项目名称: 空间随机前沿模型估计及设定检验研究
项目编号: No.71301088
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 管理科学
项目作者: 张进峰
作者单位: 山东大学
项目金额: 20.5万元
中文摘要: 随机前沿模型在经济管理科学的各个领域都有广泛的应用,而截面之间的空间相关性问题也在理论和应用研究中日益受到重视。本项目主要探讨包含空间相关性的随机前沿模型估计与检验,着重研究以下四方面的内容:(1)分别采用修正两阶段最小二乘法、广义矩法和半参数估计方法估计空间滞后随机前沿模型;(2)以广义矩估计结果为基础,构建空间滞后随机前沿模型中用来检验空间效应的C(a)检验统计量;(3)将空间滞后随机前沿模型扩展到空间面板随机前沿模型,研究随机效应空间面板随机前沿模型的广义矩估计,该估计量能有效克服极大似然估计遇到的繁琐计算问题;(4)构建空间效应参数存在局部误设时对空间面板随机前沿模型个体效应的稳健检验,该类检验兼顾了模型估计中的繁琐计算问题和检验的有效性。以外,本项目将通过大量蒙特卡罗模拟分析和比较不同估计量和检验统计量的有限样本性质。
中文关键词: 随机前沿;设定检验;半参数;空间面板;极大似然
英文摘要: Stochastic frontier models have been widely employed in many fields of economics and management science.On the other hand, the issue of spatial dependence among cross sections has received heated attentions both in theoretical and empirical literatures. In this project, we study the estimation and testing of a class of spatial stochastic frontier models which includes four emphysis:(1) propose estimation methods for spatial lag stochastic frontier model by using corrected two stage least square estimation, generalized method of moments and semi parametric estimation;(2)based on generalized method of moments estimators, constructing a C(a) type test statistics for spatial lag stochastic frontier model;(3)extend to spatial panel stochastic frontier models and propose generalized method of moments estiamtion for random effect spatial panel stochastic frontier model, which is easy to implement relative to maximum likelihood method;(4) derive a class of tests robust to spatial effect local misspecification for testing random effect of spatial panel stochastic frontier models. These robust tests are much easier to compute and have good performance. In addition, extensive Monte-Carlo simulations will be conducted to analyze and compare the finite sample performance of estimators and test statistics.
英文关键词: Stochastic frontier;Specification test;Semi-parametric;Spatial panel;Maximum likelihood