摘要
本文通过蒙特卡洛模拟试验,研究了个体间的协整关系对面板单位根检验统计量分布、实际检验水平及检验功效的影响。结果显示:个体间协整关系个数和协整向量均对检验统计量分布产生影响;同时,如果忽视个体间存在的协整关系,继续运用个体间无协整关系条件下检验统计量的临界值进行面板单位根检验,各检验统计量实际检验水平将严重失真。
This paper researches the effect of cointegration among individual units on the distributions, sizes and powers of unit root test statistics in panel data via Monte Carlo simulations. The results reveal that both the number of cointegTating relations and the cointegTating vectors have effect on the distributions of the statistics. Therefore, if cointegrations among individual units are ignored, using critical values under crosssectional dependency hypothesis will distort the test size.
出处
《统计研究》
CSSCI
北大核心
2009年第4期91-100,共10页
Statistical Research
基金
国家自然科学基金项目(批准号:70771072,70571039)的阶段性成果