基于GARCH-MIDAS的混频投资者情绪对股市波动的影响

Impact of Investor Sentiment on Stock Market Volatility in Hybrid Frequency Based on GARCH-MIDAS

采用广义自回归条件异方差的混频数据抽样模型(GARCH-MIDAS)研究投资者情绪对中国股市收益率波动的影响。实证结果表明:与投资者关注、经济政策不确定性相比,投资者情绪在单因子GARCH-MIDAS模型中优度最佳,对市场波动产生显著的正向影响,并能够解释A股中较高比例的长期波动;投资者关注对市场波动产生显著的正向影响;经济政策不确定性对A股的影响不显著。此外,投资者情绪与投资者关注组合时的双因子GARCH-MIDAS模型可以得出与单因子模型一致的结论,在模型中合理地选取解释变量进行组合可以提高模型对股市波动的解释能力。GARCH-MIDAS模型有效地解决了投资者情绪与股市波动率数据频率不一致的问题,为股市波动率影响因素的研究提供了新的研究视角。

This paper studies the impact of investor sentiment on the volatility of China’s stock market returns using the mixed data sampling model of generalized autoregressive conditional heteroscedasticity(GARCH-MIDAS).The empirical results show that compared with investor attention and economic policy uncertainty,investor sentiment is the best in the single-factor GARCH-MIDAS model,which has a significant positive impact on the market volatility and can explain the long-term volatility of A-share.Investor attention has a significant positive impact on market volatility while the impact of economic policy uncertainty on A-share is not significant.In addition,the two-factor GARCH-MIDAS model with investor sentiment and investor attention can draw the same conclusion as the single-factor model.Reasonable selection of explanatory variables in the two-factor model can improve the ability of the model to explain stock market volatility.The GARCH-MIDAS model effectively solves the problem that the data frequency of investor sentiment is inconsistent with that of stock market volatility,and provides a new research perspective for the study of the influencing factors of stock market volatility.