4.3. Empirical Analysis
First, a least square regression was carried out to test the relationship between cross-border e-commerce and international trade.
As shown in Table 2
, the growth rate of cross-border e-commerce is proportional to the total amount of international trade, and the coefficient is significant. When the growth rate of cross-border e-commerce increased 1%, total volume of international trade would increase by 0.678%.
This result shows that China’s cross-border e-commerce has a significant promotional effect on the growth of international trade. China’s cross-border e-commerce reduces traditional trade transaction costs. The reason why tax cost is not negatively affected is possibly due to China’s effective tariff reduction policy for cross-border e-commerce. The gradual implementation of the “the belt and road” initiative and the “China–EU trains” and the establishment of the “China–Russia economic corridor” significantly reduce the trade transportation cost of inland provinces of China. With the help of improvements in logistics, the positive effect of China’s cross-border e-commerce would offset the negative effect of transport cost.
Moreover, the coefficient of t is −0.364, meaning that it is negative and significant, indicating changes in the amount of international trade that cannot be explained by the independent variables listed above. The negative effect increases year by year. This may have been largely due to the impact of the market shrinking after the global financial crisis in 2008.
The absolute value of the coefficient of logpec (0.678) is only greater than the absolute value of the coefficient of t if it was in 2011 (−0.364), which means that the positive impact of cross-border e-commerce cannot sufficiently offset the negative impact of the financial crisis. Thus, cross-border e-commerce has not really become a new source of growth for international trade.
In recent years, promotional policies for cross-border e-commerce by government agencies such as the State Council, the Ministry of Commerce and the Ministry of Finance have been enacted to stimulate international trade. In our sample, we found 230 favorable policies in e-commerce and trade from every provincial government website. 60% of policies have been issued by the provinces along the land and maritime Silk Roads, 40% of policies have been issued by other provinces. On the basis of Model 1, we introduce an interaction item of cross-border e-commerce and time trend to test whether the coefficient of cross-border e-commerce has significant changes across different years. According to the regression results of Model 2 shown in Table 2
, the coefficient of the interactive item is positive, which is 0.0373, meaning that it is very small and insignificant. The result shows that, although the impact of cross-border e-commerce on international trade each year was positive, the impact does not change significantly year by year. This may imply that the intended effects of the policies did not occur due to ineffective implementation or regional disparity in trade infrastructure.
In short, the regression results of Models 1 and 2 verify Proposition 1: the growth of cross-border e-commerce enlarges the scale of international trade. However, they do not support Proposition 2: the promotional effect of China’s cross-border e-commerce on the scale of international trade will show an upward trend year by year.
In order to test whether the impact of cross-border e-commerce on international trade is different across different regions, we divided China’s 31 provinces into “the belt and road” and “non-belt and road” provinces. “The belt and road province” implies a province along the land and maritime Silk Roads, “non-belt and road province” implies a province not along the land and maritime Silk Roads. Due to the “the belt and road” initiative, favorable policies are leaning to some provinces that lie along the land and maritime Silk Roads. The impact of cross-border e-commerce on international trade in these provinces may differ from other regions. The regression results of “the belt and road” provinces are presented in Models 1 and 2 in Table 3
. Regression results of “non-belt and road” provinces are present in Models 3 and 4. The impact of cross-border e-commerce on international trade is still significant and positive in these models.
Comparing the results of Model 1 in Table 2
and Table 3
, the coefficient of growth rate of cross-border e-commerce in the “the belt and road” provinces is 0.897, which is 32% higher than the coefficient of all 31 provinces (0.678, the result in Table 2
), while the impact of cross-border e-commerce on international trade in “non-belt and road” provinces (0.536) is 31% lower than the coefficient of all 31 provinces (0.678).
According to the results of Model 2 shown in Table 3
, interaction item is positive but insignificant, which is 0.129. Meanwhile, the coefficient of the interaction item of the “non-belt and road” provinces is −0.0431. These imply that the impact of cross-border e-commerce in “the belt and road” or “non-belt and road” provinces also do not change year by year. Although both are insignificant, one is positive and one is negative.
These results demonstrate that cross-border e-commerce in “the belt and road” provinces has played a more significant role in the growth of international trade. The reason why this phenomenon appears is that transaction costs of international trade in the “the belt and road” provinces are much higher than those in other provinces. In order to match the Chinese government’s policy goals about the “the belt and road” vision, these provinces have made a lot of improvement in facilitating cross-border e-commerce. Therefore, cross-border e-commerce contributes to the reduction of transaction costs in these provinces, and plays a stronger role in promoting international trade.
The increase of international trade may improve the GDP per capita in some provinces and not in others. This implies that endogeneity may be caused by simultaneity [36
]. In order to solve this problem, we choose GDP per capita of one year and two years before the given year as an instrumental variable for “pgdp” in a generalized method of moments (GMM) regression analysis. These instruments pass the overidentification test (p
= 0.42). The results are shown in Table 4
Comparing Table 2
with Table 4
, we find the results are consistent. The results of the “the belt and road” provinces and “non-belt and Rroad” provinces also do not have substantial differences. The detailed results of GMM estimation are not reported here due to limited space.