Classical versus modern approaches to interregional input-output analysis

Abstract

The family of classical interregional input-output models associated with authors such as Isard, Chenery, Moses, Leontief, Strout, Riefler and Tiebout may be classified and compared in terms of the assumed structure of their corresponding matrix of interregional trade share coefficients. In this way, a general hierarchy of interregional trade theories may be specified in terms of the quantity and nature of the trade data so assumed or collected. Each theoretical model reduces to a statistical estimation problem based on varying degrees of available trade information, and may therefore be handled conveniently using methods of statistical inference such as the principle of minimum information gain. By simply specifying any theoretical trade hypothesis in the form of a priori restrictions or linear constraints on the trade share estimates, this methodological principle can fruitfully be used to quantify and compare all the classical interregional models as well as other semisurvey approaches.