Staggered protection: a study of the dynamic effects of protected areas

Previous estimates of the effect of the creation of protected areas (PAs) on natural conservation 9 are biased by staggered protection and confounder environmental policies. We address these 10 biases by employing a cohort-time refined estimator using Amazon Basin data from 2003 to 11 2020. We also uncover policy-relevant dynamic patterns that remained hidden in previous 12 papers’ aggregate effects. Our findings show that PAs’ effects on deforestation, fires and illegal 13 artisanal mining were biased in at least 50% by staggered protection. Failure to control for 14 confounder policies deflated the effect on deforestation in 13%, and inflated the effects on fires 15 and mining in 16% and 25%. We also observe a rise in deforestation two years before 16 protection, an evidence of forward-looking behaviour. Moreover, PAs’ effects increased with 17 ageing, suggesting that enforcement is subject to learning. Effects were heterogeneous, with 18 both moderately and severely restricted PAs mitigating fires and mining, but only the severely 19 restricted mitigating deforestation. The effects of conservation unit PAs managed by national or 20 subnational governments were unequivocal only on mining, whereas indigenous land PAs 21 successfully curbed deforestation, fires and mining. Therefore, with dynamic and heterogeneous 22 effects, PA creation should leverage the strengths of different government levels and PA types, 23 while also anticipating forward-looking reactions.