The Modeling And Applied Research Of The Volatility Of China Stock Market Based On Multiplicative Error Model

In financial markets, the volatility is a very important factor to measure the financial risk.If a trader expects to get a higher payoff and control the risk, it is necessary to study theproperties of the volatility. So it is particularly important to accurately predict the volatility.At present, the study for the high frequency data mainly concentrated on the four methodsas following: the extension of ARCH models, basing on "Realized Volatility"(RV)theoretical model, the Autoregressive Condition Duration Models (ACD) andMultiplicative Error Models (MEM). MEM as the common development model of ARCHmodels and ACD models, it is worthy of further research on it. In the existing literature inChina, the research on the volatility of high frequency data mostly concentrated in theformer three models, few multiplicative error models in this area. In view of this, this paperproposes the MEM models to use for the prediction of the volatility of high frequency datain Chinese stock market, and it is also one of the innovations of the paper.Based on the above study, this paper takes HS300index as the research object andstudies the statistical characteristics of high frequency data in Chinese stock market. Theempirical test shows that the high frequency data in Chinese stock market has suchcharacteristics: high-peak and fat-tail, self-correlation, long-memory and leverage effect,etc. According to the statistical characteristic of high frequency data in China, we establishTARCH model for leverage effect, ARFIMA model and MEM for the realizevolatility.Then we use the above three models to forecast the volatility and compare theforecasting accuracy. The result shows that the prediction effect of the MEM model is thebest, the second is ARFIMA model. Finally, we apply MEM model to the calculation ofVaR value, then return to test its risk prediction effect.