| In recent years,international financial markets have undergone enormous changes. With the daily volatility of financial markets,some catastrophic events happened frequently. The measurement of financial risks has become been one main issue in finance field. Traditional risk management is based on Markowitz portfolio theory, which assumes a multinormal distribution of such risk factor returns and the mutual relationship is linear, and uses VaR to measure financial risk. However,a lot of evidences show that the volatility of an asset possess the heteroscedasticity and clustering. These hypothesises often underestimate the probability of catastrophic events. Meanwhile, as the shortcomings of VaR, it couldn not meet the actual need of financial practitioners. For this reason, we put forward Copula Function and CVaR Method in this paper.For a single finance asset, CVaR, which considers the average of tail losses. For portfolio,by Copula Function which considers the nonlinear correlation of asset portfolio, we employ different marginal distributions to form a suitable joint distribution function of gains and losses.In this paper, we first disscuss the application of Copula Function with the character of tail correlation of finance markets. By Historical Simulation Methods,VaR-CoVaR Method, Monte Carlo Simulation Method and Extreme Value Theory Method, we calculate VaR and CVaR of Dow Jones industrial Index and Hang Seng Index. It is obvious that all of VaR can’t pass Kupiec Back Test and underestimate real risk. while CVaR can measure risk more effectively and accurately. At the confidence level of95%and99%,the probability of failure reduces a lot. Considering the time varying of returns, we build GJR-EVT model to fit a single return of stock index. For portfolio, we establish GJR-EVT-Copula and compute VaR and CVaR by Monte Carlo Simulation Method. Evidences show that CVaR based on GJR-EVT-Copula model can measure risk more accurately. |
