Research On Modeling And Forecasting Of Stock Market Volatility

The stock market is an important part of the financial capital markets of many countries and has become an important indicator reflecting the national economy.The volatility of the stock market affects the rationality of resource allocation,the investment of the majority of shareholders and the financing of enterprises,and even the financial crisis.There is a close relationship between stock market volatility and economic policy uncertainty.Changes in economic policies may cause rising risks and affect the stable development of the country's economy.Therefore,it is of great significance to study the impact of economic policy uncertainty index on stock market volatility.In traditional econometrics,GARCH models are mostly used to predict stock market volatility,but this type of model strictly restricts the data with same frequency,which has great disadvantages.The data of macro variables are mostly monthly and quarterly data,which are different from daily stock data.The GARCH-MIDAS model proposed by Engle et al.(2013)improved.It decomposes the conditional variance into long-term and short-term components,can consider both high-frequency and low-frequency data.However,the GARCH-MIDAS model still uses daily data and does not use the effective information of intra-day highfrequency data.The Realized GARCH model proposed by Hansen et al.(2011)uses a measurement equation to jointly model the daily rate of return data and the realized measurement calculated from intra-day high-frequency data.Therefore,this paper improves the GARCH-MIDAS model on the basis of existing research,combines the advantages of the GARCH-MIDAS model and the Realized GARCH model,constructs the Realized GARCHMIDAS family model,and gives the specific construction process and parameter estimation methods of the model.Furthermore,this article also considers the impact of global economic uncertainty index on stock market volatility,introduces it into the GARCH-MIDAS family model,and constructs a new type of extended GARCH-MIDAS model: Multi-mixing Realized GARCHMIDAS-X model and Multi-mixing Realized EGARCH-MIDAS-X model.The advantage of this type model is that it can jointly model three frequency data of daily return rate data,realized measurement calculated from intra-day high-frequency data,and monthly lowfrequency macroeconomic variables.Finally,the constructed model is applied to the Dow Jones Industrial Average Index volatility prediction research.The results of the in-sample effect and out-of-sample volatility prediction test show that: the multi-mixing Realized EGARCH-MIDAS-X model with highfrequency information and macroeconomic variables added Can effectively improve the stock market volatility fitting and forecasting capabilities,And a variety of robustness analysis has also verified this point.