Causal inference in conservation policy: Evaluating the effects of Protected Areas on deforestation and well-being in the Peruvian Amazon

Abstract

The Peruvian Amazon, one of the most biodiverse regions on Earth, is under increasing pressure from deforestation, threatening both ecological wealth and local livelihoods. Protected Areas (PAs) have emerged as one of the main conservation strategies; however, questions remain regarding their effectiveness in reducing deforestation and their socioeconomic impact. This study evaluates the effects of PAs, their Buffer Zones (BZs), and Regional Conservation Areas (RCAs) on deforestation and household welfare between 2001 and 2023. Using causal inference methods, geographic Regression Discontinuity Design (RDD) and staggered Difference-in-Differences (DiD), the analysis integrates high-resolution spatial data on forest loss with georeferenced household information. Results indicate that PAs significantly reduce long-term deforestation by 1.3 to 2.8 ha/km², while BZs mitigate forest loss by 1.1 to 1.5 ha/km². However, no significant deforestation effects are observed in the short term. Regarding welfare, households within BZs exhibit lower per capita income and expenditure, particularly near indirect-use PAs. The study also provides exploratory evidence on RCAs, a category rarely examined in the literature. Although their estimated effects are smaller and not statistically significant, their inclusion highlights the growing role of regional governance in conservation policy. These findings indicate that conservation outcomes strengthen over time and depend on local governance, institutional capacity, and the ability to contain pressures from extractive activities. This research contributes new empirical evidence to inform conservation policy design, emphasizing the need to integrate ecological and social dimensions into Protected Area management.