Prediction Of Chinese Stock Market Volatility Based On Herding Effect

Volatility modeling and prediction in financial markets has always been a hot academic topic in financial circles.Excessive volatility of financial market is not only the performance of unbalanced and unstable economic development,but also the inevitable process of self-regulation of financial market.Existing scholars’research on volatility is mainly carried out from the measurement,influencing factors and prediction of volatility to explore the law of volatility in the financial market.High-frequency data is widely used in the study of financial market volatility because it can more fully capture the influence of information on price.Realized Volatility（RV）calculated based on high-frequency data has also become an important index for measuring market Volatility.In this paper,the Heterogeneous auto-regressive（HAR）model proposed by Coris（2009）was used to model and forecast China’s financial market volatility.In order to improve the predictive performance of the model,the market herd effect index is further introduced into the model.The results show that:First of all,the new HAR model can significantly improve the in-sample prediction ability of the model after adding herd effect indicators into the HAR model.Secondly,the out of sample prediction ability of market herding is further investigated.The predictive power of out of sample volatility for the future 1day,2 days,1 week,2 weeks and 1 month was empirically analyzed,and the predictive effect was evaluated by using out of sample²statistics,DM test（Diebold Mariano test）and MCS test（model confidence set test）.The results show that the market sheep index significantly improves the model’s out of sample prediction ability.Moreover,the prediction effect of stock market long-term fluctuation is more obvious.Finally,this paper adopts the replacement of rolling window,recursive window test,replacement of HAR-CJ model and replacement of the index of the predicted object to carry out robustness test,and finds that the conclusions of this paper are still robust.The main significance of this study lies in the following aspects:firstly,the theory of volatility prediction is enriched,and a new index is found to improve the prediction accuracy of stock price volatility;Secondly,from the perspective of behavioral finance,it deepens the understanding of herd effect.In practice,it is beneficial for investors to understand the operation rules of the financial market and allocate assets more rationally.At the same time,it will provide policy suggestions for financial market risk regulation.