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

TL;DR

This study uses a natural experiment in Madagascar to show that income from vanilla farming can both reduce and increase deforestation depending on local conditions, highlighting the importance of targeted policies.

Contribution

It provides causal evidence on how income shocks influence deforestation, revealing heterogeneous effects based on environmental and economic contexts.

Findings

Income gains reduced deforestation by 1.7 percentage points in 2017.

Higher incomes decreased deforestation in dry, accessible areas.

Increased incomes led to more deforestation in wet, high-potential areas.

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.