On the Specification and Simulation of a Regional Econometric Model: A Model of Mississippi

Abstract

HE analogy between the economy of a region and small country is a good starting point for regional modeling since the theories of the open economy and international trade transfer in important respects to regional analysis. On the other hand, recent econometric models of regions, although stressing macroeconometric relationships, have been structured along a unique pattern, quite different from national macro models. Numerous models have been constructed for states and even smaller areas' in the search to find an effective forecasting tool linking the regional economic outlook to the national economic forecast. Regional models are constructed as satellites to national models. The structure of the regional models has been greatly influenced by the purposes for which they are to be used and by the data. At this time it is possible to consider a structure for a regional model and to provide for it a somewhat clearer theoretical basis than has been customary. This article describes a new econometric model for the state of Mississippi, its theoretical background, estimation, and simulation applications. As with national models, one is tempted to adopt an aggregate income expenditure approach in a national accounts framework for regional model building.2 The regional econometric model would resemble the model of a small country.3 And, presumably, the national model could then be seen as the aggregation of many interrelated regional models.4 This resemblance takes little account of the purpose for which regional inodels are being built or of the lack on the regional level of the statistical materials which go into typical Keynesian model structures. The regional model builder needs to predict the evolution of the regional economy in the national economic setting. Like the small country economist, he takes the external (national) environment as given. If the region is small and industrially diversified, the feedbacks between regional developments and the national outlook are likely also to be small. In this sense the regional model builder can structure his model as a satellite of the national economy, the causation running from national developments to the region but not from the region to the nation.