TD NEREUS 02-2024
André Luis Squarize Chagas and Luiza Andrade
Difference-in-differences (DID) has long been a staple in estimating treatment effects in applied econometrics, with recent advancements relaxing traditional assumptions to explore heterogeneous and spillover effects. While heterogeneous effects analysis examines causal impacts across diverse groups and periods, spillover effects analysis delves into the influence of treatments on neighboring units. Incorporating spatial dependence within the DID framework, Spatial Difference-in-Differences (SDID) models have emerged as a powerful tool for analyzing such effects, particularly in settings where observations represent fixed geographical units. This study contributes to the literature by explicitly formalizing underlying assumptions and employing an SDID model to analyze the impact of Brazil’s Priority Municipalities List on deforestation in the Amazon region. Utilizing both traditional DID and SDID methodologies, we uncover significant reductions in deforestation odds ratios within listed municipalities and neighboring unlisted municipalities. Furthermore, we introduce an event study approach linked with SDID to explore the policy’s anticipatory effects. Our findings underscore the effectiveness of the Priority Municipalities List in curbing deforestation and highlight the importance of spatially explicit methodologies in environmental policy evaluation. This article advances methodological discussions surrounding SDID estimation and provides empirical insights into the efficacy of targeted environmental policies in combating deforestation in sensitive ecosystems like the Amazon.