1. Introduction
The development of information and communications technology (ICT) has improved accessibility to vast amounts of information and has raised the level of openness around the world. This has affected not only the degree of ICT use by citizens, but also the administrative operations of governments. ICT has laid the foundation for governments to respond effectively to citizens’ needs for administrative transparency and has paved the way for the implementation of e-government systems. Along with the establishment of e-government infrastructure, most countries have enhanced national competitiveness through the development of public services, thereby laying the foundation for maturing into world-leading ICT hubs.
Past research on e-governments has mainly discussed the correlation between trust in the government and administrative transparency or the level of national corruption, which is consistent with the purpose behind e-governments. Since the occurrence of corruption varies by country and class, and [
1], studies have been confined to the aforementioned correlation and have failed to explore the causal relationship between e-government and government corruption [
2]. The correlation between e-government and government corruption has been examined from three perspectives, namely, decrease in corruption, trigger of corruption, and no relation [
2]. Most research has shown that e-governments have contributed to a decrease in government corruption. That is, the governments themselves make the administrative system more transparent using ICT and subject themselves to greater monitoring by the citizens, and this ultimately lowers the degree of corruption.
In some countries, a contradictory relationship was observed between e-government development and government corruption. For instance, Korea’s e-government ranked first in the United Nations (UN) E-Government Survey in 2012 and 2014 [
3], but the level of public trust in the government stood at 25% in 2013 and 34% in 2014 [
4]. In terms of the corruption perceptions index (CPI), it fell from 37th place in 2015 to 52nd in 2016 [
5]. What leads to this contradictory relationship?
This study was initiated based on the contradictory relationship between e-government development and government corruption. While past studies found that e-governments have had positive effects in decreasing government corruption, some factors may have been excluded in their discussions of government corruption. By examining previously unaccounted factors, this study may discover a new correlation between e-government development and government corruption. Most past studies focused on analyzing superficial factors and adopted a one-directional approach to the relationship between e-government and government corruption. From another perspective, it could be that there is little or no relationship between e-government and government corruption. This highlights the need to re-establish the relationship between the two factors, especially considering how recent advancements in ICT and the upcoming fourth industrial revolution have accelerated the development of e-governments.
The purpose of this study is to derive factors contributing to decreases in corruption based on a literature review, and to examine the relationship between e-government and government corruption through a cross-country analysis. Instead of taking a one-dimensional approach, this study first analyzed how government corruption in each country is influenced by e-government user status, governance, and regulations. With ICT and e-government development as mediating variables, the effects of various factors on government corruption were examined by country. That is, the mediating effects of ICT use and e-government development (telecommunication infrastructure, online service, e-participation) were assessed in order to define the relationship between users, governance, regulations and government corruption. This study ensured objectivity of data by utilizing statistics provided by credible organizations such as the the Transparency International (TI), Organization for Economic Cooperation and Development (OECD) and United Nations (UN) and differs from past research in that it focused on the mediating effects of e-government. Based on a multi-dimensional analysis of factors contributing to a decrease in government corruption, this study proposes anti-corruption strategies to be considered in the government policymaking process.
3. Hypothesis, Research Models & Statistical Methods
3.1. Hypothesis and Research Models
Based on the aforementioned studies, the factors contributing to a decrease in government corruption are economic and educational status of e-government system users, governance, and government regulations. The mediating factors in the decrease of government corruption are ICT use indices: telecommunication infrastructure (TII), online services (OSI), and e-government participation (EPI). The following hypotheses are proposed to examine factors affecting government corruption and the mediating effects of ICT use and e-government.
First, Hypothesis 1 was proposed to analyze the relationship between the various factors and the actual decrease in government corruption. The factors tested were the e-government user status, governance, and government regulations.
Hypothesis 1 (H1). The higher the e-government users’ status and governance quality, and the greater the relaxation of government regulations, the more positive the effect on decreasing government corruption.
Hypothesis 1-1 (H1-1). The higher the e-government users’ status, the more positive the effect on decreasing government corruption.
Hypothesis 1-2 (H1-2). The higher the quality of governance, the more positive the effect on decreasing government corruption.
- (3)
Government size
- (4)
Voice and accountability
Hypothesis 1-3 (H1-3). The greater the relaxation of government regulations, the more positive the effect on decreasing government corruption.
- (5)
Business regulations
- (6)
Credit market regulations
In examining based on Hypothesis 1, the research model shown below (see
Figure 1) was formulated to examine the influence of e-government user status, governance and government regulations on the decrease of government corruption. Among the factors affecting government corruption, the e-government users’ status was categorized by education and GDP. Governance was categorized by government size and voice and accountability, and government regulations by business regulations and credit market regulations. The control variables were world press freedom, political rights, and civil liberties.
Hypothesis 2 was proposed to analyze the mediating effects of ICT use and e-government in the decrease of government corruption. Based on past studies, the e-government development index (EDI) was divided into the telecommunication infrastructure index (TII), the online service index (OSI), and the e-participation index (EPI).
Hypothesis 2 (H2). The higher the level of ICT use and e-government, the more positive the effect on decreasing government corruption.
Hypothesis 2-1 (H2-1). The higher the level of ICT use, the more positive the effect on decreasing government corruption.
Hypothesis 2-2 (H2-2). The higher the level of telecommunication infrastructure (TII), the more positive the effect on decreasing government corruption.
Hypothesis 2-3 (H2-3). The higher the level of online services provided by the government, the more positive the effect on decreasing government corruption.
Hypothesis 2-4 (H2-4). The higher the level of e-government participation, the more positive the effect on decreasing government corruption.
In examining the influence of e-government user status, governance and government regulations on the decrease of government corruption, this study assumes that ICT use and e-government development will exhibit significant mediating effects. Based on Hypothesis 2, the research model shown below (see
Figure 2) was formulated to analyze the mediating effects of ICT use and e-government development between government users’ economic and educational levels, governance and government regulations, and government corruption. Under ICT development, only ICT use was considered, and ICT access and ICT skills were excluded to avoid repetition. Under e-government development, infrastructure was measured using the telecommunication infrastructure index (TII), government transparency using the online service index (OSI), and government accountability using the e-participation index (EPI). In particular, ICT use and e-government development have been identified as both dependent and independent variables contributing to a decrease in government corruption (Kim, T. et al., 2008; Choi, J., 2014; Moon, 2015), and play a mediating role in reducing corruption. The research model below was developed to analyze the mediating effects of ICT use and e-government development, and to test the proposed hypotheses.
3.2. Data Collection and Methodology
The final number of countries included in the analysis was 120 out of 176 after excluding those with missing values. Independent variables and mediating variables were based on 2014 reports by human development indicators, the United Nations Development Programme (UNDP) [
68], the United Nations Statistics Division [
69], World Bank Governance Matters VIII [
70], the Fraser Institute [
71], Reporters without Borders [
72], Freedom House [
73], the UN E-Government Survey [
3], and International Telecommunication Union (ITU)’s Measuring the Information Society 2015 [
74]. Dependent variables were based on 2016 reports by Transparency International (TI) [
67].
For an empirical analysis of government corruption, this study used the corruption perceptions index (CPI) of Transparency International (TI) as a dependent variable. The corruption perceptions index (CPI) reports from 2016 were utilized in consideration of the causal relationship with independent variables (2014) and the time needed for independent relationships to affect dependent variables. The e-government users’ status, adopted as an independent variable, was comprised of education level (education index, human development indicators, UNDP) and economic development (GDP per capita, UN Statistics Division). Governance was divided into government size (Fraser Institute), and voice and accountability (World Bank Governance Matters VIII). Fraser Institute’s 2014 reports were used as a reference for government regulations, comprised of business regulations and credit market regulations.
The mediating variable of ICT use was based on ITU’s Measuring the Information Society 2015. The e-government development index was based on the telecommunication infrastructure index (TII), online service index (OSI), and e-participation index (EPI) in the UN E-Government Survey 2014. The operational definition of each variable is given in
Table 2.
This study applied multiple regression analysis to test Hypothesis 1, that is, to examine the effects of e-government user status, governance and government regulations on corruption decrease. To test the mediating effects of e-government development in Hypothesis 2, three steps proposed by Baron and Kenny (1986) were implemented. The reliability of the results was enhanced through an additional Sobel test, an interval estimation method that supplements the statistical significance of three-stage point estimation.
4. Findings and Results
The descriptive statistics for independent variables, mediating variables and dependent variables are provided in
Table 3. Most variables had an absolute skewness below two and a kurtosis below 10, thus satisfying the assumption of normality.
In Step 1 of Baron and Kenny’s steps for mediation, the regression coefficient must be statistically significant when dependent variables are regressed on independent variables. In Step 2, the regression coefficient must be significant when mediating variables are regressed on independent variables. In Step 3, the regression coefficient must be statistically significant when dependent variables are regressed on mediating and independent variables. The results support full mediation when the relationship between independent and dependent variables are not significant. Partial mediation is observed when the relationship is significant, and the absolute value of the regression coefficient has decreased (Baron and Kenny, 1986).
Step 1 in
Table 4 is the result of testing Hypothesis 1. The proposed model, used to examine the effects of e-government user status, governance, and government regulations on corruption decrease, had an explanatory power (R
2) of 0.85. Economic development, business regulations, and voice and accountability were found to be significant in decreasing government corruption at the 99.9% confidence interval. Government size was significant at 95%, and education level at 90%. Hypotheses 1–3 were partially supported, as credit regulations had no significant influence on corruption. The results partially supported Hypothesis 1, which predicts that government corruption will decrease with e-government user status and governance quality, and greater relaxation of government regulations.
As can be seen from Step 2 of the three-step procedure proposed by Baron and Kenny (1986) in
Table 3, education level and economic development had a significant influence on ICT use at the 99.9% confidence interval, and government size at 90%. The results in Step 3 show the mediating effect of ICT use, which significantly mediated the relationship between the variables of education level, economic development, and government size, and a decrease in government corruption at the 99% confidence interval. More specifically, full mediation was observed between education level and corruption decrease, and partial mediation between economic development and corruption decrease. Under governance, partial mediation was observed between a reduction in government size and corruption decrease. The mediating effect on the relationship between education level and corruption decrease was significant at the 90% confidence interval in Step 1, but not significant in Step 3, thus suggesting full mediation. For e-government user status, the mediating effect between economic development and corruption decrease was significant in all three stages at the 99.9% confidence interval. The unstandardized coefficient for economic development decreased from 0.0003 in Step 1 to 0.0002 in Step 3, which indicates that ICT use partially mediates the relationship between economic development and corruption decrease. Under governance, the mediating effect between government size and corruption decrease was significant at the 95% confidence interval in Step 1, 90% in Step 2, and 95% in Step 3. The decrease in the absolute unstandardized coefficient for government size from −1.4058 in Step 1 to −1.139 in Step 3 shows that ICT use acts as a partial mediator between government size and decrease in government corruption. Since Step 2 revealed that ICT use was not significant in mediating the relationship between the dependent variables of voice and accountability and regulations, and decrease in corruption, it was excluded from the analysis of mediating effects. Thus, Hypothesis 2-1 was partially supported.
As can be seen from the results for Step 2 in
Table 5, education level had a significant effect on the telecommunication infrastructure (TII) at the 99.9% confidence interval, and business regulations at 90%. The results for Step 3 show that the telecommunication infrastructure (TII) had a full mediating effect on the relationship between education level and government corruption, and a partial effect between business regulations and government corruption. More specifically, education level had a significant effect on government corruption in Step 1 at the 90% confidence interval, but was not significant in Step 3, thus suggesting full mediation. Business regulations were significant in Step 1, but the decrease in the unstandardized coefficient in Step 3 demonstrates that the telecommunication infrastructure (TII) acts as a partial mediator between business regulations and corruption decrease. GDP and credit regulations, on the other hand, had no mediating effect on the decrease in government corruption. Thus, Hypothesis 2-2 was partially supported.
The results for Step 2 in
Table 6 show that the online service (OSI) was significantly affected by education level and GDP at the 99.9% confidence interval, by government size at 95%, and by voice and accountability at 90%. In Step 3, the online service (OSI) had a full mediating effect for education level and voice and accountability, and a partial effect for GDP and government size. More specifically, education level had a significant effect on government corruption in Step 1 at the 90% confidence interval, but not in Step 3, thus suggesting full mediation. Both GDP and government size significantly affected government corruption in Step 3, but the decrease in the unstandardized coefficient in Step 3 shows that the online service (OSI) acts as a partial mediator between GDP and governance, and corruption decrease. voice and accountability significantly affected government corruption in Step 1 at the 99.9% confidence interval, but not in Step 3, thus suggesting full mediation. Regulations did not have any mediating effect on decrease in government corruption. Thus, Hypothesis 2-3 was partially supported.
The results for Step 2 in
Table 7 show that education level has a significant effect on e-government participation (EPI) at the 99.9% confidence interval. In Step 3, e-government participation (EPI) acts as a full mediator between education level and corruption decrease.
More specifically, education level had a significant effect on corruption decrease in Step 1 at the 90% confidence interval, but not in Step 3, thus suggesting full mediation. However, GDP, governance, and government regulations had no mediating effect on decrease in government corruption. Thus, Hypothesis 2-4 was partially supported.
A summary of the mediating effects of ICT use and e-government development is presented in
Table 8. ICT use, telecommunication infrastructure (TII), snline service (OSI) and e-participation (EPI) fully mediated the relationship between the education level of e-government users and corruption decrease. ICT use partially mediated the relationship between GDP and government size, and corruption decrease. Telecommunication infrastructure (TII) was a partial mediator between business regulations and decrease in government corruption.
The online service index (OSI) was a partial mediator for GDP and government size, and a full mediator between voice and accountability and corruption decrease.
Lastly,
Table 9 shows the Sobel test results for the mediating effect analysis. Based on the three-step analysis of mediating effects proposed by Baron and Kenny (1986) [
75], this study performed a Sobel test to determine the statistical significance of mediating effects and to obtain confidence intervals. Similar to how the three-step analysis relies on the decreasing absolute value of the regression coefficient for point estimation, the Sobel test performs interval estimation under the assumption that indirect effects follow a normal distribution. The ten pairs of independent and mediating variables observed to have a full or partial mediating effect in Baron and Kenny’s three-step analysis were all found to be significant at the 90% confidence interval (α = 0.1).
First, ICT use showed significant mediating effects on the relationship between e-government user status and governance, and corruption decrease when examined using Baron and Kenny’s three-step analysis, and acted as a mediator for education level, GDP and government size at the 99% confidence interval, according to the Sobel test. Second, the telecommunication infrastructure index (TII) showed significant mediating effects on the relationship between e-government user status and government regulations, and corruption decrease, when examined using Baron and Kenny’s three step analysis. According to the Sobel test, it acted as a mediator between education level and corruption decrease at the 99% confidence interval, and as a partial mediator between business regulations and corruption decrease at the 99% confidence interval. Third, the online service (OSI) showed significant mediating effects on the relationship between e-government user status and governance, and corruption decrease, when examined using Baron and Kenny’s three-step analysis, and acted as a mediator for level of users and governance at the 99.9% confidence interval, according to the Sobel test. Fourth, the e-government participation (EPI) showed significant mediating effects on the relationship between e-government user status and corruption decrease in Baron and Kenny’s three-step analysis, and acted as a mediator between education level and corruption decrease at the 99% confidence interval, according to the Sobel test.
5. Conclusions, Research Limitations, and Implications
The development of ICT, accompanied by administrative reforms and a demand for greater involvement in government affairs by citizens, has accelerated the emergence of e-government systems. E-government has grown more widespread with the expansion of information disclosure and higher demand for transparency in the public sector. Since citizen participation, information disclosure and public transparency are factors related to government corruption, the introduction of e-government was expected to have a positive effect on decreasing corruption. Some countries, however, were associated with low public trust and transparency despite having advanced e-governments.
This study examined the relationship between ICT use, e-government and government corruption based on the aforementioned contradiction. Instead of performing a one-dimensional analysis of the correlation between e-government and government corruption, this study analyzed various countries to determine the effects of ICT and e-government on corruption and identified the causes of these differences. It examined the relationship among e-government user status, governance, regulations and government corruption, and analyzed the mediating effects of e-government development in terms of ICT use, telecommunications infrastructure, and online services.
First, e-government user status, governance and government regulations were found to have a significant effect on decreasing government corruption. The significant factors under e-government user status were education level and GDP, and those under governance were government size, and voice and accountability. Under government regulations, the factor with a significant influence was business regulations. The factors with a full mediating effect on the relationship between education level and corruption decrease were ICT use, telecommunications infrastructure, online services, and e-government participation. ICT use acted as a partial mediator between GDP and corruption decrease, and between government size and corruption decrease. Telecommunications infrastructure partially mediated between business regulations and decrease in government corruption. Online services partially mediated between GDP and governance (government size, voice and accountability), and corruption decrease. In the analysis of factors affecting decrease in government corruption and the mediating effects of ICT use and e-government development, credit regulations did not show any statistical significance.
The political implications of this study are as follows. ICT use and e-government development were found to be mediators of e-government user status, government size, voice and accountability, and market regulations. As such, the results can serve as a valuable reference in the establishment of effective anti-corruption strategies. The role of the government is emphasized in most of the results. Since education level fully mediates the e-government’s ICT use, telecommunications infrastructure, online services, and e-government participation, the level of education must be enhanced to decrease government corruption. Given that online services act as a direct mediator for voice and accountability, we can expect greater involvement of citizens and media in government affairs to have a positive effect on online service accessibility and information acquisition, thereby contributing to a decrease in government corruption. That is, government corruption can be reduced with citizens participating more actively in government affairs and with the media serving as a watchdog over the government. This will help to enhance the level of e-government development, and ultimately reduce corruption in the government. The government must relax market and business regulations in order to lower corruption. Major countries around the world are relaxing their regulations to keep up with rapid changes, including the advent of the fourth industrial revolution, and to remove obstacles in the development of new technology and the pioneering of new markets. More flexible regulations should be introduced for ICT-related private enterprises and markets. The expansion of ICT-related infrastructure will allow various entities to enjoy access to vast sources of information, thus reducing government corruption.
This study performed a cross-country analysis on the relationship among e-government user status, governance, regulations and government corruption, with e-government as the mediating variable in the contradictory relationship between e-government development and government corruption. Rather than adopting a typical one-dimensional approach, the factors affecting corruption decrease and their mediating effects were analyzed. However, this study is limited in that it did not explore the fundamental causes behind the contradictory relationship between e-government development and government corruption. The significance of this study lies in providing necessary information for the establishment of anti-corruption strategies and policies and serve as a basis for further investigation. Follow-up studies will be conducted to trace the fundamental causes behind the high level of corruption in countries with advanced e-governments, and the above results can be utilized to derive political factors and additional indicators.