We consider Bayesian estimation of the coefficients in a linear regression model, using a conjugate prior, when certain additional exact restrictions are placed on these coefficients. The bias and matrix mean squared errors of the Bayes and restricted Bayes estimators are compared when these...