Panel-Data下Granger因果检验的理论和应用发展综述

An Overview on the Theory and Applications of Granger Causality Tests in Panel-Data

Panel-Data下Granger因果检验的相关理论是最近几年才发展起来的,现有的研究提出了关于Panel-Data下Granger因果检验的四个基本假设:同质无因果关系假设(HNCH)、同质因果关系假设(HCH)、异质因果关系假设(HECH)以及异质无因果关系假设(HENCH),根据检验参数的特点给出三种类型的检验模型:固定系数模型、随机系数模型和混合固定随机系数模型。目前,还只有固定系数模型的相关理论较为完善,另外两种模型的检验还都存在一定的难度。因此,只有从理论研究和实际应用两个方面对该理论进行阐述,并对现有的理论进行简要的评述,才可指出其存在的不足及可能的改进方向。

The recently developed theory of Granger Causality Tests in Panel -Data distinguishes four kinds of causality relationships（homogenous non causality hypothesis, homogenous causality hypothesis, heterogenous causality hypothesis and heterogenous non causality hypothesis）, three kinds of models （fixed coefficients models, random coefficients models and mixed fixed and random coefficients models）. By now, just the theory of fixed coefficients models is almost perfect, there are still difficulties in testing the left two kinds of models. The article introduces the theory of granger causality tests in panel - data from theory and applications, then comment on the theory of granger causality tests in panel - data, find the limitation of this theory and point out what we can do for this theory.