Quantitative Evaluation of a Public Policy among Rural Agricultural Households in Mexico: The Case of the Producción para el Bienestar Program

Objective: To estimate the average treatment effect on the treated (ATT), under the assumptions of an observational design based on propensity scores, associated with participation in the Production for Well-Being Program and its relationship with the economic performance of rural households engaged in agricultural activities in Mexico. Design/methodology/approach: A cross-sectional observational study was conducted using microdata from Mexico’s 2024 National Household Income and Expenditure Survey. The analytical sample comprised 10,055 rural households engaged in agricultural activities. Propensity score matching was implemented using three nearest neighbors with replacement, common support, and a caliper restriction. Statistical inference was performed through subsampling, with full re-estimation of both the propensity score and the matching procedure. Multiplicity across the five evaluated outcomes was controlled using Holm’s adjustment procedure. Results: The matching procedure achieved adequate covariate balance. For the sample ATT, all five estimated effects were positive after applying a logarithmic transformation to mitigate the influence of extreme values. Following Holm adjustment, quarterly agricultural expenditure and agricultural self-consumption remained statistically significant. Quarterly agricultural income, the primary outcome, exhibited a positive estimated effect but did not reach statistical significance. The weighted sensitivity analysis revealed a broader pattern of positive associations, whereas the exclusive-treatment specification did not reproduce the same degree of statistical significance. Study limitations/implications: The cross-sectional design, self-reported treatment status, concurrent participation in other public support programs, and the possibility of unobserved confounding constrain the causal interpretation of the estimated effects. Findings/conclusions: The evidence indicates favorable but heterogeneous associations between program participation and different components of the economic performance of agricultural households. However, the conclusions are partly contingent on the estimand considered and the treatment of the survey design.