Testing competing world trade models against the facts of world trade

We carry out an indirect inference test of two versions of a computable general equilibrium (CGE) model of world trade. One of these, the 'classical' model, is well-known as the Heckscher-Ohlin-Samuelson model of world trade, in which countries trade homogeneous products in world markets and produce according to their comparative advantage as determined by their resource endowments. The other, the 'gravity' model, assumes products are differentiated by geographical origin, so that trade is determined largely by demand and relative prices differing according to distance; trade in turn affects productivity through technology transfer. These two CGE models of world trade behave in very different ways and predict quite different effects for trade policy, underlining the importance of discovering which best fits the facts of international trade. Our findings here are that the classical model fits these facts fairly well in general, while the gravity model is largely strongly rejected by them.