Empirical Research Of Mixed GARCH-Copula Model In Measuring The Risk Of Portfolio

Copula functions have unique advantages in capturing the related structure between multiple financial time series, so it is widely used in empirical research. However, financial markets are complex, then leads to the complexity of the related structure. And mixed Copula model can choose the different Copula function and adjust the weight parameters flexibly to fit the complex related structure of the markets. So it is more suitable to deal with the real problems. Given few empirical studies done with the three-dimensional mixed Copula model, this paper will try.This paper selected 2289 observations separately from the Shanghai Composite Index, CSI 300 Index and Industrial Index as the sample data for our empirical research. First of all, sufficient evidence shows that the financial time series exhibit fat tails and skew, so here choose the GARCH-t model to fit the marginal distribution. Considering the asymmetric dependence between the three assets, this paper not only select each Copula function of different types to build four models, but also build mixed Copula model to capture the related structure. Then, we assess and comparative analysis the above different models by K-S test and the failing Days of VaR. The results show that the mixed Copula model is the best. Finally, we choose the optimal model combining with the Monte Carlo simulation to calculate VaR of the specific portfolio, which help the investors to make efficient decisions.