Bilateral trade volumes, the gravity equation and factor proportions

The gravity equation has been widely used in studying the determinants of bilateral trade flows. Despite their dubious theoretical foundations gravity models have been extremely successful empirically. All theoretical attempts to provide a formal justification for the gravity equation assume complete specialization in production. This leads to a misleading impression that complete specialization is a necessary condition for deriving the gravity equation. In this paper we demonstrate formally that the gravity equation can be derived also from a variety of incomplete specialization models based on both neoclassical and monopolistic competition assumptions. The common prediction that emerges from these models is that factor proportion variables, along with the country size variables, play a key role in determination of bilateral trade volumes, however, their impact is model specific. The neglect of these variables in empirical studies employing gravity equations derived from complete specialization models might result in estimates that suffer from the omitted variable bias if trading partners differ in terms of their relative factor endowments.