摘要
运用多元的DCC-MVGARCH模型方法对股票投资组合进行VaR测度,并与J.P.Morgan银行采用的IGARCH模型计算结果进行对比。结果表明,在测度VaR方面,无论在1%或是5%置信水平下,DCC-MVGARCH模型均优于单变量IGARCH模型。以DCC-MVGARCH模型测度的VaR为基础,把峰度、流动性风险因素纳入VaR模型框架后,发现拓展后的VaR模型预测风险能力显著增强,在所有拓展模型中,同时考虑了内生性、外生性流动性风险的LAVaR3模型表现最优。
Measuring VaR correctly is the focus of risk management. This paper measures the VaRs of stock portfolio with DCC- MVGARCH model and IGARCH model adopted by J.P. Morgan bank respectively, and compared the performances of the different VaRs. Results indicate that DCC model is superior to IGARCH model in measuring portfolio＇s VaR whether in 5% or 1% credit levels. Based on DCC- MVGARCH model , after incorporating kurtosis factors and liquidity factors in traditional VaR frames, we find that the extended models significantly have stronger abilities of forecasting risk than ever. In all extended models, the LAVaR3 model that incorporating endogenous and exogenous risks simultaneously perform best.
出处
《统计与信息论坛》
CSSCI
2009年第2期64-71,共8页
Journal of Statistics and Information
作者简介
冯金余（1974-），男，江西莲花人，经济师，博士生，研究方向：证券市场与金融工程。
