Econometric Analysis of Aggregation in the Context of Linear Prediction Models.

This paper focuses on whether to use macro or micro equations to predict aggregate variables. The Grunfeld-Griliches prediction criterion is generalized to allow for contemporaneous covariances across the micro equations and for parametric restrictions on the disaggregate equations. An econometric test is proposed of the hypothesis of "perfect aggregation." An application is made to employment demand functions for the U.K. economy disaggregated by forty industries. The hypothesis of perfect aggregation is firmly rejected. The prediction criterion marginally favors the aggregate equation when aggregating over the manufacturing industries, but over all industries the disaggregated equations are strongly preferred. Copyright 1989 by The Econometric Society.