Discrete Plant-Location Decisions in an Applied General-Equilibrium Model of Trade Liberalization

Theoretical and applied work in industrial-organization approaches to international trade typically assume either that there are fixed numbers of firms, or that there is free entry and exit with a continuum of firms. This paper makes a first step toward a more realistic approach in which firms face discrete choices about the numbers and locations of their plants. The model is applied to the North American auto industry in the context of the draft North American Free Trade Agreement. Results include: (1) production appears to be excessively geographically diversified initially; (2) autos are produced in fewer locations as trade barriers are lowered; (3) a 'non-monotonicity' case is produced in which a plant is first closed and then reopened as trade barriers are progressively lowered; (4) an example of the misleading nature of marginalist analysis is presented in which plants in Canada and Mexico increase production when locations are fixed but closed down when locations are endogenous and optimized.