摘要
当检验方程含有确定性趋势时,基于残差的Panel协整检验通过全样本回归来消除确定性趋势的方法会导致检验统计量出现Nickell偏差,必须使用标准化系数进行修正,但这些标准化系数却依赖于真实DGP的设定参数。本文将Breitung和Das(2005)等使用部分样本回归消除确定性趋势的方法应用到Panel协整检验,提出了基于准残差的Panel协整检验,本文证明了这种新的检验统计量的渐近正态性,模拟的结果也表明其具有非常良好的小样本表现。
Panel cointegration test based on residual eliminates deterministic trends by regression with whole sample, which brings the Nickell Bias to its statistic and must be corrected by some standardized coefficients. Unfortunately, these standardized coefficients are dependent on the null parameters. This paper applies the method of eliminating deterministic trends by regression with partial sample, which advised by Breitung and Das (2005), to panel cointegration test, and advises a new approach for panel cointegration test which based on quasi-residual. We prove that the statistic of this new approach converges to standard normally distribution, the results of simulation also show that it has a good small sample performance.
出处
《统计研究》
CSSCI
北大核心
2009年第2期74-79,共6页
Statistical Research
基金
国家社科基金项目(项目批准号:06BJY033)
广东省普通高校人文社会科学研究重点项目(项目批准号:06JDXM790002)的阶段研究成果