局部平稳性未知条件下基于ESTAR模型的单位根检验

Unit Root Test against the Globally Stationary ESTAR Model under Unknown Partial Stationarity

文章探讨了局部平稳性未知情况下ESTAR模型的单位根检验,提出了修正的Wald统计量,通过模拟给出了其临界值,推导出了该统计量的极限分布,并分析了在有限样本下该统计量的特性。通过蒙特卡罗模拟,该检验统计量具有良好的检验水平和较高的检验功效,进一步通过模拟发现在全局平稳非线性ESTAR模型下,该修正的Wald统计量比KSS型统计量具有更高的检验功效。

This paper discusses the unit root test against the globally stationary ESTAR model when partial stationarity is un- known and a modified Wald-type test is proposed. By simulating to tabulate the critical value, the asymptotic distributions of the test statistics are derived and the properties under finite samples are examined. In a Monte Carlo study, the test statistics show good in- spection level and high inspection effects. Furthermore, compared with the popular KSS-type test proposed by Kapetanios et Al. , this modified Wald-type model examines the statistics with higher inspection effects by simulating under the non-linear ESTAR model with whole stationarity.