Abstract
Climate-smart agriculture (CSA) is widely promoted to improve smallholder livelihoods under increasing climate risks, yet evidence on impacts beyond productivity, particularly multidimensional food security, remains limited. This study examines determinants of adoption of selected CSA practices and their effects on maize productivity, farm income, food expenditure share, and dietary diversity among 510 smallholder households in Kasungu District, central Malawi. Using 2023 cross-sectional survey data, we employ a multinomial endogenous switching regression model to address potential selection bias across eight mutually exclusive adoption options based on three CSA practices. Gender of the household head, extension participation, and cooperative membership are significantly associated with adoption, with female-headed households more likely to adopt multiple-practice combinations. Average treatment effect on the treated (ATT) estimates indicate significant maize-yield and farm-income gains for certain options relative to the estimated non-adoption counterfactual. However, no practice option shows a statistically significant ATT estimate for dietary diversity measured by the Food Consumption Score, indicating a disconnect between estimated farm-income gains and dietary outcomes. The findings suggest that CSA may support productivity and income, but complementary nutrition-sensitive interventions may be needed to achieve multidimensional food security.