基于异质面板数据的格兰杰因果检验CHNC

Testing Granger Causality CHNC in Heterogenous Panel Data

本文提出一种新的基于异质面板数据的格兰杰(1969)因果检验方法CHNC,将面板数据依据数据生成过程划分为若干个类,使得类内为同质面板数据,类与类之间异质,在此分类基础上构造针对总体的格兰杰因果检验统计量CHNC,并证明统计量渐近服从卡方分布。实证分析部分,文章研究我国31个省市2001-2011年金融发展与城市化进程之间的关系,研究结果表明,金融发展与城市化之间是一种互相促进与影响的双边影响关系,此研究结论同时也表明,研究金融发展与经济问题时,不能简单依赖相关分析原理确定影响变量与响应变量,很可能变量之间存在双向因果影响关系。

This paper proposes a simple test of Granger（1969） non causality hypothesis CHNC in heterogeneous panel data,The panel data based on the data structure is divided into a number of categories,homogeneous within the class,heterogeneous between classes,and this statistic is shown to converge Chi-square distribution with N tends to infinity.Empirical analysis,We implement CHNC with a sample of 31 provinces over 2001-2011 periods,develop the relationship between financial development and the urbanization process.The results show that inclusive financial development and urbanization process is a two-way effect relationship,This study also showed that,in the study of financial development and economic development,financial development and urbanization process,model building and variable selection are important.