空间门限随机前沿模型的构建与估计  

The Construction and Estimation of Spatial Threshold Stochastic Frontier Models

作　　者：蒋青嬗 李毅君 JIANG Qing-shan;LI Yi-jun(School of Mathematics and Statistics Guangdong University of Foreign Studies,Guangzhou 510006,China;Lingnan College,Sun Yat-sen University,Guangzhou 510275,China)

机构地区：[1]广东外语外贸大学数学与统计学院,广东广州510006 [2]中山大学岭南学院,广东广州510275

出　　处：《统计与信息论坛》2019年第4期25-31,共7页Journal of Statistics and Information

基　　金：全国统计科学研究项目《内生性空间面板随机前沿模型估计方法研究》(2018LY81);教育部人文社会科学研究规划基金项目《新形势下最低工资标准的保障能力与适度性的统计综合评价方法及实证研究》(16YJA910001);广东省自然科学基金项目《"强关系"抑或"弱关系"?股东关系网络的影响机理研究:基于企业绩效的视角》(2018A030310572)

摘　　要：基于空间计量视角拓展门限随机前沿模型,从技术效率时变和非时变两个层面分别构建空间门限随机前沿模型。模型同时考虑了生产单元的异质性和空间相关性,适用性较佳。分别使用两阶段最小二乘法和极大似然方法估计非时变和时变层面下的参数,使用JLMS法估计效率。蒙特卡罗结果表明:此方法的估计精度较高。随着样本容量的增加,估计精度增加;忽略空间效应或者门限效应,估计精度较低。This paper incorporates spatial effect into threshold stochastic frontier models and constructs spatial threshold stochastic frontier models based on time-invariant and time-variant technical efficiency to get better model applicability.These models simultaneously study the spatial correlation and heterogeneity of production units which can effectively avoid biased and inconsistent estimators and be with better applicability.We use two stages least squares method and maximum likelihood method to estimate the unknown parameters,then use JLMS method to estimate technical efficiency.The simulations show that:(i) The method is with high accuracy.When sample size increases,the accuracy gets higher.(ii) Neglecting spatial effect or threshold effect will get low accuracy.

关 键 词：随机前沿模型 门限效应 空间效应 两阶段最小二乘法 极大似然估计 

分 类 号：O212[理学—概率论与数理统计]