Gravity's Rainbow: Modeling the World Trade Network

Abstract:     
The gravity model, long the empirical workhorse for modeling international trade, has recently been put on firmer theoretical foundations (Anderson and van~Wincoop, 2003; Helpman et alia 2008). These theoretical models emphasize how economic conditions in the rest of the world affect bilateral trade. Nevertheless, standard applications of the gravity model ignore the network dependencies in bilateral trade data, assuming that dyadic trade is independent, conditional on a hierarchy of covariates over country, time, and dyad. More nuanced implementations attempting to employ proxies for inherently unobservable "multilateral resistance terms'' suffer from measurement error and omitted variable problems. We propose a strategy designed to account for and estimate second- and third-order dependencies in the data. We estimate this model using bilateral trade data from 1990-2008, which substantially outperforms standard accounts in terms of both in- and out-of-sample predictive heuristics. We illustrate the model's usefulness by tracking specific trading propensities.

Number of Pages in PDF File: 1

Keywords: gravity model, trade, networks, latent factors

JEL Classification: C10, C11, C23, C52, C82, F10, F13, F17

working papers series