空间面板随机前沿模型及技术效率估计

Spatial Panel Stochastic Frontier Model and Technical Efficiency Estimation

随机前沿模型是测算技术效率的重要方法之一。通常,模型假设生产单元之间彼此独立,然而在技术扩散过程中,空间外部性起着重要作用。文章结合随机前沿模型理论与空间经济计量分析方法,构建空间面板随机前沿模型,同时考虑空间滞后因变量和空间误差自相关,并逐步放松模型设定条件,首先考虑技术效率时变,接着引入技术无效率项的异方差性,之后考虑观察数据中潜在的截面异质性,分别以引入随机截面特有项和设定随机系数的形式来表示截面异质性。针对各种模型设定提出相应的参数估计方法,最后给出技术效率的估计。

Stochastic frontier model is one of important methods for measuring technical efficiency.Generally,the model assumes that production units are independent of each other.However,spatial externalities play an important role in diffusing technical progress and points in space are co-related.By combining stochastic frontier production theory with spatial econometric analysis,we propose a spatial panel stochastic frontier model in which both of spatially lagged dependent variables and spatially autoregressive error terms are considered,at the same time,we relax the model assumptions that seem to be strong ones.Firstly,time-varying technical efficiency model is analyzed.Secondly,the consequence of heteroskedasticity in the technical inefficiency is investigated.Then,we consider some extensions that account for the latent cross unit heterogeneity.Explicit models for unmeasured cross unit heterogeneity are built into the stochastic frontier.Estimation methods for various kinds of reformulated models are presented.Finally,we focus on the estimation and comparison of technical efficiency measures.