得分驱动时变DCC模型及其应用

Score-driven Time-varying DCC Model and Its Application

收益率协方差矩阵在投资组合和风险计量中有重要应用。文章先用得分驱动方法推导出DCC模型的等价形式,以此表明DCC模型的观测驱动特性;然后再次采用得分驱动方法将DCC模型的参数时变化,得出加速DCC(aDCC)模型。随机模拟及实证研究显示,加速DCC模型不仅具有更好的波动拟合效果,而且基于其构建的最小方差组合还能实现更小的组合方差及夏普比率。

Yield covariance matrix has important applications in portfolio and risk measurement.In the paper,the equivalent form of DCC model is deduced by the score-driven method,showing the observed driving characteristics of DCC model.And then,the time-varying parameters of DCC model are changed by the score-driven method again,and the accelerated DCC(aDCC)model is obtained.Stochastic simulation and empirical study show that the accelerated DCC model not only has better fluctuation fitting effect,but also can achieve smaller combination variance and Sharpe ratio based on the minimum variance combination constructed by it.