Addressing Food Insecurity in Ultra-Poor Families: Findings from a Randomized-Controlled Trial in Burkina Faso

Background: An emerging literature suggests graduation programs— “cash plus” interventions that combine economic incentives with psychoeducational supports to promote asset accrual—can effectively lessen food insecurity among ultra-poor families. Nevertheless, psychoeducational supports largely focus on child development and sanitation, failing to address gender norms, family dynamics, and child protection issues that also shape food insecurity. Combining poverty-reduction efforts with family coaching that addresses these issues may yield greater improvements in food security than poverty-reduction alone. Moreover, few studies test whether graduation program effects are moderated by family structure or child gender. To address these gaps, the present study examines the effects of a graduation program on food insecurity experienced by children and mothers from ultra-poor households in Burkina Faso, and assesses whether the addition of a family coaching component elicits greater improvements in food security.

Method: Data are derived from a randomized-controlled trial evaluating the impact of an integrated graduation program on child protection and labor outcomes among ultra-poor families in Burkina Faso. Families were randomly assigned to one of three treatment arms. In Arm 1, participants were offered economic incentives and ongoing assistance to help mothers achieve business goals. In Arm 2, participants were given the same intervention as arm 1 and a family coaching intervention addressing gender norms, family dynamics, and child protection. Participants in Arm 3 served as waitlist controls. Food insecurity was measured in children and mothers at baseline and 12- and 24-months from baseline using the Household Hunger Scale. To assess differences in intrahousehold food distribution, we created a difference score by subtracting child scores from mother scores. Child gender and family structure (monogamous/polygynous) were tested as moderators. Intervention effects were tested using multilevel mixed-effects regression.

Results: Children in Arms 1 (24-month: B= -0.51, p< .05) and 2 (12-month: B= -0.61, p< .05; 24-month: B= -0.70, p< .001) reported less food insecurity than controls, with children in Arm 2 reporting greater improvements at 24 months (B= -0.19, p< .01). Mothers in Arms 1 (12-month: B= -0.99, p< .001; 24-month: B= -0.37, p< .05) and 2 (12-month: B= -1.09, p< .001; 24-month: B= -0.50, p< .001) reported less food insecurity than controls, with mothers in Arm 2 reporter greater improvements at 24 months (B= -0.13, p< .05). Difference scores between mothers’ and children’s food insecurity showed greater reductions in Arm 2 compared to Arm 1 at the 12-month follow-up (B= 0.18, p< .05). Marginal estimates indicated effects differed by family structure and child gender, with polygynous families (12-month: B= -0.82, p< .01) and girls (12-month: B= -0.58, p< .01; 24-month: B= -0.59, p< .01) only showing improvements in Arm 2.

Conclusion and Implications: While additional research is needed to understand the mechanisms driving intervention effects, results suggest combining poverty-reduction with family coaching that addresses gender norms, family dynamics, and child protection elicits greater improvements in food insecurity that poverty-reduction alone. Our findings allude to the potential of addressing broader social systems to enhance not only food security among ultra-poor families, but also their social protection.