Title: SHARP: A state-space HAR model with particle GIBBS sampling
Authors: Aya Ghalayini - Lancaster University (United Kingdom) [presenting]
Marwan Izzeldin - Lancaster University Management School (United Kingdom)
Mike Tsionas - Lancaster University (United Kingdom)
Abstract: The aim is to propose a general state-space autoregressive (AR) model with time-varying coefficients that follow an AR process with stochastic volatility. We implement these new specifications in the HAR framework to capture the time-varying salient feature of volatility using a two-state representation via a) allowing the time-varying coefficients to follow an AR(1) specification. b) introducing stochastic volatility for the innovations of the coefficients. Using high-frequency data of the SPY-ETF and representative NYSE stocks from 2000 to 2016, we show that the proposed model estimated using particle Gibbs sampling consistently outperforms different HAR model specifications in forecasting financial volatility.