Conference Agenda

Cross-border agricultural land acquisitions and frontier settlement are reshaping forested regions in Latin America. Low German Mennonite colonies represent a distinct form of transnational agrarian expansion—large, contiguous land purchases followed by the rapid development of commercial-scale agriculture—that has accelerated settlement in remote forest frontiers. We combine spatially explicit data on 105 Mennonite colonies established across Latin America after the year 2000 with annual satellite measures of forest loss, biomass carbon, biodiversity, and nighttime luminosity. We exploit quasi-experimental variation in establishment timing and locations of Mennonite settlements to plausibly identify the causal impacts of these frontier settlements on the local environment, ecology, and economic development. Frontier settlements lead to very large and significant forest loss with substantial ecological and environmental costs. We find no evidence of any positive development spillovers to local communities. Our findings highlight the ecological consequences of cross-border agricultural settlement in frontier regions.

Protected areas are key conservation instruments in the Amazon rainforest, yet assessing their effectiveness in avoiding deforestation remains challenging. Forest loss is driven by complex, nonlinear, and heterogeneous processes, and protected areas are established in a non-random and staggered manner. We address these challenges using a physics-aware neural network trained on unprotected forest observations to predict counterfactual deforestation in areas that become protected. Comparing observed and predicted deforestation reveals the effective reduction of deforestation following protection. Results are consistent with state-of-the-art econometric methods, while the neural network prediction captures spatial heterogeneity and highlights how conservation effectiveness varies across contexts.

Droughts are widely recognized for their adverse effects on economies, especially on agriculture. With approximately 75% of global food production occurring under rainfed conditions understanding the role of soil moisture is critical. Yet its influence on the economic impacts of drought remains poorly understood. Using global gridded data spanning 25 years, we find that locations with higher pre-shock root-zone soil moisture are associated with smaller drought-year growth losses. We also study a plausible determinant of elevated soil moisture —upstream forest cover— and find that grid cells with greater upstream forest cover correlate with higher soil moisture storage and exhibit losses about half as large as in low-forest settings. Taken together, the evidence highlights the potential value of considering soil moisture and its forest-based drivers alongside conventional “blue” water management, and climate adaptation strategies.

Over the past two decades, the Brazilian Amazon has experienced persistent forest fires, even as monitoring and law enforcement against illegal deforestation have improved. This paper argues that advances in enforcement technology have reshaped illegal behavior, favoring fire use over more detectable deforestation methods. Using cloud cover as an instrument for satellite-based detection, I find that municipalities under more intensive surveillance exhibit higher fire frequency. This suggests that farmers adapt by substituting detectable deforestation technologies with less detectable alternatives. These findings reveal unintended consequences of targeted enforcement and underscore the need for comprehensive monitoring strategies that address multiple deforestation methods.

Credibility concerns about the Verified Carbon Standard, amplified by recent causal evaluations and market fallout, have raised doubts about REDD+ carbon crediting. In response, Verra introduced a new jurisdictional baseline methodology to address past methodological issues. We evaluate this approach by comparing two consecutive six-year baselines with Synthetic Control counterfactuals using satellite forest-loss data (2001-2023) for three protected areas in Sub-Saharan Africa. Jurisdictional baselines vary across sites and periods, ranging from substantial overestimation to close alignment to increasing conservativeness. Our findings suggest that baseline credibility remains sensitive to risk maps and modelling choices, demanding rigorous evaluation of new baselines.

Understanding how crop yield changes affect forest conservation in tropical regions is central to achieving both food security and environmental sustainability. While existing evidence is primarily based on case studies, systematic evidence at broad spatial and temporal scales remains limited. We combine high-resolution (30 × 30 m) satellite data covering the entire tropics annually from 2001 to 2022 to estimate the local effects of crop yield changes on forest cover. Using a dynamic panel framework, we find modest but statistically significant forest loss following yield gains. On average, approximately eight percent of local forest loss can be attributed to local yield increases in the tropics. Effects are heterogeneous and concentrated in more accessible areas with higher initial forest cover, a certain balance between land availability and incentive for expansion, and few agri-environmental policies. Regions with mostly smallholder farming (even defined up to 20 ha) as well as those with broad agroforestry adoption do not lose any forest on net in response to crop yield gains. The strongest negative effects are observed in Sub-Saharan Africa and in Latin America, respectively, whereas we do not find an effect in Asia.