Beyond the Payout Window: Modelling the Durability of Poverty Graduation After Large Lump Sum Cash Transfers in Rural Kenya

Abstract

Cash transfer programmes have generated compelling evidence of short to medium term welfare gains in sub-Saharan Africa, yet a fundamental question persists largely unanswered: how long do these gains last, and what determines whether a household that exits poverty after a transfer stays out? Conventional impact evaluations rely on fixed follow-up windows of twelve to thirty-six months, producing outcome snapshots that cannot characterise the durability of poverty graduation as a dynamic process. This paper proposes a methodological reorientation: applying actuarial survival analysis, specifically adapted Cox proportional hazards modelling, to the study of post-transfer poverty reversion among smallholder farming households in rural Kenya. Using a cohort design that maps directly onto the operational structure of large lump sum transfer programmes, we demonstrate how time-to-event methods transform our understanding of poverty graduation from a binary outcome into a time-varying trajectory shaped by household characteristics, asset dynamics, climate shocks, and the presence or absence of complementary support. The framework produces three outputs unavailable from conventional evaluation designs: poverty reversion survival curves, covariate-specific hazard ratios for reversion risk, and a predictive scoring tool that programme managers can deploy during recipient selection to stratify households by their projected graduation durability. We argue that this actuarial reorientation is not merely a methodological refinement. It represents a fundamental shift in how we ask whether cash transfers work: from did the transfer lift this household? to how long will this household remain lifted, and why?