This paper considers a Bayesian analysis of a multivariate factor asymmetric stochastic volatility model, and proposes an efficient Markov chain Monte Carlo (MCMC) method. The basic multivariate stochastic volatility has been recently extended to consider common factors among asset returns. However, well-known leverage effects in stock markets still have not been considered in the past literature. This paper generalizes the basic multivariate stochastic model by incorporating both leverage effects and common factors. Since the maximum likelihood estimation of such a generalized model is difficult to implement due to many parameters and latent variables, we take a Bayesian approach and use MCMC estimation. The block sampler for latent volatility variables is used to accelerate the convergence of MCMC samples to the posterior distribution.
