2.2 Model construction
The relationship between human capital，material capital and output can be expressed by the production function：
Consider the heterogeneity of human capital：
So：
Natural logarithm of both sides of the equation：
The regression method is used to reduce the dimension of the input index in 2013，analysis of regional human capital efficiency by using data envelopment analysis.This parametric method（regression analysis）and non-parametric method（data envelopment analysis）can be combined with the method，which can eliminate the irrelevant factors from the input index system，improve the pertinence of the data envelopment analysis method.
First，do a logarithmic processing to the original data.Then do regression analysis on the data of 2013 with the stepwise regression method.Data process with software Eviews 7.2.Regression results in function 5.The variables are：CZ（junior middle school education），GZ（senior middle school education），DZ（college degree or above），GD（fixed assets）.
Note：＊＊＊，＊＊，＊were expressed respectively in 1%，5%and 10% levels.Inner brackets are t statistics.The regression results are shown in Table 2.
Table 2 Stepwise regression results
According to the results of stepwise regression，the variables that are better explained GDP are fixed assets，junior middle school education human capital，senior middle school education human capital，college degree and above education human capital.The primary school education human capital is excluded from the model.This conclusion is similar to that of Liu Zhiyong et al.（2008）.