Financial Risk Measurement Based On The Theory Of Copulas

With the accelerated process of financial globalization,the risk of financial market is increasingly complex and diversified.The related models of financial assets are characterized by nonlinear,asymmetric and tail correlation.The original linear correlation analysis method is no longer suitable for describing financial risk.Copula function is a good tool to describe related structure.It can well measure various complex related patterns and degree of coeerlation in financial market.This paper explores the application of Copula function in financial risk measurement.The key point of this paper is to establish a financial model based on Copula function.It can be used to depict the correlation of financial assets and quantify the risk value of financial assets.There are two parts in the empirical study.The first part studies the correlation between the Shanghai stock index return rate and the logarithmic change rate of volume.Under the normal distribution,the t distribution and the generalized error distribution,we use different GARCH family functions to fit the edge distribution.According to the AIC criterion,we choose the EGARCH(1,1)-ged model to fit the rate of return and the GJR-GARCH(1,1)-t model to fit the logarithmic change rate of volume.The hybrid model of EGARCH(1,1)-ged and GJR-GARCH(1,1)-t combined with Gumbel Copula is constructed.The hybrid model well describes the characteristics of tail correlation between the two sequences.The second part quantifies the risk value of investment portfolio in Shanghai composite index and Shenzhen Component Index.First,GARCH(1,1)-t model is established to fit the edge distribution,and then t Copula function is used to describe the related structure between two assets.Finally,t Copula-GARCH-Monte Carlo-VaR model is constructed with Monte Carlo simulation.Compared with the traditional model,the hybrid model not only can accurately depict the correlation between two assets with t Copula function,but also can use Monte Carlo simulation,and ultimately effectively quantify the value at risk.