Optimal Tariffs in the Melitz Model: A sufficient statistics approach for trade policy
(Previous title) Competition, Productivity and Trade, Reconsidered*

This paper shows that the variable nature of the trade elasticity provides new policy implications for optimal tariffs. To achieve the goal, we develop a heterogeneous firm model with (i) a general productivity distribution so that the trade elasticity is bilateral-specific to country-pairs; (ii) no outside good so that the wage rate is endogenous; and (iii) import tariffs so that tariff revenues are one of the welfare components. In this general setting, we find that optimal levels of import tariffs are the same across different trade models with a constant trade elasticity, conditional on the two sufficient statistics for welfare --- the domestic trade share and the trade elasticity. However, the equivalence of optimal tariffs across different trade models no longer holds when the trade elasticity differs across markets. Calibrating the model to US data, optimal tariffs with a variable trade elasticity are substantially lower than those with a constant trade elasticity. Moreover, using analytical solutions of comparative statics, the effect of market size on optimal tariffs is quantitatively much smaller than that of variable trade costs.

* We revised this discussion paper with the new title in December 2025. This paper was previously circulated under the title "Competition, Productivity and Trade, Reconsidered."

Forthcoming: Ara, Tomohiro. "Optimal tariffs in the Melitz model: A sufficient statistics approach for trade policy," The Japanese Economic Review.