Abstract: On the basis of the comparisons the CSSW, CSSG, BC and KSS stochastic frontier analysis estimators, this paper analysis regional technical efficiency in China. Besides, we explain the trend differences among regions technical efficiency from the perspective of determinants. The main findings are: (1) Granger causality test, Hausman-Wu test and normality test of random error term show that the model itself has the endogenous problem on the basis of Cobb-Douglas production function, the results of previous research based on BC are untrusted. Considering the superiority in the technical inefficiency and endogeneity, the KSS estimator is the relatively most reliable. (2) Global and regional average technical efficiencies present fluctuant character and different from the previous conclusions significantly. Prior to 2008, the technical efficiency of eastern region is the highest, followed by central and western regions, while the central region catch up with the eastern and the gap is going to widening after 2008. (3) Geographical location, government expenditure on technology development, the scale of high technical industry and education level of population are the key factors impacting currently.

摘要： 本文对CSSW、CSSG、BC及KSS四种技术效率测算模型进行了比较分析，并基于此对我国区域1997-2010年间技术效率水平进行了再估计，进一步重点研究了区域技术效率的区位因素。研究发现：（1）格兰杰因果检验、Hausman-Wu检验及随机误差项正态分布检验结果共同显示，利用对数型柯布-道格拉斯生产函数对区域技术效率进行测算时，模型本身存在内生性，前人采用BC模型得到的结果可信度低下；考虑到在内生性及技术非效率项处理上的完备性，认为KSS模型测算结果相对更加可信。（2）无论是整体还是地区，平均技术效率均呈现涨跌互动的波动趋势，与前人得出的单调递增（减）结论差异显著；2008年之前，东部地区平均技术效率最高，中部次之，西部最低，2008年之后中部地区平均技术效率赶上东部，且两者之间差距呈现扩大趋势。（3）地理位置、财政科技投入、高科技产业规模和人口素质是当前导致区域技术效率差异显著的关键因素。

关键词: 随机前沿分析, 技术效率, 内生性, 区位因素, stochastic frontier analysis, technical efficiency, endogeneity, determinants