CFE 2015: Start Registration
View Submission - CFE
A1442
Topic: Contributed on Modelling financial volatility Title: Bayesian semiparametric modeling of multivariate stochastic volatility Authors:  Martina Danielova Zaharieva - Erasmus University Rotterdam (Netherlands) [presenting]
Abstract: The proposed model is a multivariate stochastic volatility model (MSV), in which the errors are modeled as an infinite scale mixture of multivariate Gaussian distributions. A Bayesian non-parametric approach, in particular a Dirichlet process mixture (DPM), is adopted. This allows for highly flexible modeling of the return distribution with respect to the kurtosis. The structure of the MSV model is based on Cholesky decomposition, which simplifies the estimation of the latent volatility states and allows the use of standard filtering methods. For the DPM part of the model an efficient sampling algorithm related to the slice sampler is proposed. Finally, an empirical application regarding volatility shock transmissions between different markets is applicable.