A Value-at-Risk Analysis With Long Memory Of Volatility:Evidence From The Chinese Stock Market

摘要

This paper investigate Value at Risk (VaR) measurement based on the long memory properties for closing prices of SSEC and SZEC in Chinese stock market. The FIGARCH(1, d, 1) and HYGARCH (1, d, 1) models with normal, student t and skew student t distribution of innovations are used to calculate dynamic VaR for long and short trading positions, and apply Kupiec's LR testing to test the accuracy of insample VaR measuring and out-of-sample VaR forecast ability of these models introduced in this paper. Our empirical results show that there exists significant long memory of volatility in Chinese stock markets; the skew student t distribution is the best to model the innovation of return, but normal is the worst to capture distribution of financial series. The FIGARCHd , d, 1) and HYGARCH(1, d, 1) models with skew student t distribution measure accurately dynamic VaR for SSEC and SZEC of Chinese stock market, and also exhibits outperform forecast ability of out-of-sample VaR.