基于DCC-GARCH模型的协方差矩阵预测  

Covariance Matrix Prediction Based on DCC-GARCH Model

作　　者：周亮[1,2] Zhou Liang(Hunan University of Finance and Economics,Changsha 410205,China;Business School,Hunan Normal University,Changsha 410081,China)

机构地区：[1]湖南财政经济学院,长沙410205 [2]湖南师范大学商学院,长沙410081

出　　处：《统计与决策》2021年第20期35-38,共4页Statistics & Decision

摘　　要：文章利用DCC-GARCH模型对大类资产或股票行业指数间的协方差矩阵进行预测,并通过矩阵预测误差及组合风险跟踪误差将其与其他常见模型进行比较。结果显示:DCC模型能够显著提升协方差矩阵的预测准确性及降低最小化方差投资组合的风险跟踪误差,考虑了非对称性的ADCC模型无法进一步提升DCC模型的预测效果;DCC模型的预测能力在任何市场状态下均优于其他模型,在牛市期及低情绪期的表现更为突出。总而言之,DCC模型考虑了资产间相关性的动态变化特征,用它来预测协方差矩阵并进行投资组合构建是更为有效的。This paper uses the DCC-GARCH model to predict the covariance matrix between large classes of assets or stock industry indexes,and compares it with other common models through matrix prediction errors and portfolio risk tracking errors.The results show that DCC model can significantly improve the prediction accuracy of covariance matrix and reduce the risk tracking error of minimized variance portfolio,while the ADCC model considering the asymmetry cannot further improve the prediction effect of DCC model,and that the predictive ability of DCC model is superior to other models in any market state,especially in bull market and downturn.In a word,DCC model considers the dynamic characteristics of the correlation between assets,and it is more effective to use DCC model for the prediction of covariance matrix and the construction of portfolio.

关 键 词：DCC 协方差 风险管理 风险预测 投资组合 

分 类 号：F832.5[经济管理—金融学]