The focus of this study was to identify the impact of government policy, legal environment, socio-cultural framework, Gross Domestic Product (GDP), Internet diffusion and technology transfer on the national adoption of eCommerce in developing countries. A cross country analysis, using data sets that were based on a quantitative survey was used in this study. The study concluded that socio-cultural factors have the most influence on e-Commerce adoption. Technology transfer and Internet diffusion provided some level of mediation which improved the significance level of other variables. The results also demonstrate that the information and computer literacy problem can only be addressed by enhancing various organizing forms in order to improve eCommerce diffusion. The mediating role of the organizing forms is critical in galvanizing local community participation, for instance in capturing and accumulating cultural knowledge as a basis of developing relevant content for enhancing community participation in e-Commerce. These findings can advance policy implementation, particularly in integrated development planning in developing countries that continue to suffer from a lack of adequate knowledge regarding the digital commerce environment. Further, as the Big Data & Analytics finds traction in the e-commerce environment, the trajectory of ecommerce adoption is likely to change dramatically in developing countries.
1.1
Multiple studies of e-Commerce (e-commerce) adoption show that adoption in
developed countries is higher than in developing countries [
According to Kumar, Kumar, Dutta and Fantazy [
.H11: Government policy influences adoption of e-commerce in a country H10: Government policy does not influence adoption of e-commerce in a country H21: Legal environment influences adoption of e-commerce in a country H20: Legal environment does not influence adoption of e-commerce in a country H31: Socio-cultural factors influence adoption of e-commerce in a country H30: Socio-cultural factors do not influence adoption of e-commerce in a country H41: Internet diffusion influences adoption of e-commerce in a country H4: Internet diffusion does not influence adoption of e-commerce in a country H51 Technology transfer influences adoption of e-commerce in a country H50: Technology transfer does not influence adoption of e-commerce in a country
A positivist quantitative survey approach relying on datasets was adopted in this study. The sources of secondary data are e-commerce and IT reports from Economist Intelligence Unit (EIU), World Bank, World Economic Forum (WEF) and United Nations Conference on Trade and Development (UNCTAD), Global Information Technology Report (GITR) that provide current and historical global data on ICT related developments and the Electoral Institute for Sustainable Democracy in Africa (EISA) and other miscellaneous sources. This study focused on the SADC region particularly because its member countries are all under the developing countries’ category and most of them share common characteristics. 2.2
The dependent variable (DV) is e-commerce adoption, while the socio-cultural framework, legal and policy environments are the Independent Variables (IV), and Internet Diffusion as well as Technology Transfer are mediating variables (MV) and GDP is the control variable. Data for DV is gathered from the GITR (2012), while all other variable are taken from the World Bank online database. The data is a mixture of ratios (e.g., Literacy Rate, Unemployment and Poverty Ratio) which are all part of the socio-cultural component as well as Internet Penetration. Government Policy and Legal Environment values are also ratios and E-Commerce Adoption and Technology Transfer are rank values from a Likert scale of 1 to 7 with 1 being the lowest and 7 being highest as per the GITR description (2012, p171). The measurement items for the study are found in Table 1.
Measurement Items Government Policy
The overall objective of the analysis in this section is to explain the degree and nature of the relationship between E-Commerce Adoption (DV) and the independent and mediating variables relating to Government Policy, Legal Environment, Social and Cultural Infrastructure, Internet Diffusion, Technology Transfer and GDP. 3.2
According to the results presented in Table 2, two variables are missing a score. Multiple Regression uses only the 13 variables with complete data (e.g., Angola, Botswana, Lesotho, Malawi, Madagascar, Mauritius, Mozambique, Namibia, South Africa, Swaziland, Tanzania, Zambia and Zimbabwe). Data for Business Usage and Infrastructure plus Digital Content was missing for the Democratic Republic of Congo and Seychelles. Population had a high mean and the frequency graph shows that GDP is skewed to the left showing a large gap in GDP among the sample under review. Data from the World Bank (2010) in Millions of US Dollars shows that South Africa had a highest GDP at 276.8, Angola at 84.937, Tanzania at 22.915, Botswana at 14.838, DRC at 13.144, Zambia at 14.314, Mozambique at 9.735, Mauritius at 8.651, Namibia at 8.563, Malawi at 4.269, Swaziland at 2.618, Lesotho at 2.179, Seychelles at 0.835 and Zimbabwe at 0.641.
The skewness above 1.5(+ or -) are considered to abnormally distributed [
This study followed a four-step analysis process. Step 1 involved a Multiple Regression Analysis (MANOVA) with all variables excluding mediating variables (i.e., Internet Diffusion, Technology Transfer) in order to test for a direct relationship with ECommerce Adoption. Step 2 is using a MANOVA with all variables predicting the mediation caused by Internet Diffusion and Technology Transfer. Step 3 is using a MANOVA with Internet Diffusion, then Technology Transfer predicting ECommerce Adoption. Step 4 is using a MANOVA with all variables including mediating variables predicting E-Commerce Adoption (Table 3). 3.4
The results indicate that in Step 1, all variables excluding the mediators significantly predict E-Commerce Adoption at a model fit of 91%. There is no significant relationship in Step 2 for both Technology Transfer and Internet Diffusion. However, there is a strong relationship between Technology Transfer and E-Commerce Adoption and very little between Internet Diffusion and E-Commerce Adoption. The relationship between all variables including mediators is significant at a p-value < 0.1 with 72% explanation for the variance in E-Commerce Adoption. The observation is that some form of mediation is supported as shown by a significant value in Step 4. Adjusted R2 (%) 91% 69% 36% 72% 8% 72% Step
IV 1 2 3 4 Technology Transfer Internet Diffusion GDP, Unemployment, Government Effectiveness, Rule of Law, Regulatory Quality, Literacy Rate, Poverty Ratio, Government Usage, Individual Usage, Political Stability, Technology Transfer And Internet Diffusion
DV E-Commerce tion
AdopTechnology Transfer Internet Diffusion E-Com. Adoption E-Com. Adoption E-Commerce tion
Adop
Sig 0.070 0.233 0.722 0.000 0.181 0.126
The rest of the analysis will continue based on the results in Step 4. GDP together
with Internet Penetration violated the kurtosis rule. Mean and standard deviation
satisfied the rules at the mean of 0 and standard deviation of 1 for all variables (Table 4).
The assumptions for using a MANOVA include the following; that the relationship
between each of the predictor variables and the DV is linear and that the error, or
residual, is normally distributed and uncorrelated with the predictors [
Since multiple regression models assume linearity, the linear association between each IV and DV was tested using an ANOVA test of linearity (Table 5). The F statistic is 4.360 and significance value for the nonlinear components was above the critical value (i.e., p < 0.1). The results showed that all IVs do not have a linear relationship with the DV of Business Usage. The significance value of .126 indicates that these variables can be used to explain the DV given that the study is using secondary data, the population size is small and that there could be some other factors that could play a role in explaining E-Commerce Adoption but are not included in this model. Therefore, at the significance level of .126, the model can be accepted, meaning the predictors can marginally explain E-Commerce Adoption. Given that:
Business Usage = function (GDP, Internet Penetration, Literacy Rate, Unemployment, Poverty Ratio, Government Effectiveness, Political Stability, Regulatory Quality) Or
Business Usage = β1 + β2 x GDP + β3 x Internet Penetration+ β4 x Literacy Rate + β5 x Unemployment + β6 x Poverty Ratio + β7 x Government Effectiveness + β8 x Political Stability + β9 x Regulatory Quality + β10 x + c.
Assuming that F is testing the hypothesis that β1= β2= β3= β4= β5= β6= β7= β8= β9= β10, since F is significant, the use of the IV has assisted us in explaining the DV (Business Usage). Total Sum of Squares (TSS) is 12.000. This shows the deviations in the DV. Regression shows the Explained Sum of Squares (ESS) of 11.148, which is the amount of TSS that could be explained by the model. The Residual Sum of Squares (RSS) is 0.852, which is the amount that could not be explained by the model. RSS represents unexplained variation and a smaller RSS means that the model fits the data well.
The results for correlation analysis (Table 6) show that Literacy Rate (LR) is highly correlated with Technology Transfer (TT) with a coefficient of 0.701 at a significance level of .008; which means as people get more educated; they are likely to adopt new technologies. Poverty Ratio (PR) is highly correlated with Government Effectiveness (GE) with -0.793 at a significance level of .001, and Political Stability (PS) with 0.576 at a significance level of .039. Poverty is also highly correlated with Regulatory Quality (RQ) with -0.709 at a significance level of .007. These results show that poverty is more prevalent in countries affected by corruption and lack of proper gover nance. There will be less poverty in a more stable government since there will likely be fewer wars, political strife and other factors that consume resources that could otherwise be channeled to poverty eradication. Rule of Law can be the foundation for proper governance, which in turn will lead to overall institutional stability hence a sound business environment.
Poverty Ratio is highly correlated with Technology Transfer, with a weighting of 0.734 at a significance level of .004. Government Effectiveness is highly correlated with Political Stability (0.711); Regulatory Quality (0.961); Business Usage (-0.612) and Technology Transfer (-0.734). Political Stability is highly correlated with Regulatory Quality (0.705). Political Stability means that the government will be able to set clear and effective regulations that enable the general business environment. There will be sound and strong governance structures that will enable a conducive business environment. Regulatory Quality is highly correlated with Business Usage (0.675) and Technology Transfer (0.724). Thus, sound regulations will enhance the business environment, protect consumer rights and therefore enable people to conduct business freely whether online or not. Finally, Business Usage is highly significant to Government Usage (0.624) and Technology Transfer (0.860). As individuals increasingly take up and use technology, businesses will more likely grow due to the increase in the demand for goods and services. The improvement in the business environment will enable more research and development in a country; hence the increase in technology development will drive demand for advanced technologies. As business improve and advance in technologies, so will the government as a partner in business.
The results of the multiple regressions are presented in Table 7. The R2 measures the proportion of the variation in the DV (Business Usage) that was explained by the variations in the IV (Poverty Rate, Government Effectiveness, Regulatory Quality, Political Stability, Technology Transfer, Internet Penetration). Thus, 93% of the variation from the mean in the DV was explained. The adjusted R2 is the measure of the proportion of the variance in the DV (Business Usage) that was explained by the variations in the IVs. The standard error of estimate is 0.53, which is far less than 10% and therefore the dispersion of the DVs around the mean is low.
The model summary table shows that the multiple coefficient (R), using all the predictors simultaneously, is .964 (R2=.929) and adjusted R2 is .716 meaning that 72% of the variance in the DV (Business Usage or E-Commerce Adoption) can be predicted and explained by the IVs (Government Effectiveness, Literacy Rate, GDP, Internet Penetration (Technology Transfer) Unemployment, Regulatory Quality, Poverty Ratio and Political Stability). The higher value may be attributed to the decrease in the number of IVs. Therefore a conclusion was reached that the model is a good fit.
The Tables of Coefficients (Table 8) show the standardized beta coefficients, interpreted similarly to correlation coefficients. The t value and the sig opposite each IV indicates whether that variable is significantly contributing to the equation for predicting E-Commerce Adoption from the whole set of predictors. GDP, Literacy Rate, Unemployment and Poverty Rate are significantly contributing to the equation. However, all the variables need to be included in this result, since the F value was computed with all the variables in the equation. GDP explains the most variance at 84% (.9182), followed by at 80%, Unemployment (70%) and Literacy Rate (54%). Political Stability (38%), Regulatory Quality (26%) and finally Internet Penetration (14%) show the lowest amount of unique variance. Therefore, GDP, Literacy Rate, Unemployment, Poverty Ration explain E-Commerce Adoption more than other variables. Tolerance and VIF tell us if there is multi-collinearity problem. If tolerance is low (<1-R2), then there is a multi-collinearity problem. In this case, adjusted R2 is .842 and 1-.8422 is .29. Unemployment, Poverty Rate, Government Effectiveness and Regulatory Quality have tolerance levels lower than (1-R2), hence shows a multicollinearity problem. However, as previous stated, these variables may be important in explaining the questions this research is answering.. This section gives the analysis of results based on Table 8.
H11: Government policy influences adoption of e-commerce in a country H10: Government policy does not influence adoption of e-commerce in a country
Literature supports the view that government stimulates technology adoption (Shin, 2007; Cordeiro & Al-Hawamdeh, 2001). However, despite literature support for government influence on e-commerce adoption, our analysis reveals a weak link. In our study, government policy was represented by Government Effectiveness and Political Stability. The p-values are .277 and .352, while the beta values are -3.850 and -.811 for Government Effectiveness and Political Stability respectively. All p-values are > 0.1 which shows that there is no relationship between Government Policy and ECommerce Adoption in the context of this study. However, there is some evidence that given some level of Technology Transfer and Internet Diffusion, Government Policy can influence adoption of e-commerce in a country. Therefore, the null hypothesis H11 is rejected and the alternative hypothesis H10 (government policy does not influence adoption of e-commerce adoption in a country) is accepted. In in line with the findings from the literature, it may be argued that the alternative hypothesis maybe accepted, taking into account sample size limitations. However, the correlation analysis revealed a high correlation between Government Effectiveness variables and E-Commerce Adoption at acceptable significance levels.
H21: Legal framework influences adoption of e-commerce in a country H20: Legal framework does not influence adoption of e-commerce in a country
Regulatory Quality is the proxy for legal framework. The beta coefficient is .304 and the p-value is .797. However, when Technology Transfer and Internet Diffusion are incorporated, the beta coefficient increases to .904 and the significance level increases to .296. Again, the p-value is greater than .01 but there is an enormous difference caused by Technology Transfer and Internet Diffusion, hence we reject the null hypothesis H21 and accept the hypothesis H20, (legal framework does not have an influence on adoption of e-commerce in a country). Zhu and Thatcher (2010) show that the legal environment should reduce uncertainty by providing adequate, clear and efficient frameworks for economic exchange. If a government shows a clear commitment to e-commerce, this becomes apparent in its policy measures, which in turn can encourage e-commerce utilization (Molla & Licker 2005).
H31: Socio-cultural factors influence adoption of e-commerce in a country H30: Socio-cultural factors do not influence adoption of e-commerce in a country
From the coefficients table, Socio-Cultural Factors as represented by Literacy Rate, Unemployment and Poverty Ratio have beta values of .382, .710 and 1.396, with pvalues of .221, .040 and .022 respectively. Unemployment and Poverty Ratios show a strong positive relationship with E-Commerce Adoption. Introducing Technology Transfer and Internet Diffusion, beta values are 113, .503 and 1.075 for Literacy Rate, Unemployment and Poverty Ratio respectively, with significant values of .822, .347 and .351 respectively. This resulted in a significant decrease in the beta and p-values. This means that as the Unemployment and Poverty Ratio decrease in a country, the level of E-Commerce Adoption increases but bringing in Technology Transfer and Internet Diffusion suppresses the spread and use of e-commerce possibly due to the anticipated costs brought by new infrastructure that supports new technologies.
The results show that all the Socio-Cultural Factors may have an impact on ECommerce Adoption. Therefore, we reject the null hypothesis, H30 and accept H31 (socio-cultural factors contribute positively to the adoption of e-commerce in a country). The level of influence is quite high looking at the p-value of the significant variables. Several socio-cultural factors have been identified as barriers to e-commerce adoption in developing countries.
H41: Internet diffusion influences adoption of e-commerce in a country H40: Internet diffusion does not influence adoption of e-commerce in a country
The results in the regression analysis show that Internet Diffusion has a
significance level of .181 on e-commerce with the adjusted R2 of .080 and the beta value of
.396. Using the significance level of .10, H41 is rejected thereby inferring that at the
conceptual level, there is no relationship between Internet Diffusion and E-Commerce
Adoption. Given the small sample size and data issues, this result can be questioned.
Literature shows that there is a positive relationship between Internet Diffusion and
E-Commerce Adoption as an artefact of technology [
H51: Technology transfer influences adoption of e-commerce adoption in a country H50: Technology transfer does not influence adoption of e-commerce adoption in a country
Technology Transfer, tested in isolation from the other variables, was found to
have the most significance on E-Commerce Adoption. This was shown by the p-value
of .000 and the beta value of 5.581. However, when combined with all other predictor
variables, the p-value decreases to .126. Based on the literature, Technology Transfer
is believed to have an influence on E-Commerce Adoption in a country [
A model for e-commerce adoption incorporating Internet diffusion and technology transfer as mediating factors was developed. All variables were tested together to determine their impact on e-commerce adoption. The study found that GDP, sociocultural factors, technology transfer had a significant impact on e-commerce adoption. This study revealed that government policy, legal environment, socio-economic framework, Internet diffusion, technology transfer could affect the adoption of ecommerce in a developing country. Given the context of Sub-Saharan Africa, particularly the SADC region which has for many years been affected by political instability, poverty, corruption, riots, poor infrastructure, unemployment and high illiteracy rates, the findings reveal that all variables could contribute to e-commerce adoption to some extent as shown by a strong contribution by socio-cultural factors and GDP. This can be validated by the fact that more studies revealed the opposite in developed countries, which are affected less by the conditions prevailing in the developing world.
The proposed model has identified the link between national ecology factors (particularly the social-cultural construct), Internet diffusion and technology transfer with e-commerce adoption. It has shown that socio-cultural issues contribute more than any other factors to the adoption of e-commerce. A somewhat small effect was realised from technology transfer and Internet diffusion. More research is needed to identify other factors that could have an impact on e-commerce adoption. The analysis confirmed that the socio-cultural infrastructure is a critical factor in enabling ecommerce adoption, when technology transfer influence is taken into account. The socio-cultural context emphasizes the need to build the Social Capital or Social Resources required for stimulating e-commerce adoption.