同时考虑环境因素和统计噪声的四阶段SFA模型——在我国省际技术效率和政府管理绩效中的应用

为克服三阶段DEA模型忽略了投入-产出之间统计噪声对技术效率的影响这一不足，将环境因素细分为员工工作环境和管理环境两类，提出了一种基于随机前沿分析（stochastic frontier analysis）技术的四阶段模型. 通过对比剔除管理环境因素前后的技术效率，得到动态经营管理效率（management efficiency）. 进一步，利用四阶段SFA模型对2002-2010 年间我国省际技术效率和政府管理绩效进行了测算. 研究发现：（1） 我国整体、三大地区及各省区综合技术效率均呈现“U型”波动趋势，且东部地区最高，中部次之，西部最低；而实际技术效率呈现单调递增的变动趋势，中部最高，东部次之，西部最低. （2） 政策扶持对于地区技术效率的提升具有重大的推动作用，贡献率高达71.4%；政府管理绩效呈现“U型”波动趋势（江西，湖北，海南除外），且东部地区高出中西部约8.9% 和3.2%. 作为技术效率和经营管理绩效实证研究的技术基础，四阶段SFA 模型为优化资源配置提供了很好参考.

The four-stage SFA model accounting for environmental factors and statistical noises：An application for estimating Chinese provincial technical efficiency and government performance

For overcoming the drawback that the three-stage DEA model neglects the input-output statistical noise, we put forward a new four-stage SFA model by dividing environmental factors into employee working and management environment. By comparing the technical efficiencies before and after removing the management environmental effects, we can get the corresponding dynamic management efficiency. Further, we use it to evaluate each provincial technical efficiency and government performance covering the period 2002-2010 in China. Results show that: (1) The overall, regional and provincial comprehensive technical efficiencies present the “U-Style" trend, and the east area is the highest, which is followed by the central and west areas. In contrast, the actual technical efficiencies keep increasing, and the central area is the highest, followed by the east and west areas. (2) Government policies play an important role in promoting local technical efficiency with the contribution rate up to 71.4 percent; most provincial management efficiencies present the “U-Style" trend (except Jiangxi, Hubei and Hainan). The east area's mean management efficiency is higher than the central area about 8.9 percent and the west area about 3.2 percent. As the technical basis of empirical research on estimating the technical efficiency and management efficiency, the four-stage SFA provides a good reference for optimizing resource allocation.