一类包含不同权重函数的混频GARCH族模型及其应用研究

A Kind of GARCH-MIDAS Model with Different Weighting Functions and Their Application Research

研究目标:构建包含不同权重函数的混频GARCH族模型,给出权重分布性质证明,通过中国股市波动度量和预测进行模型比较与评估。研究方法:将多种权重函数纳入混频GARCH模型,给出离散时间下的权重分布性质,基于准极大似然法对模型进行估计。研究发现:Beta权重函数、指数阿尔蒙多项式和扩展自回归权重函数具有较好的样本外预测效果,指数阿尔蒙多项式和扩展自回归权重函数能形成更多样化的权重分布;宏观经济变量能够显著影响未来股票市场波动,其中实际GDP和工业增加值对股市波动的影响具有较长滞后性,货币供应量能持续影响未来股市波动;对于不同宏观经济变量应匹配对应的权重函数以实现最优预测效果。研究创新:构建并扩展了包含不同权重函数的混频GARCH族模型,给出了权重分布性质证明,基于中国宏观经济和股市数据讨论了模型的稳健性和适用性。研究价值:丰富并扩展了混频GARCH模型,实证研究有利于投资者更深刻理解股市波动的驱动因素,有利于监管部门更好进行风险预警和防范。

Propose GARCH-MIDAS models with different weighting functions,provide the weighting distribution properties proves,and empirically conduct the model comparisons.Research Methods:Propose the GARCH-MIDAS model with different weighting functions,and provide the distribution properties,and estimate the model with QMLE.Research Findings:Based on out-of-sample analysis results,Beta,Exponential Almon Polynomial and Augmented AR weights perform well,and Exponential Almon Polynomial and Augmented AR weights could produce various weights distributions.This paper finds that macroeconomic variables could significantly predict future market volatility by GARCH-MIDAS model with different weighting functions,and real GDP and Industrial Production have a lag effect on the market volatility,M 2 has long-term impact on the volatility.We could choose the weighting function based on the macroeconomic variable.Research Innovations:Form GARCH-MIDAS type models with different weighting functions,provide the weighting distribution properties,and provide the evidence of China on the relationships between macroeconomics and market volatility,and discuss the fitness of the models.Research Value:Help the investors better understand the driving forces of market volatility,better avoid risks and allocate the assets,and help the alarm of risks and preventions.