The distributional effects of trade: Theory and evidence from the United States

How much do consumption patterns matter for the impact of international trade on inequality? In neoclassical trade models, the effects of trade shocks on consumers' purchasing power are governed by the shares of imports in consumer expenditures, under no parametric assumptions on preferences and technology. This paper provides in-depth measurement of import shares across the income distribution in the United States, using new datasets linking expenditure and customs microdata. Contrary to common wisdom, we find that import shares are at throughout the income distribution: the purchasing-power gains from lower trade costs are distributionally neutral. Accounting for changes in wages in addition to prices in a unified nonparametric framework, we find substantial distributional effects that arise within, but not across, income and education groups. There is little impact of a fall in trade costs on inequality, even though trade shocks generate winners and losers at all income levels, via wage changes.