The estimation and forecast of portfolio market risks is always a very important aspect of risk manage-ment. This paper employs the realized co-variance matrix model, DCC-MVGARCH model, RiskMetrics model and multi-variants Orthogonal GARCH model to compare their forecast failure ratios of the value at risk of the Shanghai and Shenzhen stock index portfolio and also compare these models with a dynamic quantile test for the forecasting robust-ness. The results show that the realized co-variance matrix model based on high-frequency prices data can significantly improve the forecast accuracy of the portfolio market risk,and its failure rates are also strictly consistent with the corre-sponding confidence levels. Hence this model has achieved a good balance between high utilization of money and also its risk exposures of portfolio management.%组合风险的估计和预测一直都是风险管理中非常重要的一个方面。本文使用了利用高频数据信息的实现协方差矩阵、DCC-MVGARCH多元波动率模型、RiskMetrics模型和多元正交GARCH模型对沪深两市的指数资产组合风险在险价值的预测失败率进行了对比,并利用动态分位数检验方法对各模型的组合风险测度稳健性进行了对比研究。研究结果证明,基于高频数据的实现协方差矩阵模型能够显著提高组合风险测度的预测精度,且严格符合VaR置信区间所要求的失败率,能够很好地在提高资金使用效率与管理资产组合风险敞口间取得平衡。
展开▼