International economic integration yields large potential welfare effects, even in
a static constant returns competitive world economy. Our method is novel. The effect
of border barriers on trade flows is often inferred from gravity models. But their
rather atheoretic structure precludes welfare analysis. Computable general equilibrium
models are designed for tight welfare analysis, but lack econometric foundation.
Our method combines these approaches. Gravity models based on Anderson's (1979) interpretation
are full general equilibrium models of a special simple sort. In Anderson and van
Wincoop (NBER WP 8079, 2001) we develop and estimate this structure, then calculate
the comparative static effects on trade flows of border barriers. In this paper we
further deploy the model to explore the comparative statics of welfare with respect
to borders, to currency unions and to NAFTA. Our NAFTA exercise does a much better
job of replicating the actual trade flow changes than do computable general equilibrium
models. An interesting implication is that terms of trade changes are very important,
even for small' countries such as Mexico.