New Method Of Selecting Binary Archimedean Copula Function And Improvement Of Related Model

As the process of economic globalization speeding up,the factors affecting the of the economy is no longer single.And the relationship between the various factors become complex,so the traditional model which has based on linear correlation analysis of the financial market risk is no longer suitable for more complex economic system.When measuring the correlation between financial data,we usually assume that marginal distribution is normal distribution or t distribution.In fact,this is contradiction with financial data which has characteristics of peak thick tail.Copula connect theory is very important,which has two advantage.On the hand,Copula can separate the marginal distribution and joint distribution of variables,and no longer confined to normal distribution or t distribution of such a few kinds of probability distribution,we can choose appropriate distribution according to actual situation.On the other hand,Copula function which is derived for consistency and relevance measure not only can describe the linear correlation, but also can capture the nonlinear, non symmetric correlation,more widely in the practical application.The generation of Archimedean Copula function can effectively describe the complex nonlinear dependence relationship among multivariate variables in the financial field,and because of simple structure,convenient calculation is widely used in financial markets.But choose different Archimedean copula connect function will get different results.So, how to choose appropriate Archimedean copula connect function to describe the dependent relationship between data is important.Based on the binary Archimedean copulas connect function as the research object,starting from the copula connect distribution function,constructing a estimator which is more effective than the nonparametric estimator of the distribution function,then calculating distance between ?tk? and ?tk?,according the distance to choose the right Archimedean copula connect function.The empirical analysis shows that the new method can effectively select binary Archimedean copulas connect function.We often build Copula-GDP model and Copula-GARCH-GDP model to measure the correlation between financial assets sequence,but the two models exist some shortcomings.Copula-GPD does not consider conditional variance and volatility clustering,two models are selected for a single Copula function to analyze the correlation between the sequence,the results are not comprehensive.In this paper,on the basis of considering financial sequence exists asymmetry,Overcoming the limitation ofthe single Copula function describe the tail dependence of the sequence,M-copulas-TGARCH-GDP model is established.The empirical analysis shows that the M-copulas-TGARCH-GDP model can better reflect the tail correlation between financial sequence.