Modelling the Volatility of Stock Markets

Abstract

The main goal of this thesis is to identify sudden breaks in volatility and determine an impact of these breaks on the volatility of stock market. For the propose of this thesis, we use weekly time series of Chinese and U.S. stock markets, covering the period of 2000 ~2014 years. And then, the primary aim will be fulfilled by 3 steps. In first stage, in order to detecting sudden breaks in both stock markets, we will utilize data samples with a help of the Iterated Cumulative Sums of Squares or ICSS algorithm. In this thesis, Chinese stock market is represented by Shenzhen Component Index, while American stock market is approximated by Dow Jones Industrial Average or DJIA Index. Next step, we will evaluate an impact of sudden breaks on volatility persistence or long memory, by using General Autoregressive Conditional Heteroskedasticity (GARCH) mdoel and Fractional Integrated General Autoregressive Conditional Heteroskedasticity (FIGARCH) models. Finally, there will be an in-sample test for fitting degree. And the forecasting ability would be presented as the performance of comparison between actual variances and expected variances, which is created by above-mentioned conditional heteroskedastic variance models.

The main goal of this thesis is to identify sudden breaks in volatility and determine an impact of these breaks on the volatility of stock market. For the propose of this thesis, we use weekly time series of Chinese and U.S. stock markets, covering the period of 2000 ~2014 years. And then, the primary aim will be fulfilled by 3 steps. In first stage, in order to detecting sudden breaks in both stock markets, we will utilize data samples with a help of the Iterated Cumulative Sums of Squares or ICSS algorithm. In this thesis, Chinese stock market is represented by Shenzhen Component Index, while American stock market is approximated by Dow Jones Industrial Average or DJIA Index. Next step, we will evaluate an impact of sudden breaks on volatility persistence or long memory, by using General Autoregressive Conditional Heteroskedasticity (GARCH) mdoel and Fractional Integrated General Autoregressive Conditional Heteroskedasticity (FIGARCH) models. Finally, there will be an in-sample test for fitting degree. And the forecasting ability would be presented as the performance of comparison between actual variances and expected variances, which is created by above-mentioned conditional heteroskedastic variance models.