This thesis comprises two chapters with a focus on volatility estimating, modeling and forecasting using intraday data in the Chinese stock market. The first chapter explores the performance of two types of estimators in volatility prediction: the realized volatility (RV) type and duration-based ones. This is motivated by the theoretical and empirical support for both categories of estimators that are distinct from each other. I use intraday data for 203 component stocks in the CSI 300 index and adopt a combination of volatility models and these two types of estimators to produce 1-, 5- and 22-day ahead forecasts. I show that, although empirically more efficient with US data, the duration-based volatility estimators fail to compete statistically with the traditional RV-type although in a portfolio setting both types of estimators generate similar economic value to a mean-variance investor. A comprehensive simulation exercise is undertaken to rationalize the poorer statistical performance of duration-based estimators.
In the second chapter, I use daily and intraday data to examine the impact of crosssectional return dispersion on volatility forecasting in the Chinese equity market. I adopt traditional GARCH and HAR models and, by augmenting them with return dispersion measures, provide evidence that the return dispersion exhibits substantial information in describing the volatility dynamics by generating significantly lower forecasting errors at market and industry levels. Furthermore, the information content of the return dispersion tends to offer economic gain to a mean-variance utility investor. The findings are robust with respect to alternative volatility proxies and weighting scheme in constructing industry indices.
|Date of Award||8 Mar 2020|
- Univerisity of Nottingham
|Supervisor||Xiaoquan Liu (Supervisor), Alan Wen (Supervisor) & Xiafei Li (Supervisor)|
- Duration-based Estimator
- Return dispersion
Risk, return, and investor behavior in the Chinese equity market
FEI, T. (Author). 8 Mar 2020
Student thesis: PhD Thesis