Modeling Agglomeration and Dispersion in City and Country: Gunnar Myrdal, Francois Perroux, and the New Economic Geography - Critical Essay

American Journal of Economics and Sociology, The, Jan, 2001 by Stephen J. Meardon

1. are of a general equilibrium nature, in that they consist of a system of equations wherein prices and quantities are determined simultaneously, firms enter freely and make zero profits in equilibrium, and economy-wide resource constraints are specified;

2. feature increasing returns to scale production technologies, providing a rationale for the concentration of production in locations that may differ from the initial locations of productive factors;

3. employ a monopolistically competitive market structure, usually (but not necessarily) the Dixit and Stiglitz (1977) formulation, to overcome the difficulties in obtaining criteria (1) and (2) simultaneously;

4. have a spatial geometry that is not only expressed in notation ([x.sub.i,j] denotes good i produced in region j), but also is implicit in the parameterization of the model. For example, locations 1, 2, and 3 could be assumed to form an isosceles triangle in geographic space if transport costs between all three pairs, in any direction, are equal;

5. allow at least one productive factor to be mobile between locations, so the rationale of spatial concentration can be realized and the initial locations of productive factors can be altered.

The works that meet these criteria are sufficiently numerous to deserve some kind of collective name; yet at the same time they are sufficiently few, and written by sufficiently few authors, that Krugman (1998a) appears more comfortable referring to the new economic geography as a "genre" rather than a "subdiscipline." By far the most widely discussed article that meets the criteria is Krugman (1991a), which is cited 220 times by articles indexed by the SSCI. [5] Other influential contributions--whose citation counts, however, differ from the latter's by an order of magnitude--have been Krugman and Venables (1995: 49 citations), Krugman (1993: 38 citations), Venables (1996: 33 citations), Fujita and Krugman (1995: 28 citations), and Puga and Venables (1996, 1997: 7 and 10 citations). Fujita, Krugman, and Venables (1999) recapitulated the methods and insights of the literature produced thus far in a book, as Helpman and Krugman (1985) did for the new trade theory.

Articles about the economics of geographical space that ask different questions, use different methods, or generally have a different flavor than those cited above tend to meet some but not all of the definition's five criteria. A couple of examples should help illustrate the point. August Losch's (1950) model has an explicit spatial geometry as in criterion (4), and is, in a sense, a general equilibrium system--albeit not in the sense of (1). Within any firm's market area, other firms are assumed not to enter. The system determines quantities of goods, their prices, and the spatial dimensions of hexagonal market areas that are assumed to be spread evenly throughout the geographical plane. Losch's model thus addresses the geographical breadth of markets but not their uneven concentration.

A less well known example of spatial modeling that is in some respects closer to the new economic geography, but still distinctly different from it, is found in Bruce Benson and James Hartigan's (1984, 1987) models of spatial competition. Their models were intended to wed Losch to the emerging "new trade theory" models of Brander (1981) and Brander and Krugman (1983), whose own purpose was to explain the ubiquity of intra-industry trade (also known as "reciprocal dumping" or "cross-hauling"). A compelling theoretical explanation of intra-industry trade was a paramount objective of new trade theorists, who observed the phenomenon empirically but could not make sense of it within the confines of the Heckscher-Ohlin model. Brander and Krugman explained intra-industry trade using a model in which Gournot-competing firms in two nations served the consumers of both nations; Benson and Hartigan extended this model to make space more explicit: consumers were assumed to be distributed along a line whereupon domestic and foreign producers were located at the ends and the national border was located somewhere in the middle. Transportation costs had to be incurred to ship goods to consumers in proportion to their distance from producers. Intra-industry trade was therefore likely to be observed near the border, where substantial transport costs would be incurred in shipments from either domestic or foreign producers.