Gibbs samplers for a set of seemingly unrelated regressions

Abstract

Bayesian estimation of a collection of seemingly unrelated regressions, referred to as a 'set of seemingly unrelated regressions' is considered. The collection of seemingly unrelated regressions is linked by common coefficients and/or a common error covariance matrix. Gibbs samplers useful for estimating posterior quantities are described and one of them is applied to a set of linear expenditure functions to estimate household equivalent scales. © 2006 Australian Statistical Publishing Association Inc.