Trade and welfare : does industrial organization matter?

Many contemporary theoretic studies of trade over geography reduce to an ex- amination of constant-elasticity reactions to changes in iceberg trade costs. These impacts are readily analyzed in simple constant-returns models based on the Arm- ington (1969) assumption of regionally differentiated goods. Following the line of reasoning suggested by Arkolakis et al. (2008) one can reach the surprising conclu- sion that industrial organization does not matter. In the present paper, we show that this finding is fragile, and with a minor elaboration of their model, the rich industrial-organization features of the popular Melitz (2003) model do, in fact, gen- erate important differences for trade and welfare.