Study On Accuracy Of Measuring Of Dynamic Extreme VaR And Granger Casuality Of VaRs

Financial markets always act as very important role in economy, and become a key for financial risk manger and investor. Once financial market risk rise, economy will be destroyed badly, and maybe break out economic crisis, and maybe bring social turbulence. So it is a key for government and investor to study financial market risk management.This paper applied Extreme Value Theory (EVT) to construct the risk measurement model of financial market based on the shortcome of some risk measurement model. This paper applied ARMA(1,1)-GARCH(1,1), ARMA(1,1) -GJR(1,1) to capture some stylized facts, such as, autocorrelation,volatility clustering and leverage effect, and then construct standardized residuals, and then select 10% maximum standardized residuals as extreme values and apply EVT to model these extreme values. We measure dynamic extreme VaR by stochais process. At last, apply Back-testing method to check the accuracy of these risk measurement models. Our results show that these can measue the Chinese financial market risk accurately.This paper also applied dynamic extreme VaR measurement model to measue VaR under weekly and daily time scale to chech if there exist invariance of the accuracy of different time scales. The results show that not all the models are of time-scaling-invariance, there exist some differences of the effect of risk measurements in the three stock markets at different confidence levels and different time scales, and this paper should use the best model to measure risk for each time scale.This paper also apply multivariate GARCH to model covariance matrix of asset portfolio loss, and to calculate the standardized residual series of portfolio loss, and then applies EVT to model the extreme tail of standardized residual series, and measure portfolio dynamic extreme VaR, at last, this paper also applies Back-testing method to check the accuracy of the different risk models. Our result show that multivariate GARCH model can capture the time varying correlations of different losses effectively; the distribution of extreme tail is close to GPD; multivariate GARCH and EVT can measure the dynamic risk of asset portfolio accurately as well.This paper applied Grange-Causalty method to test relations of dynamic extreme VaR of Chinese stock market and other international stock markets, and analysis Grange-Causalty between Shanghai copper future market and New York copper future market. Our results show that there are no risk Grange-Causalty effect between Chinese stock market and other three international stock markets. Extreme dynamic risks transmit form Hong Kong stock market to Tokyo stock market, and then transmit form Tokyo stock market to New York stock market. There exists bidirectional Grange-Causalty effect between Tokyo and New York stock markets; there exist VaR Grange-Causalty effect between shanghai and shengzhen stock markets; there exist VaR Grange-Causalty effect between Shanghai copper future market and New York copper future market.Combining these empirical results to the Chinese financial market, This paper think that a new financial systeme must be constructed, set up a multi-level financial market system, strengthen risk management and optimized the quality of corporation, and try to ingrater Chinese financial market and international finance market, to realize the boom of Chinese financial market.