Regional exposure to trade shocks : reconciling theory and evidence

Abstract

A growing literature shows that shocks to international trade can have heterogenous effects across regions within a country. This literature takes two broad approaches. The first approach is reduced-form and consists of regressing changes in regional outcomes on measures of regional exposure to trade. The variation in these measures of regional exposure is primarily driven by variation in sectoral employment shares across regions. The second approach estimates structural general equilibrium models and quantifies the changes in regional outcomes in response to a trade shock through counterfactual exercises. We show that the reduced-form measures of regional exposure cannot be derived from a general equilibrium model even when only considering the partial equilibrium effect. Using Brazilian data on sub-country trade flows, we show that these analytical differences between the reduced form and theoretical measures of exposure translate into quantitative differences in the measures. We also show that the rank correlation between the different measures of regional exposure is sensitive to the source country of the import cost shock. The results presented caution against relying too heavily on reduced-form exposure measures to recover partial elasticities of regional wages to an international trade shock.