Johansen和Juselius协整检验应注意的几个问题  

Some Notes on Johansen and Juselius Cointegration Test

作　　者：钟志威[1] 雷钦礼[1] 

机构地区：[1]暨南大学经济学院,广东广州510632

出　　处：《统计与信息论坛》2008年第10期80-85,90,共7页Journal of Statistics and Information

摘　　要：Johansen和Juselius的似然比检验多变量协整关系的方法在实证分析中得到了广泛应用。在总结该方法的基础上,针对国内使用该方法存在比较混乱的状况指出了一些注意事项,譬如根据经济时间序列的数据生成过程选择确定性成分,检验临界值的使用以及协整关系个数的非唯一性等问题,还简要论述了阶数的确定、外生性与因果关系检验等问题,最后指出了该检验的一些不足。通过对上述问题的讨论,试图为实证研究人员在使用该方法时提供简单有效的指导性建议。Johansen and Juselius cointegration test has already been used in practice widely. We summarizes this method, and then arises a few attention to this test according to the fact that the confusion of application of this method in domestic, such as how to deal with deterministic components basing on time series＇ DGP, the usage of critical values and the cointegration identification problem. Besides, we are also concern with the determination of orders, the tests of exogeneity and granger causality. At the last section of this paper we point out some arguments about Johansen and Juselius cointegration test. Through such problems＇ extended discussion, we attempt to give a few practical and efficacious guidelines for cointegration analysis.

关 键 词：协整 JJ检验 识别问题 因果关系 外生性 向量误差修正模型 

分 类 号：F275[经济管理—企业管理]