The study, modelling and implications of realised volatility for Chinese stock index futures and spot markets

摘要

Realised volatility is a recently developed measure (Andersen et al., 2001), and it hasattracted the attention of numerous economic researchers. This thesis aims to explore how realised volatility can be applied in Chinese capital markets in the area of shares and stock index futures, to investigate the applicability of an optimal realised volatility forecast model and to examine the implications of realised volatility on optimal hedge ratio and Value-at-Risk (VaR) performance.The empirical results indicate four important realised volatility characteristics of the selected markets. First, the optimum data frequency intervals for applying realised volatility models are equal to five minutes. Second, there is empirical support of the hypothesis that daily volatility jumps exist, and that there are significant intraday volatility jumps and periodicity effects, where logarithm realised volatility shows resilient long memory characteristics. Two important issues (volatility transmission and Markov regime-switching) are examined before modelling realised volatility. Third, based on these results, this thesis proposes, for the first time in economic history, a Heterogeneous Autoregressive-Jumps- Markov Regime-switching (HAR-J-MS) model, which combines the daily volatility jump components and regime-switching effects. The empirical results indicate the superior forecasting power of this new proposed model. Fourth, the empirical results suggest realised volatility performs better on optimal hedge ratio and VaR compared to other models.This thesis contributes to the current literature in four respects. First, it provides fresh and timely evidence on the features of both the spot and futures financial markets in the largest emerging economy: China. Second, this thesis not only investigates China’s financial markets from the traditional perspective of conditional volatility, but also from the relative new perspective of realised volatility (Andersen et al., 2001). Third, it investigates volatility spillover by using both intraday and daily data. Fourth, in terms of methodology, this thesis proposes, for the first time, a HAR-J-MS model to combine the influence of daily volatility jumps and a Markov regime-switching based on a HAR framework, which constitutes a methodological innovation.Overall, this thesis is a comprehensive research paper on realised volatility. To the best of the author’s knowledge, there are few studies that apply realised volatility on Chinese stock index futures and spot markets. This thesis fills this gap in the literature.