The Main Drivers of E-Commerce Adoption: A Global Panel Data Analysis
Department of International Business & Economics, Faculty of International Business and Economics, Bucharest University of Economic Studies, 010374 Bucharest, Romania
*
Author to whom correspondence should be addressed.
J. Theor. Appl. Electron. Commer. Res. 2024, 19(3), 2198-2217; https://doi.org/10.3390/jtaer19030107 (registering DOI)
Submission received: 4 July 2024
/
Revised: 23 August 2024
/
Accepted: 28 August 2024
/
Published: 30 August 2024
(This article belongs to the Topic Digital Marketing Dynamics: From Browsing to Buying)
Abstract
:Digitalization has become more and more important for economic activities and economic development. E-commerce, as a part of local and international trade, is of increasing importance and is highly correlated with technological progress and innovation. Our research investigates the relevance of selected drivers that could explain the latest e-commerce adoption at global level. In our analysis, we used the UNCTAD B2C E-commerce Index (2014–2020, yearly data, covering 133 countries, 931 observations) which can be considered relevant to express an economy’s readiness to support e-commerce. E-commerce adoption is assessed in our research by the following six factors: (i) wealth, economic freedom, and economic development; (ii) access and sophistication of the financial sector; (iii) education; (iv) regulations; (v) development of ICT infrastructure; and (vi) frontier drivers (such as AI, cryptocurrencies, and blockchain technologies). In our research, we used a panel data analysis framework using hard data provided by different databases (UNCTAD, World Bank, etc.). The results we obtained confirmed that developed countries with a higher income level are higher adopters of e-commerce, financial development and the accessibility of financial services significantly help e-commerce adoption, a regulatory system (particularly economic freedom and property rights) strongly supports e-commerce adoption, education has a positive impact on e-commerce adoption, ICT infrastructure increases the adoption of e-commerce and the readiness and use of AI, and frontier technologies generate an increased adoption of e-commerce. The results we obtained are consistent with the findings of similar studies (most of them using different research methodologies) and opens the ground for interesting discussions and further research developments. The novelty of our research consists in the exhaustive perspective on e-commerce adoption drivers (including frontier technologies such as AI or blockchain) based on hard country data collected from various sources for a consistent panel of countries and a relevant number of years, providing an alternative approach to the mainstream studies on e-commerce adoption that process data from surveys, interviews, or focus groups.
1. Introduction
One of the most important effects of economic globalization is market integration. International trade has increased as trade barriers have decreased. New technologies, digitalization, and financial service sophistication have increased international trade and consumption based on a better appreciation of the competitive and comparative advantages of nations. E-commerce significantly changes the way of buying and selling goods and services, including cross-border transactions. The use of the Internet has rapidly extended to all types of transaction between businesses (B) and consumers (C), i.e., B2B, B2C, C2B, and C2C. Today, e-commerce refers not only to the effective transactions of goods and services but also to the transfer of corresponding information and money between the participants of these transactions. Moreover, the area of e-commerce is constantly expanded, today including very different activities such as online retailing, electronic marketplaces (trading platforms), online auctions, digital goods and services (e-books, digital music, etc.), banking and finance, and mobile commerce. Artificial intelligence and blockchain technologies are being proposed to add more dimensions to global e-commerce. The research goal of this paper is to test the main drivers of e-commerce adoption based on hard country data using a panel data analysis (PDA) framework. The main objectives of this research are to check if wealth, financial aspects, regulations, education, ICT infrastructure, AI, and frontier technologies (blockchain, cryptocurrencies, etc.) are relevant to explain e-commerce adoption.
In 2023, the volume of e-commerce was estimated by Statista at USD 4.2 billion (around four times higher than 10 years ago), and in 2027, e-commerce is estimated to double its volume (at USD 8.034 billion). The e-global commerce sector continues to be very concentrated, with four international trading platforms (Alibaba, Amazon, JD.com, and Pinduoduo) that together have around 52% of the total share of the global market (2023, Statista). According to data provided by UNCTAD (2023), Europe remains the most important provider for global exports of digitally deliverable services (51% of the market), followed by Asia (27%), and North America (18%); developed countries concentrate 76% of this market but with a negative trend since 2012 (−5% for developed countries, −4% for Europe).
The adoption of e-commerce is directly influenced by the capacity for consumption and the propensity of the population to consume, the wealth of nations, and specific infrastructure and technology, referring not only to ICT infrastructure but also to the infrastructure needed for financial services (payments, money transfers, etc.). Modern e-commerce has specific challenges and concerns, most of them regarding data protection, online consumer protection, cybercrime, and the security of e-transactions. According to data provided by UNCTAD Cyber Tracker (2023), the rate of the adoption of cyberlaw legislation has significantly increased, being close to 80–90% of countries in the world, the most important leaps since 2021 being made in the case of data protection and online consumer protection (from 40–50% in 2021 to 80% in 2023).
This paper is an empirical research following a panel data research methodology with the fundamental goal of analyzing the impact of the main relevant drivers for global e-commerce adoption: economic development and wealth; financial infrastructure (development, accessibility); regulatory system (regarding property rights, fundamental rights, etc.); education; ICT infrastructure; and frontier emerging technologies (such as AI or blockchain). The data of this research covers 7 years (2014–2021), 133 countries, and 21 variables, and six panel data structural models are proposed, following the considered research hypotheses. One of the most important challenges of a study like this is to find relevant and sufficient country data/indicators that could describe the dependent and explanatory variables, particularly in the field of e-commerce and in the case of emerging drivers (such as AI or blockchain). The novelty of this research consists in the exhaustive approach of e-commerce drivers that includes the main drivers identified by us from a literature review; the nature of data we used (hard country data) compared with the mainstream studies using data collected from interviews, questionnaires, focus groups, or surveys; the very consistent number of countries we included in this research that allows the possibility to extrapolate globally the results we obtained (133 countries with complete datasets); and the confirmation of the previous results, merely obtained by using other methodologies (adapted to the type of data used), by using an alternative methodology (panel data analysis). This study is developed in the following sections: review of the literature, research design, preliminary dataset tests, results, and final conclusions. The results we obtained confirmed the main research goal and the subsequent research hypotheses, which are in line with the most important findings of the economic literature and are statistically significant and robust. The conclusions section includes some useful policy recommendations following the results of our research and clarifies the limitations of this study and possible future developments of its findings.
2. Literature Review
The meta-study on the literature and research focused on e-commerce revealed the increasing importance of this field in academic research [1]: in 2017, the number of articles with this topic was only 150 in the business economics research area; in 2021, this number increased to 368. In 2021, the most important five research areas including research papers focused on e-commerce were business economics (1161 cumulated articles in the last considered five years, 2017–2021), computer science (757 articles), engineering (376 articles), operation research management science (217 articles), and science technology (206 articles). Another meta-study of the recent literature studying e-commerce supply chain management [2] identified seven main research topics: e-commerce operations, supply chain management, pricing strategy, coordination mechanism design, sustainable development management, and closed-loop supply chain management. E-commerce is a source of competitiveness, especially for small and medium enterprises, changing and improving their business model. Technology and the Internet have significantly contributed to the development of e-commerce over the last decades. The meta-study in [3] on the role of blockchain in e-commerce transactions analyzed the literature using the PRISMA methodology and presented the main research directions for e-commerce, which were merely focused on e-commerce security, privacy and challenges, and the latest research directions regarding the role of blockchain in e-commerce, such as how blockchain improves data security, increases business transparency and trust, decentralizes marketplaces, regulatory compliance improved by blockchain, etc.
By studying the literature between 2020 and 2024 (first 500 articles of 9745 articles containing the key word “e-commerce” indexed by Web of Science), we identified the following important research topics: (i) general studies focused on the identification of the most important drivers for the development and adoption of e-commerce by consumers and firms; (ii) the importance of new technologies and innovation for growth and development; (iii) the role of financial infrastructure in e-commerce development; (iv) shifts in the consumer behavior towards higher adoption of online buying and selling operations; (v) the security challenges for e-commerce, (vi) supply chain problems and e-commerce; (vii) the impact of e-commerce on the business sector and economic development; (viii) e-commerce and the environment; and (ix) new development opportunities for e-commerce (derived from blockchain and artificial intelligence). Furthermore, we identified some very interesting and original directions such as the importance of e-commerce for the internationalization of businesses (export and import activities), the role of e-commerce to improve rural development and reduce the poverty gap, and the impact of e-commerce on household financial leverage.
Ref. [4] studied the drivers for B2C e-commerce using the PLS-SEM methodology and found that it “is positively influenced by ICT access, political and regulatory environment and human resource development”. Ref. [5] used the survey methodology and identified a gap between large and SMEs regarding e-commerce adoption and found that better e-commerce literacy, greater business experience, and a more developed e-commerce business infrastructure (e-marketplaces, online trading and delivering platforms, websites, mobile applications, etc.) significantly enhance this sector. Another study focused on the barriers and key determinants for e-commerce adoption and utilization in the food and beverage retail SME sector and found that government support, e-commerce tools, and international experience positively influence the growth and development of the e-commerce sector [6]. The contribution of [7] to the identification of the main drivers contributing to the development of e-commerce is consistent. According to them, the drivers can be grouped into economic factors, technological factors, social factors, regulatory factors, psychological factors, and political factors. Following a survey methodology, they found that improved knowledge, skills, and competencies have a positive influence on the development of e-commerce. According to their findings, regulatory aspects and human skills (social aspects) are more important than economic, political, or technological aspects. Ref. [8] used a questionnaire method to study the key success-driving factors for the adoption of e-commerce by SMEs and found that “technological, organizational and environmental factors have a profound influence on the adoption of e-commerce, the role of organizational factors is limited and perceived complexity, compatibility and relative advantage, information intensity, management support have a large impact on the adoption of e-commerce”. Ref. [9] used a binomial logistic regression on data collected from 148 companies and found “top management support, learning orientation, acceptance of change, strategic orientation, IT readiness, cost and relative advantage as statistically significant determinants of e-commerce adoption”. In their research that collected data from 315 companies, [10] found that competition, business organization readiness for e-commerce, and desirability have a strong positive impact on the adoption and development of e-commerce. Moreover, they found that top management and government support have also a strong importance for e-commerce transactions, and the influence of business partners has no relevant impact on it. Ref. [11] used structural equation modeling applied on data collected through a survey of 283 companies, forming that the “CEO’s technology knowledge and relative advantage displayed a significantly positive influence on the adoption of e-commerce, while competitive pressure exhibited a negligible impact on the adoption of e-commerce”.
There is a general consensus about the positive impact of technology and innovation on e-commerce development and growth, on the fact that information communication infrastructure and technology play an essential role for e-commerce capacity, and the positive influence of the diffusion of innovation for e-commerce adoption and trust [12,13,14]. Ref. [15] found that “connectivity and technological efficacy directly influence e-commerce”, and [16] also concluded that information and communication technology empower the development of e-commerce by improving well-being and living standards.
The financial sector is considered to be decisive for the development of e-commerce. First of all, because the payments, money transfer, and even financing (as an important part of the supply chain) are important to be provided online [17,18]. Second, there is a problem of confidence and trust in the security of these financial services, the findings of [19] suggesting that “consumer knowledge about e-wallet technology relates to perceived usefulness, perceived ease of use, and trust, which are known to influence attitude and behavioral intention to adopt and use new technologies such as e-wallet”; their findings being similar to those of [20,21].
Consumers should be prepared to adopt the new way of buying and selling goods and services online, to accept paying or receiving money through the Internet. Online shopping is different from traditional retail and has specific limitations and challenges for all participants. It is generally accepted that the success and dynamic of e-commerce is strongly influenced by these required shifts in consumers’ behavior [22,23,24]. In these cases, communication with customers is fundamental for online transactions: sellers should listen to their voice and should be better connected with their opinions and perspectives, with e-WOM (electronic word of mouth) being different online than in the case of traditional business [25]. Following a structural model, [26] analyzed the intention to use mobile devices to buy products online and found that the “perceived usefulness of m-shopping and enjoyment emerged as the most informative variables that positively affect behavioral intention”.
The major concerns about privacy, data security, and intellectual property rights are influencing the adoption of e-commerce [27,28,29]. Beyond the technical solutions (communication tools, protocols, standards, etc.), the quality of regulations and institutions play a significant role in providing the required confidence and trust for the development of a favorable online environment for business [30,31]. E-trust is a key driver for the adoption of e-commerce. It is highly correlated with electronic education and technological literacy [32,33]. There are significant differences between large companies and SMEs regarding the adoption of e-commerce, between developed countries and less developed countries, and among different activity sectors. Education is one of the key factors explaining these differences. E-literacy improves e-trust in the use of new communication technologies by better understanding security, privacy, and data protection and understanding the proposed technical solutions to mitigate all of them [34,35].
The mitigation of the privacy and security problems of e-commerce through legislation and sound institutions designed to implement it has become fundamental for the adoption of online transactions [36,37]. Fundamental rights granted to consumers must be protected by adapted regulation and legislation to the specificity of online activity [38]. Moreover, companies should also be protected when deciding to explore and to expand their economic activity online. Different risks and threats should be considered in this case.
Artificial intelligence is one of the most important opportunities for the adoption and development of e-commerce. The support of AI in this sector is already deeply explored, and the main impacted areas are big data mining and processing (about clients, their behavior, prices, costs, etc.), which will significantly improve the quality of business decisions, optimizing customer experience and relationships, better predictions of online market developments, streamlining business processes, and improving security and data protection and privacy [39,40,41]. Blockchain is the other technology that already drives the development of e-commerce through enhancing the security, transparency, soundness, and efficiency of online transactions [42,43,44].
The latest research directions we can find in the economic literature regarding the emerging characteristics of e-commerce adoption are focused on measuring the positive impact of e-commerce on the environment due to the simplification of trading procedures, explaining how the COVID-19 crisis influenced the growth and development of the e-commerce sector, estimating the presumed positive impact of e-commerce on the efficiency and performance of companies (especially SMEs) due to less transaction costs, understanding and managing logistic problems in the case of global e-commerce development, and evaluating the impact of e-commerce on poverty reduction and economic growth. Following this research, we intend to continue our research in all these emerging directions by finding the relevant hard data that can be relevant proxies for all of them.
Summarizing, the research gap proposed by this research consists mainly of providing a different methodological approach to the main drivers of e-commerce adoption by using hard macroeconomic data that proxy the selected relevant drivers, by using a different methodology (panel data analysis—PCA), by considering a consistent number of countries (133) with a consistent number of complete data, and by including among the considered drivers the frontier technologies (such as AI, blockchain, cryptocurrencies, etc.). Therefore, the confirmation of the major research hypothesis from previous recent mainstream studies addressing e-commerce adoption, which use surveys, focus groups, questionnaires, or structural equations models (SEMs), by our study is an important research endeavor; in the social sciences, the consistency of findings is ensured by comparing the results obtained by alternative research methodologies and tools.
3. Research Design
3.1. Research Hypotheses
Based on the review of the studied literature, our research aims to investigate the relationship between e-commerce adoption and selected drivers considered relevant to explain it over time. Considering the aim of our research, we proposed in our study the following research hypotheses:
Hypothesis 1.
Higher economic growth and development positively influence e-commerce adoption. This research hypothesis can be found in the studies of [7,8,9,35]; the wealth of nations is said to have a positive influence on the adoption of e-commerce due to the higher consumption and investment propensity of people and companies. The level of income, the dynamic of the wealth of nations, the GDP growth rate, GDP per capital level, and GDP per capita growth rate are often used to estimate the prosperity of nations and their capacity to consume, to invest, and to produce intermediate or final goods. Moreover, economic prosperity is a good incentive to innovate and introduce and use the latest technologies in production and trade. Economic freedom is important for the ease of conducting business and contractual issues, both of which are vital for trade and production.
Hypothesis 2.
Better financial system accessibility and sophistication positively influence e-commerce adoption. The development of financial systems (banking services, payments, electronic money transfers, etc.), the accessibility of the financial sector for the business sector and consumers, and the diversification of financial products and services are positively influencing the adoption of e-commerce. This research hypothesis can be found in the studies of [17,18,19]; the positive impact of the financial system on the adoption of e-commerce is merely explained by the direct connection between electronic payments, financing tools, and money transfers and the e-commerce sector. The numbers of banks, the number of ATMs, the credit provided by the financial sector to the private sector, financial assets to GDP, the accessibility of the financial system, the interest rate gap (the affordability of financial services), the number of accounts per 1.000 inhabitants, and the number of cards per 1.000 inhabitants are the most used hard data as proxies for financial system accessibility and sophistication.
Hypothesis 3.
Higher economic freedom, better institutions, and regulations (i.e., those regarding intellectual property rights) positively influence e-commerce adoption. This hypothesis is included in other previous studies such as [36,37]. Economic freedom, regulations, and institutions are the key driving factors for any economic activity, including e-commerce. In addition, less trade barriers are considerably important for the development of international trade. Fundamental rights protection is relevant to describe the quality of institutions and regulations, which is very important for e-trust. Additionally, one of the most sensitive aspects in the digital sector is the protection of intellectual property rights and the ways to enforce the safeness of the transfers of these rights between the market participants, including at the international/global level.
Hypothesis 4.
Better education (including IT literacy) has a positive influence on the adoption of e-commerce. This research hypothesis can also be found in previous studies on the adoption of e-commerce such as [32,33,34,35], education being a key factor in the use of ICT and the adoption of e-commerce. Skills to use a computer, to operate digital applications, to set up an Internet connection, and to understand and deal with online security issues are vital in explaining the willingness of people today to trade online for various goods and services. ICT literacy is important for buyers and sellers, especially because today the solutions that facilitate their online presence are significantly improved and simplified. If you are skilled enough today, you can start an online business, or you can add this feature to an existing business, boosting your sales and diversifying the segments of your customers. Moreover, by digitalizing their sales and understanding the challenges and risks, companies can easily start today to internationalize their business, becoming better connected with customers from abroad. A lot of indicators can be used to estimate education, from the expenditure for education (as a percentage of total government expenditures, as a percentage of GDP) to the percentage of enrolled persons at the advanced education level relative to the total population (tertiary education), average grades, employment after graduation, the structure of educational infrastructure by the sciences, etc.
Hypothesis 5.
A better ICT infrastructure positively influences e-commerce adoption. This hypothesis can also be found constantly in previous studies ([12,13,14,15,16]), with e-commerce adoption not being possible without the required ICT infrastructure. The ICT infrastructure must be developed in line with the growth and development of e-commerce if we expect to stay on top of global trends. However, the development and quality of an ICT infrastructure requires significant capital investments and is also strongly connected to the financial system. A lot of indicators could be used as proxies for testing this hypothesis: the number of servers, the speed of Internet connection, number of Internet users, the use of mobile Internet and applications, etc.
Hypothesis 6.
A better AI and frontier technology readiness positively influences e-commerce adoption. This research hypothesis can be found in some very recent studies such as [39,40,41] looking to artificial intelligence as a driver for e-commerce adoption or [42,43,44] analyzing the relevance of blockchain and other frontier technologies for e-commerce adoption. The openness and readiness to these frontier technologies has become a key element in this respect. All these frontier technologies are submitted to have a positive impact on the evolution of e-commerce due to their significant improvements on security issues, customer experience and relationships, cutting trading costs, etc.
Based on these research hypotheses, the logical scheme and expected correlations between the main drivers considered and the e-commerce development index are represented in Figure 1. As we expect, following our empirical tests, e-commerce should be positively impacted by economic growth and development, positively influenced by the accessibility and sophistication of the financial system, positively correlated with better protected intellectual property rights, and positively impacted by better education, an improved ICT infrastructure, and a higher readiness for frontier technologies (AI, blockchain, cryptocurrencies, etc.).
3.2. Methodology and Model Parameters
Based on the type of indicators and the nature of the available data, we decided that the best option for this factorial analysis of the growth and development of e-commerce was a dynamic panel data analysis. For each research hypothesis tested in our paper, we used a different model. In each selected model, we tested two different aspects: (i) the development of the sector (e-commerce) based on the country–year data for selected explanatory variables (log values) and (ii) the growth of the sector based on the yearly changes of these explanatory variables (log values). In all our models, we used, as controlling variables, two common indicators describing the international openness of the country through trade and investments, one variable about total population, and one variable about final consumption (these controlling variables are common for all models). The model equations for each factor taken into consideration to explain the development and growth of the e-commerce sector are the following:
- Model 1 (Economic growth and development):where Explained variable: ECOMDEV—B2C E-commerce UNCTAD Index for each country “t”; Explanatory variables: GDPCAP—GDP per capita in each country “t” provided by the World Bank, DEMOCRACY– deliberative democracy index for each country “t”, INCOME10—pre-tax income shares of the top 10% of the population in each country “t”; Controlling variables: EXPORT—the export to GDP for each country “t”, POPULATION—the total population of each country “t” (it is relevant for the size of the country), CONSUM—total consumption to GDP for each country “t” (over time “i”).
- Model 2 (Financial sector):where Explained variable: ECOMDEV—B2C E-commerce UNCTAD Index for each country “t”; Explanatory variables: ACCESS—financing SMEs in each country “t”, BANKS—soundness of banks in each country “t”, VENTURE—venture capital availability in each country “t”; Controlling variables: EXPORT—the export to GDP for each country “t”, POPULATION—the total population of each country “t” (it is relevant for the size of the country), CONSUM—total consumption to GDP for each country “t” (over time “i”).
- Model 3 (Intellectual property rights and legal aspects):where Explained variable: ECOMDEV—B2C E-commerce UNCTAD Index for each country “t”; Explanatory variables: PROPERTY—intellectual property rights index in each country “t”, FUNDRIGHTS –fundamental rights index in each country “t”, REGENF—regulatory enforcement index in each country “t”; Controlling variables: EXPORT—the export to GDP for each country “t”, POPULATION—the total population of each country “t” (it is relevant for the size of the country), CONSUM—total consumption to GDP for each country “t” (over time “i”).
- Model 4 (Education):where Explained variable: ECOMDEV—B2C E-commerce UNCTAD Index for each country “t”; Explanatory variables: ENROL—school enrollment at the tertiary stage in each country “t”, EDEXP—total expenditure for education to total public expenditure in country “t”, EDEXPGDP—government expenditure on education to GDP in country “t”; Controlling variables: EXPORT—the export to GDP for each country “t”, POPULATION—the total population of each country “t” (it is relevant for the size of the country), CONSUM—total consumption to GDP for each country “t” (over time “i”).
- Model 5 (ICT infrastructure):where Explained variable: ECOMDEV—B2C E-commerce UNCTAD Index for each country “t”; Explanatory variables: BROADSUB—active mobile-broadband subscription per 100 people, INTUSERS– individuals using the internet (% of total population), INTHOUSE—households with internet access at home (% of total); Controlling variables: EXPORT—the export to GDP for each country “t”, POPULATION—the total population of each country “t” (it is relevant for the size of the country), CONSUM—total consumption to GDP for each country “t” (over time “i”).
- Model 6 (Frontier drivers):where Explained variable: ECOMDEV—B2C E-commerce UNCTAD Index for each country “t”; Explanatory variables: AI_READ—artificial intelligence readiness index, FT_READ—frontier technology readiness; Controlling variables: EXPORT—the export to GDP for each country “t”, POPULATION—the total population of each country “t” (it is relevant for the size of the country), CONSUM—total consumption to GDP for each country “t” (over time “i”).
3.3. Panel Data Samples and Description
In order to achieve the main research goal (to explain e-commerce development and the importance of the main drivers of it) and to test the assigned research hypotheses (from H1 to H5), we selected three types of variables: an explained variable (dependent variable), explanatory variables corresponding to each hypothesis, and common controlling variables for all panel regressions. We ran five different panel regressions for each research hypothesis. The short description of the variables included in our models is presented in Table 1 (including their data source).
Our data panels included 133 countries and a period between 2014 and 2020 (7 years, 931 country years of data). We decided to exclude years without complete country data and to exclude countries with missing year data in order to obtain a balanced and fixed panel for all models tested in our research. Our data panel is a wide panel having significantly more countries than years included in it. These options explain the limitation of our research to the selected period and to the selected countries. However, because of the high number of countries included in our research, the results can easily be extrapolated to all countries of the world and are significant for a longer period of time (including years before and during the pandemic crisis).
In addition to the relevance for the research hypotheses, another of the most important criteria for variable selection was the availability of country–year data. We started the research with a wider list of indicators selected for each research hypothesis, and we kept only the variables that provided the maximum number of observations for the countries included in the data panel.
A short statistical description of the data is presented in Table 2. According to the outputs, very few of the panel data series are normally distributed (PROPERTY, EXPORTS, and POPULATION), but this is a common feature of panels (they do not follow a normal distribution), taking into consideration that the data series are composed by country–year data. This feature of the data samples which do not follow a normal distribution does not produce significant biases in the panel data analysis and can be managed by including fixed and random effects in the estimations, following the corresponding tests.
3.4. Panel Data Preliminary Test
To avoid possible biases in the outputs of the models, the recommended tests for panel data analysis are the following:
- (i)
- Tests to check for the stationarity of variables (we used the standard four unit root tests: Levin, Lin, and Chu t test; Im, Pesaran, and Shin W-stat test; ADF—Fisher Chi-square test; and PP—Fisher Chi-square test);
- (ii)
- Tests to examine the long-term relationship among variables in each panel (we used the Pedroni test and the Kao test) and short-term relationship among variables (we used the VECM and Wald test on the coefficients);
- (iii)
- Tests to detect the presence of serial correlation (can be detected by the Wooldridge test and the Durbin–Watson Test);
- (iv)
- Tests to examine the presence of heteroskedasticity (the common tests being the White test and the Breusch–Pagan test);
- (v)
- Tests to choose between fixed effects and random effects models (both cross-section and period; the recommended tests are the Breusch–Pagan LM test and the Hausman test).
3.4.1. Unit Root Tests for Confirming the Stationary Statistical Property for Panel Data Series
The results of the unit root tests are presented in Table 3, for all variables included in the five models corresponding to the proposed research hypotheses (H1 to H6). The tests we used are standard tests for panel data analysis: Levin, Lin, and Chu t test; Im, Pesaran, and Shin W-stat test; ADF—Fisher Chi-square test; and PP—Fisher Chi-square test.
According to the results, following these unit root tests, it is clearly confirmed that all data series included in this research follow a stationary process (mean, variance, and autocorrelation are constant over time). Because stationary is confirmed for ‘level’, there is no need to transform the data by using specific methods (such as differencing or detrending).
3.4.2. Cointegration Tests
The second important panel data feature is cointegration among the selected variables. The presence of cointegration is generally beneficial for panel data analysis because it indicates a meaningful long-term deterministic relationship between the variables included in the model. For testing the presence of this long-term relationship, we looked to standard cointegration tests. Pedroni, Kao, and Fisher tests, but because the number of years is very limited, the only possible test was the Kao test (we also confirmed these results by VECM coefficients). We also tested the short-term deterministic relationship between the variables using the Wald test. The results of this test are summarized in Table 4, for the six panel data models used to test the research hypotheses of this research.
According to the results, following the cointegration tests (Kao and Wald), with only one exception (Model 3), we observed a strong and significant long-term and short-term cointegration among the variables included in our models. These results confirm that the selection of variables is in line with the economic assumptions and clearly improves the specifications and outputs of the models.
3.4.3. Serial Correlation Tests
The third preliminary test looks at the existence of possible serial correlation or autocorrelation (the correlation of a variable with itself and various lags). Serial correlation can be detected by using the Durbin–Watson test and the Wooldridge test (manually implemented by generating lagged series of residuals for each equation). The results of the Durbin–Watson tests and the Woolridge test (we used lags 1 to 3 for residuals) are summarized in Table 5.
A value for the Durbin–Watson (DW) tests close to 2 indicates the lack of autocorrelation. This test was also used for the Woolridge test. According to the results we obtained, in the case of Panel 1, there was no consistent autocorrelation in the panel data. In the case of Panels 2, 3, 4, 5, and 6, the autocorrelation is present. In order to avoid incorrect inference due to serial correlation errors, biases, or invalid hypothesis testing, a lagged residuals series was included in each panel that confirmed the presence of serial correlation.
3.4.4. Heteroskedasticity Tests
The presence of heteroskedasticity indicates the presence of non-constant variance in the error terms of a regression model generating biased standard errors, inefficient estimators, and invalid hypothesis tests. We used LR tests for cross-sectional homoskedasticity (the variance of errors is constant) and period homoskedasticity. The results of these tests are summarized in Table 6.
As we obtained following these tests, cross-section heteroskedasticity is present in all panels (the null hypothesis is rejected), and period heteroskedasticity is present in the cases of Panels 2, 4, and 5.
3.4.5. Fixed and Random Effects
Another important set of tests in the panel data analysis is the cross-sectional fixed-effects test (the potential existence of unobserved variables that differ across countries but are constant over time) and the period fixed-effects test (the potential existence of unobserved variables that might differ over time and are constant across countries). Additionally, we tested the cross-sectional random effects (to eliminate the possible biases of all time-invariant characteristics of the countries included in the panel) and the period random effect (to eliminate the potential biases due to the all-time specific characteristics that affect all countries from the panel similarly). The recommended tests are the Breusch-Pagan LM test (for fixed effects) and the Hausman test (for random effects). The results of all these tests are summarized in Table 7.
The results of these tests confirmed the existence of cross-section fixed and random effects and period fixed effects. Period effects were not tested due to the larger number of regressors than the number of years (the data panel is wide). The selection of the appropriate panel data regression took into consideration all these aspects and is presented in the summary of the outputs (Table 8).
4. Results and Discussions
Following the results of the preliminary tests (unit root, cointegration, serial correlation, heteroskedasticity, fixed and random effects), we selected the most appropriate regression model for the panels included in our analysis that follow the proposed research hypothesis. The results are summarized in Table 8 (all panels), and the parameters of the panel data models are summarized in Table 9 (all models).
These results fully confirmed Research Hypothesis 1: the wealth of a nation positively influences the dynamic and the intensity of e-commerce. Moreover, our research strongly confirmed that less income inequality has a direct impact on the dynamic and intensity of e-commerce, more people having access to financial services and the Internet, and more people having income to spend for their own benefit (Table 10). The second research hypothesis H2 testing the impact of the accessibility and sophistication of the financial system on the growth and development of e-commerce is partially confirmed. The development of the financial system and the diversification of financial products and services positively influence the dynamic and development of e-commerce, but the impact of accessibility is statistically inconsistent. The third research hypothesis H3 testing the impact of intellectual property rights, fundamental rights, and regulations on e-commerce growth and development is also partially confirmed. Property rights have a negative and statistically inconsistent impact, but fundamental rights and the regulatory system have a positive and statistically relevant impact on e-commerce growth and development.
The fourth research hypothesis H4 that estimated the impact of education on e-commerce adoption is partially confirmed by our study. There is a positive impact of the level of education and the intensity of the contribution of the economy to the educational system (education expenditure) and the dynamic and development of the e-commerce sector. More educated people mean more people who can use financial services and the Internet for their consumption. The results confirmed that ICT and financial literacy are more present in the education of people, and their economic behavior and attitude toward e-commerce and digitalization changed significantly. However, the negative relationship between e-commerce and expenditure for education to total public expenditure should be explored in order to be able to explain this result. The research hypothesis H5 is fully confirmed: there is a positive and statistically significant impact of ICT infrastructure development and e-commerce adoption. Finally, research hypothesis H6 is also fully confirmed: the frontier technologies (such as AI, blockchain, cryptocurrencies, etc.) are a positive driver for e-commerce adoption.
The impact of the controlling variables (exports, population, and consumption) was also positive and statistically significant (with very few exceptions). This means that countries with a higher international openness, higher size (in terms of population), and higher consumption registered a higher e-commerce adoption.
Typically for any research in the social sciences, the results of our study are statistically consistent but always limited to specific methodological aspects: the selection of indicators proxying the dependent and independent variables, the time period, the selection of countries included in the data panel. We tried to answer most of them by constructing research hypotheses following the latest research enquiries published in the most relevant academic journals. We selected the relevant indicators as proxies for the variables included in the model by taking into consideration similar studies and by consistently analyzing their description and methodology. We included a sufficient extended number of countries in the panel data (133 countries), and the maximum number of years provided the complete country data, and for robustness, we compared the results we obtained with the latest research on e-commerce adoption that used a different research methodology (surveys, focus groups, interviews, questionnaires, or structural equations).
It is true that the data panel includes only one year of the COVID-19 crisis because for 2021 there is still not so many hard data available. Therefore, this study is not consistent with the impact of COVID-19 on e-commerce adoption. We intend to study this problem in the near future when enough hard data will be available for a global panel of data. Moreover, in this case, we intend to use a different methodology, a counterfactual data analysis by constructing a relevant data series before COVID-19 and after COVID-19 to capture the differences in terms of e-commerce adoption.
In conclusion, the results we obtained confirmed the positive impact of the drivers considered for e-commerce adoption. The results we obtained are in line with the research of the hypotheses of our study and in accordance with the main latest findings in the review of the economic literature. The results are consistent and interesting for the further design of public policies and strategies addressed to this fundamental sector for economic growth and development.
5. Concluding Remarks
The purpose of this research was to test the main economic determinants considered for e-commerce adoption. The study provides a good perspective and understanding of the dynamics and other challenges of e-commerce by taking into consideration a global panel of data that is wide enough to provide consistency and significance for the conclusions based on it. This study included 133 countries and data covering the period between 2014 and 2020 (7 years), including 931 country–year observations. The research was developed taking into consideration five main hypotheses submitted for its main goal (to identify the main economic drivers for the e-commerce sector) and a total number of 19 variables (1 dependent variable, 3 controlling variables, and 15 explanatory variables, 3 of them for each of the 5 research hypothesis). After collecting all data for the selected variables, we obtained a balanced, fixed, and wide global panel of data.
Following the results, we obtained the following: Economic development has a positive impact on e-commerce adoption, meaning that more prosperous and wealthier nations are asked to spend more money for their online consumption, generating a higher adoption rate for e-commerce platforms (these results we obtained are in line with the findings of [7] which observed from the responses to their survey the importance of the business environment for the adoption of e-commerce, the findings of [8] which found the importance of economic freedom and reforms strengthening economic development as essential for the adoption of e-commerce, the findings of [9] which concluded that complexity negatively affects the adoption of e-commerce, the findings of [35] which observed from surveys a positive relationship between income level and e-commerce adoption, and the findings of [10] confirming the positive impact of competitive pressure on the adoption of e-commerce. We found that the financial intermediation, accessibility, and affordability of financial services determines a higher use of e-commerce, meaning that countries with a more inclusive and sophisticated financial sector provide a higher level of trust and transaction costs to the users of online services and e-commerce platforms (our findings are similar with [17], which observed a positive correlation between the implementation of biometrics, facial recognition, and QR codes by FinTechs and e-commerce adoption; [18], which confirmed the importance of the perceived usefulness, usability, and compatibility of e-payments for e-commerce adoption; and [19], which found the importance of the specific features and sophistications of e-wallets for their adoption by e-commerce users). Our research confirmed that a better regulatory system and better institutions defending the fundamental rights of people and their freedoms determine a higher e-commerce adoption rate (these findings are consistent with the findings of [36], which found that it is very important to adopt effective protection tools to prevent web-crawlers from harming e-commerce websites, and [37], which pleaded for improving and strengthening the institutional support for digital commerce “driven by cultural traditions and differences in the levels of socio-economic development” for better e-commerce adoption. Another important result we obtained is the confirmation that the education system is positively influencing e-commerce adoption. The results confirmed that the resources involved in the education system, and the higher level of education increase the trust in online transactions (similar to [32], which used a different methodology (LOGIT model) and found a positive determination between education level and PC skills and e-trust in 2019 compared to 2014; Ref. [33], which also used a different methodology (ARDL econometric model) and found a positive relationship between Internet abilities and e-commerce adoption; and [35], which used a different methodology based on surveys and also confirmed that e-commerce is more used among highly educated people). The development of the ICT sector has a positive influence on e-commerce adoption, meaning that countries with a better communication and technology infrastructure register higher online trade volumes (consistent with the similar results of [12] that clearly concluded, following a different research methodology (structural equation analysis—SEM), that innovation and innovation adoption is very important for e-commerce adoption; Ref. [13], which also used SEM methodology and found that infrastructure and information technology are essential for e-commerce adoption; Ref. [14], which found, based on a different methodology collecting data from surveys applied to SMEs, that innovation is very important for e-commerce adoption; Ref. [15], which used PLS analytical procedures and confirmed a positive influence of technological efficacy on e-commerce adoption, and, finally [16], which found that ICT affordance is definitely important for e-commerce development. Finally, the results of this study also confirmed that countries more open to AI and frontier technologies (results we obtained regarding the positive impact of frontier technologies (AI, blockchain, cryptocurrencies, etc.) on e-commerce adoption confirmed the previous results of [39], which analyzed the positive impact of AI data mining technologies for determining the probability of fraudulent behavior, increasing e-commerce safeness; Ref. [41], which proposed a multi-layer model to explore the positive impact of an AI application on the adoption of e-commerce by SMEs; Ref. [42], which explored the positive impact of blockchain technology on e-commerce architecture and found that this frontier technology could change the e-commerce business model in the directions of logistics, payments, data flow, and customs supervision; Ref. [43], which presented a model to assess the positive impact of basic cryptocurrency features on e-commerce adoption; and [44], which observed the positive impact of blockchain technologies on the disclosure risks, customer experience, and reduction in the redundancy of e-commerce flows. Furthermore, our study confirmed that the intensity of exporting activity (the international openness of the country), the size of the country, and the intensity of internal consumption are also positively influencing e-commerce adoption at the global level. The strong statistical significance of all these results is a strong point of this research; the results could be used as a relevant basis for future trends and developments and for extrapolating the results to all countries.
Following the results of our study, e-commerce adoption can be significantly improved by implementing public strategies and policies that address better education (including improved skills to deal with frontier technologies), a more developed and sophisticated financial system, better legislation, better ICT infrastructure, and more consistent support and openness to frontier technologies by all stakeholders (AI, blockchains, etc.). Reducing economic isolation, better regional and global integration, and participation in bilateral and multilateral trade and investment agreements could emphasize trade volumes for the e-commerce sector.
Besides the methodological aspects, there are a few limitations of our study derived from the limited number of years (only 7 years compared with the significantly larger number of countries), the limited selection of variables describing the e-commerce sector and each research hypothesis, and possible biases due to the selection of the controlling variables included in each panel model. Some inconsistent results that partially confirmed our research hypotheses require further analysis. Other further developments of our research will take into consideration a counterfactual analysis on e-commerce adoption drivers, comparing the period before and after the COVID-19 crisis, extending the number of years, finding other indicators that can emphasize the drivers of e-commerce adoption, testing the positive impact of e-commerce adoption on business performance and environment, and introducing a sectorial approach in the analysis for capturing the differences among relevant economic activities.
Author Contributions
Conceptualization, C.P., C.I., A.O. and D.D.; methodology, C.P.; validation, C.P., C.I., A.O. and D.D.; formal Analysis, C.P. and C.I.; investigation, C.P. and A.O.; resources, C.P. and D.D.; data curation, C.P.; writing—original draft preparation, C.P.; writing—review and editing, C.P., C.I., A.O. and D.D.; visualization, C.P.; supervision, C.P.; project administration, C.P. All authors have read and agreed to the published version of the manuscript.
Funding
This research received no external funding.
Institutional Review Board Statement
Not applicable.
Informed Consent Statement
Not applicable.
Data Availability Statement
Dataset available on request from the authors.
Conflicts of Interest
The authors declare no conflicts of interest.
References
- Özköse, H.; Kömür, G. Examination of e-commerce studies between 2016 and 2021 with science mapping and network analysis. COLLNET J. Scientometr. Inf. Manag. 2022, 16, 477–507. [Google Scholar] [CrossRef]
- He, P.; Wang, T.Y.; Shang, Q.; Zhang, J.; Xu, H. Knowledge mapping of e-commerce supply chain management: A bibliometric analysis. Electron. Commer. Res. 2022. [Google Scholar] [CrossRef]
- Albshaier, L.; Almarri, S.; Hafizur Rahman, M.M. A Review of Blockchain’s Role in E-Commerce Transactions: Open Challenges, and Future Research Directions. Computers 2024, 13, 27. [Google Scholar] [CrossRef]
- Adam, I.O.; Alhassan, M.D.; Afriyie, Y. What drives global B2C E-commerce? An analysis of the effect of ICT access, human resource development and regulatory environment. Technol. Anal. Strateg. Manag. 2020, 32, 835–850. [Google Scholar] [CrossRef]
- Amornkitvikai, Y.; Tham, S.Y.; Harvie, C.; Buachoom, W.W. Barriers and Factors Affecting the E-Commerce Sustainability of Thai Micro-, Small- and Medium-Sized Enterprises (MSMEs). Sustainability 2022, 14, 8476. [Google Scholar] [CrossRef]
- Amornkitvikai, Y.; Tham, S.Y.; Tangpoolcharoen, J. Barriers and Factors Affecting E-commerce Utilization of Thai Small and Medium-Sized Enterprises in Food and Beverage and Retail Services. Glob. Bus. Rev. 2021. [Google Scholar] [CrossRef]
- Jekabsone, I.; Gudele, I. Factors Contributing to the Development of E-Commerce by the Degree of use in Latvia. Eur. Integr. Stud. 2020, 1, 207–216. [Google Scholar] [CrossRef]
- Hossain, M.; Alhanakta, R.; Hervie, M.D.; Illes, C.B.; Nor, K.M. Exploring the Key Success Factors for sustainable e-commerce adoption in SMEs. Pol. J. Manag. Stud. 2022, 25, 162–178. [Google Scholar] [CrossRef]
- Sujatha, M.; Karthikeyan, S. Determinants of e-Commerce Adoption: Evidence from Small and Medium-Sized Enterprises in India. Int. J. Bus. Soc. 2021, 22, 574–590. [Google Scholar] [CrossRef]
- Ocloo, C.E.; Xuhua, H.; Akaba, S.; Shi, J.; Worwui-Brown, D.K. The Determinant Factors of Business to Business. B2B E-Commerce Adoption in Small- and Medium-Sized Manufacturing Enterprises. J. Glob. Inf. Technol. Manag. 2020, 23, 191–216. [Google Scholar] [CrossRef]
- Sin, K.-Y.; Sin, M.-C. Factors Influencing e-commerce adoption: Evaluation using structural equation modelling (SEM). Int. J. Bus. Soc. 2020, 21, 1192–1202. [Google Scholar] [CrossRef]
- Phumphakhawat, P.; Yodsurang, S.; Wisittigars, B.; Boonratmaitree, A. Creation of innovations and technologies that affect the effectiveness of e-commerce entrepreneurs among Internet merchants. Int. J. Early Child. Spec. Educ. 2022, 14, 2148–2155. [Google Scholar] [CrossRef]
- Zygiaris, S. The Impact of Innovation Systems on E-commerce capacity. J. Knowl. Econ. 2022, 13, 276–289. [Google Scholar] [CrossRef]
- Khsroo, I.T.N.; Burhanuddin, M.; Ali, M.A.; Shihab Ahmed, M. The Influence of Technological Factors on Adoption E-Commerce in SME: The Role of Trust. J. Glob. Inf. Technol. Manag. 2024, 27, 148–167. [Google Scholar] [CrossRef]
- Merhi, M.I. The Role of Technology, Government, Law, And Social Trust on E-Commerce Adoption. J. Glob. Inf. Technol. Manag. 2022, 25, 217–229. [Google Scholar] [CrossRef]
- Wu, W.; Zhang, Y.; Fan, Y. ICT Empowers the Formation and Development of Rural E-Commerce in China. IEEE Access 2020, 8, 135264–135283. [Google Scholar] [CrossRef]
- Amofah, D.O.; Chai, J. Sustaining Consumer E-Commerce Adoption in Sub-Saharan Africa: Do Trust and Payment Method Matter? Sustainability 2022, 14, 8466. [Google Scholar] [CrossRef]
- Wang, J. Impact of mobile payment on e-commerce operations in different business scenarios under cloud computing environment. Int. J. Syst. Assur. Eng. Manag. 2021, 12, 776–789. [Google Scholar] [CrossRef]
- Kınış, F.; Tanova, C. Can I Trust My Phone to Replace My Wallet? The Determinants of E-Wallet Adoption in North Cyprus. J. Theor. Appl. Electron. Commer. Res. 2022, 17, 1696–1715. [Google Scholar] [CrossRef]
- Sarmah, R.; Dhiman, N.; Kanojia, H. Understanding intentions and actual use of mobile wallets by millennial: An extended TAM model perspective. J. Indian Bus. Resour. 2021, 13, 361–381. [Google Scholar] [CrossRef]
- Gong, X.; Zhang, H.; Zhang, J. Digital Wallet, Happy Heart: An Analysis Based on the Economic–Social–Environmental Perspective. J. Theor. Appl. Electron. Commer. Res. 2024, 19, 1222–1242. [Google Scholar] [CrossRef]
- Liu, A.Y.; Lin, S. Service for specialized and digitalized: Consumer switching perceptions of E-commerce in specialty retail. Manag. Decis. Econ. 2022, 44, 1521–1533. [Google Scholar] [CrossRef]
- Svobodová, Z.; Rajchlová, J. Strategic Behavior of E-Commerce Businesses in Online Industry of Electronics from a Customer Perspective. Adm. Sci. 2020, 10, 78. [Google Scholar] [CrossRef]
- Lv, H. E-commerce Consumer Behavior Analysis Based on Big Data. J. Comput. Methods Sci. Eng. 2023, 23, 651–661. [Google Scholar] [CrossRef]
- Rouibah, K.; Al-Qirim, N.; Hwang, Y.; Pouri, S.G. The determinants of eWoM in social commerce: The role of perceived value, perceived enjoyment, trust, risk and satisfaction. J. Glob. Inf. Manag. 2021, 29, 75–102. [Google Scholar] [CrossRef]
- Saprikis, V.; Markos, A.; Zarmpou, T.; Vlachopoulou, M. Mobile Shopping Consumers’ Behavior: An Exploratory Study and Review. J. Theor. Appl. Electron. Commer. Res. 2018, 13, 71–90. [Google Scholar] [CrossRef]
- Kumbhakar, D.; Karforma, K.; Sanyal, S. A Secure and Efficient Authentication in E-Commerce. Biosci. Biotechnol. Res. Commun. 2021, 14, 93–99. [Google Scholar] [CrossRef]
- Cebeci, S.E.; Nari, K.; Ozdemir, E. Secure E-Commerce Scheme. IEEE Access 2022, 10, 10359–10370. [Google Scholar] [CrossRef]
- Saeed, S. A Customer-Centric View of E-Commerce Security and Privacy. Appl. Sci. 2023, 13, 1020. [Google Scholar] [CrossRef]
- Alkis, A.; Kose, T. Privacy concerns in consumer E-commerce activities and response to social media advertising: Empirical evidence from Europe. Comput. Hum. Behav. 2022, 137, 107412. [Google Scholar] [CrossRef]
- Affia, A.O.; Matulevicius, R.; Nolte, A. Security Risk Management in E-commerce Systems: A Threat-driven Approach. Balt. J. Mod. Comput. 2020, 8, 213–240. [Google Scholar] [CrossRef]
- Fernández-Bonilla, F.; Gijón, C.; De la Vega, B. E-commerce in Spain: Determining factors and the importance of the e-trust. Telecommun. Policy 2022, 46, 102280. [Google Scholar] [CrossRef]
- Jula, N.M.; Staicu, G.I.; Moraru, L.C.; Bodislav, D.A. Toward a Sustainable Development of E-Commerce in EU: The Role of Education, Internet Infrastructure, Income, and Economic Freedom on E-Commerce Growth. Sustainability 2024, 16, 3809. [Google Scholar] [CrossRef]
- Zeng, M.; Zheng, Y.; Tian, Y.; Jebbouri, A. Rural E-Commerce Entrepreneurship Education in Higher Education Institutions: Model Construction via Empirical Analysis. Sustainability 2022, 14, 10854. [Google Scholar] [CrossRef]
- Ayob, A.H. E-commerce adoption in ASEAN: Who and where? Futur. Bus. J. 2021, 7, 4. [Google Scholar] [CrossRef]
- Mei, M.; Tan, H. Data Expression and Protection of Intellectual Property Education Resources Based on Machine Learning. Complexity 2021, 2021, 5583389. [Google Scholar] [CrossRef]
- Sidenko, V. New institutional formats in the development of global e-commerce. Econ. Forecast. 2021, 2021, 85–110. [Google Scholar] [CrossRef]
- Peráček, T. E-commerce and its limits in the context of the consumer protection: The case of the Slovak Republic. Juridical Trib. 2022, 12, 35–50. [Google Scholar] [CrossRef]
- Li, J. E-Commerce Fraud Detection Model by Computer Artificial Intelligence Data Mining. Comput. Intell. Neurosci. 2022, 2022, 8783783. [Google Scholar] [CrossRef]
- Cheng, X.; Cohen, J.; Mou, J. AI-enabled Technology Innovation in e-commerce. J. Electron. Commer. Res. 2023, 24, 1–6. [Google Scholar]
- Wang, J. Using artificial intelligence to analyze SME e-commerce utilization and growth strategies. J. Comput. Methods Sci. Eng. 2024, 24, 611–621. [Google Scholar] [CrossRef]
- Zheng, L. Analysis of Computer-Based Blockchain Technology in Cross-Border E-commerce Platforms. Mob. Inf. Syst. 2022, 2022, 5083518. [Google Scholar] [CrossRef]
- Kim, S.-I.; Kim, S.-H. E-commerce payment model using blockchain. J. Ambient Intell. Humaniz. Comput. 2022, 13, 1673–1685. [Google Scholar] [CrossRef]
- Zhao, B.; Jiang, X.; Zhang, N.; Guo, Q.; Song, R. Design and Application of e-Commerce Platform System Based on Blockchain Technology on the Internet of Things. Wirel. Commun. Mob. Comput. 2022, 2022, 4448588. [Google Scholar] [CrossRef]
Figure 1.
The structured model for e-commerce adoption.
Table 1.
Variables included in the panel data models (synthesis).
| Type of Variable | Name | Short Description of Variables Extracted from the Data Providers’ Methodologies | Data Source |
|---|---|---|---|
| Dependent variable | The UNCTAD B2C E-Commerce Index (ECOMDEV) | This index estimates an economy’s capacity to support online shopping based on a combination of four relevant variables. | UNCTAD |
| H1—Explanatory variables | GDP per capita (GDPCAP) | Measure the value-added of an economy divided by the total population. | World Bank database |
| V-Dem Democracy Index (DEMOCRACY) | It is a composite index of developed variables that measures the country’s willingness to meet a few important principles of democracy, such as the electoral principle, the liberal principle, the deliberative principle, the participative principle, the majoritarian principle, the egalitarian principle, and the consensual principle. | The V-Dem Institute of the University of Göteborg | |
| Income share held by highest 10% (INCOME10) | This indicator is a measure of income inequality and is calculated as a percentage share of income that is the share that accrues to subgroups of 10% population. | World Bank database | |
| H2—Explanatory variables | Financing SMEs (ACCESS) | The values of this indicator express the response to the survey question ‘In your country, to what extent can small and medium enterprises (SMEs) access the finance they need for their business operations through the financial sector?’ [1 = not at all; 7 = to a great extent] | Competitiveness Index, World Economic Forum |
| Soundness of Banks (BANKS) | The values of this variable represent the response to the survey question ‘In your country, how do you evaluate the soundness of banks?’ [1 = extremely low banks may require recapitalization; 7 = extremely high banks generally have solid balance sheets] | Competitiveness Index, World Economic Forum | |
| Venture capital availability (VENTURE) | The values of this variable express the response to the survey question ‘In your country, how easy is it for start-up entrepreneurs with innovative but risky projects to obtain equity funding?’ [1 = extremely difficult; 7 = extremely easy] | Competitiveness Index, World Economic Forum | |
| H3—Explanatory variables | Property rights (PROPERTY) | Response to the survey question ‘In your country, to what extent are property rights, including financial assets, protected?’ [1 = not at all; 7 = to a great extent] | Competitiveness Index, World Economic Forum |
| Fundamental rights index (FUNDRIGHTS) | Measures whether individuals are free from any discrimination in society. | World Justice Project, Rule of Law Index | |
| Regulatory Enforcement Index (REGENF) | Measures whether government regulations (such as labour, environmental, public health, commercial, and consumer protection regulations) are effectively enforced. | World Justice Project, Rule of Law Index | |
| H4—Explanatory variables | Gross enrolment ratio for tertiary school (ENROL) | It is calculated by dividing the number of students enrolled in tertiary education regardless of age by the population of the age group which officially corresponds to tertiary education and multiplying by 100. | World Bank database |
| Expenditure on education to total public expenditures (EDEXP) | Divides total government expenditures on education into all public expenditures. Reflects the importance of education from the point of view of the government. | World Bank database | |
| Government expenditure on education to GDP (EDEXPGDP) | Divides total government expenditures on education into GDP and expresses the importance of education compared to the size of the economy and its contribution to value added. | World Bank database | |
| H5—Explanatory variables | BROADSUB—Active mobile broadband subscription per 100 people | It is calculated by dividing the total active mobile broadband subscription (handset and computer-based connections) by the total population. | International Telecommunication Union |
| INTUSERS—individuals using the Internet (% of the total population) | Individuals who used the Internet in the last 3 months divided to the total population. | International Telecommunication Union | |
| INTHOUSE—households with Internet access at home (% of total) | Total households with internet access via mobile of fixed connection divided to total households. | International Telecommunication Union | |
| H6—Explanatory variables | Artificial intelligence readiness index | Index based on three pillars: government pillar, data & infrastructure pillar, technology sector pillar. | Oxford Insights |
| Frontier technologies readiness index | Frontier technologies readiness index | UNCTAD | |
| H1-H5 Common controlling variables | Export to GDP | It divides the total exports of a country by its GDP and is a proxy for economic openness. | World Bank database |
| The total population | It counts the total population of a country and is an indicator of the size of the country. | World Bank database | |
| Total consumption to GDP | It is calculated by dividing the total consumption expenditures by the total GDP and measures the importance of this factor to the formation of GDP and its dependence on consumption. | World Bank database |
Table 2.
Descriptive statistics of panels’ variables.
| Dependent | Explanatory Variables Panel 1 | Explanatory Variables Panel 2 | ||||||
|---|---|---|---|---|---|---|---|---|
| ECOMDEV | DEMOCRACY | INCOME10 | GDPCAP | ACCESS | BANKS | VENTURE | ||
| Mean | 1.676 | −0.450 | −0.363 | 3.827 | 0.525 | 0.680 | 0.460 | |
| Median | 1.736 | −0.370 | −0.346 | 3.805 | 0.545 | 0.697 | 0.454 | |
| Maximum | 1.985 | −0.050 | −0.083 | 5.092 | 0.762 | 0.828 | 0.748 | |
| Minimum | 0.477 | −1.420 | −0.577 | 2.358 | 0.175 | 0.235 | 0.149 | |
| Std. Dev. | 0.262 | 0.342 | 0.089 | 0.616 | 0.122 | 0.095 | 0.111 | |
| Skewness | −1.236 | −0.844 | −0.109 | −0.017 | −0.594 | −1.080 | 0.072 | |
| Kurtosis | 4.467 | 2.841 | 2.631 | 2.060 | 2.927 | 4.534 | 2.520 | |
| Jarque-Bera | 320.395 | 111.620 | 7.118 | 34.304 | 49.990 | 247.831 | 8.858 | |
| Probability | 0.000 | 0.000 | 0.028 | 0.000 | 0.000 | 0.000 | 0.012 | |
| Explanatory Variables Panel 3 | Explanatory Variables Panel 4 | |||||||
| PROPERTY | FUNDRIGHTS | REGENF | ENROL | EDEXP | EDEXPGDP | |||
| Mean | 0.633 | −0.242 | −0.273 | 1.568 | 1.134 | 0.621 | ||
| Median | 0.625 | −0.238 | −0.293 | 1.702 | 1.140 | 0.636 | ||
| Maximum | 0.820 | −0.036 | −0.046 | 2.156 | 1.477 | 0.941 | ||
| Minimum | 0.375 | −0.652 | −0.569 | 0.481 | 0.342 | 0.192 | ||
| Std. Dev. | 0.090 | 0.124 | 0.113 | 0.383 | 0.137 | 0.146 | ||
| Skewness | −0.021 | −0.500 | 0.268 | −0.981 | −0.314 | −0.327 | ||
| Kurtosis | 2.679 | 2.941 | 2.398 | 2.968 | 3.754 | 2.827 | ||
| Jarque-Bera | 2.846 | 27.205 | 17.616 | 117.860 | 29.528 | 14.039 | ||
| Probability | 0.241 | 0.000 | 0.000 | 0.000 | 0.000 | 0.001 | ||
| Explanatory Variables Model 5 | Explanatory Variables Model 6 | Controlling Variables (all panels) | ||||||
| BROADSUB | INTUSERS | INTHOUSE | AI_READ | FT_READ | EXPORTS | POP. | CONSUM | |
| Mean | 1.721 | 1.682 | 1.626 | 1.621 | 3.625 | 1.510 | 7.140 | 1.887 |
| Median | 1.832 | 1.812 | 1.828 | 1.672 | 3.707 | 1.504 | 7.063 | 1.891 |
| Maximum | 2.421 | 2.000 | 2.000 | 1.968 | 4.000 | 2.283 | 9.147 | 2.111 |
| Minimum | −0.813 | 0.018 | −0.601 | 0.080 | 1.944 | 0.830 | 5.515 | 1.495 |
| Std. Dev. | 0.400 | 0.330 | 0.419 | 0.266 | 0.312 | 0.255 | 0.681 | 0.079 |
| Skewness | −2.153 | −1.635 | −1.697 | −2.289 | −1.800 | −0.006 | 0.131 | −0.780 |
| Kurtosis | 10.074 | 5.694 | 6.546 | 11.019 | 8.720 | 3.084 | 3.207 | 4.774 |
| Jarque-Bera | 2540.504 | 664.951 | 892.392 | 1890.032 | 1012.535 | 0.193 | 4.302 | 216.402 |
| Probability | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.908 | 0.116 | 0.000 |
Source: own estimations based on collected data.
Table 3.
Unit root tests for all variables.
| Variables/Hypothesis | Levin, Lin and Chu t | ADF—Fisher Chi-Square | PP—Fisher Chi-Square | ||||
|---|---|---|---|---|---|---|---|
| Statistics | Prob. | Statistics | Prob. | Statistics | Prob. | ||
| ECOMDEV | Dependent | −55.43 | 0.000 | 596.72 | 0.000 | 624.14 | 0.00 |
| GDPCAP | H1 | −18.83 | 0.000 | 451.03 | 0.000 | 496.52 | 0.00 |
| DEMOCRACY | −19.19 | 0.000 | 325.83 | 0.007 | 455.56 | 0.00 | |
| INCOME | −18.21 | 0.000 | 421.80 | 0.000 | 736.00 | 0.00 | |
| ACCESS | H2 | −35.36 | 0.000 | 443.75 | 0.000 | 202.11 | 0.971 |
| BANKS | −13.15 | 0.000 | 371.94 | 0.000 | 229.63 | 0.706 | |
| VENTURE | −30.46 | 0.000 | 406.36 | 0.000 | 199.77 | 0.978 | |
| PROPERTY | H3 | −22.33 | 0.000 | 268.22 | 0.000 | 452.76 | 0.000 |
| FUNDRIGHTS | −88.42 | 0.000 | 451.14 | 0.000 | 518.42 | 0.000 | |
| REGENF | −88.42 | 0.000 | 451.14 | 0.000 | 518.42 | 0.000 | |
| ENROL | H4 | −5.15 | 0.000 | 285.35 | 0.000 | 289.69 | 0.000 |
| EDEXP | −11.42 | 0.000 | 225.50 | 0.123 | 259.38 | 0.004 | |
| EDEXPGDP | −23.04 | 0.000 | 249.37 | 0.021 | 244.58 | 0.034 | |
| BROADSUB | H5 | −57.84 | 0.000 | 506.12 | 0.000 | 917.08 | 0.000 |
| INTUSERS | −4.86 | 0.000 | 226.39 | 0.893 | 420.83 | 0.000 | |
| INTHOUSE | −9.83 | 0.000 | 208.46 | 0.983 | 458.44 | 0.000 | |
| AI_READ | H6 | −10.51 | 0.000 | 372.53 | 0.000 | 465.83 | 0.000 |
| FT_READ | −116.68 | 0.000 | 750.46 | 0.000 | 1031.75 | 0.00 | |
| EXPORTS | Common Controlling variables | −69.20 | 0.000 | 385.07 | 0.000 | 313.87 | 0.00 |
| POPULATION | −46.68 | 0.000 | 468.60 | 0.000 | 587.07 | 0.00 | |
| CONSUM | −14.77 | 0.000 | 227.21 | 0.021 | 467.54 | 0.00 | |
Source: own estimations based on collected data.
Table 4.
Cointegration tests results (Kao Test for long-term relationship, Wald test for short-term relationship).
Table 4.
Cointegration tests results (Kao Test for long-term relationship, Wald test for short-term relationship).
| Models | Kao Test | Long-Term Cointegration | Wald Test | Short-Term Cointegration | ||
|---|---|---|---|---|---|---|
| T-Stat | Prob. | Chi-Square | Prob. | |||
| Panel 1 (Research Hypothesis 1) | −13.947 | 0.000 | Yes | 33,097.2 | 0.000 | Yes |
| Panel 2 (Research Hypothesis 2) | 2.952 | 0.002 | Yes | 589,097.2 | 0.000 | Yes |
| Panel 3 (Research Hypothesis 3) | −0.294 | 0.384 | No | 33,852.0 | 0.000 | Yes |
| Panel 4 (Research Hypothesis 4) | 1.296 | 0.098 | Yes ** | 463,932.8 | 0.000 | Yes |
| Panel 5 (Research Hypothesis 5) | −11.875 | 0.000 | Yes | 554,218.7 | 0.000 | Yes |
| Panel 6 (Research Hypothesis 6) | −3.941 | 0.000 | Yes | 566,727.8 | 0.000 | Yes |
Source: own estimations based on collected data.
Table 5.
Serial correlation results for all panel data.
| Autocorrelation | Panel 1 | Panel 2 | Panel 3 | |||
|---|---|---|---|---|---|---|
| Durbin-Watson Stat | 1.983 | No | 0.161 | Yes | 1.733 | Yes |
| Wooldridge Test | p-Values | D-W Test | p-Values | D-W Test | p-Values | D-W Test |
| Residuals Lag 1 | 0.035 | 2.038 | 0.000 | 2.154 | 0.043 | 1.912 |
| Residuals Lag 2 | 0.026 | 2.089 | 0.000 | 1.636 | 0.932 | 1.755 |
| Residuals Lag 3 | 0.022 | 1.966 | 0.064 | 2.174 | 0.941 | 1.743 |
| Autocorrelation | Panel 4 | Panel 5 | Panel 6 | |||
| Durbin-Watson Stat | 0.253 | Yes | 0.241 | Yes | 0.209 | Yes |
| Wooldridge Test | p-Values | D-W Test | p-Values | D-W Test | p-Values | D-W Test |
| Residuals Lag 1 | 0.000 | 2.212 | 0.000 | 2.113 | 0.000 | 2.511 |
| Residuals Lag 2 | 0.000 | 1.366 | 0.000 | 1.569 | 0.032 | 2.156 |
| Residuals Lag 3 | 0.947 | 1.901004 | 0.331 | 1.901 | 0.411 | 0.000 |
Source: own estimations based on collected data.
Table 6.
Heteroskedasticity tests for all panels (cross-section and period).
| Panels | Cross-Section Homoskedasticity (Null Hypothesis) | Period Homoskedasticity (Null Hypothesis) | ||
|---|---|---|---|---|
| Value | Probability | Value | Probability | |
| Panel 1 | 487.261 | 0.000 | 124.700 | 0.684 |
| Panel 2 | 455.030 | 0.000 | 251.904 | 0.000 |
| Panel 3 | 467.473 | 0.000 | 38.638 | 1.000 |
| Panel 4 | 511.669 | 0.000 | 294.812 | 0.000 |
| Panel 5 | 608.128 | 0.000 | 209.660 | 0.000 |
| Panel 6 | 719.943 | 0.000 | 0.793 | 1.000 |
Source: own estimations based on collected data.
Table 7.
Fixed and random effects (cross-section & period).
| Panels | Fixed Effects Tests | Random Effects Tests (Hausman) | ||||
|---|---|---|---|---|---|---|
| Cross-Section F | Period F | Cross-Section Random | ||||
| Statistic | Prob. | Statistic | Prob. | Statistic | Prob. | |
| Panel 1 | 1.170 | 0.108 | 3.848 | 0.001 | 86.671 | 0.000 |
| Panel 2 | 10.290 | 0.000 | 34.749 | 0.000 | 845.964 | 0.000 |
| Panel 3 | 1.205 | 0.113 | 0.949 | 0.449 | 70.541 | 0.000 |
| Panel 4 | 8.967 | 0.000 | 44.326 | 0.000 | 588.131 | 0.000 |
| Panel 5 | 9.881 | 0.000 | 37.883 | 0.000 | 847.394 | 0.000 |
| Panel 6 | 19.514 | 0.000 | 0.072 | 0.789 | 1065.077 | 0.000 |
Source: own estimations based on collected data.
Table 8.
Results (fitted models).
| Variables | Panel 1 (H1) | Panel 2 (H2) | Panel 3 (H3) | Panel 4 (H4) | Panel 5 (H5) | Panel 6 (H6) | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Coeff. | Prob. | Coeff. | Prob. | Coeff. | Prob. | Coeff. | Prob. | Coeff. | Prob. | Coeff. | Prob. | |
| GDPCAP | 0.269 | 0.000 | ||||||||||
| DEMOCRACY | 0.060 | 0.000 | ||||||||||
| INCOME10 | −0.470 | 0.000 | ||||||||||
| ACCESS | −0.039 | 0.563 | ||||||||||
| BANKS | 0.246 | 0.000 | ||||||||||
| VENTURE | 0.534 | 0.000 | ||||||||||
| PROPERTY | −0.003 | 0.985 | ||||||||||
| FUNDRIGHTS | 0.356 | 0.003 | ||||||||||
| REGENF | 0.672 | 0.000 | ||||||||||
| ENROL | 0.383 | 0.000 | ||||||||||
| EDEXP | −0.334 | 0.000 | ||||||||||
| EDEXPGDP | 0.163 | 0.000 | ||||||||||
| BROADSUB | 0.117 | 0.000 | ||||||||||
| INTUSERS | 0.333 | 0.000 | ||||||||||
| INTHOUSE | 0.140 | 0.000 | ||||||||||
| AI_READ | 0.209 | 0.000 | ||||||||||
| FT_READ | 0.196 | 0.000 | ||||||||||
| EXPORTS | 0.085 | 0.000 | 0.184 | 0.000 | 0.003 | 0.936 | −0.004 | 0.796 | 0.008 | 0.565 | 0.119 | 0.000 |
| POPULATION | 0.048 | 0.000 | −0.017 | 0.004 | 0.004 | 0.771 | −0.013 | 0.034 | 0.002 | 0.603 | −0.004 | 0.760 |
| CONSUM | 0.227 | 0.004 | −0.364 | 0.000 | −0.477 | 0.006 | −0.598 | 0.000 | −0.180 | 0.000 | −0.454 | 0.000 |
| C | −0.393 | 0.052 | 0.871 | 0.000 | 2.857 | 0.433 | 2.624 | 0.141 | 0.999 | 0.000 | 1.360 | 0.000 |
Source: own estimations based on collected data.
Table 9.
Panel data models’ parameters and significance.
| Model Statistics: | ||||||
|---|---|---|---|---|---|---|
| Adjusted R-sq. | 0.727 | 0.874 | 0.309 | 0.895 | 0.878 | 0.747 |
| F-statistic | 413.333 | 418.735 | 36.613 | 449.690 | 457.728 | 109.040 |
| Prob (F-statistic) | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 |
| DW stat | 1.975 | 2.227 | 1.912 | 2.167 | 2.072 | 2.234 |
| Model Specification: | ||||||
| Method | Panel EGLS | Panel EGLS | Panel EGLS | Panel EGLS | Panel EGLS | Panel EGLS |
| Coef cov. method | Period SUR | Period SUR | Period SUR | Period weights | Cross-sect. SUR | Period weights |
| GLS Weights | Period SUR | None | None | None | None | None |
| Random effects | None | Cross-section | Cross-section | Cross-section | Cross-section | Cross-section |
| Fixed effects | None | Period | Nome | Period | Period | None |
Source: own estimations based on collected data.
Table 10.
Research hypotheses and results (synthesis).
| Research Hypotheses | Expected Impact | Confirmation by the Obtained Results |
|---|---|---|
| Research hypothesis H1 | Positive impact of economic development on the adoption of e-commerce | Fully confirmed |
| Research hypothesis H2 | Positive impact of financial accessibility and sophistication on the adoption of e-commerce | Partially confirmed (“accessibility” is not confirmed) |
| Research hypothesis H3 | Positive impact of intellectual property rights and legislation on the adoption of e-commerce | Partially confirmed (“property rights” is not confirmed) |
| Research hypothesis H4 | Positive impact of education on the development on the adoption of e-commerce | Partially confirmed (“expenditures for education” partially confirmed) |
| Research hypothesis H5 | Positive impact of the development of the ICT sector on the adoption of e-commerce | Fully confirmed |
| Research hypothesis H6 | Positive impact of frontier technologies (artificial intelligence, blockchain, cryptocurrencies) in the adoption of e-commerce | Fully confirmed |
Source: own estimations based on collected data.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
MDPI and ACS Style
Paun, C.; Ivascu, C.; Olteteanu, A.; Dantis, D. The Main Drivers of E-Commerce Adoption: A Global Panel Data Analysis. J. Theor. Appl. Electron. Commer. Res. 2024, 19, 2198-2217. https://doi.org/10.3390/jtaer19030107
AMA Style
Paun C, Ivascu C, Olteteanu A, Dantis D. The Main Drivers of E-Commerce Adoption: A Global Panel Data Analysis. Journal of Theoretical and Applied Electronic Commerce Research. 2024; 19(3):2198-2217. https://doi.org/10.3390/jtaer19030107
Chicago/Turabian StylePaun, Cristian, Cosmin Ivascu, Angel Olteteanu, and Dragos Dantis. 2024. "The Main Drivers of E-Commerce Adoption: A Global Panel Data Analysis" Journal of Theoretical and Applied Electronic Commerce Research 19, no. 3: 2198-2217. https://doi.org/10.3390/jtaer19030107
