The Missing Trade Elasticities and the Floating Size of the Gains from Trade

Abstract

Multi-sector variants of standard gravity models typically predict much larger gains from trade than their one-sector counterparts. This paper explores to what extent this result is due to the relevant cross-sector variation observed in trade elasticity and to what extent it is instead an artifact of (discretionary) sectoral disaggregation. Numerical calculations are performed based on standard sufficient statistics for a specific class of gravity models (i.e. perfect competition models). Results suggest that, for benchmark values of the trade elasticity that applies to the one-sector specification, heterogeneous elasticities across sectors are, by far, the main source of the gains magnification effect, at least under standard levels of sectoral disaggregation at which input-output tables are commonly publicly available. Yet, when using more disaggregated data from tables available for the U.S. only, the effect of sectoral deconstruction per se substantially increases, leaving the issue on the true size of the gains still widely open.