两类多元GARCH模型的预测绩效和组合VaR的研究

运用V aR模型对股票组合进行风险测度的关键之一是得到组合条件协方差矩阵.而经典的多元GARCH模型来求解波动率面临着估计参数过多,计算量庞大的问题.因此,使用正交GARCH模型和CCC模型来估算波动率,并以沪深两市A股市场上四个行业的65只股票为样本,使用RM SE和M AD指标比较这些模型的预测能力,求得股票组合的V aR,得出前者效率高和后者预测能力略高的结论.

The Forecast Performance of Two Style GARCH Models and Portfolio VaR Computation

The conditional covariance matrix plays a critical role in portfolio VaR computation. However, in the classical multivariate GARCH model large-scale portfolio is highly parameterized and difficult to estimate in practice. So this paper uses O-GARCH model and CCC model to calculate the volatilities of portfolio. Then 65 samples of stocks in A stock market of Shanghai and Shenzhen is examined by these models. Finally, we follow two criteria to judge the quality of the volatility forcasts and compute portfolio VaR by the conditional covariance matrix.