Learning to use trade agreements

Meeting Rules of Origin (ROOs) in order to obtain lower tariffs in a Preferential Trading Area (PTA) is costly both in terms of production costs and fixed documentation costs. Using a model-based approach that corrects for endogeneity and a unique exporter–importer matched transaction-level customs dataset from Latin American countries, we show that preference usage patterns suggest that these fixed costs fall with exporters’ experience in preference utilization, particularly that in the same product and with the same partner, indicating both the existence and channel of learning. Exploiting a natural experiment, we also show that newly covered products have much more learning as might be expected.