Resilience pays off unevenly for food security in Senegal, study finds

Households in Senegal’s Groundnut Basin that score higher on a composite measure of resilience capacity eat better and more varied diets, but the payoff is far from uniform: it rises steeply at low to moderate resilience levels, flattens beyond an apparent optimum, and differs sharply between households headed by men and women. That is the central finding of a new study published in Agriculture & Food Security, which analyzed survey data from 3,010 households to trace how resilience capacity translates into two widely used food security measures, the Food Consumption Score and the Household Dietary Diversity Score.

The research, conducted by Adama Faye and Awudu Abdulai, draws on the 2019 National Food Security, Nutrition and Resilience Survey carried out by the Executive Secretariat of the National Council for Food Security in Senegal. The full survey covered all 14 regions and 45 departments of the country and interviewed 7,847 households, but the analysis focuses on 19 departments within the Groundnut Basin, a region that spans the western and central parts of the country and covers roughly a quarter of Senegal’s territory. The sample extracted for the study comprises 2,442 male-headed and 568 female-headed households. The region is a critical source of national food supply and export earnings, yet it faces irregular rainfall, climate shocks, land degradation and market volatility, and national survey data suggest food insecurity rates there of roughly 20 to 35 percent, well above the national average of around 8 percent.

To quantify resilience, the authors used the Resilience Index Measurement and Analysis II framework developed by the Food and Agriculture Organization. This approach treats resilience as a latent, multidimensional capacity built from four pillars: access to basic services, assets, social safety nets and adaptive capacity. Each pillar is constructed through factor analysis from observable household characteristics, and the pillars are then combined into a standardized Resilience Capacity Index scaled from 0 to 100. Access to basic services in this study includes indicators such as drinking water sources, sanitation, electricity and distance to health facilities; the asset pillar covers wealth indices and livestock holdings measured in tropical livestock units; social safety nets capture cash transfers, food aid, vouchers and association membership; and adaptive capacity reflects education, dependence ratios, coping strategies and income diversity.

The descriptive picture of the study area reveals uneven access to the building blocks of resilience. About 85 percent of households had access to an improved water source and roughly 59 percent to electricity, but only 6.3 percent had an improved evacuation mode, 29 percent had solid waste disposal and 27.7 percent had a sludge removal system. Social safety net coverage was strikingly low: fewer than 1 percent of households received food vouchers, 13 percent received cash assistance, 2.7 percent received food aid and only 0.6 percent had sickness coverage, while 38 percent belonged to an association. Nearly 97 percent of households reported using at least one coping strategy to deal with food insecurity, and the average reported monthly salary was 30,842 FCFA, with a standard deviation pointing to substantial income inequality.

When the researchers estimated which pillars actually drive the resilience index, three of the four emerged as statistically significant. A one-unit increase in the access to basic services pillar was associated with a 0.35-unit increase in the resilience index, the largest contribution, followed by adaptive capacity at 0.16 and assets at 0.09. The social safety net pillar, although positive, was not statistically significant, a result the authors note is consistent with earlier studies in Senegal. The survey data also showed that assistance variables such as cash aid and food aid were negatively associated with resilience, which the authors interpret as evidence that aid is targeted to already vulnerable households rather than that assistance undermines resilience. Interestingly, the agricultural asset index per capita showed a negative association with resilience, which the authors attribute to the fact that in predominantly rainfed systems with limited irrigation, insurance and market access, greater agricultural asset holdings may signal stronger exposure to climatic and price shocks rather than greater capacity to absorb them.

A key methodological feature of the study is its treatment of endogeneity. Because resilience and food security may be jointly determined, with reverse causality, omitted variables and measurement error all plausible, the authors used a two-stage least squares instrumental variables approach. The instruments were the distance from the household to the nearest mosque or church and the number of religious brotherhood groups in the community, variables chosen because religion and religious networks are frequently used as proxies for social capital, which has been observed to strengthen resilience. The first-stage results supported the relevance condition: the number of religious groups had a positive and significant effect on the resilience index, while greater distance to a mosque or church had a negative and significant effect. A Wu-Hausman test rejected the exogeneity of the resilience index, motivating the instrumental variables approach. The authors caution that the validity of the instruments ultimately depends on the exclusion restriction, which could be threatened if religious proximity affects food security through channels other than resilience, and that because the index is built from self-reported survey indicators, coefficient magnitudes should be interpreted cautiously.

The second-stage estimates showed that higher resilience capacity is associated with better food security: on average, the resilience index increased the Food Consumption Score by about 0.71 points and dietary diversity by about 0.03 food groups. The analysis also captured the role of shocks. A one-unit increase in the Standardized Precipitation Index, a drought indicator computed from CHIRPS satellite and station rainfall data, reduced the Food Consumption Score by 0.36 points and dietary diversity by about 0.02 food groups, while livestock theft reduced the Food Consumption Score by about 2.43 points and dietary diversity by about 0.14 food groups. Female-headed households had resilience index values averaging 58, higher than the male-headed average of 52, a pattern the authors link to the fact that 78 percent of female-headed households in Senegal live in urban areas, where resilience tends to be higher, and to lower poverty rates among female-headed households.

To move beyond average effects, the study employed a dose-response function framework based on the generalized propensity score method for continuous treatments, which allows the effect of resilience to vary across the entire distribution of the resilience index rather than at a few discrete points. The average treatment effect indicated that a percentage increase in resilience intensity raised the Food Consumption Score by approximately 5 percent and dietary diversity by about 16 percent, while among treated households the corresponding figures were 8 and 27 percent. The average treatment effect on the untreated was negative and large in magnitude, with estimated reductions of roughly 152 percent in the Food Consumption Score and 336 percent in dietary diversity, indicating substantially worse outcomes among households with very low resilience. The ordering of the treatment effects, with the effect on the treated exceeding the average effect, which in turn exceeds the effect on the untreated, points to a positive selection pattern in which households with higher resilience benefit more from further increases. The dose-response curves also revealed non-linearity: outcomes improved with resilience up to a maximum and then declined slightly, a pattern the authors suggest may reflect a resource-allocation trade-off in which limited household resources are spread more thinly across multiple investments as resilience intensity rises.

The gender-disaggregated analysis produced some of the study’s most distinctive results. Although resilience capacity had a positive and statistically significant effect on both food security measures for both genders, the effect on dietary diversity was more pronounced for female-headed households, a pattern the authors interpret through intra-household resource allocation theory, since empirical evidence indicates that women tend to allocate resources toward food quality, dietary diversity and nutrition-related expenditures when they have greater control over household resources. The average treatment effect on the treated was also larger for female-headed households, particularly for dietary diversity. At the same time, the negative treatment effects on the untreated were larger for female-headed households, suggesting that those in the lower tail of the resilience distribution are particularly vulnerable, possibly reflecting weaker access to land, credit and labor markets and fewer buffers to smooth consumption. For female-headed households, the dose-response curves showed food security outcomes improving with resilience up to an optimal range of roughly 60 to 70 on the index, after which marginal gains diminished, whereas the relationship for male-headed households was more monotonic.

The authors conclude that policy measures should strengthen gender-responsive resilience-building interventions, targeting the constraints faced by low-resilience households and ensuring a more equitable distribution of benefits. They point to enhancing access to productive assets, promoting climate-smart agricultural practices, maintaining social protection mechanisms against shocks such as droughts and livestock losses, and investing in early warning systems and index-based insurance schemes. Because the analysis is cross-sectional and relies on instrumental assumptions that cannot be fully verified, the estimates are best read as carefully identified associations rather than definitive causal quantities, but the overall pattern is clear: in this Senegalese study region, building resilience capacity is associated with better food consumption and dietary diversity, and the households that stand to gain the most, and lose the most without it, are often those headed by women.

Subject of Research: The heterogeneous impact of household resilience capacity on food and nutrition security in Senegal's Groundnut Basin, disaggregated by the gender of the household head.

Article Title: Heterogeneous impact of household resilience on food and nutrition security: evidence from Senegal

Article References: Faye, A., & Abdulai, A. (2026). Heterogeneous impact of household resilience on food and nutrition security: evidence from Senegal. Agriculture & Food Security, 15(1), Article 48. https://doi.org/10.1186/s40066-026-00630-y

Image Credits: AI Generated

DOI: 10.1186/s40066-026-00630-y

Keywords: Senegal, food security, household resilience, RIMA-II, Groundnut Basin, female-headed households, dietary diversity, Food Consumption Score, dose-response function, instrumental variables, gender, resilience capacity index