Selecting the covariance structure for the seemingly unrelated regression models

Abstract

ObjectiveThis paper is concerned with evaluating suggested approach of selecting the suitable covariance structure for fitting the seemingly unrelated regression equations (SURE) models efficiently. MethodThe paper assessed AL-Marshadi (2014) methodology in terms of its percentage of times that it identifies the right covariance structure for mixed model analysis of SURE models using simulated data. ApplicationThe simulated equations of SURE models have identical explanatory variables, the regressors in one block of equations are a subset of those in another, and different regressors in the equations with various settings of covariance structures of ∑. Moreover, the percentage of times that REML fail to converge under normal situation are reported. The application of the proposed methodology is given using a panel of data. ConclusionsIn short, AL-Marshadi (2014) methodology provided an excellent tool for selecting the right covariance structure for SURE models using restricted maximum likelihood (REML) estimation method in order to fit the SURE models more efficiently than the existing method that considering the stander unstructured covariance structure in fitting SURE models.