The Productivity Advantage In China’s Export Enterprises And The Source Identification Of The Advantage

China’s reform and opening up policy has made great achievements, and exports have become one of the engines driving China’s economic development. In the course of exports, the country has created a number of advanced technology industries and enterprises, but there are also some companies influenced by the international market fluctuations. Under the influence of heterogeneous enterprise trade theory currently, the domestic and foreign scholars used different data and measurement methods to study the self-selection effects and export learning effect in export process and got different conclusions. This article stands on a new perspective to study the issue of export and productivity and attempt to solve the flaws in formal theoretical and empirical research.Firstly this paper integrates the learning effect of export into the analytical framework under heterogeneous enterprise trade theory and makes the self-selection effects and export learning effects into a single theoretical model. This paper also points out that self-selection effect mainly plays a role in the low quantile of productivity distribution, leading to the left tails phenomenon, while the exports learning effect moves the overall productivity distribution to the right. At last two effects on productivity are shown in a graphics representation.Next this article uses propensity score matching methods to overcome the sample selection bias and endogeneity problem, measuring out that export can increase productivity for by 0.231 units during the study sample time. Regression results by year show that such promotion will get weak as year pass by and even have a negative impact in the late.Then the pattern recognition method result shows that left tails phenomenon exists in the distribution of export, which confirmed the existence of self-selection effects, while there is no right movement in the distribution, which means that export learning effect does not exist in the latter study time. In addition numerical comparison of the level of sub-sites have come to the same conclusion.The following quantile regression robustness tests, controlling for other factors, still proves that self-selection effects exist but there is no learning effect of exports. The article also study the reasons for non-existence of learning by exporting effect. the processing trade and labor-intensive enterprises lead to the existence of this phenomenon.Finally, a number of policy recommendations are given according to the empirical results.