General Expression of Best Equivariant Estimators of Regression Coefficients in Seemingly Unrelated Regression Models

ABSTRACT Seemingly unrelated regression models have been commonly employed as regression models with multiple explanatory variables and multiple response variables having cross‐correlations. This paper discusses equivariant estimation of the regression coefficient vector and derives the best equivariant estimator under a generalized quadratic loss function. The result of this paper is shown to include the results of previous best equivariant estimators under specific loss functions as special cases.