The Nether-
Dolly Chugh, and Sendhil Mullainathan, “Implicit lands: Elsevier Science, 1999).
Discrimination,” American Economic Review 95 18. Note that because PSID wage data are
(May 2005). not available for 1999, the two periods are of
6. Gary S. Becker, The Economics of Discrimina- slightly unequal length. In Blau and Kahn, “The
tion, 2nd ed. (Chicago: University of Chicago Press, US Gender Pay Gap in the 1990s,” we calculate
1971; 1st ed. 1957). decade-equivalent changes for the 1980s and the
7. Ibid. 1990s, and this does not change the conclusions that
8. Ronald G. Ehrenberg and Robert S. Smith, I describe in the text here.
Modern Labor Economics: Theory and Public Pol- 19. Ibid.
icy, 11th ed. (New York: Pearson/Addison Wesley, 20. Ibid.
2011). 21. Ibid.
9. Dan A. Black, “Discrimination in an Equilib- 22. Blau, Ferber, and Winkler, The Economics of
rium Search Model,” Journal of Labor Economics 13 Women, Men, and Work.
(April 1995). See also Janice F. Madden, The Eco- 23. Mary C. Noonan, Mary E. Corcoran, and
nomics of Sex Discrimination (Lexington, MA: Lex- Paul Courant, “Pay Differences Among the Highly
ington Books, 1973). Trained: Cohort Differences in the Sex Gap in Law-
10. See, e.g., Francine D. Blau and Lawrence M. yers’ Earnings,” Social Forces 84 (December 2005).
Kahn, “Changes in the Labor Supply Behavior of 24. Catherine J. Weinberger, “Race and Gender
Married Women: 1980–2000,” Journal of Labor Eco- Wage Gaps in the Market for Recent College Gradu-
nomics 25 (July 2007). ates,” Industrial Relations 37 (January 1998). See also
11. Models of statistical discrimination have also Dan A. Black, Amelia M. Haviland, Seth G. Sand-
considered the situation in which female workers’ ers, and Lowell J. Taylor, “Gender Wage Disparities
productivity or behavior is less reliably predicted Among the Highly Educated,” Journal of Human
than that of male workers. Resources 43 (Summer 2008).
12. Dennis J. Aigner and Glen G. Cain, “Statis- 25. David M. Neumark, “Sex Discrimination in
tical Theories of Discrimination in Labor Markets,” Restaurant Hiring: An Audit Study,” Quarterly Jour-
Industrial and Labor Relations Review 30 (1977). nal of Economics 111 (August 1996).
13. See Stefania Albanesi and Claudia Olivetti, 26. Claudia Goldin and Cecilia Rouse, “Orches-
“Home Production, Market Production and the trating Impartiality: The Impact of ‘Blind’ Auditions
Gender Wage Gap: Incentives and Expectations,” on Female Musicians,” American Economic Review 90
Review of Economic Dynamics 12 (January 2009). (September 2000).
14. Chinhui Juhn, Kevin M. Murphy, and 27. A third study, again based on a hiring
Brooks Pierce, “Accounting for the Slowdown in audit, showed that employers discriminated against
 Blau—The Sources of the Gender Pay Gap • 941

mothers but not fathers; see, Shelley J. Correll, Ste- 33. “Sex Suit Costs Morgan Stanley $54M,” July
phen Benard, and In Paik, “Getting a Job: Is There a 12, 2004, http://www.dbsnews.com/stories /2004
Motherhood Penalty?,” American Journal of Sociology /07/12/national/main628907.shtml?tag=mncol;lst;l
112 (March 2007). and Patrick McGeehan, “Discrimination on Wall
28. Gary S. Becker, The Economics of St.? Run the Numbers and Weep,” New York Times,
Discrimination. July 14, 2004, Section C, pp. 1, 7.
29. Judith K. Hellerstein, David Neumark, and 34. Bureau of National Affairs, “American Express
Kenneth Troske, “Market Forces and Sex Discrim- Financial Advisors Reach $31 Million Agreement on
ination,” Journal of Human Resources 37 (Spring Sex Bias Charges,” Employment Discrimination Re-
2002). port 18, no. 9 (February 27, 2002): 251.
30. Sandra E. Black and Philip E. Strahan, “The 35. “Government to Pay $508 Million for Sex
Division of Spoils: Rent-Sharing and Discrimination Discrimination at U.S. Information Agency,” Federal
in a Regulated Industry,” American Economic Review Human Resources Week 6 (2000): 47.
91 (September 2001). 36. Ronette King, “Women Taking Action
31. Sandra E. Black and Elizabeth Brainerd, “The Against Many Companies,” Times-Picayune, April
Impact of Globalization on Gender Discrimination,” 27, 1997.
Industrial & Labor Relations Review 57 (July 2004). 37. Ransom and Oaxaca, “Intrafirm Mobility
32. Claudia Goldin, Understanding the Gender and Sex Differences in Pay.”
Gap (New York: Oxford University Press, 1990). 38. Ibid., 219.
942 • Part VIII: Gender Inequality

How Gender Intersects

109 • Margaret L. Andersen and Patricia Hill Collins

Why Race, Class, and Gender Matter

Race, class, and gender matter because they con- that stem from the societal configuration of
tinue to structure society in ways that value some race, class, and gender relations. This structural
lives more than others. Currently, some groups pattern affects individual consciousness, group
have more opportunities and resources, while interaction, and group access to institutional
other groups struggle. Race, class and gender power and privileges (Collins 2000). Within this
matter because they remain the foundations for structural framework, we focus less on compar-
systems of power and inequality that, despite our ing race, class, and gender as separate systems
nation’s diversity, continue to be among the most of power than on investigating the structural
significant social facts of people’s lives. Thus, patterns that join them. Because of the simul-
despite having removed the formal barriers to taneity of race, class, and gender in people’s
opportunity, the United States is still highly lives, intersections of race, class, and gender
stratified along lines of race, class, and gen- can be seen in individual stories and personal
der. . . . For years, social scientists have studied experience. In fact, much exciting work on the
the consequences of race, class, and gender in- intersections of race, class, and gender appears
equality for different groups in society. [We] ex- in autobiographies, fiction, and personal essays.
plore how race, class, and gender operate together We do recognize the significance of these indi-
in people’s lives. Fundamentally, race, class, and vidual narratives, but we also emphasize social
gender are intersecting categories of experience structures that provide the context for individual
that affect all aspects of human life; thus, they si- experiences.
multaneously structure the experiences of all peo- Second, studying interconnections among
ple in this society. At any moment, race, class, or race, class, and gender within a context of social
gender may feel more salient or meaningful in a structures helps us understand how race, class,
given person’s life, but they are overlapping and and gender are manifested differently, depending
cumulative in their effects. on their configuration with the others. Thus, one
[We] emphasize social structure in our efforts might say African American men are privileged
to conceptualize intersections of race, class, and as men, but this may not be true when their race
gender. We [suggest] the approach of a matrix of and class are also taken into account. Otherwise,
domination to analyze race, class, and gender. A how can we possibly explain the particular dis-
matrix of domination sees social structure as hav- advantages African American men experience in
ing multiple, interlocking levels of domination the criminal justice system, in education, and in

Margaret L. Andersen and Patricia Hill Collins. “Why Race, Class, and Gender Matter,” in Race, Class, and
Gender: An Anthology, Seventh Edition (Wadsworth, Inc., 2010), pp. 1, 5–7, 16. Copyright © 2010 by Wads-
worth, a part of Cengage Learning, Inc. Reproduced by permission, www.cengage.com/permissions.
 Greenman and Xie—Double Jeopardy • 943

the labor market? For that matter, how can we sexuality is not inherently more significant than
explain the experiences that Native American class and ethnicity.
women undergo—disadvantaged by the unique Given the complex and changing relation-
experiences that they have based on race, class, ships among these categories of analysis, [it is
and gender—none of which is isolated from the important to] ground analyses in the historical,
effects of the others? Studying the connections institutional context of the United States. Doing
among race, class, and gender reveals that divi- so means that race, class, and gender emerge as
sions by race and by class and by gender are not fundamental categories of analysis in the U.S.
as clear-cut as they may seem. White women, for setting, so significant that in many ways they
example, may be disadvantaged because of gender influence all of the other categories. Systems of
but privileged by race and perhaps (but not nec- race, class, and gender have been so consistently
essarily) by class. And increasing class differenti- and deeply codified in U.S. laws that they have
ation within racial-ethnic groups reminds us that had intergenerational effects, on economic, po-
race is not a monolithic category, as can be seen litical, and social institutions. For example, the
in the fact that white poverty is increasing more capitalist class relations that have characterized all
than poverty among other groups, even while phases of U.S. history have routinely privileged
some whites hold the most power in society. or penalized groups organized by gender and by
Third, the matrix of domination approach race. U.S. social institutions have reproduced
to race, class, and gender studies is historically economic equalities for poor people, women,
grounded. We [choose] to emphasize the in- and people of color from one generation to the
tersections of race, class, and gender as institu- next. Thus, in the United States, race, class, and
tional systems that have had a special impact gender demonstrate visible, long-standing, mate-
in the United States. Yet race, class, and gender rial effects that in many ways foreshadow more
intersect with other categories of experience, recently visible categories of ethnicity, religion,
such as sexuality, ethnicity, age, ability, religion, age, ability, and/or sexuality.
and nationality. Historically, these intersections
have taken varying forms from one society to REFERENCES
the next; within any given society, the connec- Collins, Patricia Hill. 2000. Black Feminist Thought:
Knowledge, Consciousness, and Empowerment.
tions among them also shift. Thus, race is not
New York: Routledge.
inherently more important than gender, just as

110 • Emily Greenman and Yu Xie

Double Jeopardy

A substantial body of literature argues for “in- Intersectional perspectives argue that the mean-
tersectionality” or the recognition that group ing of gender differs across racial groups and the
identities such as race and gender cannot be meaning of race differs for men and women. In-
understood in isolation from one another. tersectionality has made valuable contributions

Emily Greenman and Yu Xie. “Double Jeopardy? The Interaction of Gender and Race on Earnings in the
United States,” Social Forces 86:3 (2008), pp. 1218–1223, 1225–1232, 1236–1242. Reprinted by permission
of Oxford University Press.
944 • Part VIII: Gender Inequality

towards understanding the lives of minority While few researchers explicitly put forth this
women, who do not necessarily experience race assumption, it is invoked implicitly whenever
in the same way as minority men or gender in researchers draw inferences about “the race gap’’
the same way as white women (Browne and or “the gender gap” from studies that focus on
Misra 2003; McCall 2005). Yet few empirical only one or the other.
studies on earnings inequality by race and gender There is already a great deal of evidence that
have adopted this perspective (Brewer, Conrad calls into question the double jeopardy charac-
and King 2002). terization. The additivity assumption is prob-
Most existing studies of earnings inequality lematic because it ignores the ways in which
focus on either racial inequality among men or minority women’s experiences are unique, com-
gender inequality among whites, often over- parable neither to those of white women nor to
looking minority women (Malveaux 1986). those of men of the same race/ethnicity (King
Work that does address the earnings of minority 1988). While minority women of most ethnici-
women often still fails to consider race and ties are clearly disadvantaged, their earnings are
gender jointly. A common research design is to often still higher than one might predict based
compare minority women either to minority on their race and gender alone. Many studies
men of the same group in the gender-centered have shown that the earnings of black women are
approach (Blau and Beller 1988) or to white higher relative to those of white women than the
female workers in the race/ethnicity-centered earnings of black men relative to those of white
approach (Bound and Dresser 1999; Corcoran men (Blau and Beller 1988, 1992; Cancio, Evans
1999). While the two approaches avoid con- and Maume 1996; Carlson and Swartz 1988;
founding race and gender, they preclude direct King 1988; Marini 1989). While few stud-
comparisons between any two groups that dif- ies have considered other races and ethnicities
fer from one another in both race and gender. (Browne and Misra 2003), several have uncov-
To overcome this limitation, two alternative ered a similar pattern among various Hispanic
practices have emerged in the literature. The and/or Asian ethnic groups in relation to whites
first is to compare all gender-race combinations (Carlson and Swartz 1988; England, Christo-
simultaneously to one reference group, usually pher and Reid 1999; Xie and Goyette 2004).
white men (Corcoran and Duncan 1979; Farley Despite the suggestiveness of these findings,
1984); the second is to understand gender effects most previous research on race and gender earn-
by race and then, sequentially, to understand ings gaps has not attempted to address the ad-
race effects by gender (Kilbourne, England and ditivity assumption directly. Even when their
Beron 1994). empirical results show clear deviations from the
These two alternative practices have an ad- double jeopardy characterization, researchers
vantage over either the gender-centered ap- frequently pay little attention to the underlying
proach or the race/ethnicity-centered approach reasons for, and sometimes even fail to com-
in avoiding a strong assumption that minority ment on, the apparent interactions between race
women incur two earnings disadvantages ad- and gender (e.g., Blau and Beller 1992; Darity,
ditively, one associated with being female and Guilkey and Winfrey 1996; Padavic and Reskin
another associated with being nonwhite. Thus, 2002). To be sure, several studies have explored
there would be no intersection of race and gen- race/gender interactions in the earnings deter-
der, and the total disadvantage faced by minority mination process. For example, studies have fo-
women relative to white men would be simply cused on variation by both race and gender in
the sum of the gender penalty and the race pen- the rewards attached to human capital and job
alty. Deborah King (1988:47) aptly referred to characteristics (England, Christopher and Reid
the additivity assumption as “double jeopardy.” 1999; Kilbourne, England and Beron 1994;
 Greenman and Xie—Double Jeopardy • 945

McGuire and Reskin 1993), the effects of local but rather as two separate dimensions of inequal-
economic structure (McCall 2001), and earnings ity, each with unique structural determinants.
trends over time (Blau and Beller 1992; Cotter, For example, although black-white relations
Hermsen and Vanneman 1999). . . . While con- have epitomized racial relations in the United
tributing valuable evidence about intersectional- States due to their historical prominence, there
ity in the earnings determination process, none are also many other racial/ethnic groups with
of these earlier studies has made racial variation varying historical experiences. The number of
in the gender earnings gap its explicit focus. racial/ethnic groups is increasing and the bound-
Hence, the extent of racial variation in the aries between some are becoming blurred, due
gender earnings gap remains to be fully docu- in part to the increasing prominence of mul-
mented and understood. In numerous studies in tiracial groups. Gender, by contrast, is fixed at
sociology and economics, the interaction effects two categories, and its distribution is relatively
between race and gender have often been ap- unchanging. Although there is little difference
parent, but they have been treated more as em- in the distribution of gender across racial/ethnic
pirical nuances than as subjects to investigate. categories, it is possible that the social meaning
This study represents a systematic effort to study attached to gender may vary by race/ethnicity.
racial patterns in the gender earnings gap and There is something else unique to gender.
draw meaningful theoretical implications from Men and women, to a much greater extent than
such patterns. individuals of different races, are frequently part
of the same families—through either marriage,
cohabitation, having children together or some
Theoretical Issues combination of these. The family is fundamen-
We know that differences in productivity-related tal to the structure of gender relations. As has
factors—such as education and work experi- long been recognized in both economics and
ence—account for some of the observed differ- sociology, an adequate explanation of gender
ences in earnings by race/ethnicity and gender. inequality in the labor force therefore requires
While disagreement lingers concerning the in- the researcher to go beyond discrimination and
terpretation of the unexplained portion of the productivity-related attributes (i.e., human capi-
observed group differences, a long tradition tal) and to consider the role of the family (Becker
treats such residuals from regression analysis as 1973, 1974, 1991; Mincer and Polachek 1974;
measures of discrimination (Cole 1979). Past many others). The family must be considered
research has shown that, net of human capital in studies of gender inequality for several rea-
factors, gender differences in earnings are con- sons. First, because resources are typically pooled
siderably larger than racial differences between across family members, gender inequality in
whites and blacks (Durden and Gaynor 1998; earnings is not necessarily reflected in inequal-
Farley 1984). Does this mean that racial dis- ity in economic well-being among married or
crimination is smaller than gender discrimi- cohabiting adults.1 That is, an adult’s economic
nation? An answer of “yes” would contradict and social position in society is affected not only
common wisdom about structural inequalities in by how well he or she does in the labor market,
the United States, where racial barriers to some but also by whether and to whom he or she is
highly valued socioeconomic resources (such as married or partnered. Second, the traditional
quality education) appear much greater than division of labor within married-couple families
gender barriers. has placed responsibility for the domestic work
To answer this question, we need to concep- and child care primarily on the wife (Brines
tualize race and gender differentials not as two 1994), generating significant barriers to success
indicators of a single underlying phenomenon, in the labor market for married women (Budig
946 • Part VIII: Gender Inequality

and England 2001; Goldin 1997; Noonan 2001; than husbands (Brines 1994; Raley et al. 2006),
Waldfogel 1997). Gender roles within the family giving them a comparative advantage, rather than
are thus intimately connected with gender in- a disadvantage, in the labor market. Hence, the
equality in the workplace. extent to which role specialization theory is ap-
This interplay between family factors and plicable can vary substantially across families.
women’s labor force outcomes is at the heart of We conjecture that the applicability of role
neoclassical economic explanations for wom- specialization theory may vary across racial/eth-
en’s lower earnings. While there is a great deal nic groups. We give three reasons for this con-
of diversity in modern family structures, the jecture. The first is that certain family-related
neoclassical explanation primarily focuses on attitudes and practices are cultural and as
married-couple families with children (or on per- such are maintained more in some racial/eth-
sons who anticipate one day being part of such a nic groups than in others (Blee and Tickamyer
family). There are three key components to this 1995; Kane 2000; McLoyd et al. 2000; Rans-
explanation. First, it is assumed that economic ford and Miller 1983). For example, researchers
resources are a family-level utility that is shared have found that African Americans and Mex-
equally between the spouses (Becker 1973, 1974, ican Americans both express greater support
1991; Lundberg and Pollack 1993; Mincer and than whites for the idea that married women
Polachek 1974). Second, it is assumed that there should contribute financially to the family
is an efficiency gain in having one spouse (typi- (Blee and Tickamyer 1995; Taylor, Tucker, and
cally the husband) specialize in market produc- Mitchell-Kernan 1999)—despite the fact that
tion, and the other spouse (typically the wife) this and other research has repeatedly found
specialize in domestic production. This efficiency that African Americans and some groups of
gain is the result of the wage rate of the spouse Hispanics tend to express more traditional (that
who specializes in the market exceeding that of is, patriarchal) gender role attitudes than whites
the other spouse. Third, due to anticipation of with respect to other issues, such as women’s
withdrawing from the labor force and/or work- role in politics or their responsibility for home
ing parttime during childrearing, women tend to and family (Blee and Tickamyer 1995; Bol-
under-invest in their human capital and receive zendahl and Myers 2004; McLoyd et al. 2000;
lower returns on their work experience (Mincer Ransford and Miller 1983; Taylor, Tucker and
and Polachek 1974). Thus, neoclassical econom- Mitchell-Kernan 1999). On the other hand,
ics provides a theoretical framework that explic- despite such attitudinal differences, research
itly links gender inequality at work with gender has also shown that black husbands do a greater
inequality at home.2 Let us refer to this explana- share of housework than white husbands do
tion as “role specialization theory.” (Kamo and Cohen 1998; John and Shelton
The theory is silent on issues of race. How- 1997). Thus, the relationship between race and
ever, we know that the theory, even if it is true, gender role attitudes and practices is probably
can only be a crude approximation of a reality quite complex. Such differences are likely to af-
that is far more complicated. The problem is that fect men’s and women’s choices about work and
not all families meet the ideal conditions assumed family, including the extent to which they spe-
by role specialization theory. First, not all women cialize according to traditional gender norms.
or men intend to marry or have good prospects The second reason, which is widely rec-
to marry. Similarly, not all married couples have ognized in the literature, is the more difficult
or intend to have children, and in the absence economic circumstances facing many minority
of children the advantages to gender role spe- groups. The higher unemployment rates and
cialization are substantially reduced. Second, in lower earnings among many groups of mi-
a growing number of families wives earn more nority men undermine the applicability of role
 Greenman and Xie—Double Jeopardy • 947

specialization theory. For example, lower rates different racial groups. We also explore whether
of marriage in some minority communities, es- racial variation in the applicability of role spe-
pecially impoverished African American com- cialization theory contributes to the race/gender
munities, are partially attributable to the lack interaction in earnings—first by looking at how
of “marriageable” men with steady, well-paying the race/gender interaction varies across marital/
jobs (Lichter et al. 1992; Wilson 1996). Even parental status groups and then by gauging the
among married couples in economically disad- extent to which gender role specialization varies
vantaged minority groups, role specialization by race.
may not be an option if the husband does not
have sufficient earnings to be the primary, if not
the sole, breadwinner for the family (Padavic Methodology
and Reskin 2002). Furthermore, higher rates The relationship between race/ethnicity and
of marital instability in economically disadvan- gender in earnings determination is examined
taged minority groups (Ruggles 1997) would with the following methodology: For each racial
make specialization in domestic production, and or ethnic group k, we compute the quantity dk,
the degree of economic dependency it entails, which represents the difference between the mi-
a very risky strategy for a woman (Edin 2005; nority gender earnings gap and that of whites.
Smock, Manning and Gupta 1999). There are Previous literature suggests that dk will be posi-
thus several reasons to suspect that role special- tive for some racial groups, but it is not known
ization theory may apply better to middle-class how generally this is true. Although there is no
whites than to economically disadvantaged mi- theoretical reason to believe that dk may be nega-
nority groups. tive for any group, such a relationship is possible
Third, it has been well documented that most and cannot be ruled out a priori. In addition to
Asian American groups actually attain higher the unadjusted dk, we compute dk after adjust-
average economic status than whites (Xie and ing for earnings-relevant characteristics. These
Goyette 2004). However, most Asian Americans include education, experience and region.
are recent immigrants or children of immigrants, We next examine dk across subpopulations.
and as newcomers to the United States, attaining Role specialization theory is a theory of the
economic security is a high priority. Thus, Asian family. If it is to explain racial variation in the
Americans’ family-level strategies for economic gender earnings gap, there should be a stronger
adaptation may render role specialization less interaction between race/ethnicity and gender
applicable to Asian Americans than to whites. among the married than among the unmarried.
This study examines gender inequality in For this reason, the sample is disaggregated by
earnings across all major racial and ethnic mi- marital status and dk is recomputed. Significant
nority groups in the United States, while previous differences in dk by parental status are also tested.
studies have examined only one or two groups at
a time. From the previous literature, one expects
a positive interaction between race and gender Data
for African American women (and a few indi- Data come from the Public Use Micro Sample
cations of a similar effect for certain groups of of the 2000 Census. PUMS provides the only
Asian American and Hispanic women), but it is data with a large enough sample size to allow
unknown whether this pattern may hold for mi- the study of smaller racial and ethnic minority
nority groups more generally. A systematic met- groups. In order to get desirable sample sizes
ric measures the extent to which the effects of for each racial group, the analytical sample is
race and gender deviate from the assumption of comprised of the following: a 10 percent sam-
additivity, which facilitates comparisons between ple of mono-racial whites from the 1% PUMS,
948 • Part VIII: Gender Inequality

all mono-racial blacks from the 1% PUMS, and racial group for men and women, respectively.
all other groups from the 5% PUMS. When ap- For both sexes, the highest-earning groups are
propriate, the data are weighted according to the Chinese, Japanese, Koreans and Indians, while
inverse probability of being in the sample. the lowest-earning group is Native Americans.
Race is measured with a system of 19 While only 4 out of 18 minority groups have
mutually exclusive categories. In addition higher average earnings than whites among
to non-Hispanic whites, blacks and Native men, the corresponding figure is 9 out of 18
Americans, the larger Asian and Hispanic eth- for women. Column 3 gives the female-to-male
nic groups are treated as distinct categories. earnings ratio within each racial group. While
The 2000 U.S. Census data identify bi-racial white women make about .7 times the earnings
or multi-racial individuals. The most com- of white men, women’s relative earnings are uni-
mon combinations of two races (Asian-white, formly higher in each of the other racial groups.
black-white, Native American–white and Columns 4 and 5 give the earnings ratio relative
black-Asian) are treated as distinct categories. to whites of the same sex for minority men and
Finally, individuals who report more than two women, respectively. Comparing the two col-
races or who do not fit into any other racial umns, it is clear that minority women’s relative
category are coded as “other.” Because Hispan- earnings are higher than those of minority men.
ics are treated as an ethnic rather than a racial Column 6 gives the antilog of the quantity dk,
category in the census, Hispanics can be of any defined above. A positive value of dk corresponds
race. Therefore, to achieve exclusivity, individu- to exp(dk) being greater than 1, while a negative
als reporting Hispanic ethnicity are coded into value corresponds to exp(dk) being less than 1.
the appropriate Hispanic category, regardless Exp(dk) represents the ratio of minority women’s
of race. Thus, all individuals in race categories observed to predicted earnings, where predicted
other than “Mexican,” “Cuban,” “Puerto Rican” earnings are based on the assumption of additiv-
or “Other Hispanic” are non-Hispanic. ity between race and gender effects. Column 6,
Because earnings determination is more com- then, quantifies the patterns that can be identi-
plex for immigrants than for the native-born fied by “eyeballing” columns 3, 4 and 5.
(Zeng and Xie 2004), only U.S.-born workers The results in Column 6 are striking. In
are examined. This restriction limits the gener- every case, exp(dk) is greater than 1. The values
alizability of the findings for many of the groups of exp(dk) indicate that the average earnings of
studied. Because of the preponderance of im- nonwhite women range from about 4 percent
migrants in many Asian and Hispanic ethnic to 21 percent higher than predicted under the
groups, we emphasize that our results apply only additivity assumption, with Native American–
to the subsets of these populations that were born white bi-racial workers having the lowest value
in the United States. To assure comparability of and Korean workers the highest. For 16 out of
workers in the analysis, the sample is restricted to our 18 minority groups (all groups other than
full-time, full-year workers ages 25–55. Black-Asians and Vietnamese), dk is also statis-
tically significant. This is strong evidence that
the effects of race and sex on earnings are not
Results additive. Instead, there is a positive interaction
The main findings of the analysis are presented between being female and being a member of a
in Table 110.1. The racial categories are listed minority group. This interaction is widespread
in order of highest to lowest earnings among across different ethnicities, with groups as diverse
men, with the exception of whites as the refer- as Mexicans, Filipinos, Koreans, black-white
ence category in the first row. Columns 1 and biracials and Native Americans all showing ev-
2 present the geometric mean earnings of each idence of such an effect.
 Greenman and Xie—Double Jeopardy • 949

Table 110.1. Earnings and Relative Earnings by Race and Sex

(3) (4) (5) (6) (7)
(1) (2) Female- Minority- Minority- Observed- Observed-
Mean Mean Male White White Predicted Predicted
Annual Annual Earnings Earnings Earnings Earnings Earnings
Earnings Earnings Ratio Ratio Ratio Ratio Ratio with
(Men1) (Women1) within Race (Men) (Women) (Women) Controls
White Only 40,600 28,700 .71 1.00 1.00 1.00 1.00
Chinese 54,600 44,100 .81 1.34 1.54 1.15*** 1.15***
Asian Indian 47,700 38,300 .80 1.17 1.34 1.14*** 1.14***
Korean 46,300 39,700 .86 1.14 1.38 1.21*** 1.20***
Japanese 48,600 38,300 .79 1.20 1.33 1.11*** 1.11***
Cuban 39,400 32,200 .82 .97 1.12 1.16*** 1.12***
Other Asian 39,500 33,400 .85 .97 1.17 1.20*** 1.20***
Asian-white 39,800 32,600 .82 .98 1.13 1.16*** 1.15***
Black-Asian 38,900 29,900 .77 .96 1.04 1.09 1.10
Filipino 37,900 32,000 .84 .93 1.12 1.20*** 1.17***
Other 35,100 27,700 .79 .86 .96 1.11*** 1.09***
Vietnamese
35,300 27,300 .77 .87 .95 1.09 1.08
Only
Black-white 34,800 27,700 .80 .86 .97 1.13*** 1.10***
Native
American– 33,300 24,500 .74 .82 .86 1.04*** 1.02*
white
Other
31,900 24,900 .78 .79 .87 1.10*** 1.08***
Hispanic
Puerto Rican 32,000 26,500 .83 .79 .92 1.17*** 1.12***
Mexican 31,600 24,700 .78 .78 .86 1.11*** 1.08***
Black 30,000 25,200 .84 .74 .88 1.19*** 1.14***
Native
29,400 23,300 .79 .72 .81 1.12*** 1.08***
American
Notes: Sample indudes full-time, full-year workers between the ages of 25–55 who were born in the United States. Statis-
tical significance refers to the significance of the race-gender interaction. Control variables are: Education, potential work
experience, potential work experience squared, hours worked per week (above 35), self-employment and region.
1
Geometric mean of annual earnings.
*p < 1 **p < .05 ***p < .01

We next test whether this interaction is ro- difference for most Asian ethnic groups, with the
bust in a multivariate setting. The log of annual exception of Filipinos, whose adjusted exp(dk) is
earnings is regressed on a series of race*sex in- 3 percentage points lower than the unadjusted
teraction dummies, with controls for education, exp(dk). For the non-Asian racial groups, adjust-
potential work experience (calculated as the in- ing for earnings-relevant characteristics lowers
dividual’s age – years of schooling – 6), potential exp(dk) by between 2 and 5 percentage points.
work experience squared, hours worked per week This reduction in exp(dk) after the inclusion of
above the 35 hour full-time cutoff, urban resi- the control variables indicates that some part of
dence, self-employment and region of residence. the observed interaction may be due to variation
The antilogs of the coefficients on the sex*race across racial groups in the gender differences in
interaction terms give adjusted estimates of dk. earnings-relevant characteristics. However, the
The results of the multivariate analysis are re- inclusion of these controls does not change
ported in column 7. The adjustments make little the general pattern we discerned in column
950 • Part VIII: Gender Inequality

6: minority women’s earnings are consistently married women are slightly more pronounced
higher than would be predicted under additivity. than those for all women. Exp(dk) is greater than
The consistently positive pattern of dk across 1 for every group, and is statistically significant
all 18 minority groups is surprising. Columns 1 for 15 out of 18 minority groups. The values of
and 2 of Table 110.1 show large differences in exp(dk) indicate that married women’s earnings
average earnings across the racial/ethnic groups. range between 2 percent and 32 percent higher
While blacks, most Hispanic groups and Native than predicted under additivity. The results for
Americans all have considerably lower earnings unmarried women, however, are very different.
than whites, several Asian groups have consid- In general, the values of exp(dk) are quite close to
erably higher earnings. Nonetheless, both “dis- 1, and fail to reach statistical significance for the
advantaged” and “advantaged” minority groups majority of groups. Only five groups (Japanese,
have positive values of dk. For groups that have Cubans, Asian-whites, Puerto Ricans and blacks)
lower earnings than whites, this pattern means have values of exp(dk) significantly greater than
an attenuation of the race disadvantage among 1. We also test to see if these differences in ex-
women compared to that among men. How- p(dk ) between married and unmarried women
ever, for women in minority groups with higher are statistically significant. The difference is
earnings than whites, this means a more pro- indeed significant for 10 out of the 18 groups.
nounced advantage among women than among Thus, the pattern of higher-than-expected earn-
men. Women of every group have lower average ings we have found for minority women applies
earnings than men. Therefore, the interpretation primarily to the married.
of the interaction effect is more straightforward The adjusted exp(dk) is also computed for
when stated in terms of the variation in the gen- each marital status group, controlling for the
der effect across racial groups than when stated same factors as in column 7 of Table 110.1. The
in terms of the variation in the race effect across addition of the control variables changes the
gender: The effect of gender is always weaker individual values of exp(dk) somewhat, but it
among minorities than among whites. We also does not change the overall pattern of positive
prefer this second interpretation because it is interaction for married women. For unmarried
directly linked to our attempt to explain the ob- persons, exp(dk) tends to be slightly larger after
served empirical pattern in terms of differences the addition of the controls, resulting in a greater
in the applicability of role specialization theory number of groups with statistically significant
across racial/ethnic groups. values. Nonetheless, it is still much closer to 1 in
general for unmarried women than for married
women. Statistical tests of the difference between
Results by Marital Status exp(dk) for married and unmarried women in-
Table 110.2 presents results analogous to those dicate that the difference is indeed statistically
in columns 4 and 5 of Table 110.1, now disag- significant for 10 groups, the same as before the
gregated by marital status. For this portion of addition of the controls.
the analysis we originally divided the sample into
four groups by both marital and parental status
(married with children, married no children, Discussion and Conclusion
etc.). Surprisingly, the results showed that chil- This study has confirmed the ubiquitous inter-
dren make little additional difference above and sectionality of race and gender in the determi-
beyond marital status. Therefore, for parsimony nation of earnings. It is clear that among U.S.
this discussion is framed around differences by workers, there is no such thing as a pure “gender
marital status only. The baseline model without effect” or “race effect” when it comes to earnings.
covariates is discussed first. The results among The two must be considered simultaneously.
 Greenman and Xie—Double Jeopardy • 951

Table 110.2. Observed-to-Predicted Earnings Ratios for Minority Women, by Marital Status
Without Controls With Controls
Married Unmarried All Married Unmarried All
White Only 1.00 1.00 1.00 1.00 1.00 1.00
Chinese 1.13*** 1.05 1.15*** 1.15*** 1.08***^ 1.15***
Asian Indian 1.10 1.01 1.14*** 1.14** 1.04 1.14***
Korean 1.20** 1.02 1.21*** 1.20** 1.08 1.20***
Japanese 1.10*** 1.06** 1.11*** 1.09*** 1.07*** 1.11***
Cuban 1.13*** 1.10*** 1.16*** 1.10*** 110*** 1.12***
Other Asian 1.32*** 1.03^^^ 1.20*** 1.25*** 1.06^^ 1.20***
Asian-white 1.14*** 1.06* 1.16*** 1.13*** 1.09*** 1.15***
Black-Asian 1.20 .98 1.09 1.13 .99 1.10
Filipino 1.25*** 1.01^^^ 1.20*** 1.20*** 1.04^^^ 1.17***
Other 1.12*** 1.00^^^ 1.11*** 1.10*** 1.02^^^ 1.09***
Vietnamese Only 1.02 1.12 1.09 1.01 1.09 1.08
Black-white 1.17*** .99^^^ 1.13*** 1.12*** 1.00^^ 1.10***
Native American–white 1.03* 1.00 1.04*** 1.01 1.01 1.02*
Other Hispanic 1.13*** 1.01^^^ 1.10*** 1.08*** 1.03**^^ 1.08***
Puerto Rican 1.23*** 1.04**^^^ 1.17*** 1.14*** 1 04**^^^ 1.12***
Mexican 1.14*** 1.01^^^ 1.11*** 1.09*** 1.02^^^ 1.08***
Black 1.25*** 1.05***^^^ 1.19*** 1.17*** 1.05***^^^ 1.14***
Native American 1.16*** 1.00^^^ 1.12*** 1.09*** 1.01^^^ 1.08***

Notes: Sample includes full-time, full-year workers between the ages of 25–55 who were born in the United States. Control
variables are: Education, potential work experience, potential work experience squared, hours worked per week (above 35),
self-employment and region.
* Race-sex interaction statistically significant at the .1 level.
** Race-sex interaction statistically significant at the .05 level.
*** Race-sex interaction statistically significant at the .01 level.
^ Statistically different from married at the .1 level
^^ Statistically different from married at the .05 level
^^^ Statistically different from married at the .01 level

Furthermore, the statistical interaction between holds true. It would be hard to argue that this
being minority and being female is consistently result could be due to any similarity across such
positive: Among groups who are disadvantaged an array of groups. Therefore, the explanation
in earnings relative to whites, the race penalty is more likely to be found in something unique
is always smaller among women than among about our comparison group—non-Hispanic
men, while for earnings-advantaged groups, whites. Up to this point, the discussion of
the advantage is greater for women than men. earnings has been framed in terms of the dis-
Conversely, for all minority groups the gender advantages associated with being female and (in
penalty is smaller for minority women than most cases) with being nonwhite. But instead of
for white women. Thus, the “double jeopardy” concluding that minority women’s earnings are
characterization proposed in the earlier litera- higher than expected under additivity, perhaps
ture is a poor description of minority women’s the results mean that white women’s earnings
earnings. are lower than expected. Such would have been
It is striking that across such a diverse array our conclusion if we had chosen blacks, for ex-
of racial groups, including Asians, Hispanics and ample, instead of whites as the reference group.
mixed-race individuals, the same basic pattern There is no way to distinguish between these
952 • Part VIII: Gender Inequality

interpretations empirically—they are equally housing—there is debate in the literature about the
consistent with the results. Reframing this dis- extent to which they pool other economic resources.
cussion in terms of unexpectedly low earnings 2. The applicability of this theory to gender seg-
among white women suggests that the explana- regation of occupations has been challenged by En-
gland (1982, 1988).
tion for the empirical pattern observed in this
study may involve something atypical about the
REFERENCES
system of gender relations among whites. Becker, Gary S. 1973. “A Theory of Marriage: Part
The answer may lie in role specialization I.” Journal of Political Economy 81(4):813–46.
theory being more applicable to whites than to ———. 1974. “A Theory of Marriage: Part II.” Jour-
other groups. Although we cannot test defini- nal of Political Economy 82(2):S11–S26.
tively whether this is the case, findings indicate ———. 1991. A Treatise on the Family. Harvard
this explanation warrants further investigation. University Press.
Results by marital status revealed that there were Blau. Francine D., and Andrea H. Beller. 1992.
“Black-White Earnings over the 1970s and
few racial differences in the gender earnings gap
1980s: Gender Differences in Trends.” The Re-
among the unmarried, while for the married,
view of Economics and Statistics 74(2):276–86.
the gender earnings gap was significantly smaller ———. 1988. “Trends in Earnings Differentials by
for almost every minority group than for whites. Gender, 1971–1981.” Industrial and Labor Rela-
These results strongly suggest that the explana- tions Review 41(4):513–29.
tion for the race/gender earnings interaction has Blee, Kathleen M., and Ann R. Tickamyer. 1995.
to do with family factors. “Racial Differences in Men’s Attitudes about
The primary contribution of this study lies Women’s Gender Roles.” Journal of Marriage and
in the documentation of the ubiquity of the Family 57(1): 21–30.
Bound, John, and Laura Dresser. 1999. “Losing
gender-race interaction in earnings determina-
Ground: The Erosion of the Relative Earnings
tion. We are unable to provide a definitive ex-
of African American Women during the 1980s.”
planation for the greater gender earnings gap Pp. 61–104. African American and Latina Women
among whites than among other racial groups. at Work. Irene Browne, editor. Russell Sage
However, results suggest that gender dynamics Foundation.
within families may be a fruitful area for future Bolzendahl, Catherine I., and Daniel J. Myers. 2004.
research. While researchers have examined racial “Feminist Attitudes and Support for Gender
differences in gender role attitudes, marital re- Equality: Opinion and Change in Women and
lationships, gender division of housework, and Men, 1974–1998.” Social Forces 83(2): 759–90.
other family processes (McLoyd et al. 2000), Brewer, Rose M., Cecilia A. Conrad and Mary C.
King. 2002. “The Complexities and Potential
none has explicitly linked these differences to ra-
of Theorizing Gender, Caste, Race, and Class.”
cial differences in labor force outcomes (Brewer Feminist Economics 8(2): 3–18.
et al. 2002). Given that another body of litera- Brines, Julie. 1994. “Economic Dependency, Gen-
ture shows that family-linked processes such as der, and the Division of Labor at Home.” Ameri-
parenthood (Budig and England 2001; Waldfo- can Journal of Sociology 100(3):652–88.
gel 1997) and the gender division of housework Browne, Irene, and Joya Misra. 2003. “The Intersec-
affect earnings, it follows that there may be such tion of Gender and Race in the Labor Market.”
a link. Hence, we suggest that the intersection of Annual Review of Sociology 29:487–513.
family and labor force outcomes may well hold Budig, Michelle J., and Paula England. 2001. “The
Wage Penalty for Motherhood.” American Socio-
the key to understanding the intersection of race
logical Review 66(2):204–25.
and gender. Cancio, A.S., T.D. Evans and David J. Maume,
Jr. 1996. “Reconsidering the Declining Sig-
NOTES nificance of Race: Racial Differences in Early
1. Although cohabiting adults necessarily Career Wages.” American Sociological Review
share some aspects of economic well-being—e.g., 61(4):541–56.
 Greenman and Xie—Double Jeopardy • 953

Carlson, Leonard A., and Caroline Swartz. 1988. King, Deborah K. 1988. “Multiple Jeopardy, Multi-
“The Earnings of Women and Ethnic Minori- ple Consciousness: The Context of a Black Fem-
ties, 1959–1979.” Industrial and Labor Relations inist Ideology.” Signs 14(1):42–72.
Review 41(4):530–46. Lichter, Daniel T., Diane K. McLaughlin, George
Corcoran, Mary. 1999. “The Economic Progress Kephart and David J. Landry. 1992. “Race and
of African American Women.” Pp. 35–60. Afri- the Retreat from Marriage: A Shortage of Mar-
can American and Latina Women at Work. Irene riageable Men?” American Sociological Review
Browne, editor. Russell Sage Foundation. 57(6): 781–99.
Cotter, David A., Joan M Hermsen and Reeve Lundberg, Shelley, and Robert A. Pollack. 1996.
Vanneman. 1999. “Systems of Gender, Race “Bargaining and Distribution in Marriage.” Jour-
and Class Inequality: Multilevel Analyses.” Social nal of Economic Perspectives 10(4): 139–58.
Forces 78(2):433–60. Malveaux, Julianne. 1986. “Research and Policy Im-
Darity, William Jr., David K. Guilkey and Wil- plications: Section Introduction.” Pp. 289–91.
liam Winfrey. 1996. “Explaining Differences in Slipping through the Cracks. Margaret C. Sims
Economic Performance among Racial and Eth- and Julianne Malveaux, editors. Transaction
nic Groups in the USA: The Data Examined.” Books.
American Journal of Economics and Sociology Marini, Margaret Mooney. 1989. “Sex Differences in
55(4):411–25. Earnings in the United States.” Annual Review of
Durden, Garey C., and Patricia E. Gaynor. 1998. Sociology 15:343–80.
“More on the Cost of Being Other than White McCall, Leslie. 2001. Complex Inequality. Routledge.
and Male: Measurement of Race, Ethnic, and ———. 2005. “The Complexity of Intersectional-
Gender Effects on Yearly Earnings.” American ity.” Signs 30(3): 1772–1800.
Journal of Economics and Sociology 57(1): 95–104. McGuire, Gail M., and Barbara Reskin. 1993. “Au-
Edin, Kathryn. 2005. Promises I Can Keep: Why Poor thority Hierarchies at Work: The Impacts of Race
Women Put Motherhood before Marriage. Univer- and Sex.” Gender and Society 7(4):487–506.
sity of California Press. McLoyd, Vonnie C., Ana Mari Cauce, David Takeu-
England, Paula, Karen Christopher and Lori L. Reid. chi and Leon Wilson. 2000. “Marital Processes
1999. “Gender Race, Ethnicity, and Wages.” Pp. and Parental Socialization in Families of Color: A
139–82. African American and Latina Women Decade Review of Research.” Journal of Marriage
at Work. Irene Browne, editor. Russell Sage and Family 62(4): 1070–93.
Foundation. Mincer, Jacob, and Solomon Polachek. 1974.
Farley, Reynolds. 1984. Blacks and Whites: Narrow- “Family Investments in Human Capital: Earn-
ing the Gap? Harvard University Press. ings of Women.” Journal of Political Economy
Goldin, Claudia D. 1997. “Career and Family: Col- 82(2):S76–S108.
lege Women Look to the Past.” Pp. 20–58. Gender Noonan, Mary C. 2001. “The Impact of Domestic
and Family Issues in the Workplace. F.D. Blau and Work on Men’s and Women’s Wages.” Journal of
R.G. Ehrenberg, editors. Russell Sage Foundation. Marriage and Family 63(4): 1134–45.
John, Daphne, and Beth Anne Shelton. 1997. “The Padavic, Irene, and Barbara Reskin. 2002. Women
Production of Gender among Black and White and Men at Work. Pine Forge Press.
Women and Men: The Case of Household Ransford, Edward H., and Jon Miller. 1983. “Race,
Labor.” Sex Roles 36(3/4): 171–93. Sex, and Feminist Outlooks.” American Sociolog-
Kamo, Yoshinori, and Ellen L. Cohen. 1998. “Di- ical Review 48(1):46–59.
vision of Household Work between Partners: A Raley, Sara B., Marybeth J. Mattingly and Suzanne
Comparison of Black and White Couples.” Jour- M. Bianchi. 2006. “How Dual are Dual-Income
nal of Comparative Family Studies 29(1): 131–45. Couples? Documenting Change from 1970 to
Kane, Emily W. 2000. “Racial and Ethnic Variation 2001.” Journal of Marriage and Family 68(1):
in Gender-Related Attitudes.” Annual Review of 11–28.
Sociology 26:419–39. Smock, Pamela J., Wendy D. Manning and Sanjiv
Kilbourne, Barbara, Paula England and Kurt Beron. Gupta. 1999. “The Effect of Marriage and Di-
1994. “Effects of Individual, Occupational, and vorce on Women’s Economic Well-Being.” Amer-
Industrial Characteristics on Earnings: Intersec- ican Sociological Review 64(6): 794–812.
tions of Race and Gender.” Social Forces 72(4): Taylor, Pamela J., M Belinda Tucker and Claudia
1149–76. Mitchell-Kernan. 1999. “Ethnic Variations in
954 • Part VIII: Gender Inequality

Perceptions of Men’s Provider Role.” Psychology of Xie, Yu, and Kimberly A. Goyette. 2004. A Demo-
Women Quarterly 23(4):741–61. graphic Portrait of Asian Americans. New York:
Waldofogel, Jane. 1997. “The Effect of Children on Russell Sage Foundation.
Women’s Wages.” American Sociological Review Zeng, Zhen, and Yu Xie. 2004. “Asian Americans’
62(2):209–217. Earnings Disadvantage Reexamined: The Role of
Wilson, William Julius. 1996. When Work Disap- Place of Education.” American Journal of Sociol-
pears: The World of the New Urban Poor. Alfred ogy 109(5): 1075–1108.
A. Knopf.
 England—The Gender Revolution • 955

A Stalling Out?

111 • Paula England

The Gender Revolution

Uneven and Stalled

We sometimes call the sweeping changes in the change, I will review trends on a number of in-
gender system since the 1960s a “revolution.” dicators. While the shape of most of the trends
Women’s employment increased dramatically is not in dispute among scholars, the explana-
(Cotter, Hermsen, and England 2008); birth tions I offer for the uneven and halting nature of
control became widely available (Bailey 2006); change have the status of hypotheses rather than
women caught up with and surpassed men in well-documented conclusions.
rates of college graduation (Cotter, Hermsen, I will argue that there has been little cultural
and Vanneman 2004, 23); undergraduate college or institutional change in the devaluation of tra-
majors desegregated substantially (England and ditionally female activities and jobs, and as a re-
Li 2006); more women than ever got doctorates sult, women have had more incentive than men
as well as professional degrees in law, medicine, to move into gender-nontraditional activities and
and business (Cotter, Hermsen, and Vanneman positions. This led to asymmetric change; wom-
2004, 22–23; England et al. 2007); many kinds en’s lives have changed much more than men’s.
of gender discrimination in employment and ed- Yet in some subgroups and arenas, there is less
ucation became illegal (Burstein 1989); women clear incentive for change even among women;
entered many previously male-dominated occu- examples are the relatively low employment rates
pations (Cotter, Hermsen, and Vanneman 2004, of less educated women and the persistence of
10–14); and more women were elected to po- traditionally gendered patterns in heterosexual
litical office (Cotter, Hermsen, and Vanneman romantic, sexual, and marital relationships.
2004, 25). As sweeping as these changes have I also argue, drawing on work by Charles and
been, change in the gender system has been un- Bradley, that the type of gender egalitarianism
even—affecting some groups more than others that did take hold was the type most compati-
and some arenas of life more than others, and ble with American individualism and its cultural
change has recently stalled. My goal in this and institutional logics, which include rights of
chapter is not to argue over whether we should access to jobs and education and the desideratum
view the proverbial cup as half empty or half full of upward mobility and of expressing one’s “true
but, rather, to stretch toward an understanding self ” (Charles 2011; Charles and Bradley 2002,
of why some things change so much more than 2009). One form this gender egalitarianism has
others. To show the uneven nature of gender taken has been the reduction of discrimination

Paula England. “The Gender Revolution: Uneven and Stalled,” Gender & Society 24:2 (2010), pp. 149–166.
Copyright © 2010 by Sage Publications. Reprinted with Permission by SAGE Publications.
956 • Part VIII: Gender Inequality

Figure 111.1. Percentage of U.S. Men and Women Employed, 1962–2007

100.0%

90.0%
Percent Employed

80.0%

70.0%

60.0%

50.0%
Men
Women
40.0%
1962 1967 1972 1977 1982 1987 1992 1997 2002 2007
Year
Source: Cotter, Hermsen, and Vanneman (2009).
Note: Persons are considered employed if they worked for pay anytime during the year. Refers to adults aged 25 to 54.

in hiring. This has made much of the gender to men, with few changes in the opposite direc-
revolution that has occurred possible; women tion. The source of this asymmetry is an aspect
can now enter formerly “male” spheres. But of society’s valuation and reward system that has
co-occurring with this gender egalitarianism, not changed much—the tendency to devalue
and discouraging such integration is a strong (if and badly reward activities and jobs traditionally
often tacit) belief in gender essentialism—the done by women.
notion that men and women are innately and
fundamentally different in interests and skills
(Charles 2011; Charles and Bradley 2002, 2009; Women’s Increased Employment
Ridgeway 2009). A result of these co-occurring One form the devaluation of traditionally female
logics is that women are most likely to challenge activities takes is the failure to treat child rearing
gender boundaries when there is no path of up- as a public good and support those who do it
ward mobility without doing so, but otherwise with state payments. In the United States, wel-
gender blinders guide the paths of both men and fare reform took away much of what little such
women. support had been present. Without this, women
doing child rearing are reliant on the employ-
ment of male partners (if present) or their own
Devaluation of “Female” Activities employment. Thus, women have had a strong
as Asymmetric Incentives for incentive to seek paid employment, and more so
Women and Men to Change as wage levels rose across the decades (Bergmann
Most of the changes in the gender system her- 2005). As Figure 111.1 shows, women’s employ-
alded as “revolutionary” involve women moving ment has increased dramatically. But change
into positions and activities previously limited has not been continuous, as the trend line
 England—The Gender Revolution • 957

flattened after 1990 and turned down slightly Shafer (2008) use data from 16 affluent coun-
after 2000 before turning up again. This turn- tries circa 2000 and show that, in all of them,
down was hardly an “opt-out revolution,” to use among women partnered with men (married
the popular-press term, as the decline was tiny or cohabiting), those with more education are
relative to the dramatic increase across 40 years more likely to be employed. Moreover, there is
(Kuperberg and Stone 2008; Percheski 2008). no monotonic relationship between partner’s
But the stall after 1990 is clear, if unexplained. earnings and a woman’s employment; at top lev-
Figure 111.1 also shows the asymmetry in els of his income, her employment is deterred.
change between men’s and women’s employ- But women whose male partners are at middle
ment; women’s employment has increased much income levels are more likely to be employed
more than men’s has declined. There was no- than women whose partners have very low or
where near one man leaving the labor force to no earnings, the opposite of what the “need for
become a full-time homemaker for every woman income” principle suggests.
who entered, nor did men pick up household In the United States, it has been true for de-
work to the extent women added hours of em- cades that well-educated women are more likely
ployment (Bianchi, Robinson, and Milkie 2006). to be employed, and the effect of a woman’s own
Men had little incentive to leave employment. education has increased, while the deterring
Among women, incentives for employment effect of her husband’s income has declined
vary. Class-based resources, such as education, (Cohen and Bianchi 1999). For example, in
affect these incentives. At first glance, we might 1970, 59 percent of college graduate women,
expect less educated women to have higher em- but only 43 percent of those with less than a
ployment rates than their better-educated peers high school education, were employed sometime
because they are less likely to be married to a during the year. In 2007, the figures were 80 per-
high-earning man. Most marriages are between cent for college graduates and 47 percent for less
two people at a similar education level (Mare than high school (calculated from data in Cotter,
1991), so the less educated woman, if she is Hermsen, and Vanneman 2009).
married, typically has a husband earning less
than the husband of the college graduate. Her
family would seem to need the money from Women Moving into “Male” Jobs and
her employment more than the family headed Fields of Study
by two college graduates. Let us call this the The devaluation of and underpayment of pre-
“need for income” effect. But the countervailing dominantly female occupations is an important
“opportunity cost” factor is that well-educated institutional reality that provides incentives for
women have more economic incentive for em- both men and women to choose “male” over “fe-
ployment because they can earn more (England, male” occupations and the fields of study that
Garcia-Beaulieu, and Ross 2004). Put another lead to them. Research has shown that predom-
way, the opportunity cost of staying at home is inantly female occupations pay less, on average,
greater for the woman who can earn more. In- than jobs with a higher proportion of men. At
deed, the woman who did not graduate from least some of the gap is attributable to sex com-
high school may have potential earnings so low position because it persists in statistical models
that she could not even cover child care costs controlling for occupations’ educational require-
with what she could earn. It is an empirical ques- ments, amount of skill required, unionization,
tion whether the “need for income” or “opportu- and so forth. I have argued that this is a form of
nity cost” effect predominates. gender discrimination—employers see the worth
Recent research shows that the opportunity- of predominantly female jobs through biased
cost effect predominates in the United States and lenses and, as a result, set pay levels for both men
other affluent nations. England, Gornick, and and women in predominantly female jobs lower
958 • Part VIII: Gender Inequality

Figure 111.2. Sex Segregation of Fields of Study for U.S. Bachelor Degree Recipients, 1971–2006

55
Index of Dissimilarity (D)

50

45

40

35

30
1971 1976 1981 1986 1991 1996 2001 2006
Year

Source: Author’s calculations from the National Center for Education Statistics (NCES) 1971–2003 and NCES
2004–2007.

than they would be if the jobs had a more heav- in previously male-dominated, business-related
ily male sex composition (England 1992; Kil- fields, such as business, marketing, and account-
bourne et al. 1994; England and Folbre 2005). ing; while fewer chose traditionally female ma-
While the overall sex gap in pay has diminished jors like English, education, and sociology; and
because more women have moved into “male” there was little increase of men’s choice of these
fields (England and Folbre 2005), there is no ev- latter majors (England and Li 2006, 667–69).
idence that the devaluation of occupations filled Figure 111.2 shows the desegregation of fields
with women has diminished (Levanon, England, of bachelor’s degree receipt, using the index of
and Allison 2009). Indeed, as U.S. courts have dissimilarity (D), a scale on which complete seg-
interpreted the law, this type of between-job dis- regation is 100 and complete integration is 0. It
crimination is not even illegal (England 1992, shows that segregation dropped significantly in
225–51; Steinberg 2001), whereas it is illegal to the 1970s and early 1980s, but has been quite
pay women less than men in the same job, un- flat since the mid-1980s. Women’s increased in-
less based on factors such as seniority, qualifica- tegration of business fields stopped then as well
tions, or performance. Given this, both men and (England and Li 2006).
women continue to have a pecuniary incentive Women have also recently increased their
to choose male-dominated occupations. Thus, representation in formerly male-dominated pro-
we should not be surprised that desegregation of fessional degrees, getting MDs, MBAs, and law
occupations has largely taken the form of women degrees in large numbers. Women were 6 per-
moving into male-dominated fields, rather than cent of those getting MDs in 1960, 23 percent
men moving into female-dominated fields. in 1980, 43 percent in 2000, and 49 percent in
Consistent with the incentives embedded in 2007; the analogous numbers for law degrees
the ongoing devaluation of female fields, desegre- (JDs) were 3, 30, 46, and 47 percent, and for
gation of fields of college study came from more MBAs (and other management first-professional
women going into fields that were predominantly degrees), 4, 22, 39, and 44 percent (National
male, not from more men entering “female” Center for Education Statistics 2004–2008).
fields. Since 1970, women increasingly majored There was no marked increase in the proportion
 England—The Gender Revolution • 959

of men in female-dominated graduate profes- realm, especially in dyadic heterosexual relation-

sional programs such as library science, social ships. It is still men who usually ask women on
work, or nursing (National Center for Education dates, and sexual behavior is generally initiated
Statistics 2009). by men (England, Shafer, and Fogarty 2008).
As women have increasingly trained for pre- Sexual permissiveness has increased, making it
viously male-dominated fields, they have also more acceptable for both heterosexual men and
integrated previously male-dominated occu- women to have sex outside committed relation-
pations in management and the professions in ships. But the gendered part of this—the dou-
large numbers (Cotter, Hermsen, and Vanneman ble standard—persists stubbornly; women are
2004, 10–13). Women may face discrimination judged much more harshly than men for casual
and coworker resistance when they attempt to sex (Hamilton and Armstrong 2009; England,
integrate these fields, but they have a strong Shafer, and Fogarty 2008). The ubiquity of ask-
pecuniary incentive to do so. Men lose money ing about height in Internet dating Web sites
and suffer cultural disapproval when they choose suggests that the convention that men should be
traditionally female-dominated fields; they have taller than their female partner has not budged.
little incentive to transgress gender boundaries. The double standard of aging prevails, making
women’s chances of marriage decrease with age
much more than men’s (England and McClin-
The “Personal” Realm tock 2009). Men are still expected to propose
“The personal is political” was a rallying cry marriage (Sassler and Miller 2007). Upon mar-
of 1960s feminists, urging women to demand riage, the vast majority of women take their
equality in private as well as public life. Yet husband’s surname. The number of women
conventions embodying male dominance have keeping their own name increased in the 1970s
changed much less in “the personal” than in and 1980s but little thereafter, never exceeding
the job world. Where they have changed, the about 25 percent even for college graduates (who
asymmetry described above for the job world have higher rates than other women) (Goldin
prevails. For example, parents are more likely to and Shim 2004). Children are usually given their
give girls “boy” toys such as Legos than they are father’s surname; a recent survey found that even
to give dolls to their sons. Girls have increased in cases where the mother is not married to the
their participation in sports more than boys have father, 92 percent of babies are given the father’s
taken up cheerleading or ballet. Women now last name (McLanahan forthcoming). While we
commonly wear pants, while men wearing skirts do not have trend data on all these personal mat-
remains rare. A few women started keeping their ters, my sense is that they have changed much
birth-given surname upon marriage (Goldin less than gendered features of the world of paid
and Shim 2004), with little adoption by men work.
of women’s last names. Here, as with jobs, the The limited change seen in the heterosexual
asymmetry follows incentives, albeit nonmate- personal realm may be because women’s in-
rial ones. These social incentives themselves flow centive to change these things is less clear than
from a largely unchanged devaluation of things their incentive to move into paid work and into
culturally defined as feminine. When boys and higher-paying “male” jobs. The incentives that
men take on “female” activities, they often suffer do exist are largely noneconomic. For example,
disrespect, but under some circumstances, girls women may find it meaningful to keep their
and women gain respect for taking on “male” birth-given surnames and give them to their
activities. children, and they probably enjoy sexual free-
What is more striking than the asymme- dom and initiation, especially if they are not
try of gender change in the personal realm is judged adversely for it. But these noneconomic
how little gendering has changed at all in this benefits may be neutralized by the noneconomic
960 • Part VIII: Gender Inequality

Figure 111.3. Sex Segregation of Middle-Class and Working-Class Occupations in the United States,
1950–2000
70
Working Class
Middle Class
Index of Dissimilarity (D)

60

50

40

30
1950 1960 1970 1980 1990 2000

Year
Source: Cotter, Hermsen, and Vanneman (2004, 14).
Note: Middle-class occupations include professional, management, and nonretail sales. All others are classified as
working-class occupations.

penalties from transgressing gender norms and nonprofessional service work). Figure 111.3
by the fact that some have internalized the shows that desegregation has proceeded much
norms. When women transgress gender barriers farther in middle-class than working-class jobs.
to enter “male” jobs, they too may be socially Middle-class jobs showed dramatic desegrega-
penalized for violating norms, but for many this tion, although the trend lessened its pace after
is offset by the economic gain. 1990. By contrast, working-class jobs are almost
as segregated as they were in 1950! Women have
integrated the previously male strongholds of
Co-Occurring Logics of Women’s management, law, medicine, and academia in
Rights to Upward Mobility and large numbers. But women have hardly gained
Gender Essentialism a foothold in blue-collar, male-dominated jobs
I have stressed that important change in the such as plumbing, construction, truck driving,
gender system has taken the form of women welding, and assembly in durable manufactur-
integrating traditionally male occupations and ing industries such as auto and steel (Cotter,
fields of study. But even here change is un- Hermsen, and Vanneman 2004, 12–14). This is
even. The main generalization is shown by roughly the situation in other affluent nations as
Figure 111.3, which divides all occupations well (Charles and Grusky 2004). This same class
by a crude measure of class, calling profes- difference in trend can be seen if we compare the
sional, management, and nonretail sales occu- degree of segregation among those who have var-
pations “middle class,” and all others “working ious levels of education; in the United States, sex
class” (including retail sales, assembly work in segregation declined much more dramatically
manufacturing, blue-collar trades, and other since 1970 for college graduates than any other
 England—The Gender Revolution • 961

group (Cotter, Hermsen, and Vanneman 2009, most available material from which to construct
2004, 13–14). aspirations and may be used even more when a
Why has desegregation been limited to job choice is seen as a deep statement about self
high-level jobs? The question has two parts: why (Charles and Bradley 2009).
women did not integrate blue-collar male jobs in Given all this, I hypothesize that if women
significant numbers, and why women did inte- can move “up” in status or income relative to
grate professional and managerial jobs in droves. their reference group while still staying in a
Why one and not the other? Many factors were job typically filled by women, then because of
undoubtedly at work, but I will focus on one gender beliefs and gendered identities, they are
account, which borrows from Charles and Brad- likely to do so. If they cannot move up without
ley (Charles 2011; Charles and Bradley 2002, integrating a male field, and demand is present
2009). In the United States and many Western and discrimination not too strong, they are more
societies today, a certain kind of gender egalitar- likely to cross the gender boundary. Applying
ianism has taken hold ideologically and institu- this hypothesis, why would women not enter
tionally. The logic is that individuals should have male blue-collar fields? To be sure, many women
equal rights to education and jobs of their choice. without college degrees would earn much more
Moreover, achievement and upward mobility are in the skilled blue-collar crafts or unionized
generally valued. There is also a “postmaterial- manufacturing jobs than in the service jobs typi-
ist” aspect to the culture which orients one to cally filled by women at their education levels—
find her or his “true self.” The common ethos jobs such as maid, child care worker, retail sales
is a combination of “the American dream” and clerk, or assembler in the textile industry. So they
liberal individualism. Many women, like men, have an economic incentive to enter these jobs.
want to “move up” in earnings and/or status, or But such women could also move “up” to clerical
at least avoid moving down. But up or down rel- work or teaching, higher status and better paying
ative to what reference group? I suggest that the but still traditionally female jobs. Many take this
implicit reference group is typically those in the path, often getting more education.
previous generation (or previous birth cohorts) In contrast, consider women who assumed
of one’s own social class background and one’s they would go to college and whose moth-
own sex. For example, women might see their ers were in female-dominated jobs requiring a
mothers or aunts as a reference, or women who college degree like teacher, nurse, librarian, or
graduated with their level of education ten years social worker. For these women, to move up in
ago. Persons of the same-sex category are the im- status or earnings from their reference group
plicit reference group because of strong beliefs options requires them to enter traditionally
in gender essentialism, that notion that men and male jobs; there are virtually no heavily female
women are innately and fundamentally different jobs with higher status than these female pro-
(Charles 2011; Ridgeway 2009). While liberal fessions. These are just the women, usually of
individualism encourages a commitment to middle-class origins, who have been integrating
“free choice” gender egalitarianism (such as legal management, law, medicine, and academia in
equality of opportunity), ironically, orienting recent decades. For them, upward mobility was
toward gender-typical paths has probably been not possible within traditional boundaries, so
encouraged by the emerging form of individu- they were more likely to integrate male fields.
alism that stresses finding and expressing one’s In sum, my argument is that one reason that
“true self.” Notions of self will in fact be largely women integrated male professions and manage-
socially constructed, pulling from socially salient ment much more than blue-collar jobs is that
identities. Because of the omnipresent nature of the women for whom the blue-collar male jobs
gender in the culture (Ridgeway 2009; West and would have constituted “progress” also had the
Zimmerman 1987), gender often becomes the option to move up by entering higher-ranking
962 • Part VIII: Gender Inequality

Figure 111.4. Percentage of All Doctoral Degree Recipients Who Were Women in Selected Large Fields,
1971–2006
80%

Psychology
70%
Education

60% Sociology

All fields
Percent Female

50% combined
Biology

40% History

Political Science
30%
Economics

20% Math

Engineering
10%

0%
1971 1976 1981 1986 1991 1996 2001 2006
Year

Source: Author’s calculations from the National Center for Education Statistics (NCES) 1971–2003 and NCES
2004–2007.
Note: All fields combined refers to all fields of doctoral study, not only the nine fields shown here. Engineering refers to
doctoral degrees in E.E. Psychology excludes clinical psychology.

female jobs via more education. They thus had up dramatically, reflecting the overall increase in
options for upward mobility without trans- women getting doctorates. But the rank order of
gressing gender boundaries not present for their fields in their percentage female changed little.
middle-class sisters. The fields with the highest percentage of women
Even women entering male-typical occu- today are those that already had a high percent-
pations, however, sometimes choose the more age of women decades ago relative to other fields.
female-intensive subfields in them. In some What explains the failure of fields of doctoral
cases, ending up in female-intensive subfields study—and thus academic departments—to de-
results from discrimination, but in others it may segregate? Following the line of argument above,
result from the gender essentialism discussed I suggest that the extreme differentiation of fields
above. An example is the movement of women of academic study allowed many women moving
into doctoral study and into the occupation of “up” to doctoral study and an academic career
“professor.” This development brought women to do so in fields that seemed consistent with
into a new arena. But within this arena, there was their (tacitly gendered) notions of their inter-
virtually no desegregation of fields of doctoral ests and “true selves.” Women academics in the
study from 1970 on (England et al. 2007, 32). humanities and social sciences thus find them-
Women have gone from being only 14 percent selves in the more female subunits (disciplines)
of those who get doctorates in 1971 to nearly of a still largely male-dominated larger unit (the
half. But, conditional on getting a doctoral professorate).
degree, neither women nor men have changed
the fields of study they choose much (England
et al. 2007). This can be seen in Figure 111.4, Conclusion
which shows the percentage of women in nine Change in the gender system has been uneven,
large fields of study in each year from 1971 to changing the lives of some groups of people
2006. The percentage female in every field went more than others and changing lives in some
 England—The Gender Revolution • 963

arenas more than others. Although many factors and gender essentialism—make it most likely
are at play, I have offered two broad explanations for women to move into nontraditional fields of
for the uneven nature of change. study or work when there is no possible female
First, I argued that, because of the cultural field that constitutes upward mobility from the
and institutional devaluation of characteristics socially constructed reference point. This helps
and activities associated with women, men had explain why women integrated male-dominated
little incentive to move into badly rewarded, tra- professional and managerial jobs more than
ditionally female activities such as homemaking blue-collar jobs. Women from working-class
or female-dominated occupations. By contrast, backgrounds, whose mothers were maids or as-
women had powerful economic incentives to semblers in nondurable manufacturing, could
move into the traditionally male domains of paid move up financially by entering blue-collar
employment and male-typical occupations; and “male” trades but often decide instead to get
when hiring discrimination declined, many did. more education and move up into a female job
These incentives varied by class, however; the such as secretary or teacher. It is women with
incentive to go to work for pay is much stronger middle-class backgrounds, whose mothers were
for women who can earn more; thus employ- teachers or nurses, who cannot move up without
ment levels have been higher for well-educated entering a male-dominated career, and it is just
women. I also noted a lack of change in the gen- such women who have integrated management,
dering of the personal realm, especially of het- law, medicine, and academia. Yet even while in-
erosexual romantic and sexual relationships tegrating large fields such as academia, women
Second, I explored the consequences of the often gravitate toward the more female-typical
co-occurrence of two Western cultural and in- fields of study.
stitutional logics. Individualism, encompassing a While discussing the uneven character of
belief in rights to equal opportunity in access to gender change, I noted that the type of gender
jobs and education in order to express one’s “true change with the most momentum—middle-class
self,” promotes a certain kind of gender egali- women entering traditionally male spheres—
tarianism. It does not challenge the devaluation has recently stalled (Cotter, Hermsen, and
of traditionally female spheres, but it encour- Vanneman 2004, 2009). Women’s employment
ages the rights of women to upward mobility rates stabilized, desegregation of occupations
through equal access to education and jobs. To slowed down, and desegregation of fields of col-
be sure, this ideal has been imperfectly realized, lege study stopped. Erosion of the sex gap in pay
but this type of gender egalitarianism has taken slowed as well (Cotter, Hermsen, and Vanneman
hold strongly. But co-occurring with it, some- 2009). While the reason for the stalling is un-
what paradoxically, are strong (if tacit) beliefs in clear, like the unevenness of change, the stall-
gender essentialism—that men and women are ing of change reminds us how contingent and
innately and fundamentally different in interests path-dependent gender egalitarian change is,
and skills (Charles 2011; Charles and Bradley with no inexorable equal endpoint. Change has
2002, 2009; Ridgeway 2009). Almost no men been as much unintended consequence of larger
and precious few women, even those who believe institutional and cultural forces as realization of
in “equal opportunity,” have an explicit commit- the efforts of feminist organizing, although the
ment to undoing gender differentiation for its latter has surely helped. Indeed, given the recent
own sake. Gender essentialism encourages tradi- stalling of change, future feminist organizing
tional choices and leads women to see previous may be necessary to revitalize change.
cohorts of women of their social class as the refer-
ence point from which they seek upward mobil- REFERENCES
ity. I concluded that the co-occurrence of these Bailey, Martha J. 2006. More power to the pill: The
impact of contraceptive freedom on women’s life
two logics—equal opportunity individualism
964 • Part VIII: Gender Inequality

cycle labor supply. Quarterly Journal of Economics England, Paula, and Nancy Folbre. 2005. Gender
121:289–320. and economic sociology. In The handbook of
Bergmann, Barbara. 2005. The economic emergence of economic sociology, edited by N. J. Smelser and
women. 2nd ed. New York: Basic Books. R. Swedberg, 627–49. New York: Russell Sage
Bianchi, Suzanne, John P. Robinson, and Melissa Foundation.
A. Milkie. 2006. Changing rhythms of American England, Paula, Carmen Garcia-Beaulieu, and
family life. New York: Russell Sage Foundation. Mary Ross. 2004. Women’s employment among
Burstein, Paul. 1989. Attacking sex discrimination Blacks, whites, and three groups of Latinas: Do
in the labor market: A study in law and politics. more privileged women have higher employ-
Social Forces 67:641–65. ment? Gender & Society 18 (4): 494–509.
Charles, Maria. 2011. A world of difference: Interna- England, Paula, Janet C. Gornick, and Emily Fitz-
tional trends in women’s economic status. Annual gibbons Shafer. 2008. Is it better at the top? How
Review of Sociology 37:355–71. women’s employment and the gender earnings
Charles, Maria, and Karen Bradley. 2002. Equal but gap vary by education in sixteen countries. Paper
separate: A cross-national study of sex segregation presented at the 2008 annual meeting of the
in higher education. American Sociological Review American Sociological Association.
67:573–99. England, Paula, and Su Li. 2006. Desegregation
———. 2009. Indulging our gendered selves: Sex stalled: The changing gender composition of
segregation by field of study in 44 countries. college majors, 1971–2002. Gender & Society
American Journal of Sociology 114:924–76. 20:657–77.
Charles, Maria, and David B. Grusky. 2004. Oc- England, Paula, and Elizabeth Aura McClintock.
cupational ghettos: The worldwide segregation of 2009. The gendered double standard of aging in
women and men. Stanford, CA: Stanford Univer- U.S. marriage markets. Population and Develop-
sity Press. ment Review 35:797–816.
Cohen, Philip N., and Suzanne M. Bianchi. 1999. England, Paula, Emily Fitzgibbons Shafer, and Ali-
Marriage, children, and women’s employment: son C. K. Fogarty. 2008. Hooking up and form-
What do we know? Monthly Labor Review ing romantic relationships on today’s college
122:22–31. campuses. In The gendered society reader, 3rd ed.,
Cotter, David A., Joan M. Hermsen, and Paula En- edited by Michael Kimmel and Amy Aronson,
gland. 2008. Moms and jobs: Trends in mothers’ 531–46. New York: Oxford University Press.
employment and which mothers stay home. In Ferree, Myra Marx. 1976. Working-class jobs: Paid
American families: A multicultural reader, 2nd work and housework as sources of satisfaction.
ed., edited by Stephanie Coontz, with Maya Social Problems 23 (4): 431–41.
Parson and Gabrielle Raley, 379–86. New York: Goldin, Claudia, and Maria Shim. 2004. Making
Routledge. a name: Women’s surnames at marriage and be-
Cotter, David A., Joan M. Hermsen, and Reeve yond. Journal of Economic Perspectives 18:143–60.
Vanneman. 2004. Gender inequality at work. Hamilton, Laura, and Elizabeth A. Armstrong.
New York: Russell Sage Foundation. 2009. Gendered sexuality in young adulthood:
———. 2009. End of the gender revolution website, Double binds and flawed options. Gender & So-
http://www.bsos.umd.edu/socy/vanneman/end- ciety 23:589–616.
ofgr/default.html (accessed December 14, 2009). Kilbourne, Barbara Stanek, Paula England, George
Edin, Kathryn, and Maria Kefalas. 2005. Promises I Farkas, Kurt Beron, and Dorothea Weir. 1994.
can keep: Why poor women put motherhood before Returns to skills, compensating differentials, and
marriage. Berkeley: University of California Press. gender bias: Effects of occupational character-
England, Paula 1992. Comparable worth: Theories istics on the wages of white women and men.
and evidence. New York: Aldine. American Journal of Sociology 100:689–719.
England, Paula, Paul Allison, Su Li, Noah Mark, Kuperberg, Arielle, and Pamela Stone. 2008. The
Jennifer Thompson, Michelle Budig, and Han media depiction of women who opt out. Gender
Sun. 2007. Why are some academic fields tip- & Society 2:497–517.
ping toward female? The sex composition of U.S. Levanon, Asaf, Paula England, and Paul Allison.
fields of doctoral degree receipt, 1971–2002. So- 2009. Occupational feminization and pay: As-
ciology of Education 80:23–42. sessing causal dynamics using 1950–2000 census
data. Social Forces 88:865–92.
 Cotter, Hermsen, and Vanneman—Anti-Feminist Backlash and Trends in Gender Attitudes • 965

Mare, Robert D. 1991. Five decades of educational Percheski, Christine. 2008. Opting out? Cohort
assortative mating. American Sociological Review differences in professional women’s employment
56:15–32. rates from 1960 to 2000. American Sociological
McLanahan, Sara. Forthcoming. Children in fragile Review 73:497–517.
families. In Families in an unequal society, edited Ridgeway, Cecilia L. 2009. Framed before we know
by Marcia Carlson and Paula England. Stanford, it: How gender shapes social relations. Gender &
CA: Stanford University Press. Society 23:145–60.
McQuillan, Julia, Arthur L. Greil, Karina M. Shref- Risman, Barbara J. 2009. From doing to undo-
fler, and Veronica Tichenor. 2008. The impor- ing: Gender as we know it. Gender & Society
tance of motherhood among women in the 23:81–84.
contemporary United States. Gender & Society Sassler, Sharon, and Amanda Miller. 2007. Waiting
22:477–96. to be asked: Gender, power and relationship pro-
Morris, Martina, and Bruce Western. 1999. Inequal- gression among cohabiting couples. Presented at
ity in earnings at the close of the twentieth cen- the annual meeting of the American Sociological
tury. Annual Review of Sociology 25:623–57. Association, New York, August.
National Center for Education Statistics. 1971– Steinberg, Ronnie J. 2001. Comparable worth in
2003. Digest of education statistics. Washington, gender studies. In International encyclopedia of
DC: Government Printing Office. the social and behavioral sciences, vol. 4, edited by
———. 2004–2007. Table numbers by year: 2004: Neil J. Smelser and Paul B. Baltes. Cambridge:
253, 2005: 252, 2006: 58, 2007: 264. 2008: Cambridge University Press.
279. http://nces.ed.gov/programs/digest (ac- West, Candace, and Donald H. Zimmerman. 1987.
cessed December 14, 2009). Doing gender. Gender & Society 1:125–51.

112 • David Cotter, Joan M. Hermsen, and Reeve Vanneman

The Anti-Feminist Backlash and Recent Trends in

Gender Attitudes

Since Myra Marx Ferree (1974) first noted the liberal gender attitudes had recently slowed,
increased acceptance of voting for women for “the liberalization of gender beliefs has not yet
president, a steady series of studies has docu- run its course” (485). Others have echoed that
mented rising egalitarianism in gender attitudes optimism.
and made guarded but optimistic predictions As more recent data became available, re-
about the future. For example, Karin Brew- searchers became more cautious: “There may
ster and Irene Padavic (2000), using General have been not only a leveling off of the egal-
Social Survey (GSS) data from 1977 to 1996, itarian trend in the late 1990s but a small re-
found that although the pace of change toward versal of a long-term pattern” (Thornton and

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: David Cotter, Joan M. Hermsen, and Reeve Vanneman, “The End of the Gender
Revolution? Gender Role Attitudes from 1977 to 2008,” American Journal of Sociology 117:1 (July 2011), pp.
259–289, published by the University of Chicago Press. The article printed here was originally prepared by
David Cotter, Joan M. Hermsen, and Reeve Vanneman for this fourth edition of Social Stratification: Class,
Race, and Gender in Sociological Perspective, edited by David B. Grusky. Copyright © 2014 by Westview Press.
966 • Part VIII: Gender Inequality

Figure 112.1. Four Gender Attitudes, 1977–2008

Source: General Social Surveys, 1977–2008.

Note: Strongly agree (disagree) and agree (disagree) are combined. Don’t know responses are coded as not egalitarian.

Young-DeMarco, 2001, 1014). Figure 112.1 racial-ethnic groups (except Asian Americans)
shows these trends for four gender attitudes from and all levels of education and income. We test
the GSS. After an egalitarian peak in 1994, gen- and reject a variety of the most readily identi-
der attitudes fell until the end of the twentieth fiable social structural causes of the reversal in
century. Two of the questions show a moderate the 1990s. Although several demographic and
resurgence since 2000, one has changed little, structural factors can help explain the rise of lib-
and the fourth has continued to decline in the eral attitudes throughout the 1970s and 1980s,
twenty-first century. Nevertheless, the overall the decline after the mid-1990s is robust to all
patterns clearly break into two broad eras: a uni- controls. Further, the change in gender attitudes
form trend toward less traditional gender roles after the mid-1990s does not appear to be part of
during the 1970s and 1980s, but only small and a broader period shift toward more conservative
shifting changes since the mid-1990s. political and family attitudes.
This chapter asks what structural and cultural We suggest an antifeminist backlash in the
changes might account for this unexpected stall- popular culture as the most likely explanation
ing of gender attitudes. We show that the shift for the attitude shift. At least since the publi-
in the mid-1990s can be observed within almost cation of Susan Faludi’s Backlash (1991), media
all cohorts, among both men and women of all and cultural critics have warned that a cultural
 Cotter, Hermsen, and Vanneman—Anti-Feminist Backlash and Trends in Gender Attitudes • 967

shift was threatening the feminist gains of the educational, income, and employment—can
1970s. In recent years the pace of these warnings help explain the attitude changes. Finally, we
has increased. The growth of “intensive moth- take a closer look at the cultural changes of the
ering” (Hays 1996) has put enormous pressures 1980s and 1990s to identify new elements in
on women to choose between careers and moth- the latter decade that better coincide with the
erhood (Blair-Loy 2003). The media (Belkin observed attitude shifts. We argue that the most
2003; Story 2005) have been quick to document promising explanation appears to be the rise in
working mothers who “opt out” of their careers. the 1990s of this new cultural frame that could
The pattern revealed in the GSS attitudes pro- claim to be egalitarian while still reinforcing tra-
vides the first quantitative evidence that the cul- ditional gender roles.
tural attacks on feminist equality have had an
impact on public consciousness.
We argue that the result has been not a rever- Methods
sion to the gender traditionalism of the 1950s, Variables
but rather the rise of a third cultural frame of Gender attitude scale. We use the General Social
“egalitarian essentialism” (Charles and Grusky Survey (GSS) data (Davis and Smith 2007) to
2004), combining support for stay-at-home analyze changes in gender attitudes. Because
mothering with a continued feminist rhetoric our main focus is changes in the last decade, we
of choice and equality (Stone 2007). We be- have developed a scale using only the four items
lieve this cultural explanation is also consistent (FEPOL, FECHLD, FEPRESCH, FEFAM)
with the broader pattern of gender changes that that were asked in 1977, 1985–1986, and then
shifted in the mid-1990s (see Cotter, Hermsen, continuously from 1988 to 2008.1 The scale is
and Vanneman 2004 for a review). constructed by taking the mean of the standard-
ized scores for each of the four items (Cronbach
alpha = 0.71).
Background
One place to look for possible causes of the Cohort. To evaluate the role of cohort replace-
mid-1990s shift is changes in the factors that ment, we include controls for year of birth. In
explained the initial liberalizing trends in the addition to a linear year of birth variable, we
1970s. Among the factors most commonly cited also include a cohort spline at 1952, which dif-
as causes of the liberalizing gender trends in the ferentiates between pre– and post–baby boom
1970s and 1980s are cohorts, suggested by earlier research, to mark
• Cohort replacement of early conservative a slowing of cohort-related change (Brooks and
cohorts by recent liberal cohorts, Bolzendahl 2004; Schnittker, Freese, and Powell
• Social structural changes such as increasing 2003). The cohort spline is coded as 0 for birth
education and declining fertility, years up to 1952 and then increases by 1 for each
• The entry of women into the labor force, subsequent birth year (e.g., 1952 = 1, 1953 = 2).
and
• The rise of the second wave of the wom- Social structural controls. We include a broad
en’s movement. range of social structural, women’s employment,
and ideological controls that have been shown
Each of these factors can plausibly be amended in past research to be related to gender attitudes.
to explain a downturn in the 1990s. Moreover, Our purpose is not so much to analyze the extent
empirical tests for several of them are readily of these associations (although several are theo-
available using existing survey data. retically interesting) as to see whether changes in
We proceed as follows. First we measure and any of these help account for the changes over
describe the trend in the GSS gender items. Then time in gender attitudes, particularly for the
we test what structural changes—demographic, turnaround beginning in 1994.
968 • Part VIII: Gender Inequality

Figure 112.2. Gender Role Scale, 1977–2008

Source: General Social Surveys, 1977–2008.

Note: The gender role scale is the mean of the four items listed in Figure 112.1, each standardized by its 1987–1998 mean
and standard deviation.

Results scale trends separately for various subgroups and

tested for significant group differences. The re-
Figure 112.2 reports the gender attitude scale sults of seven of these models are summarized in
means for each survey year in which the four gen- Table 112.1.
der role attitudes were covered. Following Hout In general, most groups show the same basic
and Fischer (2002), the figure also fits the ob- changes over time: more liberal attitudes in the
served means with the best-fitting spline function 1970s and 1980s, a downturn at 1994, and some
that has “knots” at 1994 and 2000. After 1994, rebound after 2000. There are almost no major
not only does the trend cease rising, it turns differences by gender, race/ethnicity, or cohort.
downward until 2000, when it begins to rise The gender attitude turnaround in the 1990s
again, although more slowly than in the 1980s. was sharpest for respondents with children and
for high-income households. These results argue
that the changes of the mid-1990s reflected a gen-
Group Differences eral cultural change in gender attitudes, but that
How general are these gender attitude trends, es- change was felt most strongly by high-income
pecially the turnaround in the 1990s? To begin households with working mothers, for whom
to answer this question, we computed the gender work-family stresses were most relevant.
 Cotter, Hermsen, and Vanneman—Anti-Feminist Backlash and Trends in Gender Attitudes • 969

Table 112.1. Slopes of Year and Year Splines, Separately for Selected Groups
Year Spline Year Spline
Unweighted N Year (knot = 1994) (knot = 2000)
Total Sample 22,770 0.040*** -0.065*** 0.042***
By Gender
Women 12,797 0.042 -0.068 0.047
Men 9,973 0.038 -0.062 0.038
By Race/Ethnicity
White (reference) 17,913 0.041 -0.070 0.053***
Black 2,886 0.036 -0.056 0.045
Hispanic 1,146 0.044 -0.069 0.002*
Asian 449 0.004** 0.015** -0.002
By Cohort
1925–1934 (reference) 2,297 0.027 -0.053 0.030
1935–1944 2,973 0.028 -0.079 0.076+
1945–1954 4,479 0.023 -0.062 0.042
1955–1964 5,024 0.029 -0.078 0.079*
1965–1974 3,343 0.041+ -0.068 0.023
By Education
12 years 22,770 0.034 -0.057 0.036
16 years 0.029** -0.062 0.056*
By Family Income
(log constant )
$10,000 20,283 0.040 -0.054 0.023
$200,000 0.041 -0.083* 0.069*
By Marital and
Parental Status
Married with children (ref.) 6,105 0.043 -0.087 0.065
Married no children 5,506 0.043 -0.065* 0.045
Single with children 8,255 0.040 -0.073 0.032
Single no children 2,754 0.035* -0.048*** 0.032*
By Woman/Wife in
Labor Force
Not in labor force 7,679 0.038 -0.058 0.042
In labor force 10,688 0.036 -0.071 0.055

Note: Slopes by groups are calculated from 7 separate regressions of the gender scale on survey year and
year interactions with each group. Significance tests represent differences from the reference group (the
first line in each panel). Education and income are entered in regressions as continuous variables. Slopes
are calculated at representative low and high points from the education and income distributions.
Source: General Social Surveys, 1977–2008.
+p < .10, *p < .05, **p < .01, ***p < .001
the size of the “hinge” in the spline functions.
Explanations of the Trends For example, if the cohort replacement effect
Next we use a series of structural and ideological has slowed down, then cohort controls not only
variables to explain the observed curve: the rise should explain the liberal rise in the 1970s and
in the 1970s through the 1980s, another smaller 1980s, but also should eliminate or reduce the
rise since 2000, and especially the downturn in downward turn in 1994.
the late 1990s. The logic here is that if a change Table 112.2 reports tests for these factors.
in US society accounted for the turnaround, The first column repeats the spline function co-
then controlling for such a change should reduce efficients from the first line of Table 112.1 that
970 • Part VIII: Gender Inequality

Table 112.2. Regressions of Gender Attitude Scale, Selected Coefficients

Model 3 + Model 4
Model 1 Model 2 women’s social + women’s
base +cohort structure employment
Regression Coefficients
Intercept -80.216*** -90.995*** -76.589*** -70.747***
Survey year 0.040*** 0.025*** 0.022*** 0.023***
Survey year spline (knot = 1994) -0.068*** -0.062*** -0.061*** -0.061***
Survey year spline (knot = 2000) 0.045*** 0.047*** 0.053*** 0.054***
Cohort: year of birth 0.021*** 0.016*** 0.013***
Cohort: year of birth spline (1952) -0.018*** -0.014*** -0.010***
Social structural controls No No Yes Yes
Women’s employment No No No Yes
Calculated Period Slopes
1977–1994 0.040 0.025 0.022 0.023
1994–2000 -0.027 -0.037 -0.039 -0.039
2000–2008 0.018 0.010 0.014 0.015
Calculated Cohort Slopes
Pre-1952 0.021 0.016 0.013
Post-1952 0.003 0.003 0.003
R-square 0.037 0.130 0.230 0.245
N= 21,925 21,925 21,925 21,925
Source: General Social Surveys, 1977–2008.
*p < .05, **p < .01, ***p < .001

were illustrated in Figure 112.2. Our goal in ex- surveys in Figure 112.2 is that a much-reduced
plaining the reversal of the 1990s is to reduce the cohort replacement effect continues to push up
hinge coefficient to zero (i.e., nonsignificance). the national average. In sum, only the 1977 to
The second hinge at 2000 is positive and also 1994 period shows any strong evidence that
significant, indicating that some of the decline in individual Americans’ gender attitudes became
the 1990s was reversed in the next decade. Nev- more egalitarian.
ertheless, the second spline coefficient is smaller
than the first, so not all the forward progress was Social structural controls. How much of the pe-
regained. riod (and cohort) change can be explained by
alterations in the composition of the American
Cohort controls. The differences among the co- public? We test for the effects of these structural
horts before the baby boom are quite large; from changes on the 1990s reversal in two steps. First,
the baby boom onward the differences become in column 3 we add controls for education, reli-
much more modest, although there is no indi- gion, ethnicity, family income, and family struc-
cation that the most recent cohorts are actually ture, and then in a second step, because of its
more conservative than the generation that pio- special problems of endogeneity, add the control
neered the second wave of feminism. for whether there is a working woman in the
Column 2 shows the effects of controls for family. All of these factors have their expected
birth cohort and the cohort spline. Much of and now well-documented impacts on gender
the change in the 1970s and 1980s is explained attitudes, but none of these controls changes the
by the cohort controls. However, only a small period coefficients very much.
part (8 percent) of the reversal of the 1990s is The story is quite different for the cohort
explained by cohort controls. A main reason effects. These are more easily explained by the
for the mild liberalizing trend in the post 2000 changes in American society included in Model
 Cotter, Hermsen, and Vanneman—Anti-Feminist Backlash and Trends in Gender Attitudes • 971

Table 112.3. Regressions of Gender Attitude Scale with Political Ideology, Selected Coefficients
Model 1 Model 4 Model 5
Regression Coefficients
Intercept -79.947*** -70.330*** -64.863***
Survey year 0.040*** 0.023*** 0.024***
Survey year spline (knot = 1994) -0.066*** -0.058*** -0.058***
Survey year spline (knot = 2000) 0.044*** 0.053*** 0.051***
Cohort: year of birth 0.013*** 0.009***
Cohort: year of birth spline (1952) -0.010*** -0.005***
Social structural controls No Yes Yes
Women’s employment controls No Yes Yes
Ideology and political controls No No Yes
Calculated Period Slopes
1977–1994 0.040 0.023 0.023
1994–2000 -0.025 -0.035 -0.034
2000–2008 0.018 0.018 0.017
Calculated cohort slopes
Pre-1952 0.013 0.009
Post-1952 0.003 0.004
R-square 0.055 0.253 0.308
N= 11,446 11,446 11,446
Source: General Social Surveys, 1977–2008.
*p < .05, **p < .01, ***p < .001

3. Increasing education is important in explain- after the controls for background factors. Like
ing the rising liberalism among early cohorts. the social structural changes, the ideological
Perhaps more interesting, the weakening of the changes help explain the cohort differences, but
cohort effect in the post-1952 cohorts is in part not the period effects. The negative coefficient
a result of the weakening of many of these struc- for the 1994 period spline barely changes after
tural changes in the post-1952 cohorts. adding the political and rights ideology con-
Although women’s labor force participation is trols. In contrast, the rights ideology is closely
strongly associated with gender role attitudes, it associated with the cohort differences in gender
explains almost none of the period effects. Thus, attitudes. The post-1952 birth cohorts that favor
the structural changes in American society, espe- more egalitarian gender roles are the same co-
cially higher education and more women work- horts that favor a more tolerant acceptance of
ing in higher status positions, helped give more social and political minorities. Change toward
recent cohorts more liberal gender attitudes. But this more liberal ideology was rapid before 1952
few of the period effects, and almost none of but slower after the baby boom. Period changes
the reversal of the 1990s, can be related to these in gender attitudes are quite different. Neither
structural changes. the general period rise in gender egalitarianism
in the 1970s and 1980s nor its turnaround after
Ideological changes. Our last possibility for ex- 1994 is associated with period trends in liberal
plaining the attitude reversal of the 1990s is to ideology.
look at the broader ideological climate that may
have become more conservative in the 1990s.
The results of the controls for the rights ideology Discussion
scale, political conservatism, and party identifi- Changes in gender attitudes in the last three
cation are reported in Table 112.3. Both ideology decades can be broken into two distinct phases:
variables and Democratic Party identification are (1) a steady liberalization that (2) halted in the
strongly associated with gender attitudes, even mid-1990s and has changed little in the years
972 • Part VIII: Gender Inequality

since. There has been a slight and interesting of hierarchical gender relations of submissive
rebound since 2000, but at nowhere near the wives and decision-making husbands, but as a
rate of change of the 1970s and 1990s. About sensible choice for women to maximize their
half of the increase until the mid-1990s was a own and their children’s best interests. This new
cohort replacement effect, with more recent, lib- frame would be most persuasive for those house-
eral cohorts replacing earlier, more conservative holds facing the strains of work-family conflicts;
cohorts. This weaker effect in recent cohorts ac- that is, for parents and households with working
counts for a small part of the mid-1990s turn- women—–precisely those households that the
around, but most of the post-1990s leveling is GSS shows had the strongest turnaround in atti-
unexplained. tudes during the 1990s.
The cohort differences are more easily ex- Our analyses have shown that the reversal in
plained than the period effects. Cohort differ- gender attitudes in the 1990s had little basis in
ences in gender attitudes are best understood as any structural or broader ideological changes in
part of this same pattern of growing but deceler- American society. Instead, we need to explain
ating social liberalism among recent generations. period changes with changes in period events:
The period effects, however, are not well linked the rise and decline of feminist and antifemi-
to changes in American social structure or to nist movements; the shifts in popular culture,
broader changes in social ideology. especially the rise of a new cultural frame that
What is still most difficult to understand is incorporates traditional gender roles without
the timing of the reversal in gender attitudes in implying hierarchical power relations; and the
the mid-1990s rather than in the mid-1980s, independent emergence of new cultural con-
when a cultural backlash was already well estab- cerns such as intensive parenting and career
lished and feminist mobilization had subsided. stress, whose consequences, intentionally or not,
Recent analyses of the antifeminist backlash helped reverse past trends in norms of gender
have identified new themes in popular cul- equality.
ture that might explain a reversal in the 1990s
rather than in the 1980s. During the 1970s and NOTES
for much of the 1980s, the struggle over gen- 1. The four items in the scale are “Tell me if you
agree or disagree with this statement: Most men are
der relations had been framed in a rhetoric of
better suited emotionally for politics than are most
progressive equality versus on of older family
women” (FEPOL); “A working mother can establish
hierarchy tradition. In the 1990s a third cultural just as warm and secure a relationship with her chil-
frame emerged that proclaimed itself egalitarian dren as a mother who does not work” (FECHLD);
and woman-centered, but nevertheless opposed “A preschool child is likely to suffer if his or her
some of the structural changes that had moved mother works” (FEPRESCH); and “It is much bet-
American society toward a more feminist fu- ter for everyone involved if the man is the achiever
ture. In this alternative frame, equality meant outside the home and the woman takes care of the
the right of women to choose—so choosing a home and family” (FEFAM).
stay-at-home mother role could be as much a
REFERENCES
“feminist” choice as pursuing an independent
Belkin, Lisa. 2003. “The Opt-Out Revolution.” New
career (Williams 2000).
York Times Magazine, October 26.
Thus the new gender frame could still be Blair-Loy, Mary. 2003. Competing Devotions: Ca-
technically egalitarian between husband and wife reer and Family Among Women Executives. Cam-
and even emphasize the importance of women’s bridge, MA: Harvard University Press.
autonomous choices, but nevertheless support Brewster, Karin L., and Irene Padavic. 2000.
what still looks like a traditional gender division “Change in Gender-Ideology, 1977–1996: The
of labor. Agreement with the traditional alterna- Contributions of Intracohort Change and Pop-
tives of the GSS gender items could therefore be ulation Turnover.” Journal of Marriage and the
Family 62: 477–487.
reinterpreted in the 1990s not as an endorsement
 Ridgeway—The Persistence of Gender Inequality • 973

Brooks, Clem, and Catherine Bolzendahl. 2004. Hout, Michael, and Claude S. Fischer. 2002. “Ex-
“The Transformation of U.S. Gender Role At- plaining the Rise of Americans with No Religious
titudes: Cohort Replacement, Social-Structural Preference: Politics and Generation.” American
Change, and Ideological Learning.” Social Science Sociology Review 67: 165–190.
Research 33: 106–133. Hout, Michael, Andrew Greeley, and Melissa J.
Charles, Maria, and David B. Grusky. 2004. Occu- Wilde. 2001. “The Demographic Imperative in
pational Ghettos: The Worldwide Segregation of Religious Change in the United States.” American
Women and Men. Stanford, CA: Stanford Uni- Journal of Sociology 107: 468–500.
versity Press. Schnittker, Jason, Jeremy Freese, and Brian Powell.
Cotter, David A., Joan M. Hermsen, and Reeve 2003. “Who Are Feminists and What Do They
Vanneman. 2004. Gender Inequality at Work. Believe? The Role of Generations.” American So-
New York/Washington: Russell Sage Founda- ciological Review 68: 607–622.
tion/Population Reference Bureau. Stone, Pamela. 2007. Opting Out? Why Women Really
Davis, James A., and Tom W. Smith 2007. General Quit Careers and Head Home. Berkeley: Univer-
Social Surveys, 1972–2006 [machine-readable sity of California Press.
data file]. Principal Investigator, James A. Davis; Story, Louise. 2005. “Many Women at Elite Col-
Director and Co-Principal Investigator, Tom W. leges Set Career Path to Motherhood.” New York
Smith; Co-Principal Investigator, Peter V. Mars- Times, September 20.
den, NORC ed. Chicago: National Opinion Re- Thornton, Arland, and Deborah Freedman. 1979.
search Center, producer, 2005; Storrs, CT: The “Changes in Sex Role Attitudes of Women,
Roper Center for Public Opinion Research, Uni- 1962–1977.” American Sociological Review 44:
versity of Connecticut, distributor. 831–842.
Faludi, Susan. 1991. Backlash: The Undeclared War Thornton, Arland, and Linda Young-DeMarco.
Against American Women. New York: Anchor 2001. “Four Decades of Trends in Attitudes To-
Books. wards Family Issues in the United States: The
Ferree, Myra Marx. 1974. “A Woman for President? 1960s Through the 1990s.” Journal of Marriage
Changing Responses: 1958–1972.” Public Opin- and the Family 63: 1009–1037.
ion Quarterly 38: 390–399. Williams, Joan. 2000. Unbending Gender: Why Fam-
Hays, Sharon. 1996. The Cultural Contradictions of ily and Work Conflict and What to Do About It.
Motherhood. New Haven, CT: Yale University New York: Oxford.
Press.

113 • Cecilia L. Ridgeway

The Persistence of Gender Inequality

How does gender inequality persist in an ad- use of gender as a primary means for organizing
vanced industrial society like the contemporary their social relations with others is a powerful
United States, in which many economic, legal, social process that continually re-creates gen-
institutional, and political forces work against der inequality in new forms as society changes
it? In this chapter I argue that people’s everyday (Ridgeway 2011). The use of gender as an

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Cecilia Ridgeway, Framed by Gender: How Gender Inequality Persists in the Modern
World, published by Oxford University Press in 2011. The article printed here was originally prepared by Ce-
cilia Ridgeway for this fourth edition of Social Stratification: Class, Race, and Gender in Sociological Perspective,
edited by David B. Grusky. Copyright © 2014 by Westview Press.
974 • Part VIII: Gender Inequality

initial framing device for relationships with and defining “who” the self and other are in the
others spreads gendered meanings, including as- situation, so that we can anticipate how each
sumptions about inequality embedded in those of us is likely to act and coordinate our actions
meanings, beyond contexts associated with sex accordingly.
and reproduction to all spheres of social life that
occur through social relationships. Through gen-
der’s role in organizing social relations, gender Coordination and Difference
inequality is rewritten into new economic and Systems for categorizing and defining things
social arrangements as they emerge, preserving are based on contrast and therefore difference:
that inequality in modified form throughout so- something is this because it is different from that.
cioeconomic transformations. Defining self and other to relate, then, focuses us
This persistence process turns on the fact that on finding shared principles of social difference
cultural beliefs about gender change more slowly that we can use to categorize and make sense of
than do material arrangements between men and one another. I argue that the coordination prob-
women, even though these beliefs eventually re- lem inherent in organizing social relations drives
spond to material changes. As a result of this populations of people who must regularly relate
cultural lag, at the edge of social change peo- to one another to develop shared social category
ple confront their new, uncertain circumstances systems based on culturally defined standards of
with gender beliefs that are more traditional than difference.
those circumstances. As they implicitly draw on To manage social relations in real time, some
the too-convenient cultural frame of gender to of these cultural category systems must be so
help organize their new ways of doing things, simplified that they can be quickly applied as
they reinscribe trailing cultural assumptions framing devices to virtually anyone to start the
about gender difference and inequality into the process of defining self and other in the situa-
new activities, procedures, and forms of orga- tion. In fact, studies of social cognition suggest
nization that they create, in effect reinventing that a very small number of such cultural differ-
gender inequality for a new era. In the following ence systems, approximately three, serve as the
discussion I explain this argument and some of primary categories of person perception in a so-
the evidence behind it. ciety (Brewer and Lui 1989; Fiske 1998). These
categories define the things a person in that so-
ciety must know about another person to render
Gender as a Primary Frame that someone sufficiently meaningful to relate to
What do I mean by gender being a primary cul- him or her. In the United States sex, race, and
tural frame for social relations? We all depend for age are primary categories (Schneider 2004, 96).
our very survival on social relations with others, Social cognition studies show that we auto-
but these relations pose a well-known problem: matically and nearly instantly sex categorize any
to relate to another person to accomplish a val- specific person we attempt to relate to (Stangor
ued goal, we and the other have to find some et al. 1992; Ito and Urland 2003). Our subse-
way to coordinate our behavior. Classic sociol- quent categorizations of people as, say, bosses or
ogists like Goffman (1967) and contemporary coworkers, are nested in our prior understand-
game theorists (Chwe 2001) have arrived at the ings of them as males or females and take on a
same conclusion about what it takes to solve this slightly different meaning as a result, as cogni-
coordination problem: for you and me to coor- tive research has shown (Brewer and Lui 1989;
dinate effectively, we need shared, “common” Fiske 1998; Macrae and Quadflieg 2010). This
knowledge, which of course is cultural knowl- happens not just in person, but also over the
edge, to use as a basis for our joint actions. In Internet and even in imagination, as we exam-
particular, we need a shared way of categorizing ine a person’s resume or think about the kind of
 Ridgeway—The Persistence of Gender Inequality • 975

person we would like to hire. This initial framing and makes them cognitively available to shape
by sex never quite disappears from our under- behavior and judgments (Blair and Banaji 1996;
standing of others or of ourselves in relation to Kunda and Spencer 2003). The extent to which
them. they actually do shape our behavior, however,
can range from negligible to substantial, de-
pending on the nature of the particular situation
Cultural Beliefs About Gender and our own motives or interests in it (Macrae
Primary categories of person perception, in- and Quadflieg 2010; Wagner and Berger 1997).
cluding sex category, work as cultural frames for What matters is the extent to which the infor-
coordinating behavior by associating category mation in gender beliefs is diagnostic for us,
membership with widely shared cultural beliefs in that it helps us figure out how to act in the
about how people in one category are likely to situation.
behave compared to those in a contrasting cate- As part of the primary person frame, the in-
gory—in other words, stereotypes. We all know structions for behavior encoded in gender ste-
common gender stereotypes as cultural knowl- reotypes are exceedingly abstract and diffuse. For
edge, even though many of us do not personally this very reason they can be applied to virtually
endorse them (Diekman and Eagly 2000; Glick any situation, but by the same token, they do
et al. 2004). Because we think “most people” not take an actor very far in figuring out exactly
hold these beliefs, we expect others to judge us how to behave. In contrast, the institutional
according to them. As a result, we must take frameworks that govern the contexts in which
these beliefs into account in our own behavior most everyday social relations take place are
even if we do not agree with them. In this way much more specific. They contain defined roles
these shared cultural beliefs act as the “rules” for and the expected relationships among them. For
coordinating public behavior on the basis of gen- individuals, these institutional identities and
der. And as we would expect from the “rules” rules are in the foreground of their sense of who
for gender, widely shared gender stereotypes they are and how they should behave in most
have been shown to have a prescriptive as well settings. Gender, in contrast, is almost always a
as descriptive quality (Eagly and Karau 2002; background identity for individuals. As a back-
Prentice and Carranza 2002). ground identity, gender typically acts to bias in
The content of our gender stereotypes shows gendered directions the performance of behav-
the characteristic pattern of status inequality, iors undertaken in the name of more concrete,
in which the higher status group is perceived foregrounded organizational roles or identities.
as more proactive and agentically competent Thus gender becomes a way of acting like a doc-
(“from Mars”), and the lower status group is tor or managing a team project (West and Zim-
seen as more reactive and emotionally expressive merman 1987).
(“from Venus”) (Cuddy, Fiske, and Glick 2007; A wide variety of research shows that the
Conway, Pizzamiglio, and Mount 1996; Fiske extent and direction of the biases the back-
et al. 2002; Wagner and Berger 1997). Conse- ground gender frame introduces into our ex-
quently, coordination on the basis of our shared pectations for self and others, our behavior, and
gender beliefs creates social relations of inequal- our judgments fall into a distinctive pattern (see
ity as well as difference. Ridgeway 2001; Ridgeway and Correll 2004;
Ridgeway and Smith-Lovin 1999 for reviews). In
mixed sex settings in which the task or context
How Does the Gender Frame Shape is relatively gender neutral, cultural beliefs that
Behavior? men are more agentically competent and more
Research shows that sex categorization uncon- worthy of status advantage them over otherwise
sciously primes gender stereotypes in our minds similar women, but only modestly so. In settings
976 • Part VIII: Gender Inequality

that are culturally typed as masculine, gender be- status in society. Then I turn to evidence that
liefs bias judgments and behaviors more strongly lagging beliefs infect sites of innovation.
in favor of men. In contexts culturally linked
with women, biases weakly favor women, except
for positions of authority. Men are advantaged Do Gender Beliefs Lag Behind?
for authority in all settings. When changing social circumstances, such as
women’s growing commitment to the labor
force, cause people to have more and more
The Persistence Process gender-atypical experiences, why would this
We can see how this pattern of implicit biases, not simply be reflected in changed gender ste-
acting through the everyday social relations that reotypes? The answer is that it will eventually,
make up the worlds of work and home, reinforces but the impact of atypical experiences is blunted
and helps reproduce the structures of gender in- by two processes that slow down changes in
equality with which we are familiar: the gender stereotypes. At the individual level gender
gap in wages and authority (cf. Cotter, Hermsen, stereotypes are subject to powerful confirma-
and Vanneman 2004), the sex-segregated job tion biases that cause people to resist noticing
structure (cf. Charles and Grusky 2004), and the gender-inconsistent information or to reinter-
unequal division of household labor (cf. Bian- pret it as consistent (von Hippel, Sekaquaptewa,
chi, Robinson, and Milkie 2006). The everyday and Vargas 1995; Dunning and Sherman 1997).
reproduction of inequality within established in- Even more important, however, is a social process
stitutional structures is an important part of the that derives from the use of gender as a frame for
persistence process. But the future of gender as a coordinating behavior. People’s assumption that
principle of inequality lies with sites at the edge gender stereotypes are “common knowledge”
of social and technological change, where sub- rules for gendered behavior inhibits the public
stantially new forms of work or of heterosexual display of explicitly disconfirming behavior or
unions are innovated. Some of these innovations information (Clark and Kashima 2007; Rud-
then become blueprints for new industries and man and Fairchild 2004; Seachrist and Stangor
ways of life. 2001). This further reinforces people’s assump-
I argue that the gender frame, acting through tion that “most people” hold these stereotypes.
social relations, infuses assumptions about gen- As we would expect, then, evidence suggests
der inequality even into innovative contexts that gender stereotypes have lagged behind
that are explicitly devoted to doing things dif- changes in women’s roles in society. Studies show
ferently. Sites of innovation are typically small, a slight narrowing of the gender gap in instru-
interpersonally oriented settings. Also, they are mental, agentic traits over the last few decades,
settings in which the determination to do things in that women are now seen as almost the same
differently weakens the normally powerful im- as men on the softer aspects of agency like “an-
pact of existing institutional schemas as guides alytical” or “reasoning.” Women, however, are
to behavior. Both factors increase the likelihood still rated much lower than men in the stronger
that participants will unconsciously fall back on aspects of agency like “aggressive,” “forceful,”
familiar person-based schemas like gender to “leaderlike,” or “competitive.” Women also are
help organize their inherently uncertain task. still rated much higher than men in communal
The shared gender stereotypes people have to traits like warmth (Cejka and Eagly 1999; Diek-
fall back on trail behind changes in the material man and Eagly 2000; Prentice and Carranza
circumstances they find themselves in. 2002; Spence and Buckner 2000). In common
What is the support for these arguments? First stereotypes, then, men are still from Mars and
I examine evidence that cultural beliefs about women are still mostly from Venus, despite mas-
gender lag behind changes in women’s material sive changes in women’s labor force involvement
 Ridgeway—The Persistence of Gender Inequality • 977

since the 1970s. The next question is, do these workforce in this field, cultural beliefs about
lagging stereotypes have an impact on emer- gender should be salient in biotech firms, but
gent forms of gender relations at sites of social only diffusely so, and we would expect these be-
innovation? liefs to create only modest advantages for men
in expected competence. Facing only modest bi-
ases, women scientists in biotech should have the
Sites of Innovation basic credibility with their coworkers that they
Recall that the gender frame produces a distinc- need to take effective advantage of the opportu-
tive pattern of implicit biases that vary systemat- nities offered by the flexible structure of innova-
ically by context. The most telling evidence that tive firms. In fact, Whittington and Smith-Doerr
the gender frame reestablishes gender inequality (2008) find women life scientists do better in
at sites of innovation would come from sites for these innovative biotech firms in terms of super-
which the argument makes different predictions visory positions and patents earned than they do
about the degrees of inequality likely to emerge. in more traditional research organizations like
An example of such evidence from the work pharmaceutical firms. Even in these innovative
world focuses on high-tech start-ups. firms, however, women attain fewer total patents
than comparable men. This disadvantage is not
surprising if we remember that background gen-
Hi-tech Start-ups der biases still modestly favor men even in this
Whittington and Smith-Doerr examined how innovative biotech context.
women scientists fared in innovative start-ups In contrast to the life sciences, engineer-
compared to more traditional, established re- ing and the physical sciences are still strongly
search firms (Smith-Doerr 2004; Whittington sex-typed in favor of men. Thus the background
2007; Whittington and Smith-Doerr 2008). gender frame in the IT context is more power-
Some of the start-ups were biotech firms based fully relevant and creates stronger implicit biases
in the life sciences. Others were information against women’s competence than in biotech set-
technology (IT) firms in engineering and the tings. In this situation we would expect the in-
physical sciences. All had adopted a new kind formality and flexibility of the innovative firm to
of organizational form in which work is orga- be, if anything, a disadvantage for women scien-
nized in terms of project teams that are often tists. Facing strong challenges to their credibility,
jointly constructed with other firms. Scientists in it will be more difficult for women to effectively
a firm move flexibly among these project teams, take advantage of the flexible structure. Also, in
and the hierarchies of control over their activities the context of a masculine-typed gender frame,
are relatively flat. Is this informal, flexible, less the informal work structure may lead to a “boys’
hierarchical structure advantageous or disadvan- club” atmosphere. Not surprisingly, then, Whit-
tageous for women scientists who work in these tington (2007) found that women physical sci-
high-tech firms? The gender frame argument entists and engineers were just as disadvantaged
predicts that the answer will be different for bio- in terms of patents in small, flexible start-ups as
tech firms than for IT firms, even though these in traditional industrial research firms. Another
firms have the same organizational structure. study found women engineers actually did better
Applying the distinctive pattern of biases cre- in a traditional, rule-structured aerospace firm
ated by the background gender frame to biotech than in a more informal start-up, because in the
start-ups, here is what we would expect. The life context of a disadvantaging background gender
sciences are no longer strongly sex-typed in con- frame, formal rules leveled the playing field to
temporary culture. Women earn half the PhDs some extent (McIlwee and Robinson 1992).
in this area (England et al. 2007). Therefore, be- Since the organizational form of the firms
cause of the mixed gender composition of the is held constant in these studies, it is especially
978 • Part VIII: Gender Inequality

likely that the varying effects of the gender rewrite gender inequality into the new cultural
frame are behind these differing patterns of in- norms and practices that they forge for this in-
equality between the biotech and IT start-ups. novative type of intimate encounter.
Other studies suggest that these initial patterns Studies of contemporary college students
of gender inequality at the start-up stage of a show (Armstrong, England, and Fogarty 2012;
firm or industry potentially have long-range ef- England, Shafer, and Fogarty 2008) that women
fects. Gender relations in these start-ups work are only slightly less likely than men to initiate
together with the organizational logic of the the intimate talking and dyadic focus that lead
firms to shape the workplace practices, rou- to a hook-up, but men still usually initiate the
tines, and procedures that define work in that move to sexual behavior. And both men and
firm. These workplace practices and routines women students report that sexual activities
then carry the implicit stamp of those found- during hook-ups are oriented more toward the
ing gender inequalities and, as the practices and man’s sexual pleasure than the woman’s. Further-
routines become institutionalized, they become more, traditional double standards persist in the
an independent agent by which the gender re- social reputations that men and women acquire
gime is reproduced within the firm (Baron et for having hook-ups with many partners. De-
al. 2007). These implicitly gendered procedures spite the increasingly egalitarian material terms
also spread to other firms (Phillips 2005). Thus, on which college men and women encounter
the way that the background frame of gender in- one another, lagging cultural beliefs about gen-
fuses gender inequality into the organizational der continue to cause substantial inequality in
routines of small, pioneering firms potentially close heterosexual bonds.
has long-range consequences for the future of
gender inequality in the workplace as the econ-
omy changes and new industries emerge. Conclusion
In contexts as diverse as high-tech start-ups and
college student hook-ups, our everyday use of
Innovative Intimate Unions gender as a basic framing device for social re-
The work world is important for the persistence lations is a powerful, if implicit, process by
of gender inequality, but so is the world of in- which gender as a principle of social inequality
timate heterosexual unions. Although there is reinscribed into new forms of social and eco-
has been less change in the gendered structure nomic organization as they emerge in society.
of heterosexual unions than in paid work, new, This persistence dynamic does not mean that
innovative social forms do develop even in this gender inequality can never be overcome. After
sphere. An example is the way the casual date on all, powerful economic, legal, and institutional
college campuses has increasingly been replaced forces, as well as women’s own self-interests, all
by the explicitly egalitarian “hook-up” (England, work against gender inequality in the modern
Shafer, and Fogarty 2008; Hamilton and Arm- world. But this persistence dynamic does mean
strong 2009). that the forces for change are not unopposed. As
Because of the intensely gendered nature of a result, progress toward equality is likely to be
the hook-up context, the gender frame argument both uneven and uncertain.
predicts that students will pursue their intended
egalitarian goals against a powerful if implicit REFERENCES
backdrop of lagging gender stereotypes about Armstrong, Elizabeth A., Paula England, and Alison
C. K. Fogarty. 2012. “Accounting for Women’s
gender difference and male status superiority.
Orgasm and Sexual Enjoyment in College Hook-
As students implicitly draw on these stereotypes
ups and Relationships.” American Sociological Re-
to make sense of their uncertain encounters, the view 77: 435–462.
argument suggests that they will inadvertently
 Ridgeway—The Persistence of Gender Inequality • 979

Baron, James N., Michael T. Hannan, Greta Hsu, Dunning, David, and David A. Sherman. 1997.
and Ozgecan Kocak. 2007. “In the Company “Stereotypes and Trait Inference.” Journal of Per-
of Women: Gender Inequality and the Logic of sonality and Social Psychology 73: 459–471.
Bureaucracy in Start-Up Firms.” Work and Occu- Eagly, Alice H., and Stephen J. Karau. 2002. “Role
pations 34: 35–66. Congruity Theory of Prejudice Towards Female
Bianchi, Suzanne M., John P. Robinson, and Me- Leaders.” Psychological Review 109: 573–579.
lissa Milkie. 2006. Changing Rhythms of American England, Paula, Paul Allison, Su Li, Noah Mark, Jen-
Family Life. New York: Russell Sage Foundation. nifer Thompson, Michelle Budig, and Han Sun.
Blair, Irene V., and Mahzarin R. Banaji. 1996. “Au- 2007. “Why Are Some Academic Fields Tipping
tomatic and Controlled Processes in Stereotype Toward Female? The Sex Composition of U.S.
Priming.” Journal of Personality and Social Psy- Fields of Doctoral Degree Receipt, 1971–2002.”
chology 70: 1142–1163. Sociology of Education 80: 23–42.
Brewer, Marilynn B., and Layton N. Lui. 1989. England, Paula, Emily Fitzgibbons Shafer, and Al-
“The Primacy of Age and Sex in the Structure of ison C. K. Fogarty. 2008. “Hooking-up and
Person Categories.” Social Cognition 7: 262–274. Forming Romantic Relationships on Today’s
Cejka, Mary Ann, and Alice H. Eagly. 1999. College Campuses.” In The Gendered Society
“Gender-Stereotypic Images of Occupations Reader, 3rd ed., edited by M. S. Kimmel and A.
Correspond to the Sex Segregation of Employ- Aronson, 531–547. New York: Oxford Univer-
ment.” Personality and Social Psychology Bulletin sity Press.
25: 413–423. Fiske, Susan T. 1998. “Stereotyping, Prejudice,
Charles, Maria, and David Grusky. 2004. Occu- and Discrimination.” In The Handbook of So-
pational Ghettos: The Worldwide Segregation of cial Psychology, 4th ed., edited by D.T. Gilbert,
Women and Men. Stanford, CA: Stanford Uni- S.T. Fiske, and G. Lindzey, 357–411. New York:
versity Press. McGraw-Hill.
Chwe, Michael S. 2001. Rational Ritual: Culture, Fiske, Susan T., Amy J. Cuddy, Peter Glick, and Jun
Coordination, and Common Knowledge. Prince- Xu. 2002. “A Model of (Often Mixed) Stereotype
ton, NJ: Princeton University Press. Content: Competence and Warmth Respectively
Clark, Anna E., and Yoshihisa Kashima. 2007. “Ste- Follow from Perceived Status and Competition.”
reotypes Help People Connect with Others in the Journal of Personality and Social Psychology 82:
Community: A Situated Functional Analysis of 878–902.
the Stereotype Consistency Bias in Communica- Glick, Peter, Maria Lameiras, Susan T. Fiske,
tion.” Journal of Personality and Social Psychology Thomas Eckes, Barbara Masser, Chiara Vol-
93: 1028–1039. pato, Anna Maria Manganelli, Jolynn C. X. Pek,
Conway, Michael M., Teresa Pizzamiglio, and Lau- Li-li Huang, Nuray Sakalli-Ugurlu, Yolanda
ren Mount. 1996. “Status, Communality, and Rodriguez Castro, Maria Luiza D’Avila Pereira,
Agency: Implications for Stereotypes of Gender Tineke M. Willemsen, Annetje Brunner, Iris
and Other Groups.” Journal of Personality and So- Six-Materna, and Robin Wells. 2004. “Bad But
cial Psychology 71: 25–38. Bold: Ambivalent Attitudes Toward Men Predict
Cotter, David A., Joan M. Hermsen, and Reeve Gender Inequality in 16 Nations.” Journal of Per-
Vanneman. 2004. “Gender Inequality at Work.” sonality and Social Psychology 86: 713–728.
In The American People Census 2000. New York: Goffman, Erving. 1967. Interaction Ritual: Essays on
Russell Sage Foundation and Population Refer- Face-to-Face Behavior. 1st ed. Garden City, NY:
ence Bureau. Anchor Books.
Cuddy, Amy J., Susan T. Fiske, and Peter Glick. Hamilton, Laura, and Elizabeth A. Armstrong.
2007. “The BIAS Map: Behaviors from Inter- 2009. “Gendered Sexuality in Young Adulthood:
group Affect and Stereotypes.” Journal of Person- Double Binds and Flawed Options.” Gender &
ality and Social Psychology 92: 631–648. Society 23: 589–616.
Diekman, Amanda B., and Alice H. Eagly. 2000. Ito, Tiffany A., and Geoffrey R. Urland. 2003. “Race
“Stereotypes as Dynamic Constructs: Women and Gender on the Brain: Electrocortical Mea-
and Men of the Past, Present, and Future.” sures of Attention to the Race and Gender of
Personality and Social Psychology Bulletin 26: Multiply Categorizable Individuals.” Journal of
1171–1188. Personality and Social Psychology 85 (4): 616–626.
980 • Part VIII: Gender Inequality

Kunda, Ziva, and Steven J. Spencer. 2003. “When Schneider, Donald J. 2004. The Psychology of Stereo-
Do Stereotypes Come to Mind and When Do typing. New York: Guilford Press.
They Color Judgment? A Goal-Based Theoretical Seachrist, Gretchen B., and Charles Stangor. 2001.
Framework for Stereotype Activation and Appli- “Perceived Consensus Influences Intergroup Be-
cation.” Psychological Bulletin 129 (4): 522–554. havior and Stereotype Accessibility.” Journal of
Macrae, C. Neil, and Susanne Quadflieg. 2010. Personality and Social Psychology 80: 645–654.
“Perceiving People.” In Handbook of Social Psy- Smith-Doerr, Laurel. 2004. Women’s Work: Gender
chology, 5th ed., Vol. I, edited by S. T. Fiske, D. Equality vs. Hierarchy in the Life Sciences. Boul-
T. Gilbert, and G. Lindzey, 428–463. New York: der, CO: Lynne Rienner Publishers.
Wiley. Spence, Janet T., and Camille E. Buckner. 2000.
McIlwee, Judith S., and J. Gregg Robinson. 1992. “Instrumental and Expressive Traits, Trait Stereo-
Women in Engineering: Gender, Power, and Work- types, and Sexist Attitudes: What Do They Sig-
place Culture. Albany: State University of New nify?” Psychology of Women Quarterly 24: 44–62.
York Press. Stangor, Charles, Laure Lynch, Changming Duan,
Phillips, Damon J. 2005. “Organizational Geneal- and Beth Glass. 1992. “Categorization of Indi-
ogies and the Persistence of Gender Inequality: viduals on the Basis of Multiple Social Features.”
The Case of Silicon Valley Law Firms.” Adminis- Journal of Personality and Social Psychology 62:
trative Science Quarterly 50: 440–472. 207–218.
Prentice, Deborah A., and Erica Carranza. 2002. von Hippel, Willaim, Denise Sekaquaptewa, and
“What Women and Men Should Be, Shouldn’t Patrick T. Vargas. 1995. “On the Role of En-
Be, Are Allowed to Be, and Don’t Have to coding Processes in Stereotype Maintenance.”
Be: The Contents of Prescriptive Gender Ste- Advances in Experimental Social Psychology 27:
reotypes.” Psychology of Women Quarterly 26: 177–254.
269–281. Wagner, David G., and Joseph Berger. 1997. “Gen-
Ridgeway, Cecilia L. 2001. “Gender, Status, and der and Interpersonal Task Behaviors: Status Ex-
Leadership.” Journal of Social Issues 57: 637–655. pectation Accounts.” Sociological Perspectives 40:
———. 2011. Framed by Gender: How Gender In- 1–32.
equality Persists in the Modern World. New York: West, Candace, and Don H. Zimmerman. 1987.
Oxford University Press. “Doing Gender.” Gender and Society 1: 125–151.
Ridgeway, Cecilia L., and Shelley J. Correll. 2004. Whittington, Kjersten Bunker. 2007. “Employment
“Unpacking the Gender System: A Theoretical Sectors as Opportunity Structures: Male and
Perspective on Cultural Beliefs and Social Rela- Female: The Effects of Location on Male and
tions.” Gender and Society 18: 510–531. Female Scientific Dissemination.” PhD diss.,
Ridgeway, Cecilia L., and Lynn Smith-Lovin. 1999. Department of Sociology, Stanford University,
“The Gender System and Interaction.” Annual Stanford, CA.
Review of Sociology 25: 191–216. Whittington, Kjersten Bunker, and Laurel
Rudman, Laurie A., and Kimberly Fairchild. 2004. Smith-Doerr. 2008. “Women Inventors in Con-
“Reactions to Counterstereotypic Behavior: The text: Disparities in Patenting Across Academia
Role of Backlash in Cultural Stereotype Mainte- and Industry.” Gender and Society 22: 194–218.
nance.” Journal of Personality and Social Psychol-
ogy 87: 157–176.
 PART IX

The Consequences
of Inequality

Lifestyles
114 Distinction: A Social Critique of the Judgment of Taste 982
115 The Social Stratification of Theater, Dance,
and Cinema Attendance 1004
116 Unequal Childhoods: Class, Race, and Family Life 1013
117 Income Inequality and Income Segregation 1023

Politics and Attitudes

118 What’s the Matter with Kansas? 1035
119 The Realignment of U.S. Presidential Voting 1037

Health
120 Health Inequalities and the Psychosocial Environment:
Two Scientific Challenges 1046
121 The Fundamentals of Fundamental Causality 1050
122 Why Do Mortality Disparities Persist? 1056
982 • Part IX: The Consequences of Inequality

Lifestyles

114 • Pierre Bourdieu

Distinction
A Social Critique of the Judgment of Taste

The Social Space percent of unskilled workers) from the working

The distribution of the different classes (and classes.1
class fractions) runs from those who are best pro- The differences stemming from the total vol-
vided with both economic and cultural capital ume of capital almost always conceal, both from
to those who are most deprived in both respects common awareness and also from “scientific”
(see Figures 114.1 and 114.2). The members knowledge, the secondary differences which,
of the professions, who have high incomes and within each of the classes defined by overall vol-
high qualifications, who very often (52.9 per- ume of capital, separate class fractions, defined
cent) originate from the dominant class (pro- by different asset structures, i.e., different distri-
fessions or senior executives), who receive and butions of their total capital among the different
consume a large quantity of both material and kinds of capital [economic and cultural].
cultural goods, are opposed in almost all re- Once one takes account of the structure of
spects to the office workers, who have low qual- total assets—and not only, as has always been
ifications, often originate from the working or done implicitly, of the dominant kind in a given
middle classes, who receive little and consume structure, “birth,” “fortune” or “talents,” as the
little, devoting a high proportion of their time nineteenth century put it—one has the means
to care maintenance and home improvement; of making more precise divisions and also of
and they are even more opposed to the skilled or observing the specific effects of the structure of
semi-skilled workers, and still more to unskilled distribution between the different kinds of cap-
workers or farm labourers, who have the lowest ital. This may, for example, be symmetrical (as
incomes, no qualifications, and originate almost in the case of the professions, which combine
exclusively (90.5 percent of farm labourers, 84.5 very high income with very high cultural capital)

Pierre Bourdieu. Distinction: A Social Critique of the Judgment of Taste, translated by Richard Nice (Cambridge,
Mass.: Harvard University Press, 1984), pp. 114–115, 128–129, 169–180, 183–186, 188–189, 193–197,
199–202, 206–209, 520–521, 572–575. Copyright © 1984 by the President and Fellows of Harvard College
and Routledge and Kegan Paul, Ltd. Reprinted by permission of the publisher. This chapter is based on survey
data collected by Pierre Bourdieu and on additional complementary data sources (denoted in the text by C.S.
I–C.S. V). For the purpose of conserving space, we have omitted all the special passages, set off in distinct
type in the original, that were devoted to “illustrative examples or discussion of ancillary issues” (Bourdieu,
Distinction, p. xiii).
Figure 114.1. The Space of Social Positions (Show in Black)
983
Figure 114.2. The Space of Life-Styles (Shown in Grey)
984
 Bourdieu—Distinction • 985

or asymmetrical (in the case of higher-education for greater objectivity, sociologists almost always
and secondary teachers or employers, with cul- forget that the “objects” they classify produce
tural capital dominant in one case, economic not only objectively classifiable practices but
capital in the other). One thus discovers two sets also classifying operations that are no less objec-
of homologous positions. The fractions whose tive and are themselves classifiable. The division
reproduction depends on economic capital, into classes performed by sociology leads to the
usually inherited—industrial and commercial common root of the classifiable practices which
employers at the higher level, craftsmen and agents produce and of the classificatory judge-
shopkeepers at the intermediate level—are op- ments they make of other agents’ practices and
posed to the fractions which are least endowed their own. The habitus is both the generative
(relatively, of course) with economic capital, and principle of objectively classifiable judgements
whose reproduction mainly depends on cultural and the system of classification (principium divi-
capital—higher-education and secondary teach- sionis) of these practices. It is in the relationship
ers at the higher level, primary teachers at the between the two capacities which define the hab-
intermediate level. . . . itus, the capacity to produce classifiable practices
and works, and the capacity to differentiate and
appreciate these practices and products (taste),
The Habitus that the represented social world, i.e., the space
The mere fact that the social space described of lifestyles, is constituted.
here can be presented as a diagram indicates that The relationship that is actually established
it is an abstract representation, deliberately con- between the pertinent characteristics of eco-
structed, like a map, to give a bird’s-eye view, a nomic and social condition (capital volume
point of view on the whole set of points from and composition, in both synchronic and dia-
which ordinary agents (including the sociologist chronic aspects) and the distinctive features as-
and his reader, in their ordinary behaviour) see sociated with the corresponding position in the
the social world. Bringing together in simulta- universe of life-styles only becomes intelligible
neity, in the scope of a single glance—this is its when the habitus is constructed as the genera-
heuristic value—positions which the agents can tive formula which makes it possible to account
never apprehend in their totality and in their both for the classifiable practices and products
multiple relationships, social space is to the and for the judgements, themselves classified,
practical space of everyday life, with its distances which make these practices and works into a
which are kept or signalled, and neighbours who system of distinctive signs. When one speaks
may be more remote than strangers, what geo- of the aristocratic asceticism of teachers or the
metrical space is to the “travelling space” (espace pretension of the petite bourgeoisie, one is not
hodologique) of ordinary experience, with its gaps only describing these groups by one, or even
and discontinuities. the most important, of their properties, but
But the most crucial thing to note is that also endeavouring to name the principle which
the question of this space is raised within the generates all their properties and all their judge-
space itself—that the agents have points of ments of their, or other people’s, properties. The
view on this objective space which depend on habitus is necessity internalized and converted
their position within it and in which their will into a disposition that generates meaningful
to transform or conserve it is often expressed. practices and meaning-giving perceptions; it is
Thus many of the words which sociology uses to a general, transposable disposition which carries
designate the classes it constructs are borrowed out a systematic, universal application—beyond
from ordinary usage, where they serve to express the limits of what has been directly learnt—of
the (generally polemical) view that one group the necessity inherent in the learning conditions.
has of another. As if carried away by their quest That is why an agent’s whole set of practices (or
986 • Part IX: The Consequences of Inequality

those of a whole set of agents produced by simi- apprehends differences between conditions,
lar conditions) are both systematic, inasmuch as which it grasps in the form of differences be-
they are the product of the application of iden- tween classified, classifying practices (products
tical (or interchangeable) schemes, and system- of other habitus), in accordance with princi-
atically distinct from the practices constituting ples of differentiation which, being themselves
another life-style. the product of these differences, are objectively
Because different conditions of existence attuned to them and therefore tend to perceive
produce different habitus—systems of genera- them as natural.
tive schemes applicable, by simple transfer, to While it must be reasserted, against all forms
the most varied areas of practice—the practices of mechanism, that ordinary experience of the
engendered by the different habitus appear as social world is a cognition, it is equally impor-
systematic configurations of properties express- tant to realize—contrary to the illusion of the
ing the differences objectively inscribed in con- spontaneous generation of consciousness which
ditions of existence in the form of systems of so many theories of the “awakening of class con-
differential deviations which, when perceived by sciousness” (prise de conscience) amount to—that
agents endowed with the schemes of perception primary cognition is misrecognition, recogni-
and appreciation necessary in order to identify, tion of an order which is also established in the
interpret and evaluate their pertinent features, mind. Life-styles are thus the systematic prod-
function as life-styles (see Figure 114.3).2 ucts of habitus, which, perceived in their mutual
The habitus is not only a structuring struc- relations through the schemes of the habitus,
ture, which organizes practices and the percep- become sign systems that are socially qualified
tion of practices, but also a structured structure: (as “distinguished,” “vulgar” etc.). The dialec-
the principle of division into logical classes tic of conditions and habitus is the basis of an
which organizes the perception of the social alchemy which transforms the distribution of
world is itself the product of internalization of capital, the balance-sheet of a power relation,
the division into social classes. Each class con- into a system of perceived differences, distinc-
dition is defined, simultaneously, by its intrinsic tive properties, that is, a distribution of symbolic
properties and by the relational properties which capital, legitimate capital, whose objective truth
it derives from its position in the system of class is misrecognized.
conditions, which is also a system of differences, As structured products (opus operatum)
differential positions, i.e., by everything which which a structuring structure (modus operandi)
distinguishes it from what it is not and especially produces through retranslations according to
from everything it is opposed to; social identity the specific logic of the different fields, all the
is defined and asserted through difference. This practices and products of a given agent are ob-
means that inevitably inscribed within the dis- jectively harmonized among themselves, with-
positions of the habitus is the whole structure out any deliberate pursuit of coherence, and
of the system of conditions, as it presents itself objectively orchestrated, without any conscious
in the experience of a life-condition occupying concertation, with those of all members of the
a particular position within that structure. The same class. The habitus continuously generates
most fundamental oppositions in the structure practical metaphors, that is to say, transfers (of
(high/low, rich/poor etc.) tend to establish them- which the transfer of motor habits is only one
selves as the fundamental structuring principles example) or, more precisely, systematic transpo-
of practices and the perception of practices. As sitions required by the particular conditions in
a system of practice-generating schemes which which the habitus is “put into practice” (so that,
expresses systematically the necessity and free- for example, the ascetic ethos which might be
dom inherent in its class condition and the dif- expected always to express itself in saving may,
ference constituting that position, the habitus in a given context, express itself in a particular
 Bourdieu—Distinction • 987

Figure 114.3. Conditions of Existence, Habitus, and Life-Style

way of using credit). The practices of the same Systematicity is found in the opus operatum
agent, and, more generally, the practices of all because it is in the modus operandi.3 It is found
agents of the same class, owe the stylistic affinity in all the properties—and property—with which
which makes each of them a metaphor of any of individuals and groups surround themselves,
the others to the fact that they are the product of houses, furniture, paintings, books, cars, spirits,
transfers of the same schemes of action from one cigarettes, perfume, clothes, and in the practices
field to another. An obvious paradigm would be in which they manifest their distinction, sports,
the disposition called “handwriting,” a singular games, entertainments, only because it is in the
way of tracing letters which always produces the synthetic unity of the habitus, the unifying, gen-
same writing, i.e., graphic forms which, in spite erative principle of all practices. Taste, the pro-
of all the differences of size, material or colour pensity and capacity to appropriate (materially or
due to the surface (paper or blackboard) or the symbolically) a given class of classified, classify-
instrument (pen or chalk)—in spite, therefore, ing objects or practices, is the generative formula
of the different use of muscles—present an im- of life-style, a unitary set of distinctive prefer-
mediately perceptible family resemblance, like ences which express the same expressive inten-
all the features of style or manner whereby a tion in the specific logic of each of the symbolic
painter or writer can be recognized as infallibly subspaces, furniture, clothing, language or body
as a man by his walk. hexis. Each dimension of life-style “symbolizes
988 • Part IX: The Consequences of Inequality

with” the others, in Leibniz’s phrase, and sym- smile, the measured gesture, the well-tailored
bolizes them. An old cabinetmaker’s world view, suit and the bourgeois salon of the person who
the way he manages his budget, his time or his pronounces them.
body, his use of language and choice of clothing Taste is the practical operator of the trans-
are fully present in his ethic of scrupulous, im- mutation of things into distinct and distinctive
peccable craftsmanship and in the aesthetic of signs, of continuous distributions into discon-
work for work’s sake which leads him to measure tinuous oppositions; it raises the differences
the beauty of his products by the care and pa- inscribed in the physical order of bodies to
tience that have gone into them. the symbolic order of significant distinctions.
The system of matching properties, which It transforms objectively classified practices, in
includes people—one speaks of a “well-matched which a class condition signifies itself (through
couple,” and friends like to say they have the taste), into classifying practices, that is, into a
same tastes—is organized by taste, a system of symbolic expression of class position, by perceiv-
classificatory schemes which may only very par- ing them in their mutual relations and in terms
tially become conscious although, as one rises of social classificatory schemes. Taste is thus
in the social hierarchy, life-style is increasingly the source of the system of distinctive features
a matter of what Weber calls the “stylization of which cannot fail to be perceived as a systematic
life.” Taste is the basis of the mutual adjustment expression of a particular class of conditions of
of all the features associated with a person, which existence, i.e., as a distinctive life-style, by any-
the old aesthetic recommended for the sake of one who possesses practical knowledge of the
the mutual reinforcement they give one another; relationships between distinctive signs and posi-
the countless pieces of information a person tions in the distributions—between the universe
consciously or unconsciously imparts endlessly of objective properties, which is brought to light
underline and confirm one another, offering the by scientific construction, and the no less objec-
alert observer the same pleasure an art-lover de- tive universe of life-styles, which exists as such
rives from the symmetries and correspondences for and through ordinary experience.
produced by a harmonious distribution of re- This classificatory system, which is the prod-
dundancies. The over-determination that results uct of the internalization of the structure of
from these redundancies is felt the more strongly social space, in the form in which it impinges
because the different features which have to be through the experience of a particular position
isolated for observation or measurement strongly in that space, is, within the limits of economic
interpenetrate in ordinary perception; each possibilities and impossibilities (which it tends
item of information imparted in practice (e.g., to reproduce in its own logic), the generator of
a judgement of a painting) is contaminated— practices adjusted to the regularities inherent in a
and, if it deviates from the probable feature, cor- condition. It continuously transforms necessities
rected—by the effect of the whole set of features into strategies, constraints into preferences, and,
previously or simultaneously perceived. That is without any mechanical determination, it gen-
why a survey which tends to isolate features—for erates the set of “choices” constituting life-styles,
example, by dissociating the things said from the which derive their meaning, i.e., their value,
way they are said—and detach them from the from their position in a system of oppositions
system of correlative features tends to minimize and correlations.4 It is a virtue made of necessity
the deviation, on each point, between the classes, which continuously transforms necessity into
especially that between the petit bourgeois and virtue by inducing “choices” which correspond
the bourgeois. In the ordinary situations of bour- to the condition of which it is the product. As
geois life, banalities about art, literature or cin- can be seen whenever a change in social position
ema are inseparable from the steady tone, the puts the habitus into new conditions, so that its
slow, casual diction, the distant or self-assured specific efficacy can be isolated, it is taste—the
 Bourdieu—Distinction • 989

taste of necessity or the taste of luxury—and the combination of ease and asceticism, i.e.,
not high or low income which commands the self-imposed austerity, restraint, reserve, which
practices objectively adjusted to these resources. are affirmed in that absolute manifestation of
Through taste, an agent has what he likes be- excellence, relaxation in tension.
cause he likes what he has, that is, the properties This fundamental opposition is specified
actually given to him in the distributions and le- according to capital composition. Through the
gitimately assigned to him in the classifications.5 mediation of the means of appropriation avail-
able to them, exclusively or principally cultural
on the one hand, mainly economic on the other,
The Homology Between the Spaces and the different forms of relation to works of art
Bearing in mind all that precedes, in particular which result from them, the different fractions of
the fact that the generative schemes of the hab- the dominant class are oriented towards cultural
itus are applied, by simple transfer, to the most practices so different in their style and object and
dissimilar areas of practice, one can immediately sometimes so antagonistic (those of “artists” and
understand that the practices or goods associated “bourgeois”)6 that it is easy to forget that they are
with the different classes in the different areas of variants of the same fundamental relationship to
practice are organized in accordance with struc- necessity and to those who remain subject to it,
tures of opposition which are homologous to one and that each pursues the exclusive appropria-
another because they are all homologous to the tion of legitimate cultural goods and the asso-
structure of objective oppositions between class ciated symbolic profits. Whereas the dominant
conditions. Without presuming to demonstrate fractions of the dominant class (the “bourgeoi-
here in a few pages what the whole of the rest of sie”) demand of art a high degree of denial of
this work will endeavour to establish—but lest the social world and incline towards a hedo-
the reader fail to see the wood for the trees of nistic aesthetic of ease and facility, symbolized
detailed analysis—I shall merely indicate, very by boulevard theatre or Impressionist painting,
schematically, how the two major organizing the dominated fractions (the “intellectuals” and
principles of the social space govern the struc- “artists”) have affinities with the ascetic aspect of
ture and modification of the space of cultural aesthetics and are inclined to support all artistic
consumption, and, more generally, the whole revolutions conducted in the name of purity and
universe of life-styles. purification, refusal of ostentation and the bour-
In cultural consumption, the main oppo- geois taste for ornament; and the dispositions
sition, by overall capital value, is between the towards the social world which they owe to their
practices designated by their rarity as distin- status as poor relations incline them to welcome
guished, those of the fractions richest in both a pessimistic representation of the social world.
economic and cultural capital, and the practices While it is clear that art offers it the greatest
socially identified as vulgar because they are both scope, there is no area of practice in which the
easy and common, those of the fractions poor- intention of purifying, refining and sublimating
est in both these respects. In the intermediate facile impulses and primary needs cannot assert
position are the practices which are perceived as itself, or in which the stylization of life, i.e., the
pretentious, because of the manifest discrepancy primacy of form over function, which leads to
between ambition and possibilities. In opposi- the denial of function, does not produce the
tion to the dominated condition, characterized, same effects. In language, it gives the opposition
from the point of view of the dominant, by the between popular outspokenness and the highly
combination of forced poverty and unjustified censored language of the bourgeois, between
laxity, the dominant aesthetic—of which the the expressionist pursuit of the picturesque or
work of art and the aesthetic disposition are the rhetorical effect and the choice of restraint
the most complete embodiments—proposes and false simplicity (litotes). The same economy
990 • Part IX: The Consequences of Inequality

of means is found in body language: here too, account of all the characteristics of social con-
agitation and haste, grimaces and gesticulation dition which are (statistically) associated from
are opposed to slowness—“the slow gestures, earliest childhood with possession of high or low
the slow glance” of nobility, according to Nietz- income and which tend to shape tastes adjusted
sche7—to the restraint and impassivity which to these conditions.9 The true basis of the differ-
signify elevation. Even the field of primary tastes ences found in the area of consumption, and far
is organized according to the fundamental oppo- beyond it, is the opposition between the tastes of
sition, with the antithesis between quantity and luxury (or freedom) and the tastes of necessity.
quality, belly and palate, matter and manners, The former are the tastes of individuals who are
substance and form. the product of material conditions of existence
defined by distance from necessity, by the free-
doms or facilities stemming from possession of
Form and Substance capital; the latter express, precisely in their ad-
The fact that in the realm of food the main op- justment, the necessities of which they are the
position broadly corresponds to differences in product. Thus it is possible to deduce popular
income has masked the secondary opposition tastes for the foods that are simultaneously most
which exists, both within the middle classes and “filling” and most economical10 from the neces-
within the dominant class, between the fractions sity of reproducing labour power at the lowest
richer in cultural capital and less rich in eco- cost which is forced on the proletariat as its very
nomic capital and those whose assets are struc- definition. The idea of taste, typically bourgeois,
tured in the opposite way. Observers tend to see since it presupposes absolute freedom of choice,
a simple effect of income in the fact that, as one is so closely associated with the idea of freedom
rises in the social hierarchy, the proportion of in- that many people find it hard to grasp the para-
come spent on food diminishes, or that, within doxes of the taste of necessity. Some simply sweep
the food budget, the proportion spent on heavy, it aside, making practice a direct product of eco-
fatty, fattening foods, which are also cheap— nomic necessity (workers eat beans because they
pasta, potatoes, beans, bacon, pork—declines cannot afford anything else), failing to realize
(C.S. III), as does that spent on wine, whereas that necessity can only be fulfilled, most of the
an increasing proportion is spent on leaner, time, because the agents are inclined to fulfill it,
lighter (more digestible), non-fattening foods because they have a taste for what they are any-
(beef, veal, mutton, lamb, and especially fresh way condemned to. Others turn it into a taste of
fruit and vegetables).8 Because the real principle freedom, forgetting the conditionings of which
of preferences is taste, a virtue made of necessity, it is the product, and so reduce it to patholog-
the theory which makes consumption a simple ical or morbid preference for (basic) essentials,
function of income has all the appearances to a sort of congenital coarseness, the pretext for a
support it, since income plays an important part class racism which associates the populace with
in determining distance from necessity. How- everything heavy, thick and fat.11 Taste is amor
ever, it cannot account for cases in which the fati, the choice of destiny, but a forced choice,
same income is associated with totally different produced by conditions of existence which rule
consumption patterns. Thus, foremen remain out all alternatives as mere daydreams and leave
attached to “popular” taste although they earn no choice but the taste for the necessary.
more than clerical and commercial employees, The taste of necessity can only be the basis of a
whose taste differs radically from that of manual life-style “in-itself,” which is defined as such only
workers and is closer to that of teachers. negatively, by an absence, by the relationship of
For a real explanation of the variations which privation between itself and the other life-styles.
J. F. Engel’s law merely records, one has to take For some, there are elective emblems, for others
 Bourdieu—Distinction • 991

stigmata which they bear in their very bodies. eating and drinking together, in a conviviality
“As the chosen people bore in their features the which sweeps away restraints and reticence.
sign that they were the property of Jehovah, so The boundary marking the break with the
the division of labour brands the manufacturing popular relation to food runs, without any
worker as the property of capital.”12 The brand doubt, between the manual workers and the
which Marx speaks of is nothing other than clerical and commercial employees (C.S. II).
life-style, through which the most deprived im- Clerical workers spend less on food than skilled
mediately betray themselves, even in their use of manual workers, both in absolute terms (9,376
spare time; in so doing they inevitably serve as a francs as against 10,347 francs) and in relative
foil to every distinction and contribute, purely terms (34.2 percent as against 38.3 percent);
negatively, to the dialectic of pretension and dis- they consume less bread, pork, pork products
tinction which fuels the incessant changing of (charcuterie), milk, cheese, rabbit, poultry, dried
taste. Not content with lacking virtually all the vegetables and fats, and, within a smaller food
knowledge or manners which are valued in the budget, spend as much on meat—beef, veal,
markets of academic examination or polite con- mutton and lamb—and slightly more on fish,
versation nor with only possessing skills which fresh fruit and aperitifs. These changes in the
have no value there, they are the people “who structure of spending on food are accompanied
don’t know how to live,” who sacrifice most to by increased spending on health and beauty care
material foods, and to the heaviest, grossest and and clothing, and a slight increase in spending on
most fattening of them, bread, potatoes, fats, cultural and leisure activities. When it is noted
and the most vulgar, such as wine; who spend that the reduced spending on food, especially on
least on clothing and cosmetics, appearance and the most earthly, earthy, down-to-earth foods, is
beauty; those who “don’t know how to relax,” accompanied by a lower birth-rate, it is reason-
“who always have to be doing something,” who able to suppose that it constitutes one aspect of
set off in their Renault 5 or Simca 1000 to join an overall transformation of the relationship to
the great traffic jams of the holiday exodus, who the world. The “modest” taste which can defer
picnic beside major roads, cram their tents into its gratifications is opposed to the spontaneous
overcrowded campsites, fling themselves into the materialism of the working classes, who refuse
prefabricated leisure activities designed for them to participate in the Benthamite calculation of
by the engineers of cultural mass production; pleasures and pains, benefits and costs (e.g., for
those who by all these uninspired “choices” con- health and beauty). In other words, these two
firm class racism, if it needed to be confirmed, relations to the “fruits of the earth” are grounded
in its conviction that they only get what they in two dispositions towards the future which are
deserve. themselves related in circular causality to two
The art of eating and drinking remains one of objective futures. Against the imaginary anthro-
the few areas in which the working classes explic- pology of economics, which has never shrunk
itly challenge the legitimate art of living. In the from formulating universal laws of “temporal
face of the new ethic of sobriety for the sake of preference,” it has to be pointed out that the pro-
slimness, which is most recognized at the high- pensity to subordinate present desires to future
est levels of the social hierarchy, peasants and desires depends on the extent to which this sac-
especially industrial workers maintain an ethic rifice is “reasonable,” that is, on the likelihood,
of convivial indulgence. A bon vivant is not just in any case, of obtaining future satisfactions su-
someone who enjoys eating and drinking; he is perior to those sacrificed.13
someone capable of entering into the generous Among the economic conditions of the
and familiar—that is, both simple and free—re- propensity to sacrifice immediate satisfactions
lationship that is encouraged and symbolized by to expected satisfactions one must include the
992 • Part IX: The Consequences of Inequality

Table 114.1. Yearly Spending by Teachers, Professionals, and Industrial and Commercial
Employers, 1972

Source: C.S. II (1972).

a
Includes restaurant or canteen meals.
b
Clothes, shoes, repairs and cleaning, toiletries, hairdressing, domestic servants.
c
Books, newspapers and magazines, stationery, records, sport, toys, music, entertainments.

probability of these future satisfactions which is many oppositions as there are different ways of
inscribed in the present condition. There is still asserting one’s distinction vis-à-vis the working
a sort of economic calculation in the unwilling- class and its primary needs, or—which amounts
ness to subject existence to economic calcula- to the same thing—different powers whereby
tion. The hedonism which seizes day by day the necessity can be kept at a distance. Thus, within
rare satisfactions (“good times”) of the immedi- the dominant class, one can, for the sake of sim-
ate present is the only philosophy conceivable plicity, distinguish three structures of the con-
to those who “have no future” and, in any case, sumption distributed under three items: food,
little to expect from the future.14 It becomes culture and presentation (clothing, beauty care,
clearer why the practical materialism which is toiletries, domestic servants). These structures
particularly manifested in the relation to food take strictly opposite forms—like the structures
is one of the most fundamental components of of their capital—among the teachers as against
the popular ethos and even the popular ethic. the industrial and commercial employers (see
The being-in-the-present which is affirmed in Table 114.1). Whereas the latter have exception-
the readiness to take advantage of the good times ally high expenditure on food (37 percent of the
and take time as it comes is, in itself, an affirma- budget), low cultural costs and medium spend-
tion of solidarity with others (who are often the ing on presentation and representation, the for-
only present guarantee against the threats of the mer, whose total spending is lower on average,
future), inasmuch as this temporal immanentism have low expenditure on food (relatively less
is a recognition of the limits which define the than manual workers), limited expenditure on
condition. This is why the sobriety of the petit presentation (though their expenditure on health
bourgeois is felt as a break: in abstaining from is one of the highest) and relatively high expen-
having a good time and from having it with diture on culture (books, papers, entertainments,
others, the would-be petit bourgeois betrays his sport, toys, music, radio and record-player). Op-
ambition of escaping from the common present, posed to both these groups are the members of
when, that is, he does not construct his whole the professions, who devote the same proportion
self-image around the opposition between his of their budget to food as the teachers (24.4 per-
home and the café, abstinence and intemper- cent), but out of much greater total expenditure
ance, in other words, between individual salva- (57,122 francs as against 40,884 francs), and
tion and collective solidarities. who spend much more on presentation and
representation than all other fractions, especially
if the costs of domestic service are included,
Three Styles of Distinction whereas their cultural expenditure is lower than
The basic opposition between the tastes of lux- that of the teachers (or even the engineers and
ury and the tastes of necessity is specified in as senior executives, who are situated between the
 Bourdieu—Distinction • 993

teachers and the professionals, though nearer the cuisine, rich in expensive or rare products (fresh
latter, for almost all items). vegetables, meat). Finally, the teachers, richer in
The system of differences becomes clearer cultural capital than in economic capital, and
when one looks more closely at the patterns of therefore inclined to ascetic consumption in all
spending on food. In this respect the industrial areas, pursue originality at the lowest economic
and commercial employers differ markedly from cost and go in for exoticism (Italian, Chinese
the professionals, and a fortiori from the teach- cooking etc.)17 and culinary populism (peasant
ers, by virtue of the importance they give to dishes). They are thus almost consciously op-
cereal-based products (especially cakes and pas- posed to the (new) rich with their rich food, the
tries), wine, meat preserves (foie gras, etc.) and buyers and sellers of grosse bouffe, the “fat cats,”18
game, and their relatively low spending on meat, gross in body and mind, who have the economic
fresh fruit and vegetables. The teachers, whose means to flaunt, with an arrogance perceived as
food purchases are almost identically structured “vulgar,” a life-style which remains very close to
to those of office workers, spend more than all that of the working classes as regards economic
other fractions on bread, milk products, sugar, and cultural consumption.
fruit preserves and non-alcoholic drinks, less on
wine and spirits and distinctly less than the pro-
fessions on expensive products such as meat— Unpretentious or Uncouth?
especially the most expensive meats, such as It is clear that tastes in food cannot be consid-
mutton and lamb—and fresh fruit and vegeta- ered in complete independence of the other
bles. The members of the professions are mainly dimensions of the relationship to the world, to
distinguished by the high proportion of their others and to one’s own body, through which
spending which goes on expensive products, the practical philosophy of each class is enacted.
particularly meat (18.3 percent of their food To demonstrate this, one would have to make a
budget), and especially the most expensive meat systematic comparison of the working-class and
(veal, lamb, mutton), fresh fruit and vegetables, bourgeois ways of treating food, of serving, pre-
fish and shellfish, cheese and aperitifs.15 senting and offering it, which are infinitely more
Thus, when one moves from the manual revelatory than even the nature of the products
workers to the industrial and commercial em- involved (especially since most surveys of con-
ployers, through foremen, craftsmen and small sumption ignore differences in quality). The
shopkeepers, economic constraints tend to relax analysis is a difficult one, because each life-style
without any fundamental change in the pattern can only really be constructed in relation to the
of spending (see Table 114.2). The opposition other, which is its objective and subjective nega-
between the two extremes is here established tion, so that the meaning of behaviour is totally
between the poor and the rich (nouveau riche), reversed depending on which point of view is
between la bouffe and la grande bouffe;16 the food adopted and on whether the common words
consumed is increasingly rich (both in cost and which have to be used to name the conduct (e.g.,
in calories) and increasingly heavy (game, foie “manners”) are invested with popular or bour-
gras). By contrast, the taste of the professionals geois connotations.
or senior executives defines the popular taste, Plain speaking, plain eating: the working-class
by negation, as the taste for the heavy, the fat meal is characterized by plenty (which does not
and the coarse, by tending towards the light, exclude restrictions and limits) and above all by
the refined and the delicate (see Figure 114.4). freedom. “Elastic” and “abundant” dishes are
The disappearance of economic constraints is brought to the table—soups or sauces, pasta
accompanied by a strengthening of the social or potatoes (almost always included among the
censorships which forbid coarseness and fatness, vegetables)—and served with a ladle or spoon,
in favour of slimness and distinction. The taste to avoid too much measuring and counting, in
for rare, aristocratic foods points to a traditional contrast to everything that has to be cut and
 994
Table 114.2. Annual Household Expenditures on Food: Fractions of the Dominant Class, 1972
Source: C.S. II (1972).
995
996 • Part IX: The Consequences of Inequality

Figure 114.4. The Food Square

divided, such as roasts.19 This impression of and drinking. These strongly marked differences
abundance, which is the norm on special occa- of social status (associated with sex and age) are
sions, and always applies, so far as is possible, for accompanied by no practical differentiation
the men, whose plates are filled twice (a privilege (such as the bourgeois division between the din-
which marks a boy’s accession to manhood), is ing room and the kitchen, where the servants eat
often balanced, on ordinary occasions, by restric- and sometimes the children), and strict sequenc-
tions which generally apply to the women, who ing of the meal tends to be ignored. Everything
will share one portion between two, or eat the may be put on the table at much the same time
leftovers of the previous day; a girl’s accession (which also saves walking), so that the women
to womanhood is marked by doing without. It may have reached the dessert, and also the chil-
is part of men’s status to eat and to eat well (and dren, who will take their plates and watch televi-
also to drink well); it is particularly insisted that sion, while the men are still eating the main dish
they should eat, on the grounds that “it won’t and the “lad,” who has arrived late, is swallowing
keep,” and there is something suspect about a his soup.
refusal. On Sundays, while the women are on This freedom, which may be perceived as dis-
their feet, busily serving, clearing the table, order or slovenliness, is adapted to its function.
washing up, the men remain seated, still eating Firstly, it is labour-saving, which is seen as an
 Bourdieu—Distinction • 997

advantage. Because men take no part in house- not appearing over-eager. A strict sequence is ob-
work, not least because the women would not served and all coexistence of dishes which the
allow it—it would be a dishonour to see men sequence separates, fish and meat, cheese and
step outside their rôle—every economy of effort dessert, is excluded: for example, before the des-
is welcome. Thus, when the coffee is served, a sert is served, everything left on the table, even
single spoon may be passed around to stir it. But the salt-cellar, is removed, and the crumbs are
these short cuts are only permissible because one swept up. This extension of rigorous rules into
is and feels at home, among the family, where everyday life (the bourgeois male shaves and
ceremony would be an affectation. For example, dresses first thing every morning, and not just to
to save washing up, the dessert may be handed “go out”), refusing the division between home
out on improvised plates torn from the cake-box and the exterior, the quotidian and the extraquo-
(with a joke about “taking the liberty,” to mark tidian, is not explained solely by the presence of
the transgression), and the neighbour invited in strangers—servants and guests—in the familiar
for a meal will also receive his piece of cardboard family world. It is the expression of a habitus
(offering a plate would exclude him) as a sign of of order, restraint and propriety which may not
familiarity. Similarly, the plates are not changed be abdicated. The relation to food—the primary
between dishes. The soup plate, wiped with need and pleasure—is only one dimension of
bread, can be used right through the meal. The the bourgeois relation to the social world. The
hostess will certainly offer to “change the plates,” opposition between the immediate and the de-
pushing back her chair with one hand and reach- ferred, the easy and the difficult, substance (or
ing with the other for the plate next to her, but function) and form, which is exposed in a partic-
everyone will protest (“It all gets mixed up inside ularly striking fashion in bourgeois ways of eat-
you”) and if she were to insist it would look as if ing, is the basis of all aestheticization of practice
she wanted to show off her crockery (which she and every aesthetic. Through all the forms and
is allowed to if it is a new present) or to treat her formalisms imposed on the immediate appetite,
guests as strangers, as is sometimes deliberately what is demanded—and inculcated—is not only
done to intruders or “scroungers” who never re- a disposition to discipline food consumption by
turn the invitation. These unwanted guests may a conventional structuring which is also a gen-
be frozen out by changing their plates despite tle, indirect, invisible censorship (quite different
their protests, not laughing at their jokes, or from enforced privations) and which is an ele-
scolding the children for their behaviour (“No, ment in an art of living (correct eating, for exam-
no, we don’t mind,” say the guests; “They ought ple, is a way of paying homage to one’s hosts and
to know better by now,” the parents respond). to the mistress of the house, a tribute to her care
The common root of all these “liberties” is no and effort). It is also a whole relationship to an-
doubt the sense that at least there will not be imal nature, to primary needs and the populace
self-imposed controls, constraints and restric- who indulge them without restraint; it is a way
tions—especially not in eating, a primary need of denying the meaning and primary function
and a compensation—and especially not in the of consumption, which are essentially common,
heart of domestic life, the one realm of freedom, by making the meal a social ceremony, an affir-
when everywhere else, and at all other times, ne- mation of ethical tone and aesthetic refinement.
cessity prevails. The manner of presenting and consuming the
In opposition to the free-and-easy food, the organization of the meal and setting
working-class meal, the bourgeoisie is concerned of the places, strictly differentiated according to
to eat with all due form. Form is first of all a the sequence of dishes and arranged to please the
matter of rhythm, which implies expectations, eye, the presentation of the dishes, considered as
pauses, restraints; waiting until the last person much in terms of shape and colour (like works
served has started to eat, taking modest helpings, of art) as of their consumable substance, the
998 • Part IX: The Consequences of Inequality

etiquette governing posture and gesture, ways of is felt and proved in actions; it is the free-speech
serving oneself and others, of using the different and language of the heart which make the true
utensils, the seating plan, strictly but discreetly “nice guy,” blunt, straightforward, unbending,
hierarchical, the censorship of all bodily manifes- honest, genuine, “straight down the line” and
tations of the act or pleasure of eating (such as “straight as a die,” as opposed to everything
noise or haste), the very refinement of the things that is pure form, done only for form’s sake; it
consumed, with quality more important than is freedom and the refusal of complications, as
quantity—this whole commitment to stylization opposed to respect for all the forms and formal-
tends to shift the emphasis from substance and ities spontaneously perceived as instruments of
function to form and manner, and so to deny the distinction and power. On these moralities, these
crudely material reality of the act of eating and world views, there is no neutral viewpoint; what
of the things consumed, or, which amounts to for some is shameless and slovenly, for others is
the same thing, the basely material vulgarity of straightforward, unpretentious; familiarity is for
those who indulge in the immediate satisfactions some the most absolute form of recognition, the
of food and drink.20 abdication of all distance, a trusting openness, a
Given the basic opposition between form relation of equal to equal; for others, who shun
and substance, one could re-generate each of the familiarity, it is an unseemly liberty.
oppositions between the two antagonistic ap- The popular realism which inclines working
proaches to the treatment of food and the act of people to reduce practices to the reality of their
eating. In one case, food is claimed as a material function, to do what they do, and be what they
reality, a nourishing substance which sustains the are (“That’s the way I am”), without “kidding
body and gives strength (hence the emphasis on themselves” (“That’s the way it is”), and the prac-
heavy, fatty, strong foods, of which the paradigm tical materialism which inclines them to censor
is pork—fatty and salty—the antithesis of fish— the expression of feelings or to divert emotion
light, lean and bland); in the other, the priority into violence or oaths, are the near-perfect an-
given to form (the shape of the body, for exam- tithesis of the aesthetic disavowal which, by a
ple) and social form, formality, puts the pursuit sort of essential hypocrisy (seen, for example, in
of strength and substance in the background the opposition between pornography and eroti-
and identifies true freedom with the elective as- cism) masks the interest in function by the pri-
ceticism of a self-imposed rule. And it could be macy given to form, so that what people do, they
shown that two antagonistic world views, two do as if they were not doing it.
worlds, two representations of human excellence
are contained in this matrix. Substance—or
matter—is what is substantial, not only “filling” The Visible and the Invisible
but also real, as opposed to all appearances, all But food—which the working classes place on
the fine words and empty gestures that “butter the side of being and substance, whereas the
no parsnips” and are, as the phrase goes, purely bourgeoisie, refusing the distinction between in-
symbolic; reality, as against sham, imitation, side and outside or “at home” and “for others,”
window-dressing; the little eating-house with its the quotidian and the extraquotidian, introduces
marble-topped tables and paper napkins where into it the categories of form and appearance—is
you get an honest square meal and aren’t “paying itself related to clothing as inside to outside, the
for the wallpaper” as in fancy restaurants; being, domestic to the public, being to seeming. And
as against seeming, nature and the natural, sim- the inversion of the places of food and clothing
plicity (pot-luck, “take it as it comes,” “no stand- in the contrast between the spending patterns of
ing on ceremony”), as against embarrassment, the working classes, who give priority to being,
mincing and posturing, airs and graces, which and the middle classes, where the concern for
are always suspected of being a substitute for “seeming” arises, is the sign of a reversal of the
substance, i.e., for sincerity, for feeling, for what whole world view. The working classes make a
 Bourdieu—Distinction • 999

realistic or, one might say, functionalist use of moves up the social hierarchy; the difference is
clothing. Looking for substance and function greatest for suits and costumes—expensive gar-
rather than form, they seek “value for money” ments—and smaller for dresses and especially
and choose what will “last.” Ignoring the bour- skirts and jackets. The topcoat, which is increas-
geois concern to introduce formality and formal ingly frequent among women at higher social
dress into the domestic world, the place for free- levels, is opposed to the “all-purpose” raincoat,
dom—an apron and slippers (for women), bare in the same way as overcoat and lumber jacket
chest or a vest (for men)—they scarcely mark are opposed for men. The use of the smock
the distinction between top clothes, visible, in- and the apron, which in the working classes is
tended to be seen, and underclothes, invisible virtually the housewife’s uniform, increases as
or hidden—unlike the middle classes, who have one moves down the hierarchy (in contrast to
a degree of anxiety about external appearances, the dressing-gown, which is virtually unknown
both sartorial and cosmetic, at least outside and among peasants and industrial workers).
at work (to which middle-class women more The interest the different classes have in
often have access). self-presentation, the attention they devote to it,
Thus, despite the limits of the data avail- their awareness of the profits it gives and the in-
able, one finds in men’s clothing (which is much vestment of time, effort, sacrifice and care which
more socially marked, at the level of what can be they actually put into it are proportionate to the
grasped by statistics on purchases, than women’s chances of material or symbolic profit they can
clothing) the equivalent of the major oppositions reasonably expect from it. More precisely, they
found in food consumption. In the first dimen- depend on the existence of a labour market in
sion of the space, the division again runs be- which physical appearance may be valorized in
tween the office workers and the manual workers the performance of the job itself or in profes-
and is marked particularly by the opposition sional relations; and on the differential chances
between grey or white overalls and blue dun- of access to this market and the sectors of this
garees or boiler-suits, between town shoes and market in which beauty and deportment most
the more relaxed moccasins, kickers or sneakers strongly contribute to occupational value. A
(not to mention dressing-gowns, which clerical first indication of this correspondence between
workers buy 3.5 times more often than manual the propensity to cosmetic investments and the
workers). The increased quantity and quality of chances of profit may be seen in the gap, for all
all purchases of men’s clothing is summed up in forms of beauty care, between those who work
the opposition between the suit, the prerogative and those who do not (which must also vary ac-
of the senior executive, and the blue overall, the cording to the nature of the job and the work
distinctive mark of the farmer and industrial environment). It can be understood in terms of
worker (it is virtually unknown in other groups, this logic why working-class women, who are
except craftsmen); or between the overcoat, al- less likely to have a job and much less likely to
ways much rarer among men than women, but enter one of the occupations which most strictly
much more frequent among senior executives demand conformity to the dominant norms of
than the other classes, and the fur-lined jacket beauty, are less aware than all others of the “mar-
or lumber jacket, mainly worn by agricultural ket” value of beauty and much less inclined to
and industrial workers. In between are the junior invest time and effort, sacrifices and money in
executives, who now scarcely ever wear working cultivating their bodies.
clothes but fairly often buy suits. It is quite different with the women of the
Among women, who, in all categories (except petite bourgeoisie, especially the new petite
farmers and farm labourers), spend more than bourgeoisie, in the occupations involving pre-
men (especially in the junior and senior execu- sentation and representation, which often im-
tive, professional and other high-income catego- pose a uniform (tenue) intended, among other
ries), the number of purchases increases as one things, to abolish all traces of heterodox taste,
1000 • Part IX: The Consequences of Inequality

and which always demand what is called tenue, moral value, they feel superior both in the in-
in the sense of “dignity of conduct and correct- trinsic, natural beauty of their bodies and in the
ness of manners,” implying, according to the art of self-embellishment and everything they
dictionary, “a refusal to give way to vulgarity call tenue, a moral and aesthetic virtue which
or facility.” (In the specialized “charm schools” defines “nature” negatively as sloppiness. Beauty
which train hostesses, the working-class girls can thus be simultaneously a gift of nature and a
who select themselves on the basis of “natural” conquest of merit, as much opposed to the abdi-
beauty undergo a radical transformation in cations of vulgarity as to ugliness.
their way of walking, sitting, laughing, smiling, Thus, the experience par excellence of the
talking, dressing, making-up etc.) Women of the “alienated body,” embarrassment, and the oppo-
petite bourgeoisie who have sufficient interests site experience, ease, are clearly unequally prob-
in the market in which physical properties can able for members of the petite bourgeoisie and
function as capital to recognize the dominant the bourgeoisie, who grant the same recognition
image of the body unconditionally without pos- to the same representation of the legitimate body
sessing, at least in their own eyes (and no doubt and legitimate deportment, but are unequally
objectively) enough body capital to obtain the able to achieve it. The chances of experiencing
highest profits, are, here too, at the site of great- one’s own body as a vessel of grace, a continu-
est tension. ous miracle, are that much greater when bodily
The self-assurance given by the certain knowl- capacity is commensurate with recognition;
edge of one’s own value, especially that of one’s and, conversely, the probability of experiencing
body or speech, is in fact very closely linked to the body with unease, embarrassment, timid-
the position occupied in social space (and also, ity grows with the disparity between the ideal
of course, to trajectory). Thus, the proportion of body and the real body, the dream body and the
women who consider themselves below average “looking-glass self ” reflected in the reactions of
in beauty, or who think they look older than they others (the same laws are also true of speech).
are, falls very rapidly as one moves up the so- Although it is not a petit-bourgeois monop-
cial hierarchy. Similarly, the ratings women give oly, the petit-bourgeois experience of the world
themselves for the different parts of their bodies starts out from timidity, the embarrassment of
tend to rise with social position, and this despite someone who is uneasy in his body and his lan-
the fact that the implicit demands rise too. It guage and who, instead of being “as one body
is not surprising that petit-bourgeois women— with them,” observes them from outside, through
who are almost as dissatisfied with their bodies other people’s eyes, watching, checking, correct-
as working-class women (they are the ones who ing himself, and who, by his desperate attempts
most often wish they looked different and who to reappropriate an alienated being-for-others,
are most discontented with various parts of their exposes himself to appropriation, giving himself
bodies), while being more aware of the useful- away as much by hyper-correction as by clumsi-
ness of beauty and more often recognizing the ness. The timidity which, despite itself, realizes
dominant ideal of physical excellence—devote the objectified body, which lets itself be trapped
such great investments, of self-denial and espe- in the destiny proposed by collective perception
cially of time, to improving their appearance and and statement (nicknames etc.), is betrayed by a
are such unconditional believers in all forms of body that is subject to the representation of oth-
cosmetic voluntarism (e.g., plastic surgery). ers even in its passive, unconscious reactions (one
As for the women of the dominant class, they feels oneself blushing). By contrast, ease, a sort
derive a double assurance from their bodies. of indifference to the objectifying gaze of others
Believing, like petit-bourgeois women, in the which neutralizes its powers, presupposes the
value of beauty and the value of the effort to be self-assurance which comes from the certainty
beautiful, and so associating aesthetic value and of being able to objectify that objectification,
 Bourdieu—Distinction • 1001

appropriate that appropriation, of being capable social space determined by volume and compo-
of imposing the norms of apperception of one’s sition of capital. Fully to construct the space of
own body, in short, of commanding all the pow- life-styles within which cultural practices are de-
ers which, even when they reside in the body fined, one would first have to establish, for each
and apparently borrow its most specific weap- class and class fraction, that is, for each of the
ons, such as “presence” or charm, are essentially configurations of capital, the generative formula
irreducible to it. This is the real meaning of the of the habitus which retranslates the necessities
findings of the experiment by W. D. Dannen- and facilities characteristic of that class of (rel-
maier and F. J. Thumin, in which the subjects, atively) homogeneous conditions of existence
when asked to assess the height of familiar per- into a particular life-style. One would then have
sons from memory, tended to overestimate most to determine how the dispositions of the habi-
of the height of those who had most authority tus are specified, for each of the major areas of
or prestige in their eyes.21 It would seem that the practice, by implementing one of the stylistic
logic whereby the “great” are perceived as physi- possibles offered by each field (the field of sport,
cally greater than they are applies very generally, or music, or food, decoration, politics, language
and that authority of whatever sort contains a etc.). By superimposing these homologous
power of seduction which it would be naive to spaces one would obtain a rigorous representa-
reduce to the effect of self-interested servility. tion of the space of life-styles, making it possible
That is why political contestation has always to characterize each of the distinctive features
made use of caricature, a distortion of the bodily (e.g., wearing a cap or playing the piano) in the
image intended to break the charm and hold up two respects in which it is objectively defined,
to ridicule one of the principles of the effect of that is, on the one hand by reference to the set
authority imposition. of features constituting the area in question (e.g.,
Charm and charisma in fact designate the the system of hairstyles), and on the other hand
power, which certain people have, to impose by reference to the set of features constituting
their own self-image as the objective and collec- a particular life-style (e.g., the working-class
tive image of their body and being; to persuade life-style), within which its social significance is
others, as in love or faith, to abdicate their ge- determined.
neric power of objectification and delegate it to
the person who should be its object, who thereby SOURCES
becomes an absolute subject, without an exterior The survey on which the work was based was car-
ried out in 1963, after a preliminary survey by ex-
(being his own Other), fully justified in existing,
tended interview and ethnographic observation, on
legitimated. The charismatic leader manages to
a sample of 692 subjects (both sexes) in Paris, Lille,
be for the group what he is for himself, instead and a small provincial town. To obtain a sample large
of being for himself, like those dominated in enough to make it possible to analyse variations in
the symbolic struggle, what he is for others. He practices and opinions in relation to sufficiently ho-
“makes” the opinion which makes him; he con- mogeneous social units, a complementary survey was
stitutes himself as an absolute by a manipulation carried out in 1967–68, bringing the total number
of symbolic power which is constitutive of his of subjects to 1,217. Because the survey measured
power since it enables him to produce and im- relatively stable dispositions, this time-lag does not
pose his own objectification. seem to have affected the responses (except perhaps
for the question on singers, an area of culture where
fashions change more rapidly).
The Universes of Stylistic Possibles The following complementary sources (C.S.)
were also used:
Thus, the spaces defined by preferences in food, I. The 1966 survey on “businessmen and senior
clothing or cosmetics are organized according executives” was carried out by SOFRES (Société
to the same fundamental structure, that of the française d’enquêtes par sondages) on behalf of the
1002 • Part IX: The Consequences of Inequality

Centre d’études des supports de publicité (CESP). (2) ouvriers (industrial manual workers); (3) indus-
The sample consisted of 2,257 persons aged 15 and trial and commercial employers; (4) clerical workers
over, each living in a household the head of which and junior executives; (5) senior executives and the
was a large industrial or commercial employer, “liberal professions.”
a member of the professions, a senior executive, III. INSEE, “La consommation alimentaire
an engineer, or a secondary or higher-education des Français,” Collections de l’INSEE. (The regular
teacher. The questionnaire included a set of ques- INSEE surveys on eating habits.)
tions on reading habits and the previous few days’ IV. SOFRES, Les habitudes de table des Français
reading of daily, weekly and monthly newspapers (Paris, 1972).
and magazines, use of radio and TV, standard of liv- V. SOFRES, Les Français et la gastronomie, July
ing, household equipment, life-style (holidays, sport, 1977. (Sample of 1,000.)
consumption), professional life (conferences, travel,
business meals), cultural practices and the principal NOTES
basic data (educational level, income, population of 1. The gaps are more clear-cut and certainly more
place of residence etc.). I had access to the whole set visible as regards education than income, because
of distributions by the socio-occupational category information on incomes (based on tax declarations)
of the head of household or individual. is much less reliable than information on qualifica-
II. The regular survey by INSEE (Institut na- tions. This is especially true of industrial and com-
tional de la statistique et des études economiques) mercial employers (who, in the CESP survey—C.S.
on household living conditions and expenditure was I—provided, along with doctors, the highest rate of
based in 1972 on a representative sample of 13,000 non-response to the questions about income), crafts-
households. It consists of a survey by questionnaire men, shopkeepers, and farmers.
dealing with the characteristics of the household 2. It follows from this that the relationship be-
(composition, ages, occupation of the head), accom- tween conditions of existence and practices or the
modation and facilities, major expenditure (clothing, meaning of practices is not to be understood in
fuel etc.), periodic expenditure (rent, service charges terms either of the logic of mechanism or of the logic
etc.), combined with analysis of account books for of consciousness.
current expenditure, left with each household for a 3. In contrast to the atomistic approach of social
week, collected and checked by the interviewer. This psychology, which breaks the unity of practice to
survey makes it possible to assess the whole range establish partial “laws” claiming to account for the
of expenditure (except certain major and infrequent products of practice, the opus operatum, the aim is
items such as air travel, removal expenses etc.), as to establish general laws reproducing the laws of pro-
well as items of consumption not preceded by pur- duction, the modus operandi.
chase (food in the case of farmers, items drawn from 4. Economic theory, which treats economic
their stocks by craftsmen and shopkeepers), which agents as interchangeable actors, paradoxically fails
are evaluated at their retail price to allow compari- to take account of the economic dispositions, and is
son with the other categories of households. This ex- thereby prevented from really explaining the systems
plains why consumption is considerably higher than of preferences which define incommensurable and
income in the case of farmers and small businessmen independent subjective use-values.
(categories which are always particularly prone to 5. An ethic, which seeks to impose the principles
under-declare their income). For the overall findings, of an ethos (i.e., the forced choices of a social con-
see G. Bigata and B. Bouvier, “Les conditions de vie dition) as a universal norm, is another, more subtle
des ménages en 1972,” Collections de l’INSEE, ser. way of succumbing to amor fati, of being content
M, no. 32 (February 1972). The data presented here with what one is and has. Such is the basis of the
come from secondary analysis of tables by narrow felt contradiction between ethics and revolutionary
categories produced at my request. intent.
The other surveys consulted are simply listed 6. “Bourgeois” is used here as shorthand for
below. These studies, almost always devoted to a par- “dominant fractions of the dominant class,” and “in-
ticular area of cultural activity, are generally based on tellectual” or “artist” functions in the same way for
relatively limited samples. They mostly use a classi- “dominated fractions of the dominant class.”
fication which groups the occupations into five cate- 7. F. Nietzsche, Der Wille zur Macht (Stuttgart,
gories: (1) agriculteurs (farmers and farm labourers); Alfred Kröner, 1964), no. 943, p. 630.
 Bourdieu—Distinction • 1003

8. Bananas are the only fruit for which manual is valued more at higher levels of the social hierarchy)
workers and farm workers have higher annual per is reflected in the small proportion of manual work-
capita spending (FF 23.26 and FF 25.20) than all ers who say that “there is a new life after death” (15
other classes, especially the senior executives, who percent, compared with 18 percent of craftsmen and
spend most on apples (FF 31.60 as against FF 21.00 shopkeepers, office workers and middle managers,
for manual workers), whereas the rich, expensive and 32 percent of senior executives).
fruits—grapes, peaches and nuts—are mainly eaten 15. The oppositions are much less clear-cut in the
by professionals and industrial and commercial em- middle classes, although homologous differences are
ployers (FF 29.04 for grapes, 19.09 for peaches and found between primary teachers and office workers
17.33 for nuts, as against FF 6.74, 11.78 and 4.90, on the one hand and shopkeepers on the other.
respectively, for manual workers). 16. La bouffe: “grub,” “nosh”; grande bouffe:
9. This whole paragraph is based on second- “blow-out” (translator).
ary analysis of the tables from the 1972 INSEE 17. The preference for foreign restaurants—Ital-
survey on household expenditure on 39 items by ian, Chinese, Japanese and, to a lesser extent, Rus-
socio-occupational category (C.S. II). sian—rises with level in the social hierarchy. The
10. A fuller translation of the original text would only exceptions are Spanish restaurants, which are
include: “les nourritures à la fois les plus nourrissantes associated with a more popular form of tourism, and
et les plus économiques” (the double tautology North African restaurants, which are most favoured
showing the reduction to pure economic function.)” by junior executives (C.S. IV).
(Translator’s note.) 18. Les gros: the rich; grosse bouffe: bulk food (cf.
11. In the French: “le gros et le gras, gros rouge, grossiste: wholesaler, and English “grocer”). See also
gros sabots, gros travaux, gros rire, grosses blagues, note 17 above (translator).
gros bon sens, plaisanteries grasses”—cheap red wine, 19. One could similarly contrast the bowl, which
clogs (i.e., obviousness), heavy work, belly laughs, is generously filled and held two-handed for unpre-
crude common sense, crude jokes (translator). tentious drinking, and the cup, into which a little
12. K. Marx, Capital, I (Harmondsworth, Pen- is poured, and more later (“Would you care for a
guin, 1976), 482 (translator). little more coffee?”), and which is held between two
13. One fine example, taken from Böhm-Bawerk, fingers and sipped from.
will demonstrate this essentialism: “We must now 20. Formality is a way of denying the truth of
consider a second phenomenon of human experi- the social world and of social relations. Just as popu-
ence—one that is heavily fraught with consequence. lar “functionalism” is refused as regards food, so too
That is the fact that we feel less concerned about fu- there is a refusal of the realistic vision which leads the
ture sensations of joy and sorrow simply because they working classes to accept social exchanges for what
do lie in the future, and the lessening of our concern they are (and, for example, to say, without cynicism,
is in proportion to the remoteness of that future. of someone who has done a favour or rendered a
Consequently we accord to goods which are intended service, “She knows I’ll pay her back”). Suppressing
to serve future ends a value which falls short of the avowal of the calculation which pervades social rela-
true intensity of their future marginal utility. We sys- tions, there is a striving to see presents, received or
tematically undervalue our future wants and also the given, as “pure” testimonies of friendship, respect,
means which serve to satisfy them.” E. Böhm-Bawerk, affection, and equally “pure” manifestations of gen-
Capital and Interest, II (South Holland, Ill., 1959), erosity and moral worth.
268, quoted by G. L. Stigler and G. S. Becker, “De 21. W. D. Dannenmaier and F. J. Thumin, “Au-
gustibus non est disputandum,” American Economic thority Status as a Factor in Perceptual Distortion
Review, 67 (March 1977), 76–90. of Size,” Journal of Social Psychology, 63 (1964),
14. We may assume that the deep-seated relation 361–365.
to the future (and also to one’s own person—which
1004 • Part IX: The Consequences of Inequality

115 • Tak Wing Chan and John H. Goldthorpe

The Social Stratikcation of Theater, Dance, and Cin-

ema Attendance

In current sociological literature the relationship various intermediate situations being also
between social inequality and patterns of cultural recognized. In some versions of the argu-
taste and consumption is the subject of a large ment (e.g. Bourdieu, 1984) it is further
and complex debate. This paper starts by out- claimed that the arrogation of ‘distinction’
lining the leading positions that have been taken in cultural taste and, conversely, processes
up in this debate, and then presents some illus- of ‘aesthetic distancing’ are actively used
trative findings from a research programme in by members of dominant social classes as
which the authors are currently engaged. These means of symbolically demonstrating and
findings are used chiefly as the basis for a critical confirming their superiority.
evaluation of the rival positions that have been 2. The individualization argument: this seeks
set out, although they may also be of interest in in effect to relegate the homology argument
their own right. In addition, attention is drawn to the past. In modern, relatively affluent
to the methodology on which the paper rests. Al- and highly commercialized societies, it is
though this may appear somewhat daunting, at held that differences in cultural taste and
least to readers who lack a statistical background, consumption are rapidly losing any clear
it is, we would argue, only by following such an grounding in social stratification: age, gen-
approach that the issues with which we are con- der, ethnicity or sexuality, for example, all
cerned can be adequately addressed. can, and do, serve as alternative social bases
In the sociological literature referred to, it is of cultural differentiation. And in more rad-
possible to identify three main lines of argument. ical forms of the argument (e.g. Bauman,
In their essentials, these arguments can be stated 1988; Featherstone, 1987), emphasis is
as follows, although each has variant forms. placed on the growing ability of individuals
1. The homology argument: this claims that to free themselves from social conditioning
social stratification—that is, the prevailing and influence of any kind and to choose
structure of inequality within a society— and form their own distinctive identities
and cultural stratification map onto each and lifestyles—patterns of cultural con-
other very closely. Individuals in higher so- sumption included.
cial strata are those who prefer and predom- 3. The omnivore–univore argument: this in
inantly consume ‘high’ or ‘elite’ culture, and effect challenges both the homology and
individuals in lower social strata are those individualization arguments (see especially
who prefer and predominantly consume Peterson & Kern, 1996; Peterson & Simkus,
‘popular’ or ‘mass’ culture—with, usually, 1992). As against the latter, it sees cultural

Tak Wing Chan and John H. Goldthorpe. “The Social Stratification of Theatre, Dance, and Cinema Atten-
dance,” Cultural Trends 14, issue 3 (September 2005), pp. 194–201, 203–205. Used by permission from Taylor
and Francis Ltd., http://www.tandf.co.uk/journals.
 Chan and Goldthorpe—The Social Stratikcation of Theater, Dance, and Cinema Attendance • 1005

differentiation as still mapping closely onto survey which was carried out by the Office for
social stratification; but, as against the for- National Statistics on behalf of Arts Council En-
mer, it does not see this mapping as being gland. The survey was based on a stratified prob-
on ‘elite-to-mass’ lines. Rather, it claims ability sample of individuals aged over 16 and
that the cultural consumption of individ- living in private households in England in 2001.
uals in higher social strata differs from that In all, 6,042 interviews were completed, repre-
of individuals in lower social strata in that senting a response rate of 64 per cent (for further
it is greater and much wider in its range. details, see Skelton et al., 2002). However, here
It comprises not only more ‘high-brow’ attention is restricted to respondents aged 20 to
culture, but also more ‘middle-brow’ and 64. With this limitation, and after deleting all
more ‘low-brow’ culture as well, while the cases with missing values on variables of interest
consumption of individuals in lower social here, there is an effective sample of N = 3,819.
strata tends to be largely restricted to more As regards theatre and cinema attendance, at-
popular cultural forms. Thus, the crucial tention is concentrated on the results obtained in
distinction is not between elite and mass the survey from six questions. Respondents were
but rather between cultural omnivores and asked whether or not, in the last 12 months, they
cultural univores. had attended a performance of a play/drama, a
musical, a pantomime, a ballet, some other form
In other work (Chan & Goldthorpe, 2005), of dance (including contemporary dance, Af-
we seek to evaluate these three arguments by an- rican People’s dance and South Asian dance),
alysing data on musical consumption. Far stron- or had seen a film at a cinema (or other venue
ger, even if still somewhat qualified, support is rather than at home). It should be stressed that
found for the omnivore–univore argument than interest in these results lies not primarily in the
for either the homology or individualization ar- answers given to each question taken separately,
gument, as explained further below. However, but rather in the possibility that the answers to
it has to be noted that the omnivore–univore the six questions taken together can reveal patterns
argument was initially developed with specific of theatre and cinema consumption and in turn
reference to musical consumption (Peterson & serve to identify types of consumer in this domain.
Simkus, 1992); and while, in this context, it As will be seen, this interest is reflected in the
has in fact received support additional to that of way in which the data are analysed.
the present authors (see e.g. Coulangeon, 2003; As regards social stratification, we believe it
Van Eijck, 2001), its wider application has been important to do more than treat this through
rather little studied. The present paper, therefore, some single classification or scale of an essentially
aims to examine how well it fares in another do- ad hoc kind—such as, say, the Market Research
main of cultural consumption, that of theatre, Society categories (AB, C1, C2, D) or the old
dance and cinema. This differs from music in Registrar General’s Social Classes. Therefore in-
several ways, but perhaps most obviously for formation collected in the Arts in England survey
present purposes in that consumption generally on respondents’ employment and occupation is
requires attendance at some venue and, with drawn on in order to allocate them by both social
the main exception of video film (which is dis- class and social status, which are viewed, follow-
counted in what follows), there is no equivalent ing a long-established tradition in sociology, as
to the home consumption via various media that conceptually separate forms of stratification.
is of major importance in the domain of music. The class structure overall, and likewise in-
dividuals’ positions within it, are seen as being
defined in a quite objective way by economic
Data and Concepts relations or, more precisely, by relations in la-
As in the authors’ work on musical consump- bour markets and production units. To allocate
tion, data is used from the Arts in England individuals by class, the new National Statistics
1006 • Part IX: The Consequences of Inequality

Socio-Economic Classification (NS-SEC) is cultural consumption it is status that will carry

used—in its seven-category version—which is the greater weight. This is because differences in
specifically designed to capture differences in status are typically expressed in lifestyles, and
employment relations (Rose & Pevalin, 2003). cultural consumption is one important aspect of
The categories of the classification are shown lifestyle through which status ‘markers’ can be
in Table 115.4. In contrast, the status order is readily laid down. In order to test whether this
seen as reflecting inter-subjective assessments of expectation holds good, it is of course essential
individuals’ social superiority, equality and in- that one should be able to distinguish class and
feriority as expressed most directly in relations status, conceptually and operationally.
of social intimacy. Such relations, where present In addition to treating class and status sep-
among members of different social groupings, arately, we also draw on information available
imply a basic equality of status and, where ab- from the Arts in England Survey on respondents’
sent, a recognition of inequality. To allocate in- incomes and on their educational qualifications.
dividuals by status, a 31-category occupationally The latter are coded to the six National Voca-
based scale developed by the authors from anal- tional Qualification levels that range from ‘no
yses of patterns of close friendship in contempo- qualifications’ to ‘degree-level qualification or
rary British society is used (Chan & Goldthorpe, higher’. In sociological analyses of cultural con-
2004). The closer together any two categories in sumption, income and education are often taken
the scale are, the more similar, occupationally, as substitutes or proxies for more direct measures
are their members’ friends; the further apart they of class or status of the kind to be used in the
are, the less similar are their members’ friends. present study. However, income and education
In any society, the positions of individuals are here considered along with such measures of
within the class structure and the status order class and status, so that their independent effects,
will tend to be correlated—but not perfectly if any, can be established.
so. Instances of discrepancy between class and Finally, also included in the present analyses
status position are always likely to occur. For is socio-demographic information collected in
example, the authors’ own results indicate that the Arts in England survey, in particular regard-
in present-day Britain salaried professionals and ing respondents’ sex, age, marital status, family
associate professionals tend to have higher status composition and region of residence. Given that
than do salaried managers, and especially man- the primary concern is with the social stratifica-
agers in manufacturing, construction or trans- tion of theatre, dance and cinema attendance,
port, despite holding similar class positions as these socio-demographic variables are intended
defined in terms of employment relations; or to serve primarily as ‘controls’: that is to say, they
again that routine wage workers in services, espe- are brought into analyses chiefly in order to re-
cially personal services, tend to have higher sta- move the possibility of any hidden confounding
tus than even skilled manual workers. Since class of their effects with those of class, status, income
and status are only imperfectly correlated, it is and education on which present interest centres.
possible to ask whether it is the one or the other
that exerts the greater influence on individuals’
experience and action across different areas of so- Analytical Strategy
cial life. There is, for example, evidence that class In Table 115.1 we show the percentage of re-
is the dominant influence so far as individuals’ spondents who in the last year had attended a
economic life-chances are concerned—i.e. in de- theatre, for performances of the kinds previously
termining their degree of economic security and indicated, or a cinema. As can be seen, there is
their prospects—and also in shaping their po- some wide variation in the probabilities of at-
litical orientations and affiliations. But, in con- tendance, although much on lines that might be
trast, the expectation would be that in regard to expected. Cinema attendance is by far the most
 Chan and Goldthorpe—The Social Stratikcation of Theater, Dance, and Cinema Attendance • 1007

Table 115.1. Percentage of Respondents Who Have indicators. It seeks to identify a limited number
Visited a Cinema or a Theatre for Various Kinds of of discrete classes, or categories, of respondents
Performance in the Past Twelve Months such that, conditional on their belonging to one
Ballet 1.9 or other of these classes, individuals’ responses on
the indicator items become independent of each
Other dance 12.7
other—i.e. there is no longer any association be-
Pantomime 14.6
tween them. Insofar as this can be done, it can
Musical 25.4 be said that it is individuals’ membership of the
Play/drama 29.0 latent classes that is the source of the association
Cinema 62.7 initially found among their responses, and each
latent class can be taken as representing a quite
frequently reported, while, at the other extreme, distinctive pattern of response.
going to the theatre for a ballet performance is
at a very low level.
As earlier remarked, we wish to treat the data Results
in question primarily as basis for obtaining an Latent Class Analysis of Theatre and
understanding of individuals’ patterns of theatre Cinema Attendance
and cinema attendance and of the different types It turns out in fact that, as is shown in Table
of consumer in this cultural domain. To this end, 115.2, a very simple latent class solution can be
a statistical technique known as latent class anal- obtained for our data on theatre and cinema at-
ysis is employed, which can be intuitively under- tendance. With the minor technical modification
stood as follows. that is noted in Table 115.2, a model proposing
There are six questions that serve as indica- just two latent classes fits the data satisfactorily:
tors of theatre or cinema attendance. Since each that is, just two latent classes prove sufficient to
question has a two-option (yes/no) answer, there capture virtually all of the association that exists
are in fact 26 or 64 different possible response among responses on the six indicator items. Or,
sets. The overall pattern of individuals’ responses one could say, it emerges that underlying the re-
will therefore be complex. But the answers given sults previously reported in Table 115.1 on these
by respondents to the six questions can be ex- six different kinds of attendance, a clear, essen-
pected to show some degree of association. Thus, tially dichotomous, patterning prevails.
for example, those who say that they have been On the basis of this solution we can then go
to a play are also likely to report having been on to assign each individual in the sample to
to a cinema. Conversely, those who say that one or other of the two latent classes that are
they have not been to a ballet are also likely to identified—that is, to whichever he or she has
report that they have not been to other dance the highest probability of belonging to, given his
events, and so on. What the technique of latent or her own set of responses on the six indicator
class analysis aims to do is to simplify matters items; in this way the respondents are divided
by exploiting this association among the six into two types of consumer of theatre, dance and

Table 115.2. Latent Class Models Fitted to Data on Cultural Participation in

the Domain of Theatre, Dance and Cinema

Model Number of classes G2 df p

1 1 1583.64 57 0.00
2 2 268.16 50 0.00
3 2a 53.22 49 0.31
a
A local dependence term is included in this model to allow for an association between
attendance at ballet and other dance events.
1008 • Part IX: The Consequences of Inequality

Table 115.3. Estimated Size of the Latent Classes and Conditional Prob-
abilities (Percent) of Attendance Under Our Preferred Model

Latent class
1 2
Relative size (%), initial 62.5 37.5
Relative size (%), post-assignment 64.2 35.8
Probabilities of attendance (%)
Ballet 0.1 5.0
Other dance 5.6 24.6
Pantomime 6.7 27.9
Musical 6.9 56.2
Play/drama 6.1 67.1
Cinema 48.0 87.1

cinema. In Table 115.3 it is shown, first of all, Class 2 represents the theatre and cinema om-
that this process of assignment does not result nivores, and latent Class 1 the univores, whose
in any major change in the relative size of the consumption is in fact more or less restricted to
latent classes from that initially estimated under the cinema. Certainly, we find no evidence of the
the model; or, in other words, no great degree kind that might be expected from the homology
of uncertainty appears to arise about the latent argument of a cultural elite who, in pursuit of
class with which particular respondents should ‘distinction’, attend the theatre for, say, drama
be affiliated. Second, the probabilities of indi- and ballet performances but who at the same
viduals reporting each of the six different kinds time display ‘aesthetic distancing’ in shunning
of attendance considered are shown, given their musicals and pantomime. Members of latent
latent class membership. Class 2 have the highest probability of atten-
What, then, can be discovered about the dance at not only drama and ballet but also all
two types of consumer that are derived from other kinds of theatre performance covered and
the latent class analysis? Our findings are in fact the cinema. Furthermore, the very fact that the
rather clear cut. As can be seen, latent class 1, sample divides so readily into just two types of
which accounts for almost two-thirds of the sam- consumer is in itself sufficient to throw serious
ple, comprises individuals who have a very low doubt on the individualization argument. There
probability—less than 10 per cent—of having is no evidence here of the kind of individual di-
attended a theatre in the year before the inter- versity in cultural consumption that would, were
view for any of the kinds of performance distin- it present, effectively defy latent class analysis or
guished, and whose consumption is effectively at all events require that an unmanageably large
limited to a fairly modest—48 per cent—prob- number of latent classes be distinguished, and
ability of having visited a cinema. Latent class 2, ones to which individuals could be assigned only
in contrast, which accounts for somewhat over with great uncertainty.
one-third of the sample, comprises individuals
who have a relatively high probability (i.e. as
compared to the overall rates shown in Table Theatre, Dance and Cinema
115.1) of having attended a theatre for each of Attendance and Social Stratification
the kinds of performance covered and of having As already noted, in our work on musical con-
been to the cinema as well. sumption (Chan & Goldthorpe, 2005), we also
These findings would then, so far as they go, find, with some qualification, support for the
appear highly consistent with the omnivore–uni- omnivore–univore argument. Our latent class
vore argument initially referred to. The latent analyses in this case point in fact to three types
 Chan and Goldthorpe—The Social Stratikcation of Theater, Dance, and Cinema Attendance • 1009

Table 115.4. Distribution of Univores (U) and Omnivores (O) Within NS Social Classes

NS social class U (%) O (%) N

1 Higher managerial and professional occupations 43.9 56.2 488
2 Lower managerial and professional occupations 49.4 50.6 1023
3 Intermediate occupations 63.2 36.8 574
4 Small employers and own-account workers 72.7 27.3 275
5 Lower supervisory and technical occupations 77.7 22.3 359
6 Semi-routine occupations 77.1 22.9 620
7 Routine occupations 85.8 14.2 480

64.2 35.8 3819

of musical consumer: univores, whose consump- of lower supervisory and technical, semi-routine
tion is largely restricted to pop and rock, and and routine workers.
then two kinds of omnivore—‘true’ omnivores Figure 115.1 is then analogous to Table
and omnivore listeners. The former have a high 115.4, but with the 31 categories of our status
probability both of attending musical events and scale replacing the seven NS-SEC classes. An
of listening to music across all the genres distin- obvious ‘status gradient’ exists in the chances of
guished, while the latter are omnivorous only in being an omnivore rather than a univore.
their listening to broadcast or recorded music. These results are then consistent with the
Further analysis then reveals that the chances of general idea that it is members of higher social
being an omnivore, and especially a true omni- strata who are more likely to be culturally om-
vore, rather than a univore, increase with status, nivorous, and members of lower strata who are
although—following the expectations earlier more likely to be univorous. But to test this idea
mentioned—the effects of class are negligible more rigorously against the data on theatre and
once status is included in the analysis. In addi- cinema attendance, it is necessary to move on
tion, it is shown that even when the effects of from merely two-way, or bivariate, analysis to
status (and class) are controlled, the chances of analysis of a multivariate kind. That is to say, it
being a musical omnivore rather than a univore is necessary to relate the chances of an individ-
still increase fairly steadily with level of educa- ual being an omnivore rather than a univore to
tional qualifications, but that, in contrast, these the full range of stratification variables referred
chances do not appear to be affected by income to earlier and also to the socio-demographic
when other stratification variables are controlled. variables of sex, age, marital status, family com-
How far, then, are similar results obtained in re- position and region of residence that are intro-
gard to cultural consumption in the form of the- duced as controls. Only if the effects of all these
atre and cinema attendance? variables are considered simultaneously can we
To begin with, we may examine the simple hope to gain some reliable idea of their relative
two-way relationships that exist between the importance.
chances of being in this regard an omnivore or In Table 115.5 we report results from a bi-
a cinema-only univore (according to the previ- nary logistic regression analysis. Such an analysis
ous analyses) and class and status respectively. is appropriate where the ‘dependent’ variable—
In Table 115.4 we show the distribution of that on which our explanatory interest centres—
univores (latent Class 1) and omnivores (latent has just two possible values, ‘x’ or ‘not x’: or, in
Class 2) within the seven classes of NS-SEC. It our case, being an omnivore or not being an
is evident that omnivores are most common in omnivore, and thus a univore. The ȕ coefficients
the professional and managerial classes, 1 and 2, shown in the second column of the table repre-
where they are in fact in a slight majority, while sent the estimated effects of the variables listed
univores dominate in Classes 5, 6 and 7, those in the first column. A positive coefficient implies
 1010 • Part IX: The Consequences of Inequality

Figure 115.1. The Proportion of Respondents Being Uni- a result that serves to confirm what several other
vores and Omnivores Within Status Categories by Status investigators (e.g. O’Hagan, 1999; Quine, 1999)
Score have previously reported: namely, that a highly
(a) Theatre and Cinema Omnivores significant gender effect exists in that women are
more likely to be theatre goers than men; or, in
our terms, women are more likely than men to
60 be theatre and cinema omnivores rather than
simply cinema-only univores. Second, though,
50 the further finding of other investigators that the
percentage

probability of theatre attendance increases with

40 age cannot be confirmed. What we do find is
that having a family that includes children below
30 age 5, as compared with having no children, has
a significant negative effect on the chances of
20
being an omnivore. This would then suggest that
where positive age effects on theatre attendance
10
do show up in two-way analyses, or indeed in
-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 multivariate analyses in which family composi-
status score tion variables are not included, they should be
Theatre and Cinema Univores very cautiously interpreted, and with the pos-
(b)
sibility being kept in mind that they may well
90 reflect life-cycle stage rather than generation.
A third result likewise brings out the advan-
80 tage of multivariate analysis. It can be seen that
the coefficients for living in regions outside Lon-
70 don, although often negative in sign, in no case
percentage

achieve significance (though that for the North

60 comes close). Again, then, the implication is that
where in two-way analyses living outside Lon-
50 don (or the South East) appears to have a neg-
ative effect on theatre attendance, this finding
40 could easily mislead. If stratification variables are
not simultaneously considered, it could be that
what region variables largely pick up is not any
-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6
specifically geographical effects, such as the loca-
status score
tion of venues, travelling times, etc., but rather
that the higher the value of the explanatory the (concealed) effects of stratification variables,
variable, the higher the probability of being an on account of the populations of regions differ-
omnivore rather than a univore, and a negative ing in their class and status composition and in
coefficient implies the opposite. Coefficients that their average levels of income and educational
are starred are statistically significant. attainment.
In assessing the results obtained from our re- Now turning to the effects of stratification
gression analysis, we may begin with those relat- variables in our own analysis, which is our main
ing to the socio-demographic control variables focus of interest, the following points stand out.
at the top of the table, since these turn out to be First, it can be seen that when class and status
methodologically instructive. First, we may note are included in the analysis together, the effect of
 Chan and Goldthorpe—The Social Stratikcation of Theater, Dance, and Cinema Attendance • 1011

Table 115.5. Coefficients from Binary Logistic Regres- consumption represents an aspect of lifestyle
sion Analysis for Effects of Covariates on the Probability through which status is readily expressed and
of Being an Omnivore Rather than a Univore displayed, is further borne out.
ˆ SE Second, Table 115.5 reveals that level of ed-
Femalea 0.615** 0.092 ucational qualification tends also to have a sig-
Marriedb 0.148 0.112 nificant and positive effect, over and above that
Separated 0.188 0.139 of status (and class), on the chances of being an
Agec 0.005 0.004 omnivore rather than a univore. Thus, in this
Child (0–4)d –0.562** 0.113 regard, too, our findings in the case of music are
Child (5–10) 0.070 0.100
broadly confirmed, although it should be noted
Child (11–15) 0.088 0.105
The Northe –0.231 0.124
that with theatre and cinema attendance, in
Midlands –0.207 0.123 contrast with musical consumption, the effects
South East 0.083 0.135 of education are not entirely ‘monotonic’. That
South West –0.189 0.153 is to say, the effects of having some educational
Income 0.026** 0.005 qualifications rather than none on the chance of
CSE/othersf 0.169 0.152 being an omnivore do not consistently increase
O levels 0.668** 0.128
with level of qualification. As can be seen, hav-
A levels 1.130** 0.145
Sub-degree 1.027** 0.160 ing CSE-level qualifications rather than none
Degree 1.223** 0.151 has no significant effect, and having tertiary but
Class 2g 0.078 0.126 sub-degree qualifications has a weaker effect than
Class 3 –0.161 0.160 having only A levels.
Class 4 –0.205 0.203 Third, our regression analysis also shows
Class 5 –0.134 0.218
that even when all other stratification variables
Class 6 –0.199 0.195
Class 7 –0.507* 0.230 are taken into account, a highly significant and
Status 0.631** 0.179 positive effect of income on theatre and cinema
Constant –2.118** 0.292 attendance still remains: the higher an individ-
Notes: a Male is the reference category. b Single is the reference
ual’s income the more likely he or she is to be
category. c Centred at age 20. d Not having children is the refer- an omnivore rather than a univore. This result
ence category. e London is the reference category. f No qualifica- is then that which is most at variance with what
tions is the reference category. g Class 1 is the reference category.
*
we find in the analysis of musical consumption
p < 0.05, ** p < 0.01.
in which, as earlier noted, income proved to have
status is highly significant and positive—i.e. the no significant effect on the probability of being
chances of being an omnivore rather than a uni- an omnivore rather than a univore, or in fact on
vore increase with status—while class effects are the probability of being a true omnivore rather
for the most part insignificant. Only member- than an omnivore listener.
ship of Class 7, that of routine, largely manual
wage workers, has a significant—negative—ef-
fect on the chances of being an omnivore. In Conclusions
other words, our results in the domain of the- Three rival arguments concerning the relation-
atre, dance and cinema do in this regard largely ship between social stratification and cultural
replicate those obtained in the domain of music, consumption have been advanced and widely
even if in the present case the preponderance of discussed in the sociological literature: what we
status over class effects is somewhat less marked, have labelled as the homology, individualization
and thus our general theoretical expectation that and omnivore–univore arguments. In previous
status will be more closely associated with cul- work on musical consumption in England, we
tural consumption than will class, because such have presented findings that broadly favour the
1012 • Part IX: The Consequences of Inequality

omnivore–univore argument, while lending little consumption. The chances of being a theatre
support to its competitors. In the present paper, and cinema omnivore significantly decrease for
we turn to the domain of theatre, dance and cin- those holding the least advantaged class posi-
ema, and ask how far the results of the analyses tions (Class 7), even when status is controlled
of musical consumption can be replicated. for. Further, while education is clearly an im-
In one respect, it could be said that the case portant influence on theatre and cinema atten-
of theatre, dance and cinema provides a yet dance, its effects would appear to be somewhat
more straightforward confirmation of the om- less consistent and also relatively less strong than
nivore–univore argument than that of music. they are on musical consumption. And finally,
Our latent class analyses indicate just two main and most notably, income, which appears sta-
patterns of attendance, and, in turn, two main tistically non-significant in regard to musical
types of consumer: those, around one-third of omnivorousness, exerts a highly significant and
the sample, who do appear omnivorous in hav- fairly substantial effect on the probability of
ing a relatively high probability of attending being a theatre and cinema omnivore, even when
theatre performances of all the kinds considered other stratification variables are included in the
and of going to the cinema; and those, around analysis.
two-thirds of the sample, who are univorous in
being cinema-goers only, if indeed they are con- REFERENCES
sumers in the domain of theatre and cinema at Bauman, Z. (1988). Freedom. Milton Keynes: Open
University Press.
all. That it is these two types of consumer that
Bourdieu, P. (1984). Distinction: A social critique
are empirically identifiable, and only these types,
of the judgement of taste. London: Routledge &
does then in itself serve to call both the homol- Kegan Paul.
ogy and the individualization arguments into Chan, T. W., & Goldthorpe, J. H. (2004). Is there
question. Our latent class analyses fail to reveal a status order in contemporary British society?
a cultural elite who systematically discriminate Evidence from the occupational structure of
among different kinds of theatre performance; friendship. European Sociological Review, 20(5),
but at the same time, theatre and cinema atten- 383–401.
dance clearly cannot be regarded as simply forms ——— (2005). Social stratification and cultural con-
of individual expression, devoid of all social pat- sumption: music in England. Paper presented at
a one-day workshop on ‘The social bases of cul-
terning. Moreover, when stratification variables
tural consumption’, Oxford, March 18.
are introduced into the analyses the results ob- Coulangeon, P. (2003). La stratification sociale des
tained are generally those that would be expected gôuts musicaux. Revue francaise de sociologie, 44,
under the omnivore–univore argument. Higher 3–33.
status, higher educational qualifications and a Featherstone, M. (1987). Lifestyle and consumer
higher income all increase individuals’ chances culture. Theory, Culture and Society, 4(1), 55–70.
of being an omnivore rather than a univore. In McCutcheon, A. L. (1987). Latent class analysis.
sum, theatre and cinema attendance, like mu- Newbury Park, CA: Sage.
sical consumption, is quite evidently socially O’Hagan, J. (1999). Policy analysis of theatre in the
UK. Cultural Trends, 9(34), 25–29.
stratified, but on omnivore–univore rather than
Peterson, R. A., & Kern, R. M. (1996). Changing
elite–mass lines.
highbrow taste: From snob to omnivore. Ameri-
At the same time, though, some differences can Sociological Review, 61(5), 900–907.
from our findings in the domain of music have Peterson, R. A., & Simkus, A. (1992). How musi-
also to be recognized. To begin with, while sta- cal tastes mark occupational status groups. In
tus effects do generally dominate class effects on M. Lamont & M. Fournier (Eds.), Cultivating
theatre and cinema attendance, as would be ex- differences: Symbolic boundaries and the making of
pected to be the case, class is not so completely inequality (pp. 152–186). Chicago: University of
overshadowed as it is in relation to musical Chicago Press.
 Lareau—Unequal Childhoods • 1013

Quine, M. (1999). Audiences for live theatre in Brit- (2002). Arts in England: Attendance, participation
ain: the present situation and some implication. and attitudes in 2001 (Res. report 27). London:
Cultural Trends, 9(34), 1–24. Arts Council England.
Rose, D., & Pevalin, D. J. (Eds.). (2003). A research- Van Eijck, K. (2001). Social differentiation in
er’s guide to the National Statistics Socio-economic musical taste patterns. Social Forces, 79(3),
Classification. London: Sage. 1163–1184.
Skelton, A., Bridgwood, A., Duckworth, K., Hut- Weber, M. (1922/1968). Economy and society. Berke-
ton, L., Fenn, C., Creaser, C., & Babbidge, A. ley: University of California Press.

116 • Annette Lareau

Unequal Childhoods
Class, Race, and Family Life

There are many studies that tell us of the detri- Among white and black working-class and
mental effects of poverty on children’s lives, but it poor families, childrearing strategies emphasize
is less clear what the mechanisms are for the trans- the “accomplishment of natural growth.” These
mission of class advantage across generations. parents believe that as long as they provide
I suggest that social class has an important love, food, and safety, their children will grow
impact on the cultural logic of childrearing (see and thrive. They do not focus on developing
Lareau 2003 for details). Middle-class parents, the special talents of their individual children.
both white and black, appear to follow a cultural Working-class and poor children have more free
logic of childrearing that I call “concerted culti- time and deeper and richer ties within their ex-
vation.” They enroll their children in numerous tended families than the middle-class children.
age-specific organized activities that come to Some participate in organized activities, but they
dominate family life and create enormous labor, do so for different reasons than their middle-class
particularly for mothers. Parents see these activ- counterparts. Working-class and poor parents
ities as transmitting important life skills to chil- issue many more directives to their children and,
dren. Middle-class parents also stress language in some households, place more emphasis on
use and the development of reasoning. Talking physical discipline than do middle-class parents.
plays a crucial role in the disciplinary strate- The pattern of concerted cultivation, with its
gies of middle-class parents. This “cultivation” stress on individual repertoires of activities, rea-
approach results in a frenetic pace for parents, soning, and questioning, encourages an emerging
creates a cult of individualism within the family, sense of entitlement in children. Of course, not all
and emphasizes children’s performance. parents and children are equally assertive, but the

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Annette Lareau, Unequal Childhoods: Class, Race, and Family Life, published by the
University of California Press in 2003. The article printed here was originally prepared by Annette Lareau for
inclusion in The Inequality Reader: Contemporary and Foundational Readings in Race, Class, and Gender, First
Edition, edited by David B. Grusky, pp. 537–548. Copyright © 2006 by Westview Press.
1014 • Part IX: The Consequences of Inequality

Table 116.1. Argument of Unequal Childhoods: Class Differences in Childrearing

Childrearing Approach
Concerted Cultivation Accomplishment of Natural Growth
Key Elements Parent actively fosters and assesses Parent cares for child and allows child
child’s talents, opinions, and skills to grow
Organization of *multiple child leisure activities *child “hangs out” particularly with kin
Daily Life orchestrated by adults
Language Use *reasoning/directives *directives
*child contestation of adult *rare for child to question or challenge
statements adults
*extended negotiations between *general acceptance by child of directives
parents and child
Interventions in *criticisms and interventions on *dependence on institutions
Institutions behalf of child *sense of powerlessness and frustrations
*training of child to take on this role *conflict between childrearing practices at
home and at school
Consequences Emerging sense of entitlement on Emerging sense of constraint on the part
the part of the child of the child

pattern of questioning and intervening among

the white and black middle-class parents in the Methodology
study contrasts sharply with the definitions of Study Participants
how to be helpful and effective observed among The study is based on interviews and observa-
the white and black working-class and poor fam- tions of children eight to ten years of age and
ilies. The pattern of the accomplishment of nat- their families. A team of graduate research assis-
ural growth, with its emphasis on child-initiated tants and I collected the data. The first phase in-
play, autonomy from adults, and directives, en- volved observations in third-grade public school
courages an emerging sense of constraint. Members classrooms, mainly in a metropolitan area in the
of these families, adults as well as children, tend Northeast. The schools serve neighborhoods in
to be deferential and outwardly accepting (with a white suburban area and two urban locales—
sporadic moments of resistance) in their inter- one a white working-class neighborhood and the
actions with professionals such as doctors and other a nearby poor black neighborhood. About
educators. At the same time, however, compared one-half of the children are white and about
to their middle-class counterparts, the white and one-half are black. One child is interracial. The
black working-class and poor families are more research assistants and I carried out individual
distrustful of professionals in institutions. These interviews (averaging two hours each) with all of
are differences with long-term consequences. In the mothers and most of the fathers (or guard-
a historical moment where the dominant soci- ians) of 88 children, for a total of 137 interviews.
ety privileges active, informed, assertive clients We also observed children as they took part in
of health and educational services, the various organized activities in the communities sur-
strategies employed by children and parents are rounding the schools. The most intensive part
not equally valuable. In sum, differences in fam- of the research, however, involved home obser-
ily life lie not only in the advantages parents are vations of 12 children and their families. Nine
able to obtain for their children, but also in the of the 12 families came from the classrooms
skills being transmitted to children for negotiat- I observed, but the boy and girl from the two
ing their own life paths. black middle-class families and the boy from the
 Lareau—Unequal Childhoods • 1015

poor white family came from other sites. Most whose parents were employees. Various crite-
observations and interviews took place between ria have been proposed to differentiate within
1993 and 1995, but interviews were done as this heterogeneous group, but authority in the
early as 1990 and as late as 1997. This chapter workplace and “credential barriers” are the two
focuses primarily on the findings from the obser- most commonly used. I assigned the families
vations of these 12 families since the key themes in the study to a working-class or middle-class
discussed here surfaced during this part of the category based on discussions with each of the
fieldwork. I do include some information from employed adults. They provided extensive infor-
the larger study to provide a context for under- mation about the work they did, the nature of
standing the family observations. All names are the organization that employed them, and their
pseudonyms. educational credentials. I added a third category:
families not involved in the labor market (a pop-
ulation traditionally excluded from social class
Intensive Family Observations groupings) because in the first school I studied,
The research assistants and I took turns visiting a substantial number of children were from
the participating families daily, for a total of households supported by public assistance. To
about 20 visits in each home, often in the space ignore them would have restricted the scope of
of one month. The observations were not lim- the study arbitrarily. The final subsample con-
ited to the home. Fieldworkers followed children tained 4 middle-class, 4 working-class, and 4
and parents as they took part in school activities, poor families.
church services and events, organized play, kin
visits, and medical appointments. Most field ob-
servations lasted about three hours; sometimes, Children’s Time Use
depending on the event (e.g., an out-of-town fu- In our interviews and observations of white and
neral, a special extended family event, or a long black middle-class children, it was striking how
shopping trip), they lasted much longer. In most busy they were with organized activities. Indeed,
cases, there was one overnight visit. We often one of the hallmarks of middle-class children’s
carried tape recorders with us and used the au- daily lives is a set of adult-run organized activ-
diotapes for reference in writing up field notes. ities. Many children have three and four activi-
Families were paid $350, usually at the end of ties per week. In some families, every few days
the visits, for their participation. activities conflict, particularly when one season
is ending and one is beginning. For example in
the white middle-class family of the Tallingers,
A Note on Class Garrett is on multiple soccer teams—the “A”
My purpose in undertaking the field obser- traveling team of the private Forest soccer club
vations was to develop an intensive, realistic and the Intercounty soccer team—he also has
portrait of family life. Although I deliberately swim lessons, saxophone lessons at school, pri-
focused on only 12 families, I wanted to com- vate piano lessons at home, and baseball and
pare children across gender and race lines. basketball. These organized activities provided
Adopting the fine-grained differentiation of a framework for children’s lives; other activities
categories characteristic of current neo-Marxist were sandwiched between them.
and neo-Weberian empirical studies was not ten- These activities create labor for parents. In-
able. My choice of class categories was further deed, the impact of children’s activities takes its
limited by the school populations at the sites I toll on parents’ patience as well as their time. For
had selected. Very few of the students were chil- example, on a June afternoon at the beginning of
dren of employers or of self-employed workers. summer vacation, in a white-middle-class fam-
I decided to concentrate exclusively on those ily, Mr. Tallinger comes home from work to take
1016 • Part IX: The Consequences of Inequality

Garrett to his soccer game. Garrett is not ready common and there were many children who
to go, and his lackadaisical approach to getting did not have any. Many children “hung out.”
ready irks his father: Television and video games are a major source
of entertainment but outdoor play can trump
Don says, “Get your soccer stuff—you’re either of these. No advanced planning, no tele-
going to a soccer game!” Garrett comes into phone calls, no consultations between mothers,
the den with white short leggings on under- no drop-offs or pickups—no particular effort
neath a long green soccer shirt; he’s number at all—is required to launch an activity. For in-
16. He sits on an armchair catty-corner from stance, one afternoon, in a black working-class
the television and languidly watches the family, Shannon (in 7th grade) and Tyrec (in 4th
World Cup game. He slowly, abstractedly, grade) walk out their front door to the curb of
pulls on shin guards, then long socks. His the small, narrow street their house faces. Shan-
eyes are riveted to the TV screen. Don comes non begins playing a game with a ball; she soon
in: “Go get your other stuff.” Garrett says he has company:
can’t find his shorts. Don: “Did you look in
your drawer?” Garrett nods. . . . He gets up to (Two boys from the neighborhood walk up.)
look for his shorts, comes back into the den a Shannon is throwing the small ball against
few minutes later. I ask, “Any luck yet?” Gar- the side of the row house. Tyrec joins in the
rett shakes his head. Don is rustling around game with her. As they throw the ball against
elsewhere in the house. Don comes in, says the wall, they say things they must do with
to Garrett, “Well, Garrett, aren’t you wearing the ball. It went something like this: Johnny
shoes?” (Don leaves and returns a short time Crow wanted to know. . . . (bounces ball
later): “Garrett, we HAVE to go! Move! We’re against the wall), touch your knee (bounce),
late!” He says this shortly, abruptly. He comes touch your toe (bounce), touch the ground
back in a minute and drops Garrett’s shiny (bounce), under the knee (bounce), turn
green shorts on his lap without a word. around (bounce). Shannon and Tyrec played
about four rounds.
This pressured search for a pair of shiny green
soccer shorts is a typical event in the Tallinger Unexpected events produce hilarity:
household. Also typical is the solution—a par-
ent ultimately finds the missing object, while At one point Shannon accidentally threw the
continuing to prod the child to hurry. The fact ball and it bounced off of Tyrec’s head. All the
that today’s frenzied schedule will be matched or kids laughed; then Tyrec, who had the ball,
exceeded by the next day’s is also par: went chasing after Shannon. It was a close,
fun moment—lots of laughter, eye contact,
Don: (describing their day on Saturday) To- giggling, chasing.
morrow is really nuts. We have a soccer game,
then a baseball game, then another soccer Soon a different game evolves. Tyrec is on re-
game. striction. He is supposed to remain inside the
house all day. So, when he thinks he has caught a
This steady schedule of activity—that none of glimpse of his mom returning home from work,
the middle-class parents reported having when he dashes inside. He reappears as soon as he real-
they were a similar age—was not universal. In- izes that it was a false alarm. The neighborhood
deed, while we searched for a middle-class child children begin an informal game of baiting him:
who did not have a single organized activity we
could not find one, but in working-class and The kids keep teasing Tyrec that his mom’s
poor homes, organized activities were much less coming—which sends him scurrying just
 Lareau—Unequal Childhoods • 1017

inside the door, peering out of the screen suggestion. That is not an option.” He smiled
door. This game is enacted about six times. as he turned back around to the sink. Chris-
Tyrec also chases Shannon around the street, tina says, looking at Alex: “There is a word
trying to get the ball from her. A few times for that you know, plagiarism.” Terry says
Shannon tells Tyrec that he’d better “get in- (not turning around), “Someone can sue you
side”; he ignores her. Then, at 6:50 [p.m.] Ken for plagiarizing. Did you know that?” Alex:
(a friend of Tyrec’s) says, “There’s your mom!” “That’s only if it is copyrighted.” They all
Tyrec scoots inside, then says, “Oh, man. You begin talking at once.
were serious this time.”
Here we see Alex cheerfully (though gently)
Informal, impromptu outdoor play is common goading his father by pretending to misunder-
in Tyrec’s neighborhood. A group of boys ap- stand the verbal instruction to consult a book for
proximately his age, regularly numbering four help. Mr. Williams dutifully rises to the bait. Ms.
or five but sometimes reaching as many as ten, Williams reshapes this moment of lighthearted-
play ball games together on the street, walk to ness by introducing a new word into Alexander’s
the store to get treats, watch television at each vocabulary. Mr. Williams goes one step further
other’s homes, and generally hang out together. by connecting the new word to a legal conse-
quence. Alex upstages them both. He demon-
strates that he is already familiar with the general
Language Use idea of plagiarism and that he understands the
In addition to differences by social class in time concept of copyright, as well.
use, we also observed differences in language In marked contrast to working-class and poor
use in the home. As others have noted (Bern- parents, however, even when the Williamses
stein, 1971; Heath, 1983) middle-class parents issue directives, they often include explanations
used more reasoning in their speech with chil- for their orders. Here, Ms. Williams is remind-
dren while working-class and poor parents used ing her son to pay attention to his teacher:
more directives. For example, in observations of
the African American home of Alex Williams, I want you to pay close attention to Mrs.
whose father was a trial lawyer and mother was Scott when you are developing your film.
a high-level corporate executive, we found that Those chemicals are very dangerous. Don’t
the Williamses and other middle-class parents play around in the classroom. You could get
use language frequently, pleasurably, and instru- that stuff in someone’s eye. And if you swal-
mentally. Their children do likewise. For exam- low it, you could die.
ple, one January evening, Alexander is stumped
by a homework assignment to write five riddles. Alex chooses to ignore the directive in favor of
He sits at the dinner table in the kitchen with instructing his misinformed mother:
his mother and a fieldworker. Mr. Williams is at
the sink, washing the dinner dishes. He has his Alex corrects her, “Mrs. Scott told us that we
back to the group at the dinner table. Without wouldn’t die if we swallowed it. But we would
turning around, he says to Alex, “Why don’t you get very sick and would have to get our stom-
go upstairs to the third floor and get one of those ach pumped.” Christina does not follow the
books and see if there is a riddle in there?” argument any further. She simply reiterates
that he should be careful.
Alex [says] smiling, “Yeah. That’s a good
idea! I’ll go upstairs and copy one from out Possibly because the issue is safety, Ms. Williams
of the book.” Terry turns around with a does not encourage Alex to elaborate here, as
dish in hand, “That was a joke—not a valid she would be likely to do if the topic were less
1018 • Part IX: The Consequences of Inequality

charged. Instead, she restates her directive and McAllister is told to do for the four-year-old
thus underscores her expectation that Alex will daughter of Aunt Dara’s friend Charmaine:
do as she asks.
Although Mr. and Ms. Williams disagreed on Someone tells Lori, “Go do [Tyneshia’s] hair
elements of how training in race relations should for camp.” Without saying anything, Lori
be implemented, they both recognized that their gets up and goes inside and takes the little
racial and ethnic identity profoundly shaped girl with her. They head for the couch near
their and their son’s everyday experiences. They the television; Lori sits on the couch and the
were well aware of the potential for Alexander girl sits on the floor. [Tyneshia] sits quietly
to be exposed to racial injustice, and they went for about an hour, with her head tilted, while
to great lengths to try to protect their son from Lori carefully does a multitude of braids.
racial insults and other forms of discrimination.
Nevertheless, race did not appear to shape the Lori’s silent obedience is typical. Generally,
dominant cultural logic of childrearing in Alex- children perform requests without comment.
ander’s family or in other families in the study. For example, at dinner one night, after Harold
All of the middle-class families engaged in ex- McAllister complains he doesn’t like spinach, his
tensive reasoning with their children, asking mother directs him to finish it anyway:
questions, probing assertions, and listening to
answers. Similar patterns appeared in interviews Mom yells (loudly) at him to eat: “EAT! FIN-
and observations with other African American ISH THE SPINACH!” (No response. Harold
middle-class families. is at the table, dawdling.) Guion and Runako
A different pattern appeared in working-class and Alexis finish eating and leave. I finish
and poor homes where there was simply less with Harold; he eats his spinach. He leaves
verbal speech than we observed in middle-class all his yams.
homes. There was also less speech between
parents and children, a finding noted by other The verbal world of Harold McAllister and
observational studies (Hart and Risley, 1995). other poor and working-class children offers
Moreover, interspersed with intermittent talk some important advantages as well as costs.
are adult-issued directives. Children are told Compared to middle-class children we observed,
to do certain things (e.g., shower, take out the Harold is more respectful towards adults in his
garbage) and not to do others (e.g., curse, talk family. In this setting, there are clear boundaries
back). In an African American home of a fam- between adults and children. Adults feel com-
ily living on public assistance in public housing, fortable issuing directives to children, which
Ms. McAllister uses one-word directives to co- children comply with immediately. Some of
ordinate the use of the single bathroom. There the directives that adults issue center on obliga-
are almost always at least four children in the tions of children to others in the family (“don’t
apartment and often seven, plus Ms. McAllister beat on Guion” or “go do [her] hair for camp”).
and other adults. Ms. McAllister sends the chil- One consequence of this is that Harold, despite
dren to wash up by pointing to a child, saying, occasional tiffs, is much nicer to his sister (and
“Bathroom,” and handing him or her a wash- his cousins) than the siblings we observed in
cloth. Wordlessly, the designated child gets up middle-class homes. The use of directives and
and goes to the bathroom to take a shower. the pattern of silent compliance are not universal
Children usually do what adults ask of them. in Harold’s life. In his interactions with peers, for
We did not observe whining or protests, even example on the basketball “court,” Harold’s ver-
when adults assign time-consuming tasks, such bal displays are distinctively different than inside
as the hour-long process of hair-braiding Lori the household, with elaborated and embellished
 Lareau—Unequal Childhoods • 1019

discourse. Nevertheless, there is a striking dif- that institutions function that end up convey-
ference in linguistic interaction between adults ing advantages to middle-class children. In their
and children in poor and working-class families standards, these institutions also permit, and
when compared to that observed in the home of even demand, active parent involvement. In this
Alexander Williams. Ms. McAllister has the ben- way as well, middle-class children often gain an
efit of being able to issue directives without hav- advantage.
ing to justify their decisions at every moment.
This can make childrearing somewhat less tiring.
Another advantage is that Harold has more Intervention in Institutions
autonomy than middle-class children in mak- Children do not live their lives inside of the
ing important decisions in daily life. As a child, home. Instead, they are legally required to go
he controls his leisure schedule. His basketball to school, they go to the doctor, and many are
games are impromptu and allow him to develop involved in church and other adult-organized ac-
important skills and talents. He is resourceful. tivities. In children’s institutional lives, we found
He appears less exhausted than ten-year-old Al- differences by social class in how mothers moni-
exander. In addition, he has important social tored children’s institutional experiences. While
competencies, including his deftness in negoti- in working-class and poor families children are
ating the “code of the street.”1 His mother has granted autonomy to make their own way in
stressed these skills in her upbringing, as she im- organizations, in the middle-class homes, most
presses upon her children the importance of “not aspects of the children’s lives are subject to their
paying no mind” to others, including drunks mother’s ongoing scrutiny.
and drug dealers who hang out in the neighbor- For example in an African American
hoods which Harold and Alexis negotiate. middle-class home, where both parents are col-
Still, in the world of schools, health care fa- lege graduates and Ms. Marshall is a computer
cilities, and other institutional settings, these worker and her husband a civil servant, their two
valuable skills do not translate into the same daughters have a hectic schedule of organized ac-
advantages as the reasoning skills emphasized tivities including gymnastics for Stacey and bas-
in the home of Alexander Williams and other ketball for Fern. When Ms. Marshall becomes
middle-class children. Compared to Alexander aware of a problem, she moves quickly, drawing
Williams, Harold does not gain the development on her work and professional skills and experi-
of a large vocabulary, an increase of his knowledge ences. She displays tremendous assertiveness,
of science and politics, a set of tools to customize doggedness, and, in some cases, effectiveness in
situations outside the home to maximize his ad- pressing institutions to recognize her daughters’
vantage, and instruction in how to defend his ar- individualized needs. Stacey’s mother’s proactive
gument with evidence. His knowledge of words, stance reflects her belief that she has a duty to
which might appear, for example, on future SAT intervene in situations where she perceives that
tests, is not continually stressed at home. her daughter’s needs are not being met. This
In these areas, the lack of advantage is not perceived responsibility applies across all areas
connected to the intrinsic value of the McAllis- of her children’s lives. She is no more (or less)
ter family life or the use of directives at home. diligent with regard to Stacey and Fern’s leisure
Indeed, one can argue raising children who are activities than she is with regard to their experi-
polite and respectful children and do not whine, ences in school or church or the doctor’s office.
needle, or badger their parents is a highly laud- This is clear in the way she handles Stacey’s tran-
able childrearing goal. Deep and abiding ties sition from her township gymnastics classes to
with kinship groups are also, one might further the private classes at an elite private gymnastic
argue, important. Rather, it is the specific ways program at Wright’s.
1020 • Part IX: The Consequences of Inequality

Ms. Marshall describes Stacey’s first session at sometimes teaching, learning and teaching,
the club as rocky: is a two-way proposition as far as I’m con-
cerned. And sometimes teachers have to learn
The girls were not warm. And these were lit- how to, you know, meet the needs of the kid.
tle . . . eight and nine year old kids. You know, Her style, her immediate style was not accom-
they weren’t welcoming her the first night. It modating to—to Stacey.
was kinda like eyeing each other, to see, you
know, “Can you do this? Can you do that?” Here Ms. Marshall is asserting the legitimacy of
an individualized approach to instruction. She
More importantly, Ms. Marshall reported frames her opening remark as a question (“Is
that the instructor is brusque, critical and not there a problem?”). Her purpose, however, is to
friendly toward Stacey. Ms. Marshall cannot hear alert the instructor to the negative impact she has
what was being said, but she could see the inter- had on Stacey (“She’s really upset.”). Although
actions through a window. A key problem is that her criticism is indirect (“Maybe it’s not all the
because her previous instructor had not used student . . . ”), Ms. Marshall makes it clear that
the professional jargon for gymnastic moves, she expects her daughter to be treated differently
Stacey does not know these terms. When the in the future. In this case, Stacey does not hear
class ends and she walks out, she is visibly upset. what her mother says, but she knows that her
Her mother’s reaction is a common one among wishes and feelings are being transmitted to the
middle-class parents: She does not remind her instructor in a way that she could not do herself.
daughter that in life one has to adjust, that she Although parents were equally concerned
will need to work even harder, or that there is about their children’s happiness, in working-class
nothing to be done. Instead, Ms. Marshall fo- and poor homes we observed different patterns
cuses on Tina, the instructor, as the source of of oversight for children’s institutional activities.
the problem: For example in the white working-class home of
Wendy Driver, Wendy’s mother does not nur-
We sat in the car for a minute and I said, ture her daughter’s language development like
“Look, Stac,” I said. She said, “I-I,” and she Alexander Williams’ mother does her son’s. She
started crying. I said, “You wait here.” The does not attempt to draw Wendy out or follow
instructor had come to the door, Tina. So up on new information when Wendy introduces
I went to her and I said, “Look.” I said, “Is the term mortal sin while the family is sitting
there a problem?” She said, “Aww . . . she’ll around watching television. But, just like Ms.
be fine. She just needs to work on certain Williams, Ms. Driver cares very much about
things.” Blah-blah-blah. And I said, “She’s re- her child and just like middle-class parents she
ally upset. She said you-you-you [were] pretty wants to help her daughter succeed. Ms. Driver
much correcting just about everything.” And keeps a close and careful eye on her Wendy’s
[Tina] said, “Well, she’s got—she’s gotta learn schooling. She knows that Wendy is having
the terminology.” problems in school. Ms. Driver immediately
signs and returns each form Wendy brings home
Ms. Marshall acknowledges that Stacey isn’t fa- from school and reminds her to turn the papers
miliar with specialized and technical gymnastics in to her teacher.
terms. Nonetheless, she continues to defend her Wendy is “being tested” as part of an ongoing
daughter: effort to determine why she has difficulties with
spelling, reading, and related language-based
I do remember, I said to her, I said, “Look, activities. Her mother welcomes these official
maybe it’s not all the student.” You know, efforts but she did not request them. Unlike the
I just left it like that. That, you know, middle-class mothers we observed, who asked
 Lareau—Unequal Childhoods • 1021

teachers for detailed information about every We listen to her read. We help her with her
aspect of their children’s classroom performance homework. So she has more attention here in
and relentlessly pursued information and as- a smaller household than it was when I lived
sessments outside of school as well, Ms. Driver with my parents. So, we’re trying to help her
seems content with only a vague notion of her out more, which I think is helping. And with
daughter’s learning disabilities. This attitude the two [special education] classes a day at
contrasts starkly with that of Stacey Marshall’s the school, instead of one like last year, she’s
mother, for example. In discussing Stacey’s class- learning a lot from that. So, we’re just hoping
room experiences with fieldworkers, Ms. Mar- it takes time and that she’ll just snap out of it.
shall routinely described her daughter’s academic
strengths and weaknesses in detail. Ms. Driver But Ms. Driver clearly does not have an inde-
never mentions that Wendy is doing grade-level pendent understanding of the nature or degree of
work in math but is reading at a level a full three Wendy’s limitations, perhaps because she is un-
years below her grade. Her description is vague: familiar with the kind of terms the educators use
to describe her daughter’s needs (e.g., a limited
She’s having problems. . . . They had a special “sight vocabulary,” underdeveloped “language
teacher come in and see if they could find out arts skills”). Perhaps, too, her confidence in the
what the problem is. She has a reading prob- school staff makes it easier for her to leave “the
lem, but they haven’t put their finger on it details” to them: “Ms. Morton, she’s great. She’s
yet, so she’s been through all kinds of special worked with us for different testing and stuff.”
teachers and testing and everything. She goes Ms. Driver depends on the school staff’s exper-
to Special Ed, I think it’s two classes a day . . . tise to assess the situation and then share the in-
I’m not one hundred percent sure—for her formation with her:
reading. It’s very difficult for her to read
what’s on paper. But then—she can remem- I think they just want to keep it in the school
ber things. But not everything. It’s like she till now. And when they get to a point where
has a puzzle up there. And we’ve tried, well, they can’t figure out what it is, and then I
they’ve tried a lot of things. They just haven’t guess they’ll send me somewhere else. . . .
put their finger on it yet.
Her mother is not alarmed, because “the
Wendy’s teachers uniformly praise her mother school” has told her not to worry about Wendy’s
as “supportive” and describe her as “very loving,” grades:
but they are disappointed in Ms. Driver’s fail-
ure to take a more active, interventionist role in Her report card—as long as it’s not spelling
Wendy’s education, especially given the formida- and reading—spelling and reading are like
ble nature of her daughter’s learning problems. F’s. And they keep telling me not to worry,
From Ms. Driver’s perspective, however, being because she’s in the Special Ed class. But be-
actively supportive means doing whatever the sides that, she does good. I have no behavior
teachers tell her to do. problems with her at all.

Whatever they would suggest, I would do. Ms. Driver wants the best possible outcome
They suggested she go to the eye doctor, so I for her daughter and she does not know how
did that. And they checked her and said there to achieve that goal without relying heavily on
was nothing wrong there. Wendy’s teachers:

Similarly, she monitors Wendy’s homework I wouldn’t even know where to start going.
and supports her efforts to read: On the radio there was something for children
1022 • Part IX: The Consequences of Inequality

having problems reading and this and that,

call. And I suggested it to a couple different Conclusions
people, and they were like, wait a second, it’s I have stressed how social class dynamics are
only to get you there and you’ll end up paying woven into the texture and rhythm of children
an arm and a leg. So I said to my mom, “No, and parents’ daily lives. Class position influences
I’m going to wait until the first report card critical aspects of family life: time use, language
and go up and talk to them up there.” use, and kin ties. Working-class and middle-class
mothers may express beliefs that reflect a sim-
Thus, in looking for the source of Ms. Driv- ilar notion of “intensive mothering,” but their
er’s deference toward educators, the answers behavior is quite different. For that reason, I
don’t seem to lie in her having either a shy per- have described sets of paired beliefs and actions
sonality or underdeveloped mothering skills. as a “cultural logic” of childrearing. When chil-
To understand why Wendy’s mother is accept- dren and parents move outside the home into
ing where Stacey Marshall’s mother would be the world of social institutions, they find that
aggressive, it is more useful to focus on social these cultural practices are not given equal value.
class position, both in terms of how class shapes There are signs that middle-class children bene-
worldviews and how class affects economic and fit, in ways that are invisible to them and to their
educational resources. Ms. Driver understands parents, from the degree of similarity between
her role in her daughter’s education as involving the cultural repertoires in the home and those
a different set of responsibilities from those per- standards adopted by institutions.
ceived by middle-class mothers. She responds to
contacts from the school—such as invitations to NOTES
the two annual parent-teacher conferences—but 1. Elijah Anderson, Code of the Street, New York:
W. W. Norton (1999).
she does not initiate them. She views Wendy’s
school life as a separate realm, and one in which
REFERENCES
she, as a parent, is only an infrequent visitor. Ms.
Anderson, Elijah. 1999. Code of the Street. New York,
Driver expects that the teachers will teach and NY: W. W. Norton.
her daughter will learn and that, under normal Bernstein, Basil. 1971. Class, Codes, and Control:
circumstances, neither requires any additional Theoretical Studies Towards a Sociology of Lan-
help from her as a parent. If problems arise, she guage. New York, NY: Schocken.
presumes that Wendy will tell her; or, if the issue Hart, Betty and Todd R. Risley. 1995. Meaningful
is serious, the school will contact her. But what Differences in the Everyday Experiences of Young
Ms. Driver fails to understand, is that the educa- American Children. New Haven: Yale University
Press.
tors expect her to take on a pattern of “concerted
Heath, Shirley Brice. 1983. Ways with Words: Lan-
cultivation” where she actively monitors and in-
guage, Life, and Work in Communities and Class-
tervenes in her child’s schooling. The teachers rooms. Cambridge: Cambridge University Press.
asked for a complicated mixture of deference Lareau, Annette. 2003. Unequal Childhoods: Class,
and engagement from parents; they were disap- Race, and Family Life. Berkeley, CA: University
pointed when they did not get it. of California Press.
 Reardon and Bischoff—Income Inequality and Income Segregation • 1023

117 • Sean F. Reardon and Kendra Bischoff

Income Inequality and Income Segregation

Introduction High- and low-income households may be phys-

After decreasing for decades, income inequality ically far apart or may be in economically ho-
in the United States has grown substantially in mogeneous neighborhoods that are close to each
the last forty years.1 At the same time, income other (Reardon, Matthews et al. 2008). Given
segregation has also increased (Jargowsky 1996; the strong correlation between income and race
Mayer 2001; Wheeler and La Jeunesse 2006; in the United States, income segregation is often
Watson 2009), although the details of how and empirically entangled with racial segregation,
why it has grown have been much less thoroughly implying the necessity of examining income
investigated than those related to income in- segregation separately by race as well as for the
equality. Common sense and empirical evidence population as a whole.
suggest that these trends are linked—greater Income segregation—and its causes and
inequality in incomes implies greater inequality trends—is of interest to sociologists because it
in the housing and neighborhood “quality” that may lead to inequality in social outcomes. It
families or individuals can afford—but it is less implies, by definition, that lower-income house-
clear in what specific ways income inequality af- holds will live in neighborhoods with lower
fects income segregation. average incomes than do higher-income house-
Income segregation—the uneven geographic holds. If the average income of one’s neighbors
distribution of income groups within a certain indirectly affects one’s own social, economic,
area—is a complex, multidimensional phenom- or physical outcomes (many sociological the-
enon. In particular, it may be characterized by ories predict such contextual effects; see, e.g.,
the spatial segregation of poverty (the extent to Jencks and Mayer 1990; Sampson, Raudenbush
which the lowest-income households are isolated et al. 1997; Leventhal and Brooks-Gunn 2000;
from middle- and upper-income households) Morenoff 2003; Sampson, Sharkey et al. 2008),
and/or the spatial segregation of affluence (the then income segregation will lead to more un-
extent to which the highest-income households equal outcomes between low- and high-income
are isolated from middle- and lower-income households than their differences in income
households). In addition, income segrega- alone would predict. In a highly segregated re-
tion may occur at different geographic scales. gion, then, higher-income households may be

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in
the following publication: Sean F. Reardon and Kendra Bischoff, “Income Inequality and Income Segregation,”
American Journal of Sociology 116:6 (January 2011), pp. 1934–1981, published by the University of Chicago
Press. The article printed here was originally prepared by Sean F. Reardon and Kendra Bischoff for this fourth
edition of Social Stratification: Class, Race, and Gender in Sociological Perspective, edited by David B. Grusky.
Copyright © 2014 by Westview Press.
1024 • Part IX: The Consequences of Inequality

advantaged over lower-income households not resources. If high-income households cluster to-
only by the difference in their own incomes, but gether within a small number of neighborhoods
also by the differences in their respective neigh- or municipalities, they may be able to collec-
bors’ incomes. tively better their own outcomes by pooling their
In this chapter we seek to understand if and extensive financial and social capital to generate
how variations in income inequality—including resources of which only they can take advantage.
among metropolitan areas, between racial groups, High-income neighborhoods, therefore, may
and over time—shaped patterns of income segre- have more green space, better-funded schools,
gation between 1970 and 2000. First, we describe better social services, or more of any number
a set of trends in average metropolitan area in- of other amenities that affect quality of life. In
come segregation from 1970 to 2000, including addition, high- and low-income neighborhoods
trends for all families, among white and black may differ in their social processes, norms, and
families separately, and in the segregation of pov- social environments (Sampson, Raudenbush
erty and affluence. We use a rank-order measure et al. 1997; Sampson, Morenoff et al. 1999).
of income segregation that avoids the confound- Conversely, if high-income households are not
ing of changes in the income distribution with clustered together, they may help to fund social
changes in income segregation. Second, we es- services and institutions that serve lower-income
timate the effect of metropolitan area income populations. Thus the ability of high-income
inequality on overall metropolitan area income households to self-segregate affects the welfare
segregation during this time period. Third, we of poor people and the neighborhoods in which
investigate in more detail how income inequal- they reside. Not only does this resource problem
ity affects the geographic segregation of poverty affect residents’ current quality of life and oppor-
and affluence and the extent to which it affects tunities, but it can also bridge generations—the
income segregation among white and black fam- income distribution in a community may affect
ilies differently. the intergenerational transfer of occupational
status through investment in locally financed
institutions that serve children, such as schools
Background (Durlauf 1996).
Income inequality in the United States in the
twentieth century exhibited a “U-shaped” trend
(Nielsen and Alderson 1997; Ryscavage 1999). The Relationship Between Income
The increase in income inequality over the past Inequality and Income Segregation
four decades has been driven largely by the Income inequality is a necessary condition for
growth of “upper-tail inequality”—dispersion in income segregation. By definition, if there were
the relative incomes of those in the upper half of no income inequality, there could be no income
the income distribution—rather than by growth segregation. Because all individuals would have
in “lower-tail inequality” (Piketty and Saez 2003; the same income and thus all neighborhoods
Autor, Katz et al. 2006, 2008). As we argue in would have the same income distribution.
the following discussion, this pattern of growth Nonetheless, income inequality alone is not
in income inequality has important implications sufficient to create income segregation, which
for its effects on income segregation. There are also requires the presence of income-correlated
many mechanisms through which income seg- residential preferences, an income-based hous-
regation might affect individual outcomes. The ing market, and/or housing policies that link
quality of public goods and local social institu- income to residential location. Three kinds of
tions is affected by a jurisdiction’s tax base and income-correlated residential preferences may
by the involvement of the community in the lead to income segregation in the presence
maintenance of and investment in these public of income inequality: preferences about the
 Reardon and Bischoff—Income Inequality and Income Segregation • 1025

socioeconomic characteristics of one’s neigh- on information about actual dollar income

bors, preferences about characteristics of one’s amounts. This index ranges from a minimum of
neighbors that are correlated with their income 0, indicating there is no income segregation (the
(e.g., race/ethnicity or educational levels), and income distribution in each local environment,
preferences about local public goods (e.g., pub- such as the census tract, mirrors that of the re-
lic school quality, public parks, police services). gion as a whole), to a maximum of 1, indicating
Even in the presence of sizeable income inequal- there is complete income segregation (there is no
ity, however, these income-correlated preferences income variation in any local environment).2
may be insufficient to produce income segrega- We measure income inequality within each
tion. Income segregation also requires the exis- race group-metro-year with the Gini index,
tence of a housing market based on residents’ which measures the extent to which the actual
ability to pay or housing policies that sort house- income distribution deviates from a hypothet-
holds by income. ical distribution in which each person receives
As we suggested previously, income inequal- an identical share of total income. The measure
ity may affect income segregation differently ranges from 0, indicating perfect equality (each
among black and white households, because individual receives an identical share of the dis-
of the variation in housing markets available to tribution), to 1, indicating maximum inequality
each group. Racial discrimination in the hous- (one individual holds all of the income).
ing market has meant that historically, minority Throughout this study we rely on tract-level
households (particularly black households) have family income data from the US Census (Geo-
had fewer residential options than white house- Lytics 2004; Minnesota Population Center
holds with similar income and wealth. Even if 2004). We include the one hundred metropol-
black households had the same preferences and itan areas with the largest populations in 2000;3
the same level of income inequality as white however, we retain only cases in which there
households, the racially discriminatory aspects were at least ten thousand families of a given
of the housing market likely would lead to race group in a given metro area in 1970, 1980,
lower levels of income segregation among black 1990, and 2000.
households than among white households. In
the period from 1970 to 2000, however, the
housing options available to middle-class blacks Results
greatly expanded (though housing discrimina- Patterns and Trends in Income
tion persisted throughout this period; see Farley Inequality and Segregation
and Frey 1994; Yinger 1995; Ross and Turner Figure 117.1 reports the average levels of income
2005), probably tightening the link between inequality and income segregation, by race, for
inequality and segregation among blacks over the one hundred largest metropolitan areas in
this period. the United States. Overall, metropolitan area in-
come inequality grew from 1970 to 2000, with
the greatest increase occurring in the 1980s.
Data and Methods Average metropolitan area income inequality
To distinguish income segregation (the sorting increased more rapidly for blacks than whites,
of families by income among census tracts, in- particularly in the 1970s, a pattern that reflects
dependent of the income distribution) from the continuing growth of the black middle class
income inequality (the uneven distribution of that began in the 1960s. Average metropolitan
income among families), we use the rank-order area income segregation followed a similar pat-
information theory index (HR) (Reardon 2011), tern, growing from 1970 to 2000, with the fast-
which relies only on information about the est increase occurring in the 1980s. For black
rank-ordering of incomes among families, not families, income segregation grew rapidly in
1026 • Part IX: The Consequences of Inequality

Figure 117.1. Trends in Family Income Inequality and Income Segregation, 1970–2000, by Race, 100
Largest Metropolitan Areas

0.46 0.18

INCOME SEGREGATIOIN (RANK-ORDER H)

0.44 0.16
INCOME INEQUALITY (GINI INDEX)

0.42 0.14

0.40 0.12

0.38 0.10

0.36 0.08

0.34 0.06
1970 1980 1990 2000 1970 1980 1990 2000 1970 1980 1990 2000
ALL FAMILIES WHITE BLACK

Note: Black trends based on 61 metropolitan areas with at least 10,000 black families in each census year, 1970–2000.

the 1970s and 1980s, at a rate more than three and the rich and nonrich, for example.4 Note
times faster than the corresponding growth of that in 1970 the poor were much less segregated
white income segregation. In fact, average black from the nonpoor than the rich were from the
income segregation was about one-third of a nonrich. Income segregation between the poor
standard deviation lower than white income seg- and nonpoor (segregation of poverty) increased
regation in 1970, but was about one standard sharply between 1970 and 1980, whereas in-
deviation higher than white income segregation come segregation of the rich and nonrich (seg-
by 1990. These patterns suggest a relationship regation of affluence) did not. In the 1980s,
between income inequality and income segrega- however, income segregation grew at all parts of
tion; for both black and white families, as well the income distribution. In the 1990s, income
as for the total population, changes in income segregation increased only modestly, and only
segregation appear to roughly mirror changes in between families in the middle part of the in-
income inequality. come distribution. On average, the segregation
Although Figure 117.1 shows changes in of poverty and the segregation of affluence were
the average values of the rank-order informa- relatively unchanged in the 1990s.
tion theory index from 1970 to 2000, it does Figure 117.3 shows the corresponding trends
not provide details on the extent to which for black families (the trend of income segre-
changes in income segregation are attributable gation among white families is quite similar to
to changes in the segregation of affluence and the overall patterns shown in Figure 117.2, be-
poverty. Figure 117.2 shows the average metro- cause of the size of the white population). Black
politan area segregation profiles across the one income segregation grew rapidly in the 1970s
hundred largest metropolitan areas from 1970 and 1980s at all parts of the black income dis-
to 2000, illustrating the extent to which segrega- tribution. Not only did low-income black fam-
tion has changed between the poor and nonpoor ilies become more isolated from middle- and
 Reardon and Bischoff—Income Inequality and Income Segregation • 1027

Figure 117.2. Trends in Average Metropolitan Area Income Segregation, by Income Percentile, All Fami-
lies, 100 Largest Metropolitan Areas, 1970–2000

0.28

0.24
INCOME SEGREGATION (H( P ))

0.20

0.16

0.12

0.08
0 20 40 60 80 100 1970 1980 1990 2000
PERCENTILE OF METROPOLITAN INCOME DISTRIBUTION YEAR

Note: Left panel of figure indicates estimated average between-tract segregation (as measured by the information theory
index, H) between families with incomes at or above and below each percentile of the metropolitan-wide family income
distribution. Right panel shows trends for between-tract segregation at three specific percentiles.

higher-income black families, but higher-income than the segregation of poverty in the one hun-
blacks also became increasingly segregated from dred largest metropolitan areas. The next section
lower- and middle-income black families. In the of this chapter investigates the extent to which
1990s this trend ceased abruptly. In fact, the seg- variation in income inequality can explain these
regation of lower- and moderate-income black patterns.
families from higher-income black families de-
clined slightly in the 1990s.
Our descriptive analyses reveal several impor- Estimating the Effects of Income
tant trends. First, average metropolitan area in- Inequality on Income Segregation
come inequality and segregation both increased We estimate the effect of income inequality on
from 1970 to 2000, although the growth in income segregation using a set of fixed-effects
both was much larger for black families than for regression models. The models rely on 644
white ones. Second, income segregation grew at metro-group-year cases. Because each observa-
all parts of the income distribution from 1970 tion in the data corresponds to a specific met-
to 2000, although at different times and rates ropolitan area, decade, and race group, there are
for black and white families. Most of the growth three potential sources of variation in income
in income segregation occurred between 1970 inequality: variation across decades (within each
and 1990. Nonetheless, both the segregation of metro-by-group cell), variation among met-
poverty and the segregation of affluence were ropolitan areas (within each decade-by-group
much higher in 2000 than they had been in cell), and variation between race groups (within
1970, for white and black families alike. Finally, each metro-by-decade cell).5 We use three fixed
the segregation of affluence is generally greater effects models, each relying on a different source
1028 • Part IX: The Consequences of Inequality

Figure 117.3. Trends in Average Metropolitan Area Income Segregation, by Income Percentile, Black
Families, 61 Largest Metropolitan Areas with at Least 10,000 Black Families, 1970–2000

0.28

  !#
#
 
  #

 #
0.24

INCOME SEGREGATION (H( P ))

!#



0.20


0.16


0.12


0.08


0
 20
 40
 60
 80 100
 1970
  1980
  1990
  2000

!#

#! #

"#!$# !
PERCENTILE OF METROPOLITAN INCOME DISTRIBUTION YEAR

Note: Left panel of figure indicates estimated average between-tract segregation (as measured by the information theory
index, H) between black families with incomes at or above and below each percentile of the metropolitan-wide family
income distribution. Right panel shows trends for between-tract segregation at three specific percentiles.

of variation in income inequality, to estimate area-by-year fixed effects and group-specific

the effects of income inequality on segregation dummy variables. In the final set of models
over time, across race groups, and across met- (Model 5 and 6) we estimate the effect of differ-
ropolitan areas, respectively. In addition, we ences in income inequality among racial groups
control for metropolitan demographic charac- on income segregation, using metropolitan area
teristics, housing market pressures and housing and group-by-year fixed effects. Because the
stock, intra- and inter-metropolitan mobility, three models rely on different sources of vari-
population growth, labor market characteris- ation in income inequality, each relies on a dif-
tics, and family structure (Wilson 1987; Massey ferent key assumption to support a causal claim
and Eggers 1993; Abramson, Tobin et al. 1995; about the effect of income inequality on income
Jargowsky 1996; Pendall and Carruthers 2003; segregation.6 As a result, many sources of po-
Wheeler 2006; Watson 2009). tential bias can be ruled out if the three sets of
In the first set of regression models (Models models produce similar results.
1 and 2) we estimate the effect of changes in
inequality on changes in segregation, including
both metropolitan area-by-group fixed effects Main Effects of Income Inequality on
and decade fixed effects. We compute boot- Income Segregation
strapped standard errors in all of the regression Table 117.1 reports the estimates from the mod-
models to account for the clustered nature of els. Of primary interest here are the estimated co-
the observations. In the second set of models efficients on the Gini index in Models 2, 4, and 6,
(Model 3 and 4) we estimate the effect of differ- which include the full set of covariates as well as
ences in income inequality between race groups the fixed effects. Model 2 yields an estimated as-
on income segregation, using metropolitan sociation of 0.467 (s.e. = 0.060; p < .001) between
Table 117.1. Esti-
mated Effects of In-
come Inequality on
Income Segregation,
1970–2000
1029
1030 • Part IX: The Consequences of Inequality

income inequality and income segregation, net of

the model’s fixed effects and covariates. In other Effects of Income Inequality on the
words, a change of one point in a group’s income Segregation of Poverty and Affluence
inequality is associated with a change of roughly One advantage of the information theory index
half a point in income segregation. is that it enables us to investigate whether in-
Model 4, which relies on variation in income come inequality more strongly affects income
inequality between white and black families segregation through the segregation of poverty
within the same metropolitan area and year, or the segregation of affluence. To do this, we
yields an estimated association between inequal- fit models identical to Models 2, 4, and 6 but
ity and segregation of 0.783 (s.e. = 0.125; p < using income segregation measured at a set of
.001), somewhat larger than the estimate from income percentiles (the 5th, 10th, 25th, 50th,
model 2. Finally, Model 6 yields an estimated as- 75th, 90th, and 95th percentiles) rather than
sociation between income inequality and income the rank-order information theory index as the
segregation of 0.502 (s.e. = 0.110; p < .001). outcome.
Thus, each of the three models yields estimated The estimates from these models are re-
coefficients on income inequality that imply in- ported in the left-hand section of Figure 117.4,
equality has a positive effect on income segre- which shows that income inequality has little
gation. To get a sense of the magnitude of these or no significant impact on the segregation of
effects, note that an effect of 0.500 (roughly the very poorest families in a metropolitan area
that found in Models 2 and 6) implies that the from all other families, but has large and signif-
changes in income inequality from 1970 to 2000 icant effects on the segregation of moderate- to
shown in Table 117.1 account for roughly 40 high-income families from those with lower
percent of the average change in black income incomes. In other words, income inequality ap-
segregation, 80 percent of the average change pears to be much more strongly linked to the
in white income segregation, and 60 percent segregation of affluence than to the segregation
of the average change in overall income segre- of poverty.
gation. Put differently, a one-standard deviation The same general pattern is true when we
change in income inequality leads to roughly investigate the effects of income inequality on
one-quarter of a standard deviation change in income segregation for white and black fam-
income segregation.7 ilies separately, as shown in the middle and
To ensure that the estimates from Mod- right-hand sections of Figure 117.4. However,
els 2, 4, and 6 are not driven by one particu- the effect of inequality on the segregation of af-
lar race group or decade, we fit an additional fluence is much stronger for black families than
set of models (not shown) for each race group for white ones. That is, in metropolitan areas
separately and another set of models for each and years when black income inequality is larg-
decade separately. In each model, the coeffi- est, the highest-earning 10 percent of black fam-
cient on inequality is positive and statistically ilies is much more segregated from the lower 90
significant. In the group-specific models, the percent of black families than when and where
coefficients range from 0.450 to 0.561; in the black income inequality is low.
decade-specific models, they range from 0.624
to 0.732. Thus, across all the models we find
that income inequality has a large and positive Discussion and Conclusions
association with income segregation, regardless Our analysis yields three main findings. First,
of whether we rely on temporal, between-group, we reproduce the finding in Watson (2009)
or between-metropolitan-area variation to iden- and Mayer (2001) linking income inequality to
tify this association, and regardless of which income segregation. Using a set of fixed-effects
groups or decades we use in the sample. regression models, we show that there is a strong
 Reardon and Bischoff—Income Inequality and Income Segregation • 1031

Figure 117.4. Estimated Effects of Family Income Inequality on Income Segregation by Percentile of
Income Distribution and Race, 100 Largest Metropolitan Areas, 1970–2000

Note: Bars indicate 95 percent confidence interval for estimates. Model specifications and samples are identical to those in
columns 1–3 of Table 117.1.

and robust relationship between within-race Our second main finding is that income in-
metropolitan area income inequality and equality influences income segregation primarily
within-race metropolitan area income segrega- by affecting the segregation of affluence, rather
tion, net of secular trends, stable between-race than the segregation of poverty. Although the
differences, and stable differences among met- segregation of poverty increased from 1970 to
ropolitan areas. Our estimates indicate that a 2000 for both white and black families, as well as
one-standard-deviation increase in income in- for all families, very little of this change is attrib-
equality leads to a quarter of a standard deviation utable to changes in income inequality. Given
increase in income segregation, an effect roughly this, we suspect that the segregation of poverty
half the size of that found by Watson (2009). is more a result of housing policy than of income
Nonetheless, these effects are large enough to inequality. Throughout the 1980s federal and
be substantially meaningful: they imply that in- metropolitan housing policies fostered the devel-
creasing income inequality was responsible for opment of high-density housing for low-income
40–80 percent of the changes in income seg- families. These policies may be responsible for
regation from 1970 to 2000. The strength and much of the increase in the segregation of pov-
consistency of our results across a wide range erty during that period. Likewise, the growth
of model specifications suggests this is a robust in scattered site low-income and mixed-income
relationship, at least among large metropolitan housing in the 1990s, coupled with the demoli-
areas in the period 1970–2000. It is important tion of some large, high-density public housing
to keep in mind that this analysis investigates the projects, may account for the stabilization of the
effect of income inequality in an era of rising segregation of poverty in the 1990s.
inequality. It is not clear to what extent these Third, we find that the relationship between
findings can be generalized to eras of lower, or income inequality and income segregation dif-
stable, inequality. fers for black and white families. In 1970 income
1032 • Part IX: The Consequences of Inequality

segregation among black families was lower than low-income households may have important and
among white families. This was probably the re- far-reaching consequences, particularly given
sult of the ghettoization of minorities that took that the top 10 percent of earners in the United
place, particularly in northern and midwestern States now receive 45 percent of all income. The
American cities, during the post–World War II segregation of these high-income households in
suburbanization boom. Because of housing dis- communities physically far from lower-income
crimination, black families were largely denied households may reduce the likelihood that
access to suburban areas, leaving both middle- high-income residents will have social, or even
and lower-income black families living in relative casual, contact with lower-income residents.
proximity in urban areas. The passage of housing This in turn may make it less likely that they will
and lending antidiscrimination legislation in the be willing to invest in metropolitan-wide pub-
1970s began to reduce the prevalence of housing lic resources that would benefit residents of all
discrimination, making a wider range of neigh- income levels, such as transportation networks,
borhood options available to middle-income utilities, parks, services, and cultural amenities.
black families. As a result, income segregation Moreover, the separation of the affluent and
among black families rose steeply from 1970 to poor implies that there will be few opportunities
1990, as the growing black middle class was able for disadvantaged families to benefit from local
to move into previously inaccessible suburban spillover of public goods. The distance between
areas. Although this increase in income segre- affluent and lower-income communities makes
gation among black families is a result of both it unlikely that disadvantaged families will be
the growing black middle class and reductions in able to take advantage of the local schools, parks,
housing discrimination—both signs of progress and services in which affluent communities in-
since the 1960s—it nonetheless may have neg- vest. Although most sociological theory and re-
ative consequences. Given high levels of racial search on the spatial distribution of income has
segregation in US cities, the growth of income focused on the effects of concentrated poverty on
segregation among black families results in the residents of poor neighborhoods, the findings
increasing racial and socioeconomic isolation of here suggest that a better understanding of the
low-income black families in neighborhoods of effects of concentrated affluence on residents far
concentrated disadvantage (Wilson 1987). from affluent communities is also needed. The
In sum, our analyses show that income in- segregation of affluence may directly affect the
equality has a strong and robust effect on income resources available to residents of both poor and
segregation, but it is more nuanced in form than low-income neighborhoods.
one might expect. In fact, additional analyses
(included in the original, unabridged version NOTES
of this chapter; see Reardon and Bischoff 2011) 1. http://www.census.gov/hhes/www/income/
histinc/p60no231_tablea3.pdf (accessed September
show that income inequality appears to be re-
2, 2009).
sponsible for a specific aspect of income segre-
2. Refer to Reardon and Bischoff (2011) for more
gation: the large-scale separation of the affluent details about this measure.
from lower-income households and families. 3. These 100 metropolitan areas together were
It does not, however, appear to be responsible home to 173 million residents in 2000, 62 percent of
for patterns of segregation of poverty (for that, the total U.S. total population, including 70 percent
housing policy is probably to blame). Nor is it of non-Hispanic blacks (23.6 million), 78 percent of
responsible for patterns of small-scale income Hispanics (27.6 million), and 89 percent of Asians
segregation, such as those resulting from the (9 million). The metropolitan areas range in pop-
gentrification of urban neighborhoods adjacent ulation from 561,000 (Scranton–Wilkes-Barre, PA)
to 11.3 million (New York–White Plains, NY-NJ).
to poor, nongentrifying neighborhoods.
4. The income segregation profile shows the ex-
The macroscale spatial segregation of tent of segregation of those below a given income
high-income households from middle- and percentile from those above that percentile. The
 Reardon and Bischoff—Income Inequality and Income Segregation • 1033

curved “profile” traces this value for all income REFERENCES

percentiles. So, for example, the leftmost dots on Abramson, A. J., M. S. Tobin, et al. 1995. “The
the curves indicate the segregation of the bottom Changing Geography of Metropolitan Opportu-
10 percent of families from the top 90 percent; nity: The Segregation of the Poor in U.S. Met-
dots in the middle of the curves indicate the seg- ropolitan Areas, 1970 to 1990.” Housing Policy
regation level of those with below-median income Debate 6 (1): 45–72.
from those with above-median income. For a full Autor, D. H., L. H. Katz, et al. 2006. “The Polar-
description of the income segregation profiles, see ization of the U.S. Labor Market.” American Eco-
the unabridged version of this chapter (Reardon nomic Review 95 (2): 189–194.
and Bischoff 2011). Autor, D. H., L. H. Katz, et al. 2008. “Trends in U.S.
5. Only 61 of the 100 metropolitan areas have Wage Inequality: Revising the Revisionists.” Re-
this latter variation, because we include observations view of Economics and Statistics 90 (2): 300–323.
for black families in the sample only for the 61 met- Durlauf, S. 1996. “A Theory of Persistent Income In-
ropolitan areas where there are at least 10,000 black equality.” Journal of Economic Growth 1: 75–93.
families in each of the four Census years. Farley, R., and W. H. Frey. 1994. “Changes in the
6. The models that include metro-by-group fixed Segregation of Whites from Blacks During the
effects rely on the assumption that changes in income 1980s: Small Steps Toward a More Integrated
inequality within a metropolitan area and race group Society.” American Sociological Review 59 (Feb-
over time are exogenous, conditional on secular ruary): 23–45.
trends common to all groups and metropolitan areas GeoLytics. 2004. Neighborhood Change Database.
and the set of included covariates in the model. The East Brunswick, NJ, GeoLytics, Inc.
models that include metro-by-year fixed effects rely Jargowsky, P. A. 1996. “Take the Money and Run:
on the assumption that differences between white Economic Segregation in U.S. Metropolitan
and black income inequality within the same met- Areas.” American Sociological Review 61 (6):
ropolitan area and decade are exogenous, once we 984–998.
have accounted for differences in income inequality Jencks, C., and S. Mayer. 1990. “The Social Con-
between white and black families that are common sequences of Growing Up in a Poor Neighbor-
across metropolitan areas and time and differences in hood.” In Inner-city Poverty in the United States,
inequality that are associated with differences in the edited by L. E. Lynn Jr. and M. G. H. McGeary,
covariates included in the model. Finally, the models 111–186. Washington, DC: National Academy
that include group-by-year and metropolitan area Press.
fixed effects rely on the assumption that differences Leventhal, T., and J. Brooks-Gunn. 2000. “The
in income inequality within a given year and for a Neighborhoods They Live in: The Effects of
given race group are exogenous, once we have ac- Neighborhood Residence on Child and Adoles-
counted for stable differences among metropolitan cent Outcomes.” Psychological Bulletin 126 (2):
areas and differences associated with the covariates in 309–337.
the model. None of these three assumptions is likely Massey, D. S., and M. L. Eggers. 1993. “The Spatial
to be perfectly true, but each model is somewhat in- Concentration of Affluence and Poverty During
sulated against threats to any other. the 1970s.” Urban Affairs Quarterly 29: 299–315.
7. These are computed from the 1970–2000 Mayer, S. E. 2001. “How the Growth in Income In-
changes in inequality and segregation shown in Table equality Increased Economic Segregation.” Joint
117.2. For example, income inequality among all Center for Poverty Research Working Paper 230.
families grew by 0.048 from 1970 to 2000. If the ef- Minnesota Population Center. 2004. National
fect of income inequality on segregation were 0.500, Historical Geographic Information System:
this would imply a change in income segregation of Pre-release Version 0.1. Minneapolis: University
0.048 · 0.5=0.024, which is roughly 70 percent of of Minnesota.
the observed total change (0.033) in income segre- Morenoff, J. D. 2003. “Neighborhood Mechanisms
gation from 1970 to 2000. Likewise, the standard and the Spatial Dynamics of Birth Weight.”
deviation of income inequality within a given year American Journal of Sociology 108 (5): 976–1017.
was roughly 0.025, on average, whereas the standard Nielsen, F., and A. S. Alderson. 1997. “The Kuznets
deviation of income segregation was roughly 0.050. Curve and the Great U-turn: Income Inequality
This implies that an effect of 0.500 corresponds to in U.S. Counties, 1970 to 1990.” American So-
an effect size of 0.25. ciological Review 62 (1): 12–33.
1034 • Part IX: The Consequences of Inequality

Pendall, R., and J. I. Carruthers. 2003. “Does Den- Sampson, R. J., S. W. Raudenbush, et al. 1997.
sity Exacerbate Income Segregation? Evidence “Neighborhoods and Violent Crime: A Multi-
from U.S. Metropolitan Areas, 1980 to 2000.” level Study of Collective Efficacy.” Science 277:
Housing Policy Debate 14 (4): 541–589. 918–924.
Piketty, T., and E. Saez. 2003. “Income Inequality in Sampson, R. J., P. Sharkey, et al. 2008. “Durable
the United States, 1913–1998.” Quarterly Journal Effects of Concentrated Disadvantage on Verbal
of Economics 118 (1): 1–39. Ability Among African-American Children.” Pro-
Reardon, S. F. 2011. “Measures of Income Segrega- ceedings of the National Academy of Sciences 105
tion.” Working paper, Stanford Center for Edu- (3): 845–852.
cation Policy Analysis. Watson, T. 2009. “Inequality and the Measurement
Reardon, S. F., and K. Bischoff. 2011. “Income of Residential Segregation by Income.” Review of
Inequality and Income Segregation.” American Income and Wealth 55 (3): 820–844.
Journal of Sociology 116 (4): 1092–1153. Wheeler, C. H. 2006. Urban Decentralization and
Reardon, S. F., S. A. Matthews, et al. 2008. “The Income Inequality: Is Sprawl Associated with Ris-
Geographic Scale of Metropolitan Racial Segre- ing Income Segregation Across Neighborhoods? St.
gation.” Demography 45: 489–514. Louis, MO: Federal Reserve Bank of St. Louis.
Ross, S. L., and M. A. Turner. 2005. “Housing Dis- Wheeler, C. H., and E. A. La Jeunesse. 2006. Neigh-
crimination in Metropolitan America: Explain- borhood Income Inequality. St. Louis, MO: Fed-
ing Changes Between 1989 and 2000.” Social eral Reserve Bank of St. Louis.
Problems 52: 152–180. Wilson, W. J. 1987. The Truly Disadvantaged: The
Ryscavage, P. 1999. Income Inequality in America: An Inner City, the Underclass, and Public Policy. Chi-
Analysis of Trends. New York: M. E. Sharpe. cago: University of Chicago Press.
Sampson, R. J., J. D. Morenoff, et al. 1999. “Be- Yinger, J. 1995. Closed Doors, Opportunities Lost: The
yond Social Capital: Spatial Dynamics of Collec- Continuing Costs of Housing Discrimination. New
tive Efficacy for Children.” American Sociological York: Russell Sage Foundation.
Review 64 (5): 633–660.
 Frank—What’s the Matter with Kansas? • 1035

Politics and Attitudes

118 • Thomas Frank

What’s the Matter with Kansas?

The poorest county in America isn’t in Appa- monuments. The backlash is what has made
lachia or the Deep South. It is on the Great possible the international free-market consen-
Plains, a region of struggling ranchers and dying sus of recent years, with all the privatization,
farm towns, and in the election of 2000 the deregulation, and deunionization that are its
Republican candidate for president, George W. components. Backlash ensures that Republicans
Bush, carried it by a majority of greater than 80 will continue to be returned to office even when
percent.1 their free-market miracles fail and their libertar-
This puzzled me when I first read about it, as ian schemes don’t deliver and their “New Econ-
it puzzles many of the people I know. For us it is omy” collapses.
the Democrats that are the party of workers, of The Great Backlash has made the laissez-faire
the poor, of the weak and the victimized. When revival possible, but this does not mean that it
I told a friend of mine about that impoverished speaks to us in the manner of the capitalists of
High Plains county so enamored of President old, invoking the divine right of money or de-
Bush, she was perplexed. “How can anyone who manding that the lowly learn their place in the
has ever worked for someone else vote Republi- great chain of being. On the contrary; the back-
can?” she asked. lash imagines itself as a foe of the elite, as the
This derangement is the signature expression voice of the unfairly persecuted, as a righteous
of the Great Backlash, a style of conservatism protest of the people on history’s receiving end.
that first came snarling onto the national stage That its greatest beneficiaries are the wealthiest
in response to the partying and protests of the people on the planet does not give it pause.
late sixties. While earlier forms of conservatism In fact, backlash leaders systematically down-
emphasized fiscal sobriety, the backlash mobi- play the politics of economics. The movement’s
lizes voters with explosive social issues—sum- basic premise is that culture outweighs econom-
moning public outrage over everything from ics as a matter of public concern—that Values
busing to un-Christian art—which it then mar- Matter Most, as one backlash title has it. On
ries to pro-business economic policies. Cultural those grounds it rallies citizens who would once
anger is marshaled to achieve economic ends. have been reliable partisans of the New Deal to
And it is these economic achievements—not the standard of conservatism.2 Old-fashioned
the forgettable skirmishes of the never-ending values may count when conservatives appear on
culture wars—that are the movement’s greatest the stump, but once conservatives are in office

Thomas Frank. What’s the Matter with Kansas? How Conservatives Won the Heart of America (New York: Met-
ropolitan Books, 2004), pp. 1, 5–6, 14, 16–17, 254–255.
1036 • Part IX: The Consequences of Inequality

the only old-fashioned situation they care to re- what it was about and went Republican, by a
vive is an economic regimen of low wages and margin in square miles of four to one.3
lax regulations. Over the last three decades they The attraction of such a scheme for conser-
have smashed the welfare state, reduced the tax vatives was powerful and obvious. The red-state
burden on corporations and the wealthy, and narrative brought majoritarian legitimacy to a
generally facilitated the country’s return to a president who had actually lost the popular vote.
nineteenth-century pattern of wealth distri- It also allowed conservatives to present their
bution. Thus the primary contradiction of the views as the philosophy of a region that Amer-
backlash: it is a working-class movement that has icans—even sophisticated urban ones—tradi-
done incalculable, historic harm to working-class tionally venerate as the repository of national
people. virtue, a place of plain speaking and straight
The leaders of the backlash may talk Christ, shooting.
but they walk corporate. Values may “matter The red-state/blue-state divide also helped
most” to voters, but they always take a backseat conservatives perform one of their dearest rhe-
to the needs of money once the elections arc torical maneuvers, which we will call the latte
won. This is a basic earmark of the phenome- libel: the suggestion that liberals are identifiable
non, absolutely consistent across its decades-long by their tastes and consumer preferences and
history. Abortion is never halted. Affirmative ac- that these tastes and preferences reveal the es-
tion is never abolished. The culture industry is sential arrogance and foreignness of liberalism.
never forced to clean up its act. While a more straightforward discussion of poli-
The great dream of conservatives ever since tics might begin by considering the economic in-
the thirties has been a working-class movement terests that each party serves, the latte libel insists
that for once takes their side of the issues, that that such interests are irrelevant. Instead it’s the
votes Republican and reverses the achievements places that people live and the things that they
of working-class movements of the past. In drink, eat, and drive that are the critical factors,
the starkly divided red/blue map of 2000 they the clues that bring us to the truth. In particu-
thought they saw it being realized: the old Dem- lar, the things that liberals are said to drink, eat,
ocratic regions of the South and the Great Plains and drive; the Volvos, the imported cheese, and
were on their team now, solid masses of uninter- above all, the lattes.
rupted red, while the Democrats were restricted
to the old-line, blueblood states of the North- NOTES
east, along with the hedonist left coast. . . . 1. I am referring to McPherson County, Ne-
braska. “Trampled Dreams: The Neglected Economy
Just by looking at the [electoral] map, [pun-
of the Rural Great Plains,” a study by Patricia Funk
dits could] easily tell that George W. Bush was
and Jon Bailey (Walthill, Neb.: Center for Rural Af-
the choice of the plain people, the grassroots fairs, 2000), p. 6.
Americans who inhabited the place we know 2. Ben J. Wattenberg, Values Matter Most: How
as the “heartland,” a region of humility, guilt- Republicans or Democrats or a Third Party Can Win
lessness, and, above all, stout yeoman righteous- and Renew the American Way of Life (New York: Free
ness. The Democrats, on the other hand, were Press, 1995).
the party of the elite. Just by looking at the map 3. I am referring here to the county-by-county re-
we could see that liberals were sophisticated, sults, in square miles. Bush won the votes of counties
wealthy, and materialistic. While the big cities occupying 2,427,039 square miles, while Gore only
took the votes of 580,134 square miles.
blued themselves shamelessly, the land knew
 Hout and Laurison—The Realignment of U.S. Presidential Voting • 1037

119 • Michael Hout and Daniel Laurison

The Realignment of U.S. Presidential Voting

Introduction voting in US presidential elections in the 1960s

American presidential elections since the 1960s were more important in the 2000s, even as
have supplied ample material for study by polit- gender and church attendance emerged as new
ical scientists and political sociologists, who have cleavages (Greeley and Hout 2006, ch. 3; Bartels
contended that the stable class politics of the in- 2008). In Britain class voting fluctuated without
dustrial era—roughly 1932–1964 (or perhaps as trend (Goldthorpe 1999). In the United States
late as 1976)—gave way to new, “postmaterial” classes realigned (Hout, Brooks, and Manza
politics (e.g., Inglehart 1977; Lipset 1981; Clark 1995), and the effects of income on the vote in-
and Lipset 1991). Scholars who subscribe to this creased (Bartels 2008). In a six-country compar-
view point to newer cleavages based on gen- ison of elections from the 1960s to the 1990s,
der, identity, concern for the environment, and Brooks, Nieuwbeerta, and Manza (2006) found
family values, which they argue have displaced small but significant changes in overall cleav-
class from its central place in American politics. age—increases in the United States and Aus-
When Democrats appeal to middle-class voters tralia, decreases in Germany and Britain—that
and the British Labour Party touts a “third way” made politics in the six countries more similar
in twenty-first-century politics, some of these in the 1990s than in the 1960s.
ideas ring true.
Other scholars have countered the postmate-
rialists. Hout, Brooks, and Manza (1995) noted Class Voting and Class Politics
that political cleavages need not sum to zero. Interest in class voting goes back to the dawn
Voters typically balance many facets of their po- of modern understandings of class, to the grand
litical identity when they vote, including class, theorists of the nineteenth century. The roots of
race, gender, religion, and region. Nothing in the contemporary debates about class voting, how-
political arithmetic of social cleavages mandates ever, are planted in data, not in class theory. The
that the emergence of a new factor lowers the national election surveys that have been taken in
weight given to others. The empirical evidence many countries since the 1960s and the World
supports this insight. Research has shown that Values Surveys that began in the 1970s yielded
nearly all those factors that were important for a harvest of class-voting time series. Findings

This commissioned piece was prepared by Michael Hout and Daniel Laurison for this fourth edition of Social
Stratification: Class, Race, and Gender in Sociological Perspective, edited by David B. Grusky. Copyright ©
2014 by Westview Press. The article updates the work of Hout, Brooks, and Manza (1995). The authors owe
a huge debt to Jeff Manza and Clem Brooks, who established the research templates used in this work and to
Benjamin Moodie, who coauthored the 2006 update. In addition, they thank the participants of the 2005
Stanford-Berkeley Inequality Colloquium, especially Annette Lareau, for helpful discussion.
1038 • Part IX: The Consequences of Inequality

through the early 1970s showed a strong, if vari- among education, occupation, and income (r ≈
able, relationship between class (as indicated by 0.4) mean many people can easily justify more
occupation) and voters’ choices of candidates than one answer (Hout 2006). So the broader
and parties (Lipset 1981; Alford 1963; Rose appeal to middle-class values and interests has
1974). Lipset and Rokkan’s (1967) influential been effective for Democrats in the last twenty
synthesis of the empirical record pointed to two years. Rhetoric aside, the working class is too
nineteenth-century revolutions, the national small in postindustrial America to be an effective
revolution and the Industrial Revolution, which political base; even winning every working-class
they argued initiated processes of social differen- vote would not add up to victory (Manza and
tiation and conflict everywhere. As democracy Brooks 1999).
spread, the four axes of social differentiation— Unions used to bind blue-collar Americans to
(1) churches versus the secular state, (2) domi- a Democratic Party that mimicked labor party
nant versus subject subcultures, (3) rural areas features. From 1936 to 1968 (and again in 1984)
versus cities, and (4) employers versus workers— Democrats gave unions an important voice in
came to influence voting behavior. The details the selection of candidates. In exchange, the
varied from country to country, depending on unions delivered their share of the working-class
history, the presence and influence of churches vote to the Democrats. That link was snapped
and minority subcultures, and timing. But ev- by electoral reforms in the 1970s. The popular
erywhere the advance of industrialization pushed party primary system of selecting candidates
the first three axes to the side and the fourth— moved politics from back rooms to TV screens,
class—to the fore. and unions lost out. At the same time, they were
In many Western European nations (and losing membership and the trust of members
Australia) the cleavage structure was “frozen” in and nonmembers alike (Lipset and Schneider
the dominance of class-based political parties 1983; Hout, Manza, and Brooks 1999). In the
(see Rose and Urwin 1970; Mair 1999; Lipset age of Jimmy Hoffa, identification with unions
and Rokkan 1967). Where this happened, it became more a political liability than an asset,
makes sense to speak of “class politics” (Mair and unions never regained the clout they once
1999). Elsewhere the association between class had at the ballot box. Union members still lean
and voting behavior is better thought of as “class strongly toward Democratic candidates (Hout,
voting.” Without a party structure that freezes Manza, and Brooks 1999), but the heyday of
both classes and voters in place, we face the identification between the Democrats and the
prospect of shifting alliances, new coalitions, labor movement is over. Unions are not exclu-
and realignment. In the United States the lack sively working-class anymore, either. The teach-
of explicit links between classes and parties frees ers’ unions have been essential to Democrats at
voters to weigh each party’s appeal and frees each least since Jimmy Carter’s campaign in 1976.
party to revise its strategies from time to time. As unions declined, Republican appeals to a
This process has been explicit in US presiden- “right to work” and less industrial regulation at-
tial elections since 1992. First Bill Clinton and tracted blue-collar voters. From Reagan to Rom-
then John Kerry dropped the Democrats’ famil- ney, Republican candidates’ promises of relief
iar appeals to “working people” in favor of di- from regulations and taxes appealed to a grow-
rect calls to the middle class. Barack Obama has ing minority of blue-collar workers as well as to
maintained that emphasis, even while touting managers and the self-employed. In particular,
the revival of auto manufacturing in Michigan lower taxes and less regulation made the greatest
and Ohio. Most American blue-collar workers inroads among small businesspeople and man-
still identify with the working class when sur- ual workers who were not in unions. To these
veyed, but many Americans have trouble de- people, Reagan’s firing of striking air-traffic con-
scribing their class position; modest correlations trollers in 1981 may have seemed like a dramatic
 Hout and Laurison—The Realignment of U.S. Presidential Voting • 1039

rebuke to special pleaders who wanted to dodge politics, whereas the traditional approach mis-
the apparently inexorable economic changes that cast the same trends as dealignment (see Hout,
everyone else was facing. If we are reading these Manza, and Brooks 1999).
signs correctly, the persistence of union member- Another advantage of the total class voting
ship as a factor in American politics is as likely approach is that it incorporates the class differ-
to be a reflection of rightward tilt among people entials in turnout as well as class differences in
without union protection as a reflection of left- votes. Middle-class Americans are more likely
ward stance among those with it. to vote than working- or lower-class Americans
Following Hout, Brooks, and Manza (1995), (Schlozman, Verba, and Brady 2012). The class
we distinguish between “traditional” and “total” skew in participation is likely to affect the party
class voting and use statistical models appropri- system and public policy (Piven and Cloward
ate to each. Traditional class voting hinges on 1986; Rosenthal 2004). Setting it aside, as the
the correspondence between the working class traditional class voting approach does, leaves out
and parties of the Left and the middle classes an important element of class politics per se. Be-
and parties of the Center or Right. This corre- cause of space limitations, we cannot explore this
spondence can be embedded in statistical models advantage in this chapter; it is sufficient to note
of class voting that identify a “natural” party for that our turnout equation shows no significant
each class (Weakliem 1995). A slightly weaker change in class differences in turnout since the
version of the traditional approach is to array 1992 election; the patterns Hout, Brooks, and
classes and parties as ordered points on latent Manza (1995) describe for 1992 are still evident
continua and examine the strength of associa- in the data on the 2008 election.
tion between the two latent variables (Weakliem Our distinction between traditional and total
and Heath 1999). However the researcher ap- class voting is related to Mair’s (1999) distinction
proaches the data, the traditional model is really between “class politics” and “class voting.” Ac-
only appropriate for understanding the histor- cording to Mair, class politics signifies the kind of
ically significant but specific identification of institutionalized class alliances that we call tradi-
classes and parties. tional. Class voting, for Mair, signifies a tendency
Total class voting, on the other hand, is not for classes to ally themselves with parties in a
tied to any specific correspondence of classes more ad hoc endorsement for a specific election.
and parties. This fact is useful and informative The deal, far from being institutionalized in party
when traditional class voting alliances may not structure, must be renegotiated for each election
be the strongest or most evident. For example, cycle. Total class voting, as we have defined it,
the recent emergence of affinities between Dem- only requires class voting, not class politics.
ocrats and teachers may be as strong and signifi- Our class voting perspective has supplanted
cant as the former links between Democrats and the traditional class politics approach since
blue-collar workers were. Throw in trial lawyers, 1995. Most contemporary research assumes the
college professors, nurses, and half of the doc- class voting perspective and treats traditional
tors, and the Democrats’ courting of profession- class politics as a special case, if at all (Jansen,
als yields more votes than traditional appeals Evans, and de Graaf 2012).
to blue-collar workers could. A traditional ap-
proach would miss this kind of class voting; a
total approach can capture it. As long as tradi- Data and Methods
tional class voting is but one of several patterns The data used are from the General Social
of association between classes and parties, the Survey (GSS), a succession of cross-sectional
total approach is to be preferred for its flexibil- surveys dating from 1972 (Smith, Marsden,
ity. In previous work the total approach allowed and Hout 2011). The GSS is a full-probability
us to see the class realignment in US presidential sample of households representative of the
1040 • Part IX: The Consequences of Inequality

English-speaking adult US population through for i = 1, . . . , N indexes persons; j = 1, . . .

2002 and the English- or Spanish-speaking adult , 4 indexes outcomes [1 = Republican, 2 =
population since. Sample sizes range from 4,100 Democrat, 3 = other, 4 = did not vote]; T1 . . .
to 6,700 respondents per election, except for T10 are dummy variables for elections, X1 . . .
2008, which has a sample size of 1,882. X6 are dummy variables for classes, Z1 . . .
The dependent variable for our analysis is ZP are continuous or dummy variables for
self-reported vote for president in the most re- other independent variables, ȕ0 . . . ȕP +21,
cent election, recoded to four categories: (1) the Ȗ1 . . . Ȗ6, and į1,1 . . . į10,P are parameters to
Democratic candidate, (2) the Republican can- be estimated. For identification, we constrain
didate, (3) some other candidate, and (4) did not ∑ 10k =1 β k = ∑ 16k =11β k = ∑ 6k =1 ∑ 10k '=1γ kk ' = Ø . For
vote. Self-reports exaggerate turnout but repro- comparison, we also fit two logit regression
duce the partisan split very well (e.g., Abramson, models in which the dependent variables were
Aldrich, and Rhode 1994). Democrat versus any other vote and Republi-
We include election, class, family income, can versus any other vote (excluding nonvoters
education, gender, racial ancestry, age, region, as missing data). Because the two logit models
and religion as independent variables. Income implied the same substantive conclusions as the
(recoded to constant dollars and logged), edu- more comprehensive multinomial model, we
cation, and attendance at religious services are only report multinomial results here. For statisti-
treated as scores; the other variables are coded in cal efficiency, we dropped interaction effects that
conventional categories, using dummy variables.1 failed to reach the conventional significance level
The class scheme used by Hout, Brooks, (p < .05) from further calculations after checking
and Manza (1995) is applied; homemakers and to see if a simpler set of time contrasts might
retirees are assigned the class of their most re- be significant. For income that proved to be the
cent occupation.2 The six substantive classes case, so we estimated only three income effects:
are (1) professionals, (2) managers, (3) routine the main effect, an interaction for the 1980s, and
white-collar workers (mostly clerical, sales, and a second interaction for elections in the 1990s
white-collar service workers), (4) self-employed or 2000s.
(other than professional and blue collar), (5)
skilled blue-collar workers, and (6) less-skilled
blue-collar workers (including manual service The Realignment of Class Voting
workers). By Barack Obama’s election in 2008, profession-
We test for changes across elections in the als were supporting Democrats by a significant
association between vote and the independent margin, the self-employed were supporting Re-
variables by generating interaction effects be- publicans by a nearly identical margin, and man-
tween the election and each of the independent agers were almost as strongly Republican as the
variables. Following Hout, Brooks, and Manza self-employed were. White-collar and blue-collar
(1995), we fit the following model as a multino- workers clustered in the middle, giving Obama
mial logistic regression: an edge over McCain within sampling error of
his edge among the electorate as a whole.
⎛ pij ⎞
yij = ln ⎜ ⎟ That class voting pattern was substantially
⎝ pi1 ⎠ different from the one in the 1970s, when it was
14 20 P + 20
= β0 + ∑ β k Tki + ∑ β k X ( k −14) i + ∑β Z k ( k − 20) i
blue-collar workers who were strong Democrats
k =1 k =15 k = 21
6 14 6 14
and professionals were in the middle, and even
+ ∑ ∑ γ kk ' X kiTk ' i + ∑ ∑ δ k,k ' Z kiTk ' i more different from the 1950s, when profession-
k =1 k ' =1 k =1 k ' = 1
als were strong Republicans (Hout, Brooks, and
Manza 1995). The traditional pattern of class
voting held in US presidential elections from
 Hout and Laurison—The Realignment of U.S. Presidential Voting • 1041

Figure 119.1. Net Correlation of Class with Partisan Choice by Election Year, Class, and Model: Persons 25 Years of
Age and Older, 1972–2008

Professionals Managers Self-employed

Net e
ect of class on voting for the Democrat

0.50

0.25

0.00

-0.25

-0.50

1972 1980 1988 1996 2004 2012 1972 1980 1988 1996 2004 2012 1972 1980 1988 1996 2004 2012

White collar Skilled Unskilled

0.50

0.25

0.00

-0.25

-0.50

1972 1980 1988 1996 2004 2012 1972 1980 1988 1996 2004 2012 1972 1980 1988 1996 2004 2012

Election year

Note: Points in these graphs represent multinomial logistic regression coefficients, normed to sum to zero within election year. The four
sloping lines show the simple regression of those coefficients on time; the two flat lines show the mean of the coefficients. The models
control for election, family income, education, gender, racial ancestry, age, region, religion, and statistically significant changes in the
correlations between vote and those factors.
Source: Authors’ calculations from General Social Surveys, 1973–2010.

1948 to 1960, then gave way to one dominated The most significant feature of these trends
by a cleavage separating business and the pro- for political sociology is the reversed role of
fessions, with white-collar and blue-collar voters the professions and the working class since
up for grabs. Figure 119.1 summarizes the re- the 1960s. From the 1960s to the 1990s pro-
alignment of class voting across the presidential fessionals switched from being the mainstay of
elections covered by the GSS. The dots show the Republican coalition to being that of the
our statistical estimates in each election, and the Democratic coalition; from the 1960s to the
lines emphasize the long-term trend. 2000s skilled and unskilled blue-collar workers
Figure 119.1 displays all these trends. Each went from being the Democratic mainstay to
data point in the figure is the value of ȕk + Ȗkk' splitting between Democrats and Republicans.
for all possible combinations of k = 1, . . . , 6 and Brooks and Manza (1997) and Brooks (2000)
kl = 1, . . . , 10, divided into six panels (one for provide convincing evidence that professionals’
each value of k'). The lines in most panels were fit rising commitments to civil rights for blacks
by ordinary least squares. For managers and the (and to some extent for other groups) prompted
self-employed, the lines were very nearly flat, so their partisan shift. Professionals, when picking
we simply drew horizontal lines at the mean ȕk + candidates, began to give greater weight to is-
Ȗkk in those two panels. sues of equal opportunity than they had in the
1042 • Part IX: The Consequences of Inequality

past, even bypassing their personal economic in- 1972 to 1992, but has leveled off since then. The
terests to do so. Working-class disaffection with dots in Figure 119.2 show that there are sub-
federal spending (especially welfare spending), stantial fluctuations, but recalculating kappa for
a preference for a tougher approach in criminal the data with linear trends shows twenty years
justice, and support for the military are possi- at roughly 0.20 after a 25 percent decline from
ble explanations for the blue-collar shift to the 1972 to 1992.
middle (Manza and Brooks 2012). Parties mim-
icked these preferences and adapted their appeals
to the new reality (Jansen, Evans, and de Graaf Income and Voting
2012). Most discussion of class and voting in political
In one sense the triumph of civil rights (and sociology focuses on the role of occupation in
civil liberties and the environment) over personal defining social class. This reflects the sociolo-
material interests is a point in favor of postma- gist’s proclivity to privilege the source over the
terialist accounts of electoral change. But the amount of income when considering class and
postmaterialists assert that class no longer cor- class-based behavior and attitudes. Yet as Bartels
relates with vote. The correlation is evident here; (2008) and others have shown, family income
it just departs significantly from the traditional has long been a significant factor in American
middle-class versus working-class cleavage and political behavior; it became even more impor-
rhetoric. Here we see that a class moved from tant during the Reagan era (roughly the 1980s)
one party to another in response to these rising and has been so since then (see also Greeley and
concerns and did so as a class. Hout 2006, 50–54).
Less challenging to political sociology is the Figure 119.3 quantifies the trend toward a
way “traditional,” that is, self-interested, class steeper income cleavage in American politics.
voting appears in much of the rest of the data. The proportion voting Democratic (adjusted for
Managers voted in recent elections more or less other correlates) decreased as income increased
as they did in the 1970s. They changed their at- in each of the three political eras shown, but it
titudes toward civil rights throughout the 1970s, dropped significantly more in the last five elec-
but unlike professionals, their changing atti- tions than in earlier ones. All else being equal, we
tudes failed to influence their votes (Brooks and expect roughly 60 percent of voters from families
Manza 1997). Personal and class interest in cut- with incomes around $12,000 per year to pick
ting taxes and deregulating the economy proved the Democrats in a very close election, whereas
to be more crucial to how they voted. Similarly, less than 40 percent of those from families
the voting pattern of the self-employed main- with incomes approaching $200,000 per year
tained a steady preference for Republicans. In would do so. That 20 percentage point income
the 1950s the self-employed were evenly divided gap in voting is significantly larger than the 15
between Democrats and Republicans. They percentage point gap in the 1970s. The adjust-
moved strongly toward the Republicans in the ments mean that the income patterns statistically
1960s and were their most reliable supporters control for occupation, education, gender, race,
throughout the 1972–2008 period under obser- age, church attendance, region, and differences
vation here. We suspect that some of the change among elections.
reflects issues and some reflects a shift in compo-
sition; farmers were a more significant share of
the self-employed in 1952 than by 1980. Conclusions
All of these changes might suggest that total American class coalitions realigned after the
class voting may have disappeared. That is not civil rights movement in the 1960s. Voters
the case. The kappa measure, introduced by changed their expectations of presidential pol-
Hout, Brooks, and Manza (1995), declined from itics, and parties changed what they promise.
 Hout and Laurison—The Realignment of U.S. Presidential Voting • 1043

Figure 119.2. Total Class Voting (Kappa) by Election Year: Persons 25 Years of Age and Older, 1972–2008

0.40

0.30
Kappa

0.20

0.10

0.00
1972 1980 1988 1996 2004 2012
Election year
Note: Points show calculation based on points in Figure 119.1; line shows calculation based on lines in Figure 119.1.
Source: Authors’ calculations from coefficients in Figure 119.1.

The Republican platform of lower taxes and American politics is a general trend that extends
deregulation deepened the loyalty of managers to other Western democracies or another in-
and won over the half of the self-employed who stance of “American exceptionalism.” We note
used to back the Democrats. Blue-collar workers that English and German parties have started
also prefer low taxes. Republicans courted and to adopt some American-style appeals. In Brit-
won their votes in most elections after 1972. ain “New Labour’s” platform is known as “the
But Democrats countered with strong plays for third way”; Schröder moved the German Social
professionals’ interests in many kinds of social Democrats in that direction, also. The statisti-
spending and their commitment to civil rights. cal evidence makes us cautious. In Tony Blair’s
“New Democrats”—most notably Clinton—ap- first win, he drew proportionally higher sup-
pealed directly to middle-class voters and shored port in each class, leading to no change in the
up support among professionals by balancing the association between class and vote in Britain
budget in the 1990s. On net, the Republicans (Goldthorpe 1999; Brooks, Nieuwbeerta, and
gained and the Democrats lost among employed Manza 2006).
people. But Democrats’ defense of Social Secu- The 2012 election probably continued
rity and Medicare allowed them to gain support these trends. Discussion of the so-called 1 per-
among retirees and pensioners, who are out cent, the video of Romney dismissing the “47
of the labor force and thus outside this analy- percent,” and Obama’s promise to extend the
sis. These changes helped balance the political Bush tax cuts only for families making less than
arithmetic. $250,000 made class issues the most salient ele-
We do not have enough evidence to deter- ment of presidential politics in recent memory.
mine whether the class voting realignment in Data published by the New York Times show
1044 • Part IX: The Consequences of Inequality

Figure 119.3. Relationship Between Family Income and Partisan Choice by Election: Persons 25 Years of
Age and Older, 1972–2008

65

60
Democrat (%)

55

50

45 Elections
1972-1976
1980-1988
40 1992-2008

12.5 25 50 100 200

Family income ($000s)
Note: Incomes adjusted for inflation and shown on ratio scale; that is, each value displayed is double the previous one.
Source: Authors’ calculations from General Social Surveys, 1972–2010.

Alford, R. 1963. Party and Society: The

that 62 percent of voters whose incomes fell Anglo-American Democracies. Westport, CT:
below $30,000 voted for Obama, compared to Greenwood Press.
44 percent of voters whose incomes exceeded Bartels, L. M. 2008. Unequal Democracy. Princeton,
$100,000—an eighteen-point gap in line with NJ: Princeton University Press.
our estimates of the class voting gap (no data on Brooks, C. 2000. “Civil Rights Liberalism and the
Suppression of a Republican Political Realign-
occupations are available yet).
ment in the United States, 1972–1996.” Amer-
ican Sociological Review 65: 483–505.
NOTES
Brooks, C., and J. Manza. 1997. “Class Politics and
1. The age categories are less than twenty-five
Political Change in the United States, 1952–
years, twenty-five to thirty-four, thirty-five to forty-
1992.” Social Forces 76: 397–408.
four, forty-five to fifty-four, fifty-five to sixty-four,
Brooks, C., P. Nieuwbeerta, and J. Manza.
sixty-five to seventy-four, and seventy-five years and
2006. “Cleavage-based Voting Behavior in
over.
Cross-national Perspective: Evidence from Six
2. People with no labor force experience and
Postwar Democracies. Social Science Research 35:
those whose occupational report could not be coded
88–128.
are excluded.
Clark, T. N., and S. M. Lipset. 1991. “Are So-
cial Classes Dying?” International Sociology 6:
REFERENCES
397–410.
Abramson, P. R., J. H. Aldrich, and D. W. Rhode.
Goldthorpe, J. H. 1999. “The Persistence of Class
1994. Change and Continuity in the 1992 Elec-
Voting in Recent British Elections.” In The End
tions. Washington, DC: Congressional Quarterly
of Class Politics?, edited by Geoffrey Evans, 63–
Press.
82. Oxford: Oxford University Press.
 Hout and Laurison—The Realignment of U.S. Presidential Voting • 1045

Greeley, A. M., and M. Hout. 2006. The Truth About Mair, P. 1999. “Critical Commentary.” In The End
Conservative Christians. Chicago: University of of Class Politics?, edited by Geoffrey Evans, 308–
Chicago Press. 312. Oxford: Oxford University Press.
Hout, M. 2006. “How Class Works: Subjective As- Manza, J., and C. Brooks. 1999. “Group Size, Turn-
pects of Social Class Since the 1970s.” In Social out, and Alignments in the Making of U.S. Party
Class: How Does It Work?, edited by A. Lareau Coalitions, 1960–1992.” European Sociological
and D. Conley, 25–64. New York: Russell Sage Review 15: 369–390.
Foundation. ———. 2012. “How Sociology Lost Public Opin-
Hout, M., C. Brooks, and J. Manza. 1995. “The ion: A Genealogy of a Missing Concept in the
Democratic Class Struggle in U.S. Presidential Study of the Political.” Sociological Theory 30:
Elections.” American Sociological Review 60: 89–113.
805–828. Piven, F. F., and R. A. Cloward. 1986. Why Ameri-
Hout, M., J. Manza, and C. Brooks. 1999. “Classes, cans Don’t Vote. New York: Pantheon.
Unions, and the Realignment of U.S. Presiden- Rose, R., ed. 1974. Electoral Behavior: A Comparative
tial Voting, 1952–1992.” In The End of Class Handbook. New York: Free Press.
Politics?, edited by G. Evans, 83–95. Oxford: Rose, R., and D. Urwin. 1970. “Persistence and
Oxford University Press. Change in Western Party Systems Since 1945.”
Inglehart, R. 1977. The Silent Revolution: Changing Political Studies 18: 287–319.
Values and Political Styles Among Western Publics. Rosenthal, H. 2004. “Politics, Public Policy, and In-
Princeton, NJ: Princeton University Press. equality: A Look Back at the 20th Century.” In
Jansen, Giedo, Geoffrey Evans, and Nan Dirk de- Social Inequality, edited by Kathryn Neckerman.
Graaf. 2012. “Class Voting and Left–Right Party New York: Russell Sage Foundation.
Positions: A Comparative Study of 15 Western Schlozman, K. L., S. Verba, and H. E. Brady. 2012.
Democracies, 1960–2005.” Social Science Re- The Unheavenly Chorus: Unequal Political Voice
search 42: 376–400. and the Broken Promise of American Democracy.
Lipset, S. M. 1981. Political Man: Second, Expanded Princeton, NJ: Princeton University Press.
Edition. Baltimore, MD: Johns Hopkins Univer- Smith, T. W., P. V. Marsden, and M. Hout. 2011.
sity Press. General Social Survey Cumulative Codebook,
Lipset, S. M., and S. Rokkan. 1967. “Cleavage Struc- 1972–2010. Storrs CT: Roper Center.
tures, Party Systems, and Voter Alignments.” In Weakliem, D. 1995. “Two Models of Class Voting.”
Party Systems and Voter Alignments, edited by S. British Journal of Political Science 25: 254–270.
M. Lipset and S. Rokkan, 1–64. New York: Free Weakliem, D., and A. F. Heath. 1999. “The Secret
Press. Life of Class Voting: Britain, France, and the
Lipset, S. M., and W. S. Schneider. 1983. The Confi- United States Since the 1930s.” In The End of
dence Gap: Business, Labor, and Government in the Class Politics?, edited by Geoffrey Evans, 97–136.
Public Mind. New York: Free Press. Oxford: Oxford University Press.
1046 • Part IX: The Consequences of Inequality

Health

120 • Johannes Siegrist and Michael Marmot

Health Inequalities and the Psychosocial

Environment
Two Scientific Challenges

Introduction and an increase in health care spending (Marmot

At the turn of the 21st century, social inequali- & McDowall, 1986).
ties in health continue to be a key public health Importantly, there is a social gradient of
problem in advanced societies, including Euro- mortality and morbidity revealed in the Brit-
pean countries. Evidence of strong variations in ish Whitehall studies, among others (Marmot,
life expectancy between and within countries Shipley, & Rose, 1984). Results of the first
has accumulated over the past three or four Whitehall study show that, at younger ages,
decades (Drever & Whitehead, 1997; Elstad, men in the lowest employment grades have a
2000; Fox, 1989; Kunst, 1997; Mackenbach & four times higher mortality rate than men in
Bakker, 2002; Marmot & Wilkinson, 1999). the highest grade. As striking as the difference
With regard to mortality, mean difference in life between the top and bottom is the gradient. Po-
expectancy between those at the top and at the sition in the hierarchy shows a strong correlation
bottom of a society’s social structure (as defined with mortality risk. Men second from the top
by education, income, employment status) are have higher mortality than top-grade civil ser-
anywhere from 4 to 10 years. For instance, in vants; clerical officers have higher mortality rates
Finland, life expectancy at age 35 for men em- than the men above them in hierarchy. A social
ployed in white-collar jobs was 6.9 years higher gradient was seen not only for total mortality,
than that of manual workers in 1995 (Lahelma, but for all the major causes of death, including
2001). Marked socioeconomic differences are coronary heart disease (CHD) and stroke (Mar-
reported in the United States and across Europe mot et al., 1984).
(see, for example, Pappas et al. 1993; Kunst The steepest social gradients in health are
1997). Several epidemiological investigations observed at two stages of the life course: early
observed a widening of social inequalities in childhood and midlife. Less inequality is ob-
health during the final quarter of the last century served in adolescence and in older age (Kuh &
despite considerable progress in medical science Ben Shlomo, 1997). It should be noticed that

Johannes Siegrist and Michael Marmot. “Health Inequalities and the Psychosocial Environment—Two Scien-
tific Challenges,” Social Science & Medicine 58:8 (2004), pp. 1463–1473. Used with permission from Elsevier.
 Siegrist and Marmot—Health Inequalities and the Psychosocial Environment • 1047

the size of these social variations depends on the effects are not confined to the ‘material’ dimen-
measures of social status. Most frequently used sion, but include a ‘psychosocial’ dimension of
measures of social status are employment or oc- adverse environments.
cupational class, education or income. Although A third framework addresses macrosocial de-
interrelated, these measures of social inequality terminants of morbidity and mortality. Research
point to different phenomena: income serves as findings support the notion that powerful de-
an indicator of material resources versus depri- terminants of health are inherent characteristics
vation whereas educational attainment reflects of the society at large, such as a pronounced
different levels of skill and qualification. Oc- inequality of a country’s income distribution,
cupational characteristics cover most relevant a high degree of regional disparities, or a lack
aspects of socioeconomic inequalities, but their of social coherence and poor social capital
application is limited to employed populations (Diez-Roux, 1998; Kawachi, Kennedy, Lochner,
(Fox, 1989; Lahelma, 2001; Marmot & Wilkin- & Prothrow-Stith, 1997; Wilkinson, 1996).
son, 1999). This chapter deals with the second explan-
Previous research has posited several expla- atory framework, social variations of health in
nations for these empirical findings. The first midlife. Its focus is on the study of adverse ef-
explanation concerns life course influences on fects on health produced by an unfavourable
health. The central hypothesis states that cumu- psychosocial environment.
lative, differential exposure to health damaging
or health promoting environments is the main
determinant of the observed variations. Partic- Defining the Psychosocial
ular emphasis is put on early life influences as Environment with Relevance
there is increasing evidence of the long-term to Health
health effects of the foetal and early child en- We underline two general (and perhaps univer-
vironment (Barker, 1995; Power & Hertzman, sal) patterns of human motivations in their re-
1997; Wadsworth, 1999). Moreover, parents’ lation to the social environment: first, the need
deprived social conditions are transmitted to for physical and mental well-being as a prereq-
offspring’s life chances (e.g. educational attain- uisite for the organism’s reproduction and the
ment) via primary socialization. As a result, individual’s productivity, and secondly, the need
offspring’s exposure to unfavourable social cir- of experiencing a positive self. Positive experi-
cumstances during adolescence and early adult- ence of self, while closely linked to well being,
hood increases the risk of disease during midlife is contingent on a social environment that pro-
(Kuh & Ben Shlomo, 1997). vides opportunities of belonging, acting or con-
Social variations in midlife are the focus of tributing and of receiving favourable feedback.
a second explanatory framework. More specif- Conversely, a social structure that excludes indi-
ically, health adverse behaviours of individuals viduals from belonging, acting or contributing
or groups and negative emotions may result in and that prevents them from receiving favour-
enhanced stress responses. These behaviours and able feedback diminishes or even destroys posi-
emotions are analysed in the context of adverse tive experience of self.
living and working conditions. It has been found We propose to define the term ‘psychosocial
that midlife is the period of life, after the first environment’ as the sociostructural range of
year of life, during which social inequalities in opportunities that is available to an individual
mortality manifest themselves most strongly person to meet his or her needs of well being,
(Marmot, Shipley, Brunner, & Hemingway, productivity and positive self-experience. Two as-
2001). In addition, employment status and pects of positive self-experience are of particular
quality of work exert strongest effects on health importance for well being and health: self-efficacy
during this period. As will be argued, these and self-esteem. Self-efficacy has been defined as
1048 • Part IX: The Consequences of Inequality

the belief a person has in his or her ability to reciprocated (Siegrist, 2000). Feelings of being
accomplish tasks (Bandura, 1985). This belief is excluded or of not getting anywhere despite
based on a favourable evaluation of one’s compe- one’s efforts may become generalized and trigger
tence and of expected outcomes. A psychosocial a sense of being locked in an unrewarding psy-
environment conducive to self-efficacy enables chosocial environment (see below).
the person to practice his or her skills, to experi- In summary, we maintain that the psychoso-
ence control in terms of successful agency. At the cial environment serves as a concept that bridges
psychological level, self-efficacy induces feelings the social opportunity structure with the indi-
of mastery (Pearlin & Schooler, 1978) and con- vidual’s strong need of favourable self-efficacy
trol beliefs (Rotter, 1966). At the psychobiologi- and self-esteem. Furthermore, we maintain that
cal level, control of aversive stimuli and positive health and well being are to a considerable extent
outcome expectancy are considered crucial trig- influenced by the quality and intensity of these
gers of activation of the hypothalamic-pituitary processes.
adrenocortical stress axis and its long-term health Research findings that aim at explaining part
effects (Henry & Stephens, 1977; Levine & of the social gradient of disease in midlife by
Ursin, 1991; McEwen, 1998; Steptoe & Appels, a mediating or modifying role of indicators of
1989). It is therefore assumed that a psychosocial adverse psychosocial environment raised some
environment that offers options of experiencing criticism. For instance, Macleod et al. (2001)
self-efficacy produces favourable effects on health claimed that these psychosocial predictors may
and well being whereas opposite effects are ex- operate as indicators of some unmeasured con-
pected in individuals who are confined to restric- dition of low socioeconomic position that ac-
tive, control-limiting psychosocial environments counts for an increased risk of illness (residual
(see below). confounding). In particular, these authors point
A similar relationship is assumed with respect to unfavourable conditions during each child-
to self-esteem, i.e. the continued positive expe- hood and adverse material living conditions.
rience of a person’s self-worth. A psychosocial Yet, several epidemiological studies have
environment conducive to self-esteem enables shown that adjustment for socioeconomic con-
the person to connect him- or herself with sig- ditions in early life (e.g. height, father’s socio-
nificant others and to receive appropriate feed- economic position, regional disparity) does not
back for well accomplished tasks. Self-esteem remove the association between measures of an
strengthens feelings of belonging, approval and adverse psychosocial work environment and car-
success. At the psychobiological level, there is diovascular disease (Kivimäki et al., 2002; Mar-
now evidence on a powerful regulatory role of mot et al., 2001).
the mesolimbic dopamine system, a brain system Convincing evidence of an independent role
implicated in motivation, reinforcement and re- of psychosocial constructs in explaining social
ward in personal and interpersonal well-being inequalities in health is provided by laboratory
(Blum, Cull, Braverman, & Comings, 1996; investigations that are conducted in the frame of
Rolls, 2000). Reward-sensitive neuronal struc- socioepidemiologic studies (see e.g. Kunz-Ebrecht
tures in the orbitofrontal cortex modulate the et al., 2003). For instance, lower occupational
quality and intensity of emotions associated groups showed more prolonged blood pressure
with the experience, expectancy or frustration of and clotting factor stress responses and higher
reward. If a psychosocial environment prevents concentrations of inflammatory markers in com-
the person from experiences of belonging and parison with higher status participants when
approval, recurrent feelings of disappointment exposed to a standardized psychosocial stressor
and frustration emerge. Reward deficiency is (Steptoe et al., 2002, 2003). Furthermore, pro-
particularly stressful if invested efforts are not longed blood pressure elevations and haemostatic
 Siegrist and Marmot—Health Inequalities and the Psychosocial Environment • 1049

risk factors were shown to be associated with low Henry, J., & Stephens, P. (1977). Stress, health and
control at work (Schnall et al., 2000; Steptoe et the social environment: A sociobiologic approach to
al., 2003) and with components of the effort–re- medicine. New York: Springer.
ward imbalance model (Vrijkotte, van Doornen, Kawachi, I., Kennedy, B. P., Lochner, K., &
Prothrow-Stith, D. (1997). Social capital, in-
& de Geus, 1999, 2000). Most recently, with
come inequality and mortality. American Journal
the identification of downstream signals convert- of Public Health, 87(9), 1491–1498.
ing psychosocial stress into cellular dysfunction, Kitagawa, E. M., & Hauser, P. M. (1973). Differ-
a new dimension of psychobiological stress re- ential mortality in the United States: A study in
search with relevance to human health was dis- socio-economic epidemiology. Cambridge, MA:
covered (Bierhaus et al., 2003). Harvard University Press.
In conclusion, in this paper we have sought Kivimäki, M., Leino-Arjas, A. P., Luukkonen, R., Ri-
to make progress in explaining the links between ihimäki, H., Vahtera, J., & Kirjonen, J. (2002).
adverse psychosocial environments within and Work stress and risk of cardiovascular mortality:
Prospective cohort study of industrial employees.
beyond work and health. Future research must
British Medical Journal, 325, 857–860.
work to tackle a scientific challenge, the elucida- Kuh, D. L. J., & Ben Shlomo, Y. (Eds.). (1997). A
tion of upstream processes that relate the health life course approach to chronic disease epidemiology.
effects of adverse psychosocial environments to a Oxford: Oxford University Press.
society’s structure of social inequalities. Kunst, A. E. (1997). Cross-national comparisons of
socio-economic differences in mortality. Ph.D. The-
REFERENCES sis, Erasmus University, Rotterdam.
Bandura, A. (1985). Social foundations of thought and Kunz-Ebrecht, S. R., Kirschbaum, C., & Steptoe,
action. New Jersey: Englewood Cliffs. A. (2003). Work stress, socioeconomic status,
Barker, D. J. (1995). Fetal origins of coronary heart and neuroendocrine activation over the working
disease. British Medical Journal, 311, 171–174. day. Social Science & Medicine, this issue (doi:
Bierhaus, A., Wolf, J., Andrassy, M., Rohleder, N., 10,1016/S0277–9536(03)00347–2).
Humpert, P. M., Petrov, D., Ferstl, R., Eynat- Lahelma, E. (2001). Health and social stratification.
ten, M. v., Wendt, T., Rudofsky, G., Joswig, In W. C. Cockerham (Ed.), The Blackwell com-
M., Morcos, M., Schwaninger, M., McEwen, panion to medical sociology (pp. 64–93). Oxford:
B., Kirschbaum, C., & Nawroth, P. P. (2003). Blackwell Publishers.
A mechanism converting psychosocial stress into Levine, S., & Ursin, H. (1991). What is stress? In M.
mononuclear cell activation. Proceedings of the R. Brown, C. Rivier, & G. Koob (Eds.). Stress,
National Academy of Science, 100(4), 1920–1925. neurobiology and neuroendocrinology (pp. 3–21).
Blum, K., Cull, J.G., Braverman, E. R., & Com- New York: Marcel Decker.
ings, D. E. (1996). Reward deficiency syndrome: Mackenbach, J., & Bakker, M. (Eds.). (2002). Re-
Addictive, impulsive and compulsive disor- ducing inequalities in health. A European perspec-
ders—including alcoholism, attention-deficit tive. London: Routledge.
disorders, drug abuse and food bingeing—may Macleod, J., Davey Smith, G., Heslop, P., Met-
have common genetic basis. American Scientist, calfe, C., Carroll, D., & Hart, C. (2001). Are
84, 132–145. the effects of psychosocial exposures attributable
Diez-Roux, A. (1998). Bringing context back to to confounding? Evidence from a prospective
epidemiology: Variables and fallacies in multi- observational study on psychological stress and
level analysis. American Journal of Public Health, mortality. Journal of Epidemiology and Commu-
88(2), 216–222. nity Health, 55, 878–884.
Drever, F., & Whitehead, M. (Eds.). (1997). Health Marmot, M. G., & McDowall, M. E. (1986). Mor-
inequalities. London: The Stationary Office. tality decline and widening social inequalities.
Elstad, J. I. (2000). Social inequalities in health and Lancet, 2, 274–276.
their explanations. Rapport 9/00. Oslo: NOVA. Marmot, M., Shipley, M., Brunner, E., & Hem-
Fox, J. (Ed.). (1989). Health inequalities in European ingway, H. (2001). Relative contribution of
countries. Aldershot: Gower. early life and adult socioeconomic factors to
1050 • Part IX: The Consequences of Inequality

adult morbidity in the Whitehall II study. Jour- Siegrist, J. (2000). Place, social exchange and health:
nal of Epidemiology and Community Health, 55, Proposed sociological framework. Social Science
301–307. & Medicine, 51, 1283–1293.
Marmot, M., & Wilkinson, R. (Eds.). (1999). Social Steptoe, A., & Appels, A. (Eds.). (1989). Stress, per-
determinants of health. Oxford: Oxford Univer- sonal control and health. Chichester, UK: Wiley.
sity Press. Steptoe, A., Feldman, P. M., Kunz, S., Owen, N.,
McEwen, B. (1998). Protective and damaging effects Willemsen, G., & Marmot, M. (2002). Stress
of stress mediators. New England Journal of Medi- responsivity and socioeconomic status: A mech-
cine, 338(3), S171–S179. anism for increased cardiovascular disease risk?
Pearlin, L. I., & Schooler, C. (1978). The structure European Heart Journal, 23, 1757–1763.
of coping. Journal of Health and Social Behavior, Vrijkotte, T. G. M., van Doornen, L. J. P., & de
22, 337–356. Geus, E. J. C. (1999). Work stress and metabolic
Power, C., & Hertzman, C. (1997). Social and bio- and hemostatic risk factors. Psychosomatic Medi-
logical pathways linking early life and adult dis- cine, 61, 796–805.
ease. British Medical Bulletin, 53, 210–221. ——— (2000). Effect of work stress on ambulatory
Rolls, E. T. (2000). The orbitofrontal cortex and re- blood pressure, heart rate, and heart rate variabil-
ward. Cerebral Cortex, 10, 284–294. ity. Hypertension, 35, 880–886.
Rotter, J. B. (1966). Generalized expectancies for Wadsworth, M. (1999). Early life. In M. Marmot, &
internal versus external control of reinforcement. R. Wilkinson (Eds.), Social determinants of health
Psychological Monographs, 80, 1–28. (pp. 44–63). Oxford: Oxford University Press.
Schnall, P. L., Belkic, K., Landsbergis, P., & Baker, Wilkinson, R. (1996). Unhealthy societies: The afflic-
D. (Eds.). (2000). The workplace and cardiovas- tions of inequality. London: Routledge.
cular disease. Occupational Medicine: State of the
Art Reviews, 15, 1–334.

121 • Karen Lutfey and Jeremy Freese

The Fundamentals of Fundamental Causality

Weber’s ([1921] 1968) concept of “life chances” and this association has persisted across historical
highlights both the diverse consequences of so- periods in which risk factors and disease profiles
cial standing and their probabilistic character. have changed radically (Link et al. 1998). We use
The most poignant kind of life chance affected ethnographic data to examine the association be-
by socioeconomic standing may also be the most tween SES and adverse health outcomes among
literal: the probability of staying alive or dying. persons with diabetes.
Lower socioeconomic status (SES) is associated We draw specifically on Link and Phelan’s
with worse health and higher mortality rates at (1995; Phelan, Link, and Tehranifar 2010) con-
virtually every age (Robert and House 1994), cept of SES as a “fundamental cause” of health.
The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in the
following publication: Karen Lutfey and Jeremy Freese, “Toward Some Fundamentals of Fundamental Causal-
ity: Socioeconomic Status and Health in the Routine Clinic Visit for Diabetes,” American Journal of Sociology
110:5 (March 2005), pp. 1326–1372, published by the University of Chicago Press. The article printed here
was originally prepared by Karen Lutfey and Jeremy Freese for this fourth edition of Social Stratification: Class,
Race, and Gender in Sociological Perspective, edited by David B. Grusky. Copyright © 2014 by Westview Press.
 Lutfey and Freese—The Fundamentals of Fundamental Causality • 1051

Fundamental causality is not just about what inquiry is simplified. Conditions that probabi-
causes an outcome like ill-health but about what listically affect patients’ capacities for controlling
causes the causes of ill-health. That is, the concept glucose levels can be expected to likewise prob-
is not about the specific proximate mechanisms re- abilistically affect long-term health outcomes.
sponsible for a persistent association, but rather More aggressive treatment regimens—increasing
about the fact that some metamechanism(s) are in complexity and potential glucose control—
responsible for the generation of multiple con- may be employed as patients become more adept
crete mechanisms that reproduce a particular at managing diabetes, whereas concessionary reg-
relationship in different places and times. The imens—decreasing in complexity and effectively
metamechanism provides what we refer to as accepting that patients will have weaker glucose
a durable narrative (Freese and Lutfey 2011) control and greater risk of long-term complica-
about why the SES-health relationship should tions—may be deployed to accommodate any
be robust to changes in health threats and treat- of several sources of resistance to implementing
ments—an explanation of why a similar associa- more effective regimens.
tion would be observed in diverse sociohistorical
contexts.
A fundamental relationship between two Data and Analytic Strategy
variables such as SES and health implies the po- Data are from a yearlong ethnographic study
tential for a massive multiplicity of mechanisms conducted by the first author in 1997–1998.
connecting the two. No individual mechanism The fieldwork sites were two weekly, four-hour
is so dominant that it alone is responsible for the endocrinology clinics at two hospitals that are
bulk of the observed association between SES both part of the same university-based medical
and health. Rather, as proximate causes of health center located in a large, midwestern city. The
change, the standing conjecture is that these will, clinics were selected to provide an optimal con-
on balance, sustain the overall relationship be- trast of the socioeconomic diversity of persons
tween SES and health. Differential resources with diabetes: “Park” Hospital serves a primar-
provide a possible metamechanism (Link and ily white, upper- and middle-class population,
Phelan 1995). The ways resources can influence whereas “County” Hospital has a largely mi-
health are flexible and varied, allowing the possi- nority, working-class, and underinsured clien-
bility that, when one uses ethnographic methods tele. Our data include approximately 250 hours
to consider in a concrete setting how different re- of observation at these clinics and semistruc-
sources might lead to different outcomes, many tured interviews conducted with twenty-five
potential mechanisms will be revealed. practitioners. We pursue systematic connections
between the regimen design we observed in the
clinics and the socioeconomic conditions of pa-
Diabetes tients’ lives. We organize our results around dif-
Diabetes is a major cause of morbidity and mor- ferent senses of the “location” of the mechanism,
tality in the United States, and its prevalence including physical location of the clinic, factors
is increasing dramatically (Centers for Disease inside and outside the clinic, and those internal
Control and Prevention 2011). Diabetes inci- and external to individual patients (see Figure
dence, complications (Booth and Hux 2003; 121.1).
Robbins et al. 2001), and mortality have all been
shown to be related to SES. Because diabetes
complications are known to be linked to average Differences Between Park and County
glucose levels (Diabetes Control and Complica- Clinics
tions Trial Research Group 1993), identifying Because Park clinic serves a much higher SES
potential mechanisms within an ethnographic clientele than the County clinic, we begin by
1052 • Part IX: The Consequences of Inequality

Figure 121.1. “Location” of Mechanisms as an Organizing Device

considering whether differences in the organiza- medical histories and their family and work
tion of the two clinics may affect regimen design. situations. One physician complained that the
Any differences that place County clinic patients low continuity of care at County required physi-
at a systematic, probabilistic disadvantage rela- cians to rely too heavily on objective indicators,
tive to Park clinic patients can be considered po- which though important are limited and even
tential mechanisms by which differential access potentially deceptive without a broader array of
to health care may contribute to preserving the information:
fundamental cause relationship among diabetes
patients. Although several such differences are The outcome [should not be] based on a
discussed in Lutfey and Freese (2005), here we single test. Nor a single blood sugar, nor
focus on one key finding related to Park clinic a single week, nor a single visit, nor a sin-
offering higher continuity of care than County gle anything. . . . The outcome [should be]
clinic does. based on the totality of the things that you
Park clinic is similar to many private medical assess over the period of time that you see the
organizations in that the continuity of care—the patient. The focus that you see in the clinic
extent to which patients see the same providers like we have [County] is that the totality of
over time—is extremely high. Park has two en- the assessment comes down to “What’s your
docrinologists and one nurse practitioner who last Hemoglobin A1c [a test estimating av-
see most patients, and all diabetes education oc- erage glucose levels over the previous three
curs in a separate center housing several full-time months]?” Which doesn’t mean jack.
educational staff. By contrast, County clinic is
usually supervised by four attending physicians, A hemoglobin A1c result can be potentially
but with residents playing a large role. Diabe- misleading, insofar as it can produce results that
tes education occurs through a certified diabetes make it appear the diabetes has been well con-
educator who attends the clinic weekly on a vol- trolled when actually the patient has undergone
unteer basis. extreme highs and lows. In one example from
High continuity of care enables providers our field notes, residents at County were pleased
to become much more familiar with patients’ by one patient with a seemingly reasonable
 Lutfey and Freese—The Fundamentals of Fundamental Causality • 1053

hemoglobin A1c, but additional probing by how much do each of those things vary, how
a suspicious attending physician revealed that sensitive are they to exercise and to diet, how
this patient regularly had severe episodes of hy- compliant are they with their diet, how much
poglycemia because he drank a six-pack of beer does their diet vary from day to day. . . . You
every night. For avoiding risks associated with can address that a lot more effectively with
hypoglycemia, hemoglobin A1c may not provide continuity or follow-up over a longer period
all the information physicians need. High con- of time than to try and just see a patient cold
tinuity of care may thus allow the physicians at and do all that on one visit.
Park to better assemble a “totality” of informa-
tion, which in turn allows them to provide more Under these circumstances, providers in County
helpful treatment recommendations. often favored conservative treatment regimens,
Continuity of care also facilitates open which they regarded as the safest option in the
patient-provider communication, increasing the absence of close follow-up. In the aggregate, be-
likelihood that useful information is obtained cause these regimens afford only weaker glucose
during the visit. Providers in County have much control, they might be expected to increase the risk
less opportunity to personally monitor or follow of long-term complications and thus preserve the
patients they treat. One County physician we fundamental relationship.
interviewed complained about having to make
regimen decisions while knowing virtually noth-
ing about how the patient would behave outside External Constraints on Regimen
the clinic. Another explained: Design
Although we observed two very different diabe-
You have to glean everything from the chart. tes clinics, we do not suggest that all mechanisms
You don’t get a nice letter written to you would be eliminated if the patients we observed
[about a patient’s status], it’s a note on a chart, were treated at the same clinic. On the con-
and depending on how busy the [providers trary, a series of additional phenomena would
who saw the patient last] are in the clinic, preserve the negative relationship between SES
they write a longer or shorter note, usually a and health outcomes. Next we consider occupa-
shorter one because everybody’s pretty busy. tional influences on regimen design (see Lutfey
And that’s sometimes a little bit of a struggle and Freese 2005 for supplementary discussion of
to really get a comprehensive picture of what finances and social networks).
[a patient’s] complication status is and what Apart from having less money, low SES pa-
other doctors who are taking care of them tients might also be more likely to work at jobs
are thinking about how their med[ication]s that are less hospitable to implementing effective
should be adjusted and these kinds of things. plans for managing glucose. Physicians noted
that patients working swing shifts have schedules
Beyond the challenges of interpreting informa- that make it very difficult to implement a regi-
tion from patients’ charts, low continuity of care men that permits tight control, because they are
generates the additional problem of learning sleeping during the day. Manual labor jobs pose
about a patient’s habits and how those habits are additional challenges, because these patients
connected to his or her diabetes management, as often use glucose somewhat intensely and irregu-
another County physician described: larly while working and so are much more likely
to have problems with hypoglycemia. Worse,
I think that’s where continuity of care is im- they are more likely to be in physical danger if
portant, because the more you know some- they do become hypoglycemic, especially if op-
body’s habits—and there’s so many different erating heavy machinery or working alone, and
habits to learn about with diabetes, exercise, may be at risk for losing operating licenses. To
1054 • Part IX: The Consequences of Inequality

accommodate these risks, regimens for these Hospital patients were always seen by either a
people usually favor higher glucose levels (and resident or fellow before the attending physician,
therefore higher risk of complications) to avoid their actual appointments were also much lon-
hypoglycemia but preserve daily functioning. ger. Patients reported differences in flexibility for
People working in white-collar jobs in offices taking time off work for appointments, whether
were seen as better able to maintain regular eat- such time off is paid (and the personal need for
ing and exercising schedules, to be at lower risk such pay), and the availability of a personal ve-
of physical danger, and to be better candidates to hicle for driving to appointments compared to
receive assistance from nearby and knowledge- time-consuming use of public transportation.
able coworkers in the event of hypoglycemia. As Patients were also often viewed as unmo-
one physician attests: tivated if they allowed prescriptions to lapse.
The social program subsidizing medications for
Some patients don’t wanna have more than many County patients required them to fill their
one or two shots a day. For example, if you prescriptions at the hospital pharmacy, whereas
have a truck driver . . . you may be able to patients with private insurance could not only
convince them if they’re local drivers to take use their regular local pharmacies but also call
a shot at bedtime, but they [are] hardly ever in refills beforehand. As one County physician
going to have more than two shots a day and complained:
they are not gonna adjust their insulin dose
while they’re on the road. . . . [By contrast,] If you gave a businessman a prescription that
I’ve got patients that have a predictable life- had to be refilled every month, and he had to
style . . . they go to the office and they come stop what he was doing and go to the store
back and they have their lunch times. They and stand there in front of a pharmacist for
can handle complex regimens much better. thirty minutes, forty minutes, he’d say, “Ei-
ther you give me something that’s appropri-
ate, or I’m firing you as my physician.” And
(Apparent?) Motivation here [at County] we give patients their pre-
Next we consider the influence of perceived scription and say, “Come back every month
patient motivation on regimen design and the and stand here. Come back on the bus and
strong SES character we observed in assessments get your prescriptions filled.” Gimme a break.
of motivation. For low SES patients, the relative If that doesn’t interfere with compliance, I
costs of complying with particular features of treat- don’t know what does.
ment regimens are often greater than for high
SES patients, yet providers often attribute these Patients’ apparent motivation to comply with
differences to psychological differences in patient medical instructions may also be affected by the
motivation. immediacy and transparency of benefits associ-
Showing up for appointments might seem ated with compliant behavior. For example, the
the simplest expectation of all. Practitioners more discretion patients have over their insulin
regarded missed visits as strong indicators that dosage (as in sophisticated regimens), the more
patients were probably not following their treat- incentive there is to maintain a log, because
ment regimens well. For poorer patients, how- that information is critical for determining dos-
ever, the personal costs of making a clinic visit ages. In contrast, patients with basic regimens
often may be higher than for middle-class pa- were directed to check their glucose levels but
tients. In terms of time, County patients usu- do nothing other than write them down for the
ally had to wait sixty to ninety minutes for their doctor. Consequently, with the cost of glucose
appointments, whereas Park patients usually testing approaching $2.00–3.00/day, the point
waited less than ten minutes. Because County of keeping a log may be obscured for patients
 Lutfey and Freese—The Fundamentals of Fundamental Causality • 1055

with simple regimens, especially since—unlike require much broader observations. Obviously
missing a shot of insulin—it has no immediately diabetes is only one kind of chronic disease, and
observable effect on their actual health. chronic diseases are not the only health out-
comes for which an inverse relationship between
SES and health is observed. An eventual goal
Discussion might be a comprehensive schematic that draws
This chapter articulates some mechanisms by connections and contrasts among many condi-
which the inverse relationship between SES and tions, but that is far in the future and will require
health outcomes might be produced among per- many additional studies of specific conditions by
sons with diabetes. Comparing patients in two other investigators.
clinics that serve very different clienteles, we
identify numerous ways in which SES influences REFERENCES
the design or successful implementation of a reg- Booth, Gillian L., and Janet E. Hux. 2003. “Rela-
tionship Between Avoidable Hospitalizations for
imen. That we can specify such a large number of
Diabetes Mellitus and Income Level.” Archives of
candidate mechanisms is consistent with the ani-
Internal Medicine 163: 101–106.
mating idea of fundamental causality: that dura- Centers for Disease Control and Prevention. 2011.
ble relationships between encompassing variables “National Diabetes Fact Sheet: National Esti-
like SES and health may represent an accumu- mates and General Information on Diabetes and
lation of many small, pervasive advantages that Prediabetes in the United States.” Atlanta, GA:
can be expected to be renewed as the particulars U.S. Department of Health and Human Services,
of disease treatment change over time. Centers for Disease Control and Prevention.
We view this compendium of potential mech- The Diabetes Control and Complications Trial Re-
anisms as a contribution to the social epidemiol- search Group. 1993. “The Effect of Intensive
Treatment of Diabetes on the Development and
ogy of diabetes in its own right. An ethnographic
Progression of Long-Term Complications in
approach to fundamental cause relationships also Insulin-Dependent Diabetes Mellitus.” New En-
illustrates the value of describing concrete causal gland Journal of Medicine 329: 977–986.
pathways with a depth of observational detail. Freese, Jeremy, and Karen E. Lutfey. 2011. “Funda-
Apart from generating proposals about specific mental Causality: Challenges of an Animating
causal pathways that might be tested using quan- Concept for Medical Sociology.” In Handbook
titative methods, ethnographic investigations of Medical Sociology, edited by B. Pescosolido, J.
like ours may also expand social epidemiology’s Martin, J. McLeod, and A. Rogers, 67–81. New
understanding of the multiplicity of mechanisms York: Springer.
Link, Bruce G., Mary E. Northridge, Jo C. Phelan,
that can underlie the causal role of single vari-
and Michael L. Ganz. 1998. “Social Epidemiol-
ables such as “continuity of care.”
ogy and the Fundamental Cause Concept: On
We have purposely limited the scope of our the Structuring of Effective Cancer Screens by
inquiry to capitalize on the strengths of our data Socioeconomic Status.” Milbank Quarterly 76
and of ethnographic methods—to articulate the (3): 375–402.
concrete experiences of people with the health Link, Bruce G., and Jo C. Phelan. 1995. “Social
care system in ways that elude large-scale sta- Conditions as Fundamental Causes of Disease.”
tistical analyses. As is common in ethnography, Journal of Health and Social Behavior (Extra
we reap these benefits of depth at the expense of Issue): 80–94.
Lutfey, Karen E., and Jeremy Freese. 2005. “Toward
generalizability. By restricting our study to pa-
Some Fundamentals of Fundamental Causality:
tients whose regimens include insulin and who
Socioeconomic Status and Health in the Routine
are treated in subspecialty clinics, the average Clinic Visit for Diabetes.” American Journal of
case we observed is more serious than the average Sociology 110 (5): 1326–1372.
case among the general population of diabetes Nicolucci, Antonio, Fabrizio Carinci, and Anto-
patients. A fuller treatment of diabetes would nio Ciampi. 1998. “Stratifying Patients at Risk
1056 • Part IX: The Consequences of Inequality

of Diabetic Complications.” Diabetes Care 21: Robbins, J., V. Vaccarino, H. Zhang, and S. Kasl.
1439–1444. 2001. “Socioeconomic Status and Type 2 Dia-
Phelan, Jo C., Bruce G. Link, and Parisa Tehranifar. betes in African American and Non-Hispanic
2010. “Social Conditions as Fundamental Causes White Women and Men: Evidence from the
of Health Inequalities: Theory, Evidence, and Third National Health and Nutrition Examina-
Policy Implications.” Journal of Health and Social tion Survey.” American Journal of Public Health
Behavior 51: S28–S40. 91: 76–84.
Robert, Stephanie A., and James S. House. 1994. Weber, Max. [1921] 1968. Economy and Society.
“Socioeconomic Status and Health Across the Translated and edited by Guenther Roth and
Life Course.” In Aging and Quality of Life, ed- Claus Wittich. New York: Bedminster Press.
ited by R. P. Abeles, H. C. Gift, and M. G. Ory,
253–274. New York: Springer.

122 • Richard Miech, Fred Pampel, Jinyoung Kim, and Richard G. Rogers

Why Do Mortality Disparities Persist?

Inequality is a major cause of death in the United killers in the United States were tuberculosis,
States. In 2007 death rates were 2.7 times as high diarrhea, and pneumonia, which were concen-
for people with less than a high school diploma trated in the lower social strata. A person living a
than for those with at least some college educa- century ago might reasonably have thought that
tion (Xu et al. 2010). Evidence indicates that a substantial reduction in the influence of these
mortality disparities by education, a key measure top killers would also lead to substantial reduc-
of socioeconomic status (SES) (see Mirowsky tion in mortality disparities by socioeconomic
and Ross 2003; Smith 2004; Williams 1990), strata. In the subsequent century, the influence
have widened in recent decades (e.g., Ahmedin et of these top killers diminished substantially, but
al. 2008; Preston and Elo 1995). We expect that mortality disparities did not. The old killers were
in recent decades mortality disparities by educa- replaced by new ones—specifically, cardiovascu-
tion have persisted and may even have widened. lar disease, cancer, and stroke (Omran 2005; Xu
The “fundamental cause” perspective predicts et al. 2010)—which are similarly concentrated
that mortality disparities by education persist be- in the lower socioeconomic strata (Avendano
cause a continual stream of widening and newly et al. 2006; Clark et al. 2009). This perspective
emergent disparities counteracts the effects of predicts that when the United States successfully
any disparities diminishing. Link and Phelan counters the top killers of today, disparities will
(1995) note that one hundred years ago the top shift to the ones that replace them.

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in the
following publication: Richard Miech, Fred Pampel, Jinyoung Kim, and Richard G. Rogers, “The Enduring
Association between Education and Mortality: The Role of Widening and Narrowing Disparities,” American
Sociological Review 76 (December 2011), pp. 913–934, published by Sage Publications. The article printed here
was originally prepared by Richard Miech, Fred Pampel, Jinyoung Kim, and Richard G. Rogers for this fourth
edition of Social Stratification: Class, Race, and Gender in Sociological Perspective, edited by David B. Grusky.
Copyright © 2014 by Westview Press.
 Miech et al.—Why Do Mortality Disparities Persist? • 1057

Embedded within this perspective are at least have completed their education. The upper limit
two empirically testable hypotheses. The first is of age sixty-four is standard practice in mortality
the prediction that mortality disparities widen studies and is a rough proxy for participation in
substantially for some causes of death while the workforce.
narrowing for others. The “fundamental cause” The first main source of data is Vital Sta-
perspective posits that levels of flexible resources, tistics of the United States (CDC n.d.), which
such as money, power, prestige, knowledge, and includes yearly data from all death certificates
beneficial social connections, allow people in the issued in the United States. In 1989—the first
higher social strata to better take advantage of year in which the decedent’s educational attain-
ever-changing developments in health (Link and ment was included on the death certificate—US
Phelan 1995). The health advantage of people Vital Statistics included about 2.1 million re-
in the upper social strata may come from use of cords, and by 2007 the number of deaths had
the latest health technologies (such as uptake of increased to about 2.5 million. For analysis of
statins), a more successful use of conventional mortality by causes of death, we focus on the
health practices long known to improve health period from 1999 to 2007, when all deaths were
(such as avoiding smoking), or better positioning coded to the same classification system, the tenth
to protect against new health threats that enter a revision of the International Statistical Classifica-
population (such as the obesity epidemic). This tion of Diseases and Related Health Problems, or
health advantage persists over time, because ICD-10 (World Health Organization 1992). We
health developments are constantly changing. use the 113 “Selected Causes of Death” (NCHS
When people in the lower social strata eventu- 1999b), which aggregate the 12,420 codes in the
ally close the gap for one outcome, individuals in ICD-10 to 113 major causes. We have excluded
the upper social strata have already moved on to seven “residual” causes that had no clear inter-
improvements in other outcomes. This fluidity pretation (e.g., one excluded code was 111, “all
in disparities among causes of death implies that other diseases”).
substantial widening and narrowing of dispari- The second main source of data for this anal-
ties should occur over time. ysis is the US Census, which provided informa-
The second prediction highlights a key but tion for the mortality rate denominator. We used
widely unrecognized component of the expla- Census data to derive estimates of the total US
nation for the persistence of health disparities population, including those incarcerated or in
by the fundamental cause perspective. For so- nursing homes. All analyses are age standardized
cioeconomic health disparities to endure, they to the population in year 2000, a process that
must continually emerge and widen in the causes controls the potential influence on mortality
of death that are on the rise. These causes are rates of changing age distributions over time.
among the pool of potential outcomes that are
destined to become the major causes of death
in the future. If not present in rising causes of Results
death, major disparities would have disappeared Figure 122.1 presents the trend in all-cause mor-
many decades ago when the top killers of the tality disparities from 1989 to 2007. In accor-
previous era diminished. dance with past trends, the mortality disparity
increased over time, as evidenced by a widening
distance between the mortality rates of people at
Methods the lowest and highest educational levels. Fur-
The analysis uses two main data sets to calcu- thermore, the trends are monotonic, with the
late death rates for adults aged forty to sixty-four mortality rates highest among those with the
by causes of death. Restricting the age group to least education and steadily declining among
those forty and over ensures that most adults will groups with increasing education.
1058 • Part IX: The Consequences of Inequality

Figure 122.1. Trends in US Mortality Levels by Education for Men and Women Aged 40–64, 1989–2007
1390

1190

990
Deaths per 100,000

790

590

390

190
1989 1991 1993 1995 1997 1999 2001 2003 2005 2007
Year

Legend: Less than 12 12 years 13–15 years 16+ years Overall

years educ. educ. educ. educ. trend

Source: CDC (n.d.) and US Census.

Notes: Age standardized to year 2000 US population. In 1999 classification of cause of death changed to the ICD-10 from
the ICD-9. The detailed analyses in this chapter focus on the period 1999–2007; earlier years are presented in this graph
to put the 1999–2007 trends in context.

The analysis next investigates how this overall Additional analysis illustrates that widening
widening disparity was distributed among major disparities were concentrated among causes of
causes of death. We focus on trends in disparities death with increasing mortality rates; likewise,
from 1999 to 2007 because the widening trend narrowing disparities were concentrated among
continued during this time (as indicated in Fig- causes of death with decreasing mortality rates.
ure 122.1) and because all causes of death were Among men 94 percent of the causes of death
coded to the same ICD classification system. with increasing overall mortality rates had wid-
Results indicate substantial heterogeneity in ening disparities. For women the analogous
disparity levels across causes of death over the percentage is 92 percent. More generally, the
study period. For women disparities widened for correlation between change in mortality rate and
76 percent of the causes of death and narrowed in disparity level is high, ranging from .86 for
for the other 24 percent, and for men disparities men to .58 for women.
widened for 65 percent of the causes of death Table 122.1 provides detailed information
and narrowed for the other 35 percent. The on the causes of death. The left half of the table
overall widening in all-cause mortality presented presents the causes of death for which disparities
in Figure 122.1 thus results from a process in widened and narrowed most over the study pe-
which the influence of widening disparities out- riod, revealing that these disparities are driven
weighs that of narrowing disparities. The widen- by a wide variety of health outcomes, not a
ing in all-cause mortality was not the result of a single type. For example, among white women
uniform increase in disparities among all causes the causes of death that contributed most to
of death. the widening of mortality disparities included
 Miech et al.—Why Do Mortality Disparities Persist? • 1059

Figure 122.2. Predicted and Simulated Mortality Trends

White Women White Men

1800
1120 1600

Deaths per 100,000

1400
Deaths per 100,000

920
1200
720 1000

520 800

600
320
400

120 200
1999 2000 2001 2002 2003 2004 2005 2006 2007 1999 2000 2001 2002 2003 2004 2005 2006 2007
Year Year

Legend: < than 12 12 years 13+ years Projected Overall Mortality Rate
years educ. educ. educ If No Widening Disparities

Source: US Vital Statistics and US Census.

Notes: Age standardized to year 2000 US population. The analysis pool excludes causes of death that are residual catego-
ries, and consequently the total number of deaths in these predicted models is less than the total number of deaths in the
observed data.

accidental poisoning (almost entirely overdoses educated groups. In the absence of growing dis-
of prescription and illegal drugs; see Paulozzi, parities, this difference would have diminished
Ballesteros, and Stevens 2006), chronic lower 26 percent for white women (from 829 to 615
respiratory diseases, and malignant neoplasms. deaths per 100,000). Growing disparities had
Figure 122.2 presents predicted and simu- less influence on the difference in mortality
lated mortality trends that help to further gauge rates across the middle and highest education
the influence of changing causes of death. The groups; without growing disparities, this dif-
predicted results depicted with solid lines show ference would have diminished 16 percent for
the same trends of widening disparities displayed white women. A similar but smaller effect exists
in Figure 122.1. Importantly, the simulated re- for men.
sults depicted with dashed lines show what the
group’s mortality rate would have been if no dis-
parities had widened or emerged over the study Discussion
period. To arrive at these estimates, the analysis This study examines the proposition that mor-
considered the predicted rate of disparity growth tality disparities by education persist over time
for each cause of death in the analyses for each because of their ability to continually emerge
demographic group. All causes of death with a and/or widen in new outcomes. We used the
predicted widening disparity by education were “fundamental cause” perspective to develop the
reset to zero growth, so that the level of dispar- predictions that 1) underlying this persistence
ity at the baseline was constant throughout the are substantial widening and narrowing dispari-
study period. Causes of death with predicted ties across differing causes of death, and 2) wid-
constant or declining disparities by education ening disparities should be concentrated among
were left unaltered in the analysis, and the pre- causes of death with increasing mortality rates,
dicted mortality rates by education were then which are the causes of death primed to predom-
reestimated. inate in the future. Our results provide evidence
The simulated results show that disparity to support these two predictions. Substantial
levels today would be substantially smaller if widening and narrowing of disparities indicate
disparities had not emerged or widened, partic- that waxing and waning of disparities among
ularly the disparity between the least and most causes of death is substantial, continuous, and
 1060

Table 122.1. Changes in Disparities and Changes in Overall US Mortality Rate: Top 3 and Bottom 3 Causes
of Death by Sex and Race/Ethnicity, Ages 40–64, 1999–2007
Mortality Top Increasing and Mortality
Top Widening and Narrowing Disparity Rate Decreasing Causes of Death Rate Disparity
Disparities (ICD-10 Codes) Change Change (ICD-10 Codes) Change Change
White Women (n = )
Accidental poisoning
Accidental poisoning (X40–X49)* 12.9 10.9 10.9 12.9
(X40–X49)*
Chronic lower respiratory diseases Suicide by means other
other than bronchitis, emphysema, 7.15 0.082 than firearms (U03, X60, 2.39 1.85
or asthma (J44, J47) X71, X75, X84,Y87.0)
Malignant neoplasms of trachea, Alcoholic liver disease
5.99 -6.12 1.66 3.11
bronchus, and lung (C33–C34) (K70)
Malignant neoplasms of
Emphysema (J43) -0.835 -1.50 trachea, bronchus, and -6.12 5.99
lung (C33–C34)
Chronic ischemic heart disease
Malignant neoplasm of
other than atherosclerotic -3.05 -5.53 -6.99 0.58
breast (C50)
(I20, I25.1–I25.9)
Acute myocardial infarction Acute myocardial infarction
-4.77 -7.97 -7.97 -4.77
(I21–I22) (I21–I22)
Black Women (n = )
Accidental poisoning
HIV (B20–B24) 13.5 1.37 6.45 7.92
(X40–X49)*
Accidental poisoning (X40–X49)* 7.92 6.45 HIV (B20–B24) 1.37 13.5
Malignant neoplasms of trachea,
5.56 -4.84 Septicemia (A40–A41) 1.06 1.69
bronchus, and lung (C33–C34)
Diabetes mellitus (E10–
Alcoholic liver disease (K70) -2.49 -2.14 -8.10 -5.32
E14)
Chronic ischemic heart
disease other than
Diabetes mellitus (E10–E14) -5.32 -8.10 -9.80 -1.67
atherosclerotic (I20,
I25. 1–I25.9)
Acute myocardial infarction Acute myocardial infarction
-9.77 -13.9 -13.9 -9.77
(I21–I22) (I21–I22)
Hispanic Women (n = )
Malignant neoplasm of breast Accidental poisoning
2.98 -2.48 1.92 -0.201
(C50) (X40–X49)*
Malignant neoplasms of colon,
2.97 -0.275 Septicemia (A40–A41) 0.896 0.435
rectum, and anus (C18–C21)
Renal failure (N17–N19) 2.09 0.576 Viral hepatitis (B15–B19) 0.849 0.351
Chronic ischemic heart
Congenital malformations,
disease other than
deformations, and chromosomal -1.06 -0.578 -3.98 0.224
atherosclerotic (I20,
abnormalities (Q00 Q99)
I25.1–I25.9)
Diabetes mellitus (E10–
HIV(B20–B24) -2.59 -2.05 -4.01 -0.843
E14)
Acute myocardial infarction Acute myocardial infarction
-2.79 -5.51 -5.51 -2.79
(I21–I22) (I21–I22)
(continued on page 1061)
 1061

Table 122.1 (continued)

Mortality Top Increasing and Mortality
Top Widening and Narrowing Disparity Rate Decreasing Causes of Death Rate Disparity
Disparities (ICD-10 Codes) Change Change (ICD-10 Codes) Change Change
White Men (n = )
Accidental poisoning
Accidental poisoning (X40–X49)* 16.7 13.6 13.6 16.7
(X40–X49)*
Chronic lower respiratory diseases Suicide by means other
other than bronchitis, emphysema, 6.40 -0.484 than firearms (U03,X60 4.62 4.30
and asthma (J44,J47) X71,X75 X84,Y87.0)
Hypertensive heart disease
Viral hepatitis (B15–B19) 5.28 2.89 3.70 3.38
(I11)
Chronic ischemic heart
Malignant neoplasms of trachea, disease other than
-5.22 -16.49 -13.9 -6.41
bronchus, and lung (C33–C34) atherosclerotic (I20,
I25.1–I25.9)
Chronic ischemic heart disease Malignant neoplasms of
other than atherosclerotic (I20, -6.41 -13.9 trachea, bronchus, and -16.49 -5.22
I25.1–I25.9) lung (C33–C34)
Acute myocardial infarction Acute myocardial infarction
-13.5 -23.8 -23.8 -13.5
(I21–I22) (I21–I22)
Black Men (n = )
Malignant neoplasms of liver and Hypertensive heart disease
11.0 5.32 6.41 7.73
intrahepatic bile ducts (C22) (I11)
Malignant neoplasms of
Hypertensive heart disease (I11) 7.73 6.41 liver and intrahepatic bile 5.32 11.0
ducts (C22)
Accidental poisoning
Accidental poisoning (X40–X49)* 7.60 5.20 5.20 7.60
(X40–X49)*
Malignant neoplasms of trachea,
-5.32 -26.9 HIV (B20–B24) -19.86 6.76
bronchus, and lung (C33–C34)
Acute myocardial infarction Acute myocardial infarction
-7.33 -25.6 -25.6 -7.33
(I21–I22) (I21–I22)
Malignant neoplasms of
Alcoholic liver disease (K70) -7.53 -6.42 trachea, bronchus, and -26.9 -5.32
lung (C33–C34)
Hispanic Men (n = )
Viral hepatitis (B15–B19) 2.89 2.66 Viral hepatitis (B15–B19) 2.66 2.89
Malignant neoplasms of liver and Hypertensive heart disease
1.29 2.08 2.29 .0248
intrahepatic bile ducts (C22) (I11)
Water, air, and space, and other Malignant neoplasms of
transport accidents and their 1.24 0.173 liver and intrahepatic bile 2.08 1.29
sequelae (V90 V99,Y85) ducts (C22)
Motor vehicle accidents (**) -5.98 -3.27 HIV (B20–B24) -9.54 -6.00
Chronic ischemic heart
disease other than
HIV (B20–B24) -6.00 -9.54 -11.1 -6.08
atherosclerotic (I20,
I25.1–I25.9)
Chronic ischemic heart disease
Acute myocardial infarction
other than atherosclerotic -6.08 -11.1 -11.64 -3.66
(I21–I22)
(I20, I25.1–I25.9)

*Deaths due to accidental poisoning consist primarily of lethal drug overdoses (see Paulozzi and Annest 2007).
**ICD-10 codes for motor vehicle accidents are V02-V04,V09.0,V09.2,V12-V14,V19.0-V19.2, V19.4- V19.6,V20-V79,V80.
3-V80.5,V81.0-V81.1,V82.0-V82.1,V83-V86, V87.0-V87.8,V88.0-V88.8,V89.0, V89.2.
1062 • Part IX: The Consequences of Inequality

not restricted to a few outcomes. Furthermore, Another way to interpret the association be-
widening and emerging disparities play an im- tween increasing mortality and widening dis-
portant role in the perpetuation of mortality parities is that causes of death on the upswing
disparities by education; the results indicate are driven primarily by increased mortality rates
that today mortality disparities across the least among people in the lower socioeconomic strata,
and most educated groups would be about 25 those most likely to suffer the consequences of
percent smaller if it were not for disparities that new health threats that enter a society. Such
have widened or emerged since 1999. threats may be more prevalent and have a greater
Additional evidence for the fundamental impact in the lower socioeconomic strata be-
cause perspective comes from the finding that cause people in these strata already have poorer
widening disparities by education were concen- health (NCHS 1999a) and therefore have fewer
trated among causes of death with increasing health reserves to resist the effects of new threats;
mortality rates. If mortality disparities by edu- have higher levels of unhealthy behaviors, such
cation remain strong over time by continually as smoking and inactivity (see Pampel, Krueger,
jumping to new outcomes, it then logically and Denney 2010), which may interact with
follows that they jump to outcomes that are new threats and amplify their negative effects;
becoming increasingly prominent and are the and in general are less likely to be early adopt-
future top killers. In support of this proposition, ers of new technologies and disease prevention
the analysis indicates that almost all causes of and health promotion efforts (Rogers 2003) that
death with increasing mortality rates had in- could protect against new threats.
creasing disparities by education. Specifically, Mortality disparities have endured in the
among white decedents, 93 percent of the causes United States because they shift to new outcomes
of death with increasing mortality rates had in- over time. As disparities in the major health out-
creasing disparities; the analogous numbers for comes of today eventually diminish, new ones
blacks and Hispanics were 88 percent and 70 emerge or widen in the outcomes that come to
percent, respectively. predominate in the future—a process this study
Consistent with the fundamental cause per- shows is continual and ongoing. Identifying the
spective, causes of death with the greatest widen- upstream processes that make this shift possible
ing of disparities were associated with outcomes offers a unique opportunity to better specify the
that experienced substantial change in their risk macro-micro link between social inequality and
factors. The cause of death with the largest wid- individual health. Such knowledge will lay the
ening of disparities by education was accidental groundwork for policies and interventions aimed
poisoning, which was top ranked among white at disparity prevention, which this study indi-
men and white women, and in the top three cates will be required of any comprehensive ef-
among black men and black women. The risk fort to achieve long-lasting reductions in health
for accidental poisoning has increased tremen- disparities.
dously in recent years as scripts for prescription
opioids have more than quadrupled, from 44 REFERENCES
million in 1991 to 179 million in 2009 (Miech, Ahmedin, Jemal, Elizabeth Ward, Robert N. Ander-
son, Taylor Murray, and Michael J. Thun. 2008.
Koester, and Holliman 2011; Volkow 2010). As
“Widening of Socioeconomic Inequalities in U.S.
predicted by the fundamental cause perspective,
Death Rates, 1993–2001.” PloS ONE 3: e2181.
the consequences of this changing risk factor fell Avendano, Mauricio, Ichiro Kawachi, Frank Van
disproportionately on people with lower educa- Lenthe, Hendriek C. Boshuizen, Johan P. Mack-
tion, as indicated by a widening disparity in ac- enbach, G. A. M. Vand den Bos, Martha E. Fay,
cidental poisoning mortality. Given substantial and Lisa F. Berkman. 2006. “Socioeconomic Sta-
changes in risk factors and its high preventabil- tus and Stroke Incidence in the US Elderly: The
ity, this cause of death is indeed a strong candi- Role of Risk Factors in the EPESE Study.” Stroke
date for widening disparities. 37: 1368–1373.
 Miech et al.—Why Do Mortality Disparities Persist? • 1063

Centers for Disease Control and Prevention (CDC). Behaviors.” Annual Review of Sociology 36:
n.d. Vital Statistics of the United States. http:// 349–370.
www.cdc.gov/nchs/products/vsus.htm (accessed Paulozzi, L. J., M. F. Ballesteros, and J. A. Stevens.
May 20, 2012). 2006. “Recent Trends in Mortality from Unin-
Clark, Alexander M., Marie DesMeules, Wei Luo, tentional Injury in the United States.” Journal of
Amanda S. Duncan, and Andy Wielgosz. 2009. Safety Research 37: 277–283.
“Socioeconomic Status and Cardiovascular Dis- Paulozzi, Leonard J., and J. Annest. 2007. “Unin-
ease: Risks and Implications for Care.” Nature tentional Poisoning Deaths—United States,
Reviews Cardiology 6: 712–722. 1999–2004.” Morbidity and Mortality Weekly
Link, Bruce G., and Jo Phelan. 1995. “Social Con- Report 56: 93–96.
ditions as Fundamental Causes of Disease.” Jour- Preston, Samuel H., and Irma T. Elo. 1995. “Are
nal of Health and Social Behavior (Extra Issue): Educational Differentials in Adult Mortality In-
80–94. creasing in the United States?” Journal of Aging
Miech, Richard, Steve Koester, and Brooke Dorsey and Health 7: 476–496.
Holliman. 2011. “Increasing U.S. Mortality Due Rogers, Everett M. 2003. Diffusion of Innovations.
to Accidental Poisoning: The Role of the Baby New York: The Free Press.
Boom Cohort.” Addiction 106: 806–815. Smith, James P. 2004. “Unraveling the SES-Health
Mirowsky, John, and Catherine E. Ross. 2003. Ed- Connection.” Population and Development Re-
ucation, Social Status, and Health. New York: Al- view 31 (supplement): 108–132.
dine de Gruyter. Volkow, Nora D. 2010. “Report on Scientific Re-
National Center for Health Statistics (NCHS). search on Prescription Drug Abuse.” http://olpa.
1999a. Health, United States, 1999 with Health od.nih.gov/hearings/110/session2/Testimonies/
and Aging Chartbook. Hyattsville, MD: National Nora_Volkow.asp.
Center for Health Statistics. Williams, David. 1990. “Socioeconomic Differences
———. 1999b. Instruction Manual Part 9: ICD-10 in Health: A Review and Redirection.” Social Psy-
Cause of Death Lists for Tabulating Mortality Sta- chology Quarterly 53: 81–99.
tistics, Effective 1999. Hyattsville, MD: National World Health Organization. 1992. International
Center for Health Statistics. Statistical Classification of Diseases and Related
Omran, Abdel R. 2005. “The Epidemiological Tran- Health Problems. 10th rev. Geneva: World Health
sition: A Theory of the Epidemiology of Popula- Organization.
tion Change.” Milbank Memorial Fund Quarterly Xu, Jiaquan, Kenneth D. Kochanek, Sherry L. Mur-
83: 731–757. phy, and Betzaida Tejada-Vera. 2010. “Deaths:
Pampel, Fred C., Patrick M. Krueger, and Justin Final Data for 2007.” National Vital Statistics
T. Denney. 2010. “SES Disparities in Health Reports 58 (19): 1–135.
 PART X

The Future of Inequality

Industrialism and Post-Industrialism

123 The Coming of Post-Industrial Society 1066
124 Social Foundations of Post-Industrial Economies 1078
125 The Labor Force and the Great Recession 1091

Post-Socialism
126 Post-Socialist Stratification 1098
127 Making Capitalism Without Capitalists 1104
128 Elite Opportunity in Transitions from State Socialism 1110

Post-Modernity and High Modernity

129 The Evolution of Modern Stratification Systems 1116
130 Social Justice and Social Divisions 1125

Globalization and Inequality

131 Globalism’s Discontents 1132
132 The New Geography of Global Income Inequality 1139
1066 • Part X: The Future of Inequality

Industrialism and Post-Industrialism

123 • Daniel Bell

The Coming of Post-Industrial Society

The concept of the post-industrial society deals fortunes, the political order necessarily becomes
primarily with changes in the social structure, paramount. Since the post-industrial society
the way in which the economy is being trans- increases the importance of the technical com-
formed and the occupational system reworked, ponent of knowledge, it forces the hierophants
and with the new relations between theory and of the new society—the scientists, engineers,
empiricism, particularly science and technology. and technocrats—either to compete with poli-
These changes can be charted, as I seek to do ticians or become their allies. The relationship
in this [chapter]. But I do not claim that these between the social structure and the political
changes in social structure determine correspond- order thus becomes one of the chief problems
ing changes in the polity or the culture. Rather, of power in a post-industrial society. And, third,
the changes in social structure pose questions for the new modes of life, which depend strongly on
the rest of society in three ways. First, the social the primacy of cognitive and theoretical knowl-
structure—especially the social structure—is edge, inevitably challenge the tendencies of the
a structure of roles, designed to coordinate the culture, which strives for the enhancement of
actions of individuals to achieve specific ends. the self and turns increasingly antinomian and
Roles segment individuals by defining limited anti-institutional.
modes of behavior appropriate to a particular In this [chapter], I am concerned chiefly with
position, but individuals do not always willingly the social structural and political consequences
accept the requirements of a role. One aspect of of the post-industrial society. In a later work I
the post-industrial society, for example, is the shall deal with its relation to culture. But the
increasing bureaucratization of science and the heart of the endeavor is to trace the societal
increasing specialization of intellectual work into changes primarily within the social structure.
minute parts. Yet it is not clear that individuals “Too large a generalization,” Alfred North
entering science will accept this segmentation, Whitehead wrote, “leads to mere barrenness. It
as did the individuals who entered the factory is the large generalization, limited by a happy
system a hundred and fifty years ago. particularity, which is the fruitful conception.”1
Second, changes in social structure pose It is easy—and particularly so today—to set
“management” problems for the political sys- forth an extravagant theory which, in its histor-
tem. In a society which becomes increasingly ical sweep, makes a striking claim to originality.
conscious of its fate, and seeks to control its own But when tested eventually by reality, it turns

Daniel Bell. The Coming of Post-Industrial Society, pp. xvi–xix, 13–14, 358–364, 374, 378, 408–410, 445–446.
Copyright © 1973 by Daniel Bell. Foreword copyright © 1976 by Daniel Bell. Reprinted by permission of
Basic Books, a member of the Perseus Books Group.
 Bell—The Coming of Post-Industrial Society • 1067

into a caricature—viz. James Burnham’s theory 1975. By the year 2000, the technical and pro-
of the managerial revolution thirty years ago, or fessional class will be the largest single group in
C. Wright Mills’s conception of the power elite, the society.
or W. W. Rostow’s stages of economic growth. I 4. The change from goods to services. In the
have tried to resist that impulse. Instead, I am United States today more than 65 out of every
dealing here with tendencies, and have sought to 100 persons are engaged in services. By 1980,
explore the meaning and consequences of those the figure will be about 70 in every 100. A
tendencies if the changes in social structure that large service sector exists in every society. In a
I describe were to work themselves to their log- pre-industrial society this is mainly a household
ical limits. But there is no guarantee that they and domestic class. (In England, it was the single
will. Social tensions and social conflicts may largest class in the society until about 1870.) In
modify a society considerably; wars and recrim- an industrial society, the services are transpor-
inations can destroy it; the tendencies may pro- tation, utilities, and finance, which are auxil-
voke a set of reactions that inhibit change. Thus iary to the production of goods, and personal
I am writing what Hans Vahinger called an “as service (beauticians, restaurant employees, and
if,” a fiction, a logical construction of what could so forth). But in a post-industrial society, the
be, against which the future social reality can new services are primarily human services (prin-
be compared in order to see what intervened to cipally in health, education and social services)
change society in the direction it did take. and professional and technical services (e.g., re-
The concept of the post-industrial society is search, evaluation, computers, and systems anal-
a large generalization. Its meaning can be more ysis). The expansion of these services becomes a
easily understood if one specifies [eleven] dimen- constraint on economic growth and a source of
sions, or components, of the term: persistent inflation.
5. A change in the character of work. In a
1. The centrality of theoretical knowledge. pre-industrial world, life is a game against nature
Every society has always existed on the basis in which men wrest their living from the soil, the
of knowledge, but only now has there been a waters, or the forests, working usually in small
change whereby the codification of theoretical groups, subject to the vicissitudes of nature. In
knowledge and materials science becomes the an industrial society, work is a game against fab-
basis of innovations in technology. One sees this ricated nature, in which men become dwarfed
primarily in the new science-based industries— by machines as they turn out goods and things.
computers, electronics, optics, polymers—that But in a post-industrial world, work is primarily
mark the last third of the century. a “game between persons” (between bureaucrat
2. The creation of a new intellectual technology. and client, doctor and patient, teacher and stu-
Through new mathematical and economic tech- dent, or within research groups, office groups,
niques—based on the computer linear program- service groups). Thus in the experience of work
ming, Markov chains, stochastic processes and and the daily routine, nature is excluded, arti-
the like—we can utilize modeling, simulation facts are excluded, and persons have to learn how
and other tools of system analysis and decision to live with one another. In the history of human
theory in order to chart more efficient, “rational” society, this is a completely new and unparalleled
solutions to economic and engineering, if not so- state of affairs.
cial, problems. 6. The role of women. Work in the industrial
3. The spread of a knowledge class. The fast- sector (e.g., the factory) has largely been men’s
est growing group in society is the technical work, from which women have been usually ex-
and professional class. In the United States this cluded. Work in the post-industrial sector (e.g.,
group, together with managers, made up 25 per- human services) provides expanded employment
cent of a labor force of eight million persons in opportunities for women. For the first time,
1068 • Part X: The Future of Inequality

one can say that women have a secure base for centers), and the military. My argument is that
economic independence. One sees this in the the major interest conflicts will be between the
steadily rising curve of women’s participation in situs groups, and that the attachments to these
the labor force, in the number of families (now situses might be sufficiently strong to prevent the
60 percent of the total) that have more than one organization of the new professional groups into
regular wage earner, and in the rising incidence a coherent class in society.2
of divorce as women increasingly feel less depen- 9. Meritocracy. A post-industrial society, being
dent, economically, on men. primarily a technical society, awards place less
7. Science as the imago. The scientific com- on the basis of inheritance or property (though
munity, going back to the seventeenth century, these can command wealth or cultural advan-
has been a unique institution in human society. tage) than on education and skill. Inevitably the
It has been charismatic, in that it has been rev- question of a meritocracy becomes a crucial nor-
olutionary in its quest for truth and open in its mative question. In this [chapter] I attempt to
methods and procedures; it derives its legitimacy define the character of meritocracy and defend
from the credo that knowledge itself, not any the idea of a “just meritocracy,” or of place based
specific instrumental ends, is the goal of science. on achievement, through the respect of peers.
Unlike other charismatic communities (prin- 10. The end of scarcity? Most socialist and
cipally religious groups and messianic political utopian theories of the nineteenth century as-
movements), it has not “routinized” its creeds cribed almost all the ills of society to the scarcity
and enforced official dogmas. Yet until recently, of goods and the competition of men for these
science did not have to deal with the bureau- scarce goods. In fact, one of the most common
cratization of research, the subordination of its definitions of economics characterized it as the
inquiries to state-directed goals, and the “test” art of efficient allocation of scarce goods among
of its results on the basis of some instrumental competing ends. Marx and other socialists ar-
payoff. Now science has become inextricably in- gued that abundance was the precondition for
tertwined not only with technology but with the socialism and claimed, in fact, that under so-
military and with social technologies and soci- cialism there would be no need to adopt nor-
etal needs. In all this—a central feature of the mative rules of just distribution, since there
post-industrial society—the character of the new would be enough for everyone’s needs. In that
scientific institutions—will be crucial for the fu- sense, the definition of communism was the ab-
ture of free inquiry and knowledge. olition of economics, or the “material embod-
8. Situses as political units. Most of sociolog- iment” of philosophy. Yet it is quite clear that
ical analysis has focused its attention on classes scarcity will always be with us. I mean not just
or strata, horizontal units of society that exist the question of scarce resources (for this is still
in superior-subordinate relation to each other. a moot point) but that a post-industrial soci-
Yet for the post-industrial sectors, it may well ety, by its nature, brings new scarcities which
be that situses (from the Latin situ, location), a nineteenth- and early-twentieth-century writers
set of vertical orders, will be the more important had never thought of. The socialists and liberals
loci of political attachment. In Schema 123.1 on had talked of the scarcities of goods; but in the
page 1073, I sketch the possible situses of the post-industrial society there will be scarcities of
post-industrial order. There are four functional information and of time. And the problems of
situses—scientific, technological (i.e., applied allocation inevitably remain, in the crueler form,
skills: engineering, economics, medicine), ad- even, of man becoming homo economicus in the
ministrative and cultural—and five institutional disposition of his leisure time.
situses—economic enterprises, government 11. The economics of information. Information
bureaus, universities and research complexes, is by its nature a collective, not a private, good
social complexes (e.g., hospitals, social-service (i.e., a property). In the marketing of individual
 Bell—The Coming of Post-Industrial Society • 1069

Table 123.1. Stratification and Power

goods, it is clear that a “competitive” strategy ideologically (as they have in the recent ABM
between producers is to be preferred lest enter- debate), and different groups of scientists will
prise become slothful or monopolistic. Yet for align themselves with different segments of
the optimal social investment in knowledge, we other elites. In the nature of politics, few groups
have to follow a “cooperative” strategy in order are monolithic (“the” military, “the” scientists,
to increase the spread and use of knowledge in “the” business class), and any group contending
society. This new problem regarding information for power will seek allies from different groups.
poses the most fascinating challenges to econo- (Thus, in the Soviet Union, for example, where
mists and decision makers in respect to both the- the interest groups are more clear-cut in func-
ory and policy in the post-industrial society. . . . tional terms—factory managers, central plan-
ners, army officers, party officials—and the
power struggle more naked, any faction in the
Who Holds Power? Politburo seeking power will make alliances across
Decisions are a matter of power, and the crucial group lines. Yet once in power, the victors will
questions in any society are: Who holds power? have to make decisions between groups and affect
And how is power held? How power is held is a the relative distribution of power of the func-
system concept; who holds power is a group con- tional units and shift the weights of the system.)
cept. How one comes to power defines the base In the change of the system in the post-industrial
and route; who identifies the persons. Clearly, society, two propositions become evident:
when there is a change in the nature of the sys- 1. As a stratum, scientists, or more widely the
tem, new groups come to power. (In the tableau technical intelligentsia, now have to be
of pre-industrial, industrial, and post-industrial taken into account in the political process,
societies, the major differences can be shown though they may not have been before.
schematically—see Table 123.1 on Stratification 2. Science itself is ruled by an ethos which is
and Power.) different from the ethos of other major so-
In the post-industrial society, technical skill cial groups (e.g., business, the military), and
becomes the base of and education the mode of this ethos will predispose scientists to act in
access to power; those (or the elite of the group) a different fashion, politically, from other
who come to the fore in this fashion are the sci- groups.
entists. But this does not mean that the scientists
are monolithic and act as a corporate group. In Forty-five years ago Thorstein Veblen, in his
actual political situations scientists may divide Engineers and the Price System, foresaw a new
1070 • Part X: The Future of Inequality

society based on technical organization and in- industrial system will not work. It is a mechan-
dustrial management, a “soviet of technicians,” ically organized structure of the technical pro-
as he put it in the striking language he loved to cesses designed, installed, and conducted by the
use in order to scare and mystify the academic production engineers. Without them and their
world. In making this prediction, Veblen shared constant attention to the industrial equipment,
the illusion of that earlier technocrat, Henri de the mechanical appliances of industry will foot
Saint-Simon, that the complexity of the indus- up to just so much junk.”
trial system and the indispensability of the tech- This syndicalist idea that revolution in the
nician made military and political revolutions a twentieth century could only be an “industrial
thing of the past. “Revolutions in the eighteenth overturn” exemplifies the fallacy in so much of
century,” Veblen wrote, “were military and po- Veblen’s thought. For as we have learned, no
litical; and the Elder Statesmen who now believe matter how technical social processes may be,
themselves to be making history still believe the crucial turning points in a society occur in a
that revolutions can be made and unmade by political form. It is not the technocrat who ulti-
the same ways and means in the twentieth cen- mately holds power, but the politician.
tury. But any substantial or effectual overturn The major changes that have reshaped Amer-
in the twentieth century will necessarily be an ican society over the past thirty years—the cre-
industrial overturn, and by the same token, any ation of a managed economy, a welfare society,
twentieth-century revolution can be combat- and a mobilized polity—grew out of political
ted or neutralized only by industrial ways and responses: in the first instances to accommo-
means.” date the demands of economically insecure and
If a revolution were to come about in the disadvantaged groups—the farmers, workers,
United States—as a practiced skeptic Veblen was blacks and the poor—for protection from the
highly dubious of that prospect—it would not hazards of the market; and later because of the
be led by a minority political party, as in Soviet concentration of resources and political objec-
Russia, which was a loose-knit and backward tives following the mobilized postures of the cold
industrial region, nor would it come from the war and the space race.
trade-union “votaries of the dinner pail,” who, All of this opens up a broader and more the-
as a vested interest themselves, simply sought to oretical perspective about the changing nature of
keep prices up and labor supply down. It would class and social position in contemporary society.
occur, he said, along the lines “already laid down Class, in the final sense, denotes not a specific group
by the material conditions of its productive in- of persons but a system that has institutionalized
dustry.” And, turning this Marxist prism to his the ground rules for acquiring, holding, and trans-
own perceptions, Veblen continued: “These ferring differential power and its attendant priv-
main lines of revolutionary strategy are lines of ileges. In Western society, the dominant system
technical organization and industrial manage- has been property, guaranteed and safeguarded
ment; essentially lines of industrial engineering; by the legal order, and transmitted through
such as will fit the organization to take care of a system of marriage and family. But over the
the highly technical industrial system that con- past twenty-five to fifty years, the property sys-
stitutes the indispensable material foundation of tem has been breaking up. In American society
any modern civilized community.” today, there are three modes of power and so-
The heart of Veblen’s assessment of the cial mobility, and this baffles students of society
revolutionary class is thus summed up in his who seek to tease out the contradictory sources
identification of the “production engineers” of class positions. There is the historic mode of
as the indispensable “General Staff of the in- property as the basis of wealth and power, with
dustrial system.” “Without their immediate inheritance as the major route of access. There is
and unremitting guidance and correction the technical skill as the basis of power and position,
 Bell—The Coming of Post-Industrial Society • 1071

Table 123.2. Reduced Model

with education as the necessary route of access to program budgeting), have now become essential
skill. And finally there is political office as a base to the formulation and analysis of decisions on
of power, with organization of a machine as the which political judgments have to be made, if
route of access. not to the wielding of power. It is in this broad
One can, in a simplified way, present these sense that the spread of education, research, and
modes in Table 123.2. administration has created a new constituency—
The difficulty in the analysis of power in the technical and professional intelligentsia.
modern Western societies is that these three While these technologists are not bound
systems [in Table 123.2] co-exist, overlap, and by a sufficient common interest to make them
interpenetrate. While the family loses its im- a political class, they do have common charac-
portance as an economic unit, particularly with teristics. They are, first, the products of a new
the decline of family-firms and the break-up of system in the recruitment for power (just as
family capitalism, family background is still ad- property and inheritance were the essence of
vantageous in providing impetus (financial, cul- the old system). The norms of the new intelli-
tural and personal connections) for the family gentsia—the norms of professionalism—are a
member. Ethnic groups, often blocked in the departure from the hitherto prevailing norms of
economic access to position, have resorted to economic self-interest which have guided a busi-
the political route to gain privilege and wealth. ness civilization. In the upper reaches of this new
And, increasingly, in the post-industrial society, elite—that is, in the scientific community—men
technical skill becomes an overriding condition hold significantly different values, which could
of competence for place and position. A son may become the foundation of the new ethos for such
succeed a father as head of a firm, but without a class.
the managerial skill to run the enterprise, the Actually, the institution of property itself is
firm may lose out in competition with other, undergoing a fundamental revision, in a signif-
professionally managed corporations. To some icant way. In Western society for the past sev-
extent, the owner of a firm and the politician eral hundred years, property, as the protection
may hire technicians and experts; yet, unless the of private rights to wealth, has been the eco-
owner or politician themselves know enough nomic basis of individualism. Traditionally the
about the technical issues, their judgments may institution of property, as Charles Reich of the
falter. Yale Law School has put it, “guards the trou-
The rise of the new elites based on skill derives bled boundary between individual man and the
from the simple fact that knowledge and plan- state.” In modern life property has changed in
ning—military planning, economic planning, two distinctive ways. One of these is elementary:
social planning—have become the basic requi- Individual property has become corporate, and
sites for all organized action in a modern soci- property is no longer controlled by owners but
ety. The members of this new technocratic elite, by managers. In a more subtle and diffuse way,
with their new techniques of decision-making however, a new kind of property has emerged,
(systems analysis, linear programming, and and with it a different kind of legal relationship.
1072 • Part X: The Future of Inequality

To put it more baldly, property today consists strategic assessments, took on the highest im-
not only of visible things (land, possessions, ti- portance in the shaping of subsequent policy.
tles) but also of claims, grants, and contracts. Even a reworking of the economic map of the
The property relationship is not only between United States followed as well, with Texas and
persons but between the individual and the gov- California gaining great importance because of
ernment. As Reich points out, “The valuables the electronics and aerospace industries. In these
dispensed by government take many forms, instances technology and strategy laid down the
but they all share one characteristic. They are requirements, and only then could business and
steadily taking the place of the traditional forms local political groups seek to modify, or take ad-
of wealth—forms which are held as private prop- vantage of, these decisions so as to protect their
erty. Social insurance substitutes for savings, a own economic interests.
government contract replaces a businessman’s In all this, the technical intelligentsia holds
customers and goodwill. . . . Increasingly, Amer- a double position. To the extent that it has in-
icans live on government largess—allocated by terests in research, and positions in the universi-
government on its own terms, and held by re- ties, it becomes a new constituency—just as the
cipients subject to conditions which express ‘the military is a distinct new constituency, since this
public interest.’”3 country has never before had a permanent mil-
While many forms of this “new property” itary establishment seeking money and support
represent direct grants (subsidies to farmers, cor- for science, for research and development. Thus
porations, and universities) or are contracts for the intelligentsia becomes a claimant, like other
services or goods (to industry and universities), groups, for public support (though its influence
the most pervasive form is claims held by indi- is felt in the bureaucratic and administrative
viduals (social security, medical care, housing al- labyrinth, rather than in the electoral system or
lowances) which derive from a new definition of mass pressure). At the same time, the techni-
social rights: claims on the community to ensure cians represent an indispensable administrative
equality of treatment, claims legitimately due a staff for the political office holder with his public
person so that he will be able to share in the so- following. . . .
cial heritage. And the most important claim of If one turns, then, to the societal structure of
all is full access to education, within the limits of the post-industrial society considered along these
one’s talent and potential. two historical axes [of class and power], two con-
The result of all this is to enlarge the arena of clusions are evident. First, the major class of the
power, and at the same time to complicate the emerging new society is primarily a professional
modes of decision-making. The domestic po- class, based on knowledge rather than property.
litical process initiated by the New Deal was in But second, the control system of the society is
effect a broadening of the “brokerage” system— lodged not in a successor-occupational class but
the system of political deals between constituen- in the political order, and the question of who
cies—although there are now many participants manages the political order is an open one. (See
in the game. But there is also a new dimension in Schema 123.1.)
the political process, which has given the techno- In terms of status (esteem and recognition,
crats a new role. Matters of foreign policy have and possibly income), the knowledge class may
not been a reflex of internal political forces, but be the highest class in the new society, but in
a judgment about the national interest, involv- the nature of that structure there is no intrinsic
ing strategy decisions based on the calculation reason for this class, on the basis of some co-
of an opponent’s strength and intentions. Once herent or corporate identity, to become a new
the fundamental policy decision was made to economic interest class, or a new political class
oppose the communist power, many technical which would bid for power. The reasons for this
decisions, based on military technology and are evident from an inspection of the Schema.
 Bell—The Coming of Post-Industrial Society • 1073

Schema 123.1. The Societal Structure of the Post-Industrial Society (U.S. Model)

The professional class as I define it is made policies of the different situses they are obedi-
up of four estates: the scientific, the technologi- ent to. The administrative estate is concerned
cal, the administrative, and the cultural.4 While with the management of organizations and is
the estates, as a whole, are bound by a common bound by the self-interest of the organization
ethos, there is no intrinsic interest that binds itself (its perpetuation and aggrandizement) as
one to the other, except for a common defense well as the implementation of social purposes,
of the idea of learning; in fact there are large dis- and may come into conflict with one or an-
junctions between them. The scientific estate is other of the estates. The cultural estate—artistic
concerned with the pursuit of basic knowledge and religious—is involved with the expressive
and seeks, legitimately, to defend the conditions symbolism (plastic or ideational) of forms and
of such pursuit, untrammeled by political or ex- meanings, but to the extent that it is more in-
traneous influence. The technologists, whether tensively concerned with meanings, it may find
engineers, economists, or physicians, base their itself increasingly hostile to the technological
work on a codified body of knowledge, but in and administrative estates. As I noted in the
the application of that knowledge to social or introduction, the axial principle of modern cul-
economic purposes they are constrained by the ture, in its concern with the self, is antinomian
1074 • Part X: The Future of Inequality

and anti-institutional, and thus hostile to the advantages they could not obtain in the eco-
functional rationality which tends to dominate nomic order. This is what has been taking place
the application of knowledge by the techno- in recent years, and it will continue. The second,
logical and administrative estates. Thus in the and structurally more pervasive, shift is that in
post-industrial society one finds increasingly a the post-industrial society the situses rather than
disjunction between social structure and culture the statuses would be the major political-interest
which inevitably affects the cohesiveness if not units in the society. To some extent this is ev-
the corporate consciousness of the four estates.5 ident in the familiar phenomenon of pressure
While the classes may be represented, hori- groups. But in the post-industrial society it is
zontally, by statuses (headed by the four estates), more likely that the situses will achieve greater
the society is organized, vertically, by situses, corporate cohesiveness vis-à-vis one another and
which are the actual loci of occupational ac- become the major claimants for public support
tivities and interests. I use this unfamiliar so- and the major constituencies in the determina-
ciological word situses to emphasize the fact tion of public policy.6 And yet the very forces
that in day-to-day activities the actual play and which have re-emphasized the primacy of the
conflict of interests exist between the organiza- political order in a technical world make it im-
tions to which men belong, rather than between perative to define some coherent goals for the so-
the more diffuse class or status identities. In a ciety as a whole and, in the process, to articulate
capitalist society, the property owner or busi- a public philosophy which is more than the sum
nessman, as a class, is located exclusively in the of what particular situses or social groups may
business firm or corporation, so that status and want. In the efforts to forge some such coherence
situs are joined. In the post-industrial society, one may find the seeds of the cohesiveness of the
however, the four estates are distributed among professional class in the post-industrial society.
many different situses. Scientists can work for A new social system, contrary to Marx, does
economic enterprises, government, universities, not always arise necessarily within the shell of
social complexes, or the military (though the an old one but sometimes outside of it. The
bulk of the “pure” scientists are to be found in framework of feudal society was made up of no-
the university). And the same distributions hold blemen, lords, soldiers, and priests whose wealth
for the technologists and the managers. Because was based on land. The bourgeois society that
of this “cross-cutting,” the likelihood of a pure took hold in the thirteenth century was made
“estate” consciousness for political purposes up of artisans, merchants, and free professionals
tends to diminish. whose property lay in their skills or their will-
Finally, if the major historical turn in the ingness to take risks, and whose mundane val-
last quarter-century has been the subordination ues were far removed from the fading theatrics
of the economic function to societal goals, the of the chivalric style of life. It arose, however,
political order necessarily becomes the control outside the feudal landed structure, in the free
system of the society. But who runs it, and for communes, or towns, that were no longer sei-
whose (or what) ends? In one respect, what the gnorial dependencies. And these self-ruling small
change may mean is that traditional social con- communes became the cornerstones of the fu-
flicts have simply shifted from one arena to an- ture European mercantile and industrial society.7
other, so that what the traditional classes fought So, too, the process today. The roots of
out in the economic realm, where men sought post-industrial society lie in the inexorable in-
comparative advantage in place, privilege and fluence of science on productive methods, par-
domination, is now transferred to the political ticularly in the transformation of the electrical
realm, and as that arena widens, the special foci and chemical industries at the beginning of the
and ethnic groups (the poor and the blacks) now twentieth century. But as Robert Heilbroner
seek to gain through politics the privileges and has observed: “Science, as we know it, began
 Bell—The Coming of Post-Industrial Society • 1075

well before capitalism existed and did not ex- the household because of the need to nurture
perience its full growth until well after capital- high-IQ children, the activist women had de-
ism was solidly entrenched.” And science, as a manded equality between the sexes, a movement
quasi-autonomous force, would extend beyond that was then generalized into the demand for
capitalism. By this token, one can say that the equality for all, and for a classless society. Life
scientific estate—its ethos and its organiza- was not to be ruled by “a mathematical measure”
tion—is the monad that contains within itself but each person would develop his own diverse
the imago of the future society. 8 . . . capacities for leading his own life.10 The Popu-
lists won. After little more than half a century,
the Meritocracy had come to an end.
Meritocracy and Equality Is this, too, the fate of the post-industrial
In 1958, the English sociologist Michael Young society? The post-industrial society, in its initial
wrote a fable, The Rise of the Meritocracy.9 It pur- logic, is a meritocracy. Differential status and
ports to be a “manuscript,” written in the year differential income are based on technical skills
2033, which breaks off inconclusively for rea- and higher education. Without those achieve-
sons the “narrator” failed to comprehend. The ments one cannot fulfill the requirements of the
theme is the transformation of English society, new social division of labor which is a feature
by the turn of the twenty-first century, owing of that society. And there are few high places
to the victory of the principle of achievement open without those skills. To that extent, the
over that of ascription (i.e., the gaining of place post-industrial society differs from society at the
by assignment or inheritance). For centuries, the turn of the twentieth century. The initial change,
elite positions in the society had been held by the of course, came in the professions. Seventy years
children of the nobility on the hereditary prin- or so ago, one could still “read” law in a lawyer’s
ciple of succession. But in the nature of modern office and take the bar examination without a
society, “the rate of social progress depend[ed] college degree. Today, in medicine, law, account-
on the degree to which power is matched with ing, and a dozen other professions, one needs
intelligence.” Britain could no longer afford a college degree and accrediting, through exam-
a ruling class without the necessary technical ination, by legally sanctioned committees of the
skills. Through the successive school-reform profession, before one can practice one’s art. For
acts, the principle of merit slowly became estab- many years, until after World War II, business
lished. Each man had his place in the society on was the chief route open to an ambitious and
the basis of “IQ and Effort.” By 1990 or there- aggressive person who wanted to strike out for
abouts, all adults with IQs over 125 belonged to himself. And the rags-to-riches ascent (or, more
the meritocracy. accurately, clerk-to-capitalist, if one follows the
But with that transformation came an unex- career of a Rockefeller, Harriman, or Carnegie)
pected reaction. Previously, talent had been dis- required drive and ruthlessness rather than edu-
tributed throughout the society, and each class or cation and skills. One can still start various kinds
social group had its own natural leaders. Now all of small businesses (usually, now, by franchise
men of talent were raised into a common elite, from a larger corporation), but the expansion of
and those below had no excuses for their fail- such enterprises takes vastly different skills than
ures; they bore the stigma of rejection, they were in the past. Within the corporation, as mana-
known inferiors. gerial positions have become professionalized,
By the year 2034 the Populists had revolted. individuals are rarely promoted from shop jobs
Though the majority of the rebels were members below but are chosen from the outside, with a
of the lower classes, the leaders were high-status college degree as the passport of recognition.
women, often the wives of leading scientists. Only in politics, where position may be achieved
Relegated during the early married years to through the ability to recruit a following, or
1076 • Part X: The Future of Inequality

through patronage, is there a relatively open lad- politics from the start. The group process—
der without formal credentials. which was the vaunted discovery of the “realists”
Technical skill, in the post-industrial society, of American political science—consisted largely
is what the economists call “human capital.” An of economic bargaining between functional or
“investment” in four years of college, according pressure groups operating outside the formal
to initial estimates of Gary Becker, yields, over structure of the party system. What we now find
the average working life of the male graduate, are ethnic and ascriptive groups claiming formal
an annual return of about 13 percent.11 Grad- representation both in the formal political struc-
uation from an elite college (or elite law school ture and in all other institutions of the society
or business school) gives one a further differ- as well.
ential advantage over graduates from “mass” or These claims are legitimated, further, by the
state schools. Thus, the university, which once fact that America has been a pluralist society, or
reflected the status system of the society, has has come to accept a new definition of plural-
now become the arbiter of class position. As ism rather than the homogeneity of American-
the gate-keeper, it has gained a quasi-monopoly ism. Pluralism, in its classic conceptions,13 made
in determining the future stratification of the a claim for the continuing cultural identity of
society.12 ethnic and religious groups and for the institu-
Any institution which gains a quasi-monopoly tional autonomy of cultural institutions (e.g.,
power over the fate of individuals is likely, in universities) from politics. Pluralism was based
a free society, to be subject to quick attack. on the separation of realms. But what we have
Thus, it is striking that the populist revolt, today is a thoroughgoing politicizing of society
which Michael Young foresaw several decades in which not only the market is subordinated
hence, has already begun, at the very onset of to political decision but all institutions have to
the post-industrial society. One sees this in the bend to the demands of a political center and
derogation of the IQ and the denunciation of politicize themselves in group representational
theories espousing a genetic basis of intelligence; terms. Here, too, there has been another change.
the demand for “open admission” to universi- In functional group politics, membership was
ties on the part of minority groups in the large not fixed, and one could find cross-cutting alle-
urban centers; the pressure for increased num- giances or shifting coalitions. Today the groups
bers of blacks, women, and specific minority that claim representation—in the political par-
groups such as Puerto Ricans and Chicanos in ties, in the universities, in the hospitals and the
the faculties of universities, by quotas if neces- community—are formed by primordial or bio-
sary; and the attack on “credentials” and even logical ties, and one cannot erase the ascriptive
schooling itself as the determinant of a man’s nature of sex or color.
position in the society. A post-industrial society And yet, once one accepts the principle of re-
reshapes the class structure of society by creat- dress and representation of the disadvantaged in
ing new technical elites. The populist reaction, the group terms that were initially formulated,
which has begun in the 1970s, raises the demand it is difficult for the polity to deny those later
for greater “equality” as a defense against being claims. That is the logic of democracy which has
excluded from that society. Thus the issue of always been present in the ambiguous legacy of
meritocracy versus equality. . . . the principle of equality.
The claim for group rights stands in forma
contradiction to the principle of individualism, NOTES
with its emphasis on achievement and univer- 1. Alfred North Whitehead, Science and the Mod-
ern World (New York, 1960; original edition, 1925),
salism. But in reality it is no more than the ex-
p. 46.
tension, to hitherto excluded social units, of the
2. What is striking is that in the communist
group principle which has undergirded American world, it is quite clear that situses play the major role
 Esping-Andersen—Social Foundations of Post-Industrial Economies • 1077

in politics. One analyzes the play of power, not in and executive branches, rather than work directly
class terms, but on the basis of the rivalries among through the electoral process. Reality complicates
the party, the military, the planning ministries, the immeasurably any ideal-type schemas.
industrial enterprises, the collective farms, the cul- 7. Paradoxically, the growth of that society came
tural institutions—all of which are situses. about only after the self-contained economic life of
3. Charles Reich, “The New Property,” The Pub- the commune—its roots—was broken by the rise of
lic Interest, no. 3 (Spring 1966), p. 57. larger-scale industry which, in branching out, could
4. The suggestion of four estates is derived, of buy its raw materials in one town and sell in an-
course, from Don K. Price’s fruitful book The Scien- other, and which made its way, against both the older
tific Estate (Cambridge, Mass., 1965). Price defines feudal society and the regulative restrictions of the
four functions in government—the scientific, profes- commune, in alliance with the monarchical central-
sional, administrative, and political—and converts ization of the newly emerging national state.
each function, as an ideal type, into an estate. My 8. This is, indeed, Heilbroner’s suggestion. See
differences with Price are twofold: I think the estates Robert Heilbroner, The Limits of American Capital-
can be represented more accurately as social groups, ism (New York, 1966), p. 115.
rather than functions; more importantly, I do not 9. Michael Young, The Rise of the Meritocracy,
consider the political function coeval logically with 1870–2033 (London, 1958).
the others, for I see the political as the control system 10. A theoretician of the Technicians party, Pro-
of the entire societal structure. Terminologically, I fessor Eagle, had argued that marriage partners, in
have substituted the word “technological” (for the the national interest, should consult the intelligence
applied skills) where Price uses “professional,” since register, for a high-IQ man who mates with a low-IQ
I would reserve “professional” for the larger meaning woman is wasting his genes. The activist women, on
of the entire class, and I have added a cultural estate, the other hand, took romance as their banner and
where Price has none. Nonetheless, my indebtedness beauty as their flag, arguing that marriage should
to Price is great. be based on attraction. Their favorite slogan was
5. One might note that the more extreme forms “Beauty is achievable by all.”
of the “new consciousness” such as Theodore Roz- 11. Gary S. Becker, Human Capital (New York,
sak’s The Making of a Counter-Culture and Charles 1964), p. 112. Later writers have suggested this fig-
Reich’s The Greening of America manifest a distinct ure may be too high; the point remains that a college
hostility not only to scientism, but to science as well. degree does provide an investment “yield.”
6. The limitation of this analysis is that while the 12. For a comprehensive discussion of this major
post-industrial society, in its societal structure, in- social change, see Jencks and Riesman, The Academic
creasingly becomes a functional society, the political Revolution (New York, 1968). For a survey of the
order is not organized in functional terms. Thus the reaction, see Stephen Graubard and Geno Ballotti,
continuing existence of the traditional geographical eds., The Embattled University (New York, 1970).
districts and the dispersal of persons in this fashion 13. See, for example, the work of R. M. MacIver,
means that the political issues at any one time are The More Perfect Union: A Program for the Control
much more diffuse than the interests of the partic- of Inter-group Discrimination (New York, 1948), and
ular statuses or situses. It would also indicate that on the religious side, John Courtney Murray, We
the situses would, like the pressure groups, operate Hold These Truths: Reflections on the American Propo-
primarily through the lobbying of the legislative sition (New York, 1960).
1078 • Part X: The Future of Inequality

124 • Gøsta Esping-Andersen

Social Foundations of Post-Industrial Economies

A social trend is basically a projection from the is clearly related to job performance: in North
past. Many of the early postindustrial theorists, America labour market exclusion is less dra-
such as Daniel Bell (1976), predicted a future in matic than is growing pay inequality, declining
which most of what they deemed positive in the real wages, and a swelling army of the working
era of the ‘democratic class struggle’ would come poor. Europe’s social safety net manages to stem
to full fruition. Bell’s vision was a coming soci- the tide of inequality but its inferior job perfor-
ety of professionals and technicians, one where mance induces mass exclusion.
‘situs’ rather than class conflict would reign. How do we understand the postindustrial
This was a radical reinterpretation in so far as employment problem? Is there indeed any
postindustrial society would do away with class problem? And, if new class configurations are
altogether. Also Lipset has now embraced this emerging, what is driving them? Besides the
position (Clark and Lipset, 1991; Clark, Lipset, radical optimists, such as Lipset et al., who ba-
and Rempel, 1993). sically deny that there is any problem worthy of
Today’s visionaries find much less cause for discussion, the reigning answers all fall back on
optimism; their projections are most likely to the great ‘equality-jobs’ trade-off. According to
range from the sombre to the outright gloomy. mainstream analysis, exclusion in Europe and
The sombre view insists that little of substance inequality in America are two sides of the same
has changed (Wright, 1989; Erikson and coin, namely the inevitable consequence of tech-
Goldthorpe, 1992; and Hout et al., 1993); that nology and the new global economy. . . .
the cleavages of the past remain pretty much
intact. The gloom comes from those who see
a new era of polarization. American or British The Dilemmas of Globalization and
observers see a world with a ‘declining middle’, Technological Change
job polarization, and a new underclass (Harri- Global trade and capital mobility are often
son and Bluestone, 1988; Jencks and Peterson, seen as the culprits of de-industrialization. The
1991; Levy, 1988; Burtless, 1990). Europeans, Asian tigers can produce steel, scooters, stereos,
in contrast, see a two-thirds society with social or shoes much cheaper than Europe or North
exclusion, marginalization, and outsider classes America. Even if their regimes are not author-
(Van Parijs, 1987; Offe, 1985; Esping-Andersen, itarian, unionism is weak and labour discipline
1993; Brown and Crompton, 1994). arguably high. Regardless, they hold a massive
What the new pessimists see, in brief, is the competitive edge in terms of wage costs in
possibility of a resurgent proletarian underclass mass-production industries as well as in many
and, in its wake, a menacing set of new ‘class services. Less-skilled workers in advanced econ-
correlates’. The transAtlantic difference of accent omies are no longer cushioned by Keynesianism

Gøsta Esping-Andersen. Social Foundations of Postindustrial Economies, pp. 147–160, 168. Copyright © 1999
by Oxford University Press. Reprinted by permission of Oxford University Press.
 Esping-Andersen—Social Foundations of Post-Industrial Economies • 1079

or import protection. They must now compete Golden Age capitalism could absorb masses
in a global labour market. of low-skilled workers on simple assembly-line
We might ask ourselves what has de facto production, churning out mass-production
changed to make equality and social protection goods for which there was massive demand. It
problematic now, when once a strong welfare is these kinds of jobs that are rapidly disappear-
state was viewed as the precondition for success- ing within the advanced economies and, as we
ful international performance? It was precisely know, virtually all net new job growth will have
due to their extreme degree of international to come from services.
vulnerability that the small European econo- That simple, routine industrial jobs are van-
mies spearheaded strong worker protection and ishing must be considered a ‘Paretian’ welfare
welfare guarantees after World War Two. Global gain: the ‘South’ benefits from taking over sim-
trade competition is not novel to Belgium or ple mass production manufacturing, thus cre-
Denmark; it is to the large economies, such as the ating domestic jobs and wealth and paving the
American or British, or the erstwhile protected way for its own process of de-ruralization and,
ones, such as Australia. Perhaps the real problem eventually, post-industrialization; the ‘North’
is that the late-comers to globalization have cho- gains by eliminating its most unpleasant jobs,
sen the ‘low road-low wage’ strategy of compe- but only in so far as these are being substituted
tition, thus forcing the hand of the vanguards? by more pleasant and better-paid jobs. The op-
Regardless, the facts point to technological timistic postindustrial theorists believed this to
change as the more potent source of falling de- be the case. Current opinion is sceptical, and the
mand for less-qualified workers. Employment data seem to confirm it. We must therefore ex-
lost due to trade competition from the ‘South’ amine more closely the workings of the service
(i.e. Asia or the Third World) is rather trivial in economy.
comparison with the volume of job loss due to
‘structural change’. In a few countries (France
and Japan), jobs gained from exporting to the Dilemmas of the New Service
‘South’ actually outnumber those lost to import Economy
penetration. Only the UK and the USA, two The decline of industrial employment began in
countries traditionally dominated by compara- earnest in the 1980s. Between 1979 and 1993,
tively low-skilled, mass-production industry, ex- the OECD countries lost an (unweighted) av-
hibit a substantial loss of domestic employment erage of 22 per cent of their manufacturing
due to competition from Asia or elsewhere. jobs. Some countries (Belgium, France, Nor-
The really powerful impulse, then, may come way, Sweden, Spain, and the UK) have been
from rapid structural and technological change hit especially hard, with a net loss between
whether or not this was initially propelled by one-third and one-half. Such magnitudes echo
global trade. Whatever the root cause, the em- post-war de-ruralization. But today’s equivalent
ployment effect ends up being fairly similar: to the assembly lines, namely low-end tertiary
it raises the returns to education and reduces employment, has greater difficulty absorbing
demand for lower-skilled and less-experienced the joint impact of industrial job losses, wom-
workers. Hence the unfavourable labour market en’s rising labour supply, and the baby-boom
position of the unskilled in general, but also of cohorts. Why should this be so? In this chap-
youth and women to the extent that they lack ex- ter, I focus on three dilemmas. The first derives
perience and practical skills. In either case, there- from a rarely recognized, inherent characteristic
fore, we seem unable to dodge the evil choice of services themselves: the more we expand the
between heightened pay and job inequalities, on tertiary labour market, the larger is the share of
one hand, or unemployment and exclusion, on low-skilled services. The second comes from the
the other hand. well-known ‘Baumol cost-disease’ problem in
1080 • Part X: The Future of Inequality

services (Baumol, 1967). And the third derives Table 124.1. Industrial and Postindustrial Occupa-
from households’ economic choices and, in par- tional Hierarchies
ticular, from women’s choice to work for pay.
The Industrial Hierarchy The Service Hierarchy
Today, almost all net job creation occurs in
services but even with bouyant growth, their Managers and executives Professionals
ability to absorb masses of redundant industrial Administrators, Semi-professionals,
supervisors technicians
workers cannot be taken for granted. Many, like
Skilled manuals Skilled service
business and health services, are skill-intensive, Unskilled manuals Unskilled service
and the more routine and labour-intensive ser-
vices (such as social care, waiting, or personal
services generally) often require a modicum of self-employed and entrepreneurs. Omitting
social or cultural skills that a redundant steel self-employment may actually be problematic be-
worker is unlikely to possess. Labour-intensive cause it is becoming one means of gaining access
services, be they in the public or private sector, to the labour market when the supply of ‘regular’
are very female-dominated; not merely because employment contracts is scarce. This classifica-
they are easy-entry jobs, but also because they tory exercise must be seen as an heuristic device.
typically represent a marketized version of con-
ventional domestic tasks. In any case, services
and female participation have grown in tandem. Lousy Jobs or Outsiders?
To better understand contemporary employ- Our present preoccupation with the equality-jobs
ment dilemmas, we need to come to grips with trade-off may have blinded us to a second di-
the logic of services. First, what precisely is the lemma, i.e. professionalization with exclusion
service economy? Second, what drives it? or full employment with job polarization. The
highly professionalized scenario depicted by
postindustrial optimists can be attained only at
Service Occupations the price of massive exclusion, while a massive
Apart from the pioneering work of Renner growth of services is likely to produce a mass of
(1953) and, more recently, Bell (1976) and ‘lousy’ jobs. We need therefore to gain a better
Goldthorpe (1982), little attention has been paid picture of postindustrial job trends.
to occupational hierarchies in services. Since they The postindustrial pessimists, like Braverman
were primarily concerned with the new knowl- (1974), err on one count: tertiarization undoubt-
edge class, the lower orders have been largely edly implies occupational upgrading. Over the
ignored. As in Braverman (1974) and Erikson 1980s, professional-technical jobs rose 3–4 times
and Goldthorpe (1992), they are simply lumped as fast as employment overall in Belgium, Ger-
together with manual, industrial workers. many, Sweden, and Japan, and an astounding 7
Such superimposition is not necessarily war- times as fast in France. Their relative growth was
ranted. Even low-end service work is in typically less spectacular in Canada and the United States
less strict command hierarchies, it is less difficult (1.7 and 1.9, respectively), two nations with un-
to monitor, and orders will come more from cus- usually strong aggregate service expansion. Indi-
tomers than from bosses. Most servicing work cations are that the professional bias is inversely
is quite individualized. For heuristic reasons, it related to service economy growth. This comes
may therefore be useful to distinguish between out in Table 124.2.
the traditional skill and authority hierarchy The statistical relationship is in fact quite
of ‘industrial society’ and that of the servicing strong: for each additional percentage point of
society. In Table 124.1, I propose one such service job growth, the professional ratio declines
classification. by 1.2 points.
The two hierarchies exclude agricultural Since we must rely almost exclusively on
occupations and, more importantly, also the services to furnish new jobs and thus reduce
 Esping-Andersen—Social Foundations of Post-Industrial Economies • 1081

Table 124.2. Service Growth and the Professional-Tech- On some counts there is convergence.
nical Bias of Job Growth, 1980–1990a Manual workers decline, and the occupations
that constituted the major source of post-war
Ratio of
Annual Service Professional to ‘middle-class growth’, such as managers, clerical,
Growth Total Growth and sales jobs, are now stagnant. Today’s growth
is undisputably dominated by professionals and
Australia 3.3 1.8
Canada 2.8 1.7
semi-professionals. This is all more or less con-
UK 2.1 2.8 sistent with the prognosis of Bell (1976). Ac-
USA 2.7 1.9 cordingly, chances of upward class mobility in
Denmark 0.8 2.0 postindustrial societies will depend primarily on
Sweden 1.2 4.3 how much the skilled and professional service
Austria 1.9 2.5
occupations grow. On the other hand, if the
Belgium 1.4 3.8 objective is full employment there will also be
France 1.6 7.3 substantial growth of less-qualified service jobs,
Germany 1.8 3.1 something that Daniel Bell did not envisage.
Japan 1.2 3.8 A postindustrial paradise would combine full
a
employment with a slim but highly skilled indus-
Years vary according to data availability.
Source: ILO, Yearbook of Labour Statistics (various years). trial labour force in unison with a professionally
dominated service economy. No country comes
unemployment, the contours of postindustrial even close to this. In Germany, which often epit-
employment pose an unpleasant trade-off: we omizes skill-upgrading (see, for example, Kern
will come close to the professionalized world of and Schumann, 1984), skilled manual workers
Daniel Bell, but only when services are relatively have actually disappeared faster than the un-
stagnant. And this entails unemployment and la- skilled—although from the 1980s onwards, this
bour market exclusion. Vice versa, we can min- has reversed. None the less, Germany’s services
imize exclusion but here we must pay the price are heavily professional and ‘non-proletarian’.
of accepting a less favourable, more ‘proletarian’ High labour costs crowd out private social ser-
mix of service occupations. This comes out quite vices; low levels of female participation make
clearly when we examine the occupational hier- them less demanded. One reason for Germany’s
archies in greater detail. lower female employment rates is the absence of
Table 124.3 divides the active population, c. public care services. And, besides costs, one rea-
mid-1980s, into our ‘industrial’ and ‘postindus- son for Germany’s stagnant personal services is
trial’ hierarchy and, as somewhat of a residual, it that families engage in more self-servicing.
provides an estimate of the excluded, ‘outsider’ As a consequence, German postindustri-
population (here measured as a percentage of alization provides no substantial employment
the working-age population). The three-way outlet for either laid-off manual workers or
comparison between Germany, Sweden, and the less-qualified women. Instead, both these groups
United States is chosen because these nations have been managed primarily through labour
are prototypical examples of three types of wel- reduction strategies: early retirement or unem-
fare regimes—conservative, social democratic, ployment insurance for industrial workers, and
and liberal, respectively. But they also represent discouragement of female careers. Germany may
three distinct employment trajectories: Ger- come closest to the postindustrial ideal as regards
many stands as a best-case version of European job structure, but this comes at the expense of
jobless growth; Sweden is the epitome of the employment exclusion.
Nordic welfare state-led model of service expan- Sweden does conform to the industrial
sion; and the United States is, par excellence, the skill-upgrading thesis but suffers from strong
leading example of unregulated, market-driven polarization in the service occupations. The
employment. phenomenally high levels of female employment
1082 • Part X: The Future of Inequality

Table 124.3. The Distribution of Occupations and the Size of the Outsider Population (Percentage
Change, 1960–1980s, in Parentheses)

Germany Sweden United States

1985 1985 1988

‘Industrial’ society
Unskilled manual 16.5 (−0) 12.4 (−42) 14.4 (−33)
Skilled manual 17.3 (−32) 15.2 (−18) 8.7 (−34)
Clerical and sales 29.6 (+30) 18.6 (+16) 28.3 (+21)
Managersa 4.5 (+36) 4.0 (−15) 9.1 (+17)
Total 67.9 (−0) 50.2 (−18) 60.4 (−8)
‘Servicing’ society
Unskilled service 4.5 (−48) 16.9 (+78) 11.7 (−0)
Skilled service 5.0 (+194) 4.4 (−0) 6.6 (+57)
Professional-technical 17.3 (+121) 21.9 (+89) 18.1 (+56)
Total 26.8 (+47) 43.2 (+70) 36.4 (+31)

‘Outsider’ society
Not employed 35.2 16.7 29.5
Long-term unemploymentb 46.3 5.0 5.6

a
Includes also self-employed (non-professional).
b
As a percentage of all unemployed (1990).
Note: The table excludes primary sector occupations.
Sources: Recomputations from Esping-Andersen (1993: Tables 2.3 and 2.4); and OECD (1992; 1994).

go hand-in-hand with huge numbers of (often industrial workers were all made possible by the
low-skilled) public-sector jobs. And Sweden’s sheer volume of job growth.
formidable investment in retraining and active The juxtaposition between Germany’s more
manpower programmes has, until the crash in favourable skill mix, and the large share of un-
the 1990s, helped recycle redundant industrial skilled servicing jobs in both Sweden and the
workers into alternative jobs. The combination United States highlights the basic dilemma:
of these two factors implies few labour market today, substantial employment expansion re-
outsiders, be it in the form of early retirees, mass quires heavy growth in consumer and social
unemployment, or discouraged women. services, both characteristic of a large unskilled
The United States, finally, exhibits skill po- quotient the more they grow—hence an appar-
larization in both industry and services. It is ent trade-off between either joblessness or a mass
an economy biased towards unqualified jobs. of inferior jobs.
De-industrialization has reduced manual em- The real issue lies in the correlates of class.
ployment equally between the skilled and un- It makes a notable difference whether inferior
skilled; declining wages have helped maintain jobs provide inferior welfare. In Sweden, they
their relative share in industry and services. Of are relatively well-paid and secure welfare state
course, our data disguise the fact that, in ab- jobs; indeed, private sector ‘Mcjobs’ hardly exist.
solute terms, the number of net new jobs has While relative pay in Swedish welfare service
risen phenomenally. Hence, the absorption of jobs has been in modest decline, job security has
immigrant masses, the huge growth in women’s not. A notable aspect of Sweden’s gargantuan
employment, and the relocation of redundant welfare spending cutbacks in the 1990s is that
 Esping-Andersen—Social Foundations of Post-Industrial Economies • 1083

Table 124.4. The Growing Productivity Gap: Percentage Change in the Ratio of Service-to-Manufacturing Produc-
tivity, 1983–1995, for Select Countriesa

Ratio of Personal Ratio of Business Ratio of Social

Ratio of Restaurants Services to Services to Services to
to Manufacture Manufacture Manufacture Manufacture

United States −7 −47 −55 −62

Denmark −35 −30 −11 −2
Sweden −82 −60 −71 −54
France −56 −41 −38 +4
Netherlands −40 −19 −38 −28
a
Productivity is measured as GDP (in constant prices) per person employed.
Source: OECD Data File on Services Statistics on Value Added and Employment.

public employment is safeguarded. In the United There are three possible responses to the
States, the low-end service workers are mainly productivity gap. One would be to allow labour
in the private sector, typically poorly paid and costs to adjust to productivity differentials—the
excluded from occupational welfare entitlements market-clearing approach. The downside of un-
and basic job security. In Sweden, the concen- regulated wage adjustment is that this would
tration of women in low-end servicing jobs is push earnings in the more stagnant services to-
extreme (women account for roughly 80 per wards zero. Many services, such as music con-
cent of the total), while in America the bias is certs, psychotherapy, or aged care, are capable of
tendentially ethnic—Hispanics—rather than almost no productivity enhancement (at least not
gender-based (Esping-Andersen, 1990; ch. 8). without a quality loss), and they would therefore
In summary, there are undisputably forces most likely disappear, simply because no one
that pull economies away from job homoge- would be willing to perform them at earnings
neity and wage equality, and towards polariza- that correspond to relative productivity. The
tion. It is therefore doubtful whether the great potential limit to the market-clearing strategy is
‘equality-jobs’ trade-off overshadows everything poverty: aggregate service consumption may, on
else. It is, as we shall now see, more likely that the one hand, be stimulated by low prices but if,
trade-offs overlap. on the other hand, this means that a large pop-
ulation mass is employed at poverty-level wages,
aggregate demand will suffer.
The Cost-Disease and Service The second and, in most countries, typical re-
Expansion sponse has been to allow service earnings to fol-
Globalization, new technologies, and tertia- low general wage developments in the economy.
rization all seem to produce heightened labour This may matter little for engineering, product
market polarization. The productivity lag within design, or financial consultancy, but it risks
many services only helps reinforce this problem. pricing out of the market labour-intensive, low
As originally argued by Fourastiér, and subse- value-added activities such as personal services,
quently formalized by Baumol (1967), services entertainment, or private daycare. Both the Nor-
face a long-run ‘cost-disease’ problem. This will dic countries and Continental Europe have in
come about because, in the long haul, produc- common an institutional framework which has
tivity grows on average much faster in manu- this effect: high wage costs across the board and a
facturing than in (most) services. As shown in more compressed earnings distribution. Hence,
Table 124.4, most services lag—sometimes dra- personal services grow very sluggishly, if at all,
matically—behind manufacturing. and many services become almost extinct. To
 1084 • Part X: The Future of Inequality

Table 124.5. The Cost-Disease Illustrated by Laundry Servicing: The Comparative Cost of Washing and Ironing
One Man’s Shirt, Employment, and Self-Service Equipment, mid-1990s

Working-Age Population
Cost ($US) 1996 per Laundry Workera Washing-Machinesb

Denmark 5.20 3,500 74

Sweden 4.25 727 87
France 4.50
Germany (West) 3.70 667 88
Italy 3.25 n.a. 96
Spain 3.90 905 87
UK 2.20 750 87
USA 1.50 391 75

a
Data for the UK are 1993; for the USA, Spain, and Sweden, 1990; for Germany, 1987.
b
Percentage households equipped with a washing-machine, 1991.
Sources: Employment: United States Statistical Abstract, 1995, Table 668; Sveriges Statistiska Aarsbok, 1993, Table 204; German Cen-
sus of 1987, Part 2, Fachserie 1, Haft 10; Government Statistical Office, Employment Gazette, Historical Supplement, no. 4, October
1994; Ministero de Economia y Hacienda, Prospectiva de las Occupaciones y la Formacion en la Espana de los Noventa, 1991, App. Table
2. Washing-machines: United States Statistical Abstract, 1995, Table 1376. Prices: own data collection in Italy, Spain, and the United
States, and special thanks to Francis Castles, Kevin Farnsworth, Joakim Palme, Jon Kvist, Karl Ulrich Mayer, John Myles, Kari Salmin-
en, and Lucy Roberts for help on price and employment figures.

illustrate the problem, let us examine a prototyp- uniquely biased towards welfare state jobs and
ical labour-intensive service: laundries. away from market services. In the United States,
As we see in Table 124.5, time-saving house- the lion’s share of caring services is marketized
hold goods (washing-machines) are almost while, in Continental Europe, it is basically
universal everywhere. What varies is the cost familialized. The limits of subsidization are a
of out-servicing and, therefore, laundry jobs. question of political economy: of the balance
Laundries literally clutter American streets be- of political power and citizens’ willingness to be
cause they are cheap. They are almost impossible taxed.
to find in Copenhagen and Stockholm because The cost-disease problem and the tension
they are priced out of the market. between professionalized or more polarized
In a service-led economy, this kind of employment growth are clearly closely related.
cost-disease translates easily into jobless growth. Where the cost-disease in services is severe, and
But there is, of course, a third possible solution, public servicing is scarce, the size of the ‘outsider
namely to subsidize services—either directly via classes’ is likely to be large; where it is countered
government production, or indirectly via sub- by low-wage labour markets or by public-service
sidies to consumers. All advanced nations have provision, exclusion will be less of a problem
to a greater or lesser degree adopted the subsidy than growing occupational polarization. It is
strategy, especially for vital collective goods such this nexus which defines the varying employ-
as health and education, or for culturally valued ment scenarios within the three principal welfare
goods such as opera and theatre. The European regimes.
nations stress direct public provision, the United
States favours tax-subsidized private provision.
The uniqueness of Scandinavia lies in govern- The Micro-Foundations of
ment’s huge role in furnishing labour-intensive, Postindustrial Employment
and otherwise unaffordable, care services to As we leave behind us the era of mass produc-
families. Hence, Scandinavian tertiarization is tion, there is one thing that has not changed.
 Esping-Andersen—Social Foundations of Post-Industrial Economies • 1085

The household still remains the sine qua non of misunderstanding in much contemporary
how much, and what kind of, employment will economics. It is too often assumed that, since
grow. How much families save, consume, or pro- they are largely protected from international
duce affects the probability that their members competition, services can provide a safe haven
will find themselves employed or unemployed. of employment—service workers do not com-
Household market purchase is subordinate pete with, say, Malaysians (except in the case
to three principal factors: levels of income, rel- of mass-immigration). True, in most econo-
ative prices, and time constraints. But also the mies the lion’s share of services are sheltered
distribution of riches matters greatly. A very from global competition. However, they face
skewed income distribution may result in a an even more ferocious competitor, namely
‘Latin American’ consumption scenario: a nar- family self-servicing. And, as Gershuny (1978)
row, hyper-serviced élite being waited upon by has stressed, this competition stiffens as house-
a mass of impoverished servants. At the other holds acquire time-saving household machinery
extreme, a very egalitarian income distribution, (washing-machines, dishwashers, or microwave
as in Sweden, may very well impede mass con- ovens). Yet, as we saw, Americans are massive
sumption of market services because equal wages consumers of laundries, Europeans wash and
mean that service costs will be high. iron at home.
Rising incomes might also push demand The problem is that the marginal cost of
in the direction of ‘positional’ services, that is, services such as laundry, daycare, or home-help
services that are bought for their status value. services to the elderly affects not only wom-
Following the argument of Hirsh (1976), the ‘de- en’s ability to pursue careers, but also society’s
mocratization’ of consumption means that tradi- welfare-employment nexus. As women’s edu-
tional status-enhancing goods, like automobiles cational attainment increasingly matches (and
or cellular phones, eventually lose their prestige surpasses) males’, the social cost of housewifery
value. Hence, as mass consumption spreads so is lost human capital, productive potential, and
will demand for ‘positional’ goods (Armani suits jobs. The high costs of out-servicing in Europe
or Porsches) and services (haircuts from ‘Chez compels families to self-servicing which trans-
Pierre’, rather than the local barbershop). lates into fewer market services.
Secondly, household demand is sensitive to Here we come to the third, fundamental con-
price relativities (the Baumol cost-disease prob- dition, namely households’ time constraints. The
lem). Often the lure of Third-World travel lies in emerging ‘postindustrial’ household types, be
the amazing amount of servicing that our ‘tourist they single person, single parent, or dual-earner
dollar’ can buy. Price relativities in services are households, have one thing in common: a scar-
intimately linked to wage differentials and, as we city of time, great difficulties in harmonizing
have seen, high wages in low-productivity ser- paid employment with family obligations, do-
vices might price them out of the market. mestic duties, and leisure.
Of course, where a service is consumed for The trade-off between the burden of domes-
its ‘positional’ value the logic of relative prices tic unpaid work and paid employment can be
is altered, and may even operate in reverse. It is quite severe. Since it is unrealistic to expect a
possible that status-seeking clients choose ‘Chez major relaxation of this trade-off through male
Pierre’ precisely because he is expensive. The substitution (husbands do not generally com-
status-hungry may follow outrageous pricing pensate significantly with more unpaid work to
practices as their main signal. offset wives’ declining hours), the solution must
Aside from the more esoteric world of posi- come from outside the household itself. This
tional goods, the cost-disease problem should means either the market or the state.
be most acute in low-skilled, labour-intensive The employment potential of new house-
services that compete head-on with household hold forms comes out in consumer expenditure
self-servicing. And herein lies a fundamental data. In a pioneering study, Stigler (1956: 88)
1086 • Part X: The Future of Inequality

suggested that restaurant employment should states, Continental Europe and East Asia with
rise with the rate of women’s labour force par- familialism and corporatist social insurance, and
ticipation. Today, the dual-earner family norm the Anglo-Saxon world with targeted assistance
is much stronger and, in some countries, almost and maximum markets? There is certainly no
universal. This should show up not only in fam- want of answers. Some point to nation-building
ily consumption behaviour, but ultimately also traditions and church–state conflicts, some to
in service-sector job growth. . . . etatiste initiative, and some to the configuration
Job growth in services is, accordingly, caught of class power. These are all compelling expla-
between rival forces. There are, on the one hand, nations but they beg the question of why, in
the constraints that come from the inherent fea- the first place, did state building move in one
tures of tertiarization or globalization and, on or another direction? Why was working-class
the other hand, the potential catalyst that comes unity and mobilization so thorough in Sweden
from changing family life. If changing household and not in America? And why, once formed and
forms can have positive employment effects, our forged, did the institutions they built last and
focus should logically turn to which conditions then reproduce themselves so successfully? The
would help maximize just such an effect. Women popular answer in latter-day social science is that
with careers are likely to reduce fertility, unless institutions, once forged, are overpowering and
the tension between job and children is eased. If, rarely permit the kind of sweeping change that
then, access to child care is one constraint, society would be observable to the human eye. We may
may find itself locked into a low-fertility equilib- live in a world of path dependencies where so-
rium as seems to be the case in Southern Europe. cial creations, once cemented, are incomparably
And this will affect long-term job and growth tougher than the city of Troy. Very well, but who
prospects because, after all, aggregate demand creates and sustains the hegemony of whatever
depends on the size and growth of our popula- path dependency that obtains?
tion. Hence an ample social service infrastructure Economists often draw large benefits from
is one precondition for service-job growth. basic simplicity. Their homo oeconomicus is cer-
The alternative scenario is that women are tainly not believable, but he can be a convenient
discouraged from economic activity. In this tool with which to proceed and eventually make
case, of course, a beneficial job spiral is blunted. things more complicated. I shall, for the mo-
The service economy may flourish with more ment, strive for similar minimalism. Sociologists,
inequality but also with two-earner households. of course, crave more complexity, and one Homo
In either case, we face institutional constraints. will therefore not do. Let us allow for three ideal
Some welfare regimes, and some modes of reg- typical homines: Homo liberalismus, Homo fami-
ulating labour markets and the distribution of lius, and Homo socialdemocraticus.
welfare constrain or ease more than others. It is Homo liberalismus resembles Mister Econom-
to this that we turn in the next section. ics because he follows no loftier ideal than his
own personal welfare calculus. The well-being of
others is their affair, not his. A belief in noble
Recasting Welfare Regimes for a self-reliance does not necessarily imply indiffer-
Postindustrial Era ence to others. Homo liberalismus may be gener-
If there is one great question that unites all wel- ous, even altruistic. But kindness towards others
fare state researchers, it is this: why is it that na- is a personal affair, not something dictated from
tions respond so differently to a set of social risks above. His ethics tell him that a free lunch is
that, all told, is pretty similar whether you are an amoral, that collectivism jeopardizes freedom,
American, a Spaniard, or a Swede? that individual liberty is a fragile good, easily
Why is it that Scandinavia responds with sabotaged by sinister socialists or paternalis-
social democracy and comprehensive welfare tic authoritarians. Homo liberalismus prefers a
 Esping-Andersen—Social Foundations of Post-Industrial Economies • 1087

welfare regime where those who can play the in individual, personal empowerment. Collectiv-
market do so, whereas those who cannot must ism is not pursued for its own sake, but in order
merit charity. to bring out the utmost in each and every indi-
Homo familius inhabits an altogether differ- vidual soul. That is why collective solutions are
ent planet. He abhors atomism and impersonal- always the best. But his individualism is caution-
ity and, hence, markets and individualism. His ary because no one should be granted favours,
worst enemy is the Hobbesian world of elbows, advantages, special recognition. Homo socialdem-
because self-interest is amoral; a person will find ocraticus must constantly live with a moral clash
his equilibrium when he puts himself at the between individualism and conformity. He loves
service of his family. Freedom, to Homo fami- the idea that we all be equally endowed; he hates
lius, means that he and his kin are immunized the idea that someone may rise above others,
from the ceaseless threats that the greater world particularly above himself. Solidarity is therefore
around him produce. He is not a go-getter with fragile if someone desires to move beyond the
an irresistible urge to challenge the world around common denominator. Homo socialdemocraticus
him. He is a satisficer, not maximizer, because is fully convinced that the more we invest in the
what really counts is stability and security; a job public good, the better it will become. And this
for life in the postal service is heaven on earth; will trickle down to all, himself especially, in
it will guarantee him and his kin a good life, se- the form of a good life. Collective solutions are
curity and, incidentally, also the means to land therefore the single best assurance of a good, if
a postal job for his daughters and sons. Both perhaps dull, individual life.
Homo familius and femina familia see patriarchy We all combine the instincts of these ideal
as a good thing. The family would be unruly typical homines; we all have moral conflicts.
without authority and, in any case, respect and And all societies combine them in one mix or
status is due to the father, on whose shoulders so another. Sweden has its Homines liberalismi (al-
much responsibility weighs. To them, the family though many emigrate to Monte Carlo), and
is the unrivalled source of solidarity and com- America its Homines socialdemocratici (many of
munity because it alone knows what its members which show up regularly in Sweden). But how
need. Alas, the family is fragile since so many did a sufficient mass manage to profile itself in
individuals must depend on the pater familias. collective expression, and thereby sway society
Homo familius is therefore quite happy with the towards its preferred welfare regime?
idea that the state—or some higher body—elim- Genetics clearly do not create preferences and
inates whatever risk of misfortune may arise. Yet, beliefs. What might account for this is society
higher bodies are there to service the family, not itself with all its institutions, incentive systems,
to command his loyalties. Homo familius wants a and inscribed norms of proper conduct. Society’s
welfare regime that tames the market and exalts ‘median Homo’ is created, and this is obviously
the virtues of close-knit solidarities. where labour movements or other collective ac-
Homo socialdemocraticus is, like a boy scout or tors have played a key role. If, originally, a labour
good Christian, inclined to believe that he will movement garnered its strength (or failed to do
do better when everybody does better. Doing so) from the available raw material, so it also
good to others is not an act of charity and can, played a constant historical role in reactivating
indeed, be coolly calculative. Homo socialdemo- that very same raw material. Institutional path
craticus plans his life around the one basic idea dependency means that one society is likely to
that he, and everybody else, will be better off in a reproduce itself in the image of Homo socialdem-
world without want but also without free-riders. ocraticus, another in the image of Homo familius,
Society is something that we all are compelled to generation after generation. . . .
share, and so we had better share it well. Homo If core institutional traits are so unyield-
socialdemocraticus is, none the less, also a believer ing to change, it remains unlikely today—as
1088 • Part X: The Future of Inequality

always—that the contemporary welfare state allowances). Working mothers may have ‘nega-
crisis will produce an avalanche of revolution- tive productivity’ in the sense that their earnings
ary change, no matter how urgent such change (plus public subsidies) exceed their output. But
is claimed to be. Any blueprint for reform is these costs are potentially recovered via their
bound to be naïve if it calls for a radical depar- higher lifelong earnings, the smaller depreciation
ture from existing welfare regime practice. The of their human capital, and the resulting fertility
IMF is naïve when it asks that European welfare dividend.
states adopt, lock, stock, and barrel, the Chil- A fundamental ‘postindustrial’ dilemma—
ean pension system; American leftists, or emer- generally overlooked in contemporary debates—
gent democratic governments in ex-communist is that families seem no longer inclined to assume
countries, are equally naïve when they call for the costs of bearing children. As a comparison
the importation of the ‘Swedish model’. One between fertility in Southern Europe (with
thing is to redesign or reform the ways in which Japan) and Scandinavia suggests, the social price
the state delivers welfare, another to recast the of children may be high indeed. Immigration is
entire welfare regime. Academically speaking, certainly an alternative, and herein lies a major
there may exist a blueprint for a ‘win-win’ strat- reason behind North America’s (and Australia’s)
egy, for an ideal postindustrial welfare regime. younger age profile. Yet, besides being a rather
But unless it is compatible with existing welfare unrealistic scenario for Europe today, the immi-
regime practice, it may not be practicable. As I gration option poses hard dilemmas of its own.
shall argue below, optimizing welfare in a postin- If one of our chief problems is how to absorb
dustrial setting will, none the less, require radical masses of low-skilled workers, this would only
departures. . . . be compounded by Third World immigration.
The advanced economies would clearly prefer
highly educated immigrants, yet encouraging
A Third Way? such a flow can hardly be considered solidaristic
Up to a point, but not further, a ‘social dem- towards Third World nations.
ocratic’ approach might furnish the basis for Universalizing the double-earner household
a new, positive equilibrium. A politics of col- (with lots of children) holds yet another prom-
lectivizing families’ needs (de-familialization) ise of welfare improvement. When we add its
frees women from unpaid labour, and thereby greater purchasing power to its desperate search
nurtures the dual-earner household. And this for free time, it is a truly promising source of
reduces child poverty, and makes households service consumption, from restaurants and lei-
better equipped to weather the storms of flex- sure parks to child care and home-help services
ibilization, since they will usually have one for their aged parents. Thereby families create
member’s earnings to fall back on if the other jobs for waiters, park-keepers, child-minders,
is made redundant, needs temporary retrain- and home-helpers. Of course, access to child
ing, or suffers wage decline. Two-earner house- care is a pre-condition for dual-income families
holds have stronger social networks, and are less in the first place. This is exactly the point: ser-
likely to run them down if one partner becomes vices beget services; the double-earner household
unemployed. plays the role of employment multiplier. The
A social democratic de-familialization strat- employment multiplier of working mothers can
egy can reverse fertility decline if it helps em- be quite substantial, especially in those kinds of
ployed mothers square the caring-work circle services that are labour intensive.
(mainly via daycare), and if it is willing to cover Encouraging families to consume more ex-
a good part of the opportunity costs of having ternal services is therefore part of a potential
children (which means expensive maternity ‘win-win’ strategy. Yet, as we know, relative costs
and parental leave as well as generous child can be a major constraint. This is why, despite
 Esping-Andersen—Social Foundations of Post-Industrial Economies • 1089

the virtual universalization of two-earner house- guaranteeing all citizens against entrapment: no
holds, lower-end personal services in Scandi- one should find him- or herself in an omnibus
navia grow very little. They would doubtlessly with locked doors.
grow if costs were lower, but this implies greater Our search for a postindustrial welfare opti-
wage inequality. mum requires, therefore, some kind of a mobil-
This is where Homo socialdemocraticus must ity guarantee. What this will look like depends,
yield to Homo liberalismus. Substantial service in turn, on what are the chief causes of social
employment growth outside the public sector exclusion and inferior life chances. A room full
depends on flexibility and low wages, some- of academics and experts would, no doubt, draft
thing that social democracy cannot easily ac- an endlessly long list of causes. Many standard
cept. The social democratic strategy, as I have reasons, such as physical or mental handicaps,
outlined it so far, will therefore not escape the do not concern us here because they are not in-
fundamental ‘equality-jobs’ trade-off. We need herently part of the equality-jobs trade-off. Two
therefore to resolve one last problem: the very sources of substantial life chance problems stand
same ‘equality-jobs’ trade-off with which we out: one, the risks associated with marital insta-
started. . . . bility and poverty in childhood; two, inadequate
There is in truth only one way out of the skills.
impasse, namely to redefine what kind of equal- Diminishing family-induced risks calls for a
ity we desire. Homo socialdemocraticus must be standard ‘social democratic’ package of what the
convinced that we cannot aspire for all kinds of feminists call women-friendly policy: child-care
equality at once; that some inequalities can be services, incentives for mothers to work, and ad-
made compatible with some equalities. equate income maintenance to take into account
The principle of equality that must go is ex- mothers’ reduced labour supply and the cost of
actly the same that emerged—most forcefully in children. Diminishing labour market induced
Scandinavian social democracy—when welfare risks calls for a rethinking of education, training,
states sought to respond to the equality-crisis of and marketable skills.
the 1960s, namely the promise of equality for This is not the place to explore the broad
all ‘here-and-now’. In practice, this may not be issues of education. But there are a number of
so difficult a task. Homo socialdemocraticus, like basic facts that all can agree upon: the returns to
many of his rivals, has surely held any number of skills are rising, as we would expect in a world
lousy jobs in his youth. Like all Scandinavians, increasingly dominated by complex technology;
he left the parental home very early and lived for the low-skilled are in rapidly declining demand.
years on bread and water (this is what he tells Closing the skill gap is therefore an extremely
his children). Yet, he is now a respectable citizen effective way of catapulting people out of entrap-
with a respectable career. Temporary depriva- ment, of assuring good life chances. And it also
tion is unimportant if it does not affect our life pays off to society in the form of a more produc-
chances. tive workforce.
We can return to Schumpeter’s omnibus: There is clearly nothing earth-shatteringly
always full, but always with different people. novel about a call for more education. There
Everybody gets off at the next stop, or at least is virtually no government or international or-
where desired. If, like Homo socialdemocraticus, ganization today that does not advocate ‘active
we cling to a notion of equality for all, here and labour market policy’ or ‘life-long learning’.
now, we shall never resolve the fundamental di- But there is widespread scepticism about their
lemma of our times. The kinds of inequalities effectiveness. Active labour market policy with
that are inevitable in the world of Homo liber- its ‘activation’ and training programmes does
alismus can become acceptable, even welcomed, not always appear to pay off if by this we mean
if they coincide with a welfare regime capable of that the unemployed eventually find a stable,
1090 • Part X: The Future of Inequality

promising career. Too often, activation looks to lifelong low-wage employment or, possibly,
more like a temporary parking-lot, or one in- crime.
terlude in a never-ending roundabout of unem- It is this kind of result that must be elimi-
ployment, training, occasional jobs, and then nated from any kind of society if we seriously de-
unemployment again. sire an optimal welfare regime. I therefore close
Effective or not, many believe that training this chapter inviting education experts to design
is useless since there are no jobs for the newly a workable system of skilling entitlements, one
trained. In a static sense this is undeniably often that would befit an ideal postindustrial welfare
true. But they miss the basic point of the omni- regime. And I invite our political leaders to forge
bus analogy: for any given individual, skills are a new coalition of our assorted homines, one ca-
the single best source of escape from underprivi- pable of breaking the deadlock of median-voter
lege. What we are trying to resolve is a dynamic, support for anachronistic modes of welfare
life-course issue and not where to place every- production.
body today. A new welfare optimum is, in fact,
compatible with the possibility that many of us REFERENCES
will experience a spell of unpleasantness. Bairoch, P. (1996), ‘Globalization, Myths and Reali-
ties: One Century of External Trade and Foreign
What we really need to understand better is
Investment’, in R. Boyer and D. Drache (eds.),
what kinds of skills and what kind of education
States Against Markets. London: Routledge,
to promote? We are generally aware of the rising 173–93.
demand for multi-skilling, flexible adaptation, Baumol, W. (1967), ‘The Macroeconomics of Un-
and capacity for life-long learning. A worker balanced Growth’, American Economic Review,
with a good theoretical base in his vocational 57: 415–26.
training is much easier to upgrade than someone Bell, D. (1976), The Coming of Postindustrial Society.
with a high-school diploma unable to under- New York: Basic Books.
stand instructions on an aspirin bottle. This is Bjorn, N. H. (1995), ‘Causes and Consequences of
why German employers use their skilled workers Persistent Unemployment’, Ph.D. diss., Depart-
ment of Economics, Copenhagen University.
to run computerized production systems while
Braverman, H. (1974), Labor and Monopoly Capital:
in England they import engineers. We also face The Degradation of Work in the Twentieth Cen-
a reality where ‘social skills’ are more fundamen- tury. New York: Monthly Review Press.
tal than muscle because of the direct face-to-face Brown, P. and Crompton, R. (1994), Economic Re-
mode of service production. What these are is structuring and Social Exclusion. London: UCL
not easy to define, but they undoubtedly include Press.
more than the routine ‘have a nice day’. A major Burtless, G. (1990) (ed.), A Future of Lousy Jobs? The
problem in the contemporary unemployment Changing Structure of U.S. Wages. Washington:
structure is that laid-off miners and steel workers Brookings Institute.
Clark, T. and Lipset, S. M. (1991), ‘Are Social Classes
are unlikely to possess the kinds of social skills
Dying?’, International Sociology, 4: 397–410.
that sell a service.
Clark, T. Lipset, S. M., and Rempel, M. (1993),
We know from the OECD literacy studies ‘The Declining Political Significance of Class’,
(OECD, 1997) that education—even completed International Sociology, 3: 293–316.
secondary education—may guarantee very little Erikson, R. and Goldthorpe, John H. (1992), The
if it ends up producing 15 or 20 per cent Amer- Constant Flux: Class Mobility in Industrial Societ-
icans incapable of even rudimentary reading, ies. Oxford: Clarendon Press.
writing, and arithmetic. These do not even pos- Esping-Andersen, G. (1990), The Three Worlds of
sess the minimal level of qualifications needed to Welfare Capitalism. Cambridge: Polity Press.
———. (1993) (ed.), Changing Classes: Stratification
be trainable. They are a de facto human capital
and Mobility in Postindustrial Societies. London:
waste, a stratum only too obviously condemned
Sage.
 Hout and Cumberworth—The Labor Force and the Great Recession • 1091

Gershuny, J. (1978), After Industrial Society: Kern, H. and Schumann, M. (1984), Das Ende der
The Emerging Self-service Economy. London: Arbeitsteilung? Munich: C. H. Beck.
Macmillan. Levy, Frank (1988), Dollars and Dreams: The Chang-
Goldthorpe, J. (1982), ‘The Service Class, its For- ing American Income Distribution. New York: W.
mation and Future’, in A. Giddens and G. W. Norton.
Mackenzie (eds.), Social Class and the Division of OECD (1992), Employment Outlook. Paris: OECD.
Labour. Cambridge: Cambridge University Press, ———. (1997), Literacy, Skills and the Knowledge
162–87. Society. Paris: OECD.
Harrison, B. and Bluestone, B. (1988), The Great Offe, C. (1985), Disorganized Capitalism. Cam-
U-Turn, 2nd edn. New York: Basic Books. bridge, Mass.: MIT Press.
Hirsh, F. (1976), Social Limits to Growth. Cam- Renner, K. (1953), Nachgelassene Werke, iii. Wand-
bridge, Mass.: Harvard University Press. lungen der Modernen Gesellschaft. Wien: Wiener
Hout, M., Brooks, C., and Manza, J. (1993), ‘The Volksbuchhandlung.
Persistence of Classes in Postindustrial Societies’, Stigler, G. (1956), Trends in Employment in the Ser-
International Sociology, 3: 259–78. vice Industries. Princeton: Princeton University
ILO (1943), Yearbook of Labour Force Statistics. Ge- Press.
neva: International Labour Office. van Parijs, P. (1987), ‘A Revolution in Class Theory’,
Jencks, C. and Peterson, P. E. (1991) (eds.), The Politics and Society, 15/4: 453–82.
Urban Underclass. Washington: Brookings Wright, E. O. (1989), The Debate on Classes. Lon-
Institute. don: Verso.

125 • Michael Hout and Erin Cumberworth

The Labor Force and the Great Recession

Americans work for their living. For most of intensity. The “housing bubble” burst, Wall
them, having a job is an economic and moral Street stumbled, banks stopped lending, con-
imperative. The wages they earn fuel the rest of struction workers lost their jobs, sales of build-
the economy. Employment begets the spending ing materials and appliances plummeted, tax
that begets more employment. In good times, it revenues fell, governments furloughed police
is a virtuous cycle reinforcing consumer-driven and teachers, and the downward spiral spun
capitalism. Events like the financial crisis of ever lower. Federal stimulus broke the fall, but
2007 and 2008 reverse the cycle, spinning the in 2010 almost 10 percent of the labor force re-
economy downward with a momentum that can mained unemployed, and the current rate (as of
be hard to break. Job losses reduce spending, May 2012) remains over 8 percent.
which kills more jobs, reducing spending even In this chapter, our aim is to put the recession
more. into historical context, looking first at its over-
The Great Recession of 2007 to 2009 played all impact on the labor market and then at how
out these general principles of recession eco- the economic harm has been distributed across
nomics in every aspect, but with an uncommon the population by gender, level of education,

Michael Hout and Erin Cumberworth. “The Labor Force and the Great Recession,” in The Recession Trends
Report (2012), pp. 1–5. Copyright © 2012 by the Russell Sage Foundation, 112 East 64th Street, New York,
NY 10065. Reprinted with permission.
1092 • Part X: The Future of Inequality

Figure 125.1. Official Unemployment Rate, 1900–2011

Recession years



UNEMPLOYMENT RATE (%)






           
YEAR

Men Women

Source: U.S. Bureau of Labor Statistics.

and race and ethnicity. There has been much surprisingly, perhaps, the recovery, such as it is,
popular commentary suggesting that the bur- has reduced these disparities slightly.
dens of unemployment and job loss have been
widely shared across all population groups. The
shared-burden account applied in early 2009 Historical Context
but became less accurate as the recession deep- We begin by providing the larger context behind
ened. At that point layoffs at surviving firms the Great Recession. Since 1900, the American
replaced the failure of whole firms as the main economy has experienced twenty-two recessions
source of new unemployment. Typically in a and recoveries, all of which are shown in Figure
recession selective layoffs affect less-educated, 125.1. This figure plots unemployment rates
African-American, younger, and foreign-born with recession months shaded gray.
workers more than others. In 2010 less-educated, The Great Depression of the 1930s stands
African-American and younger workers saw out as the only time when unemployment ex-
their unemployment rise faster than others did. ceeded 20 percent. Almost inconceivably, unem-
Foreign-born unemployment actually increased ployment exceeded 10 percent every year from
less than that of native workers; most demog- 1930 to 1942. It was, by far, the worst economic
raphers have concluded that was because many period in the twentieth century.
Mexican immigrants returned to Mexico during Prior to the 1929 collapse, the economy grew
the recession. When new jobs were created in the and recessed in rapid succession; periods of ex-
aftermath, educated and experienced applicants pansion averaged less than two years from reces-
were favored, but enough less educated, African sion low point to recovery high point. But since
American, and young workers have found work 1960, ups have won out over downs; periods of
to lower the differences in unemployment rates growth have lasted longer while the recessions
to prerecession patterns. have been far less frequent. Macroeconomic pol-
The purpose of this chapter, then, is to pro- icy, regulatory powers to carry out that policy,
vide a broad profile of labor market outcomes and the ever-greater sophistication of the scien-
during the Great Recession and its aftermath, tific management tools available to the Federal
highlighting disparities and inequalities by Reserve Board tamed the boom-bust tendencies
gender, race, and education. We will show, of market capitalism.
perhaps not surprisingly, that the Great Reces- The recent recession of 2007 to 2009 has
sion affected vulnerable groups the most. More been exceptional in both the suddenness of the
 Hout and Cumberworth—The Labor Force and the Great Recession • 1093

economic collapse and the long duration of its how different population groups fared during
employment consequences. Unemployment this exceptional period and the halting recovery
more than doubled, from 4.5 percent to 10.6 thereafter.
percent, in twenty-six months from the onset of
recession to the peak unemployment in January
2010. In contrast, it took thirty-eight months for Gender
unemployment to double in the recession of the Men were particularly hard-hit by the recession,
early 1980s. And a year after the peak joblessness prompting some commentators to use the term
of the recent recession, the unemployment rate “man-cession.” Much of the gender difference is
had fallen only 1.5 percentage points below the explained by occupation and industry; men were
peak, whereas by January 1984 it had fallen 2.6 more heavily concentrated in the construction
percentage points below the peak. and building trades, which were hit the hard-
Unemployment is persisting not only for est by the recession. In Figure 125.2, we put the
the economy as a whole but also for individ- trend in men’s and women’s unemployment into
ual workers. In fact, long-term unemployment broader historical context.
might well be the defining difference between A drop in men’s employment ratio is to be ex-
the Great Recession and others since World War pected during a recession; it has dropped during
II ended. In the four recessions between 1977 each recession in the postwar era. The surprise
and 2001, workers were out of work an aver- in Figure 125.2 is in the recovery periods. Only
age of nine weeks at the depths of the recession. once since 1948 did men’s employment recover
In January 2010, they had been out of work an to prerecession levels—and never since 1970.
average of twenty-one weeks. The consequences Each successive recession has been like a ham-
of long-term unemployment are far greater than mer driving men’s employment down another
short-term. Research on past recessions shows notch. Men got back to work during most recov-
that the unemployed, especially those just start- ery periods, but even during the longest growth
ing their careers, bear a “scar of unemployment” periods, employment did not fully recover to
that lasts for years. They fall behind while oth- where it had been at the peak of the previous
ers gain experience, many desperately settle for growth period. In November 2010, 80 percent
work that requires less skill than they have, and of working-age men were employed, down from
few ever catch up with their peers who were not 88 percent before the recession. At no other time
unemployed. since 1948 has male employment been so low,
Since the recession ended, women’s employ- nor has it ever fallen as much as eight percentage
ment has not recovered at all while men’s has. points during a recession. Thus for men, the re-
The main reason has been sectoral. Women are cession was a particularly severe turn of a down-
concentrated in the public sector where employ- ward trend that goes back decades.
ment was sustained by stimulus spending until In contrast, the recession brought about a
the spring of 2010. Since then austerity measures reversal in the long-term trend for women’s em-
at the state and local levels have offset gains in ployment. From 1948 to early 2000, women’s
private sector employment with losses in public employment ratio rose from 33 to 74 percent.
sector employment. These disproportionately af- Recessions during that half-century led to brief
fect women because that is where you find more pauses in the upward trend, but growth peri-
of women’s jobs and fewer of men’s jobs. ods resulted in gains beyond the previous high
Calling the recent recession the “Great Re- points. The historic rise in women’s employ-
cession” thus attests to how exceptional it has ment only halted during the first decade of the
been in both the suddenness of the economic twenty-first century, and then it fell during the
collapse and the long duration of the employ- 2007-to-2009 recession. Women’s employment
ment consequences. We turn next to describing decreased four percentage points between 2000
1094 • Part X: The Future of Inequality

Figure 125.2. Employment to Population Ratio (Ages 25–54) by Gender, 1975–2011

Recession years
100

90

80

70

60
PERCENT

50

40

30

20

10

0
1975 1979 1983 1987 1991 1995 1999 2003 2007 2011
YEAR
Men Women

Source: U.S. Bureau of Labor Statistics.

and 2010, after rising three to twelve percentage Contrary to this expectation, accounts of the re-
points in every other decade since the 1950s. cession in the popular media frequently have a
Although men were especially hard hit, storyline that this recession is hurting everyone,
then, by virtue of being concentrated in the including the college-educated. The data suggest
ailing construction and building trades, the that this storyline is not totally without founda-
downturn was widespread enough that it also tion, but is misleading and overstated.
affected the economic sectors in which women Figure 125.3 shows that the risk of being
were concentrated. This in turn led to a rever- unemployed declines sharply as education rises.
sal in the long-term upward trend in women’s Prior to the recession, unemployment for peo-
employment. ple with less than a high school degree hovered
around 7 percent, while unemployment for
college graduates was only about 2 percent. As
Education unemployment spread, the rate for each edu-
Employers strongly prefer educated workers, cational category rose more or less proportion-
a point they have demonstrated over and over ally. At peak unemployment in 2010, the rate
by paying an ever-growing premium for college for people without a high school degree had
graduates, seeking H-1 visas for foreign college increased from 7 to nearly 15 percent and the
graduates, and seizing on other tactics to locate rate for college graduates had increased from 2
and hire educated workers. Thus we would to about 4.7 percent. The baseline differences
expect them to protect their college-educated were so large that proportional increases raised
workers from being laid off in the recession. unemployment most for the least-educated and
 Hout and Cumberworth—The Labor Force and the Great Recession • 1095

Figure 125.3. Unemployment Rate by Level of Education (ages 25+), 1992–2011

Recession years
16

14

12

10
PERCENT

0
1992 1995 2000 2005 2010

Less than High School Some Bachelor·s Degree

High School Degree College or More
Degree

Source: U.S. Bureau of Labor Statistics.

least for the most-educated. Even though un- between education groups, the racial and ethnic
employment rose for everyone, people without groups of Figure 125.4 all increase more or less
a high school degree bore a much greater unem- proportionally. However, since African Ameri-
ployment burden. cans and Hispanics had such a high unemploy-
ment rate before the recession, they suffered far
more in terms of absolute numbers than other
Race groups.
Like people with less education, African Amer- These results make it clear that African Amer-
icans and Hispanics are exposed to more un- icans and Hispanics were hard hit because they
employment than other groups even in good experienced a proportionate increase in unem-
times. But how have these groups fared in bad ployment off of an especially high pre-recession
times? We address this question in Figure 125.4 base. This result, important though it is, doesn’t
by plotting the unemployment rate by race and fully represent how dire the situation for African
ethnicity. Americans is. There are two additional points
This figure shows that unemployment among that should in particular be stressed. First, un-
African Americans and Hispanics was substan- employment data omit people who do not reside
tially higher than among other groups prior to in households, which is important for assessing
the recession and rose to the greatest height as racial differences because these groups differ a
the recession progressed. As with the differences great deal in the propensity to live somewhere
1096 • Part X: The Future of Inequality

Figure 125.4. Unemployment Rate by Race/Ethnicity, 1975–2011

Recession years
22
20
18
16
14
PERCENT

12
10
8
6
4
2
0
1975 1980 1985 1990 1995 2000 2005 2010

Black Hispanic White Asian

Source: U.S. Bureau of Labor Statistics.

other than a household. We know, for example, unemployment that African Americans get. It
that African Americans are at high risk of liv- follows that an African American, when con-
ing in a prison. Roughly 80 percent of prison- fronted with the dire data of figure 125.4, can-
ers would be in the labor force if they were not not count on education as providing the same
incarcerated. Because African American men are relief against the risk of unemployment that it
especially likely to be in prison, their labor force provides to other groups.
statistics are especially affected by omitting those
living in institutions. The simple upshot: the data
of figure 125.4, stark as they are, understate how Conclusions
exceptionally high black non-employment is. The Great Recession was a jobs disaster that took
The second point of caution is that education unemployment to historic heights. Any recession
has substantially less return for African Amer- brings some job loss and boosts unemployment,
icans than for other American workers. There but the number of jobs lost, the portion of the
is of course still some return: unemployment is labor force unable to find work, and the du-
lower for African American college graduates ration of layoffs and job search all hit postwar
than for African Americans with some college highs in the Great Recession. The evidence to
or only a high school diploma. But a multivari- date suggests that it is very likely that another
ate analysis of the data, not reported here, indi- major restructuring of the U.S. economy will be
cates that a college degree gives white and Asian the long-term historical significance of the Great
American workers twice the protection against Recession.
 Hout and Cumberworth—The Labor Force and the Great Recession • 1097

The popular media has often suggested that all of the prospects for a major job-creating in-
this has been a “democratic recession” in which novation are very technical and likely to favor
everyone, not just the poor and disadvantaged more-educated workers. If, however, such an in-
groups, has been hard hit. The Occupy move- novation does not present itself and economic
ment might be read as offering a slight modifi- restructuring does not happen, the United States
cation to that thesis, one that suggests the “one risks sliding into the kind of “lost decade” that
percent” has been protected and has perhaps affected Japan in the 1990s.
flourished during the downturn, whereas the
rank-and-file 99 percent has largely suffered. ADDITIONAL RESOURCES
There is no doubt some merit to this account. Gangl, Markus. 2006. “Scar effects of unemploy-
ment: an assessment of institutional comple-
But our results indicate that there are also clear
mentarities.” American Sociological Review 77(6),
winners and losers even within the “99 percent.”
986–1013.
Although there is, to be sure, widespread suffer- Jesse Rothstein, “The Labor Market Four Years into
ing in the rank-and-file labor market, we have the Crisis: Assessing Structural Explanations.”
shown that men, the less-educated, and African NBER Working Paper No. 17966. Available at:
Americans have been especially hard hit. http://www.nber.org/papers/w17966.
The duration of the Great Recession and Niall O’Higgins, “This Time It’s Different?, Youth
its differential impacts suggest that the Ameri- Labour Markets During ‘the Great Reces-
can economy will have to restructure in order sion,’” IZA Discussion Paper No. 6434. Avail-
to make a full recovery. A major innovation in able at: http://papers.ssrn.com/sol3/papers.
cfm?abstract_id=2032002.
goods production along the lines of the auto-
Rinne, Ulf, and Klaus F. Zimmermann. 2011 “An-
mobile and computer innovations of 1983 and other Economic Miracle? The German Labor
1984 would be very welcome. No expert can Market and the Great Recession.” IZA Discus-
say with confidence where innovation will come sion Paper No. 6250. Available at: http://www.
from—a popular new device, a bio-technology iza.org/en/webcontent/publications/papers/
breakthough, green energy. Noteworthy is that viewAbstract?dp_id=6250.
1098 • Part X: The Future of Inequality

Post-Socialism

126 • Victor Nee

Post-Socialist Stratikcation

The transitions from state socialism in China, incentives and thereby establish the constraints
Russia and Eastern Europe provide natural ex- within which rational actors identify and pursue
periments involving change in the stratification their interests (Nee and Ingram 1998; Cook and
order on a scale reminiscent of that experienced Levi 1990). It builds on Polanyi’s (1944; North
in the West during the rise of capitalism. Com- 1977, 1981) insight that economies are embed-
parative stratification research builds on mod- ded in definite institutional arrangements. In
ernization theory in assuming that industrial Polanyi’s view, the problem with neoclassical
development, whether socialist or capitalist, economics is that its formal models assume the
leads to convergence (Lipset and Zetterberg existence of a market economy. Yet compara-
1959; Treiman 1970; Grusky and Hauser 1984; tive studies of human societies, past and pres-
Hauser and Grusky 1984). Institutional the- ent, have demonstrated a variety of institutional
orists, however, are skeptical about theories of forms giving rise to economies organized around
convergence driven by industrialism. If conver- fundamentally different operating principles
gence pointed the way to state socialism’s future, (Polanyi 1957). Hence, the movement of goods
the sweeping measures to institute a market and services in an economy is not simply a prod-
economy in industrially developed Eastern Eu- uct of the aggregation of individual maximizing
rope and Russia would not have been necessary. behavior, as assumed in neoclassical economics.
Rather than focusing on the effects of industrial In societies integrated by redistribution, goods
growth, institutionalists insist that research on and services are collected and distributed from
state socialism needs also to take into account a center in accordance with the customs, regu-
underlying differences in institutional forms. lation, ideology, and ad hoc decisions of those
Such a focus on alternative institutional forms social groups that hold redistributive power. Pro-
that provide a deep structure for economic ac- ducers pass on a larger share of the economic
tion underlies my argument that the shift from surplus to the state precisely because they have
redistribution to markets gives rise to different less bargaining power over the terms of ex-
mechanisms of stratification. change than they would in a market economy.
New institutionalist theory in sociology Instead, the state requisitions their products
maintains that institutions shape the structure of and compels them to work for prices and wages

The ideas, issues, and theories considered in this brief commissioned piece are examined in greater depth in the
following publication: Victor Nee, “The Emergence of a Market Society: Changing Mechanisms of Stratifica-
tion in China,” American Journal of Sociology 101:4 (January 1996), pp. 908–949, published by the University
of Chicago Press. The article printed here was originally prepared by Victor Nee for Social Stratification: Class,
Race, and Gender in Sociological Perspective, Second Edition, edited by David B. Grusky, pp. 846–852. Copyright
© 2001 by Westview Press.
 Nee—Post-Socialist Stratikcation • 1099

fixed by central bureaus (I. Szelényi 1978). As state (North 1981; Nee 1992). Such action at the
a consequence, the institutional logic of redis- margins is cumulative and gradually results in
tributive economies differs substantively from transformation of the institutional environment.
market economies where goods and services are Market transition theory (Nee 1989, 1991,
exchanged directly between buyer and seller (Po- 1996) maintains that as power—control over
lanyi 1957; Kornai 1980), and from economies resources—shifts progressively from political
based on reciprocity where trust and cooperation disposition to market institutions, there will be
in local social orders allow for balanced exchange a change in the distribution of rewards favoring
(Homans 1950; Sahlins 1972). Redistribution, those who hold market rather than redistributive
market, and reciprocity, as alternative institu- power. Compared to nonmarket allocation, mar-
tional forms, incorporate different structures of ket exchange enhances the bargaining power of
incentives and constraints and therefore distinct producers. Incentives are improved as producers
parameters of choice. Elements of each coexist retain a greater share of the economic surplus.
in every society, at various levels, but Polanyi Opportunities for gain depend less on the per-
claimed that only one can constitute the dom- sonal discretion of cadres in positional power.
inant integrative mechanism of an economy. Consequently, the growth of market institutions
In this sense, the “economy” is not narrowly (i.e., labor markets, subcontracting arrange-
economic but is embedded in customs, social ments, capital markets, and business groups)
norms, laws, regulations, and the state. Funda- causes a decline in the significance of socialist
mentally, differences in institutional framework redistributive power even in the absence of fun-
are reflected in contrasting structures of property damental change in the political order. In sectors
rights (Pryor 1973). of the socialist economy where a decisive shift
The transition from one dominant insti- to markets has occurred, officials are less likely
tutional form to another entails remaking the to gain a dominant advantage from positional
fundamental rules that shape economies, from power in the state socialist redistributive bu-
formal regulations and laws to informal con- reaucracy. Contingent on continuing allocative
ventions and norms. The shift to markets—well control over key factor resources, their relative
underway but far from complete in postsocialist advantage declines as a function of the extent
societies—involves changing structures of op- to which markets replace redistribution as the
portunity (Merton 1949). Whereas opportuni- coordinating mechanism of the economy.
ties for advancement were previously centered In the state socialist redistributive economy,
solely on decisions made by the redistributive officials act as monopolists who specify and en-
bureaucracy and within the economy controlled force the rules of exchange by administrative fiat
by it, markets open up alternative avenues for and exclude private entrepreneurs from taking
mobility through emergent entrepreneurship part in legitimate economic activities. Because
and the selling of labor power. Under conditions economic actors depend on resources allocated
of expanding markets, economic actors strive to from above, they strive to secure favorable access
institute new rules of competition and coop- to these decision makers in order to maximize
eration that serve their interests, both through their access to scarce resources (Walder 1986).
informal arrangements and through formal insti- By contrast, the shift to the market mechanism
tutional channels. This entails efforts to change reduces dependence on superordinate bureau-
the structure of property rights in a manner that cratic agencies, as producers and consumers in-
enables entrepreneurs and producers to capture creasingly get needed goods and services through
a greater proportion of the economic surplus, markets. Market-based exchanges stimulate the
previously transferred to the state by administra- development of horizontal ties between buyers
tive fiat. Political actors contribute to instituting and sellers, in labor and production markets,
change in the formal rules of the game insofar resulting in an incremental decline in the rel-
as gains in productivity increase revenues to the ative value of vertical connections—political
1100 • Part X: The Future of Inequality

capital—and an increasing importance of net- likely to proceed without fundamental changes

work ties between economic actors in society.1 in the mechanism of stratification. According
In sum, changes in the mechanisms of stratifi- to proponents of this perspective, the initial ad-
cation stem from the expansion of opportunities vantages of the redistributive elite are such that
for gain and profit centering on market institu- this elite will come to dominate the postsocial-
tions. Opportunities are more broadly based and ist stratification order (Rona-Tas 1995; Walder
diverse when markets replace and augment the 1996). The old redistributive elite, they argue,
opportunity structure controlled by the state. occupy strategic positions in political markets,
Moreover, markets provide powerful incentives which enable them to adapt flexibly to emerging
for direct producers and entrepreneurs, whereas markets and capture a disproportionate share
state socialist redistributive economies depress of the rewards (Parish and Michelson 1996). In
incentives because administratively set prices for Russia, for example, officials and their cronies
labor lack sensitivity to differential performance. gained, by means of state-sponsored privatiza-
The foregoing line of reasoning may be rep- tion, ownership rights over public assets—oil-
resented as four closely related hypotheses. These fields, mines, utility companies, state-owned
are as follows: enterprises, and so forth. The key issue here is the
1. Political capital hypothesis: In market trans- extensiveness of such power conversion relative
actions, economic actors have the right to to the frequency with which new elites emerge
withhold their labor power or product in in the market economy. Market transition the-
emergent markets until a mutually agreed ory maintains that because an expanding market
upon price is set, and a greater share of the economy activates a substantially broader base of
economic surplus is therefore retained by economic and social participation, even though
producers than in a redistributive economy. a substantial element of the political elite suc-
The relative returns to political capital ac- ceed in converting political capital to economic
cordingly decline. capital, cadres as a social group still experience a
2. Human capital hypothesis: There are also relative decline. A recent study of the Hungar-
stronger incentives for individual effort ian elite confirms a very substantial decline in
because rewards are more closely related to the representation of the nomenklatura in the
individual productivity. This is likely to be composition of the postsocialist elite (S. Szelényi
reflected in higher returns to human capi- 1998). In urban China, where the communist
tal, which is among the best indicators of party remains firmly in power, increasing relative
human productivity. returns to human capital are revealed not only in
3. Sector mobility hypothesis: Producers who regressions on income (Zhou 2000), but in the
shift into the marketized sectors of the lower chances of promotion for party members
transition economy are likely to experience even in state-owned enterprises oriented to mar-
higher returns to their labor and productiv- ket action (Cao 1999).
ity than those who remain in the sectors of Table 126.1 provides a meta-analysis of re-
the transition economy still controlled by sults reported in empirical studies testing mar-
redistribution. ket transition theory. The findings confirm the
4. Entrepreneurship hypothesis: Finally, expand- theory’s predictions of increasing relative returns
ing markets open up alternative avenues of to human capital, and its claim that emergent
socioeconomic mobility, giving rise to en- market economies open alternative pathways
terpreneurship as a mechanism for upward for upward mobility through entrepreneurship
mobility into the postsocialist elite. and labor markets in the private/hybrid sectors
of the transition economy. However, empirical
Critics of the theory advance the claim that tests of the prediction of declining relative sig-
departures from a centrally planned economy are nificance of political capital show mixed results,
Table 126.1. Summary of Empirical Studies Relevant to Market Transition Theory

Data Findings
Political Human Sector Entrepre-
Study Nation Population Year Dependent Variable Capital Capital Mobility neurship
Szelényi 1988 Hungary Rural, whole nation 1982–83 Agricultural goods Mixed Yes — —
Nee 1989 China Rural, Fujian Province 1985 Income Yes Yes — Yes
Nee 1991 China Rural, Fujian Province 1985 Income Yes — — Yes
Peng 1992 China Urban & rural, selected areas 1986 Income — Yes Yes —
Rona-Tas 1994 Hungary Urban & rural, whole nation 1989,91 Employment Inconclusive* Yes — —
Income Inconclusive* Yes — Yes
Damanski and Heyns 1995 Poland Urban & rural, whole nation 1987,91 Income Yes Yes Incon- —
clusive
Parish, Zhe, and Li 1995 China Rural, eastern two-thirds of China 1993 Employment Mixed Yes — —
Income Yes Yes Yes —
Nee 1996 China Rural, whole nation 1989–90 Employment Mixed Yes — —
Income Yes No — Yes
Parish and Michelson 1996 China Rural, whole nation 1988 Employment Inconclusive# — — —
Income Inconclusive# — — —
Xie and Hannum 1996 China Urban, whole nation 1988 Income Inconclusive# Incon- — —
clusive#
Bian and Logan 1996 China Urban, Tianjin 1988,93 Income Yes* Yes Yes —
Gerber and Hout 1998 Russia Urban & rural, whole nation 1991–95 Income — No — Yes
S. Szelényi 1998
(1998) Hungary Whole nation 1993 Elite membership Yes No — —
Brainerd 1998 Russia Urban & rural, whole nation 1991–94 Income — Yes — —
Cao and Nee 1999 China Urban, two southern cities 1994–95 Income Yes Yes Yes Yes
Zhou 2000 China Urban, selected areas 1995 Income Yes* Yes* Yes —

Notes: “Yes” means that the finding is consistent with the market transition theory, “No” means inconsistent and “—” means inapplicable.
“*” indicates that the finding reported here is based on our own interpretation of the results, which differs from the author’s.
“#” indicates that we disagree with the study’s research design and hence regard the results as inconclusive.
Source: Nee and Cao (1999, pp. 10–11).
1101
1102 • Part X: The Future of Inequality

suggesting that elites in some postsocialist strati- Stratification in China.” American Sociological
fication orders have circulated more successfully Review 61:739–58.
than in others. Overall, the cumulative evidence Brainerd, Elizabeth. 1998. “Winners and Losers in
from empirical studies points to path-dependent Russia’s Economic Transition.” American Eco-
nomic Review 88:1094–116.
changes in the mechanisms of stratification
Cao, Yang. 1999. “Careers Inside Organizations:
largely in line with predictions proffered by mar- Comparative Study of Promotion Determination
ket transition theory (Nee and Cao 1999). in Urban China.” Unpublished manuscript, De-
The institutional changes that accompany the partment of Sociology, Cornell University.
shift from a redistributive to a market economy Cao, Yang, and Victor Nee. 1999. “Remaking In-
comprise the causal mechanisms that cumu- equality: Market Dependency and Organiza-
latively transform the stratification order. Al- tional Adaptation in Urban China.” Working
though the transition entails hybrid stratification paper (Ithaca: Department of Sociology, Cornell
orders in which elements of the old elite retain University).
———. 2000. “Comment: Controversies and Evi-
and even augment their advantages, the secular
dence in the Market Transition Debate.” Ameri-
trends of increasing returns to human capital can Journal of Sociology 105:1175–95.
and declining relative returns to political capital Cook, Karen Schweers, and Margaret Levi, eds.
signal the decline of the redistributive elite. The 1990. The Limits of Rationality. Chicago: Uni-
more developed the market economy, the greater versity of Chicago Press.
the breadth and diversity of opportunities that Damanski, Henrik, and Barbara Heyns. 1995. “To-
develop outside the boundaries of the old redis- ward a Theory of the Role of the State in Mar-
tributive economy. Groups and individuals who ket Transition: From Bargaining to Markets in
were formerly barred from advancement in the Post-Communism.” European Journal of Sociology
36:317–51.
state socialist bureaucracy and economy gain
Emerson, Richard A. 1962. “Power-Dependence Re-
chances for social mobility through emergent
lations.” American Sociological Review 27: 31–41.
labor markets and private entrepreneurship. To Gerber, Theodore, and Michael Hout. 1998. “More
the extent that new bases of opportunity expand, Shock than Therapy: Market Transition, Em-
resources become embedded in alternative net- ployment, and Income in Russia, 1991–1995.”
works and institutions, dependence on the es- American Journal of Sociology 104:1–50.
tablished elite declines, and excluded groups Grusky, David B., and Robert M. Hauser. 1984.
gain in power relative to the established elite. “Comparative Social Mobility Revisited: Models
Importantly, such shifts in power need not entail of Convergence and Divergence in 16 Coun-
tries.” American Sociological Review 49: 19–38.
a direct transfer of power as in a regime change
Hamilton, Gary G., ed. 1991. Business Networks and
but occur as an unintended by-product of insti-
Economic Development in East and Southeast Asia.
tutional change. As in the rise of capitalism in Hong Kong: University of Hong Kong, Centre
the West, such change in the stratification order of Asian Study.
occurs gradually over time. Hechter, Michael. 1987. Principles of Group Solidar-
ity. Berkeley: University of California Press.
NOTES Homans, George C. 1950. The Human Group. New
1. This argument is consistent with power depen- York: Harcourt Brace Jovanovich.
dence theories (Emerson 1962; Hechter 1987) and Kornai, János. 1980. The Economics of Shortage. Am-
social resource theory (Lin 1982). As China shifts sterdam: North-Holland.
to market coordination, family firms and social net- Lin, Nan. 1982. “Social Resources and Instrumen-
works will grow in importance in economic transac- tal Action.” Pages 131–46 in Social Structure and
tions (Hamilton 1991). Network Analysis, edited by P. Marsden and N.
Lin. Beverly Hills, CA: Sage.
REFERENCES Lipset, Seymour Martin, and Hans L. Zetterberg.
Bian, Yanjie, and John Logan. 1996. “Market Transi- 1959. “Social Mobility in Industrial Societies.”
tion and the Persistence of Power: The Changing Pp. 11–75 in Social Mobility in Industrial Society,
 Nee—Post-Socialist Stratikcation • 1103

edited by S. M. Lipset and R. Bendix. Berkeley: Polanyi, Karl. 1944. The Great Transformation: The
University of California Press. Political and Economic Origins of Our Time. New
Merton, Robert K. 1949. Social Theory and Social York: Rinehart.
Structure. New York: Free Press. ———. 1957. “The Economy as Instituted Pro-
Nee, Victor. 1989. “A Theory of Market Transition: cess.” Pages 243–69 in Trade and Market in Early
From Redistribution to Markets in State Social- Empires, edited by K. Polanyi, C. Arensberg, and
ism.” American Sociological Review 54: 663–81. H. Pearson. Glencoe, IL: Free Press.
———. 1991. “Social Inequalities in Reforming Pryor, Frederick L. 1973. Property and Industrial Or-
State Socialism: Between Redistribution and ganization in Communist and Capitalist Nations.
Markets in China.” American Sociological Review Bloomington: Indiana University Press.
56:267–82. Rona-Tas, Akos. 1994. “The First Shall Be Last?
———. 1992. “Organizational Dynamics of Mar- Entrepreneurship and Communist Cadres in the
ket Transition: Hybrid Forms, Property Rights, Transition from State Socialism.” American Jour-
and Mixed Economy in China.” Administrative nal of Sociology 100:40–69.
Science Quarterly 37:1–27. Sahlins, Marshall. 1972. Stone Age Economics. New
———. 1996. “The Emergence of a Market Society: York: Aldine.
Changing Mechanism of Stratification in China.” Szelényi, Iván. 1978. “Social Inequalities in State So-
American Journal of Sociology 100:908–949. cialist Redistributive Economies.” International
Nee, Victor, and Yang Cao. 1999. “Path Dependent Journal of Comparative Sociology 19: 63–87.
Societal Transformation: Stratification in Hybrid ———. 1988. Socialist Entrepreneurs. Madison, WI:
Mixed Economies.” Theory and Society 28:1–37. University of Madison Press.
Nee, Victor, and Paul Ingram. 1998. “Embedded- Szelényi, Szonja. 1998. Equality by Design: The
ness and Beyond: Institutions, Exchange, and Grand Experiment in Destratification in Social-
Social Structure.” Pages 19–45 in The New In- ist Hungary. Stanford, CA: Stanford University
stitutionalism in Economic Sociology, edited by Press.
M. Brinton and V. Nee. New York: Russell Sage Treiman, Donald J. 1970. “Industrialization and So-
Foundation. cial Stratification.” Pages 207–34 in Social Strati-
North, Douglass C. 1977. “Markets and Other Al- fication: Research and Theory for the 1970s, edited
location Systems in History: The Challenge of by E. O. Laumann. Indianapolis: Bobbs-Merrill.
Karl Polanyi.” Journal of European Economic His- Walder, Andrew G. 1986. Communist
tory 6:703–16. Neo-Traditionalism: Work and Authority in Chi-
———. 1981. Structure and Change in Economic nese Industry. Berkeley and Los Angeles: Univer-
History. New York: Norton. sity of California Press.
Parish, William L., and Ethan Michelson. 1996. ———. 1996. “Markets and Inequality in Transi-
“Politics and Markets: Dual Transformations.” tional Economies: Toward Testable Theories.”
American Journal of Sociology 101:1042–59. American Journal of Sociology 101: 1060–73.
Parish, William L., Xiaoye Zhe, and Fang Li. Xie, Yu, and Emily Hannum. 1996. “Regional
1995. “Nonfarm Work and Marketization of Variation in Earnings Inequality in Reform-Era
the Chinese Countryside.” China Quarterly Urban China.” American Journal of Sociology
143:697–730. 101:950–992.
Peng, Yusheng. 1992. “Wage Determination in Rural Zhou, Xueguang. 2000. “Economic Transformation
and Urban China: A Comparison of Public and and Income Inequality in Urban China.” Ameri-
Private Industrial Sectors.” American Sociological can Journal of Sociology 2000: 1135–74.
Review 57:198–213.
1104 • Part X: The Future of Inequality

127 • Gil Eyal, Iván Szelényi, and Eleanor Townsley

Making Capitalism Without Capitalists

There have been a number of schools of thought possessors of material property (the economic
about which social groups would be best placed bourgeoisie) and possessors of culture or knowl-
to take advantage of market reform after the fall edge (the cultural bourgeoisie)—then one can
of communism. Market transition theory1 and claim—and we do so in this [chapter]—that
research on socialist entrepreneurs2 suggested post-communist capitalism is being promoted
that those who had been successful in the sec- by a broadly defined intelligentsia which is com-
ond economy during the late communist period mitted to the cause of bourgeois society and cap-
would benefit most from the transition to capi- italist economic institutions.
talism. In contrast, theorists of political capital- This approach to analyzing the transition to
ism argued that former cadres were best placed capitalism in Central Europe necessarily differs
to convert state-socialist privilege into economic both from the classical social and economic the-
capital.3 Against both these theories we contend ories of Adam Smith and Karl Marx and from
that cultural capital became the dominant form twentieth-century visions of corporate, man-
of capital in post-communism. The coalition agerial, or other kinds of post-capitalist societ-
that governs post-communist societies is com- ies, such as those proposed by Ralf Dahrendorf,
prised of technocrats and managers—many of Berle and Means, or Daniel Bell. On one hand,
whom held senior positions in communist insti- classical theorists assumed that there must have
tutions—and former dissident intellectuals who been capitalists before capitalism. For this reason,
contributed to the fall of communist regimes at these theorists expended much scholarly effort
the end of the 1980s. investigating the process of the ‘original’ or
This [chapter] offers a new theory of the ‘primitive accumulation’ of capital. The rationale
transition to capitalism, by telling the story of was that both logically and historically, private
how capitalism is being built without capitalists capital accumulation must have occurred before
in post-communist Central Europe. We theorize market institutions could operate.4 On the other
capitalism without capitalists as a distinctive new hand, a recurrent theme in the study of existing
strategy of transition adopted by technocratic– capitalist systems, particularly since the 1930s,
intellectual elites in societies where no class of has been that the importance of individual pri-
private owners existed prior to the introduction vate owners is waning. Observing the growth
of market mechanisms. Note, however, that of large corporate organizations, the increasing
capitalism without capitalists is not necessarily role of financial institutions, or the growing im-
capitalism without a bourgeoisie. If one thinks of portance of science/knowledge, varied theorists
the bourgeoisie as plural—thus, if one conceives of ‘late’ capitalism have argued that the role of
bourgeoisies as a social group composed of both capital as the main source of economic growth is

Gil Eyal, Iván Szelényi, and Eleanor Townsley. Making Capitalism Without Capitalists: The New Ruling Elites
in Eastern Europe, pp. 1–2, 6–7, 32–33, 160–163, 229–231, 247–248. Copyright © 1998 by Verso Press.
Reprinted by permission of Verso Press.
 Eyal, Szelényi, and Townsley—Making Capitalism Without Capitalists • 1105

ending.5 There is little doubt within this group sharing their managerial power (even temporar-
that the capitalist system itself remains robust; ily) with a class of individual owners. . . .
what is in contention is the shape of capitalism
after capitalists. Our theory of the transition to
capitalism in Central Europe borrows from both Cultural Capital and Class in
these theoretical traditions, but differs from and Post-Communism
builds upon them by imagining the historical Post-communist society can be described as a
possibility of capitalism without capitalists. unique social structure in which cultural capital
First, like ‘capitalists before capitalism’, ours is the main source of power, prestige, and privi-
is primarily a theory of transition. Our central lege. Possession of economic capital places actors
aim is to understand and explain how capital- only in the middle of the social hierarchy, and
ism can emerge in an economic system with no the conversion of former political capital into
propertied bourgeoisie. We want to know what private wealth is more the exception than the
agents are building post-communist capitalism, rule. Indeed, the conversion of former commu-
and on whose behalf and for what purposes they nist privilege into a post-communist equivalent
act. One possibility is that the technocratic elite happens only when social actors possess the right
of former state-socialist societies may turn itself kinds of capital to make the transition. Thus,
into a new propertied bourgeoisie and thereby those who were at the top of the social hierarchy
fulfil the classical condition for capitalist devel- under state socialism can stay there only if they
opment. Indeed, some analysts claim—rather are capable of ‘trajectory adjustment’, which at
prematurely, in our view—that such a transfor- the current juncture means if they are well en-
mation has already occurred. Another possibility dowed with cultural capital. By contrast, those
is that the liberal intelligentsia will act as an ‘in- who relied exclusively on now devalued political
tellectual vanguard of the economic bourgeoisie’, capital from the communist era are not able to
creating a new class of proprietors from agents convert this capital into anything valuable, and
other than itself. Having fulfilled its historic are likely to be downwardly mobile. . . .
mission, the intelligentsia may then return to Our point of departure is Bourdieu’s the-
creative writing, research, or teaching, or it may ory of social structure, which we reconstruct
keep managing capitalist enterprises owned by using Weber’s distinction between rank and
others. class societies.7 In this way, we hope to bring
Within this field of possibilities, however, it a comparative-historical dimension into Bour-
is not inconceivable that capitalism will be built dieu’s framework, which was designed initially to
without a class of individual proprietors. In this explain reproduction rather than social change.
case, we would expect an economic system which (No criticism is implied here: Bourdieu’s research
looks like those envisioned by twentieth-century site is contemporary French society, where repro-
theorists of ‘capitalism after capitalists’. If it is duction is the overwhelming trend; we study the
true that the future of capitalist economies will recent social history of Central Europe where
be systems in which individual proprietors do not turbulence and large-scale social change have
play a major role (and this proposition has been predominated.) From Bourdieu we borrow the
forcefully challenged by Zeitlin, Domhoff and ideas of three forms of capital, social space, and
others6), it is possible that Central European so- habitus. On this basis, we distinguish between
cieties will emerge as corporate or post-capitalist social, economic, and cultural capital, and we
societies without the historical intervention of conceptualize social structures as ‘spaces’ which
a grande bourgeoisie. In other words, it is con- are differentially stratified by various distribu-
ceivable that post-communist elites will take a tions of these types of capital. Individuals ‘travel’
historic short cut and move directly to the most in these spaces, and if changes in the relative
‘advanced’ stage of corporate capitalism, never importance of one or another type of capital
1106 • Part X: The Future of Inequality

occurs in their lifetimes, they try to reshuffle class-stratified societies—communist societies

their portfolio of different types of capital and were examples of modern rank order. With this
convert devalued forms of capital into revalued understanding, we can argue that the transition
forms in order to stay ‘on trajectory’. On this from communism to capitalism is, in principle,
basis, there are several specific innovations we a transition from rank order to class society, and
propose to explain the nature of social change in thereby introduce the dynamism of Weber’s
Central Europe. historical sociology into Bourdieu’s structural
analysis. Actually existing post-communism,
1. We conceptualize pre-communism, com- however, as a system of ‘capitalism without cap-
munism, and post-communism as three different italists’, occupies a middle position on this scale
stratification regimes defined by the dominance of from rank to class. The Bildungsbürgertum [i.e.,
different types of capital. While Bourdieu used the the educated middle classes] is neither a rank nor
three types of capital to describe contemporary a class. Rather, it combines the characteristics
French society, where economic capital is domi- and contains the possibilities of both logics of
nant and the other forms of capital are subordi- social stratification. . . .
nate, in the ever-changing landscape of Central
Europe during the past fifty years we distinguish
three qualitatively different social spaces, each of The ‘New Class’ Project for the
which is defined by the dominance of a different Fourth Time?
form of capital. We have been careful not to call Central Euro-
2. We claim that post-communism is a his- pean post-capitalism a class society, and in par-
torically unique system of stratification in which ticular we have explicitly avoided referring to
cultural capital is dominant. This enables us to technocrats and managers, or the intelligentsia
develop a new theory of post-communist social as a whole, as a dominant class. Rather, the key
structure which is consistent with our argument to our theory of post-communist managerial-
about the crucial roles played by the technoc- ism and the conception of ‘capitalism without
racy and former dissident intellectuals in the capitalists’ is that the formation of a propertied
transition from socialism to post-communism. bourgeoisie has been relatively slow: market in-
We define capitalism as a class-stratified system stitutions have developed much more rapidly
in which economic capital is dominant, and in Central Europe than a capitalist class. It is
communism as a system in which social capi- precisely this weakness of a domestic proper-
tal—institutionalized as political capital—was tied bourgeoisie which has made it possible for
the major source of power and privilege. With a power bloc, composed of different fractions
the decay of state socialism and the rise of of intellectual elites, to retain a hegemonic
post-communism, the importance of political position. At the same time, our answer to the
capital is declining, the role of cultural capital ‘Whodunit’? question—that is to say, ‘Whose
is increasing, and economic capital is sufficient project is post-communism’?—is that it has
only to locate its possessors in the middle of the been the project of the intelligentsia, or at least
social hierarchy. certain fractions of the intelligentsia. So it seems
3. The transition to post-communism is a shift that the idea of the ‘New Class’ haunts us—the
from socialist rank order to capitalist class stratifica- post-communist power bloc is neither a new nor
tion. Drawing on Weber, we conceptualize com- an old class, but it is composed of intellectuals,
munism as a society based on rank order. Social and these intellectuals are pursuing a power
capital was dominant, and resulted in a social- project.
ist form of patron-client relations. Compared During the last century—starting with Ba-
to societies where economic capital was domi- kunin in the early 1870s, and stretching to
nant—which we understand, with Weber, to be Gouldner in the late 1970s—the idea of a ‘New
 Eyal, Szelényi, and Townsley—Making Capitalism Without Capitalists • 1107

Class’ of intellectuals dominating and leading without insight. From our perspective, it helps
society has haunted the social sciences. In the to explain why Central European intellectuals
West, it was thought that intellectuals would re- abandoned their Bildungsbürgertum project at
place the propertied bourgeoisie. In Central Eu- the turn of the century, and why so many of
rope, it was believed that they would replace the them were attracted to left-wing radicalism.
old-guard bureaucracy of communist regimes. The second wave of New Class theories—
And while successive predictions of the immi- fashionable in a period roughly from 1930 to the
nent rise of the intelligentsia to power proved 1950s—consisted of ‘bureaucratic-technocratic
false, social scientists never seemed to learn their class’ theories. In the version which addressed
lesson. They kept creating new theories about the social conditions of advanced Western capi-
societies in which intellectuals would dominate. talism, New Class theorists argued that the tech-
Such theorizing came in waves, was fashion- nocratic and/or managerial stratum would fill
able for a few years, and then discredited for a the gap created by the decline of family capital-
while, only to be reborn in a somewhat altered ism and individual private property.9 We think
form some years later. In an essay published a these theorists were probably correct in identi-
few years ago, one of us suggested that the stub- fying ‘New Class’ aspirations among managers
born return of the idea of the New Class requires and technocrats at this time, but they erred in
explanation.8 We argued that while neither the taking the ideologues of managerialism at their
intelligentsia as a whole, nor any of its strata, word, and underestimating the power of the old
ever succeeded in establishing itself as a new moneyed class. Critics of managerialist New
dominant class, it was probably historically true Class theories, like Zeitlin and Domhoff, were
that intellectuals formulated such ‘projects’ for correct when they observed that managerialism
power. On this basis, we interpreted the differ- in the United States and Western Europe failed
ent waves of New Class theorizing as critical or because it confronted a strong propertied bour-
apologetic—albeit premature—generalizations geoisie.10 We tend to side with the critics in this
of such power projects. It may be instructive debate, since there is strong evidence that the old
to think about post-communist managerialism propertied bourgeoisie is alive and well. Indeed,
from this perspective, and to contrast it with ear- it seems that today the concentration of wealth
lier intellectual class projects. in the United States and Western Europe is even
Between 1870 and 1970 we can identify more marked, and fewer people own more of the
at least three distinct waves of New Class the- productive assets than ever before.
orizing, and possibly three intellectual power The third wave of New Class theories, created
projects. in the 1960s and 1970s, was comprised of theo-
We identify the first wave as the ‘intellectual ries of ‘the knowledge class’. In their right-wing
class’ theory advanced by anarchists of the late version, these theories of the new class inter-
nineteenth and early twentieth centuries. This preted the radical movements of the 1960s as a
theory contended that Marxism was really the power plot by the counter-cultural intelligentsia
ideology of an intellectual elite, which was try- to dominate society. By contrast, social scientists
ing to use the working-class movement to smug- on the political Left emphasized that as science
gle itself into a position of class power. When became an increasingly important factor of pro-
a Marxism-inspired social order was established duction, scientists would begin replacing prop-
in the post-revolutionary Soviet Union—taking erty owners as the dominant group in advanced
form as a dictatorship of the Stalinist bureaucracy Western societies.11 Some of these theorists, most
rather than class rule by ‘socialist scholars’—this notably Gouldner, were not thrilled by the pros-
theory died a peaceful death. However, this does pect of a society ruled by a new class coalition
not mean that the anarchist analysis of Marx- of the technical intelligentsia and humanistic
ism as an intellectual ideology was completely intellectuals merged together by the ‘culture of
1108 • Part X: The Future of Inequality

critical discourse’, but given the alternatives, nationalization of property has been a process
they believed that it might be ‘our best card with far-reaching implications, and is possibly ir-
in history’. As the left radicalism of the 1960s reversible. As the old joke puts it: we know how
faded away, however, and as counter-cultural to make fish soup from an aquarium, but how
intellectuals were replaced by Yuppies preaching can we build an aquarium from a bowl of fish
neo-liberalism, this wave of theorizing also with- soup? East European economists enjoyed citing
ered away. this joke before 1989, but they quickly forgot
What about post-communist managerialism? it after 1989. Indeed, one major underpinning
Is it possible that we are witnessing a fourth New of managers’ power is their claim to be the only
Class project this century? Is it possible that in- ones with the knowledge of how to conduct the
tellectuals are indeed in power, even though it complex task of privatization. And the more
is ‘by default’ this time, since managers have no complex the task, the more it is practically im-
enemies able to resist their newly found domi- possible to convert former public property into
nance? Have the prophecies of Berle and Means identifiable individual private property; thus the
actually turned out to be correct—not for the more power managers have. Thinking about
advanced West but ironically, for Central Eu- post-communist managerialism as the ‘fourth
rope? Or is post-communism, as we have ana- wave’ of an intellectual project for power is use-
lyzed it, simply a step in the transition to market ful, then, since it helps us to ask these intriguing
capitalism as we know it from the advanced questions. Moreover, thinking in this way means
West? Or are we witnessing the transformation that we do not simply assume that managerial
of existing elements into a new Gestalt, of which power is a transitional phenomenon; it allows us
we have no adequate conception? We know that to ask: what will society be like if managerialism
we are shooting at a rapidly moving target, and reproduces itself?
social scientists are poor snipers even when the To push our luck even further, we may
target does not move. even wonder whether or not Central European
In this context, it is difficult to answer the post-communist developments have any rele-
question of whether or not managerialism will vance for the rest of the world. To put it differ-
last. It is difficult to predict the chances that the ently: is Central Europe the future of the West?
post-communist power bloc will reproduce itself After all, Central European managerialism may
and the social and economic order it governs. It is not be all that different from the way Western
unclear whether or not the post-communist elite capitalism operates today: property rights are
will resist the encroachments of international diffuse, managers exercise a lot of power, and
capital, and the attempts of propertied middle monetarist ideologies are powerful and wide-
classes to accumulate capital and take hold of the spread. As we have already noted, however, these
command posts of the economy. Reproduction is questions about the future are tricky ones, and
a possibility, however. It is conceivable that man- social scientists are notoriously bad at answering
agers will consolidate their powerful institutional them. Perhaps it would be best to pose the ques-
base by allying with national and international tion in a different way and ask: ‘If managerialism
fiscal institutions and, with monetarism as their was a project during the interwar years, as Berle
ideology, continue to govern post-communist and Means suggested, and if it failed, why did it
capitalism as managers. The appropriate posi- fail’? Approaching the issue in this way is likely
tions are ready, the adequate consciousness has to make us skeptical about the spread of mana-
been prepared, and the actors feel comfortable gerialism around the world. Since we argued that
in their positions and are deeply committed to managerialism was successful in Central Europe
their ideology. Furthermore, in post-communist because it did not face powerful enemies like a
societies the alternatives seem to be difficult. The large propertied bourgeoisie, we would predict
 Eyal, Szelényi, and Townsley—Making Capitalism Without Capitalists • 1109

that the chances for managerialism in the rest York: Random House; and G. William Domhoff
of the world, and especially in the countries of (1986) Who Rules America Now? A View from the
advanced Western capitalism, are slim indeed. 1980s. New York: Simon & Schuster.
7. Pierre Bourdieu ([1983] 1986) ‘The Forms
NOTES of Capital’. pp. 241–58 in Handbook of Theory and
1. Victor Nee (1989) ‘A Theory of Market Tran- Research for the Sociology of Education, ed. John G.
sition: From Redistribution to Markets in State So- Richardson. New York: Greenwood Press; Pierre
cialism’. American Sociological Review 54(5): 663–81. Bourdieu (1984) Distinction: A Sociological Critique
2. Iván Szelényi (1988) Socialist Entrepreneurs. of the Judgement of Taste. Cambridge, MA: Harvard
Madison: University of Wisconsin Press. University Press; Max Weber ([1915–21] 1978)
3. Ákos Rona-Tas (1994) ‘The First Shall Be The Economy and Society. Berkeley: University of Califor-
Last? Entrepreneurship and Communist Cadres in nia Press, pp. 926–38.
the Transition from Socialism’. American Journal of 8. Iván Szelényi and Bill Martin (1988) ‘The
Sociology 100(1): 40–69. Three Waves of New Class Theories’. Theory and
4. Karl Marx (1919) Das Kapital: Kritik der Society 17(4): 645–67.
politischen Ökonomie. Hamburg: Meissner; Adam 9. See, for example, Thornstein Veblen ([1919]
Smith ([1779] 1976) An Enquiry into the Nature and 1963) The Engineers and the Price System. New York:
Causes of the Wealth of Nations. Oxford: Oxford Uni- Harcourt & Brace; Adolf Berle and Gardiner Means
versity Press. ([1931] 1968) The Modern Corporation and Private
5. For the classic statements on the growth of the Property. New York: Harcourt Brace Jovanovich;
large corporation and its consequences, see Adolf James Burnham ([1941] 1962) The Managerial Rev-
Berle and Gardiner Means ([1931] 1968) The Mod- olution. Bloomington: Indiana University Press; Ralf
ern Corporation and Private Property. New York: Har- Dahrendorf (1972) Class and Class Conflict in Indus-
court Brace Jovanovich; and Ralf Dahrendorf (1959) trial Society. Stanford, CA: Stanford University Press;
Class and Class Conflict in Industrial Society. Stanford, John K. Galbraith (1967) The New Industrial State.
CA: Stanford University Press. For a discussion of New York: Houghton Mifflin.
the increasingly important role of financial institu- 10. Maurice Zeitlin (1974) ‘Corporate Own-
tions, see Neil Fligstein (1990) The Transformation ership and Control: The Large Corporation and
of Corporate Control. Cambridge, MA: Harvard Uni- the Capitalist Class’. American Journal of Sociology
versity Press; and for arguments which suggest that 79(5):1073–1119; G. William Domhoff (1967)
the role of capital is ending as the primary motor Who Rules America? Englewood Cliffs, NJ: Pren-
of economic growth in modern capitalism, see, for tice Hall; G. William Domhoff (1970) The Higher
example, Daniel Bell ([1973] 1976) The Coming of Circles: The Governing Class in America. New York:
Post-industrial Society. New York: Basic Books. Random House.
6. See, for example, Maurice Zeitlin (1974) ‘Cor- 11. Galbraith, in The New Industrial State, took
porate Ownership and Control: The Large Corpo- this position, and to some extent, so did Daniel Bell
ration and the Capitalist Class’. American Journal of ([1973] 1976) The Coming of Post-Industrial Society.
Sociology 79(5): 1073–1119; G. William Domhoff New York: Basic Books. See also Alvin Gouldner
(1967) Who Rules America? Englewood Cliffs, NJ: (1979) The Future of Intellectuals and the Rise of the
Prentice Hall; G. William Domhoff (1970) The New Class. New York: The Continuum Publishing
Higher Circles: The Governing Class in America. New Corporation.
1110 • Part X: The Future of Inequality

128 • Andrew G. Walder

Elite Opportunity in Transitions from

State Socialism

There are today more than thirty independent implications for elite advantage seems anachro-
countries that have abandoned state socialism nistic. In retrospect it seems clear that the early
for some form of a market economy, a process literature on this subject neglected the crucial
that began more than two decades ago. Early role of politics and government policies that
research on these developments was preoccu- shaped the distribution of power and property.
pied with testing predictions about the generic During transitions to a market economy, the
impact of market competition on returns to po- opportunities available to communist-era elites
litical credentials and connections versus educa- depended heavily on the features of a country’s
tion, experience, and entrepreneurship. Would a political trajectory. Market economies did not
market economy erode the privileges of the for- unfold naturally under uniform political circum-
mer political elite, devaluing political status and stances. It mattered greatly whether communist
connections (Nee 1989)? Or would incumbent party hierarchies were dismantled along with
elites be well positioned to reap disproportionate the command economy. In two notable Asian
benefits from their powers of office and connec- cases—China and Vietnam—party hierarchies
tions (Róna-Tas 1994)? Because the first wave of survived largely unchanged. In all others, old re-
research was about the initial stages of reform, gimes were replaced by successor states in which
analysts scrutinized survey data for hints about the continuing influence of communist-era elites
future trends. Debates focused on whether elite varied greatly. At one extreme, challengers to the
advantages were merely a symptom of “partial” old regime quickly established electoral democ-
reform, destined to shrink as a genuine market racies and assumed the balance of power. At the
economy was more fully established (Bian and other extreme, communist parties withdrew
Logan 1996; Gerber 2002; Gerber and Hout from multinational federations and continued to
1998; Nee 1991, 1996; Walder 2002; Wu and rule as nationalist dictatorships. Between these
Xie 2003; Xie and Hannum 1996). At one level, extremes was a mix of dictatorships and democ-
the debate appeared to be about who would win racies dominated by neither challengers nor old
and who would lose. More fundamentally, it was regime elites (McFaul 2002). These political cir-
about whether there were any predictable out- cumstances either magnified or diminished the
comes directly connected to the extent to which opportunities for old regime elites.
a market economy had been fully established Also important in these transitions is whether
(Walder 1996). public assets were transferred rapidly to new
After more than two decades we no longer owners, and if so, whether regimes created bar-
need to speculate about future trends, and the riers to old elite appropriation of ownership or
idea that market mechanisms alone have clear managerial control. How state property was

Original article prepared for this fourth edition of Social Stratification: Class, Race, and Gender in Sociological
Perspective, edited by David B. Grusky. Copyright © 2014 Westview Press.
 Walder—Elite Opportunity in Transitions from State Socialism • 1111

allocated in the early stages of reform created the accumulation of personal wealth disappear, and
initial conditions for the subsequent operation access to public property and public resources
of market forces. This was a highly variable pro- attains new value, along with official discretion
cess.1 In the reform communist regimes of China and personal influence in regulatory decisions.
and Vietnam and in several former Soviet repub- Because virtually all capital, real estate, and nat-
lics, privatization was delayed and slow. In most ural resources are initially under public owner-
postcommunist regimes privatization was rapid, ship, and incumbent elites exercise control over
sometimes completed in less than a decade. these assets at the outset of market transition,
To the extent that privatization was rapid and they have inherent advantages over other groups.
poorly regulated, the ability of communist-era Whether these are realized is contingent on the
elites to seize assets was enhanced. political circumstances of the transition.
These political circumstances were largely
ignored in the first wave of research, a flaw that
was inevitable in the many publications that The Extensiveness of Regime Change
analyzed data from a single country. Today a Extensive regime change altered political institu-
credible theory about the impact of markets tions in two ways. First, the old party hierarchy
would have to incorporate the role of political lost its ability to appoint officials in all govern-
transitions in emerging market economies. Such ment agencies, public institutions, and publicly
a theory would start from a different premise: owned enterprises. Second, the hierarchy itself
that the impact of markets is contingent on vari- disintegrated, leading to the disappearance of a
ations in the distribution of power and property national system of offices that had paralleled the
(Walder 2011b). governmental and enterprise hierarchies, within
which careers were organized and which itself
controlled large concentrations of property and
A Theory of Elite Opportunity other assets. If the former ruling party survived,
Bureaucratic allocation of goods and services was it did so as a smaller electoral party that offered
the foundation of elite privilege in Soviet-style few career opportunities and controlled only
economies. But the shift to market allocation meager assets. This pattern was most common
does not necessarily imply a reduction of elite in Central Europe and the Baltic states.
opportunity. Bureaucratic power, after all, did Where market reform proceeded under com-
not provide communist elites with large private munist rule (China, Laos, Vietnam), party hi-
incomes or significant personal wealth. Except at erarchies and their control over appointments
the very apex of the political system, the income and public assets were unchanged. But some
and material advantages of communist elites of the newly independent states that emerged
were modest. The advantages that accompanied from the collapse of the Soviet Union initially
rank were based on preferential access to pub- exhibited little more regime change than did
lic property: larger and better housing in more China and Vietnam. Party hierarchies in these
desirable neighborhoods, use of vacation resorts, new states frequently survived for years, adopt-
and the personal use of automobiles owned by ing a nationalist rather than socialist orientation
an organization (Walder 1992). Power and rank and continuing to rule as dictatorships.2 In these
yielded access to scarce consumer goods in an regimes market reforms were implemented by
economy of shortage. By the standards of mar- single-party dictatorships that initially preserved
ket economies—and today’s transitional econ- much of their former organizational structures,
omies—the advantages of elites in command appointment powers, and assets, yet abandoned
economies were modest. their commitment to socialist principles.
A transition to a market economy creates The most extreme case of regime change is the
the potential for unprecedented new forms of German Democratic Republic, whose institu-
elite enrichment. Political constraints on the tions were rapidly dismantled as it was absorbed
1112 • Part X: The Future of Inequality

into the Federal Republic of Germany. Extensive The first is in the reform communist regimes
regime change occurred more commonly where and the postcommunist dictatorships that re-
communist parties disintegrated after strong op- sist privatization and retain strong prohibitions
positional movements brought about decisive against the theft of state property. The relatively
electoral defeats. This created procedural de- limited asset appropriation under these circum-
mocracies in which challengers to the old regime stances is not due to the strength of legal and
held the balance of political power.3 Some of regulatory structures, which are notoriously
these new governments adopted anticommunist weak, but to the delayed onset and slow pace of
laws designed to bar those with ties to the former privatization. Under these circumstances, state
regime from elite positions (Eyal, Szelényi, and agencies and public firms may engage in a form
Townsley 1998, 108–111, 128–131). In other of asset conversion in which they transfer public
regimes—most notably Russia—initially nei- assets to private entities that are under their own
ther challengers nor communist-era elites clearly organizational control. These strategies permit
dominated postcommunist governments.4 state firms to evade state regulation and taxes by
earning larger incomes off the books (Lin 2001).
The proceeds may be used for a variety of pur-
Constraints on Asset Appropriation poses, including larger executive compensation
Elite opportunity is also affected by the extent (salaries and fringe benefits) and the (corrupt)
to which there are political or legal constraints diversion of funds into private hands. In these
on the appropriation of public assets. Asset ap- cases, incumbent officials may extract larger in-
propriation occurs when incumbent elites keep comes from such arrangements, but they do not
managerial control of public assets as they are assume outright ownership of these still-public
privatized or convert them into personal owner- assets (Walder 2011a). Asset appropriation also
ship; how this occurs depends on the policy and takes place in these settings, but it proceeds at a
regulatory environment. The first policy variable slow pace. Over time it can lead to considerable
is the pace of privatization. Many transitional elite enrichment, as individuals and families ex-
economies resisted the rapid conversion of state tract long-term financial gain from their posi-
assets to some form of private ownership. By the tions (Ding 2000a, 2000b, 2000c).
end of the 1990s the private sector’s estimated The second setting in which asset appropri-
share of gross national product in transitional ation is relatively constrained is paradoxically
economies ranged from 20 to 80 percent (Eu- very different from the first: extensive regime
ropean Bank for Reconstruction and Develop- change coupled with orderly privatization that
ment 1999, 23–24). State policies that delayed transfers assets under transparent rules. Under
privatization created barriers to asset appropri- these circumstances, incumbent elites either lose
ation. If privatization is rapid, however, asset their positions of influence too quickly to appro-
appropriation is constrained only if there are ef- priate state assets, or the process of privatization
fective regulatory prohibitions against the prac- is too well regulated and monitored. The most
tice, something that varied widely among those extreme case in this regard is the former East
transitional economies that privatized quickly. Germany, but Poland, the Czech Republic, and
For this reason former communist officials fared Hungary also exhibit these characteristics (Eyal,
much better in Russia than in Central Europe Szelényi, and Townsley 1998, ch. 4; King 2001).
(Eyal, Szelényi, and Townsley 1998). In countries that have sold large proportions of
state enterprises directly to foreign corporations,
Policy and regulatory environment. Although incumbent elites have been unable to maintain
asset appropriation occurs in all transitional control over public assets, and elite strategies
economies, it is less prevalent under two very dif- to perpetuate managerial control have been de-
ferent kinds of policy/regulatory environments. feated (Hanley, King, and Janos 2003).
 Walder—Elite Opportunity in Transitions from State Socialism • 1113

Constraints on asset appropriation are weak from the Soviet intelligence services (Treisman
under two different circumstances. The first oc- 2007, 2011, 115–119).
curs when a communist party survives largely
intact but rules under a new name after aban-
doning its commitment to public property and Conclusion
central planning. These circumstances prevailed The long-familiar claim that opportunities for
in many of the newly independent republics that old regime elites eventually decline with further
abruptly withdrew from the Soviet Union with reform does not actually conflict with the argu-
little internal opposition or regime change. This ment that we have sketched here. As one sum-
permitted the widespread transfer of state assets mary of a decade of research observed:
into the hands of officials, their kin, and their
associates, or the extraction of large incomes Insiders and oligarchs benefit immediately
from the discretionary powers of office. In these from liberalization and privatization because
settings the primary question is how fully those they can convert their existing control over
at the top of the political hierarchy monopolize state assets into substantial gains. Moreover,
opportunities for enrichment. insiders and oligarchs can reap further gains
The second circumstance occurs when a from rent seeking, arbitrage, and asset strip-
nation in the midst of extensive regime change ping. . . . As discipline is imposed and further
rapidly privatizes assets without establishing reforms encourage competition from new en-
barriers to prevent insiders from seizing control. trants and the rule of law, these initial gains
This process typically permits incumbent man- are dissipated. (World Bank 2002, 92–93)
agers of public enterprises and industrial bureaus
to retain their posts as they privatize public as- Note that this statement is silent about what cre-
sets. They emerge as modern corporate exec- ates the “rule of law,” “further reforms,” and the
utives, freed of the restraints of the command “imposition of discipline,” or whether these are
economy, who may now allocate to themselves inevitable or even likely. These are outcomes of
vastly increased pay and benefits, including own- a country’s political trajectory, and after many
ership stakes, typical of managerial practices in years it is obvious that this ideal remains elusive
corporate capitalism. The process may also per- in many countries. From a sociological perspec-
mit some managers and economic bureaucrats tive, to state that further implementation of re-
to directly assume personal ownership of large form policies eventually reduces elite advantages
concentrations of newly privatized public assets. is largely beside the point. If the opportunity to
Those who are able to complete such maneu- appropriate assets eventually declines as a result
vers move into a new private business class with of further reform, this does not retroactively ex-
origins in the communist-era elite. Such out- propriate elites who have already appropriated
comes were relatively common in Russia, where assets or accumulated large incomes. It simply
by 1993 regime insiders had acquired majority means that opportunities for further asset ap-
shares in two-thirds of privatized and privatizing propriation and extraction of income from of-
firms, and a small number of wealthy oligarchs fice will decline. The social structure has already
had assumed control of certain key sectors of the been transformed. By the time more complete
economy (Barnes 2006; Blasi, Kroumova, and reforms are implemented, old regime elites will
Kruse 1997; McFaul 1995, 210). The phenom- already have seized property and amassed private
enon evolved under Vladimir Putin, after the fortunes. Subsequent reform does not turn back
properties of many of the original oligarchs were the clock. Without further political acts to re-
confiscated and placed under the control or out- cover them, property appropriated and income
right ownership of Putin’s political allies, a strik- accumulated by former elites in the initial stages
ing number of whom are his former colleagues remain in their hands.
1114 • Part X: The Future of Inequality

NOTES Hanley, Eric, Lawrence King, and Istvan Tóth Janos.

1. This essay updates a longer presentation 2003. “The State, International Agencies, and
(Walder 2003). Property Transformation in Post-Communist
2. This includes Belarus, Kazakhstan, Kyrgyz- Hungary.” American Journal of Sociology 108:
stan, Turkmenistan, and Uzbekistan. Serbia followed 129–167.
a similar path after the initial dissolution of Yugosla- King, Lawrence P. 2001. The Basic Features of Post-
via (McFaul 2002). communist Capitalism in Eastern Europe: Firms in
3. This occurred in the Czech Republic, Estonia, Hungary, the Czech Republic, and Slovakia. West-
Hungary, Latvia, Lithuania, Poland, Slovakia, Slove- port, CT: Praeger.
nia, and Croatia (McFaul 2002). Lin, Yimin. 2001. Between Politics and Markets:
4. This describes the trajectory of Bulgaria, Mon- Firms, Competition, and Institutional Change in
golia, Tajikistan, Moldova, Russia, Ukraine, Albania, Post-Mao China. Cambridge, UK: Cambridge
Azerbaijan, and Macedonia (McFaul 2002). University Press.
McFaul, Michael. 1995. “State Power, Institutional
REFERENCES Change, and the Politics of Privatization in Rus-
Barnes, Andrew. 2006. Owning Russia: The Struggle sia.” World Politics 47: 210–243.
over Factories, Farms, and Power. Ithaca, N.Y.: ———. 2002. “The Fourth Wave of Democracy
Cornell University Press. and Dictatorship: Noncooperative Transitions
Bian, Yanjie, and John R. Logan. 1996. “Market in the Postcommunist World.” World Politics 54:
Transition and the Persistence of Power: The 212–244.
Changing Stratification System in Urban China.” Nee, Victor. 1989. “A Theory of Market Transition:
American Sociological Review 61: 739–758. From Redistribution to Markets in State Social-
Blasi, Joseph R., Maya Kroumova, and Douglas ism.” American Sociological Review 54: 663–681.
Kruse. 1997. Kremlin Capitalism: The Privatiza- ———. 1991. “Social Inequalities in Reforming
tion of the Russian Economy. Ithaca, NY: Cornell State Socialism: Between Redistribution and
University Press. Markets in State Socialism.” American Sociological
Ding, Xueliang. 2000a. “The Illicit Asset Stripping Review 56: 267–282.
of Chinese State Firms.” China Journal 43:1–28. ———. 1996. “The Emergence of a Market So-
———. 2000b. “Informal Privatization Through ciety: Changing Mechanisms of Stratification
Internationalization: The Rise of Nomenklatura in China.” American Journal of Sociology 101:
Capitalism in China’s Offshore Business.” British 908–949.
Journal of Political Science 30: 121–146. Róna-Tas, Ákos. 1994. “The First Shall be Last?
———. 2000c. “Systemic Irregularity and Spon- Entrepreneurship and Communist Cadres in the
taneous Property Transformation in the Chi- Transition from Socialism.” American Journal of
nese Financial System.” China Quarterly 163: Sociology 100: 40–69.
655–676. Treisman, Daniel. 2007. “Putin’s Silovarchs.” Orbis
European Bank for Reconstruction and Develop- 51: 141–153.
ment. 1999. Transition Report 1999: Ten Years of ———. 2011. The Return: Russia’s Journey from Gor-
Transition. London: Stationery Office. bachev to Medvedev. New York: Free Press.
Eyal, Gil, Iván Szelényi, and Eleanor R. Towns- Walder, Andrew G. 1992. “Property Rights and
ley. 1998. Making Capitalism without Capi- Stratification in Socialist Redistributive Econo-
talists: Class Formation and Elite Struggles in mies.” American Sociological Review 57: 524–539.
Post-Communist Central Europe. New York: Verso. ———. 1996. “Markets and Inequality in Transi-
Gerber, Theodore P. 2002. “Structural Change and tional Economies: Toward Testable Theories.”
Post-Socialist Stratification: Labor Market Tran- American Journal of Sociology 101: 1060–1073.
sitions in Contemporary Russia.” American Socio- ———. 2002. “Markets and Income Inequality in
logical Review 67: 629–659. Rural China: Political Advantage in an Expand-
Gerber, Theodore P., and Michael Hout. 1998. ing Economy.” American Sociological Review 67:
“More Shock Than Therapy: Market Transition, 231–253.
Employment, and Income in Russia, 1991– ———. 2003. “Elite Opportunity in Transitional
1995.” American Journal of Sociology 104: 1–50. Economies.” American Sociological Review 68:
899–916.
 Walder—Elite Opportunity in Transitions from State Socialism • 1115

———. 2011a. “From Control to Ownership: Chi- Former Soviet Union. Washington, DC: The
na’s Managerial Revolution.” Management and World Bank.
Organization Review 7: 19–38. Wu, Xiaogang, and Yu Xie. 2003. “Does the Mar-
———. 2011b. “Transitions from State Socialism: ket Pay Off? Earnings Returns to Education in
A Property Rights Perspective.” In The Sociology Urban China.” American Sociological Review 68:
of Economic Life, 3rd ed., edited by Mark Gra- 425–442.
novetter and Richard Swedberg, 503–525. Boul- Xie, Yu, and Emily Hannum. 1996. “Regional
der, CO: Westview. Variation in Earnings Inequality in Reform-Era
World Bank. 2002. Transition: The First Ten Years. Urban China.” American Journal of Sociology 101:
Analysis and Lessons for Eastern Europe and the 950–992.
1116 • Part X: The Future of Inequality

Post-Modernity and High Modernity

129 • John W. Meyer

The Evolution of Modern Stratikcation Systems

The sociological tendency, well represented in of Nee (Chapter 126) and of Eyal and his col-
the chapters in this volume on “The Future of leagues (Chapter 127), but now it plays out as
Inequality,” has been to see modern stratification communist regimes break down rather than as
systems in a realist vein. Real component indi- post-industrialism proceeds. The postmodernist
viduals and groups, competing and cooperating vision, which represents social classes as breaking
in real interdependencies, create and change a down (e.g., Pakulski and Waters 2008), has sim-
system of inequalities. Real systems—and thus ilarly functional assumptions, but represents the
some sort of functionalism, broadly defined— postmodern stratification system as built much
are involved: To be sure, some theories may place more around the subjective and cultural than the
particular stress on competitive interdependence realist model of industrial society. Culture, here,
(as in class-conflict models), and others may em- means taste, lifestyle, status, and subjective pref-
phasize economic over other interdependencies, erences. It is broadly linked to the expanding im-
but nearly all share a vision of society as a system portance of education and the rise of the service
made up of interdependent parts. sector, emphasized in all the chapters.
For example, classic theories of industrialism More critical writers, often on the left, use
(e.g., Kerr et al. 2008) represent the modern exactly the same lines of argument, but for them
economy as evolving toward greater complex- the functional requirements are those of some
ity and thus as functionally requiring changing economic elites or political forces rather than
forms of inequality (including emphases on ed- society as a whole. Esping-Andersen (Chapter
ucation and merit). The core themes here are 124) sees these basic functional processes as op-
picked up and amplified by other authors in this erating to generate actual or potential increases
volume. Bell (Chapter 123) sees both economic in inequality and breakdowns in social welfare.
and social complexity as creating (1) functional Despite their disagreements and conflicts, the
requirements for new levels of knowledge and authors of these lines of thought all share a gen-
competence and (2) a new set of strata linked eral description and analysis of the evolution of
to education. Although Gouldner (Chapter modern stratification, a view rooted in a realist
34) has a similar vision, it is generated by class vision of modern society as a functioning sys-
conflict as well as by differentiation, thus pro- tem. The postmodernists see realist modernity
ducing a recognizably similar, if darker, story. as having evolved in more subjective and cultural
Likewise, this vision again appears in the work ways, but retain the strong sense of an integrated

John W. Meyer. “The Evolution of Modern Stratification Systems,” in Social Stratification: Class, Race, and
Gender in Sociological Perspective, Second Edition, edited by David B. Grusky, pp. 881–890. Copyright © 2001
by Westview Press, Inc.
 Meyer—The Evolution of Modern Stratikcation Systems • 1117

functioning system. In all the arguments here, collective action, creating and changing their
expansion produces differentiation and complex- stratification system and rooting it more deeply
ity—these require more reliance on education in education, is not given much emphasis. “Class
and ultimately new professional strata. Every consciousness” is of interest as a dependent vari-
story told in these chapters amounts, in essence, able or as an intervening variable helping to
to either emphasizing or explaining the rising account for various outcomes such as political
stratificational importance of education and its mobilization. But this is class consciousness in
credentials. As the system expands around the the sense of Marx or Mannheim, who assume
world, modern forms of stratification rooted in an underlying sociological reality and see some
education are found everywhere. The story is form of consciousness as its product. In research
a standard sociological one, told with varying terms, it is class consciousness as a property of
emphases on conflict and consensus; it is also individuals and groups within society, not a
very similar to the stories modern societies tell property of the collective itself. More respect
about themselves. We emphasize the latter point than imitation is given to Ossowski (1963), for
as of substantive importance in understanding whom the cultural aspects of stratification can
modern societies, not so much as a criticism of be (1) independent variables and (2) collective
stratification theories. If stratification theories or institutional models.
function as cultural ideologies, this may help Inattention to collective and cultural aspects
explain the spreading dominance of education. of stratification has been a substantial limitation
The conventional story, conveyed repeatedly in the field. It has led to extraordinarily mistaken
in this fashion, has some pronounced weak- theoretical analyses and predictions (as with the
nesses: If expansion is the driving force, why families of theories that have overemphasized the
does so much differentiation occur (e.g., in the causal and descriptive role of economic dimen-
modern Third World) even without expansion? sions in stratification). In addition, it has led to
If education and the new class are relied upon for inattention to some obviously cultural aspects
technically efficient solutions, why can we not of stratification. In this chapter, corrective re-
find better evidence for their actual efficiency flections are suggested. We offer cultural inter-
(e.g., Berg 1971)? Above all, if stratification re- pretations of modern differentiation, the rise of
flects real social requirements, why are the forms education and the service sector, and the overall
of stratification and mobility so strikingly simi- homogeneity of modern stratification systems.
lar across all the extreme variations of modern The cultural perspective we emphasize is an
society? alternative to what we are calling sociological re-
There is a strong sociological tendency to alism on several dimensions that in a broader
leave cultural matters out of the equation, aside discussion could be distinguished. We may note
from some notions that individuals in society two of them here. First, modern sociological re-
may have some socialized tastes in common. alists see society as made up of individuals and
When culture is brought in, as in postmodernist groups that have prior and natural properties
accounts, it is seen as subjective and individual (e.g., motives and interests). Their strivings for
in character—matters of taste and status—rather equality and their inequality result from these
than institutionalized collective rules and mod- properties and their interaction in social life.
els. Indeed, in a strain to see education as func- A more cultural view treats the existence and
tional, these authors almost all deemphasize it as standing (and motives and interests) of these
a carrier of cultural content. The modern strati- entities as highly collectively constructed—as
fication theorist talks about education as human the product of meaning and interpretation in
or cultural or social capital, not as culture per se. modern schemes and scripts. The construction
The idea that modern societies may with some of modern citizens as formally equal, for exam-
collective self-consciousness be, through direct ple, is seen as an evolution in culture and theory
1118 • Part X: The Future of Inequality

rather than as either natural equalities or a result ignore inequalities of various sorts. Stratification
of interests in interaction (e.g., mystifying tactics theorists may understate the importance of the
or compromises of dominant elites or function- cultures of modern systems because stratification
ally necessary responses to social complexity). theories (though not necessarily only narrowly
Second, sociological realism tends to empha- academic ones) are the central cultural elements
size social systems as affected by high levels of involved. Recognizing the cultural character of
interactive interdependence among their com- stratification would weaken the realist vision of
ponents—as really social systems of interdepen- society as a real system. It would also undercut
dent parts. A more collective and cultural view some of the scientific claims of stratification the-
calls attention to the impact of cultural theories orists: If stratification is cultural theory, the the-
and ideologies on this putative interdependence: orists (as classic phenomenological critics have
The high status and importance of many groups, had it) are studying themselves.
such as the schooled professions, in modern Our point here is not to support the
stratification systems arise from cultural ideolo- often-noted special importance of some intel-
gies (which in the modern world commonly take lectuals and scientific professions in modern
the form of functional theories) as much as from society. It is to note that explicit theories and
“real” interactive dominance and dependence. ideologies about licit and illicit inequalities are
deeply ingrained in the public discussions of the
modern system. Most great political conflicts in
Conceptions of Culture this system are explicitly and articulately about
Part of the problem lies with a very limited equality and inequality: Contemporary opinion
conception of culture as individual attitudes or surveys suggest that even ordinary persons are
vague collective sentiments. In fact, most sociol- able to go on at length about inequalities and
ogy deeply shares the contemporary bias that the conditions under which they are legitimate.
modern systems (unlike the traditional ones that Indeed, it is for this reason that the system tends
anthropologists study) are so overwhelmingly to pick out for great attention some of the ideas
articulated and real that they do not have very of intellectuals (e.g., Smith or Marx) who might
much culture: In other words, the rationalism be quite obscure in other traditions. The culture
and functionalism of the cultures of modern so- of the system tends to make stratification theo-
ciety are taken as true by the analysts. The post- ries important (rather than the theorists simply
modernists, such as Pakulski and Waters (2008), having some sort of mysterious dominance over
do see the postmodern system as having greatly the culture); carriers include everyone from or-
increased scope for cultural expression and strat- dinary members to political and economic elites
ification. They see culture as playing a more in- and specialized intellectuals.
dependent and autonomous role in the future, Our emphasis on a cultural reinterpreta-
although the culture they envision tends to be tion of stratification can be pursued in either
individual and subjective rather than a system of a strong or weak form; the difference between
collective rules. the two is immaterial here, for the most part.
Another way to put this is to note that The strong form (Meyer 1989; Mann 1986;
social-scientific stratification theories are them- Hall 1986) treats the modern (Western and now
selves core cultural elements of modern soci- worldwide) stratification system as having dis-
ety. They are institutionalized as constitutional tinctive cultural roots in such themes as equal-
principles, but they also may provide cultural ity and progress. A weak form might argue that
bases for criticism and opposition. Although more narrowly “real” social forces produced such
this is obvious with Marxist theories, it is also Western structures as high capitalism or polit-
the case with most others, which provide nor- ical democracy, after which institutionalization
mative bases with which to defend, attack, or turned such arrangements into cultural recipes
 Meyer—The Evolution of Modern Stratikcation Systems • 1119

to be repeated throughout the world. Either is required to promote more investment, or

form is good enough for present arguments. more pay and status are needed for teachers if
We briefly review the cultural postures built we wish to select better ones and improve the
into the stratification theories of modern soci- performance of all, or the complex society re-
eties. That is, in discussing what seem to be the quires more educated intellectuals at its center.
most prominent substantive themes in stratifica- The idea of social progress is obviously involved
tion theories, we recognize that modern systems in the principle.
are much affected by their cultures and conceive The idea that inequality is justified only
of stratification theories as prominent parts of or mainly by functional considerations is so
these cultures. One important theme is that per- firmly established (e.g., Rawls 1971) that so-
sons are formally and ultimately equal; another ciological critics of particular inequalities must
is that inequalities can be justified only by con- argue that they are not really functional but are
tributions to collective progress. Many lines of rather products of tradition or power (as seen in
sociological thought incorporate such matters Esping-Andersen). In such analyses, socially use-
as aspects of social reality; we treat them as the ful achievement is set in opposition to two de-
cultural rules of modern collective actors. We go fective sources of inequality; discussing the issues
on to consider the consequences these cultural involved is a modern (and a sociological) obses-
postures may have—the things that can be better sion. One opposition is mainly with the past and
explained when we recognize the cultural charac- with tradition; achievement is set against ascrip-
ter of stratification theory in the modern system. tion in a drama that combines issues of social
Sociological thinking routinely recognizes efficiency with those of social justice. Thus there
that stratification systems in the premodern past is great attention to whether favorable evalua-
were cultural constructions. Further, there is a tions or opportunities are provided to students
tendency to imagine that in some future utopia or workers on merit or whether there is a direct
(or dystopia) where inequalities are put right effect of status background. Even if the analy-
(or wrong), cultural principles could again hold ses suggest merit is mainly involved, analysts
sway. This is, obviously, a dramatic counterpoint go further—perhaps the standards of merit are
to the aggressive realism about the present that status-biased, or perhaps the resources of those
is characteristic of Marxian theory. We apply the higher in status permit easier or more achieve-
same perspective to contemporary modern soci- ment. A second opposition is to set socially use-
ety, seeing its stratification system as in part the ful and legitimate achievement against power
realization of its cultural visions. that may pervert goals or means (as when he-
gemonic capitalism is thought to substitute the
requirements of capital for the good of society).
Stratification Principles The issues involved here infuse the stratifica-
Clearly a first cultural principle built into the tion literature as well as modern social discussion
stratification models of modern theories and with much normative excitement. Minor details
societies is the functional one. Society is a sys- of the distribution of income or education be-
tem—a rationally analyzable set of components come salient, and analyses of mobility give great
organized as a purposive project—and inequal- attention to causal pathways distinguishable ac-
ities among its parts are justified by inequalities cording to their justice and efficiency. On the
in their contributions to collective goods and one hand, there are many conflicts about what
goals. For example, one can defend or criticize is really functional: Radical differences in the in-
enhanced inequalities mainly by functional ar- terpretation of the functions of private capital or
guments. The chapters cited above (particularly of education can produce very different assess-
those by Bell, Gouldner, Nee, and Eyal et al.) ments. On the other hand, broad bands of agree-
do just that: A greater return to private capital ment can be found: There are few defenders of
1120 • Part X: The Future of Inequality

extreme inequalities, of high levels of status con- of the chapters in this book). Both actors and ad-
sistency, or of high levels of status inheritance. visors, of course, are concerned about the princi-
No such defenses occur in the chapters here. ples of functionality for the collective good and
All this takes on more momentum given the the ultimate equality of individuals. This is not
second crucial principle of modern stratification so unrealistic an image: Most modern stratifica-
(and theory): moral individualism. Individuals tion theorists, including the authors here, work
are to be ultimately morally equal, and justice is with just these standards, and most (including
to be assessed in terms of inequality among indi- most of the authors here) aspire to affect pub-
viduals. The notion goes unquestioned in these lic policy; more to the point, decision-making
chapters, and little reference is made to any pos- elites in modern systems operate in about the
sibility of natural inequalities between persons. same way.
There is considerable tension, naturally, between
this principle and the legitimation of inequality
on functional grounds; we later suggest that this Theory of Progress
tension is a source of a good deal of modern so- Some effects of the legitimation of inequality
cial change. in terms of the rational pursuit of progress are
The individualism of modern stratification noteworthy. First, it seems obvious that ratio-
theory and research is striking. The important nalistic and progressive ideas of the sort devel-
inequality is, for instance, individual income oped in the chapters here have played a role in
inequality. We do not much care if the genders supporting a great deal of modern social differ-
(or races or families or ethnic groups or groups entiation. Activities can be pulled out of social
of differing class origins) have equal incomes. life and bundled into rationalized elements or
In fact, calculations of the income of a gender, roles; their inequalities thus are legitimated. The
race, ethnic group, or family as a group are very modern system has been profligate in its creation
rare, in contrast to stratificational thinking in of new occupations, tasks, organizational forms,
societies in which corporate groups (e.g., clans and so forth. Not all of this is easy to explain in
or villages) are more real than individuals. We terms of standard arguments about the real func-
care only, on a per capita basis, if individuals tional requirements of modern development: It
of each gender (or class background or race or is still utterly unclear what actual (as opposed
ethnicity) have equal incomes. When we attend to culturally defined) functional requirements
to cross-national comparisons, we calculate not call for such occupations as sociologist or such
income differences among countries but income organizations as therapeutic ones. Rationalistic
differences per capita. On some questions, other models not only helped justify the evolution of
units have moral standing (as when each country these minor institutions but also played a role in
has certain rights in international relations, or as the historically crucial early Western institution-
when each family or town may be represented in alization of the capitalist class. Thus, the modern
certain activities), but the utterly dominant unit tendency toward extreme differentiation of roles
is the individual: Justice is assessed in terms of may reflect cultural legitimating pressures as
the equality or inequality of individuals. much as instrumental functional requirements;
this might help explain the exceptional differen-
tiation characteristic of egalitarian U.S. society
The Impact of Stratification in contrast to some other modern systems.
Ideology on the Development Second, strong and highly culturally theo-
of the Modern System rized notions of progress and the collective good
Imagine, thus, that the social changes of the help explain the homogeneity of modern stratifi-
modern period are going on under the continu- cation regimes (across country and time) despite
ing scrutiny of collective actors and the intellec- great heterogeneity in immediate circumstances.
tual advisors of those actors (such as the authors The finding that occupational prestige systems
 Meyer—The Evolution of Modern Stratikcation Systems • 1121

and mobility regimes are relatively homoge- (especially those by Bell, Gouldner, and Nee).
neous has been an overpowering surprise in the However, critics have noted the utter absence of
research literature. One explanation has been evidence of such requirements and have seen the
a very broad functional image, but this makes expansion as resulting instead from conflict pro-
little sense in light of the extreme variations in cesses (Collins 1979). The conflict explanation
any plausible functional requirements among is incomplete. Why would so many forces make
modern societies. A more realistic explanation education, as opposed to other institutional set-
is to see the cultural (functional) theories of tings, the new arena of competition and conflict?
modernity as widespread and as generating a Education becomes the competitive arena in part
good deal of stratificational isomorphism: Na- because myths of rational progress have made
tional communities vary enormously, but mod- educational institutions culturally central.
ern stratification is legitimated in terms of ideal Finally, we may note the extraordinary—and
pictures of society, and these models are highly in the main, unpredicted—rise of the service sec-
homogeneous (e.g., Anderson 1991 calls most tor and its new classes: This, along with the rise
modern societies “imagined communities”). of education, is the main theme of the chapters
This would also help explain why stratificational above, and in fact integrates them all. By and
change (e.g., in the status of women or ethnic large, the essays attribute the expansion of the
minorities, in the welfare correctives for the class service sector to the functional requirements of
structure, or in educational opportunity) tends the complex society. This is unconvincing be-
to be global in character (Thomas et al. 1987). cause of the absence of empirical evidence that
Third, the institutionalization of ideas of ra- expanded service sectors are indeed functional
tionality and progress can help explain the sur- in this sense. It becomes yet more unconvincing
prising worldwide rise of education as the critical when one realizes that the expansion of the ser-
dimension in all modern stratification systems. vice sector is worldwide—little predicted by the
Education provides a clear legitimating account evolutionary development of complexity. Some
of sources of improved capacity and also locates social theories attribute great feats of power and
this capacity in specific persons and groups manipulation to the professions and the state but
(Meyer 1977). Thus, ascriptive and income/ give no real explanation about where the power
power considerations and even some of the func- and capacity come from. The cultural commit-
tional ones of occupation tend to be replaced ments to rational progress previously noted may
by education as the main dimension of status: provide some explanation for why societies sup-
Education tends to correlate more highly with posedly dominated by their economies (Marx)
more outcomes than the other dimensions do, or their state bureaucracies (Weber) in fact turn
and intergenerational educational correlations out to have the professions (analogous to, of all
are usually higher than the others. Empirical things, a priesthood) as their leading groups.
researchers routinely build education into their
measures, but both researchers and theorists are
slow to let go of their nineteenth-century (cul- Effects of Individualism
tureless) models of society and continue to use There have been several effects of the institution-
terms like class or socioeconomic status when edu- alization of individualism and equality on strat-
cation is the main phenomenon at issue in their ification. A first consequence is the expansion of
measures. This peculiarity of labeling variables in social citizenship and of the dimensions of social
modern stratification research is quite revealing life it comprehends. Stratification theorists have
of its ideological base in modern culture. been so engrossed in the normative search for
The worldwide rise in educational certifica- illicit inequality as to have avoided noticing the
tion for all sorts of social positions has been dis- rise and expansion of a most important social
cussed as resulting from functional requirements; status in terms of which all are formally equal.
this is the approach taken in the chapters above For example, Parsons (1970), until his last
1122 • Part X: The Future of Inequality

discussion of stratification, essentially omitted coordination) and the equality of persons gener-
citizenship as an element. With the exception ates a great deal of role differentiation. As already
of Esping-Andersen (who treats the standard of noted, theorists such as the authors of the chap-
equality as culturally constructed), the scholars ters included here have tended to understate the
represented here simply take for granted that cultural bases of such differentiation. However,
persons are or should be naturally equal, and differentiation clearly has served to protect cen-
that inequalities are problematic and require tral myths in the modern system in two ways:
justification. The chapters are written as if this (1) Elaborate role differentiation and the tight
point of view is special to the authors, rather organization of activity systems (e.g., in such
than one of the most widespread cultural prin- arrangements as modern formal organizations)
ciples in the whole modern world, in which the help to sustain functional myths and to define
banner of human rights has the highest standing. activity as primarily linked to social tasks rather
Clearly, citizenship is now one of the more than personal status, and (2) more important,
important dimensions of status throughout the the modern system notoriously differentiates ac-
world—and on such dependent variables as tivities and roles from persons.
income perhaps the most important one. Ob- The second point is critical because all sorts
viously, claims for its expansion (in more wel- of functional roots of this in efficiency pressures
fare arenas) have been continuous throughout (for the most part, unproven) are alleged by
the modern period (and have tended to increase the theorists. It is more important to note how
inequality among citizens of different countries much such individual/activity differentiations
and thus to redefine world stratification). Sec- help sustain the principle of individual equality.
ond, the importance of individual equality has In a modern system, we can have a societywide
tended to expand education. Not only is educa- coordination of activity (e.g., public tax policy or
tion the crucial certifier of inequality, but mass other political and economic decision making)
education is a most important constructor of implying the utter domination of the activities
equality as well. Third, the importance of indi- of millions or billions of people by the activi-
vidual equality has produced considerable ex- ties of a few, without any serious violation of the
pansion of protection of young and old persons normative stratificational principle of equality.
(including extensions of personhood to prenatal In a system of this kind, the activities control
periods). In this area, as in others, the rise of pro- each other but not the persons: Differentiation
fessional protectors and the growth of organiza- is crucial to sustaining the principle of individual
tions with some type of protective agenda have equality.
been the most prominent developments. Consider how this works out in terms of
measures of income inequality. As Lenski (1966)
noted, agrarian societies (e.g., medieval states)
Equal Persons, Unequal Roles had very high levels of income inequality; the
We next consider the impact of simultaneous disparity is greatly lowered by the modern sys-
emphases on equal individuals and unequal con- tem. However, in the medieval world all the
tributions to rational progress. Many observers income of the state was attributed to the king
have noted that the dualistic Western (and now personally, so that the king appeared to be very
worldwide) emphasis on both individual and rich. In the modern world, a differentiation is
collective good has been a mobilizing dynamic established: A president gets only a modestly
in modern history. We consider a few ways in high salary (higher than the best sociologists
which this works out for stratification. but lower than the best economists) personally
but may control a monumental budget as a
Role Differentiation. The simultaneous stress property of the presidential role. If this budget
on the inequality of activities (i.e., societywide were attributed to the president personally, as in
 Meyer—The Evolution of Modern Stratikcation Systems • 1123

medieval accounting, the modern system would or legitimating significance. A sociologically

show enormously high levels of inequality (a typ- realist perspective treats all this differentiation
ical head of a modern state would be recorded as as arising step-by-step out of the pressures of
having up to half the national income). Neither organizational life. Such a perspective has great
the medieval accounting nor the modern differ- difficulty explaining why similar differentiation
entiated one is wrong: The point is that because occurs in many different places (regardless of
a great change in organizational accounting has variation in local problems or needs), is hard
occurred, enormous societywide political control pressed to explain the fact that much of the dif-
can result with only modest levels of accounted ferentiation seems only symbolic in character,
interpersonal inequality. and must resort to unsubstantiated arguments
As a result of the modern differentiation be- about the presumed efficiencies involved in dif-
tween person and role, decreasing proportions ferentiation. The idea employed here—that the
of monetary flows are recorded as individual in- functional theory involved is acting as a cultural
come, and the well-known growth in all sorts of principle—simplifies the problem.
corporate actors occurs. Our point is that one What are the mechanisms by which cultural
pressure in this direction arises from the mod- commitments produce this sort of differentia-
ern demand for individual equality, combined tion? In our argument, they occur through the
with the modern pressures for rationalization relatively articulate and explicit activity of col-
and progress. Differentiation permits increas- lective actors: public political elites, intellectual
ingly equal individuals in a system of increas- and professional theorists, and the like. Modern
ingly unequal and ordered activities: It involves, systems are constantly under scrutiny by cul-
of course, an elaborate modern differentiation tural theorists looking for precisely such illicit
between the tastes and motives attributed to inequalities. Both tax laws and common opinion
persons and those attributed to organizations may penalize the insufficient differentiation of
(Meyer 1986). personal and organizational resources and reward
Consider how both forms of differentia- instances of clear and “rational” differentiation.
tion—between persons and roles and between
different roles—help sustain equality in a mod- The Rise of Educational Credentialism. The
ern organization like a university. Some equal- tension between equal individuals and un-
ities in status and salary among professors are equal roles can be seen as one of the sources
maintained, but some professors may control of ever-expanding educational credentialism
huge research grants, hiring many servants and and its related institutions. More and more so-
much equipment; in both social custom and in cial boundaries and mechanisms are installed
tax law, this is not attributed to their persons but to keep the structural inconsistency involved
to their roles. Further, if the inequalities among maintained. There are the educational creden-
professorial roles become too great, a further tials themselves, which can be used to legitimate
differentiation occurs, such that the wealth of the allocation of formally equal individuals to
one professor is located in a specialized insti- unequal roles. The legitimation problem then
tute and controlled by the professor as director reduces to accounting for inter-individual vari-
of the institute. This control can produce social ations in the “investment” in education. There
accountings reflecting great equality of income is no shortage of ancillary social theories with
and power among the professors as persons but this orientation. Moreover, there are also theories
great inequality among their roles or activities. of variations in abilities as well as in tastes that
The role differentiation of the modern system treat role differentiation as a matter of individual
(between roles and between roles and persons) choice. Such effects work through the mecha-
solves cultural problems related to stratifica- nism of relatively articulate collective conscious-
tion theory and can be seen as having cultural ness and action. Clearly, many groups advocate,
1124 • Part X: The Future of Inequality

and states often require, the replacement of less service sector, their high levels of differentiation
legitimated and more ascriptive criteria of role among roles (and especially between roles and
entry by educational credentials. This is argued persons), and the worldwide expansion in citi-
to indicate merit, to deserve public respect and zenship statuses.
trust, and thus to enhance legitimate status. We can illustrate these points by noting that
in recent decades, the world itself has come to
The Personnel Society. All of these characteristics be conceptualized as a unitary social system or
are institutionalized in an expanding modern society. The shift to such a conception is rap-
personnel bureaucracy. A surprising feature of idly changing public and scientific stratifica-
modern organizational development, from the tion analyses: Distinctions among individuals
point of view of most theories, has been the in different countries now require rectification
elaboration of the citizenship of persons within (from the point of view of both dependency and
organizations. Many individual rights have ex- human rights ideologies), and a differentiated set
panded to provide organizational protection of of world institutions is rapidly developing. The
individuals from potentially threatening inequal- ultimate goal of stratification theories—both
ities. Such bureaucracy is seen as both more pro- lay and scientific—is no longer the articula-
gressive and more just. tion of equal individual personhood combined
with social progress at the nation-state level; a
bigger arena will presumably undercut some of
Conclusions the structures of the older one. For example, it
Two closely related themes arise from our discus- may delegitimate some efforts to produce equal-
sion. One conclusion is about modern societies: ity within developed nation-states at the cost of
Many aspects of their stratification systems re- enhanced international inequality.
flect cultural properties of the modern system,
as with the expansion of education, the service NOTES
sector, the personnel bureaucracy, and so forth. This paper benefited from useful comments
on earlier drafts by David Grusky, Ron Jepperson,
This was mostly unexpected by theorists empha-
David Strang, Marc Ventresca, and members of the
sizing the economic or political realism of the
Stanford Stratification Seminar.
modern system: From their point of view, much
of the modern social structure reflects distortion, REFERENCES
false consciousness, or mystification. The theo- Anderson, Benedict. 1991. Imagined Communities.
ries tend to be mistaken, precisely in their inat- 2nd ed. London: Verso.
tention to the modern system’s sensitivity to and Berg, Ivar. 1971. Education and Jobs: The Great
cultural analysis of its own stratification system. Training Robbery. Boston: Beacon.
A second conclusion is that the underlying Collins, Randall. 1979. The Credential Society: A
culture of the modern stratification system is Historical Sociology of Education and Stratifica-
tion. New York: Academic Press.
closely tied to the social-scientific analyses we
Hall, John. 1986. Powers and Liberties. New York:
call stratification theory. The obsessions of the-
Penguin.
ory (e.g., with individual inequality and with the Kerr, Clark, John T. Dunlop, Frederick H. Harbison,
distinction between just and functional inequal- and Charles A. Myers. 2008. “Industrialism and
ities and unjust or power and ascription-ridden Industrial Man.” Pages 954–965 in Social Stratifi-
ones) are the main cultural themes of modern cation: Class, Race, and Gender in Sociological Per-
stratification. spective, edited by David B. Grusky and Szonja
These cultural commitments in fact drive Szelényi. Boulder, CO: Westview Press.
social change. This can help explain many of Lenski, Gerhard. 1966. Power and Privilege. New
York: McGraw-Hill.
the puzzling aspects of modern stratification re-
Mann, Michael. 1986. The Sources of Social Power.
gimes: their homogeneity across time and space,
Cambridge: Cambridge University Press.
their constant shifts toward education and the
 Giddens—Social Justice and Social Divisions • 1125

Meyer, John. 1977. “The Effects of Education as Stratification: Class, Race, and Gender in Sociolog-
an Institution.” American Journal of Sociology ical Perspective, edited by David B. Grusky and
83(July):55–77. Szonja Szelényi. Boulder, CO: Westview Press.
———.1986. “The Self and the Life Course.” Pages Parsons, Talcott. 1970. “Equality and Inequality
199–216 in Human Development and the Life in Modern Society, or Social Stratification Re-
Course, edited by A. B. Sørensen, F. E. Weinert, visited.” Pages 13–72 in Social Stratification,
and L. R. Sherrod. Hillside, NJ: Erlbaum. edited by Edward Laumann. Indianapolis:
———.1989. “Conceptions of Christendom.” Pages Bobbs-Merrill.
395–413 in Cross-National Research in Sociology, Rawls, John. 1971. A Theory of Justice. Cambridge:
edited by M. Kohn. Newbury Park: Sage. Belknap.
Ossowski, Stanislaw. 1963. Class Structure in the So- Thomas, George, John Meyer, Francisco Ramirez,
cial Consciousness. New York: Free Press. and John Boli. 1987. Institutional Structure.
Pakulski, Jan, and Malcolm Waters. 2008. “The Newbury Park: Sage.
Death of Class.” Pages 1022–1029 in Social

130 • Anthony Giddens

Social Justice and Social Divisions

The class structure and class divisions associated generated by the new economy are skilled—they
with the knowledge/service economy are quite demand a technical knowledge of information
different from those of the industrial age. The technology (IT) and other skills too. Such jobs
blue-collar working class was the largest class are becoming more plentiful in relative terms.
grouping in the old social order. Marx labelled Over the period 1995–2004, the proportion of
the working class the ‘universal class’, and of jobs in the EU151 demanding advanced qualifi-
course expected it to be the agent of revolution- cations in IT—‘Apple Mac’ jobs—went up from
ary change. Today, the manual working class is 20 percent to 24 percent. But many people have
very much a minority, and its numbers are set to work in much more routine ‘Big Mac’ jobs—
to reduce further as the proportion working in serving in cafés, shops, supermarkets or petrol
manufacture continues to diminish. The old stations. The proportion of such jobs may be de-
working-class communities, that used to be the clining compared to more skilled occupations,
source of much local solidarity, have largely bro- but it remains substantial.
ken up. What used to be the ‘middle class’ has The class structure of a post-industrial society
become much more differentiated, while the looks as portrayed in Figure 130.1. This picture
land-owning upper class has largely disappeared. is a generalized one, since different societies vary
Separate agrarian classes have more or less com- between themselves to some considerable degree.
pletely evaporated too. The percentages given are only rough guides,
New occupational divisions have come into and refer to individuals, not households. The
existence, based upon the social and techno- group at the top is an amorphous one of elites
logical changes associated with the knowledge/ whose power and outlook are as much transna-
service economy. About two-thirds of the jobs tional as national, especially in Europe’s ‘global

Anthony Giddens. “Social Justice and Social Divisions,” in Europe in the Global Age, pp. 61–71. Copyright ©
2007 by Polity Press. Used by permission.
1126 • Part X: The Future of Inequality

Figure 130.1. Classes in Post-Industrial Society (% of Population)

cities’. They may include substantial proportions become reduced almost everywhere. Moreover,
of immigrants. The French banker working in at the bottom end women are actively doing bet-
the City of London leaves his office in the early ter than men—declining wage levels and job se-
evening. As he departs, a cleaner from the Phil- curity are much more marked among low-skilled
ippines, who is regularly sending money to her working-class men than among women of com-
family there, comes in to tidy up after him. parable skill levels. Crucially, the role of women
At least 50 percent of jobs in the knowledge/ in providing household income is increasing
service economy demand a high level of cogni- dramatically. In Denmark, it is reaching parity
tive and/or personal skills, stretching down into with men, at 42 percent, although at the other
work performed by ‘wired workers’—those who end of the scale, in Spain and Italy, it is as low
use computers much of the day, but are not as 27 percent.2
themselves IT specialists. All class categories Low pay does not necessarily mean that a
can involve those who work for the state or in person lives in hardship if he or she is a sup-
state-controlled enterprises. The distinction be- plementary earner in the household. Household
tween those who work for the state and those inequality is substantially lower than individual
who don’t tends to coincide with divisions in la- inequalities in most EU15 countries. Households
bour markets; but job security is normally high- most likely to be in poverty are single-earner
est for those in classes 1, 2 and 3. households and especially households where no
One of the distinguishing features of the one is in work. The proportion of households
knowledge/service economy is the high propor- in the EU15 with no one working ranges from
tion of women in the labour force—although 15 percent in the UK to 6 percent in Denmark.
few work in manual occupations. Women are The changing class structure has altered the
heavily represented in ‘Big Mac’ jobs and among nature of politics—and much else besides. In the
wired workers. They are also more likely to work old society, politics was largely shaped around the
part time than men are. Only 7 percent of men dividing lines between the manual working class
in the EU15 countries work part time, while 30 and the rest. There were always ‘working-class
percent of women do. Women’s wages are on av- Tories’, and social democratic parties needed
erage about 15 percent lower than those of men to appeal to other groups besides the working
doing the same or a comparable job. class. But political activity correlated strongly
These gross figures mask some important with class, and voting preferences tended to be
trends. The gender gap in pay is tending to quite stable from election to election. Now the
 Giddens—Social Justice and Social Divisions • 1127

major political parties have to appeal to diverse the working population are in stable jobs, 40
constituencies, while many more voters are percent are in more insecure work, while the
‘dealigned’—they do not automatically stick to remaining 10 percent consists of the socially ex-
the same party. cluded, either unemployed or shifting in and out
The working class was (and is) very much of the labour market (although the percentages
divided along skill lines. Skilled workers, ei- varied in different versions). An underclass, in
ther craftsmen or those who worked in various other words, has replaced the traditional work-
areas of industrial technology, normally had ing class, or the lower levels of it. A certain per-
quite stable jobs, and often were more affluent centage of the population is cut adrift from the
than clerical workers and others in the lower wider society.
reaches of the middle class. Skilled workers in However, this notion has proved to be in-
established trades that continue to thrive can correct.4 Sociologists are agreed that there is no
still be in much demand—consider the fabled distinct underclass in this sense in the EU coun-
Polish plumber. However, the position of the tries—although the notion may have more appli-
unskilled has worsened. Unskilled workers—es- cability in the US. Thus the government Strategy
pecially men—have poor work opportunities, Unit in the UK studied four measures of social
above all in areas where manufacturing indus- exclusion—those who are not in employment,
tries have closed down. Even ‘Big Mac’ jobs education or training; those with a low income
involve face-to-face social skills that those from (below 60 percent of median income); those
traditional working-class backgrounds may find who have few established social ties with others;
it hard to master. Many men from such back- and those who perceive themselves as living in an
grounds are in any case unwilling to do what area marked by high levels of crime, vandalism
they see as ‘women’s work’. Unskilled men are or material dilapidation. Only 1 percent of the
thus highly vulnerable to spells of poverty or un- UK population is excluded on all four of these
employment. This situation often applies with measures. Multiple deprivation does exist, but it
particular force to migrants or those from mi- tends to be concentrated in specific neighbour-
nority groups. hoods rather than affecting a ‘class’ of people.5
To refer to the situation of groups at the bot- Social exclusion is a concept with some value,
tom of the socioeconomic scale, the term social but it must not be used loosely and generically.
exclusion has been widely popularized. The point It has essentially the same meaning as ‘multiple
of the concept is to recognize that factors other deprivation’, which tends to affect pockets of
than poverty alone might prevent individuals neighbourhoods and certain groups of individ-
or groups from playing a full part in society. uals in particular (such as the homeless).
The notion is usually traced to the work of the Class divisions in post-industrial society by
French author René Lenoir. The ‘excluded’ in his and large are no longer behavioural, but are
analysis included a variety of groups, making up determined by differential life-chances. This
about 10 percent of the population—not only change is a very significant one. At one time in
the poor, but also the disabled, mentally ill, older most countries people in different class groups
people, drug users and so forth. The introduc- could be distinguished from one another quite
tion of the concept was important, because it visibly. Lifestyle variations still exist between
drew attention to the fact that the existing wel- those in different class groups, but they are often
fare provisions often did not reach the people in more influenced by taste and custom than by
these categories; and because it stressed that it is sheer financial constraints. Physical mobility,
not only sheer economic deprivation that pre- including foreign travel, is available to virtually
vents individuals from realizing their potential.3 everyone. The substantive freedoms most people
At one time it became common to speak of have, centred around freedom of lifestyle choice,
the ‘50/40/10’ society in Europe: 50 percent of today are greater than for past generations. I shall
1128 • Part X: The Future of Inequality

call this phenomenon everyday democratization. of gays being taken as a proxy of openness.
Everyday democratization tends to stretch both Some cities, such as San Francisco, Austin or
‘downwards’ and ‘upwards’ in the life-course. In San Diego, rank high on all measures. Others
the age of the Internet, childhood can no longer are cities bypassed by the creative class, such as
be as sheltered as it once was; and older peo- Memphis or Pittsburgh.
ple feel free to experiment with lifestyles just as Those cities near the top of the Florida list
much as younger ones do. head the country in terms of economic prosper-
Everyday democratization does not necessar- ity and job creation. The liveliest local econo-
ily bring greater security, or feelings of security. mies are characterized by the ‘three Ts’—talent,
In fact a series of new insecurities come into technology and tolerance. They are above all
being alongside it. Some are directly economic, cosmopolitan. The members of the creative class
others more social in nature. Most people want, are very mobile, and will gravitate towards those
and expect, more from their lives than previous cities in areas that offer what they want in terms
generations did, leading to aspirations that can- of lifestyle. They prefer active, participatory rec-
not always be realized. reation and street-level culture—a blend of cafés,
The degree of experienced security of dif- restaurants, galleries and theatres. They are into
ferent class groups tends to produce new ideo- a variety of active sports. They involve a high
logical rifts in post-industrial societies. Such proportion of qualified migrants. According to
cleavages are partly based on rational anxieties one study, almost a third of all businesses set up
(for example, fears of job loss) and partly on in Silicon Valley in the 1990s were started by
more free-floating worries. Those who flourish Indian and Chinese-born entrepreneurs.
in the new economy tend to be happy with di- Some cosmopolitans in Florida’s sense may be
versity, and embrace cosmopolitan lifestyles. found in other class groups too, given the avail-
Some may actively court what is seen negatively able range of possible lifestyles. However, many
as insecurity. For instance, they may relish the who feel vulnerable to unwanted change are
prospect of moving on from job to job, and nei- better characterized as locals, who rather want to
ther expect nor want to have a hierarchical career stick to the existing order of things, or who look
of the traditional sort. The creative industries, back nostalgically to the past—real or imagined.
high-tech jobs, finance and banking and profes- They might look for scapegoats, for example im-
sional occupations are where such groups tend to migrants, upon whom to blame their troubles;
be found in highest density. and they might be attracted by political popu-
What Richard Florida calls the ‘creative class’ lism and economic protectionism. Immigrants
now comprises well over 20 percent of the labour themselves are not necessarily cosmopolitans. In
force, with a high concentration in certain met- their attitudes they may become locals, just as
ropolitan areas. It is mainly made up of people hostile to further immigration as groups in the
in classes 2 and 3 in my categories (see Figure native population.
130.1). The members of the creative class are di- Coupled to the other changes, especially in
verse in terms of background—it includes peo- the sphere of the family, the changing class struc-
ple of all ages, from different ethnic groups and ture alters the distribution of ‘at risk’ groups, as
of differing sexual orientations.6 well as the nature and form of inequalities. The
To measure the distribution of the creative conditions generating ‘at risk’ groups are struc-
class in the US, Florida developed a ‘creativity tural, but how far they translate into real vul-
index’, based on four factors: the percentage of nerabilities depends upon the policy mix of a
the creative class in the workforce; the percent- particular society, as well as specific blockages it
age working in high-tech industry; innovation, might have. The other side of risk is opportunity.
as measured by patents per capita; and diversity, We must not make the mistake of supposing that
as assessed by the ‘Gay Index’, the proportion
 Giddens—Social Justice and Social Divisions • 1129

risk is always a negative factor. Some of the most matter a great deal for those who have few
important changes are the following: qualifications to allow them to move on.
1. There are on average fewer jobs with secure 6. In the knowledge/service society, creden-
tenure for people in most work situations, tials—certificates, diplomas, degrees—be-
although there are variations, depending come of the first importance for career
upon the strength of insider/outsider divi- mobility. It is harder for affluent parents
sions in labour markets. to pass on their advantages in a ‘direct’ way
2. Risk (and opportunity) are distributed dif- to their children. Hence they concentrate
ferently across the life-span from how they heavily on education. For example, univer-
were in the past. Risk and opportunity do sity graduates on average earn significantly
not just ‘happen’ to people. More and more more over the course of their careers than
people think strategically about their lives those who do not go to university.
in terms of future possibilities—including 7. Ethnic minorities may be significantly at
the decision to have a child/children. Tran- risk where they lack qualifications. Preju-
sitions at different phases of life are both less dices can be reinforced, and to some extent
predictable and less mechanical than they taken over by the members of the minority
used to be. Heightened rates of divorce and themselves. There can be additional prob-
separation mean that transition can happen lems for women where traditional beliefs
at diverse times. confine their role to a domestic one.
3. The intensity of technological change, con- 8. Not only do many (most) women work, and
joined to a more globalized division of la- for much of their lives, but their incomes
bour, creates new vulnerabilities for some are quite often crucial for sustaining the
groups. Young men with no qualifications, standard of living of a family. In an increas-
as mentioned, are likely to fare especially ing proportion of cases they are the prime
poorly. Older workers in manufacture, earners. This situation is not the cause of
whose jobs disappear, risk long spells of Europe’s low birth rates. On the contrary,
unemployment, or the prospect of never those families with the highest proportion
working again, unless appropriate policy of women in work are also ones with the
interventions are made. most elevated birth rates. However, work/
4. On average, older people hold a greater life issues become of great importance.
share of overall wealth and income than Women are still the main carers and their
they did in the past, compared to the careers are interrupted by having children
young. Many, however, are still at risk of much more than men’s are.
poverty on retirement. Older women liv- 9. There is a great deal of fluidity in contem-
ing on their own are the most vulnerable. porary society, but structural sources of
However, risks have cascaded down more mobility are different from a generation
towards the young rather than the old.7 ago. There will not be as clear a ‘direction’
Child poverty has become commonplace to mobility between the generations as
where child care facilities and opportunities there used to be when many people from
for work for women are underdeveloped. working-class backgrounds moved into
5. ‘Big Mac’ jobs tend to offer little chance of a white-collar and professional jobs. Struc-
career. This situation may not matter where tural mobility will depend upon a continu-
they are carried out by groups who have ing overall upgrading of knowledge-based
the capability to move into other sectors— jobs at the expense of less-skilled service
such as students who work in coffee-shops occupations. There are likely to be many
during vacations or in gap years. It can more voluntary and involuntary career tran-
sitions, often involving a move sideways to
1130 • Part X: The Future of Inequality

a different job area, or taking time out from priorities of social justice in post-industrial social
work for education or retraining. conditions:8
10. Ageing is changing its nature, not just be- 1. The fight against poverty—not just because
cause more people will work longer, but of economic inequality itself, but on the
also because there will not be the same grounds that poverty (above all, enduring
discontinuities as before between working poverty) limits the individual’s capacity for
and non-working life. The old-age ‘retire- autonomy and self-esteem.
ment ghetto’—a form of social exclusion if 2. Creating the highest possible standards of
ever there was one—is being broken down. education and training, rooted in equal and
The very notions of retirement, and even fair access for all.
pensions, might disappear in the future. In 3. Ensuring employment for those who are
their place will come more flexible attitudes willing and able to work.
towards work and more orthodox sources 4. A welfare system that provides protection
of social support for those considered to and dignity.
be significant at-risk groups among older 5. The limiting of inequalities of income and
people, including especially many older wealth if they hinder the realization of the
women. first four goals or endanger the cohesion of
society.
‘Social justice’ is a notoriously controversial
notion. How far does it imply the redistribution Obviously, the devil is in the detail, especially
of wealth and income, as compared to enhanc- in respect of point 5. But the formula provides
ing equality of opportunity? What can it actually a down-to-earth scheme that is both simple and
mean in the context of a post-industrial society? clear. It recognizes the role of equality of op-
The first of these questions can actually be quite portunity in a differentiated society, against the
easily answered on the level of principle. In a so- backdrop of current economic imperatives. It
ciety that depends upon a highly dynamic mar- quite rightly gives pre-eminence to the struggle
ketplace for its prosperity, aspiration, ambition to reduce levels of poverty, since poverty limits
and opportunity have to be central. Equalizing life-chances and the capacity for self-realization.
opportunities is important because it makes the It follows from the formula that targeting child
best use of available talents. However, reducing poverty is of especial significance. The higher the
inequalities of opportunity necessarily involves proportion of those who suffer poverty as chil-
redistribution, since otherwise those who are dren, the more likely it is that all five goals will
successful in one generation could simply hold be compromised.
on to the fruits of their success. In the EU coun-
tries, post-tax inequality is significantly lower NOTES
than pre-tax inequality, a desirable and necessary 1. EU15 comprises the following 15 countries:
Austria, Belgium, Denmark, Finland, France, Ger-
outcome since it sets a framework for other mea-
many, Greece, Ireland, Italy, Luxembourg, Nether-
sures. Further redistribution occurs through the
lands, Portugal, Spain, Sweden, United Kingdom.
welfare system and through the direct effects of They are member countries of the European Union
policy measures, depending on the welfare mix prior to the accession of 10 candidate countries on
a society has. May 1, 2004.
How social justice should be defined is a diffi- 2. Gøsta Esping-Andersen, ‘Inequality of in-
cult issue, about which large philosophical tomes comes and opportunities’, in Anthony Giddens and
have been written. However, a very useful work- Patrick Diamond (eds.), The New Egalitarianism.
ing definition has been provided by the German Cambridge: Polity, 2005, pp. 22–4.
political scientist, Wolfgang Merkel. He lists five 3. René Lenoir, Les exclus: un français sur dix.
Paris: Seuil, 1974.
 Giddens—Social Justice and Social Divisions • 1131

4. See John Goldthorpe and Abigail McKnight, 8. Wolfgang Merkel, ‘How the welfare state can
The Economic Basis of Social Class. London: Centre tackle new inequalities’, in Patrick Diamond and
for Analysis of Social Exclusion, London School of Matt Browne (eds.), Rethinking Social Democracy.
Economics, 2004. London: Policy Network, 2004. Merkel’s standpoint
5. Prime Minister’s Strategy Unit, Strategic Audit draws upon that of Amartya Sen. Sen argues that
of the UK. London, 2003. policies concerned with furthering equality should
6. Richard Florida, The Rise of the Creative Class. centre upon the ‘capability set’ of an individual—the
New York: Perseus, 2002. For the author’s subse- overall freedom a person has to pursue his or her
quent reflections, see Richard Florida, The Flight of well-being. Amartya Sen, Inequality Re-examined.
the Creative Class. New York: HarperBusiness, 2005. Oxford: Clarendon Press, 1992.
7. Esping-Andersen, ‘Inequality of incomes and
opportunities’.
1132 • Part X: The Future of Inequality

Globalization and Inequality

131 • Joseph E. Stiglitz

Globalism’s Discontents

Few subjects have polarized people throughout The international financial institutions have
the world as much as globalization. Some see it pushed a particular ideology—market funda-
as the way of the future, bringing unprecedented mentalism—that is both bad economics and
prosperity to everyone, everywhere. Others, bad politics; it is based on premises concerning
symbolized by the Seattle protestors of Decem- how markets work that do not hold even for
ber 1999, fault globalization as the source of developed countries, much less for developing
untold problems, from the destruction of native countries. The IMF has pushed these economics
cultures to increasing poverty and immiseration. policies without a broader vision of society or
In this [chapter], I want to sort out the differ- the role of economics within society. And it has
ent meanings of globalization. In many coun- pushed these policies in ways that have under-
tries, globalization has brought huge benefits to mined emerging democracies.
a few with few benefits to the many. But in the More generally, globalization itself has been
case of a few countries, it has brought enormous governed in ways that are undemocratic and
benefit to the many. Why have there been these have been disadvantageous to developing coun-
huge differences in experiences? The answer is tries, especially the poor within those countries.
that globalization has meant different things in The Seattle protestors pointed to the absence of
different places. democracy and of transparency, the governance
The countries that have managed globaliza- of the international economic institutions by
tion on their own, such as those in East Asia, and for special corporate and financial interests,
have, by and large, ensured that they reaped and the absence of countervailing democratic
huge benefits and that those benefits were equi- checks to ensure that these informal and pub-
tably shared; they were able substantially to con- lic institutions serve a general interest. In these
trol the terms on which they engaged with the complaints, there is more than a grain of truth.
global economy. By contrast, the countries that
have, by and large, had globalization managed
for them by the International Monetary Fund Beneficial Globalization
(IMF) and other international economic insti- Of the countries of the world, those in East Asia
tutions have not done so well. The problem is have grown the fastest and done most to reduce
thus not with globalization but with how it has poverty. And they have done so, emphatically,
been managed. via “globalization.” Their growth has been based

Joseph E. Stiglitz. Globalization and Its Discontents, pp. 78–80, 254. Copyright © 2002 by Joseph E. Stiglitz.
Used by permission of W. W. Norton and Company, Inc.; “Globalism’s Discontents,” The American Prospect
13:1 (January 1, 2002–January 14, 2002), pp. A16–A21. Used by permission of Joseph E. Stiglitz and The
American Prospect.
 Stiglitz—Globalism’s Discontents • 1133

on exports—by taking advantage of the global work for them; it was when they succumbed to
market for exports and by closing the technol- the pressures from the outside that they ran into
ogy gap. It was not just gaps in capital and other problems that were beyond their own capacity
resources that separated the developed from to manage well.
the less-developed countries, but differences in Globalization can yield immense benefits.
knowledge. East Asian countries took advantage Elsewhere in the developing world, globaliza-
of the “globalization of knowledge” to reduce tion of knowledge has brought improved health,
these disparities. But while some of the coun- with life spans increasing at a rapid pace. How
tries in the region grew by opening themselves can one put a price on these benefits of global-
up to multinational companies, others, such as ization? Globalization has brought still other
Korea and Taiwan, grew by creating their own benefits: Today there is the beginning of a glo-
enterprises. Here is the key distinction: Each of balized civil society that has begun to succeed
the most successful globalizing countries deter- with such reforms as the Mine Ban Treaty and
mined its own pace of change; each made sure as debt forgiveness for the poorest highly indebted
it grew that the benefits were shared equitably; countries (the Jubilee movement). The global-
each rejected the basic tenets of the “Washing- ization protest movement itself would not have
ton Consensus,” which argued for a minimalist been possible without globalization.
role for government and rapid privatization and
liberalization.
In East Asia, government took an active role The Darker Side of Globalization
in managing the economy. The steel industry How then could a trend with the power to have
that the Korean government created was among so many benefits have produced such opposi-
the most efficient in the world—performing far tion? Simply because it has not only failed to
better than its private-sector rivals in the United live up to its potential but frequently has had
States (which, though private, are constantly very adverse effects. But this forces us to ask,
turning to the government for protection and why has it had such adverse effects? The answer
for subsidies). Financial markets were highly can be seen by looking at each of the economic
regulated. My research shows that those regula- elements of globalization as pursued by the in-
tions promoted growth. It was only when these ternational financial institutions and especially
countries stripped away the regulations, under by the IMF.
pressure from the U.S. Treasury and the IMF, The most adverse effects have arisen from the
that they encountered problems. liberalization of financial and capital markets—
During the 1960s, 1970s, and 1980s, the which has posed risks to developing countries
East Asian economies not only grew rapidly but without commensurate rewards. The liberal-
were remarkably stable. Two of the countries ization has left them prey to hot money pour-
most touched by the 1997–1998 economic cri- ing into the country, an influx that has fueled
sis had had in the preceding three decades not speculative real-estate booms; just as suddenly, as
a single year of negative growth; two had only investor sentiment changes, the money is pulled
one year—a better performance than the United out, leaving in its wake economic devastation.
States or the other wealthy nations that make Early on, the IMF said that these countries
up the Organization for Economic Cooperation were being rightly punished for pursuing bad
and Development (OECD). The single most im- economic policies. But as the crisis spread from
portant factor leading to the troubles that several country to country, even those that the IMF had
of the East Asian countries encountered in the given high marks found themselves ravaged.
late 1990s—the East Asian crisis—was the rapid The IMF often speaks about the importance
liberalization of financial and capital markets. In of the discipline provided by capital markets.
short, the countries of East Asia benefited from In doing so, it exhibits a certain paternalism,
globalization because they made globalization a new form of the old colonial mentality: “We
1134 • Part X: The Future of Inequality

in the establishment, we in the North who run discipline. Today, a country that does not treat
our capital markets, know best. Do what we tell capital well will find capital quickly withdraw-
you to do, and you will prosper.” The arrogance ing; in a world of free labor mobility, if a coun-
is offensive, but the objection is more than just try did not treat skilled labor well, it too would
to style. The position is highly undemocratic: withdraw. Workers would worry about the qual-
There is an implied assumption that democracy ity of their children’s education and their fami-
by itself does not provide sufficient discipline. ly’s health care, the quality of their environment
But if one is to have an external disciplinarian, and of their own wages and working conditions.
one should choose a good disciplinarian who They would say to the government: If you fail to
knows what is good for growth, who shares one’s provide these essentials, we will move elsewhere.
values. One doesn’t want an arbitrary and capri- That is a far cry from the kind of discipline that
cious taskmaster who one moment praises you free-flowing capital provides.
for your virtues and the next screams at you for The liberalization of capital markets has not
being rotten to the core. But capital markets are brought growth: How can one build factories or
just such a fickle taskmaster; even ardent advo- create jobs with money that can come in and
cates talk about their bouts of irrational exuber- out of a country overnight? And it gets worse:
ance followed by equally irrational pessimism. Prudential behavior requires countries to set
aside reserves equal to the amount of short-term
lending; so if a firm in a poor country borrows
Lessons of Crisis $100 million at, say, 20 percent interest rates
Nowhere was the fickleness more evident than in short-term from a bank in the United States,
the last global financial crisis. Historically, most the government must set aside a corresponding
of the disturbances in capital flows into and out amount. The reserves are typically held in U.S.
of a country are not the result of factors inside Treasury bills—a safe, liquid asset. In effect,
the country. Major disturbances arise, rather, the country is borrowing $100 million from
from influences outside the country. When the United States and lending $100 million to
Argentina suddenly faced high interest rates in the United States. But when it borrows, it pays
1998, it wasn’t because of what Argentina did a high interest rate, 20 percent; when it lends,
but because of what happened in Russia. Argen- it receives a low interest rate, around 4 percent.
tina cannot be blamed for Russia’s crisis. This may be great for the United States, but it
Small developing countries find it virtually can hardly help the growth of the poor coun-
impossible to withstand this volatility. I have try. There is also a high opportunity cost of the
described capital-market liberalization with a reserves; the money could have been much bet-
simple metaphor: Small countries are like small ter spent on building rural roads or construct-
boats. Liberalizing capital markets is like setting ing schools or health clinics. But instead, the
them loose on a rough sea. Even if the boats are country is, in effect, forced to lend money to the
well captained, even if the boats are sound, they United States.
are likely to be hit broadside by a big wave and Thailand illustrates the true ironies of such
capsize. But the IMF pushed for the boats to set policies: There, the free market led to invest-
forth into the roughest parts of the sea before ments in empty office buildings, starving other
they were seaworthy, with untrained captains sectors—such as education and transportation—
and crews, and without life vests. No wonder of badly needed resources. Until the IMF and
matters turned out so badly! the U.S. Treasury came along, Thailand had re-
To see why it is important to choose a dis- stricted bank lending for speculative real estate.
ciplinarian who shares one’s values, consider The Thais had seen the record: Such lending is
a world in which there were free mobility of an essential part of the boom-bust cycle that has
skilled labor. Skilled labor would then provide characterized capitalism for 200 years. It wanted
 Stiglitz—Globalism’s Discontents • 1135

to be sure that the scarce capital went to cre- folly in liberalizing capital markets. Belatedly, so
ate jobs. But the IMF nixed this intervention in too has the IMF—at least in its official rhetoric,
the free market. If the free market said, “Build though less so in its policy stances—but too late
empty office buildings,” so be it! The market for all those countries that have suffered so much
knew better than any government bureaucrat from following the IMF’s prescriptions.
who mistakenly might have thought it wiser to But while the case for trade liberalization—
build schools or factories. when properly done—is quite compelling, the
way it has been pushed by the IMF has been
far more problematic. The basic logic is simple:
The Costs of Volatility Trade liberalization is supposed to result in re-
Capital-market liberalization is inevitably ac- sources moving from inefficient protected sectors
companied by huge volatility, and this vola- to more efficient export sectors. The problem
tility impedes growth and increases poverty. It is not only that job destruction comes before
increases the risks of investing in the country, the job creation—so that unemployment and
and thus investors demand a risk premium in the poverty result—but that the IMF’s “structural
form of higher-than-normal profits. Not only is adjustment programs” (designed in ways that al-
growth not enhanced but poverty is increased legedly would reassure global investors) make job
through several channels. The high volatility creation almost impossible. For these programs
increases the likelihood of recessions—and the are often accompanied by high interest rates that
poor always bear the brunt of such downturns. are often justified by a single-minded focus on
Even in developed countries, safety nets are weak inflation. Sometimes that concern is deserved;
or nonexistent among the self-employed and often, though, it is carried to an extreme. In the
in the rural sector. But these are the dominant United States, we worry that small increases in
sectors in developing countries. Without ade- the interest rate will discourage investment. The
quate safety nets, the recessions that follow from IMF has pushed for far higher interest rates in
capital-market liberalization lead to impoverish- countries with a far less hospitable investment
ment. In the name of imposing budget discipline environment. The high interest rates mean that
and reassuring investors, the IMF invariably de- new jobs and enterprises are not created. What
mands expenditure reductions, which almost happens is that trade liberalization, rather than
inevitably result in cuts in outlays for safety nets moving workers from low-productivity jobs
that are already threadbare. to high-productivity ones, moves them from
But matters are even worse—for under the low-productivity jobs to unemployment. Rather
doctrines of the “discipline of the capital mar- than enhanced growth, the effect is increased
kets,” if countries try to tax capital, capital flees. poverty. To make matters even worse, the unfair
Thus, the IMF doctrines inevitably lead to an in- trade-liberalization agenda forces poor countries
crease in tax burdens on the poor and the middle to compete with highly subsidized American and
classes. Thus, while IMF bailouts enable the rich European agriculture. . . .
to take their money out of the country at more
favorable terms (at the overvalued exchange
rates), the burden of repaying the loans lies with An Unfair Trade Agenda
the workers who remain behind. The trade-liberalization agenda has been set
The reason that I emphasize capital-market by the North, or more accurately, by special
liberalization is that the case against it—and interests in the North. Consequently, a dis-
against the IMF’s stance in pushing it—is so proportionate part of the gains has accrued to
compelling. It illustrates what can go wrong the advanced industrial countries, and in some
with globalization. Even economists like Jagdish cases the less-developed countries have actually
Bhagwati, strong advocates of free trade, see the been worse off. After the last round of trade
1136 • Part X: The Future of Inequality

negotiations, the Uruguay Round that ended in because the developing countries would have a
1994, the World Bank calculated the gains and comparative advantage in these sectors.
losses to each of the regions of the world. The Consider also intellectual-property rights,
United States and Europe gained enormously. which are important if innovators are to have
But sub-Saharan Africa, the poorest region of incentives to innovate (though many of the
the world, lost by about 2 percent because of corporate advocates of intellectual property
terms-of-trade effects: The trade negotiations exaggerate its importance and fail to note that
opened their markets to manufactured goods much of the most important research, as in basic
produced by the industrialized countries but science and mathematics, is not patentable).
did not open up the markets of Europe and the Intellectual-property rights, such as patents and
United States to the agricultural goods in which trademarks, need to balance the interests of pro-
poor countries often have a comparative advan- ducers with those of users—not only users in de-
tage. Nor did the trade agreements eliminate the veloping countries, but researchers in developed
subsidies to agriculture that make it so hard for countries. If we underprice the profitability of
the developing countries to compete. innovation to the inventor, we deter invention.
The U.S. negotiations with China over its If we overprice its cost to the research commu-
membership in the WTO displayed a double nity and the end user, we retard its diffusion and
standard bordering on the surreal. The U.S. beneficial effects on living standards.
trade representative, the chief negotiator for the In the final stages of the Uruguay negotia-
United States, began by insisting that China was tions, both the White House Office of Science
a developed country. Under WTO rules, devel- and Technology Policy and the Council of Eco-
oping countries are allowed longer transition nomic Advisers worried that we had not got the
periods in which state subsidies and other de- balance right—that the agreement put produc-
partures from the WTO strictures are permitted. ers’ interests over users’. We worried that, with
China certainly wishes it were a developed coun- this imbalance, the rate of progress and innova-
try, with Western-style per capita incomes. And tion might actually be impeded. After all, knowl-
since China has a lot of “capitas,” it’s possible to edge is the most important input into research,
multiply a huge number of people by very small and overly strong intellectual-property rights
average incomes and conclude that the People’s can, in effect, increase the price of this input.
Republic is a big economy. But China is not only We were also concerned about the consequences
a developing economy; it is a low-income de- of denying lifesaving medicines to the poor. This
veloping country. Yet the United States insisted issue subsequently gained international attention
that China be treated like a developed country! in the context of the provision of AIDS medi-
China went along with the fiction; the negotia- cines in South Africa. The international outrage
tions dragged on so long that China got some forced the drug companies to back down—and
extra time to adjust. But the true hypocrisy was it appears that, going forward, the most adverse
shown when U.S. negotiators asked, in effect, for consequences will be circumscribed. But it is
developing-country status for the United States worth noting that initially, even the Democratic
to get extra time to shelter the American textile U.S. administration supported the pharmaceu-
industry. tical companies.
Trade negotiations in the service industries What we were not fully aware of was another
also illustrate the unlevel nature of the playing danger—what has come to be called “biopiracy,”
field. Which service industries did the United which involves international drug companies
States say were very important? Financial ser- patenting traditional medicines. Not only do
vices—industries in which Wall Street has a they seek to make money from “resources” and
comparative advantage. Construction industries knowledge that rightfully belong to the devel-
and maritime services were not on the agenda, oping countries, but in doing so they squelch
 Stiglitz—Globalism’s Discontents • 1137

domestic firms who long provided these tradi- there had to be active programs to help the poor.
tional medicines. While it is not clear whether And when I left the White House to go to the
these patents would hold up in court if they World Bank, I brought with me the same skep-
were effectively challenged, it is clear that the ticism of trickle-down economics; if this had not
less-developed countries may not have the legal worked in the United States, why would it work
and financial resources required to mount such in developing countries? While it is true that sus-
a challenge. The issue has become the source of tained reductions in poverty cannot be attained
enormous emotional, and potentially economic, without robust economic growth, the converse
concern throughout the developing world. This is not true: growth need not benefit all. It is not
fall, while I was in Ecuador visiting a village in true that “a rising tide lifts all boats.” Sometimes,
the high Andes, the Indian mayor railed against a quickly rising tide, especially when accompa-
how globalization had led to biopiracy. . . . nied by a storm, dashes weaker boats against the
shore, smashing them to smithereens.
In spite of the obvious problems confronting
Trickle-Down Economics trickle-down economics, it has a good intellec-
We recognize today that there is a “social con- tual pedigree. One Nobel Prize winner, Arthur
tract” that binds citizens together, and with their Lewis, argued that inequality was good for de-
government. When government policies abro- velopment and economic growth, since the rich
gate that social contract, citizens may not honor save more than the poor, and the key to growth
their “contracts” with each other, or with the was capital accumulation. Another Nobel Prize
government. Maintaining that social contract winner, Simon Kuznets, argued that while in
is particularly important, and difficult, in the the initial stages of development inequality in-
midst of the social upheavals that so frequently creased, later on the trend was reversed.1
accompany the development transformation. In The history of the past fifty years has, how-
the green eye-shaded calculations of the IMF ever, not supported these theories and hypothe-
macroeconomics there is, too often, no room for ses. East Asian countries—South Korea, China,
these concerns. Taiwan, Japan—showed that high savings did
Part of the social contract entails “fairness,” not require high inequality, that one could
that the poor share in the gains of society as it achieve rapid growth without a substantial in-
grows, and that the rich share in the pains of crease in inequality. Because the governments
society in times of crisis. The Washington Con- did not believe that growth would automatically
sensus policies paid little attention to issues of benefit the poor, and because they believed that
distribution or “fairness.” If pressed, many of greater equality would actually enhance growth,
its proponents would argue that the best way governments in the region took active steps to
to help the poor is to make the economy grow. ensure that the rising tide of growth did lift
They believe in trickle-down economics. Even- most boats, that wage inequalities were kept in
tually, it is asserted, the benefits of that growth bounds, that some educational opportunity was
trickle down even to the poor. Trickle-down eco- extended to all. Their policies led to social and
nomics was never much more than just a belief, political stability, which in turn contributed to
an article of faith. Pauperism seemed to grow in an economic environment in which businesses
nineteenth-century England even though the flourished. Tapping new reservoirs of talent pro-
country as a whole prospered. Growth in Amer- vided the energy and human skills that contrib-
ica in the 1980s provided the most recent dra- uted to the dynamism of the region.
matic example: while the economy grew, those Elsewhere, where governments adopted the
at the bottom saw their real incomes decline. Washington Consensus policies, the poor have
The Clinton administration had argued strongly benefited less from growth. In Latin Amer-
against trickle-down economics; it believed that ica, growth has not been accompanied by a
1138 • Part X: The Future of Inequality

reduction in inequality, or even a reduction in policies on other countries that, had they been
poverty. In some cases poverty has actually in- advocated for the United States, would have
creased, as evidenced by the urban slums that been strongly contested within the administra-
dot the landscape. The IMF talks with pride tion, and almost surely defeated. The reason for
about the progress Latin America has made in this seeming inconsistency was simple: The IMF
market reforms over the past decade (though and the World Bank were part of Treasury’s turf,
somewhat more quietly after the collapse of the an arena in which, with few exceptions, they
star student Argentina in 2001, and the recession were allowed to push their perspectives, just as
and stagnation that have afflicted many of the other departments, within their domains, could
“reform” countries during the past five years), push theirs. . . .
but has said less about the numbers in poverty.
Clearly, growth alone does not always im-
prove the lives of all a country’s people. Not Global Social Justice
surprisingly, the phrase “trickle-down” has disap- Today, in much of the developing world, glo-
peared from the policy debate. But, in a slightly balization is being questioned. For instance, in
mutated form, the idea is still alive. I call the new Latin America, after a short burst of growth in
variant trickle-down-plus. It holds that growth is the early 1990s, stagnation and recession have
necessary and almost sufficient for reducing pov- set in. The growth was not sustained—some
erty—implying that the best strategy is simply might say, was not sustainable. Indeed, at this
to focus on growth, while mentioning issues like juncture, the growth record of the so-called
female education and health. But proponents of post-reform era looks no better, and in some
trickle-down-plus failed to implement policies countries much worse, than in the widely crit-
that would effectively address either broader icized import-substitution period of the 1950s
concerns of poverty or even specific issues such and 1960s when Latin countries tried to indus-
as the education of women. In practice, the advo- trialize by discouraging imports. Indeed, reform
cates of trickle-down-plus continued with much critics point out that the burst of growth in the
the same policies as before, with much the same early 1990s was little more than a “catch-up”
adverse effects. The overly stringent “adjustment that did not even make up for the lost decade
policies” in country after country forced cutbacks of the 1980s.
in education and health: in Thailand, as a result, Throughout the region, people are asking:
not only did female prostitution increase but “Has reform failed or has globalization failed?”
expenditures on AIDS were cut way back; and The distinction is perhaps artificial, for global-
what had been one of the world’s most successful ization was at the center of the reforms. Even in
programs in fighting AIDS had a major setback. those countries that have managed to grow, such
The irony was that one of the major propo- as Mexico, the benefits have accrued largely to
nents of trickle-down-plus was the U.S. Treasury the upper 30 percent and have been even more
under the Clinton administration. Within the concentrated in the top 10 percent. Those at
administration, in domestic politics, there was a the bottom have gained little; many are even
wide spectrum of views, from New Democrats, worse off. The reforms have exposed countries
who wanted to see a more limited role for gov- to greater risk, and the risks have been borne dis-
ernment, to Old Democrats, who looked for proportionately by those least able to cope with
more government intervention. But the central them. Just as in many countries where the pacing
view, reflected in the annual Economic Report and sequencing of reforms has resulted in job
of the President (prepared by the Council of destruction outmatching job creation, so too has
Economic Advisers), argued strongly against the exposure to risk outmatched the ability to
trickle-down economics—or even trickle-down- create institutions for coping with risk, including
plus. Here was the U.S. Treasury pushing effective safety nets.
 Firebaugh—The New Geography of Global Income Inequality • 1139

In this bleak landscape, there are some posi- countries have been given disproportionate
tive signs. Those in the North have become more weight (in matters whose effects go well beyond
aware of the inequities of the global economic finance), then the prospects for success in the
architecture. The agreement at Doha to hold a current discussions of reform, in which the same
new round of trade negotiations—the “Devel- parties continue to predominate, are bleak. They
opment Round”—promises to rectify some of are more likely to result in slight changes in the
the imbalances of the past. There has been a shape of the table, not changes in who is at the
marked change in the rhetoric of the interna- table or what is on the agenda.
tional economic institutions—at least they talk September 11 has resulted in a global alliance
about poverty. At the World Bank, there have against terrorism. What we now need is not just
been some real reforms; there has been some an alliance against evil, but an alliance for some-
progress in translating the rhetoric into reality— thing positive—a global alliance for reducing
in ensuring that the voices of the poor are heard poverty and for creating a better environment,
and the concerns of the developing countries an alliance for creating a global society with
are listened to. But elsewhere, there is often a more social justice.
gap between the rhetoric and the reality. Serious
reforms in governance, in who makes decisions NOTES
and how they are made, are not on the table. If 1. See W. A. Lewis, “Economic Development
with Unlimited Supplies of Labor,” Manchester
one of the problems at the IMF has been that the
School 22 (1954), pp. 139–91, and S. Kuznets, “Eco-
ideology, interests, and perspectives of the finan-
nomic Growth and Income Inequality,” American
cial community in the advanced industrialized Economic Review 45(1) (1955), pp. 1–28.

132 • Glenn Firebaugh

The New Geography of Global Income Inequality

The new geography of inequality—not to be stages of Western industrialization. Put in the

confused with the “new economic geography” perspective of an individual, the new geography
that arose in economics in the 1990s (Fujita, of global income inequality means that national
Krugman, and Venables 1999)—refers to the location—while still paramount—is declining
new pattern of global income inequality caused in significance in the determination of one’s
by the recent phenomenon of declining inequal- income.
ity across nations accompanied by (in many Despite a recent surge of interest in global
places) rising inequality within nations. This inequality, researchers have largely overlooked
phenomenon, which began in the last third of its changing contour. Studies of global income
the twentieth century and continues today, re- inequality over the last decades of the twentieth
sults in a “new geography” because it represents century have been preoccupied with the prob-
the reversal of trends that trace back to the early lem of global divergence, that is, the presumed

Glenn Firebaugh. The New Geography of Global Income Inequality, pp. 15–30, 228. Copyright © 2003 by the
President and Fellows of Harvard College. Reprinted by permission of Harvard University Press.
1140 • Part X: The Future of Inequality

problem of worsening income inequality for Until last week, not so many Americans had
the world as a whole (for example, Wade 2001; even heard of the WTO. Fewer still could
Milanovic 2002). This preoccupation with have identified it as the small, Geneva-based
global divergence is misguided, first, because bureaucracy that the United States and 134
global income inequality almost certainly de- other nations set up five years ago to referee
clined over this period and, second, because the global commerce. To Bill Clinton, it is a
focus on the level of global income inequality mechanism that can allow America to do well
has diverted attention from the changing nature and good at the same time. But to many of
of global income inequality in recent decades. the 40,000 activists and union members who
Global income inequality is no worse today than streamed into Seattle—a clean, scenic city
it was in the 1960s and 1970s, but global in- that has grown rich on foreign trade—the
come inequality is nevertheless changing—it is WTO is something else again: a secretive tool
gradually shifting from inequality across nations of ruthless multinational corporations. They
to inequality within nations. The rising impor- charge it with helping sneaker companies to
tance of within-nation inequality and declining exploit Asian workers, timber companies to
importance of between-nation inequality rep- clear-cut rain forests, shrimpers to kill sea tur-
resents a historic change, since it involves the tles and a world of other offenses.
reversal of a trend that began with the uneven
industrialization of the world that started more Media accounts grappled with the sheer di-
than two centuries ago. versity of the protesters, from leaders of U.S.
This chapter contrasts the New Geogra- labor to members of environmental groups to a
phy Hypothesis with the popular view that leading Chinese dissident. The common thread
globalization—by which I mean the increased seemed to be, as the New York Times (1999) put
interconnectedness of localities, particularly it, the view that the WTO is a “handmaiden of
the deepening of economic links between corporate interests whose rulings undermine
countries—has led to growing global income health, labor and environmental protections
inequality: around the world.” According to The Economist
(1999), “The WTO has become a magnet for
Globalization ĺ global inequality. resistance to globalisation by both old-fashioned
protectionists and newer critics of free trade.”
For short, I call this popular view the Trade Some of the protest groups emphasized
Protest Model, because the protests against rising inequality as among the most noxious
the World Trade Organization in Seattle and consequences of increasing trade globalization.
elsewhere were driven at least in part by the For example, Ralph Nader’s Public Citizen
assumption that global trade is exacerbating group portrays the WTO as a tool of big busi-
global inequality. To place the New Geography ness “which is harming the environment and
Hypothesis in context, it is useful first to exam- increasing inequality” (The Economist 1999),
ine five myths that underlie the globalization ĺ and representatives of 1,448 nongovernmen-
global inequality model. tal organizations protesting the WTO signed
a statement claiming that “globalisation has
three serious consequences: the concentration of
Myths of the Trade Protest Model wealth in the hands of the multinationals and
Under the heading “Siege in Seattle,” the De- the rich; poverty for the majority of the world’s
cember 13, 1999, issue of the U.S. magazine population; and unsustainable patterns of pro-
Newsweek gave this account of the protests sur- duction and consumption that destroy the en-
rounding the meeting of the World Trade Orga- vironment” (New Scientist 1999). Note that two
nization in Seattle, Washington: of the three consequences—concentration of
 Firebaugh—The New Geography of Global Income Inequality • 1141

wealth and impoverishment of the majority—tie intra-country income inequality, and sub-
globalization directly to growth in inequality. stantial growth of poverty and misery in
Global income inequality is the result of the the world at large” (Manuel Castells, 1998,
interplay of multiple causes, of course, so serious p. 82, emphasis omitted).
analyses are unlikely to give an unqualified en-
dorsement to the notion that globalization has What I have found (Firebaugh 2003) is that
automatically resulted in an explosion in global global income inequality has not exploded but
income inequality. Global inequality existed in fact leveled off and then declined in the last
before the recent growth in world trade, and it part of the twentieth century. Although income
would persist if nations suddenly stopped trading. inequality rose somewhat in the average nation,
Nonetheless, popular literature on globalization income inequality declined across nations. Since
has tended to fuel the belief in a globalization-led between-nation inequality is the larger compo-
explosion in global income inequality by making nent of global income inequality, the decline in
claims that purport to be grounded in the find- between-nation income inequality more than
ings of serious scholarly analyses. Upon closer offset the rise in within-nation income inequal-
inspection, however, many of the claims fly in ity. As a result, global income inequality declined
the face of available empirical evidence. This sec- in the last years of the twentieth century. Sher-
tion examines the key myths that underlie the lock Holmes was right: it is a capital mistake to
globalization ĺ global inequality model. theorize in advance of the facts (Doyle 1955, p.
507). With respect to global income inequal-
Myth 1. The Myth of Exploding Global Income ity, much mischief has been done by theorizing
Inequality. A steady drumbeat of reports and about global income inequality on the basis of
articles claims that the world’s income is becom- the views expressed above. Theorizing based on
ing more and more unequally distributed. Here the widespread view of exploding global income
is a sample: inequality is theorizing based on facts that aren’t.
• “Globalization has dramatically increased
inequality between and within nations” Myth 2. The Myth of Growing Income Inequal-
(Jay Mazur, 2000, in Foreign Affairs). ity Across Nations, as Rich Nations Surge Ahead
• “The very nature of globalization has an and Poor Nations Fall Further Behind. The first
inherent bias toward inequality. . . . One myth—exploding global inequality—is based
would have to be blind not to see that on a second myth, the myth that inequality is
globalization also exacerbates the disparity growing across nations. The second myth is as
between a small class of winners and the widespread as the first, and it has been fueled
rest of us” (Paul Martin, Canada’s prime by widely circulated reports of international
minister, June 1998, quoted in Eggertson agencies:
1998). • “Figures indicate that income inequal-
• “Along with ecological risk, to which it is ity between countries has increased
related, expanding inequality is the most sharply over the past 40 years” (World
serious problem facing world society” (An- Bank 2000b, World Development Report
thony Giddens, 1999). 2000/2001, p. 51).
• “Thus, overall, the ascent of informational, • “The average income in the richest 20
global capitalism is indeed characterized countries is 37 times the average in the
by simultaneous economic development poorest 20—a gap that has doubled in the
and underdevelopment, social inclusion past 40 years” (International Monetary
and social exclusion. . . . There is polar- Fund 2000, p. 50).
ization in the distribution of wealth at • “Gaps in income between the poorest
the global level, differential evolution of and richest people and countries have
1142 • Part X: The Future of Inequality

continued to widen. In 1960 the 20% of I have debunked myth number 2 first by rep-
the world’s people in the richest countries licating the United Nations/World Bank results
had 30 times the income of the poorest and then by demonstrating how they have been
20%—in 1997, 74 times as much. . . . misinterpreted (Firebaugh 2003). In addition
Gaps are widening both between and to the weighting problem just mentioned, the
within countries” (United Nations Devel- claims of growing inequality often ignore na-
opment Program 1999, Human Develop- tions in the middle of the income distribution,
ment Report 1999, p. 36). focusing instead on selected nations at the tails.
• “It is an empirical fact that the income When the entire income distribution is used,
gap between poor and rich countries has and when individuals are given equal weight, we
increased in recent decades” (Ben-David, find that—far from growing—income inequal-
Nordström, and Winters 1999, World ity across nations declined in the late twentieth
Trade Organization special study, p. 3). century.
• “In 1960, the Northern countries were 20
times richer than the Southern, in 1980 46 Myth 3. The Myth That Globalization Historically
times. . . . [I]n this kind of race, the rich Has Caused Rising Inequality Across Nations. Con-
countries will always move faster than the trary to this myth, the trend in between-nation
rest” (Sachs 1992, The Development Dictio- inequality historically has not followed changes
nary, p. 3). in the trend in world economic integration.
First, although it is true that between-nation
The myth of growing income inequality income inequality increased dramatically over
across nations is based in large part on a mis- the nineteenth and early twentieth centuries,
interpretation of the widely cited finding (for Peter Lindert and Jeffrey Williamson (2000)
example, World Bank 2000b, p. 50) that income argue that the period of rising inequality across
growth has tended to be slower in poor nations nations began before the period of true globaliza-
than in rich nations. As we shall see, this pos- tion started, so globalization apparently did not
itive cross-country association between income cause the upturn. Second, the sharp decline in
level and income growth rate conceals the critical globalization between World War I and World
fact that the poor nations that are falling badly War II did not result in declining inequality
behind contain no more than 10 percent of the across nations (to the contrary, between-nation
world’s population, whereas the poor nations income inequality shot up rapidly over the pe-
that are catching up (largely in Asia) contain riod). Finally, income inequality across nations
over 40 percent of the world’s population. When has declined in recent decades, during a period
nations are weighted by population size—as when globalization has presumably reached new
they must be if we want to use between-nation heights. (I say “presumably” because globaliza-
inequality to draw conclusions about global in- tion is itself a contentious issue: see Guillén
come inequality—we find that income inequal- 2001. Nonetheless virtually all agree that the
ity across nations peaked sometime around 1970 world has become more economically integrated
and has been declining since. This peaking of over recent decades, even if the degree of glo-
between-nation income inequality circa 1970 balization is often overstated, as Chase-Dunn,
is particularly interesting in light of Manuel Kawano, and Brewer 2000, among others, have
Castells’s (1998, p. 336) well-known claim that noted.) In short, the rise in global inequality pre-
a “new world” originated in the late 1960s to dates the rise in globalization, global inequality
mid-1970s. Ironically, though, Castells charac- has risen while globalization was declining, and
terizes the world born in this period as a world currently global inequality is declining while glo-
of sharply increasing global inequality, and many balization is rising. It is hard then to make the
other globalization writers make the same error. case historically that globalization is the cause of
 Firebaugh—The New Geography of Global Income Inequality • 1143

rising income inequality across nations (O’Ro- and understood, the effect of globalization on
urke 2001). global income inequality at any point in time is
an open question to be settled empirically.
Myth 4. The Myth of a Postindustrial World Econ- But if international exchange is inherently
omy. In reading the globalization literature it is exploitative, as some theories of world stratifi-
easy to lose sight of the fact that, until recently, cation insist, then rising exchange implies rising
most of the world’s people were engaged in exploitation, and the Trade Protest Model is true
agriculture. So the world’s workforce is barely virtually by definition. The Trade Protest Model
postagricultural, much less postindustrial. This then becomes:
chapter makes the point that the primary engine
still driving the growth in world production is Globalization ĺ more exploitation of poor
more manufacturing. A new information age nations by rich nations ĺ greater global
might be on the way, but it is not here yet—at inequality
least it is not here for most of the world’s peo-
ple. It is important to look ahead, of course, Note that this elaboration of the Trade Protest
and it is hard to argue against the view that the Model reveals how high the theoretical stakes are
world will eventually be postindustrial. The with regard to empirical tests of the globaliza-
death of industrialization is nonetheless much tion ĺ global income inequality model, since
exaggerated, as is the view that we are rapidly the failure of globalization to lead to rising global
approaching an information-based global econ- income inequality would undermine exploita-
omy (Quah 1997). Estimates of the composition tion theories (for example, dependency theories)
of global output, albeit rough approximations, as well as undermining the Trade Protest Model.
rule out the claims of some globalization writers I have presented evidence that increasing
that we live in a new economic era quite unlike international exchange over recent decades has
the era of the last generation. Industrialization been accompanied by declining—not rising—
was important in the nineteenth century, it was income inequality across nations (Firebaugh
important in the twentieth century, and it re- 2003). Other studies also document declining
mains important in most regions of the world in between-nation income inequality over recent
the twenty-first century. A preoccupation with decades. The decline in between-nation inequal-
postindustrialization in the face of the continu- ity has significant theoretical implications. The
ing diffusion of industrialization results in an assumption of inherent exploitation favoring
incomplete and distorted story of global income rich nations in international exchange is the
inequality that deemphasizes the critical role of linchpin of some theoretical schools. But if inter-
the continuing spread of industrialization to all national exchange were inherently exploitative,
regions of the world. Computers are important, we would not expect to observe declining in-
but they are not all-important. In accounting equality across nations during a period of rising
for recent trends in global income inequality, international trade. Yet the assumption persists,
the bigger story is industrial growth in Asia, not suggesting that in some theories the notion of
technological growth in the West. inherent exploitation is so essential that it enjoys
creedal status as a doctrine to be believed rather
Myth 5. The Myth of International Exchange as than as a hypothesis to be tested.
Inherently Exploitative. Globalization involves
increased exchange over national boundaries.
One might posit that increased exchange wors- Causes of the Reversal: An Overview
ens global inequality under some historical con- I argue that the world’s spreading industrializa-
ditions and reduces it under other conditions. tion and growing economic integration in the
Until those historical conditions are identified late twentieth century and the early twenty-first
1144 • Part X: The Future of Inequality

have reversed the historical pattern of uneven in the future as well. The fourth is the collapse of
economic growth favoring richer nations. The communism, which also boosted within-nation
conventional view, just elaborated, is that glo- inequality. This is a nonrecurring event, how-
balization has exacerbated global income in- ever, so its effect on within-nation inequality is
equality. The evidence challenges that view. In limited to a specific point in history, the 1990s.
reality globalization has offsetting effects—by In short, the decline in between-nation in-
spurring industrialization in poor nations, glo- come inequality that began in the late twentieth
balization raises inequality within many nations century was caused by deepening industrializa-
and compresses inequality across nations. The tion of poor nations, by growing economic in-
net effect has been a reduction in global income tegration that dissolves institutional differences
inequality in recent decades, since the reduction between nations, by technological change that
in between-nation income inequality has more reduces the effects of labor immobility across
than offset the growth in within-nation income national boundaries, and by a demographic
inequality. windfall that has benefited some poor nations
The new pattern of rising within-nation and and promises to benefit others in the future.
falling between-nation income inequality has The growth in within-nation income inequality
multiple causes. The most important cause is was caused by the deepening industrialization of
spreading industrialization—the diffusion of in- poor nations, by the growth of the service sector,
dustrialization to the world’s large poor nations. and by the collapse of communism.
The diffusion of industrialization to poor regions
compresses inequality across nations and boosts
inequality within them. The effect of spreading The Inequality Transition
industrialization on between-nation inequality is The industrialization of richer nations in the
reinforced by the effect of the growing integra- nineteenth century and first half of the twenti-
tion of national economies. Growing economic eth caused income inequality across nations to
integration tends to dissolve institutional differ- explode. As a result, global income inequality
ences between nations. The convergence of insti- shifted from inequality within nations to in-
tutional economic goals and policies compresses equality across nations. Now, however, poorer
inequality across nations by (in some instances at nations are industrializing faster than richer
least) removing impediments to growth in poor ones are, and between-nation inequality is de-
nations. clining while within-nation inequality appears
There are at least four other significant causes to be rising. If this turnaround continues, future
of the new geography of global inequality. The historians will refer to an inequality transition
first is technological change that reduces the tyr- that accompanied world industrialization. That
anny of space in general and more particularly transition is from within-nation inequality to
reduces the effect of labor immobility across between-nation inequality back to within-nation
national boundaries. This technological change inequality, with the late twentieth century as the
works to reduce inequality across nations. The period when the shift back to within-nation in-
second is a demographic windfall that has bene- come inequality began.
fited some poor Asian nations in recent decades
and promises to benefit other poor nations
in the near future. This effect also operates to Phase 1 of the Transition: From
compress global income inequality, by reducing Within- to Between-Nation Inequality
between-nation inequality. The third is the rise Phase 1 of the inequality transition coincides
of the service sector, especially in richer nations. with the period of Western industrialization that
Growth in this sector has boosted income in- began in the late eighteenth century and ended
equality within nations, and it is likely to do so in the second half of the twentieth century. The
 Firebaugh—The New Geography of Global Income Inequality • 1145

first phase of the inequality transition was char- From these 660 data points (33 nation groups
acterized by unprecedented growth in income x 20) it is a straightforward matter to apply the
inequality across nations. As Lant Pritchett population-weighted formulas for the Theil
(1996, p. 40) puts it, “the overwhelming fea- index and the mean logarithmic deviation
ture of modern economic history is a massive (MLD)—two measures of inequality—to calcu-
divergence in per capita incomes between rich late summary measures of the world’s total in-
and poor countries.” The evidence is incontro- equality for different years. By collapsing the 199
vertible. First, it is clear that current levels of nations into 33 groups, Bourguignon and Mor-
between-nation income inequality would not risson are able to extend their inequality series
have been possible earlier in human history. back to 1820. Note that, to the extent that their
Again quoting Pritchett (1997, pp. 9–10): “If grouping strategy introduces bias, the bias is in
there had been no divergence, then we could ex- the direction of underestimating between-nation
trapolate backward from present income of the inequality and inflating within-nation inequality,
poorer countries to past income assuming they since some of the inequality within the nation
grew at least as fast as the United States. How- groups is actually between-nation inequality. But
ever, this would imply that many poor countries that bias should not affect our basic conclusions
must have had incomes below $100 in 1870 [in about the relative growth in between-nation and
1985 U.S. dollars]. Since this cannot be true, within-nation income inequality over the past
there must have been divergence.” two centuries.
Second, Pritchett’s conclusion that “there The results are striking (Figure 132.1). Two
must have been divergence” is supported by facts stand out. First, the B/W ratio—the ratio
estimates of between-nation income inequal- of between-nation to within-nation income in-
ity in the nineteenth century. Consider the re- equality—is much higher now than it was in
cent estimates of Bourguignon and Morrisson the early stages of Western industrialization.
(1999). Bourguignon and Morrisson use the The increase in the B/W ratio reflects both a
Maddison (1995) data to estimate changes in rise in between-nation income inequality and
the level of between-nation income inequality a decline in within-nation income inequality
from 1820 to 1992. Because their objective is since 1820 (Table 132.1). By far the greater
to estimate total world income inequality—not change is in between-nation income inequality,
just between-nation income inequality—Bour- however. The Theil index for between-nation
guignon and Morrisson begin by disaggregating income inequality (actually, inequality between
national income data into vintiles (5 percent nation groups) shot up from 0.061 in 1820 to
groups, that is, twenty income groups per na- 0.513 in 1992, and the MLD shot up from
tion). National boundaries have changed over 0.053 in 1820 to 0.495 in 1992. As anticipated,
the past two centuries, of course, and nations then, the Industrial Revolution of the past two
have come and gone over the past two centu- centuries has increased income inequality across
ries. Even in nations where boundaries remained nations, but the magnitude of the increase is
constant, we do not always have income data for stunning. There has been a metamorphosis
the entire period. To overcome these problems, from a world where poverty was the norm in all
Bourguignon and Morrisson grouped the 199 nations to a richer world with much lower pov-
nations with income data in 1992 into 33 ho- erty rates (Bourguignon and Morrisson 1999,
mogeneous groups, each of which represented Table 1) but also with much greater income in-
at least 1 percent of the world population or equality across nations. Because the steep rise in
world GDP in 1950. The 33 groups include between-nation income inequality was not ac-
single nations (such as China and the United companied by an increase in inequality within
States) as well as large groups of small nations nations, where you live—your nation—is much
and small groups of medium-sized nations. more important in determining your income in
1146 • Part X: The Future of Inequality

Figure 132.1. Ratio of Between-Nation to Within-Nation Income Inequality for

Thirty-Three Nation Groups, 1820–1992
1.8

1.6 Theil
MLD
1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0
1820 1850 1870 1890 1910 1929 1950 1970 1992
Based on Table 132.1. Theil and MLD are measures of inequality.

Table 132.1. Trends in Income Inequality Between and Within Thirty-Three Homogeneous Nation
Groups, 1820–1992

Between-nation groups Within-nation groups

Year Theil MLD Theil MLD
1820 .061 .053 .472 .388
1850 .128 .111 .477 .393
1870 .188 .162 .485 .399
1890 .251 .217 .498 .408
1910 .299 .269 .500 .413
1929 .365 .335 .413 .372
1950 .482 .472 .323 .309
1970 .490 .515 .324 .330
1992 .513 .495 .351 .362

Source: Bourguignon and Morrisson (1999), table 3.

Estimates: Between-nation groups: From Bourguignon and Morrisson (1999), based on Maddison (1995) data set.
Within-nation groups: From Bourguignon and Morrisson (1999), based on updating of Berry, Bourguignon, and
Morrisson (1983a,b) for the post–World War II period and on various sources (for example, Lindert 1999; Morrisson
1999) for the pre–World War II period. Bourguignon and Morrisson (2002) published modestly revised estimates for
inequality. By collapsing vintiles to deciles, they report slightly lower estimates of inequality within the nation groups,
but the fundamental conclusions are the same.
Note: Income measures are adjusted for purchasing power parity, and inequality measures are based on income vintiles
(see Bourguignon and Morrisson 1999 for elaboration).

today’s world than it was in the preindustrial the nation groups has slowed dramatically since
world. 1950. In the four decades after 1950, income
The second fact that stands out is that the inequality across the nation groups increased by
growth in the B/W ratio stalled in the second 6 percent using the Theil index and by 5 percent
half of the twentieth century. The B/W ratio using the MLD. Over the four decades prior to
stopped growing in the second half of the twen- 1950, the Theil had grown about 60 percent
tieth century because growth in inequality across and the MLD had grown about 75 percent.
 Firebaugh—The New Geography of Global Income Inequality • 1147

Apparently the most dramatic effects of Western sufficient theory and evidence that we can plau-
industrialization on between-nation inequality sibly forecast that the B/W ratio will continue to
are over. After more than a century of sharp di- decline in the twenty-first century.
vergence in national incomes, the trend has been The prediction of a declining B/W ratio is
much more stable in recent years. This finding is based on two separate conjectures. The first con-
in line with the findings of others (for example, jecture is that between-nation income inequality
Schultz 1998; Firebaugh 1999; Melchior, Telle, will decline, and the second conjecture is that
and Wiig 2000; Goesling 2001) who, drawing within-nation income inequality will rise, or at
on data for individual nations instead of nation least will not decline. These conjectures are based
groups, find that between-nation income in- on the causes of the current trends, which I ex-
equality is no longer rising. As we shall see later, pect to continue and in some cases to intensify.
between-nation income inequality declined in (I am assuming that there will be no cataclysmic
the late twentieth century when income data for upheaval in the twenty-first century, such as a
the 1990s are added to data for the 1970s and global war or a worldwide plague.) Recall the
1980s. causes listed earlier for the inequality turnaround
Finally, it should be emphasized that the re- in the late twentieth century. I expect the major
sults here are so strong that the historical story causes to continue, so between-nation income
they tell of increasing inequality across nations inequality will decline because of the continued
and of a rising B/W ratio cannot be dismissed industrialization of poor nations, because of the
as due to error in the data. To be sure, income continued convergence of national economic
estimates for the nineteenth century are gross policies and institutions arising from growing
approximations for many nations. But even if economic integration of national economies,
we make the extreme assumption that incomes because of the declining significance of labor im-
are so drastically overstated for poorer nations mobility across national borders, and because of
in 1820 (or so drastically understated for richer a demographic windfall for many poor nations.
nations in 1820) that the Theil and the MLD Within-nation income inequality will rise—or
estimates understate between-nation income in- at least not decline—because of the continued
equality in 1820 by a factor of three, that would industrialization of poor nations and because of
still mean that between-nation income inequal- continued growth in the service sector. Because
ity tripled from 1820 to 1992 (from 0.18 to 0.51 between-nation inequality is the larger compo-
based on the Theil index and from 0.16 to 0.50 nent, global income inequality will decline.
for the MLD), and the B/W ratio still would have The conjecture of declining global income
more than tripled for both inequality measures. inequality is out of step with much of the
globalization literature. A recurring theme in
that literature is that we have entered a new
Phase 2 of the Transition: information-based economic era where pro-
From Between-Nation Back to ductive activity is becoming less dependent on
Within-Nation Inequality physical space, as a rising share of the world’s
The second phase of the inequality transition economic output is produced in electronic space
began in the second half of the twentieth cen- that knows no national borders (Sassen 2000).
tury, with the stabilization of between-nation This phenomenon is possible because of the
income inequality in the 1960–1990 period and emergence of a global economy where income—
the decline in between-nation inequality begin- and hence income inequality—is becoming
ning in earnest in the 1990s. Social scientists increasingly rooted in knowledge rather than
hardly have a stellar track record for predictions, in capital goods (Reich 1991). What do these
especially with regard to sweeping predictions developments imply for global income inequal-
such as the one made here. Nonetheless there is ity? For many globalization writers, the answer
1148 • Part X: The Future of Inequality

Figure 132.2. Industrialization and Between-Nation Income Inequality: Historically

and in the Late Twentieth Century

growth in
Historically: spread of industrialization
industrialization of richer nations income equality
across nations

reduction in
Late Twentieth spread of industrialization income inequality
Century: industrialization of poorer nations across nations

is clear: global inequality is bound to worsen in today’s world is industrialization for the many,
because of the growing “global digital divide” not digitization for the few. Historically the
that enlarges the gap between the “haves” and spread of industrialization has been the primary
the “have-nots” (Campbell 2001; Ishaq 2001; force driving the growth in between-nation in-
Norris 2001). come inequality. The initially richer nations of
The theoretical argument that a shift to the West were the first to industrialize, and the
a knowledge-based global economy would poorer nations of Asia and Africa lagged behind.
worsen global income inequality is shaky. It is The new geography of inequality is also driven
not hard to think of reasons why the shift from by the spread of industrialization, but the effects
an industrial-based to a knowledge-based global are different today: now the spread of industri-
economy would reduce, not increase, inequality alization means the diffusion of manufacturing
across nations. Knowledge is mobile, especially technology to the world’s largest poor regions.
with today’s telecommunication technologies In recent decades inequality has declined across
that permit virtually instant worldwide codi- nations as industrialization has been an engine
fication and distribution of knowledge. In ad- of growth in the most populous poor regions
dition, because knowledge can be given away of the world, especially East Asia. That growth
without being lost, the notion of property rights has worked both to compress inequality across
is more problematic in the case of knowledge, nations (Figure 132.2) and to boost inequality
so it is harder to concentrate and monopolize within the industrializing nations.
knowledge across nations than it is to concen- The significance of this continuation of
trate and monopolize capital goods across na- world industrialization has been lost in much of
tions. Hence the switch to a knowledge-based the literature on globalization, because of pre-
global economy should mean that one’s income occupation with the idea that we are witnessing
is increasingly determined by how much knowl- the emergence of a new knowledge-based tech-
edge one obtains and uses as opposed to where nology regime. To be sure, in the categories used
one lives. The tighter link between knowledge to classify world production, the output of the
and income in turn implies declining income so-called service sector is estimated to exceed the
inequality across nations and rising inequality output of the industrial sector for the world as a
within nations since—absent institutional bar- whole (World Bank 1997, table 12). Yet much of
riers . . . —the variance in individuals’ ability the service sector—an amorphous sector that in-
to obtain and benefit from knowledge is greater cludes wholesale and retail trade, the banking in-
within nations than between them. dustry, government, the transportation industry,
The issue of how the new information age the commercial real estate industry, and personal
will affect global inequality in the near term is services (including health care and education)—
not as decisive as often imagined, since as already has arisen to grease the wheels of industry. Aside
noted, the coming of the information age is often from the growth in personal services and govern-
much exaggerated. What is still most important ment, much of the growth in the service sector
 Firebaugh—The New Geography of Global Income Inequality • 1149

has been for services for producers, not con- not offset the effect of rising within-nation in-
sumers—for example, the rise of an engineer- equality in Asia, given the relative sizes of the
ing industry to design better machines, and the two regions. Few expect within-nation income
growth of a banking industry and a commercial inequality to decline sharply in Latin America in
real estate industry for commercial transactions. the near future, even though Latin America cur-
In addition, many of the other so-called service rently exhibits the highest level of within-nation
industries—the transportation industry that dis- income inequality of the world’s major regions.
tributes manufactured goods, for example, and Indeed, because of the advantage enjoyed by
the specialized retailing industry that sells the North American farmers with respect to some
goods—benefit producers as well as consumers. types of produce, economic integration in the
In short, a significant portion of the growth in Americas could exacerbate inequality by remov-
service industries over the past century can be ing protections for farmers in Latin America. If
seen as ancillary to the industrialization process. that reasoning is correct, there is merit in the
With regard to income inequality within concern of WTO protesters that globalization
nations, we expect the continuing spread of may exacerbate income inequality within some
world industrialization to boost inequality along poor nations by driving down incomes in the
the lines of the classic argument, from Simon lower-income agricultural sector. But the bigger
Kuznets (1955), that industrialization boosts in- story is that industrialization (not the collapse
equality (at least initially) as workers move from of farm prices) will tend to drive up inequality
the lower-wage but larger agricultural sector to within nations, at least initially. Because many
the higher-wage industrial sector. Importantly, nations are still on the part of the Kuznets
this argument suggests that income inequality (1955) curve where migration from farm to fac-
increases in poor nations as they industrialize tory boosts inequality, we can anticipate further
because of income gains, not income losses. In growth in within-nation income inequality in
other words, inequality grows not because some the early decades of the twenty-first century.
people are becoming poorer but because some
people are becoming richer, so the growth in REFERENCES
income inequality in poor nations as they in- Ben-David, Dan, Håkan Nordström, and L. Alan
Winters. 1999. Trade, Income Inequality, and Pov-
dustrialize reflects rising rather than declining
erty. WTO Special Study no. 5. Geneva: World
fortunes. If the industrialization of large poor
Trade Organization.
nations does boost income inequality in those Berry, Albert, François Bourguignon, and Christian
nations, at this juncture in history the continued Morrisson. 1983a. “Changes in the world dis-
spread of industrialization implies growth in in- tribution of income between 1950 and 1977.”
come inequality in the average nation, because Economic Journal 93:331–350.
those nations are home to such a large fraction ———. 1983b. “The level of world inequality: How
of the world’s population. And there are no obvi- much can one say?” Review of Income and Wealth
ous counterforces on the horizon. Although the 29:217–241.
jump in income inequality following the collapse Bourguignon, François, and Christian Morrisson.
1998. “Inequality and development: The role
of communism in Eastern Europe and the for-
of dualism.” Journal of Development Economics
mer USSR is not likely to be repeated, there is no 57:233–257.
good reason to expect income inequality in those ———. 1999. “The size distribution of income
nations to fall in the near future. Nor is there among world citizens: 1820–1990.” Draft. June.
any good reason to expect income inequality to ———. 2002. “Inequality among world citi-
decline notably in the West in the near future, zens: 1820–1992.” American Economic Review
either. We lack reliable inequality data for many 92:727–744.
African nations, but even if inequality were Campbell, Duncan. 2001. “Can the digital divide
falling rapidly (which is unlikely), the effect of be contained?” International Labour Review
140:119–141.
falling within-nation inequality in Africa would
1150 • Part X: The Future of Inequality

Castells, Manuel. 1998. End of Millennium. Malden, Mazur, Jay. 2000. “Labor’s new internationalism.”
Mass.: Blackwell. Foreign Affairs 79:79–93.
Chase-Dunn, Christopher, Yukio Kawano, and Melchior, Arne, Kjetil Telle, and Henrik Wiig. 2000.
Benjamin D. Brewer. 2000. “Trade globaliza- Globalization and Inequality: World Income Distri-
tion since 1795: Waves of integration in the bution and Living Standards, 1960–1998. Studies
world-system.” Special millennium issue, edited on Foreign Policy Issues. Oslo: Royal Norwegian
by Glenn Firebaugh, of the American Sociological Ministry of Foreign Affairs.
Review 65:77–95. Milanovic, Branko. 2002. “True world income dis-
Doyle, Arthur Conan. 1955. A Treasury of Sherlock tribution, 1988 and 1993: First calculation based
Holmes. Selected and with an introduction by on household surveys alone.” Economic Journal
Adrian Conan Doyle. Garden City, N.Y.: Ha- 112:51–92.
nover House. Morrisson, Christian. 1999. “Historical perspectives
The Economist. 1999. “The battle in Seattle.” U.S. on income distribution: The case of Europe.” In
edition. November 27. Handbook of Income Distribution, ed. A. B. Atkin-
Eggertson, Laura. 1998. “Rich-poor gap next issue son and F. Bourguignon. Amsterdam: Elsevier.
for Martin.” Toronto Star, June 3, p. A6. New Scientist. 1999. Editorial. December 4.
Firebaugh, Glenn. 1999. “Empirics of world in- New York Times. 1999. “National guard is called to
come inequality.” American Journal of Sociology quell trade talk protests.” December 1.
104:1597–1630. Norris, Pippa. 2001. Digital Divide: Civic Engage-
———. 2003. The New Geography of Global Income ment, Information Poverty, and the Internet World-
Inequality. Cambridge: Harvard University Press. wide. Cambridge and New York: Cambridge
Fujita, Masahisa, Paul Krugman, and Anthony J. University Press.
Venables. 1999. The Spatial Economy: Cities, Re- O’Rourke, Kevin H. 2001. “Globalization and in-
gions, and International Trade. Cambridge: MIT equality: Historical trends.” National Bureau of
Press. Economic Research. Draft. April.
Giddens, Anthony. 1999. “Globalization: An irre- Pritchett, Lant. 1996. “Forget convergence: Diver-
sistible force.” The Daily Yomiuri, June 7, p. 8. gence past, present, and future.” Finance and De-
Goesling, Brian. 2001. “Changing income inequali- velopment (June):40–43.
ties within and between nations: New evidence.” ———. 1997. “Divergence, big time.” Journal of
American Sociological Review 66:745–761. Economic Perspectives 11:3–17.
Guillén, Mauro F. 2001. “Is globalization civilizing, Quah, Danny T. 1997. “Increasingly weightless
destructive, or feeble? A critique of five key de- economies.” Bank of England Quarterly Bulletin
bates in the social science literature.” Annual Re- (February):49–55.
view of Sociology 27:235–260. Reich, Robert B. 1991. The Work of Nations: Prepar-
International Monetary Fund (IMF). 2000. “How ing Ourselves for Twenty-first-Century Capitalism.
we can help the poor.” Finance and Development New York: Knopf.
(December). (http://www.imf.org/external/pubs.) Sachs, Wolfgang, ed. 1992. The Development Dic-
Ishaq, Ashfaq. 2001. “On the global digital divide.” tionary: A Guide to Knowledge as Power. London:
Finance and Development 38:44–47. Zed Books.
Kuznets, Simon. 1955. “Economic growth and in- Sassen, Saskia. 2000. “Territory and territoriality
come inequality.” American Economic Review in the global economy.” International Sociology
45:1–28. 15:372–393.
Lindert, Peter, and Jeffrey G. Williamson. 2000. Schultz, T. Paul. 1998. “Inequality in the distribu-
“Does globalization make the world more un- tion of personal income in the world: How it is
equal?” Paper presented at the “Globalization in changing and why.” Journal of Population Eco-
Historical Perspective” preconference, National nomics 11:307–344.
Bureau of Economic Research, Cambridge, United Nations Development Program. 1999.
Mass. November. Human Development Report 1999. New York:
Maddison, Angus. 1995. Monitoring the World Econ- Oxford University Press.
omy, 1820–1992. Paris: OECD.
 ABOUT THE EDITORS

David B. Grusky is Professor of Sociology at Stanford University, Director of the Stanford Center
on Poverty and Inequality, founder and coeditor of Pathways Magazine, and coeditor of the Stanford
University Press Social Inequality Series. His recent books include Occupy the Future (with Doug
McAdam, Robert Reich, and Debra Satz), The New Gilded Age (with Tamar Kricheli-Katz), The Great
Recession (with Bruce Western and Christopher Wimer), The Inequality Reader, 2nd Edition (with
Szonja Szelényi), The Inequality Puzzle (with Roland Berger, David B. Grusky, Tobias Raffel, Geoffrey
Samuels, and Christopher Wimer), Poverty and Inequality (with Ravi Kanbur), Mobility and Inequality
(with Stephen Morgan and Gary Fields), Occupational Ghettos (with Maria Charles), and The Declining
Significance of Gender? (with Francine D. Blau and Mary C. Brinton).

Katherine R. Weisshaar is a PhD candidate in the sociology department at Stanford University. Prior
to arriving at Stanford, she graduated from Northwestern University. Her research focuses on gender,
families, and income inequality. In particular, she studies how macrolevel economic patterns influence
gendered arrangements in both occupations and families.

1151
 NAMES INDEX

Acland, Henry, 517–523 Bourguignon, François, 1145, Daley, Richard J., 348
Aigner, Dennis J., 932 1146(table) Dannenmaier, W.D., 1001
Akresh, Ilana Redstone, 780–787 Bowles, Samuel, 618 Danziger, Sheldon, 357–364
Alba, Richard D., 721–728, 789 Brand, Jennie E., 587–595 Davidson, Donald, 268
Allport, Gordon, 850 Breen, Richard, 13, 464–479, Davis, James A., 654–655
Alon, Sigal, 569–577 524–535, 551–562 Davis, Kingsley, 4, 28–31, 35,
Andersen, Margaret L., 942–952 Brewster, Karin, 965 38
Anderson, Elijah, 355 Brooks, David, 310–315 Davis, Ron, 682
Aristotle, 268, 277 Brooks-Gunn, Jeanne, 423–430 DeLuca, Stefanie, 395–402
Aronson, Joshua, 752 Brown, Gordon, 368 Denton, Nancy A., 376–385
Atkinson, Anthony B., 59–72 Brown, Joseph, 755–756 Dew-Becker, Ian, 101
Aurelius, Marcus, 86 Broyard, Antole, 688 Di Carlo, Matthew, 480–495
Burckhardt, Jacob, 283 DiMaggio, Paul, 671–680
Baltzell, Digby, 312–313 Burnham, James, 1067 DiNardo, John, 651
Bane, Mary Jo, 517–523 Burt, Ronald S., 659–663 DiPrete, Thomas A., 4, 104–112
Barber B., 240 Burtless, Gary, 358 Domhoff, G. William, 297–302,
Baron, James, 866, 868, 871 Bush, George W., 1035–1036 1107
Baumol, W., 1079–1080, 1083, Du Bois, W.E.B., 780
1085 Cain, Glen G., 932 Duncan, Greg J., 417–423
Bebchuk, Lucian, 99 Carchedi, G., 198 Duncan, Otis Dudley, 241, 248,
Becker, Gary, 496, 659, 865–866, Castells, Manuel, 1142 446, 506–517, 596–597,
868, 930–931, 938, 1076 Chan, Tak Wing, 202–215, 224, 603, 657–658
Begue, Brian, 682 1004–1013 Durkheim, Émile, 197, 217–222
Belkin, Lisa, 807–811 Charles, Maria, 902–911
Bell, Daniel, 1066–1078, 1081, Chavez, Koji, 357–364 Edin, Kathryn, 339–346
1104, 1116, 1119 Cherlin, Andrew B., 339–340 Ehrenberg, Ronald G., 930
Bertrand, Marianne, 747–751 Child, Irving, 730, 738 Ehrenreich, Barbara, 330–338
Bhagwati, Jagdish, 1135 Clark, Kenneth B., 377 Eirich, Gregory M., 4, 104–112
Bielby, Denise, 865–875 Clinton, Bill, 1038 Elwert, Felix, 403–411
Bielby, William T., 865–875 Cohen, David, 517–523 Engels, J.F., 304, 990
Bischoff, Kendra, 1023–1034 Collins, Patricia Hill, 942–952 England, Paula, 919–923,
Black, Dan A., 931 Cook, Philip, 87 955–965
Blair, Tony, 368, 1044 Cook, Tim, 88 Erikson, Robert, 247, 453–464,
Blau, Francine, 446, 929–941 Correll, Shelley J., 831–842 474–478, 552, 555–556
Blau, Peter M., 506–517, Cotter, David, 965–973 Ermisch, John, 501–505
596–597, 603, 657–658 Cox, Oliver, 696 Esping-Andersen, Gøsta, 52–57,
Bonacich, Edna, 696–707, 768 Cumberworth, Erin, 357–364, 1078–1091, 1116, 1119,
Bound, John, 650 1091–1097 1122
Bourdieu, Pierre, 12–13, 203, Evans, Gary W., 423–430
617–618, 982–1003, Dahrendorf, Ralf, 20, 143–149, Eyal, Gil, 316–321, 1104–1109,
1105–1106 151, 186, 195, 252, 1104 1116, 1119

1153
1154 • Names Index

Faludi, Susan, 966 Grant, Kathryn, 424–425 Klebanov, Pamela Kato, 423–430
Featherman, David L., 244–245, Greenman, Emily, 943–952 Kmec, Julie A., 890–901
443–453, 456–457 Greenspan, Alan, 86 Kopczuk, Wojciech, 64
Fernandez, Roberto M., 663–671 Griliches, Zvi, 78 Kornrich, Sabino, 542
Fernandez-Mateo, Isabel, Grusky, David B., 1–25, 44–50, Kuznets, Simon, 1149
663–671 113–122, 222–232,
Ferree, Myra Marx, 965 480–495, 902–911 Lareau, Annette, 542,
Firebaugh, Glenn, 1139–1150 1013–1022
Fischer, Claude S., 4, 39–42, 587 Hacker, Jacob S., 86–87, 90–103 Laumann, E.O, 203–204,
Fitzgerald, F. Scott, 312 Haller, ARchibald O., 596–621 206–207, 214
Florida, Richard, 114, 1128 Hankiss, Elemér, 316, 318–319 Laurison, Daniel, 1037–1045
Frank, Reanne, 780–787 Harris, Marvin, 771 Lee, David S., 651
Frank, Robert, 86–90 Hartmann, Heidi I., 912–913 Leibniz, G.W., 987–988
Frank, Thomas, 1035–1036 Hauser, Robert M., 244–250, Lenin, Vladimir, 304
Frankfurt, Harry, 261 443–453, 456–457 Lewis, Arthur, 1137
Frazier, E. Franklin, 774 Heckman, James J., 412–416 Lewis, Oscar, 378
Freeman, Lance, 777 Hermsen, Joan M., 965–973 Lichter, Daniel T., 788–799
Freeman, Richard, 650 Herriott, Robert E., 599 Lieberson, Stanley, 881
Freese, Jeremy, 1050–1056 Heyns, Barbara, 517–523 Lin, Nan, 657–659
Fried, Jesse, 99 Hochschild, Arlie Russell, Lincoln, Abraham, 41
803–807 Lindert, Peter, 1142
Gambetta, Diego, 627 Hope, Keith, 237–243 Link, Bruce G., 1050–1051,
Ganzeboom, H., 474–475 Hout, Michael, 39–42, 560, 587, 1056
Garip, Filiz, 671–680 1037–1045, 1091–1097 Lipset, S.M., 455, 1038
Gates, William Sr., 96–97 Liu, Yujia, 113–122
Geiger, Theodor, 145–146 Jackson, Michelle, 562–569 London, Howard, 610
Gerstner, Louis J., 88 Jacobs, Jerry A., 876–880 Lu, Bo, 780–787
Gibson, M.A., 736 James, Sherman, 755 Lucas, Samuel R., 39–42
Giddens, Anthony, 183–192, Jankowski, Martín Sánchez, Luijkx, Ruud, 474–475,
292–296, 326, 39–42 551–562
1125–1131 Jäntti, Markus, 501–505 Lutfey, Karen, 1050–1056
Ginsburg, Ruth Bader, 933 Jencks, Christopher, 517–523
Gintis, Herbert, 517–523, 618 Jiménez, Tomás R., 740–746 MacLeod, Jay, 608–621
Giroux, Henry, 617 Johnson, Lyndon B., 359, Magnuson, Katherine, 417–423
Glazer, Nathan, 771 377–378, 518 Mair, P., 1039–1040
Goldin, Claudia, 73–79, Johnson, Ross, 87 Mankiw, Gregory, 93
820–830 Johnston, David Cay, 96 Manning, Robert D., 710–721
Goldthorpe, John H., 13, Jones, G., 456–457 Mao Tse-tung, 322–323
202–215, 224, 237–243, Jonsson, Jan O., 480–495, 552, Marks, Stephen, 866
247, 259, 453–464, 567 Marmot, Michael, 1046–1050
466(table), 474–478, Jonsson, Janne, 475–476 Marshall, Alfred, 254, 257
524–535, 555–556, 560, Marshall, Thurgood, 684
1004–1013 Kahn, Lawrence M., 81, Marx, Karl, 6, 9, 12, 15, 47,
Goldwater, Barry, 312 833–837 127–145, 148, 158, 177,
Goodman, Leo A., 444–447 Kalleberg, Arne L., 632–646 183–192, 192(n24),
Gordon, Robert, 101 Katz, Lawrence F., 73–79 193–201, 251, 259–260,
Gorman, Elizabeth H., 890–901 Kenworthy, Lane, 365 304, 306, 314, 633–634,
Gouldner, Alvin W., 6, 302–310, Kerry, John, 1038 769, 991, 1104
1107, 1116, 1119 Kim, Jinyoung, 1056–1063 Mason, Phlip, 706–707
Gourevitch, Peter, 99 King, Cameron H., Jr., 704 Massey, Douglas S., 376–385
Granovetter, Mark S., 653–657, King, Larry, 319 Matute-Bianchi, M.G., 735–736
661 King, Martin Luther Jr., 247, 725 McCain, John, 1041
 Names Index • 1155

McWilliams, Carey, 706 Percheski, Christine, 811–819 Schumpeter, Joseph, 2, 188,

Mead, Lawrence, 379–380 Persson, Torsten, 41 192(n17), 291–292, 1089
Mehta, Zubin, 824 Peterson, Trond, 858–864, Sen, Amartya K., 260–273
Merton, Robert K., 662 912–919 Sewell, William H., 596–621
Meyer, John W., 1116–1125 Pettit, Becky, 431–440 Sharkey, Patrick, 403–411
Michelson, Stephan, 517–523 Phelan, Jo C., 1050–1051, 1056 Shinn, James, 99, 103(n27)
Miech, Richard, 1056–1063 Phipps, Susie Guillory, 682 Siegrist, Johannes, 1046–1050
Mill, John Stuart, 280 Pierson, Paul, 86–87, 90–103 Simmel, Georg, 662, 850
Millis, H.A., 700 Piketty, Thomas, 59–72, 94–95, Smeeding, Timothy, 501–505
Mills, C. Wright, 144, 282–292, 98 Smith, Adam, 1104
1067 Piore, Michael J., 629–632 Smith, Marshall, 517–523
Mincer, Jacob, 872–873, 929 Pittinsky, Matthew, 4, 104–112 Smith, Robert S., 930
Monea, Emily, 364 Polacheck, Solomon, 929 Solon, Gary, 496–501
Moore, Wilbert E., 4, 28–31, Polanyi, Karl, 1098 Sørensen, Aage B., 251–260,
35, 38 Pollak, Reinhard, 478, 480–495, 632–640
Morgan, Laurie A., 912–919 551–562 Sorokin, P.A., 455–456
Morgan, Stephen L., 621–628 Portes, Alejandro, 596–621, Spencer, Steven, 752
Morrisson, Christian, 1145 710–721, 729–739 Stalin, Joseph, 49
Mosca, Gaetano, 188, 276–282 Pritchett, Lant, 1145 Staniszkis, Jadwiga, 316, 318–319
Moynihan, Daniel Patrick, 379, Putin, Vladimir, 1113 Stark, David, 319
386–387 Steele, Claude, 752–756
Mullainathan, Sendhil, 747–751 Qian, Zhenchao, 788–799 Stiglitz, Joseph E., 1132–1139
Müller, Walter, 478, 551–562 Swidler, Ann, 39–42
Murray, Charles, 379–381, 684 Rawls, John, 261, 268, 1119 Szalai, Erzsébet, 316, 318–319
Myles, John, 52–57 Reagan, Ronald, 95, 350, 609, Szelény, Szonja, 17–25
Myrdal, Gunnar, 377 1039 Szelényi, Iván, 316–321,
Reagn, 97 1104–1109
Nee, Victor, 721–728, 789, Reardon, Sean F., 536–550,
1098–1103, 1116, 1119 1023–1034 Tabellini, Guido, 41
Nelson, Timothy, 339–346 Reed, Joanna Miranda, 339–346 Tam, Tony, 924–928
Neumark, David M., 938 Reich, Charles, 1071 Thumin, F.J., 1001
Newcomb, T.M., 654 Reingold, David A., 664 Thurow, Lester, 633, 637, 881
Newman, Katherine S., 369–375 Reskin, Barbara F., 849–858, Tilcsik, András, 842–848
Nietzsche, Friedrich, 171 881–890 Torche, Florencia, 578–586
Riddell, W. Craig, 97–98 Townsley, Eleanor, 316–321,
Obama, Barack, 1038, 1041, Ridgeway, Cecilia L., 973–980 1104–1109
1044 Robinson, Jackie, 755 Treiman, Donald J., 233–236,
O’Brien, Rourke L., 369–375 Roemer, John, 151–156, 161(n9) 461–462, 474–475,
O’Connor, Sandra Day, 933 Rogers, Richard G., 1056–1063 912–913
Omi, Michael, 682–686 Rokkan, S., 1038 Tullock, Gordon, 258
Ossowski, Stanislaw, 188 Romney, Mitt, 1044 Tumin, Melvin M., 4, 31–38
Roosevelt, Franklin D., 727 Tyree, Andrea, 506–517
Paavic, Irene, 965 Rosenbaum, James E., 395–402
Pager, Devah, 757–763 Rosenfeld, Jake, 80–85, 645–652 Vahinger, Hans, 1067
Pamepl, Fred, 1056–1063 Rostow, W.W., 1067 Vanneman, Reeve, 965–973
Pareto, Vilfredo, 188, 291(n3), Rouse, Cecilia, 820–830 Veblen, Thorstein, 35,
312 Rovere, Richard, 312, 314 1069–1070
Park, Robert Ezra, 722 Venkatesh, Sudhir, 355
Parkin, Frank, 193–201 Saint-Simon, Henri de, 1070 Voss, Kim, 39–42
Parsons, Talcott, 38, 178(n1), Sampson, Robert, 386–394
296(n4), 1122 Saperstein, Aliya, 687–695 Walder, Andrew G., 322–327,
Penner, Andrew M., 687–695 Sawhill, Isabel, 364 1110–1115
1156 • Names Index

Wallerstein, Immanuel, 162–164 Western, Bruce, 80–85, 431–440 Wright, Erik Olin, 149–161
Warren, John Robert, 246–250 Whitehead, Alfred North, 1066
Weber, Max, 146–147, 165–192, Williamson, Jeffrey, 1142 Xie, Yu, 587–595, 943–952
195–196, 198–199, 203, Wilson, Charles Erwin, 88
634, 639, 988, 1050, 1105 Wilson, James A., 501–505 Young, Michael, 1075–1076
Weeden, Kim, 222–232 Wilson, William Julius, 12–13,
Weill, Sanford, 101 347–357, 379–381, Zetterberg, H.L., 455
Weisshaar, Katherine, 1–16, 765–779 Zhou, Min, 729–739
44–50 Winant, Howard, 682–686
 SUBJECT INDEX

Absolute mobility rates, See also Discrimination; Race immigration boom, 712
457–460, 467–472 and ethnicity; Segregation income distribution behavior,
Academic performance, Age 69, 71
educational success and, economic inequality and, 262 Japanese immigrant enclaves,
563–568 intergenerational mobility, 714–716
Achievement gap, 535–550 480–495 managing globalization, 1132
Acquisition, social classes and, poverty rates as a function of, post-socialist party hierarchies,
176–178 358–361 1110
Adaptation, educational, 573, professional women’s labor See also China
576 force participation, 814 Asian Americans
Advanced industrialism, class stratification model, 509–510 boundary change among
structure of, 183–192 Agrarian societies. See immigrants, 726–727
Advanced placement (AP) Horticultural/agrarian importance of networks in
courses, 672 stratification hiring, 666–668
Affirmative action programs, 771, Agricultural labor, 715, 736–737, intermarriage of immigrants,
778 741–745 790, 792–798
African Americans AIDS epidemic, 351 prioritizing economic security,
class antagonism, 176 Alienated labor and the social 947
exclusion from social clubs, classes, 127–131 Asiaticism, 45(table), 46–47
300–301 Allocation rules, 2, 15(n5) Aspirations, 608–621
incarceration statistics, 388, allocation of work effort, Assets
432–434 866–868 advantaged and disadvantaged
intervention for disadvantaged institutionalization of reward groups, 3(table)
children, 413–415 allocation, 37–38 distributional structure and
leveled aspirations in a low- postindustrial end of scarcity, effects, 982, 985
income neighborhood, 1068 equalization stemming from,
608–621 tribal systems, 46 18–19
life expectancy and economic wage inequality in female institutionalization of, 15(n2)
inequality, 263, 265 occupations, 927–928 labor exploitation, 161(n12)
minority intermarriage, 790, Allocative discrimination, 912, measuring inequality, 10–11
792–798 920 post-socialist asset
racial formation in the United An American Dilemma (Myrdal), appropriation, 1111–1113
States, 682–686 377 rent benefits, 254
social transmission of poverty, Animus discrimination, 849, social position, 31
389–390 859, 861 status exploitation, 155–158
stereotype threat and student Anticapitalist revolutions, 155, Assimilation of immigrants,
achievement, 752–756 157–158 710–739, 788–799
union decline among black Anti-feminist backlash, 965–973 ‘At risk’ groups, 1128–1129
females, 650 Asia Attribution error (social cognitive
Wilson on American black exclusion movements of labor, process), 852–853
poverty, 347–357 703–706 Audit, employment, 759–760

1157
1158 • Subject Index

Authority relations, 146–148, activities, 956–957 declining significance of race

194–196 early childhood poverty, among blacks, 777–779
417–423 educated elite, 315
Benchmarking, compensation, effect of gender revolution educational differentials,
105–110, 111(n2) on, 959 524–535, 552–562,
Benign narratives, 18–21, 25(n1) effect of neighborhoods on 570–576
Bias, cognitive, 853–855 poverty, 395–396 Goldthorpe class schema,
Big-class trend in inequality, health inequalities, 466(table)
223–231, 482, 483(fig.), 1046–1047 habitus construction, 986,
484–485, 489–493 impact of incarceration, 989
Blind orchestra auditions, 437–438 incarceration statistics, 432
820–830 inherited social mobility, 41 inequality trends, 223–224,
Bobos, 310–315 investing in disadvantaged 227–229
Boundary-centered framework of children, 412–416 judgment of taste, 982, 985,
race, 743–745, 780–786 Latino immigrants’ racial self- 990–1001
identification, 785 knowledge class in
Cambridge stratification scale low-income urban fathers, postindustrial society,
(CAMSIS), 204 339–346 1067
Capital, 143–144, 1105–1106. multigenerational New Class, 150, 161(n13),
See also Cultural capital; disadvantages, 403–411 190–191, 302–310,
Human capital; Social of two working parents, 1106–1109
capital 803–807 postmodernist vision, 1116
Capitalism, 18–19 postindustrial dilemma of power elite, 286–287,
class conflict in the capitalist the shrinking family, 291–292(n4)
world economy, 162–164 1088–1090 race relations, 767–772
elites and power, 293–296 poverty rates as a function of reinterpreting Weber’s view,
Marxian theory on classes age, 358–359, 364 186–188
in capitalism and pre- poverty reduction in welfare rent-based analysis of,
capitalism, 131–141 regimes, 56(table) 251–260
post-Communist poverty-achievement link, rise of the New Class,
managerialism, 316–321 423–424 302–310
post-socialist capitalism professional women’s labor role in black poverty,
without capitalists, force participation, 818 379–384
1104–1109 segmented assimilation of ruling class, 276–282
upper class institutions, 301 immigrants, 729–739 significance of acquisition,
Weberian theory of class, SES determining unequal 176–177
status, and party, 165–171 childhoods, 1013–1022 situses in postindustrial society,
Caste societies, 45(table), 48, social network homophily, 1068
170–171, 280–281, 294, 679(n2) social importance of, 942–943
696–707 socialization, 36 socioeconomic indexes for
Categorization (social cognitive welfare benefits, 53 occupations, 246–250
process), 850–851, 854 China status and, 175–178, 184,
Children power of the ruling class, 280 203, 211–215
advantages and disadvantages, the evolving oligarchy, the black experience, 773–775
39 322–327 unequal childhoods,
Chicago’s urban black poverty, WTO membership, 1136 1015–1022
387–388 Cinema attendance, 1005–1012 Veblen’s view of the
class differentials in education, Citizenship rights, 2, 1122 revolutionary class, 1070
525–526 Civil assets, 3(table) voting and politics,
class-based raising, 22–23 Civil Rights Act (1964), 725, 920 1038–1042
deciding to continue in or Class Weberian critique of,
leave school, 526–528 as life condition, 254–256, 183–186
devaluation of “female” 259 See also Social class
 Subject Index • 1159

Class conflict in the capitalist racial categorization and, Death. See Mortality/morbidity
world economy, 162–164 688–689 Death rituals in tribal systems, 46
Class formation, 20–21 College selectivity, 579–580, 582 Decision tree model of
capitalism and pre-capitalism, College wage premium, 75–76, commitment, 623–627
131–141 76(table), 78(n4) Deindustrialization, 23–24
race and ethnicity hindering, Colleges and universities. See Demographic changes, global
191–192 Higher education inequality and, 1144
rents determining, 246–247 Communal actions, 167–169 Deprivation, 518
Class interests Communal relationships, Devaluation of skills, 924–928
communal action, 167–168 179–181 Diabetes, 1051–1055
property ownership, 175–176 Commuting patterns, 352–353 Diamond market, 705
Class mobility, 443–464. See also Comparative Analysis of Social Differential reward, 33–35
Social mobility Mobility in Industrial Diffusion process, 673
Class relationships, structuration Nations (CASMIN), Discrimination
of, 188–192 462, 464, 466, 555–556, American social clubs,
Class structures 555(table), 560 300–301
absolute mobility rates, Competition, 570–576, 636–638 as conscious or nonconscious
467–472 Condition, equality of, 461–462 process, 858–863
advanced societies, 183–192 Conscious discrimination, dual labor market, 629–632
class reproduction, 200–201 858–863 employment discrimination
divisions in the knowledge/ The Constant Flux (Erikson and among openly gay men,
service economy, Goldthorpe), 462(n1), 842–848
1125–1131 474–475 hiring of women, 955–956
intergenerational social Consumption functions, 2, labor market, 747–751
mobility, 480–495 15(n3), 52–53 Latino immigrants’ perception
music and arts consumption, Corporate governance, 98–100, of racial stratification,
1005–1012 104–105, 110–111, 784–785
status order and, 202 219–222, 284–285, penalizing mothers in the
Class struggle, types of, 168–169 297–302 labor market, 832
Class theory, 193–201 Cost-disease, 1083–1084 perpetuating racial and ethnic
Class-Exploitation Credentialism, 198–200, 1121, disadvantages, 8
Correspondence Principle, 1123–1124 redlining neighborhoods, 348
153 Crime, 439–440, 757–763 remedies for employment
Closed relationships, 179–182 Cuban Revolution, 737 discrimination, 849–858
Clothing, taste in, 998–1001 Cultural assets, 3(table), 306–307 statistical discrimination,
Cognitive development and Cultural beliefs, gendering, 629–630
skills 974–976 stereotype threat and
arguments for and against Cultural capital, 484(table), African-American student
equality, 522–523 989–990, 1105 achievement, 752–756
conscious and nonconscious Cultural forces, 20, 354–355 the impact of “blind”
gender discrimination, Cultural perspective, auditions on female
859–863 postmodernist, musicians, 820–830
differential investments and 1117–1120 urban hiring practices, 353
the achievement gap, Cultural taste, 1004–1012 See also Race and ethnicity
540–542 Culturalist phase of inequality, Disparate impact discrimination,
early childhood poverty, 12–14 860
417–423 Culture, conceptions of, Domination, exploitation and,
increasing level of 1118–1119 147–148, 150–151, 158,
competence, 519 Culture of critical discourse 278, 282–292, 942
parents’ investments in (CCD), 307–309 Downward mobility, 455–456,
children’s, 501–502 Culture of poverty, 378–379 469(table)–470(table), 688
poverty-achievement link, Cumulative inequality, 435–436 Dropout status, 507, 752
425, 520 Cycle of poverty, 507, 515 Drug trafficking, 351, 620
1160 • Subject Index

Dual labor market, 629–632 Education minority intermarriage,

ability and expectations of 790–791
Early childhood programs, success, 529–530 mortality disparities,
413–414 advanced placement (AP) 1056–1064
Earned Income Tax Credit courses, 672 occupational prestige, 235,
(EITC), 364, 367–368, America’s upper class, 298–299 236(n8), 241–242
373, 420 arguments for and against of the New Class, 303–304
Earnings determination, 34, equality, 522–523 positional rank, 29–30
635–639 Chicago’s Gautreaux postindustrial society,
Eastern Europe, 1098–1109 residential mobility 1089–1090
Economic conditions program, 397 postwar class models, 11–12
effect on labor migration, 730 class, gender, and economic poverty elimination theory,
Great Recession, 357–364, differentials, 524–535 520
1091–1097 class-and race-based Punjabi Sikh immigrants,
habitus construction, 985 differences in children’s 736–737
post-socialist stratification, experiences, 1019–1022 range of talent in a social
1098–1103 class-educational attainment population, 32–33
social foundations of link, 562–569 rent-based class analysis, 257
postindustrial economies, class-specific cultures, 13–14 resource distribution
1078–1091 college wage premium, differentials, 530–531
status stratification, 172–173 76(table) rise of the educated class,
taste in food, 991–992 converting talents to skills, 310–315
Economic crisis (2008), 86, 33–37 rising stratificational
1092–1093, 1134 credentialism, 15(n3), importance of, 1117
Economic growth 1123–1124 skill-biased technological
globalization, 1132–1133, deciding to continue in or change, 114
1137–1138 leave school, 526–527 standardized testing, 542
inequality retarding, 40–41 decline of unions, 80 stereotype threat and African
wealthy countries’ poverty early childhood programs, American student
policies, 365–369 413–414 achievement, 752–756
Economic inequality, 18–19 economic inequality, stratification model, 507–512,
education, technology and, technology and, 73–79 516
73–79 educational and occupational wealth as life condition,
educational differentials, attainment, 244–245, 254–256
524–550 536–562, 596–607 women moving into “male”
effect on equality of effects of the Great Recession fields, 957–962
opportunity and on employment, See also Higher education
intergenerational mobility, 1094–1095 Effectively declining inequality
501–505 equality of opportunity, 40, (EDI), 571, 574(fig.)
erroneous assumptions and 501–505 Effectively expanding inequality
explanations, 520–522 female musicians, 824 (EEI), 571, 574(fig.)
impact of higher education on gender gap among blacks, Egalitarianism, 18–19, 261,
socioeconomic outcomes, 778–779 902–911, 955–960,
587–595 gender revolution, 955 965–972
income inequality versus, gender wage gap, 937 Elites, 33
260–273 incorporation of immigrants, agrarian society, 46–47
taxing the poor, 369–375 732, 734–735 China’s oligarchic dual elites,
top incomes, 59–73 Internet service inequality, 323–327
tribal systems, 45–46 673–677 Mills on the power elite,
wage equalization, 521–522 leveled aspirations in a low- 282–292
Economic security with precarious income neighborhood, new upper class, 312
employment, 644 608–621 plantation economy, 770
Economic theory, 632–640, maximally maintained post-Communist
750–751 inequality, 531–532 managerialism, 316–321
 Subject Index • 1161

postindustrial society, 1071 See also Labor; Labor markets; Executive salaries, 86–103,
post-socialist elite Unemployment; Unions; 103(n23), 104–112,
opportunities, 1110–1115 Wages and salaries 111(n3)
power and, 292–296 Enlightenment, 15, 18, 20, 48 Expectations, aspirations and,
top income distribution, 60 Entitlement norms, 866–867 609–610
use of middleman minorities, Entrepreneurship hypothesis, Exploitation
712–713 1100 class as, 251
Employment Environmental hazards of domination and, 150–154,
capital market liberalization, poverty, 425–426 156–158
1134–1135 Equal opportunity, 518, 749 international exchange as,
college wage premium, Equal Pay Act (1953), 920 1143
76(table) Equity, norms of, 81–82 rent-based class analysis,
conscious and nonconscious Error discrimination, 861 251–258
gender discrimination, Esteem, 36–37 typology of class and,
858–863 Ethnic antagonism, 696–707 159–160, 159(table)
crime, incarceration, and Ethnicity. See Race and ethnicity workers in capitalism,
work opportunities, Europe 161(n11)
757–763 assimilation of immigrants,
diabetes management in the 726–727 Fair Housing Act (1968), 378,
workplace, 1053–1054 big-class inequality trends, 383
effects of early childhood 229–230 Family
poverty, 421–423 decomposition of capital, allocation of work effort,
Ehrenreich’s low-wage 143–144 866–868
workforce experience, Durkheim’s view of the class reproduction, 200–201
330–338 division of labor, 221 division of labor within a
employment discrimination health inequalities, 1046 society, 219
among openly gay men, immigrant enclaves, 713–714 early childhood poverty,
842–848 income distribution behavior, 417–423
employment discrimination 71 economic equality within,
and its remedies, 849–858 income distribution in Nordic 262
equality of opportunity, 40 countries, 66(fig.), 69–70, erosion of family
factors influencing 92 environments, 412–413
postindustrial society, inequality in educational gender divisions of labor, 9
1084–1086 attainment, 551–562 gender inequality in the labor
gender revolution, 955 intergenerational mobility, 503 force, 945–946
hopelessness in a low-income lack of underclass, 1127 hereditary status groups, 178,
neighborhood, 608–621 long-term social mobility 280–281
invisible inequality imposed findings, 464–479 home costs, 89
by incarceration, 434–435 post-Communist homeless population, 351
matching persons to jobs, managerialism, 316–321 impact of incarceration,
632–640 postwar class mobility, 437–438
Nordic welfare systems, 453–464 increasing scarcity of
53–54 poverty policy, 365–369 household time, 1085
Rust Belt’s decline, 352 rise of the New Class, 303 intergenerational income
socioeconomic indexes for social fluidity, 472–477 mobility, 496–501,
occupations, 246–250 top income share, 66(fig.), 67, 515–516
status order in British society, 68(fig.), 92 investment in children’s
205–211 union density increase and opportunity, 501–502
the impact of “blind” decrease, 102(n16) low-income urban fathers,
auditions on female Exclusion, social, 196–197, 339–346
musicians, 820–830 572–573, 574(fig.), mothers and the labor market,
urban black poverty, 354–356 631–632, 696–707, 1127 807–819, 831–842
within-gender wage gap, Exclusion movements of labor, postwar European class
912–918 703–706 mobility, 453–464
1162 • Subject Index

Family (continued ) Gautreaux program (Chicago), Glass ceiling, 890–900

reproduction of social 396–401, 409 Globalization, 6–7, 121–122,
mobility, 480–495 Gay and lesbian population, 1078–1079, 1132–1150
rise of the educated class, employment Governance, 105–106, 276–284,
310–311 discrimination among, 288–291. See also Political
status privileges, 171–172 842–848 forces; Regime type
time bind of dual working Gay Index, 1128 Gradational measurement
parents, 803–807 Gender approaches, 11–12,
transmission of conscious and nonconscious 225–227, 481–482,
multigenerational discrimination, 858–863 483(fig.), 484–485
disadvantages, 403–411 educational differentials, Gradational trends in inequality,
welfare regimes, 52–53 524–535 5–6, 225, 229
widening academic effects of the Great Recession, Great Backlash of conservatism,
achievement gap between 1093–1094 1035–1036
rich and poor, 535–550 higher education benefit, Great Depression, 59–60,
See also Children; 592(table) 518–519, 740, 1092
Intergenerational mobility; incarceration statistics, 432 Great Recession (2007–2009),
Marriage occupational achievement 62–63, 357–364,
Fatherhood, 339–346 measures, 248–249 1091–1097
Federal Housing Administration social importance of, 942–943 Group behavior, 277–278,
(FHA), 348 status order in British society, 850–851
Feedback effects of labor market 205–211
discrimination, 933 See also Women Habitus, Bourdieu’s concept of,
Feudal society, 45(table), 47, Gender attitude scale, 967–972 617–619, 985–989, 1105
132–141 Gender essentialism, 905–906, Haitian immigrants, 729–735
assets, exploitation, and 960–963 Head Start, 413
classes, 156–158, Gender inequality, 1, 8–9, 22–23 Health concerns
157(table) changing occupational sex class-based access to services,
exploitation of labor, composition, 881–888 22–23
161(n11) devaluation and the pay of fundamental causality
power elite, 286–287 comparable male and between SES and,
rise of the New Class, 303 female occupations, 1050–1056
social system, 1074 919–922 health inequalities and the
Financial deregulation, 100–101, egalitarianism and, 902–911 psychosocial environment,
103(n26) glass ceilings in corporate law 1046–1050
Flexicurity, 643–646 firms, 890–900 mortality disparities by SES,
Food, judgment of taste and, penalizing mothers in the 1056–1064
990–998 labor market, 831–842 Heterosexual relationships, effect
Food resources, 45–46 persistence of, 973–978 of gender revolution on,
Food taxes, 370, 373–374 persistent wage inequality 959–960
French Revolution, 139–140, in female occupations, Higher education
303 924–928 equalizing effects, 578–586
Friendship, 204–206, 653–657 sex segregation and women’s gender revolution, 955
Functionalism: evolution of careers, 876–880 gender wage gap, 937
modern stratification sources of the gender pay gap, impact on socioeconomic
systems, 1116–1125 929–939 outcomes, 587–595
Fundamental causality, within-gender wage gap, inequality in, 569–577
1050–1063 912–918 Japanese immigrants,
See also Women 715–716
Game theory, labor exploitation Gender revolution, 955–965 low-income population’s
and, 153–154 Gentlemen’s Agreement (1908), expectations and
Garden City, Kansas, 741–743 704, 714 aspirations, 616
 Subject Index • 1163

movement of Jewish occupational segregation by Immigration Act (1924),

immigrants into, 714 sex, 865–874 714–715
social psychological model of persistent wage inequality Immigration Act (1965), 712,
educational attainment, in female occupations, 721–722, 725–726
621–628 924–925, 929–930, Implicit Association Test, 861
women opting out of the 933–934 Incarceration, 388, 431–440,
workforce, 807–819 Hunting and gathering societies, 757–763
Hiring practices, 19, 752–756, 45–46, 45(table) Income
842–848 Hurricane Katrina, 347 defining, 60–61
Hispanic population Hypersegregated occupational effects of early childhood
immigrant incorporation, structure, 903–904 poverty, 420–423
735–736 Internet service inequality,
intermarriage of immigrants, Identity 673–677
790, 792–798 immigration patterns shaping voting patterns, 1042–1043
mortality disparity by sex and ethnic identity formation, See also Poverty; Wages and
race, 1060–1061(table) 740–745 salaries
organized labor and the Latino immigrants’ racial self- Income inequality, 23
immigrant population, identification, 782–784 changing forms of, 222–223
648–650 racial categorization, 689–690 decline of American unions,
Homeless population, 351 Ideographic phenomena, 1, 9, 17 80–85
Homines types, 1086–1089 Ideology, 141–142, 971–972, economic inequality and
Homology argument of 1120–1124 intergenerational mobility,
stratification, 1004–1012 Immigrant enclaves, 713–719 503–504
Honor, status, 169–173 Immigrant populations, 519 economic inequality versus,
Hopelessness through leveled adaptation to new societies, 260–273
aspirations, 608–621 710–721 executive salaries, 86–103,
Horticultural/agrarian assimilation theory, 721–728 103(n23), 104–112,
stratification, 18, determining the price of labor, 111(n3)
45(table), 46–48, 69, 698–700 global geography of,
1122–1123 ethnic identity formation, 1139–1150
Household inequality, 1126 740–745 income achievement gap and,
Housing Gentlemen’s Agreement 539–540
Chicago’s Gautreaux preventing Japanese labor income segregation and,
residential mobility migration, 704 1012–1034
program, 396–401 intermarriage and assimilation intergenerational income
failure of 1968 reform bill, theory, 788–799 mobility, 496–501
383 Latino immigrants’ place in knowledge/service economy,
income segregation, the racial order, 780–787 1126–1131
1024–1025 organized labor and Hispanic role differentiation,
low-income population, immigrants, 648–650 1122–1123
333–334 postindustrial society, top incomes, 59–73,
redlining, 348, 378 1088–1089 70(table), 91–103, 92(fig.)
segregation creating the prioritizing economic security, union power, 96–98
American underclass, 377 947 US executive pay, 86–90
Human assets, 3(table) racial self-identification, Income mobility, 496–501
Human capital, 484(table), 499 783(table) Income segregation, 1012–1034
education level reflecting segmented assimilation, Incorporation of immigrants,
investment, 588–589 729–739 710–721, 729–739,
human capital hypothesis, Thailand’s internal migration, 733(fig.)
1100 678–679 Increasing-disadvantage model,
inequality stemming from unauthorized immigration, 890–900
unequal ability, 659 743–745 Independent contracting, 641
1164 • Subject Index

Individualism, 40–41 racial fluidity and, 687–695 Kerner Commission report on

gender revolution and gender social networks increasing, segregation and poverty,
egalitarianism, 963 671–680 378
institutionalization of, See also Discrimination; entries Knowledge and culture,
1121–1122 beginning with Class; 279–280, 1067
moral individualism, 1120 Gender inequality; Income Knowledge/service economy,
postindustrial society, 1076 inequality; Race and 1125–1131
power elite, 282–292 ethnicity Kuznets curve, 18–19, 23, 69
ruling class and minority rule, Inequality transition,
277–278 1144–1149 Labor, 50, 76–77
social location of, 21–22 Inheritance, occupational, agricultural labor, 715,
Individualization of inequality, 444–445 736–737, 741–745
14, 1004–1012 Institutionalization of class college wage premium,
Individuals, inequality and, 5–6 conflict, 20 76(table)
Industrial relations, 82. See also Intellectualism, 141–142, decomposition of, 144–145
Unions 302–310, 1072 deindustrialization, 23–24
Industrialized societies, 45(table), Intergenerational elasticity, division of, 9, 46, 135,
48–49 498–499, 502–503, 580, 217–222
class conflict in, 143–149 580–581, 622 economic basis of race
income distribution behavior, Intergenerational inequality, 437 relations, 767–772
69 Intergenerational mobility, gendered stereotypes of
income inequality, 91–92 480–505 women in the labor force,
Marxian theory on classes educational and early 976–978
in capitalism and pre- occupational attainment ideology and class, 141–142
capitalism, 131–141 process, 596–607 income inequality, education,
occupational prestige, equalizing effects of higher technology, and, 73–79
233–236 education, 578–586 knowledge class in
post-socialist stratification, poverty elimination theory, postindustrial society,
1098 520 1067
postwar European class SES determining unequal knowledge/service economy,
mobility trends, 453–464 childhoods, 1013–1022 1125–1131
race relations, 765–766 stratification model, 507–511 manual and non-manual,
status order in British society, Intermarriage, assimilation and, 193–194
202–215 788–799 Marxian theories of class and
third industrial revolution, International Monetary Fund alienation, 127–131
113–122 (IMF), 1133–1135, 1139 Marxist stratification, 48–49
union labor and, 81–82 Internet adoption, 672–678, measuring class-based
Inequality 679(n3) inequality, 10–11
class-educational attainment Internment of Japanese mobility tables, 7–8
link, 562–569 Americans, 715–716 occupational prestige,
comparative perspective, 4–5 Intersectional perspectives of 233–236
conscious versus nonconscious gender and race, 943–952 post-occupationalism, 224
discrimination, 862 Invisible inequality, 434–435 rent-based class analysis,
functions and dysfunctions, 4 251–261
globalization and, 1132–1139 Japan: income distribution, secondary labor market, 620
mortality disparities, 66(fig.), 67, 70–71 skill-biased technological
1056–1064 Jews: immigrant enclaves, change increasing income
nonpersistent inequality in 713–714 inequality, 91–103
educational attainment, Jim Crow segregation, 770–771 the Great Recession,
551–562 Job ladders, 638 1091–1097
persistence of urban racial Job security, 641–643 third industrial revolution,
stratification, 386–394 Job-referral networks, 666–670 113–122
 Subject Index • 1165

top income distribution, immigrant assimilation spread of economic crisis,

59–73 theory, 724–725 1134
welfare regime, 52–57 immigrants’ racial self- Weberian theory of class,
See also Employment; identification, 783(table) 186–188
Marxism; Occupations labor market discrimination, See also Capitalism
and occupational status; 747–748 Market transition theory,
Professions; Unions power of the ruling class, 280 1100–1102
Labor force participation rise of the New Class, 303 Marriage
conscious versus nonconscious Latino immigrants, 780–787, 790 caste systems, 48
gender discrimination, Law firms, glass ceilings in, effect of gender revolution
858–863 890–900 on, 959
occupational segregation by Leapfrogging, 106–111 employment restrictions on
sex, 865–874 Legally privileged groups, women, 938
sex segregation and women’s 199–200 gender earnings gap,
careers, 876–880 Leisure time, 34, 204, 220–221, 950–952, 951(table)
time bind generated by, 803–807, 1015–1019 intermarriage of immigrants,
803–807 Liberal market economies, 5, 726–727, 788–799
women opting out of, 1134–1135. See also new upper class, 310–311,
807–819 Market economies 314
Labor markets, 20–21 Liberal theory of industrialism, role specialization theory,
changing occupational sex 453, 477 946–947
composition, 881–888 Life chances concept, 1050–1056 status order in British society,
corporate executive pay, Life conditions, class as, 254–256 204
104–112 Life expectancy, 262–263, 1046 urban black poverty, 379–380
discrimination within, Lifestyles, Bourdieu’s space Marxian models, 48–49
747–751 of, 984(fig.), 987–988, elites and power, 292–296
dual labor market, 629–632 987(fig.) ethnic antagonism, 696–697
ethnic antagonism and Low-income labor sex segregation in the
the split labor market, dual labor market, 629–632 workplace, 872
697–708 impact of the Great Recession, Veblen’s view of the
gender discrimination and the 361–362 revolutionary class, 1070
gender wage gap, 930–933 leveled aspirations in a low- Marxian typologies of inequality,
matching persons to jobs, income neighborhood, 44
632–640 608–621 Marxism
meritocracy hypothesis, See also Working class class theory, 193–201
581–584 Luxury, tastes of, 992–993 economic basis of race
networks, race, and hiring, relations, 767–772
664–670 Macrolevel externalities account, labor caste system, 705
precarious employment, 17 revolutionary organization,
640–646 Managerial groups, 194–196, 304
relationship ties, 655–656 316–321 See also Socialism
sacrifice of education, 34 Managerialism, post-Communist, Matrix of domination, 942
skilled immigrants, 712 316–321, 1108 Maximally maintained inequality
US income inequality, 64–65 Market economies (MMI), 570
See also Employment; China’s evolving market Meritocracy, 48–49, 572,
Occupations; Unions reforms, 326–327 578–584, 1068,
Labor transfers, 152–153 class determination in, 1075–1076
Laissez faire racism, 354 166–167 Mesoclass effects of social
Language communal action and class mobility, 492
American private schools, 299 interest, 167–168 Mexicans and Mexican
class differences in children’s post-socialist exchanges, Americans, incorporation
use of, 1017–1019 1098–1099 of, 735–736, 740–745
1166 • Subject Index

Miami, Florida, immigrants in, Narratives, 19–20, 23–25 early occupational attainment
729–735, 737–738 National Longitudinal Surveys process, 596–607
Microclass trend in inequality, (NLS), 498, 587–588, 689 field of academic study
224–229, 482–483, Native Americans: minority influencing mobility,
483(fig.), 484–485, intermarriage, 790–798 579–584
486(table), 489–491 Nativism, 742–743 gender inequality and
Middle Ages, 168–169, 174, 218 Natural inequality, 40–41 the hypersegregated
Middle class Necessity, tastes of, 992–993 occupational structure,
acquisition, 176–177 Neighborhoods, poverty and, 902–9910
children’s organized activities, 395–411 glass ceilings in corporate law
1015–1017 Neo-assimilation theory, firms, 890–900
class voting patterns, 1042 725–726 Haitian immigrants, 732–734
contradictory class locations, Neoclassical economic theory, intergenerational mobility,
159–160 633, 635–637, 865–874 480–495
deindustrialization, 23–24 Neo-Marxism, 149–152 mobility to first jobs,
growth of the black middle Network externalities, 671, 446–450
class, 772–773 673–676, 679(n1) post-communist
outmigration from the ghetto, Networks. See Social networks nomenklatura, 317–318,
380–381 New Class, 150, 161(n13), 319(table)
postwar European class 190–191, 302–310, postindustrial service
mobility trends, 454 1106–1109 economy, 1080–1083
privilege of scholars, 24–25 New Deal, 64–65, 101, 375, 1072 professional women’s labor
top income distribution, 60 New Geography Hypothesis, force participation, 816
See also Class formation; 1140–1149 queuing approach to
entries beginning with New inequalities account, 17, occupational sex
Class; Social class 21–22 composition, 881–888
Middleman minorities, 712–713 New Orleans, Louisiana, 347–348 social mobility, 489
Mobility mechanisms, 2, 7–8, Newsworthiness, 24 socioeconomic indexes for,
490–491, 496–501. See No Child Left Behind Act, 413 246–250
also Social mobility Nomenklatura, 317–318 the new educated class,
Modernity: egalitarian values and Nomothetic phenomena, 1, 9 310–311
institutional reforms, 902 Noncognitive skills, 413 women’s asymmetric change,
Monopoly rents, 257–258 Nonconscious discrimination, 955
Monopsony power, 931 858–863 See also Employment; Labor
Moral credentials, 24–25 Nonunion wage distribution, markets; Stratification
Moral economy, 81–82 83–85 models and systems
Moral individualism, 1120 Occupy Wall Street movement,
Moral life: division of labor Obesity, 1051 1, 86
within a society, 219 Occupational grading, 237–243 Oligarchy, 174, 322–327
Moral problem, inequality as, 18 Occupational prestige, 233–243, Omnivore-univore argument of
Mortality/morbidity, 1046, 1051, 247–249 stratification, 1004–1012
1056–1064 Occupations and occupational Open and closed social
Moving to Opportunity (MTO) status, 244–245 relationships, 179–182,
program, 396–401, 409 American private school 634–640
Multidimensional approach to graduates, 299 Opportunity, 491, 517–523
inequality, 2, 21–22 devaluation and pay of affirmative action programs,
Multigenerational disadvantage, comparable male and 771
403–411 female occupations, distribution of, 7–8, 18–20
Music: the impact of “blind” 919–922, 924–928, economic inequality effecting
auditions on female 956–960 equality of opportunity,
musicians, 820–830, 854 divisions in the knowledge/ 501–505
Musical consumption, service economy, equality of, 517–523, 749
1005–1012 1125–1131 functional importance, 33
 Subject Index • 1167

leveling aspirations of low- situses in postindustrial society, low-income urban fathers,

income youth, 611–621 1068 339–346
politics shaping, 40 union decline, 650–652 market volatility, 1135
post-socialist elite union role in the moral multigenerational
opportunities, 1110–1115 economy, 82 disadvantage, 403–411,
risk associated with, urban poverty, 349–352 520
1128–1130 winner-take-all politics, persistence of urban racial
social capital and human 91–103 stratification, 386–394
capital, 659–660 See also Governance popular and political
Outsourcing jobs, 76–77 Political participation, 519–520, definitions of, 518–519
955 rich countries’ policies for
Panel Study of Income Dynamics Political resources determining combating, 365–369
(PSID), 405, 498, 502 the price of labor, segregation and the American
Paternalism, 706–707, 716–717, 698–699 underclass, 376–385
1133–1134 Political rewards structures, skill formation in
Pay equity. See Wages and salaries 39–42 disadvantaged children,
Peasant class, 46–47 Positional capital, 927–928 412–416
Peer groups, 108(table), 618–619 Positional rank, 29–30 social transmission of,
Pension plan benefits, 642 Post-capitalism, 151–152 389–390
Perry Preschool Program, Post-communism/post-socialism, structural intervention, 387
413–414 1098–1115, 1144 taxing the poor, 369–375
Persistence account, 17 Postindustrial society, welfare states, 55–57
Persistent Inequality (Shavit and 1066–1097, 1125–1126 widening educational
Blossfeld), 551–552 Postindustrial world economy, achievement gap, 536–550
Personnel society, 1124 myth of, 1143 Wilson on American black
Philadelphia/Camden Fathers Postmodernism: big-class trend in poverty, 347–357
Study, 340–345 inequality, 223–224 See also Status, socioeconomic
Plantation economy, 770 Postmodernist phase of Poverty traps, 387, 389–391
Pluralism, 1076 inequality, 14, 16(n6), 21 Power
Policy. See Political forces Postmodernity, 902, 1116–1125 agrarian societies, 46–47
Political capital hypotheses, 1100 Post-occupationalism, 224 class conflict in industrialized
Political capitalism theory, Poverty, 329–364 systems, 146–148
317–318 as relative and absolute differentially valued strata, 36
Political class. See Power elite; condition, 518–520 economically determined, 165
Ruling class continuing inequalities among egalitarian ideologies, 48
Political economy, 101–102 blacks, 777–779 elites and, 292–296
Political forces, 20 cycle of, 515 ordering the labor queue,
agrarian society, 46–47 dual labor market, 629–632 885–887
big-class and microclass early childhood poverty, parties, 173–174
inequality trends, 229 417–423 postindustrial society,
deindustrialization, 23–24 Ehrenreich’s low-wage 1069–1075
Durkheim’s view of the workforce experience, separation of powers,
division of labor, 217–218, 330–338 141–142
220–221 Great Recession, 357–364 social power of parties,
government shaping the health hazards of early child 173–174
American economy, 94–95 poverty, 423–430 Weber’s analysis of class and
Great Backlash of importance of neighborhood status, 185
conservatism, 1035–1036 context, 395–402 Power elite, 282–296, 295(fig.),
in a postindustrial society, income segregation, 296–302, 325–326
1074 1012–1033 Precarious employment, 640–646
Jewish migration, 713–714 intergenerational mobility, Pre-communism, 1106
race relations, 772–773 504 Preindustrial period, race
racial formation, 685 intransigence of, 22–23 relations in, 765–766, 772
1168 • Subject Index

Presidential elections, voting Psychosocial environment, inequality, 943–952

patterns in, 1035–1045 1046–1050 Latino immigrants’ place in
Prestige attainment, 28–29, Public health, poverty and, the racial order, 780–787
31–32, 34, 36–37, 351–352, 423–430 mortality disparity by sex and
233–245, 286–287, 522, Public housing, 349–351 race, 1060–1061(table)
596 Punjabi Sikh population, networks, race, and hiring
Primary effects, 563–569 736–737 practices, 663–670
Primitive communism, 45–46 persistence of discrimination,
Private welfare, 53 Quality of life, 263, 265–270 22–23
Privilege, 175–176 Queuing, occupational sex poverty rates during the Great
acquisition and social class, composition and, Recession, 360–361
176–177 881–888 racial fluidity and inequality,
intergenerational nature of, 687–695
480–495 Race and ethnicity, 8 redlining neighborhoods,
legally privileged groups, assimilation of immigrants, 348–349
199–200 710–728 role of segregation, 377
of status, 171–172 caste systems, 48 stereotype threat and African
of the New Class, 307–308 defining race, 683 American student
studying race, class, and effects of the Great Recession, achievement, 752–756
gender intersections, 1095–1096 tangled pathology of poverty
942–943 ethnic antagonism, 696–707 in Chicago’s black
Professions, 195, 198–199 immigration patterns shaping neighborhoods, 391–392
Durkheim’s view of the ethnic identity formation, unequal childhoods,
division of labor, 217–222 740–745 1013–1022
penalizing mothers in the intermarriage of immigrants, Recession, 7, 62–63, 69,
labor market, 831–842 788–799 357–364, 1091–1097
postindustrial service Jewish migration, 713–714 Redlining, 378
economy, 1080–1083 racial formation in the United Regime change, 1098–1115
social mobility, States, 682–686 Regime type
486–487(table) social importance of, 942–943 China’s evolving oligarchy,
women moving into “male” the declining significance of 322–327
fields, 957–962 race, 765–779 post-Communist
women opting out of the the price of labor, 700–701 managerialism, 316–321
work force, 807–819 See also African Americans; postindustrial society,
Progress, theory of, 1120–1121 Discrimination; Hispanic 1086–1087
Property ownership, 166–169 population; Segregation ruling class, 277
China’s evolving oligarchy, Race relations, 765–776 See also Governance
324–325 Racial inequality Religion
class structure and, 175–176, crime, incarceration, and agrarian society, 46
183–184, 188–192 work opportunities, Great Backlash of
in a postindustrial society, 757–763 conservatism, 1035–1036
1071–1072 education, academic justifying the power elite, 288
managerial authority and achievement, and income power of the ruling class, 279
control, 195–198 inequality, 544–546, secularization of the New
post-Communist transition, 592(table) Class, 302–303
316, 318–319 ethnic segregation and caste, status and dignity, 171
power elite, 285–286 170–171 Remaking the American
rent-based class analysis, incarceration statistics, 432 Mainstream (Alba and
251–260 income segregation, Nee), 789
ruling class, 278–279 1023–1032 Rent-seeking, 111, 111(n1),
taxing the poor, 370 Internet service inequality, 251–260
Weberian analysis of class, 673–677 Replenishment, immigrant,
186–188 intersectional perspective on 744–745
 Subject Index • 1169

Representation security, 644 income segregation, 546, postindustrial society, 1071,

Reproduction theory, 524–535, 1012–1034 1076, 1090
617–618 Jim Crow, 770–771 price trends, 117–120
Residential mobility programs, new upper class, 312 technical innovation
396–402, 409–410 redlining neighborhoods, 348 replacing, 121–122
Residential segregation, 22–23, residential mobility programs, transmission to children, 482,
378 396–402 484
Resource distribution sex segregation and women’s union decline, 647–648
benign orientation to careers, 876–880 Slave societies, 44, 45(table)
inequality, 18 structure and process of sex Social background, educational
educational differentials, segregation, 865–874 success and, 563–568
530–531 the making of the American Social boundaries, assimilation
Haitian immigrants, 731–735 underclass, 376–385 theory and, 726–727
top income distribution, Self-development, 34 Social capital
59–73 Self-esteem, health inequality post-communist societies,
tribal systems, 46 and, 1048 1105–1106
union labor and, 81–82 Self-ownership, 18 social networks and status
welfare regimes, 55–57 Semiautonomous employees, attainment, 657–663
Retirement, 53–54, 57, 366–367 150–152 the strength of weak ties,
Reverse discrimination, 749–750 Service class. See Working class 653–657
Revolution, gender, 955–965 Service economy, 1079–1080 Social class
Revolution, Veblen’s view of, Service sector jobs, 620, 1121, agrarian society, 46–48
1070 1125–1131, 1144 analysis of class structure,
Reward allocation, 7–8, 15(n1), Sexual activity, effect of gender 149–161
33–42 revolution on, 959 basic typology of exploitation
The Rise of the Meritocracy Significant others’ influence and, 159–160, 159(table)
(Young), 1075–1076 (SOI), 598–603 caste societies, 48
Ruling class, 47, 276–282, Sikh population, 736–737 childraising, 22–23
292–302. See also Class Skill reproduction security, class conflict in industrial
formation; Social class 644–645 society, 143–149
Russian Revolution, 155 Skill upgrading, 79(n9) class-specific cultures, 13–14
Skill-biased technological change, defining and determining,
Sacrifice, 33–37 23–24, 74–79, 78(n2), 140–141
Santa Maria, California, 741–743 91–103, 113–122 exploitation and domination,
Science: role in postindustrial Skilled labor, 34, 74–76, 150–154
society, 1068–1069, 1073 113–122 ideology and, 141–142
Seattle, Washington: WTO Skills and talents industrial society, 48–49
protests, 1140–1141 commanding wealth, 255–256 inherited, 41
Secondary effects, 563–569 decomposition of labor, Marxian theory of alienation,
Secondary labor market, 620 144–145 127–131
Sector mobility hypothesis, 1100 earnings determination, measuring inequality, 10–11
Secularization, 302–303 635–637 phases of inequality, 11–12, 14
Segmented assimilation, 24, exploitation, 154–158 postcapitalist societies,
729–739 gender differences in job 151–152
Segregation qualifications, 929–930, social order and, 28–29
changing occupational sex 933–934 status and status groups,
composition, 881–888 gender inequality through 177–178
ethnic segregation and caste, wage discrimination, 920 stratification parameters,
170–171 measuring skill and its 45(table)
gender inequality and determinants, 116–117 Weberian theory of status,
the hypersegregated persistent wage inequality party, and, 165–171
occupational structure, in female occupations, See also Class; entries beginning
902–910 924–925 with Class
1170 • Subject Index

Social closure, 196–200, 572 urban blacks’ exclusion from, Social welfare. See Welfare
Social clubs, private, 299–301 353–356 systems
Social cognitive processes: women in the labor force, Socialism, 44, 45(table), 47
employment 876–877 China’s evolving oligarchy,
discrimination and, Social networks, 484(table) 322–327
850–853 immigrant populations, 724, decline in inequality, 18–19
Social cohesion, 656 734–735 post-socialism, 1098–1115
Social contract: precarious increasing inequality, South Africa: ethnic caste system,
employment, 643–646 671–680 705
Social control, sex segregation Jewish migration and Split labor market, 697–707,
and, 877–880 commercial enterprises, 767–769
Social democracy, 1087–1088 713–714 Standardized testing, 542, 573,
Social divisions, 1125–1131 network structure of social 575(fig.), 778
Social engineering, 20 capital, 660–662 Start-up firms, 977–978
Social environment determining race and hiring, 663–670 Statistical discrimination,
success and failure, 39–42 social networks and status 629–630, 750–751,
Social exclusion, 196–197, attainment, 657–659 849–850, 859–861, 866,
572–573, 574(fig.), the strength of weak 868–874, 931–932
631–632, 696–707, 1127 friendship ties, 653–657 Status, socioeconomic (SES), 203
Social fluidity, 467–468, Social order, 29–31 academic achievement gap,
472–478 Social psychological models, 536–550
Social formation 596–607, 621–628, automatic stereotyping, 852
elites and power, 294–296 687–695 class inequality in higher
racial formation in the United Social relationships, 179–182 education, 572–573
States, 682–686 economic inequality and, 262 crossing racial boundaries,
Social inclusion, 17–18 employment relationships 688
Social inequality, 431–440, in capitalist production, early occupational attainment,
1046–1050 634–635 598–603
Social insurance, 53 gendering, 973–977 educational test scores,
Social justice, 1125–1131, social networks and status 573–574, 575(fig.)
1138–1139 attainment, 657–659 in a postindustrial society,
Social location of the individual, the strength of weak ties, 1072–1074
21–22 653–657 income, status, and children’s
Social mobility, 41 Social reproduction, 387, 393, achievement, 542–546
American black underclass, 493, 524–535, 617–618 insecurity of the new upper
382–383 Social resources, theory of, 658 class, 315
caste systems, 48 Social Security, 366, 375 life chances, 1050–1055
downward mobility, 455–456, Social space, Bourdieu’s measures of, 537
469(table)–470(table), 688 description of, 982–985 mortality disparities,
immigrants, 790 Social standing, 32, 248, 1056–1064
long-term European study, 983(fig.) occupational achievement,
464–479 Social structure 244–245
measurement models, anti-feminist backlash, power elite, 291–292(n4)
480–495 967–972 race and class, 777–779
models of, 444–445 in a postindustrial society, racial categorization, 690–692
occupational prestige, 1073(fig.) social networks and, 657–659
233–236, 241–242 postindustrial society, 1066 socioeconomic indexes for
social order and, 28–29 race, class, and gender, occupations, 246–250
through higher education, 942–943 status-achievement gradients,
569–570 racial dimensions of, 684–685 537–539
upward mobility, 469(table)– Social transmission of poverty, stratification process,
470(table), 960–962 389–390 506–517
 Subject Index • 1171

transmission of Takeoff in inequality, 5, 17 market liberalization,

multigenerational Talents and skills, 32–33, 37–38, 1134–1135
disadvantages, 403–411 40 Theatre and cinema attendance,
See also Economic conditions; conversion of, 33–37 1005–1012
Poverty; Prestige income inequality, education, Threshold models of diffusion,
Status characteristic, motherhood technology, and, 74 673
as, 833 skill formation and investing Trade Protest Model, 1140–1149
Status groups, 165–171, 175–178, in disadvantaged children, Trade-liberalization agenda,
183–192, 202–215 412–416 1135–1137
Status order, cultural technological growth, 89 Trade-off thesis, 4
consumption and, tribal systems, 46 Tribalism, 45(table), 278
1005–1012 Tangle of pathology, 379, Trickle-down economics, 91,
Status socialization theory, 622 386–388 365–367, 1137–1138
Stereotyping Targeted social assistance, 52–53 Typology of inequalities, 44
employment discrimination Taste, judgment of, 982–1003
among openly gay men, Taxes Unauthorized immigration,
844–845 average rate for top income 744–745
gendered, 975–976 groups, 95(fig.) Underclass, 6–7
racial categorization and, distribution of top incomes, class-specific cultures, 13–14
688–689 59 economic basis of race
racial stereotypes in an era of economic growth for the poor, relations, 771–772
mass incarceration, 758 366–367 educational outcomes, 527
social cognitive process, executive pay, 90, 103(n23), EU lack of, 1127
851–852 111(n3) incarceration statistics, 432
social cognitive process income distribution, 72 non-black populations,
used in employment increasing poverty by taxing 381–382
discrimination, 854–855 the poor, 369–375 racial significance, 765
stereotype threat and policy shaping income segregation and the American
academic achievement, distribution, 94–96 underclass, 376–385
752–756 Taxi medallion system, 258 See also Class formation;
Stratification models and systems, Technocratic revolution, Poverty; Social class
506–517, 1116–1125 113–114, 304–305, 316, Unemployment, 53–54
elements of, 2–3 319–320 black/white ratio, 779
functional necessity of, 28–29 Technology, 89 Chicago’s black
ideal-typical systems, education, economic neighborhoods, 391–392
45(table) inequality and, 73–79 Great Depression, 1092–1093
range of talent in a social new geography of global Great Recession, 360
population, 32–33 inequality, 1144 role specialization theory,
social class and, 16(n7) postindustrial society, 1067, 946–947
social order and, 28–29 1073 Social Security benefits, 367
Stress response, health inequality women in start-up firms, Uniform elites, 294
and, 1047–1049 977–978 Unions, 64–65, 79(n6)
Structural-functional theory of Temporary Assistance for Needy class voting and class politics,
stratification, 243(n9), Families (TANF), 368, 1038–1039
722–723 374 density increase and decrease
Structuralist phase of inequality, Temporary work, 641 in Europe, 102(n16)
11–12 Terrorism, 1139 effect of union decline on
Structure of inequality, 5–10 Tertius gaudens (the third who American labor markets,
Subfields of inequality, 4 benefits) type, 662 646–652
Suffrage, 280–281 Thailand gender inequality, 978
Supplemental Poverty Measure internal migration in, increasing wage inequality
(SPM), 362, 364 678–679 with the decline of, 80–85
1172 • Subject Index

Unions (continued ) executive salaries, 86–103, wealth and class as life

post-occupationalism and the 103(n23), 104–112, conditions, 254–256
decline of, 224 111(n3) whitening effect of, 688
wage and salary workers in the income composition, 64 Welfare systems, 40, 52–57
US and Canada, 97(fig.) income inequality, education, culture of poverty theory,
winner-take-all inequality, technology, and, 74 378–379
96–98 intersectional perspective, devaluation of “female”
United Kingdom, 553–562 943–952 activities, 956–957
class voting, 1037 mean and median total dual labor market, 630–631
inequality in educational compensation of Ehrenreich’s low-wage
attainment, 553, 556 executives, 109(fig.) workforce experience,
occupational achievement, nonunion wage distribution, 330–338
244–245 83–85 equality of opportunity, 19
poverty policy, 365–369 Nordic welfare systems, 53 expenditure profiles, 55(table)
status order in, 202–215 occupational achievement, individual utilities and
See also Europe 248 individual incomes,
Universal class, New Class as, occupational segregation by 267–270
308–309 sex, 865–874 postindustrial societies,
Unskilled labor, 34, 75–76, open and closed employment 1086–1088
552–553, 704 relationships, 636–637 poverty programs, 366–367
Upper class, 297–302, 306, penalizing mothers in the rise of the black underclass,
310–315 labor market, 832 381
Upper-tail income inequality, persistent wage inequality White Australia policy, 696
1024 in female occupations, White ethnics, 745
Upward mobility, 469(table)– 924–928 Winner-take-all economics, 86–90
470(table), 960–962 rent-based class analysis, Winner-Take-All Politics (Hacker
Usurpation, social, 196–197 252–253 and Pierson), 86
Utilities, interpersonal skill change affecting, 113–122 The Winner-Take-All Society
comparisons of, 266–267 sources of the gender pay gap, (Frank and Cook), 89
929–939 Wisconsin: educational and
Vacancy competition, 637–638 union share of wage and occupational aspirations
Valuations, real income analysis salary workers in the US study, 600–605
and, 270–271 and Canada, 97(fig.) Within-job wage discrimination,
Valuative discrimination, 912, university professors, 313–314 912, 920
920–921 vacancy competition, 638 Women
Values, 19, 25, 1035–1036 wage equalization as anti- absolute mobility, 470(table)
Vertical mobility, 469(table)– poverty strategy, 521–522 aggregate class structure by
470(table) within-gender wage gap, country, 468(table)
Visibility account, 17 912–918 anti-feminist backlash,
Voting patterns, 1035–1045 See also Income inequality 965–973
Vulnerability of immigrant War on poverty, 518–519 children of two working
households, 731–735 Wealth and wealth distribution, parents, 803–807
64, 91–103 demand-led mechanism of
Wages and salaries, 23–24, China’s evolving oligarchy, wage increase, 115–116
34–37, 919 323–325 effect of residential mobility
college wage premium, 75–76 Marxian and Weberian programs, 398–399
devaluation and the pay of thought on class and, Ehrenreich’s low-wage
comparable male and 251–252 workforce experience,
female occupations, political definitions of poverty, 330–338
919–922 518 increasing wage inequality
Ehrenreich’s low-wage power of the ruling class, 279 with the decline of unions,
workforce experience, 331 the new upper class, 310–315 80–85
 Subject Index • 1173

knowledge/service economy, the impact of “blind” Great Backlash of

1126 auditions on female conservatism, 1035–1036
life expectancy and economic musicians, 820–830 low-income urban fathers,
inequality, 263, 265 union decline among black 339–346
low-income urban fathers, females, 650 mobility to first jobs,
340–345 See also Gender; Gender 446–450
mortality disparity by sex and inequality status order in British society,
race, 1060–1061(table) Working class, 49 214
Nordic welfare systems, 53 class voting patterns, 1042 tastes in food and clothing,
opting out of the workforce, class-specific cultures, 13–14 993–998
807–819 cultural egalitarianism, 18–19 See also Class formation;
post-Recession educational achievement, Labor; Social class
unemployment, 552–553 World Bank, 1136
1093–1094 Ehrenreich’s low-wage World Trade Organization
role in postindustrial society, workforce experience, (WTO), 1136,
1067–1068 330–338 1140–1141