Income, not health habits, drives Nigerian smallholder farmers’ willingness to pay for safer food

Smallholder farmers in Southwest Nigeria who earn the least are almost never willing to pay a premium for safe and nutritious food, while income emerges as the single strongest predictor of that willingness, according to a cross-sectional survey of 294 farmers published in Agriculture & Food Security. The study, guided by Dual System Theory, collected responses from Osun, Ondo, and Ekiti states between March 8 and 25, 2025, using structured questionnaires that included an adapted Health-Promoting Lifestyle Profile and Gabor-Granger-based willingness-to-pay tiers.

The researchers found that income dominated the willingness-to-pay outcome. In an ordinal regression model, farmers in the lowest income bracket had dramatically reduced odds of paying a premium compared with the highest income group, with odds reduced by 99.9 percent (odds ratio 0.001, 95% CI 0.000–0.003). Income was the strongest predictor of willingness to pay, with a Wald χ² of 87.969 and p < 0.001 reported in the abstract, and lower-income groups had near-zero odds of paying premiums. Mean annual income rose progressively across willingness-to-pay categories, from ₦78,068.72 among farmers unwilling to pay any premium, to ₦155,504.23 among those willing to pay up to 10 percent, ₦201,910.11 for up to 20 percent, and ₦258,208.07 for those willing to pay more than 20 percent.

Health-related behaviour, by contrast, showed no significant main effect on willingness to pay in the moderation model (odds ratio 1.589, p = 0.210). Yet it significantly moderated the influence of age and education. The interaction between health-related behaviour and age was highly significant (odds ratio 2.228, p < 0.001), meaning that for each year of age, the effect of health behaviour on willingness to pay more than doubled. Older farmers with strong health-promoting habits were over twice as likely to pay premiums. In contrast, the interaction between health-related behaviour and tertiary education was significantly negative (odds ratio 0.251, p = 0.023), suggesting that highly educated farmers with strong health behaviours may paradoxically be less willing to pay, possibly because they distrust formal certifications or over-rely on personal judgment about food safety.

Household size also played a meaningful role. Each additional household member increased the odds of being in a higher willingness-to-pay category by 24.5 percent (odds ratio 1.245, p = 0.009), a modest but statistically significant effect that may reflect greater collective concern for family health or pooled financial resources in larger households. Age and education, on their own, did not significantly predict willingness to pay in the ordinal regression (age p = 0.170; education all p > 0.271 across bins), indicating that formal schooling does not translate directly into heightened food safety valuation within this population.

An analysis of variance confirmed the income gradient. Income differed highly significantly across willingness-to-pay groups (F(3, 290) = 199.84, p < 0.001), as did health-related behaviour (F(3, 290) = 51.33, p < 0.001). Household size showed a statistically significant variation (F(3, 290) = 3.21, p = 0.024). Years spent in school did not significantly differ across groups (F(3, 290) = 1.27, p = 0.287), and age yielded a borderline result (F(3, 290) = 2.60, p = 0.052). Post-hoc tests showed that mean income increased progressively with higher willingness-to-pay levels, with the difference between the lowest and highest groups approximately ₦180,139 (95% CI ₦159,681 to ₦200,598). Health-related behaviour scores also rose across groups, with the largest gap between the non-paying group and the highest-premium group (mean difference −1.01), though the difference between the two highest-premium groups was not significant (p = 0.413), suggesting a plateau at higher willingness-to-pay levels.

The study operationalized safe and nutritious food through a consumer-centric, behavioural lens rather than direct biochemical measurement. Farmers’ willingness to pay a price premium for food described as safe and nutritious served as the direct, quantifiable proxy for their valuation of this attribute bundle, using the Gabor-Granger technique with four price tiers: not willing to pay, up to 10 percent, up to 20 percent, and more than 20 percent. The adapted Health-Promoting Lifestyle Profile, originally developed by Walker et al. (1987), was modified with three context-specific items related to occupational health and safety in farming, such as using protective gear when applying agrochemicals. The adapted scale demonstrated good reliability with a Cronbach’s alpha of 0.82, and a principal component analysis supported unidimensionality with a single dominant factor accounting for 58.4 percent of total variance.

Sampling followed a multi-stage design across 10 Agricultural Development Project zones, with random selection of 10 Local Government Areas and 10 villages, and 39 farmers sampled per village across crop, livestock, and fish production enterprises. Although 390 questionnaires were distributed and a minimum sample of 384 was determined using Cochran’s formula, the study achieved a 77 percent response rate, yielding 294 usable responses. Post-hoc power analysis confirmed at least 89 percent power to detect small-to-medium effects. The ordinal regression model demonstrated strong statistical validity, with a highly significant overall model χ² of 185.76 (p < 0.001), a Nagelkerke pseudo R² of 0.53, and a non-significant test of parallel lines (χ² = 14.32, p = 0.160) confirming the proportional odds assumption. The interaction model improved fit further, with a Nagelkerke pseudo R² of 0.62 and a lower AIC of 355.98 compared with 379.08 for the main-effects-only model.

The researchers identified potential intervention tipping points through significant interaction effects rather than separate variables. Education emerged as a leveraging tipping point: significant negative coefficients for secondary-by-health-behaviour (B = −1.09, p = 0.033) and tertiary-by-health-behaviour (B = −1.38, p = 0.023) interactions suggest that farmers with at least secondary education represent a cohort where health behaviour interventions could yield super-additive effects, though the direction of these interactions is negative, meaning higher health behaviour scores actually reduce willingness to pay among the educated. Age acted as an amplifying tipping point, with the strong positive age-by-health-behaviour coefficient (B = 0.80, p < 0.001) identifying older farmers with high health behaviour as a key demographic where combined factors synergize. Income served as a prerequisite tipping point, with a marginally significant negative interaction for the lowest income group (p = 0.104) suggesting that below approximately ₦144,682 annual income, economic constraints dominate decision-making and must be addressed before behavioural strategies can exert full influence.

The authors frame these findings as an extension of Becker’s income-deterministic model of household behaviour, positioning health-related behaviour as a conditional driver that operates independently and interactively with income rather than as a standalone factor. They recommend tiered subsidies targeting households earning below ₦221,197, behavioural nudges for older farmers through community health campaigns, family-centric interventions for larger households, transparent safety labeling to rebuild trust with educated consumers, and dual-pathway policies combining economic support such as microloans with behavioural tools. The study’s cross-sectional design prevents causal inference about the direction of these relationships, and the authors note that the sample size provides limited power for subgroup analysis within specific tipping point combinations. They call for experimental or longitudinal designs to test whether interventions tailored to these identified tipping points actually produce the predicted super-additive effects in practice, and for qualitative investigation of the cognitive and social mechanisms underlying these statistical interactions.

Subject of Research: Willingness to pay for safe and nutritious food among smallholder farmers in Southwest Nigeria

Article Title: Health-related behaviour and willingness to pay for safe and nutritious food among smallholder farmers in Southwest Nigeria

Article References: Ikuemonisan, E. S. (2026). Health-related behaviour and willingness to pay for safe and nutritious food among smallholder farmers in Southwest Nigeria. Agriculture & Food Security, 15(1), Article 30. https://doi.org/10.1186/s40066-026-00612-0

Image Credits: AI Generated

DOI: 10.1186/s40066-026-00612-0

Keywords: willingness to pay, food safety, smallholder farmers, Nigeria, health-related behaviour, income, ordinal regression, Dual System Theory, foodborne illness, nutrition security, Gabor-Granger, cross-sectional survey