1. Introduction

Farmer-managed natural regeneration (FMNR), the protection and pruning of naturally regenerating trees on farmland, has re-greened millions of hectares of the Sahel [1], [2]. The environmental gains are well documented, but rigorous estimates of the effect on household welfare are scarce, and the sign is contested: trees compete with crops for water yet also provide fodder, fuel, and a buffer against shocks [3]. This paper estimates the household-level effect of adoption on food security and income diversification.

2. Data

We use a three-wave panel of 900 households across four departments of Niger, surveyed in 2019, 2021, and 2023. Each round records FMNR adoption, on-farm tree density, food-shortfall months in the preceding year, and income by source. Sites span a rainfall gradient from 300 to 600 mm per year, letting us test whether effects are largest where water is scarcest.

3. Empirical strategy

Because adopters may differ from non-adopters in unobserved ways, we exploit the panel and estimate a household fixed-effects model

$$y_{it} = \beta\,A_{it} + \gamma^{\top} X_{it} + \mu_i + \lambda_t + \varepsilon_{it},$$

where $y_{it}$ is the outcome for household $i$ in wave $t$, $A_{it}$ indicates FMNR adoption, $X_{it}$ are time-varying controls (household size, rainfall, livestock), $\mu_i$ absorbs all fixed household characteristics, and $\lambda_t$ are wave effects. Identification comes from within-household changes in adoption over time, and $\beta$ is the parameter of interest.

4. Results

Adoption reduced food-shortfall months and raised the share of income from tree products, with the largest effects at the driest sites (Table 1). The point estimate of a 1.6-month reduction corresponds to 22 percent of the non-adopter mean.

Effects were robust to dropping any one department and to controlling for a village-level rainfall trend.

5. Discussion and conclusion

Within households, adopting FMNR is associated with materially fewer months of food shortfall and a more diversified income, and the benefit is largest exactly where drought risk is highest. Secure tree tenure predicted sustained adoption, echoing evidence that rights over regenerated trees are decisive for scale [2]. The main limitation is that adoption is not randomly assigned; while fixed effects remove time-invariant confounders, time-varying shocks correlated with adoption could remain. The findings support policies that clarify tree tenure alongside extension for FMNR.