基于GJR模型的EVT动态风险测度研究  

Study on dynamic risk measurement based on G JR and EVT

作　　者：林宇[1] 魏宇[1] 黄登仕[1] 

机构地区：[1]西南交通大学经济管理学院,四川成都610031

出　　处：《系统工程学报》2008年第1期45-51,共7页Journal of Systems Engineering

基　　金：国家自然科学基金(70501025;70572089;70771097)

摘　　要：通过运用ARMA-GJR模型捕获上证综指的损失序列的自相关、波动集聚性和杠杆效应特征,用极大似然估计(MLE)估计模型参数以求出条件均值和条件方差以及标准残差序列;然后假设沪市指数损失标准残差序列近似满足EVT条件,分别取175、105和35个极值数据并运用MLE来估计广义帕累托分布(generalized Pareto distribution,GPD)的参数,进而估计出q分位数对应的动态风险值Valt(value at risk)和ES(expected shortfall);最后对风险测度方法的估计效果进行分析.实证结果表明:标准残差序列的极值尾部近似服从GPD,ES是相对于VaR更保守的风险测度方法.In order to measure dynamic VaR and ES, firstly, use ARMA-GJR model and MLE to obtain estimators of the conditional mean and conditional volatility and standardized residuals based on the indices losses with heavy-tails, leptokurtic, heteroscedasticity, autocorrelation and leverage effect and so on in Shanghai security integration index. And then select 175,105 and 35 extreme datum to estimate GPD parameters through MLE, further calculate daily dynamic VaR and ES, at last check the effect of risk measurement methods. Empirical results show that the standardized residual of the indices losses in Shanghai is near independent identically distributed, and fit EVT; ES is a more conservative way to measure security market risk than VaR.

关 键 词：GJR EVT 极值尾部 动态风险 测度 

分 类 号：F830.91[经济管理—金融学]