The Comparison And Empirical Study Of VaR Parametric Models

Value-at-Risk is wildly used to valuate the market risk. This thesis deeply describes parametric methods of VaR. Eight VaR models are used to calculate the VaR number, using ShangHai 180(Code: 1B0007) index and ShenCheng(Code: 399001) index in Chinese stock market. Then, in order to compare and evaluate the performance of these VaR models, A comprehensive backtesting system will be constructed to evaluate a VaR model from three aspects——accuracy, conservatism and efficiency. Using this system and the recent data of the two indexes, we compare the out-of-sample predictive performance among the eight VaR models in the confidence lever 95%. The result of backtesting tells us that none of the available methods can produce uniformly superior risk forecasts for all performance criteria.However, it is possible to draw some broad conclusions. The first conclusion is that, the performance of the models considered in this paper are not greatly dissimilar across most of the performances criteria, excluding GARCH-M model. Secondly, Leverage effect and spillover effect become more and more important in VaR modeling. Thirdly, inconsistently with common conclusion, compared with the models with normal distribution,models with other fat-tail distribution don't perform perfectly in 95% confidence level, and GARCH-GED is more attractived to negative mangers. Lastly, compared with the three phases in front. the performance of the forth phase in all the models are incongruous.It may be the result of the agelong rising stock market.Taking one with another, in 95% confidence level, risk managres can use EWMA model for mangaging risk, for EWMA model has good performance in accuracy, conservatism and efficiency and is operated simply. Besides, GARCH-t is a good choice for risk manager for its perfect performance generally.