Research On Asset Portfolio Risk Based On Time-varying Vine-Copula Model

As the relevant structure between financial markets changes over time,this paper uses the two methods to study the nonlinear dynamic correlation structure between multiple financial time series based on the Vine Copula theory.The first is to use Patton's time-varying Copula theory to model the time-varying parameters through the historical correlation and the average cumulative probability of the product of historical observations,that is,the time-varying parameters of the Pair-Copula function obey the ARMA(1,1 0)model.The second is to learn from Creal's generalized autoregressive scoring theory and use data to drive time-varying parameters,that is,a generalized autoregressive scoring model is established for the time-varying parameters of the Pair-Copula function.Combining the daily closing price data of Hong Kong Hang Seng Index,Dow Jones Index,Shanghai Stock Index and Nikkei 225 Index,under different Vine structures,this paper constructs a time-varying Vine-Copula model based on Patton's idea and GAS(1,1)theory to analyze the correlation between asset portfolios.The empirical analysis of the risk of portfolio return rate is carried out by Monte Carlo simulation.This paper uses a two-stage maximum likelihood estimation method to estimate the model parameters.The first step is to use the ARMA-GARCH-GED model to fit the edge distribution of the sequence according to the characteristics of the financial return sequence.In the second step,different Vine structures are selected,and the Pair-Copula function is selected according to the AIC criterion to establish a joint distribution model of asset portfolios.The empirical results show that compared with the fitting effect of models under different Vine structures,the static Vine Copula model,the time-varying Vine Copula model based on Patton's idea,and the time-varying Vine Copula model based on GAS(1,1)theory have a better fitting effect on the financial rate series under the C-Vine structure than the D-Vine structure.Comparing the fitting effect of the model under different parameter forms,the time-varying Vine Copula model based on GAS(1,1)theory has the best fitting effect on the financial return series.According to the Kupiec failure rate test results,at three confidence levels(99%,95%,and 90%),the static Vine Copula model,the time-varying Vine Copula model based on Patton thought,and the time-varying Vine Copula model based on GAS(1,1)theory all have the robustness and reliability of sample measures,and can better fit the fluctuation of asset portfolio risk.In summary,at a 95%confidence level,the time-varying D-Vine Copula model based on GAS(1,1)theory can better measure the VaR of asset portfolios than other models.