Some remarks on decomposition of partitioned multivariate models into two seemingly unrelated submodels

Abstract

Procedure on testing decomposition of partitioned multivariate models into two seemingly unrelated submodels in order to obtain more efficient estimators is presented. The multiresponses are independently normally distributed with the same covariance matrix. Moreover, the partitioned multivariate model is considered either with, or without an intercept. The elimination transformation of the intercept that leads to the same BLUEs of parameter matrices in mul- tivariate models with and without intercept is stated. Due to the elimination transformation the same procedure for testing decomposition can be applied for models with and without the intercept.