It is well-known that international trade flows can be well described by a “gravity equation” in which bilateral trade flows are a log-linear function of the incomes of and distance between trading partners. Indeed, the gravity equation is one of the greater success stories in empirical economics. However, the theoretical foundations for this finding are less clearly understood. The gravity equation is not implied by a plausible many-country Heckscher-Ohlin model (which has nothing to say about bilateral trade flows). An equation of this type does arise, however, from a model in which countries are fully specialized in differentiated goods.

While specialization might characterize manufacturing goods, it is presumably not a feature of homogeneous primary goods. Despite this theoretical presumption, the gravity equation seems to work empirically for both OECD countries and developing countries (Hummels and Levinsohn, 1995). Since developing countries tend to sell more homogeneous goods, it seems puzzling that the gravity equation works well for these countries. Thus, it is hard to reconcile the special nature of the theory behind this equation with its empirical performance.

In this paper, we argue that conventional wisdom should be reversed: the theory behind the gravity equation is general, but its empirical performance depends on the particular sample. On the theoretical side, we show how a gravity equation can arise even with homogeneous goods produced by all countries. We model the market structure in the homogeneous good as Coumot-Nash competition, and use the “reciprocal dumping” model of trade described by Brander (1981), Brander and Krugman (1983) and Venables (1985). A two-country version of this model is developed in section 2, and used to solve for trade flows using a simple graphical technique.
Getting money for anything is not easy, especially when you have to ask other people about it. You do not have to ask us: applying for a speedy cash payday loan with is all about sending your request and waiting for it to be approved. We give you the money without asking any questions.
We use our model to derive the gravity equation in section 3. We find that the implications of the homogeneous goods model are similar to those obtained from a differentiated-products, monopolistic competition model. One important common feature is the “home market effect” (Krugman, 1980): larger countries tend to be exporters of a product, ceteris paribus, since the larger market attracts firms to locate there. Krugman established this result for a model with a monopolistically competitive sector producing differentiated-products, subject to transportation costs in trade.

We show that the home market effect – an elastic supply of exports with respect to domestic income – also characterizes a homogeneous-product sector with free entry. Since both differentiated and homogeneous goods models have a home market effect, it might seem that they do not have distinct empirical implications. But if homogeneous goods have greater barriers to entry (due to resource-dependency, for example), then the home market effect is reversed. These theoretical results will be important for interpreting our empirical findings which we turn to in section 4.