基于混频因子模型的高阶矩投资组合策略研究  

Research on the Higher-order Moments Portfolio Strategy Based on Mixed-frequency Factor Model

作　　者：杨冬 赵爽 Yang Dong;Zhao Shuang(School of Statistics,Southwestern University of Finance and Economics,Chengdu 611130,China;Joint Laboratory of Data Science and Business Intelligence,Southwestern University of Finance and Economics,Chengdu 611130,China;School of Economics,Southwest Minzu University,Chengdu 610225,China)

机构地区：[1]西南财经大学统计学院,四川成都611130 [2]西南财经大学数据科学与商业智能联合实验室,四川成都611130 [3]西南民族大学经济学院,四川成都610225

出　　处：《数量经济研究》2023年第4期161-179,共19页The Journal of Quantitative Economics

基　　金：教育部人文社会科学研究青年基金项目“基于混频数据高阶矩波动模型的下行风险预测研究”(20YJC790160);西南财经大学中央高校基本科研业务费专项资金引进人才科研启动资助项目重点项目“基于混频数据的金融高阶矩成分波动建模及其应用研究”(JBK21YJ17)的联合资助。

摘　　要：高阶矩的准确估计对提升包含高阶矩的投资组合表现无疑具有重要意义。传统样本矩估计方法未考虑“维数灾难”对高阶矩参数带来的不良影响,大大限制了高阶矩投资组合在实践中的应用,因此更为精确的高阶矩估计方法有待提出。针对高阶矩矩阵,本文基于混频数据抽样(MIDAS)技术,构建混频因子模型高阶矩估计量,并利用其构建包含高阶矩的投资组合策略,探讨其实际经济价值。在实证分析中,通过采用效用函数的高阶泰勒展开,将基于混频因子模型得到的高阶矩估计量作为最优估计进行投资组合优化,利用回测检验方法对不同相对风险厌恶系数下混频因子模型高阶矩投资组合在收益和下行风险方面的表现进行评价。研究结果发现:基于混频因子模型的高阶矩估计可以有效改善高阶矩投资组合在收益率和下行风险方面的表现,相较于传统的基于样本矩估计和同频因子估计构建的高阶矩投资组合具有显著的优势。Accurate estimation of higher-order moments is undoubtedly of great significance in improving the performance of portfolios containing higher-order moments.Traditional estimation methods do not consider the adverse effects of the“curse of dimensionality”on higher-order moment parameters,which greatly limits the application of high-order moment portfolios in practice.Therefore,more accurate higher-order moment estimation methods are needed.For the high-order moment matrix,this paper constructs the high-order moment estimator of the mixed-frequency factor model based on the mixed-frequency data sampling(MIDAS)technolo-gy,uses it to establish a portfolio strategy containing high-order moments,and discusses its actual economic value.In the empirical analysis,using the high-order Taylor expansion of the utility function,the high-order moment estimator obtained based on the mixed-frequency factor model is used as the optimal estimate for portfolio optimization,and the back-test method is used to evaluate the performance of the high-order moment portfolio.The results show that the estimation of high-order moments based on the mixed-frequency factor model can effectively im-prove the return rate and downside risk of the high-order moment portfolio.Compared with the traditional high-order moment portfolio based on sample moment estimation and traditional fac-tor estimation,it has significant advantages.

关 键 词：混频因子模型 高阶矩 投资组合策略 

分 类 号：F224.0[经济管理—国民经济]