关键词: 结构突变, 单位根检验 DF统计量, 蒙特卡罗模拟, 检验功效

Abstract: “Perron Phenomenon” is that the conventional DF unit root tests incorrectly tend to conclude the unit root process which actually is a trend stationary process with a structural break. This article considers the following four types of structural breaks, i.e. level shift, intercept shift, slope shift, mixed shifts with intercept and slope. We deduce the asymptotic distributions of the level shift and intercept shift, and then analyze the characters in finite sample of the four types of structural breaks by Monte Carlo simulations. The article is the further research on the “Perron Phenomenon”, which is the complement for DF unit root test.

Key words: Structural break, unit root test, DF statistics, Monte Carlo simulation, Test power