论文部分内容阅读
本文将非线性压缩方法运用到DCC和BEKK模型中,用非线性的压缩估计量代替MMLE估计中初始的样本协方差矩阵,大大提高了高维DCC和BEKK模型的估计效率,并突破性地使得横截面维度大于时间维度时,DCC和BEKK模型的有效估计成为可能。蒙特卡洛模拟发现:非线性压缩方法对于DCC和BEKK模型估计的优化作用显著,且优化程度随着横截面维度和时间维度的比值增大而增加。实证分析进一步说明了非线性压缩方法对于准确估计高维条件协方差矩阵、从而提高组合选择效率的重要作用。
In this paper, the nonlinear compression method is applied to the DCC and BEKK models, and the nonlinear covariance estimator replaces the initial covariance matrix in the MMLE estimation, which greatly improves the estimation efficiency of the high dimensional DCC and BEKK models. When the cross-sectional dimension is larger than the time dimension, a valid estimation of the DCC and BEKK models is possible. Monte Carlo simulation shows that the nonlinear compression method has significant optimization for DCC and BEKK model estimation, and the degree of optimization increases with the ratio of cross-sectional dimension to time dimension. The empirical analysis further illustrates the important role of nonlinear compression methods in accurately estimating covariance matrices of high-dimensional conditions and improving the efficiency of combination selection.