中文摘要：

本文引入半参数和完全非参数随机前沿模型，克服模型设定误差，利用局部线性估计方法，由1978—2012年间中国30个省市区的面板数据，对改革开放以来我国粮食生产随机前沿面和技术效率进行半参数和非参数模型估计和测算。估计结果表明，时间和省区效应非参数模型估计的拟合优度高，不合其它模型（参数和半参数模型）估计中出现的伪非效率现象，可以较全面地反映中国粮食生产随机前沿面非线性特征和技术效率的时变特征。非参数模型的估计结果显示，劳动力和农业机械动力的平均产出弹性为负，但不显著；播种面积和化肥使用量的平均产出弹性显著为正。技术效率影响因素Tobit模型的估计分析表明，我国粮食生产技术效率的时间效应关于连续型影响因素变量具有显著的权变效应，这些因素对技术效率的边际影响也具有显著的时间效应。

英文摘要：

In this paper, in order to overcome the model specification bias, we introduce semiparametric and fully nonparametric stochastic frontier models for the estimation of the food stochastic production frontier of China＇ s food production after the economic refoms in China by using the panel data of the 30 provinces during 1978- 2012. The locally linear estimation method is applied in the study. The test results show that the nonparametric model with time variation has a higher goodness -of- fit with no spurious inefficiencies which appear in the parametric and semiparametric alternative models, and hence it is more suitable for the study of the time -variant stochastic frontier and technical efficiency （TE） in China＇ s food production. The estimation results show that the average output elasticity of labor and total power of agricultural machinery are negative, though insignificantly. While the average output elasticity of some areas and chemical fertilizer are significantly positive. The factor analysis on the TE based on the estimation of the Tobit model shows that the time effect of TE in China＇ s food production is contingent on the continuous factors which affects TE. The effects on TE of those factors are also significantly time - variant.