|
Activity Number:
|
497
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Thursday, August 10, 2006 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #305924 |
|
Title:
|
Bayesian Selection of Multivariate Stochastic Volatility Models
|
|
Author(s):
|
Antonello Loddo*+
|
|
Companies:
|
University of Missouri-Columbia
|
|
Address:
|
Department of Statistics, Columbia, MO, 65211,
|
|
Keywords:
|
multivariate stochastic volatility models ; stochastic search model selection ; Gilks-Wild adaptive sampler ; particle-filters ; modified Cholasky decomposition
|
|
Abstract:
|
We consider a multivariate regression model with exogenous and endogenous explanatory variables and time-varying volatilities in the error term. Volatilities are of unknown nature and may be deterministic or stochastic. We propose Bayesian stochastic search Markov chain Monte Carlo (MCMC) algorithms for restrictions on the regression and volatility equations. Efficient parameterization of the time-varying covariance matrices is obtained through modified Cholesky decomposition. We compare two algorithms for volatility simulation, the Gilks-Wild algorithm, and particle filters. We propose a hierarchal approach for selection of the volatility equation's variance components. Numerical simulations show the proposed methods are effective and that they improve the forecasting performance of the model. Empirical applications shed new light on two macroeconomic problems.
|
