Abstract:

The superiority of Bayes estimation of regression coeffcients and error variance in linear model was studied based on normalinverse Gamma priors. The existed results were complemented without the additive conditions. It was shown that the Bayes estimation of regression coefficients is superior to the least squares estimator (LSE) under the Pitman closeness criterion. And the superiority of the Bayes estimation of error variance over LSE was also investigated in terms of the mean square error criterion. Finally, a Monte Carlo simulation was carried out to verify the theoretical results.

Key words: Bayes estimation, least squares estimation, Bayes Pitman , criterion, mean square error criterion