结构突变对KPSS检验水平与检验功效的影响——基于有限样本情形的模拟及实证研究

Structural Change on KPSS Test for Stationarity of Finite Sample by Monte Carlo Simulations and Empirical Study

本文从理论模拟和实证分析两个角度研究了以平稳过程为原假设的KPSS检验在存在结构突变的情形下该检验的有限样本性质,包括实际检验水平和检验功效的分析。结果表明当忽略数据生成过程中的结构突变时,KPSS检验的实际检验水平会发生严重扭曲,从而出现"Perron现象"。本研究针对的是有限样本情形,这不但补充和完善了相关理论,同时对实证分析工作具有很强的借鉴和指导意义。

This paper analyzes the characters of KPSS unit root tests in finite sample where exists structural breaks, including size and power. By Monte Carlo simulations, we draw a conclusion that real test size will be distorted and result in Perron Phenomenon if structural break in Data Generation Process （GDP） doesn＇t be considered. Because the purpose of this study is the limited sample of cases, so the conclusions not only supplement and perfect the relevant theory, but also have a strong reference and guidance to empirical analysis.