A0489
Title: A Bayesian analysis in long- and short-term financial volatility components with mixture distributions
Authors: Edward Meng-Hua Lin - Tunghai University (Taiwan) [presenting]
Abstract: In forecasting market volatility studies, the consideration of incorporating external information associated with volatility components is an important and challenging issue. From related literature on volatility models, there is much evidence to show that using intraday range data as the volatility measure will construct more accurate volatility forecasts than using daily returns. Therefore, we will investigate whether using the mixed frequency data to incorporate the threshold conditional autoregressive range model with mixture distributions, which could capture the long- and short-term volatility components can improve the prediction ability of the volatility model. We conduct the following issues: (1) to propose a nonlinear range-based volatility model incorporating the long- and short-term volatility components to take volatility forecasting. (2) Considering the mixture GB2 distribution is ordered to capture more features of the implied volatility. (3) We employ the Bayesian approaches to estimate the different types of unknown parameters of the proposed nonlinear model simultaneously. (4) The GB2 density could be represented in various flexible distributions, then we conduct simulation studies to explore the effect of forecasting among its change of shape parameters. (5) We explore whether the long- and short-term volatility components and mixture distribution can improve the forecast performance through empirical analysis.
