Protected Areas under Weak Institutions: Evidence from Colombia.

• We provide new evidence on the effectiveness of PAs and the role of institutions. • We use high-resolution satellite imagery and regression discontinuity methods. • PAs significantly reduce deforestation in both national PAs and collective lands. • National PAs are only effective near human settlements, in more developed regions. • In contrast, Collective lands avoid deforestation in remote, less developed regions. Protected areas (PAs) are a cornerstone of conservation policy. While evidence shows that, overall, PAs have contributed to preserving forests, their impact varies greatly depending on the institutional context. This paper provides new evidence on the mechanisms through which local institutions shape the effectiveness of PAs. We use high-resolution satellite imagery of deforestation and illicit activities in Colombia and spatial regression discontinuity methods to estimate the causal effect of PAs in different institutional contexts. Our main results indicate that PAs significantly reduce deforestation, with larger effects for collective lands than national (strict-use) PAs, and no impact for regional (multiple-use) PAs. However, national PAs are only effective near human settlements, in municipalities that provide more public goods and are less violent. In remote areas, national PAs are particularly vulnerable to the expansion of coca crops and gold mining. In contrast, collective lands reduce coca crops and avoid deforestation in remote, less developed regions. These results highlight the extent to which natural PAs rely on the institutional capacity of the national and local governments, while collective lands protect forests even when state presence is weak. [ABSTRACT FROM AUTHOR]