Based On The Research On The Asymmetric Relationship Between High-frequency Data Return And Volatility In China's Stock Market

The asymmetry between return and volatility is one of the typical characteristics of stock return,which is manifested in the negative correlation between return and volatility.It is quite challenging to provide a convincing explanation for the asymmetry of return and volatility,and the explanation of this asymmetry has been controversial.This paper introduces two classical explanations about the asymmetric relationship between return and Volatility: leverage effect and volatility feedback effect.The first is leverage effect,that is,the decline of asset price increases the possibility of financial leverage and bankruptcy,making assets more risky,thus increasing volatility.When applied to stock index,this idea is transformed into dynamic leverage effect;the second is volatility feedback effect,the expected growth of volatility will increase the required return,which means that stock price will have a An immediate decline to allow for higher future returns.In this paper,the vector auto-regression model and realized volatility are considered,and the short-term and long-term causality measures proposed by Dufour and tamouti(2009)are used to measure leverage effect and volatility feedback effect.In this paper,we use the 5-minute high-frequency data of Shanghai 50 ETF index to calculate the hourly and daily returns,realized volatility and second power variation.When using daily yield and realized volatility data,significant leverage effect is found in the first six days,and volatility feedback effect is significantly weaker than leverage effect in all time spans.In the prediction of volatility,this paper first adds leverage effect factor and volatility feedback effect factor to the traditional HAR-RV model to predict the volatility,and compares the results of this model with that of the traditional HAR-RV model.It is found that the new model with leverage effect factor and volatility feedback effect factor is significantly better than the traditional model in the prediction of volatility OK.Then,this paper considers the linear realized GARCH model and the realized GARCH model with leverage effect variables.In the empirical results of daily data,it is found that the realized GARCH model with leverage effect variables is superior to the original realized GARCH model in both the fitting effect of the model and the prediction performance outside the sample of volatility.