Title: High-frequency stochastic volatility models for the Japanese stock index
Authors: Jouchi Nakajima - Bank for International Settlements (Switzerland)
Toshiaki Watanabe - Hitotsubashi University (Japan) [presenting]
Abstract: A high-frequency stochastic volatility (SV) model is proposed for the Japanese stock index. Apart from the standard daily-frequency SV models, high-frequency SV models are fit to intraday returns by extensively capturing intraday volatility patterns. The proposed model consists of the persistent autoregressive stochastic volatility process, seasonal components of the intraday volatility patterns, and correlated jumps in prices and volatilities. A Bayesian method for the analysis of this model is developed using Markov chain Monte Carlo (MCMC) with the exact multi-move sampler for the SV process. Using this method, the proposed model is applied to the 5-minute returns of Nikkei 225 index. It is also examined whether the high-frequency SV model improves the predictive ability of volatility compared with the commonly-used realized volatility.