Analysis Of Financial Market Volatility Based On High Frequency Data

Volatility as a foundamental attribute in the financial market once again becomes the focus of practice and academic circles since the global financial crisis in 2008.In recent years,the rapid development of computer and communication technology leads to the increase in the availability of high frequency data,which allows people to further study the dynamic characteristics and causes of volatility,and to better predict volatility.In general,the information in the financial market continuously affect the securities’ price,the lower the data collection frequency,the more information loss;and vice versa.China as the second largest economy in the world,its volatility in financial markets has a significant impact on donemic and even international economic markets.Since the share reform in 2005,the pace of financial reform in China speeds up,and a series of reform activities directly promote the development of the financial market and national economy.However,the current development situation of our country stock market shows that the volatility is still large,and there is a big market risk.In this context,how should we better model and forecast the volatility of the stock market? Under the T+1 trading system in the Chinese stock market,how does the information flow outsider trading hours affect the stock price volatility? China’s capital market is in a stage of rapid development,is the refinancing beneficial to the stability of the stock market? The stock index future prices have a role in the expection of relation between supply and demand and the stock price trend in the stock market,how is the volatility spillover effect beween the stock index future and spot markets? Financial volatility is closely related to the risk,and moveover,how should we predict and assess the financial risk? The goal of this paper is to do with these questions.In this thesis,we will address these questions in two ways.First,we will investigate the volatility forecasting and what causes the changes in volatility.Second,we investigate the volatility transimission and risk in financial markets.In Chapter 3,we propose MS-HAR familly model and use it to predict the volatility of the Shanghai and Shenzhen 300 index futures.The advantage of this model is that it not only can describe the long memery characteristics of volatility,but also can describe the state transition of volatility,and allows the state transition probability to be time-varying.The empirical results show that,compared with the GARCH-type,HAR-type and ARFIMA models,MS-HAR family model has better prediction performance.In Chapter 4 and Chapter 5,we respectively study what causes the changes in volatility for the stock index futures market and the stock market.The findings show that in the stock index futures market and the stock market,there are strong leverages effects and extra leverage effects captured by trading yields outsider trading hours,and trading volume has significantly positive effects on volatility;besides,securities lending can inhibit the volatility in financial markets.It is worth noting that in the stock markets the squared yields can capture the additional volatility,and the inhibitory action of refinancing reflects the differences of enterprise and industry;finally,the refinancing can significantly reduce the possibility of the price jumps.In Chapter 6,we use VHAR model to study the volatility transmission relationship between stock index futures market and the spot market.The empirical results show that there is a two-way transmission relationship in volatilities between the stock index futures market and the spot market.Generally,the existing research only analyze static volatility transmission relationship.However,in this chapter we analyze this relationship by the DCC model and Andrews test,and find that there exists time-varying correlations and structure breaks in the volatility transmission relationships.In Chapter 7,we investage the risk prediction and evaluation in stock index futures market by MS-HARJ model.First,we compare the simulation parameter method and filtering history simulation method and get that the normal distribution assumption of standardized yields is more reasonable.Second,we compare the ARFIMA and FIGARCH model with MS-HARJ model and get that MS-HARJ model has better risk prediction performance.These results have important reference value to risk managers.