Trade with Correlation

We develop a trade model in which productivity presents an arbitrary pattern of correlation. The model approximates the full class of factor demand systems consistent with Ricardian theory. In particular, our framework formalizes Ricardo’s insight that countries gain more from trade with partners that have relatively dissimilar technology. Incorporating this insight entails a simple correction to the sufficient-statistic approach used for macro counterfactuals, and enables a general aggregation result that links macro demand systems to micro estimates. In our quantitative application, we estimate a multi-sector trade model which captures the possibility that nearby countries may share technology, and, hence, have correlated productivity draws. Our estimates suggest that accounting for correlation is key to calculating the gains from trade.