单位根检验统计量M^(GLS)的有限样本性质与应用  

Finite Sample Property and Application of the Unit Root Test M^(GLS) Statistics

作　　者：聂巧平[1] 

机构地区：[1]南开大学国际经济研究所

出　　处：《数量经济技术经济研究》2007年第4期103-114,共12页Journal of Quantitative & Technological Economics

基　　金：①本文研究得到国家社会科学基金资助;编号:03BJY014。

摘　　要：Ng-Perron单位根检验是值得在实际分析中广泛应用的检验方法。本文在详述M^(GLS)统计量的构造方法基础上,应用蒙特卡罗模拟试验分析了该统计量的有限样本分布特征,并拟合了常用显著性水平下临界值关于样本容量的响应面函数;此外,针对实际应用中检验回归式中差分变量滞后阶数的确定,本文总结了几种常用的方法,从而完善了依据M^(GLS)统计量进行单位根检验的程序和步骤。NG and Perron （2001） developed the M^GLS statistics for test unit root ,which is worth popularizing in the actual analysis. Based on the specification about the constructing method of the M^GLS statics, author analyses the finite sample distrbution characters of the statistic by Monte Carlo simulation, and fits the response face function of critical value about the sample size, where the significant level is in common use. What＇s more, about determining the lagged order of the difference variables in the test regression, some methods used usually are introduced. What have done in this paper will perfect the program and steps which are used to test the unit root by M^GLS statistics.

关 键 词：单位根检验 Ng-Perron检验 蒙特卡罗模拟 响应面函数 

分 类 号：F224[经济管理—国民经济]