A0150
Title: Bayesian estimation of multivariate stochastic volatility models using dynamic factors
Authors: Yasuhiro Omori - University of Tokyo (Japan) [presenting]
Abstract: In the stochastic volatility models for multivariate daily stock returns, it has been found that the estimates of parameters become unstable as the dimension of returns increases. We first describe various multivariate stochastic volatility models and discuss how we can overcome the difficulties in the estimation of model parameters and latent variables using Markov chain Monte Carlo simulation. Then, we focus on the model based on the factor structure of multiple returns with two additional sources of information: first, the realized stock index associated with the market factor, and second, the realized covariance matrix calculated from high-frequency data. How to remove biases of realized volatilities and realized correlation matrices are also illustrated. The proposed dynamic factor model with the leverage effect and realized measures is applied to ten of the top stocks composing the exchange-traded fund linked with the investment return of the S\&P500 index and the model is shown to have a stable advantage in portfolio performance. The relative weight of the measurement equation for the realized covariances is found to be larger than that for the daily returns. The estimates of leverage effect and the correlation coefficients among asset returns are found to be higher without realized covariances, which is considered to be the bias due to insufficient information.
