Quantile gravity : economic integration agreements, least traded goods, and less developed economies

Abstract

Gravity-equation estimates of the elasticity of trade with respect to bilateral trade costs – or of coefficient estimates of binary variables for the presence or absence of economic integration agreements (EIAs) – are central to determining quantitatively economic welfare impacts of trade-policy liberalizations. Despite decades of study, trade economists have largely focused on conditional mean estimates of a (constant) trade elasticity or of an EIA dummy variable's (common) effect in a “gravity-equation” specification. In this paper, we provide a novel panel-data quantile regression approach to estimating EIAs' partial effects across (conditional) quantiles that avoids Jensen's Inequality, avoids the incidental parameters problem associated with three-way fixed effects, and allows zeros. To motivate the potential economic usefulness of our approach, we examine two distinct “cases.” First, using quantile regressions across a broad swath of country-pairs and EIAs at the disaggregated trade-flow level, we provide systematic evidence that supports the Arkolakis (2010) proposition; trade-flow growth effects of any type of EIA are larger for goods with lower initial sales. Second, we show that the partial effects of EIAs on trade flows are considerably larger for developing countries' exporters across quantiles.