Range-based Volatility Modelling, Forecasting and Spillovers

Given the advantages of using the high-low price range to reflect price movement information, this thesis aims to investigate the forecast performance of univariate and multivariate range-based volatility models and applies these models to study volatility spillovers during Chinese stock market turbulence. The first study compares the predictive ability of six parametric range-based volatility estimators, including the vanilla and asymmetric conditional autoregressive range(CARR) model, the simple, component and fractionally integrated range-based exponential generalised autoregressive conditional heteroscedasticity (REGARCH) models, and the cyclical model. I assess the performance of these models in three dimensions: 1) three assets classes (i.e., currencies, stocksand commodities); 2) six forecast horizons (i.e.,1, 5, 20, 60, 120 and 240 days); and 3) four evaluation methods. The fractionally integrated REGARCH (FIREGARCH) model outperforms the competitors in most of the cases. The second study extends the REGARCH model into the multivariate context by combiningit with the co-range model. Two portfolios, currencies (i.e., GBP/USD and EURO/USD) and E-mini futures (i.e., S&P 500 futures and crude oil), over the period 2002-2015 are considered. The novel co-range estimator is superior to the hybrid covariance estimators, dynamic conditional correlation (DCC) and hybrid exponential weighted moving average (EWMA). In particular, the co-range based REGARCH has striking performance in forecasting assets covariation over one-, two-and four-week horizons. Finally, I investigate the impact of Chinese stock market turbulence on seventeen equity markets across the world over the period 12 June 2015 to 29 January 2016. I employ the FIREGARCH model to be incorporated into the Diebold and Yilmaz (2012)index to study the volatility transmission from China to the global markets. In addition, the DCC-FIREGARCH model is applied to study the time-varying correlation of international markets. Results show that the Chinese stock market is still primarily a volatility receiver, whereas US is the largest volatility contributor. Meanwhile, Hong Kong and Taiwan play important roles as intermediates by transmitting volatility from mainland China to the rest of the world.