论文部分内容阅读
通过运用带宽非参数方法、AR-GARCH模型对时间序列的条件均值、条件波动性进行建模估计出标准残差序列,再运用L-Moment与MLE(maximum Likelihood estimation)估计标准残差的尾部的GPD参数,进而运用实验方法测度出风险VaR(value at Risk)及ES(ExpectedShortfall),最后运用Back-Testing方法检验测度准确性。结果表明,基于带宽的非参数估计模型比GARCH簇模型在测度ES上具有更高的可靠性;基于非参数模型与L-Moment的风险测度模型能够有效测度沪深股市的动态VaR与ES。
By using the nonparametric bandwidth method, the AR-GARCH model is used to estimate the standard residual sequence by modeling the conditional mean and conditional volatility of the time series. Then, the L-Moment and MLE (maximum likelihood estimation) GPD parameters, and then use the experimental method to measure the VaR (value at Risk) and ES (ExpectedShortfall), and finally use the Back-Testing method to test the measurement accuracy. The results show that the bandwidth-based nonparametric estimation model has higher reliability than the GARCH clustering model in measuring ES. The risk measure model based on non-parametric model and L-Moment can effectively measure the dynamic VaR and ES of Shanghai and Shenzhen stock markets.