The Theoretical And Empirical Research Of DGP Misspecificaton On Johansen Cointegration Test

Cointegration theory was developed as an econometric tool by Granger and others in the 1980s, but in China it was not applied widely before 2003. But this situation reversed when Granger was granted 2003 Nobel Prize in Economics for his creation of cointegration theory, a large number of literature introducing the theory itself or its application emerged, and even to the extent of flooding. Johansen's method is often recognized as the second generation of this method, because it is not dependent on least-squares method partly, but directly established on the maximum likelihood estimator. This approach has been made a standard module and almost included in all econometric software. However, a majority of cointegration analysis overlooked the DGP identification, and just chose one subjectively from the results under five DGP settings; which ignored the general attributes of the research object, so the test result could be wrong.This paper describes the Johansen's Cointegration Test and reviews the development process of Cointegration Theory firstly, a specific introduction about Vector Autoregression Cointegration Test Model, as well as a detailed analysis of five DGP (Data Generating Process), are presented too. After that, I realized these five DGP using the MATLAB programming language. And through Monte Carlo Simulations, focusing on the most commonly used GDP (5) , I respectively implemented the Johansen Test when the model was configured correctly and not; by comparative analysis of the two results, the influence of DGP misspecification on Johansen Test was disclosed. Then, the relationship between urban consumption and inflation in China was studied by means of Johansen Test and Error Correction Model. Finally, this research came to a conclusion and some possible development were discussed simply.