Econometric profiles of testing of statistical hypotheses: model specification tests

Abstract

The paper, within the frame of econometric specification testing – after an excursus on criteria based on likelihood theory –, focuses on a pair of emblematic and virtually specular facets of the model specification and validation process. Indeed, within a mean-dispersion linear specification framework, the paper faces the issues of testing on the one hand, the overall significance of the regression and on the other, the sphericalness of disturbances and, making use of an elegant algebraic toolkit, derives the ad hoc Lagrange multiplier test statistics.