基于GARCH模型的VaR方法对我国开放式基金风险的分析

本文从我国开放式基金收益率序列的分布与波动性两方面建立了一个估计基金风险的VaR-GARCH模型，在正态分布和能够刻画收益率的尖峰厚尾特征的t分布GED分布三种不同的分布假设下，对基金的VaR值进行估计，并应用Kupiec失败频率检验方法对VaR模型的准确性进行了返回检验。结果显示，基于GED分布的GARCH模型计算的VaR值比基于正态分布和t分布GARCH模型计算的VaR值更真实地反映了基金的风险。

Analysis of the Risk of Mutual Fund Using VaR Method Based on GARCH Model in China

The paper sets up the VaR-GARCH Model for the risk of mutual fund based on the consideration of volatility and distribution of the return series, and estimates the VaR of mutual fund in China using the model under normal distribution, t-distribution and GED-distribution separately. Finally using Kupiec' s back-testing we test the veracity of the VaR-GARCH Model. The result shows the VaR estimated using the model under GED-distribution is better than the one estimated using the model under normal distribution and t-distribution in reflecting the risk of mutual fund.