Efficient Bayesian Estimation of a Multivariate Stochastic Volatility Model with Cross Leverage and Heavy-Tailed Errors

An efficient Bayesian estimation using a Markov chain Monte Carlo methodis proposed in the case of a multivariate stochastic volatility model as anatural extension of the univariate stochastic volatility model with leverageand heavy-tailed errors. Note that we further incorporate cross-leverageeffects among stock returns. Our method is based on a multi-move samplerthat samples a block of latent volatility vectors. The method is presentedas a multivariate stochastic volatility model with cross leverage and heavytailederrors. Its high sampling efficiency is shown using numerical examplesin comparison with a single-move sampler that samples one latent volatilityvector at a time, given other latent vectors and parameters. To illustrate themethod, empirical analyses are provided based on five-dimensional S&P500sector indices returns