STAR模型下初始条件对退势单位根检验统计量的影响研究  

Impacts of the Initial Condition on Unit Root Test Using De-trending Procedures in STAR Models

作　　者：欧阳敏华[1] 

机构地区：[1]暨南大学统计学博士后科研流动站,广州510632

出　　处：《统计与决策》2017年第10期35-38,共4页Statistics & Decision

基　　金：中国博士后科学基金资助项目(2014M562245)

摘　　要：在STAR模型框架下,文章分析了不同初始条件下OLS和GLS退势KSS统计量检验水平和检验势的特征,发现OLS退势KSS统计量的检验势随初始条件的增大而上升,GLS退势KSS统计量的检验势随初始条件的增大而下降,这与通常忽略初始条件影响下GLS退势比OLS退势KSS统计量有更高检验势的结论不一致。为此,进一步探讨了考虑初始条件情况下STAR模型中的退势单位根检验策略。This paper investigates the power and size performance of the KSS unit root test using OLS and GLS de-trending procedures under uncertain initial conditions in STAR models. Results show that the KSS test using GLS is more powerful than the OLS-based test for the case without consideration of initial condition. However, the power of GLS-based KSS test decreases drastically as the initial deviation gets large, while the power of OLS-based KSS test increases. The paper further discusses the unit root testing strategy under initial conditions in STAR models.

关 键 词：STAR模型 单位根 退势 KSS统计量 

分 类 号：F224[经济管理—国民经济]