University of Melbourne

Bayesian Model Averaging in Consumer Demand Systems with Inequality Constraints

Email address: mchua@unimelb.edu.au

Keywords: Marginal likelihood; Metropolis Hastings Approximation; Modified Harmonic Mean; Kleibergen Procedure.

JEL Classifications: C11, C32, E21

Abstract:
Share equations for the translog and almost ideal demand systems are estimated using Markov Chain Monte Carlo. A common prior on the elasticities and budget shares evaluated at average prices and income is used for both models. It includes equality restrictions (homogeneity, adding up and symmetry) and inequality restrictions (monotonicity and concavity). Posterior densities on the elasticities and shares are obtained; the problem of choosing between the results from the two alternative functional forms is resolved by using Bayesian model averaging. The application is to USDA data for beef, pork and poultry.

PDF file of paper: chua.pdf

Session: Bayesian Econometrics

Time: Friday, 6 July, 3:30pm - 5pm

Room: C