Drought Exposure and Food-Security Transitions in Sub-Saharan Africa: A Hidden Markov Approach

This paper examines food-security dynamics in Sub-Saharan Africa using a Hidden Markov Model and harmonized LSMS-ISA panel data for Ethiopia, Malawi, and Nigeria across four survey waves. We model household dietary diversity as an observed indicator of an underlying food-security regime and study how early drought exposure shapes transitions between latent states. Climate stress is measured using an early drought-warning indicator based on the 3-month Standardized Precipitation Evapotranspiration Index. Model selection identifies two regimes based on low and high dietary diversity. Transitions are strongly asymmetric: households in the high-diversity state show high persistence, with a probability of 0.92 of remaining there, whereas one in three food-insecure households fails to transition upward, pointing to a structurally disadvantaged subset requiring targeted intervention. Early drought exposure is associated with a 9.8 percentage-point lower probability of belonging to the high-diversity state. Female-headed households are 5.9 percentage points less likely to belong to that state, whereas urban households are 20.6 percentage points more likely. The results are robust to different estimators, alternative SPEI accumulation periods, a drought-onset indicator, crop-adjusted drought exposure, and an alternative binary HDDS outcome defined relative to the country-specific median. Subnational estimates reveal substantial spatial heterogeneity in drought effects. The findings support food policies that link early-warning systems to adaptive safety nets, rural market access, and geographically targeted interventions.