Realized Volatility Modeling,forecasting And Application In International Stock Markets

Economic globalization has brought continuous growth of the international stock market and intensification of international capital flows.Changes in the international financial market have attracted more and more attention from policy makers,investors and financial researchers.The international stock market,closely related to global economic volatilities,can provide investors and fund seekers with a wide range of trading platforms.In particular,a better understanding of volatility patterns in developed and emerging markets is of great significance for financial decision-making and portfolio diversification.At the same time,the modeling and forecasting of volatility is an important branch in the field of financial econometrics.Volatility has an important impact on option pricing,asset allocation and risk management.It has vital theoretical and practical significance for maintaining the healthy development of the capital market.Moreover,how to further improve the prediction accuracy of volatility and the application research related to volatility prediction have always been one of the hotspots and difficulties in the academic and practical circles.What is more serious is that,under the effect of the global financial crisis,the aggravation of geopolitical risks between regions and the outbreak of global emergencies,the volatility of international stock markets has been increasing.Against this background,the following questions are still worth exploring: can the volatility of international stock markets be predicted? What factors are helpful for the prediction of international stock market volatility? How to effectively apply predictors? Can the prediction of international stock market volatility optimize portfolio? This paper will focus on these important issues.The research content mainly has three parts.The first part is the research on the predictors of realized volatility in international stock markets.The second part is the research on the new volatility prediction models.The third part is the application research based on the volatility prediction.In Chapter 2,we explore in detail the role of leverage,jumps and overnight information in predicting the realized volatility(RV)of 21 international stock indexes.The classic heterogeneous autoregressive realized volatility model and the extended model have been implemented for in-sample and out-of-sample analysis.The empirical results of the in-sample show that these three factors have a significant negative impact on most international stock indexes.The out-of-sample forecast performance based on the MCS test find that leverage and overnight information have stronger predictive power than jumps.The HAR-LJO model can obtain the largest MCS test p-value in most international stock indexes,which means that using these three predictors at the same time can get the best results for RV forecast of the international stock market.In addition,this chapter also examines the long-term forecasting performance of the forecasting model.In almost all international stock indexes,the HAR-LJO model has the strongest forecasting ability.In Chapter 3,we study whether the volatility information of US crude oil and stock markets has predictive power for the RV forecast of the international stock market.Through in-sample analysis,in most international stock indexes,the predictability of RV based on US crude oil and stock market volatility information is statistically significant.Second,adding US-based crude oil(OVX)and stock market volatility information(VIX)to the benchmark model can significantly improve out-of-sample forecasting performance.In addition,we established the HAR-RV-AVERAGE model through the equal-weight combination method.The empirical results show that,except for a few Asian indexes,the HAR-RV-AVERAGE model has almost the best out-of-sample prediction performance.Obviously,the stock market volatility of developed countries benefited more from information of US crude oil and stock market volatility,while the stock markets of emerging countries benefited the least,especially Asian stock indexes.This part the thesis also provides new evidence for the spillover effect of the crude oil market on the international stock market volatility.In Chapter 4,we investigate whether information of implied volatility is more predictive than information of realized volatility.To this end,we constructed eight sets of forecasting models based on the five combination methods and principal component analysis.Based on the empirical results of the out-of-sample forecasts,we find that the information of implied volatility is better than the information of realized volatility.It has a stronger forecasting ability,whether it is the forecast range of the next day or a longer forecast range.This part of the thesis provides a new perspective for the forecast of international stock market volatility.In Chapter 5,we combine Markov switching regime theory and two prevailing shrinkage methods(LASSO and Elastic Net),and two novel volatility prediction shrinkage models are designed,namely MS-HAR-LASSO model and MS-HAR-Elastic model.The purpose of the models is to consider the transition between the high and low volatility of the international equity markets.The forecasting outcomes indicate that the MS-HAR-LASSO and MS-HAR-Elastic models perform better than many competing models(namely,five combination models,principal component models such as HAR-PCA and HAR-PLS,the traditional reduction model such as HAR-LASSO and HAR-Elastic)in most international stock market RV predictions.This part of the thesis supplements the volatility prediction model.In Chapter 6,we use a newest portfolio method proposed by Bollerslev et al.(2018)to test the economic value of volatility prediction.Most of the existing volatility prediction literature is based on a statistical perspective.Compared with the existing literature,this chapter firstly applies this new method to the international stock market volatility forecast research.The empirical results confirm that the use of this forecast model can achieve better economic benefits for investors.Once again proved the theoretical value and practical value of this thesis.