基于DMA方法和异质市场理论的中国股市波动性研究  

Chinese Stock Market Volatility Research Based on DMA Method and Heterogeneous Market Theory

作　　者：朱奇锋 吴恒煜[2] ZHU Qi-feng;WU Heng-yu(School of Economics and Information Engineering,Southwestern University of Finance and Economics,Chengdu 611130,China;School of Management,Jinan University,Guangzhou 510632,China)

机构地区：[1]西南财经经济信息工程学院,四川成都611130 [2]暨南大学管理学院,广东广州510632

出　　处：《数理统计与管理》2021年第2期366-380,共15页Journal of Applied Statistics and Management

基　　金：国家自然科学基金项目(71171168,71601125)。

摘　　要：根据Raftery等提出的动态模型平均(DMA)方法结合Müller的异质市场理论,同时考虑市场的量价关系,构建了DMA-HAR-RV、DMA-LHAR-RV和DMA-LHAR-RV-T模型,利用以上模型对上证综合指数不同期限的已实现波动率进行预测,并进行了模型置信集(MCS)检验.发现对于不同的损失函数和预测期限DMA-LHAR-RV-T模型预测效果最稳定;异质结构的波动率、正、负收益和换手率对未来波动都有显著的解释效果,减少上述任一类型的异质预测变量都会降低模型的预测效果;根据预测变量系数的时变性发现上证综指在三个时间段杠杆效应不显著或者存在反向杠杆效应,不同期限的滞后波动率对未来波动率的预测能力呈现此消彼长的特点.According to DMA(Dynamic Model Averaging)model proposed by Raftery,Miller's heterogeneous markets theory and the relationship between volume and price,we construct DMA-HAR-RV、DMA-LHAR-RV and DMA-LHAR-RV-T model,and then use these models to forecast volatilities of SSEC,use MCS(Model Confidence Set)approach to test the forecasting performance.The result shows that DMA-LHAR-RV-T model is a robust model for diferent loss functions and forecast horizons.Heterogeneous volatilities,positive,negative returns and turnovers have significant explanatory effects on future volatilities.Reducing any type of heterogeneous predictor reduces the forecast performance of the model.In our models,coefficients of the predictors are time-varying,for SSEC,we find that leverage effect is not significant or inverse during 3 periods,and the forecast abilities of the heterogeneous volatilities is that,one falls,others rise.

关 键 词：波动率 动态模型平均 异质市场理论 杠杆效应 

分 类 号：F830[经济管理—金融学] O212[理学—概率论与数理统计]