基于GARCH-EVT模型的证券投资基金动态风险测度

Dynamic risk measure of securities investment fund via GARCH-EVT model

文章考虑了一类极值风险特征更为明显的金融资产,以股票基金、混合基金和债券基金为研究对象,比较研究了RiskMetrics、GARCH和GARCH-EVT 3类模型在动态极端风险测度上的表现,分别采用似然比检验和Bootstrap方法对3类模型给出的VaR和ES动态风险测度效果进行了返回测试。实证研究表明:基于GARCH-EVT模型给出的各类基金动态VaR和ES风险测度结果更为准确,意味着将GARCH模型与极值理论相结合,能够实现极端风险的准确测度;ES风险测度比VaR风险测度更保守,在测度极端风险时,应采用ES风险测度作为VaR风险测度的补充。

Considering a class of financial assets which have more obvious extreme risk characteristics, and tak- ing equity fund, hybrid fund and bond fund as research subjects, the performance of dynamic extreme risk measure estimated by RiskMetrics model, GARCH model and GARCH-EVT model are comparatively stud- ied, and the likelihood ratio test and Bootstrap method are used to backtest dynamic VaR risk measure and ES risk measure estimated by three models respectively. The empirical results show that the results of dynamic VaR risk measure and ES risk measure of each fund estimated by GARCH-EVT model are more accurate, which implies that extreme risk may be measured through the integration of GARCH model and extreme value theory; ES risk measure is more conservative than VaR risk measure and should be used as a supplement to VaR risk measure when measuring the extreme risk.