Résumé
We estimate the additional effects of a REDD+ pilot project offering Payments for Envi-ronmental Services to reduce deforestation by smallholders in the Brazilian Amazon. We collected original data from 181 individual farmers. We use DID-matching and find evi-dence that supports the parallel trend assumption. We estimate that an average of 4 ha of forest have been saved on each participating farm in 2014, at the expense of pastures versus croplands. This amounts to a decrease in the deforestation rate of about 50 percent. We find no evidence of leakage effects. Finally, we find that the project is cost-effective.