Agro-industry Location: Theory and Test

This paper poses and tests a new model of the location of plants that process ubiquitously supplied ‘von Thünen’ inputs such as raw farm output. The new model rationalizes why agro-industry is city-located but resource-oriented. It demonstrates the significance of relative transport costs in agroindustry location. Furthermore, we present a unprecedented direct test/empirical confirmation of establishment location theory. We develop this agro-industrial plant location model in two stages. In the macro-spatial stage, we show why an initial agro-industrial plant chooses a site that becomes a city, and how that initial site can establish a market center and von Thünen rent gradient. In the micro-spatial stage we show that even though subsequent plants can choose the extent of their input market areas as well as their site, costs are convex (not concave) in distance. Thus, in contrast with canonical location theory, intermediate sites can be optimal for agro-industrial plants, even in the absence of uncertainty. This new testable model rationalizes the observed variety in spatial configurations of U.S. agroindustry. It justifies the use of normative math programming models of food and kindred processing plant location widely used by agricultural economists. The model is tested using data on transport cost rates, crop yields, acreages, and existing agro-industrial plant locations. The null hypothesis is strongly rejected for all agro-industries modeled. In the U.S. these types of plants systematically display location patterns that are concentrated, co-located, or dispersed, depending on relative transport costs in a predictable way.