| Abstract | As of counterfactuals missing, problems such as sample selection bias and individual heterogeneity arise when evaluating schooling return. The traditional OLS is biased and IV is inefficient when the return is heterogeneous. Based on a general micro-effect evaluation framework of economic policy considering heterogeneity, this paper gives an empirical analysis on the trend of china's higher education return since 1990s using three cross-sectional datasets (year 1992, 2000 and 2009) by semi-parametric estimation. The annualized schooling return rate is 5.5%, 9.9% and 11.4% for year 1992, 2000 and 2009 respectively, which shows an increasing trend of schooling return and proves the positive effects of China's labor market reform, but the amplification in the second decade is smaller than that in the first decade. Significant evidence of heterogeneity is found, and the annualized TT is 6.0%, 11.1% and 11.9%, while the annualized TUT is 5.1%, 8.4% and 10.2% for the three years. For all three cross-sectional datasets, we get i.e. schooling return of the college is higher than the counterfactual return of the high school if the individual selected college. Sorting gain is positive and selection bias is negative, which proves the individual selected education based on comparative advantage. The traditional OLS underestimates schooling return while IV overestimates the return. |