基于尾部指数回归方法的CVaR估计以及实证研究

Estimation of CVaR and Empirical Analysis Based on Tail Index Regression Model

在险价值VaR是一种非常重要的金融风险度量方法,近期也有很多关于动态VaR以及条件VaR(CVaR)等方面的研究。根据金融资产的收益率具有重尾特征这一事实,本文假定金融资产收益率服从重尾分布,并假定重尾分布的尾部指数随着收益率发生变化。本文基于尾部指数回归模型对重尾分布的尾部指数进行估计,进而得到收益率尾部数据所服从的条件分布,并首次运用该方法对条件VaR进行估计。本文对沪深300指数进行了实证研究,得到CVaR的估计,对估计得到的CVaR的预测效果作出检验,并与传统VaR估计方法进行了对比,实证结果发现本文方法的预测效果更好。

Value at Risk（VaR） is an important financial risk measurement method,and there is a lot of research on dynamic VaR and conditional VaR recently.According to the fact of financial returns＇ heavy-tail,the return of financial assets is assumed to obey heavy-tailed distribution,and the heavy-tailed index is assumed to change with returns.Based on the tail index regression model,tail index and conditional distribution are estimated.And,for the first time,CVaR is estimated by the proposed method.An empirical study of Shanghai and Shenzhen 300 Index is presented,and CVaR is estimated and compared with the traditional VaR estimation methods.The test results of risk prediction effect reveal that our method is better.