基于参数模型的EVaR风险度量计算方法及实证研究

针对EVaR(Expectile-based Value at Risk)风险度量提出了基于GARCH类和SV波动率模型的EVaR风险度量计算方法,即EVaR计算的参数模型方法.并基于模拟学生t分布时间序列数据,给出EVaR样本外预测的失败率检验方法:Kupiec失败率检验和动态分位数(DQ)检验法.与采用CARE(Conditional Autoregressive Expectile)模型的EVaR计算方法进行了对比研究,结果表明基于GARCH类模型和SV模型相对于基于CARE模型有更优的EVaR预测效果.选取2004年1月5日到2009年12月30日的国内外五个股票市场指数数据,针对日对数收益率进行了EVaR风险度量的实证研究,得出在金融危机期间,基于参数模型的EVaR预测要比基于CARE模型的EVaR预测更接近市场实际风险.

Research on EVaR computation based on parametrical models and its applications

This paper proposed a computational method of EVaR based on GARCH and SV models. For the simulated time series data, the method estimates the EVaR based on CARE model, GARCH model, EGARCH model and SV model respectively. The results are evaluated and compared with two back-tests: Kupiec fail-ratio test and Dynamic Quantile (DQ) regression test. It shows that the estimated EVaR based on the parametrical models is superior to the EVaR estimation based on the CARE model. Empirical studies with daily log returns of 5 stock market indexes over the period 2004-01-05 to 2009-12-30 are illustrated. The result shows that during the period of worldwide financial crisis, the estimated EVaR based on the parametrical models performs better than that based on the CARE model.