A Research Of Realized Volatility Based On High-frequency With Macroeconomic Voriables

There has been a long history for the research of the volatility of the financial market. There also has been a lot of research on the relationship between macroeconomic variables and stock market volatility, but the models of realized volatility based on the industry index and stocks are still rare. As the importance of the yield and volatility in the financial market, it is good for the development fo financial market to study the volatility of stock market. This paper choose Industry Index-Shanghai Industrial Index and its constituent stocks based on the five minutes of high frequency datas, calculate its realized volatility, and then isolate the continuous fluctuations and jumps volatility, analyze the characteristics of the jumps volatility of the Industry Index and its constituents. At the same time, we choose the most relevant industrial sectors and macroeconomic variables, after standardizing and taking the main component of them,we take the two main component fo the macroeconomic variables into the realized volatility model-HAR-RV-CJ model,get the HAR-RV-CJ-MEV model. Then we compare the fitting results and predictive capability of these two models. The results show that, in the empirical period,the number of days of the jump in Industry Index is far less than the average number of days of that in constituents stocks. The capability of fitting and forecasting of HAR-RV-CJ-MEV model is better than the HAR-RV-CJ model both in the short term(h=1), medium(h=5) and long-term(h=22), this results confirm that macroeconomic variables take a significant effect on the realized volatility of industrial index.Then, we model for the different realized volatility of different stocks. Due to the differences in the impact of individual information, we take the trading volume into the HAR-RV-CJ-MEV model. The results show that the model which has trading volume take a better capability in fiiting and forecasting of realized volatility.Finally, we predict the realized volatility of industrial index and individual stocks with HAR-RV-CJ, HAR-RV-CJ-MEV, HAR-RV-CJ-MEV-V and HAR-RV-CJ-V models. The results show that HAR-RV-CJ-MEV model has the best ability of forecasing in the four models, using SPA test.While for 20 stocks, there are different performance for the four models to predict the realized valotility, microeconomic variables play a positive role in forcasting the realized valitility of 20 stocks. HAR-RV-CJ-V would be a better choice in some cases, in which they are effected by its own information.