ADF与PP单位根检验法对非线性趋势平稳序列的伪检验  

Spurious Tests to Nonlinear Trend Stationary Series with ADF and PP Unit Root Test Methods

作　　者：刘田[1] 

机构地区：[1]西南财经大学统计学院

出　　处：《数量经济技术经济研究》2008年第6期137-145,共9页Journal of Quantitative & Technological Economics

摘　　要：ADF与PP单位根检验对无趋势或线性趋势平稳过程可给出正确的结果。但蒙特卡罗实验表明,对非线性趋势而言,它们趋向于将平稳过程误判为有单位根。在一定条件下,各种非线性趋势可以看成准线性的,从而利用ADF与PP检验得出正确的结论。信噪比小于15倍时,PP检验可得出正确的结果;信噪比小于4倍时,ADF检验可得出正确的结果。对非线性趋势平稳序列的检验而言,PP优于ADF检验。随着干扰相对强度的增加,正确检验的可能性也大大增加。To non-trend or linear trend stationary process, ADF or PP unit root test can give right result. But to nonlinear trend, Monte Carlo test shows that they tend to take a stationary series as a unit root process. Under certain conditions, we can treat nonlinear trend as quasi-linear and draw right conclusion using ADF or PP test method. PP can give right result when SNR〈15, and ADF can work right when SNR〈4. PP test is better than ADF to a nonlinear trend stationary process. Along with the increase of the ratio of disturbance, the possibility grows rapidly to give right result.

关 键 词：单位根 伪检验 非线性趋势 ADF PP 

分 类 号：F224.0[经济管理—国民经济]