基于CARR-EVT整体方法的动态日VaR和CVaR模型研究

Modeling Daily VaR and CVaR by Integrating CARR and EVT Models

本文同时使用日VaR和CVaR模型可实现对市场风险的双重监控,其估计一直是风险管理的重点。科技的发展使获得高频数据成为可能,由于其包含丰富波动信息,学者们开始利用它来研究日VaR和CVaR的估计问题。本文通过整合基于高频数据的CARR模型和非参数的极值理论EVT,实现对日VaR和CVaR的动态估计。上证和深圳综指的实证结果表明,与基于日度数据GARCH类模型和未与极值理论整合的CARR模型的VaR和CVaR相比,本文方法极大提高了估计的准确性,同时所得估计具有不受新息分布影响的稳健性。

Using daily VaR and CVaR simultaneously can doubly control mar ket risks. Their estimations are the key point of risk management. The rapid devel opment of science and technology makes it possible to obtain high frequent data. Because they have much information about volatility, scholars begin to study daily VaR and CVaR based on high frequent data. In the paper, we are to estimate daily VaR and CVaR by integrating the CARR model and extreme value theory. The em pirical analysis on Shanghai and Shenzhen composite indexes showes that, com pared with type of GARCH models based on daily data and CARR model not in tegrating extreme value theory, our method remarkably improves the accuracy of VaR and CVaR estimators. Moreover, the two estimators are robust to the distribu tion of new information.