The Correlation Analysis Of Chinese Stock Market Based On Time Varying T-Copula Model

Globally,financial industry with the fastest growing is gaining much more concerns in recent years,which is especially true in China.With constant reform and innovation of financial institution,the relationship among different financial markets is becoming closer.As we enter the new century,a series of policies are implemented and the focus is the reform of split-share structure,which indicates the opening of the financial market in China.And the stock market of our country is growing mature with the launch of GEM board and Stock Index Feature.In the past decades,great changes have made a difference in international financial market,such as the Asian Financial Crisis in 1990 s,the European Debt Crisis in2011 and the boom and slump of Chinese stock market in 2015.We can find that stock market in our nation is easily influenced by the volatility of global stock markets.As a result,shareholders could hardly earn any money due to the continual fluctuation.Naturally,the correlation analysis of stock-market-risk measurement has become a heated topic,and there are some analysis models of stock-market-risk measurement,including nonlinear analysis,non-normal analysis and tail correlation analysis.However,the former analysis methods,based on normal distribution hypothesis,are unable to cope with the analysis of complex stock market.Thus,we need seek more effective approach to solve problems in stock market.Thus,here is the introduction of Copula Function.Since the function shows distinct advantages in the correlation analysis of stock-market-risk measurement compared with other functions.In real life,the relations of most financial-asset combination are not linear correlation,thus the linear-correlation coefficient is unable to analyze the relation accurately.Considering the large amounts of data distribution,t-Copula Function is introduced in this paper to analyze the correlation of different variables,whether the relation of financial assets belongs to linear distribution or not.Besides,concerning the rapid development of stock market,influenced by external factors,and the nonlinear correlation in stock market,we decide to build a dynamic and nonlinear model to describe some related structure.There are mainly two kinds of dynamic copula models of correlation analysis: time-varying copula model and changeable-structure copula model,so I introduce them in this research.This paper mainly discusses the following parts.First,some related research background,overseas and domestic research status and the research advance of this function are introduced here.Then,fundamental definitions,theorems,inference and feature are demonstrated.Next,some related function classification,main function forms,correlation measurement and the application are described specifically.At last,it's about the buildup of model and its testing.Concerning the estimation of parameter in the model,there are quite a few explanations.Furthermore,we give detailed introduction about the buildup,estimation of model and its combination with time varying t-distribution in the aspect of empirical analysis.In the part of empirical study,we start with constant correlation and time-varying correlation and choose exchange rate of closing price of Shanghai Composite Index and Shenzhen Component Index as the object of data research.We apply software tool,Eviews 7.0,to preprocess the two-group data of closing-price,and the access the descriptive statistics of exchange rate.Next,we get standardized residual through describing exchange rate.Data-processing about constant correlation refers to generating two-group standardized residual as the following research object.As for time-varying correlation,we group logarithmic return according to different time,so as to get the relevant standardized residual.Through marginal distribution of research object,we can build relevant function model and get estimated value of parameters,and conduct testing to draw the conclusion.