Granger因果关系检验的模型选择与相关策略研究

运用Granger因果关系检验识别确定经济变量间因果关系是经济研究中极为常见的分析模式,然而在具体应用时,Granger因果关系检验的功效会受到模型形式选择与检验策略因素的影响,为此,解析了Granger因果关系检验的水平型VAR、差分型VAR、VEC三种模型形式选择的基本原理,探讨了与模型选择相关的四大检验策略,即变量个数选择、滞后阶数选择、变量单整性检验、协整空间维数选择,并给出了Granger因果关系检验相对稳妥的实践操作程序。

A Research on Model Selection and Related Strategies of Granger Causality Test

It＇s a very common mode of analysis in empirical research of economics to use Granger causality test to identify causality relationship between variables. However, factors like model selection and detection strategy usually have an effect on the power of Granger causality test. Therefore, it is very important to search robust operating procedures for Granger causality test and to find scientific result for causality relationship. This paper firstly analyzes basic principles of three models of Granger causality test, i. e. , maximize reduce horizontal VAR model, differential VAR model and VEC model. Then explores four test strategies related to model selection, including selection of number and lag order of variables, test of stationarity and selection of dimensions of cointegration space. Finally it puts forward a relatively robust operating procedures for Granger causality test.