Traditional theory of international trade has been developed on the assumption of homogeneous country in which identical consumers and production factors are all uniformly distributed. In the real world, however, most countries are composed of heterogeneous regions, and in accordance the above assumption is obviously unrealistic. Then a natural question is whether the main conclusions of the theory are critically affected if we relax the assumption. Speaking in the most general context, it is nearly impossible, as Nataf's theorem asserts, that the nonlinear relations about trade patterns are relevant for both regions and countries when a country consists of heterogeneous regions. The purpose of the present study is to examine whether and under what conditions we can consistently assume that both interregional and international trades follow the patterns predicted by the comparative advantage theory. In particular, relevancy of the Rybczynski, the Stolper-Samuelson, and the Heckscher-Ohlin theorems are investigated for countries with three different sources of regional heterogeneity; i.e., (1) consumer amenity difference, (2) transport advantage difference, and (3) agglomeration economies. The results are different from case to case. For a country with regional amenity difference, we can show that the Rybczynski and the Stolper-Samuelson theorems are relevant if the amenity density function is non-decreasing. However, for the Heckscher-Ohlin theorem, condition is much more restrictive; i.e., the density functions for all countries must be identical except for the scale. For a country with the regional difference in transport advantage to the international ports, the Rybczynski and the Stolper-Samuelson theorems do not hold but the Heckscher-Ohlin theorem is conditionally relevant. Finally, in a country with city agglomerations, the Rybczynski theorem is relevant if the labor requirement in the agglomerated firm is sufficiently small. On the other hand, the Heckscher-Ohlin theorem does not hold as far as the number of cities is given. However, if the number of cities changes as the country size differs, relevancy of the theorem increases. Especially, in the case that new cities appear at distant places from each other, it provides approximately relevant predictions in the sense that countries with the same labor-land endowment ratios exhibit similar patterns of international trade.
