When Does Agroforestry Income Reduce Deforestation? Evidence from a Natural Experiment in Madagascar

2026-03-13

AI Summary

This study investigates whether agroforestry income can mitigate tropical deforestation, particularly in contexts where rural poverty and forest loss are intertwined. Leveraging the 2015 vanilla price surge as an exogenous income shock and employing a matching-augmented synthetic control method in Madagascar’s primary vanilla-producing region, the analysis identifies the causal effect of income increases on deforestation. The findings reveal that higher income significantly reduced deforestation in areas with low agricultural opportunity costs—lowering the annual deforestation rate by 1.7 percentage points in 2017, equivalent to avoiding approximately 701 hectares of forest loss—yet exacerbated deforestation in regions with high agricultural potential. This heterogeneity, including a reversal in the direction of the income effect, underscores the necessity of context-specific policy design and provides causal evidence supporting income-based interventions in forest conservation.

Technology Category

• Multiagent Systems: Mechanism Design
• Machine Learning: Causal Learning
• Game Theory and Economic Paradigms: Mechanism Design

Application Category

• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Security and Privacy: Privacy-enhancing technologies
• Search and Retrieval-Augmented AI: Agentic search

Abstract

Tropical deforestation and rural poverty are deeply intertwined, yet isolating the causal effect of income on forest loss remains challenging. We use the 2015 global vanilla price boom, triggered by food-industry shifts toward natural flavoring, as an exogenous income shock affecting Madagascar's primary vanilla-producing region. Using a matching-augmented synthetic control design, we estimate that income gains reduced annual deforestation by 1.7 percentage points in 2017, equivalent to approximately 701 hectares of avoided forest loss. Under a monotonicity assumption linking the price boom to farmers' income, the sign of this reduced-form effect is informative about the causal direction of income on deforestation. However, effects were strongly heterogeneous: higher incomes reduced deforestation in drier, more accessible municipalities but increased clearing in wetter, low-elevation areas with high agricultural potential. These divergent patterns suggest that income simultaneously relaxes subsistence pressures driving forest dependence and raises the opportunity cost of conservation where agricultural returns are high. Our findings indicate that commodity-based agroforestry can align poverty alleviation with forest conservation under conditions of low agricultural opportunity cost. Still, policies must anticipate contexts where rising incomes amplify deforestation in agriculturally suitable land. The strategic targeting of livelihood interventions based on local agricultural potential may help reconcile development and conservation objectives in tropical forest frontiers.

Problem

• agroforestry
• deforestation
• income
• rural poverty
• causal effect

Innovation

• natural experiment
• synthetic control
• income-deforestation causality
• heterogeneous effects
• agroforestry