Gains from trade and their quantification: Does sectoral disaggregation matter?

Multi-sector variants of gravity models typically predict much larger gains from trade (losses from protectionism) than their one-sector counterparts. This result –corroborated by several model-based quantification studies and commonly ascribed to Jensen’s inequality– has been recently questioned by studies that use micro price data to obtain sector-level estimates of the trade elasticity, a key parameter for the quantification of the gains. We reassess this issue by using a novel set of estimates of the trade elasticity at various levels of sectoral disaggregation, exploiting a recently proposed identification strategy based on tariffs. In our baseline 24-sector model specification, we find that the cross-country average size of the gains amounts to 12% and that this number is 21% larger than the one delivered by the one-sector specification. Overall, our results suggest that the effects of magnification of the gains associated with greater sectoral disaggregation (which we confirm to be mainly driven by cross-sector variation in trade elasticity) are significant, not negligible, but considerably smaller than generally quantified in the previous literature.