Study On The Impact Of Macroeconomics Factors On The Volatility Of Stock Price Based On Improved GARCH-MIDAS Model

There are two phenomena in classical econometric research methods and contents.On the one hand,the data used in the research are in the same frequency.On the other hand many researches take the same frequency data as the research indicators when studying the impact of macroeconomics factors on the volatility of stock price.It makes the same frequency data between low frequency data and high frequency data.When processing the different frequency data,researches often convert the high frequency data to the low frequency data,which likely loses the effectiveness of mixed data containing useful information.This paper takes the mixed frequency data as the research indicators in this case which contains both low frequency and high frequency data.Traditional model of same frequency can't explain the causal relationship between macroeconomic variables and Chinese stock market volatility due to the limitation of data frequency.The classic model of generalized auto regressive conditional heteroskedasticity proposed by Engle et al.(2013)that contains the hybrid Sample technique can decompose the high-frequency volatility into short-term and long-term components.And it ingeniously deals with the macroeconomic factors as long-term volatility factor,which makes the data more useful.However the improved GARCH-MIDAS model in this paper is different from the classic GARCH-MIDAS model.The improved GARCH-MIDAS model considers the exchange rates as high frequency data into the model and uses the realized volatility of exchange rates to explain the stock market long-term fluctuations,as expands the classic GARCH–MIDAS model.In the empirical study,it establishes the single factor and multi-factor GARCH-MIDAS model to analyze Chinese stock market price volatility respectively from level and volatility aspects.This paper chooses the monthly M1 and CPI and daily exchange rates to research the stock volatility.The research results show that both of the level and volatility of money supply value have significant positive influence on Chinese stock market volatility which matches the market performance in our country.The level of the consumer price index has a significant negative correlation with Chinese stock market volatility while the volatility of it has no correlation with Chinese stock market volatility.The horizontal and volatility of the exchange rates both have the significant negative correlation relationship with the stock market fluctuations.The results of multi-factor model are the same as the results of single-factor model.But since the multi-factor models have more parameters,there may be an excessive parametric problem that possibly leads to some coefficients not significant.At the same time,the single factor and multi-factor model both have very strong prediction ability and the multi-factor level model is stronger than single factor level model in prediction.In addition,the comparison between multi-factor model and single factor model finds that the multi-factor model can explain the long-term Chinese stock market fluctuations better.It puts forward the following policy suggestions,including improving macro-control policy,developing healthy stock market,guiding rational investment and perfecting disclosure mechanism.