A common assumption in the quantitative Ricardian international trade literature is that within a country, import shares are equalized across sectors. This assumption is at odds with the data, which show within-country heterogeneity in sectoral import behavior. I build a multi-country, multi-sector general equilibrium Ricardian trade model, in which I include a new extensive and intensive international trade margin at the importing sector level. Counterfactual analysis shows that accounting for within-country sector-specific import behavior is significant for the level of welfare gains from trade. Calibrations based on two cross-country data sources show that a benchmark Ricardian model with equalized import shares across sectors underestimates welfare gains from trade by 13 to 24% on average compared to the model accounting for within-country sectoral import patterns. The benchmark model underestimates the productivity gains of sectors which account for most country-level imports and the spillovers of their productivity gains on other sectors through sectoral linkages.
