Activity Number:
|
106
|
Type:
|
Contributed
|
Date/Time:
|
Monday, August 12, 2002 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Bayesian Stat. Sciences*
|
Abstract - #301191 |
Title:
|
A Gibbs Sampler for Estimating Sets of SURs with Cross-Set Restrictions on the Coefficients
|
Author(s):
|
Ma. Valenzuela*+ and William Griffiths
|
Affiliation(s):
|
Monash University and University of Melbourne
|
Address:
|
Chisholm Towers (Bldg S), 26 Sir John Monash Drive, Caulfield East, Victoria, International, 3166, Australia
|
Keywords:
|
Bayesian Inference ; Expenditure Functions ; Input Demand Functions
|
Abstract:
|
This paper introduces a Bayesian procedure for estimating sets of seemingly unrelated regression equations where the error covariance matrix is allowed to be different for each set and where there are linear restrictions on the coefficients across the sets. First, derivation of the Bayesian methodology for a particular class of linear statistical models is outlined in detail. This is followed by a step-by-step description of the numerical algorithm that was developed in association with the methodology. The Bayesian procedure is then applied to two cases. In the first, sets of household expenditure functions with a common coefficient vector are estimated. The second is concerned with imposing homogeneity and symmetry on input demand equations within a production setting.
|