Research On Some Problems Of Multiple Cointegration In Economic Application

Cointegration theory,especially the theory of multiple cointegration,has been widely used.Multiple cointegration theory has become a powerful tool from micro economy to macro economy,from agriculture to finance.However,because of the lack of understanding of the theory of multiple cointegration,many scholars have made a lot of mistakes in the process of application.In this paper,we review the literature on cointegration theory and application,and then introduce the theory of co integration,including the stationary nature of the time series,the unit root test,cointegration and error correction model,VAR model and so on,which lays the theoretical foundation of the whole paper.First of all,we discuss DGP,that is the process of the data generation process identification.DGP error can lead the consequences include obtaining the error of the estimated value of the cointegration coefficient,excessive rejection of the original hypothesis,excessive acceptance of the original hypothesis.The article introduces the 5 kinds of DGP and how to determine the correct DGP.Secondly,we introduce the matters that should be paid attention to in the application of multiple co integration.Including the selection of the data is not easy to use the logarithm;cointegration is only for the non-stationary sequence,if the sequence is stable,then there is no cointegration;the choice of the lag order is very important,this article from the empirical and mathematical basis point of view should be how to choose the lag order.Then,the selection of co integration vector is introduced.In the multiple co integration model,it is easy to be a phenomenon that there are many co integration relationships among variables,and how to determine the optimal relationship is a very important problem when there are many co integration relationships.In empirical study,if we do not seek to determine the cointegration relationship,as long as know variables exist cointegration relationship can be,you can not selected cointegrating vectors,or just select a cointegrating vectors,or select the variable contains the most of the cointegrating vector;If the variables are not constrained,then the corresponding co integration vectors can be selected;and in most of the empirical studies,economic variables will have relevant economic theory as a support,so it can be by adding constraints of the way to get the most consistent with the actual cointegrating vectors.Finally,the identification of structural vector auto regression model is discussed.The structure vector autoregressive model includes the effects of the current period,so it is more economical than the VAR model.In this paper,we introduce the identification of structural vector auto regression,and how to use Choleski to decompose,and then introduce how to determine the order of decomposition in two cases of stationary and non-stationary sequences.At the end of the article,the paper summarizes the whole paper and points out the future research direction.