Dynamic Models, New Gains from Trade?

Yes. We state closed-form expressions for steady state gains from trade that apply in a class of dynamic trade models that includes dynamic versions of the Krugman (1980), Melitz (2003), and customer capital (e.g., Arkolakis, 2010) models. As in Arkolakis, Costinot, and Rodríguez-Clare (2012), the gains are a function of the domestic trade share and the long-run elasticity of trade with respect to iceberg trade costs. This elasticity, however, cannot be recovered from the long-run tariff elasticity alone, because the dynamic trade adjustment mechanism present in this class of models increases the difference between the iceberg and the tariff elasticity in the long run. Our main substantive finding is that the dynamic gains from trade are unambiguously greater than those implied by static models, conditional on the same long-run tariff elasticity. We propose an approach to implement the formula by combining tariff elasticity estimates at multiple horizons and demonstrate that the difference relative to static models can be substantial. Accounting for the transition path has a modest impact on the magnitude of the gains from trade, relative to simply comparing steady states using the formula.