Gravity, globalization and time-varying heterogeneity

Abstract

As data availability improves, the gravity model of international trade is used to study the flow of trade over increasingly long periods of time. This puts pressure on the standard assumption that constant bilateral fixed effects are sufficient to control for unobserved heterogeneity. When looking across decades rather than years, changes in the economic landscape will affect both explanatory and dependent variables, thus biasing the results. The paper begins by providing evidence of time variation in the unobserved heterogeneity using a novel econometric test. It then presents a new way of estimating the gravity model that can identify and control for unobserved heterogeneity as it varies over country pairs and time. This approach substantially enhances the precision of the estimated parameters, enabling a detailed analysis of the impact of different types of economic integration agreements on trade on an annual basis. This heightened precision also demonstrates the diminishing importance of distance and allows our model to accurately depict the evolution of trade over time. The pattern in the unobserved heterogeneity closely resembles the ascent and the deceleration of globalization. The paper closes with some preliminary analysis into the factors that underlie the unobserved heterogeneity.