The Systematic Risk Measurement In Chinese Stock Market Based On T-Copula-GJR-VaR Model

Since China's reform and opening up, great changes have taken place in our financial markets, the market risk which financial institutions face is increasing constantly. In order to measure the risk of market more accurately and prevent risk event, make investment decisions timely, it is necessary to research the subject in-depth. The main work In this paper has the following several aspects:First of all on the basis of a large number of domestic and foreign related literature and cases are consulted, we carry on beneficial review on the traditional theory and methods of calculating VaR. Different theories are compared and discussed.At the same time the shortcomings of traditional methods are summarized.Secondly, GARCH, EGARCH and GJR model as financial time series models are used to characterize the conditions marginal distribution. An Contrast analysis is made that the order(1, 1) model has more advantage of the order(1, 2) model. It is proved that logarithm return of Shanghai Composite Index and Shenzhen Component Index obey the peak-fat-tailed distribution empirically. GARCH effect of market volatility is obvious, and Shanghai composite index has significant leverage effect and GJR(1,1) model is suitable. While the Shenzhen component index of the leverage effect was not significant, therefore the GARCH(1, 1) model is still used.Thirdly, Method of CML is used to estimate the parameters of Gaussian Copula and t-Copula function. Two marginal t distributions are connected and the related structure is described by Copula function between Shanghai Composite Index and Shenzhen Component Index. It is reveal that the related level of Shanghai composite index and Shenzhen component index up to 91.47% by linear correlation matrix. Two markets has the phenomenon of up and down together obviously, it is reveal that systematic risk is very great.Fourthly, Monte Carlo simulation method is used to create a distribution of average weighted index return, based on that VaR is calculated. Kupiec test is conducted for retrospective test. It can't be refused that t-Copula-GJR-VaR model in all the degrees of confidence. It shows that the model on the portfolio risk measurement has higher accuracy.Finally by t-Copula-GJR-VaR model, we analyze Chinese stock market volatility in recent several years in detail, and found that in the market when relatively stable period, the prediction of VaR is very good, but when structural changes occurs in the market, model predictions are no longer accurate, simulation VaR is punctured, at the same time accompanied by the phenomenon of VaR increases sharply. So the time when the market shifts its style could be found by us for these phenomenons. This has very important practical value that Financial institutions are guided for investment decision making in the stock market by this judgment when market structure shift.