one illness undoes the ladder

one illness undoes the ladder

Problem statement

BRAC's Targeting the Ultra-Poor (TUP) graduation model — now replicated in over 50 countries — provides a bundled intervention (productive asset transfer, skills training, consumption support, savings facilitation, and coaching) to move extremely poor households above a sustainable livelihood threshold. The model works: the landmark 6-country RCT showed durable gains at 3+ years. But BRAC's own internal data reveals that health shocks (illness, injury, disability, or death of a household member) are the single largest cause of graduation failure and post-graduation relapse. Among non-graduating households in BRAC's Bangladesh program, 40–60% cite health crises as the primary reason. Yet health is not a core pillar of the graduation model — it appears as an add-on referral, not as an integrated program component with the same intensity as asset transfer or skills training.

Why this matters

The graduation approach has become the dominant anti-extreme-poverty strategy globally, endorsed by the World Bank, CGAP, BRAC, and dozens of implementing agencies. Over 100 graduation programs operate across 50+ countries. If health shocks are the primary failure mode, the entire global portfolio of graduation programs is systematically under-investing in the component that most determines success or failure. The ultra-poor population is, by definition, the most health-vulnerable: no savings buffer, no insurance, limited access to healthcare facilities, and labor-intensive livelihoods where even minor illness produces income loss. A single hospitalization can cost 3–6 months of household income, wiping out the asset base that the graduation program transferred.

What’s been tried and why it hasn’t worked

BRAC has experimented with linking TUP participants to government health services, providing basic health education, and including health referral in coaching visits. These approaches fail because they assume the health system will serve ultra-poor participants — but the ultra-poor face the highest barriers to health system access: distance, cost, stigma, informal fees, and time costs that exceed what they can bear. Referral to a health system that effectively excludes you is not health protection. Community-based health insurance schemes have been piloted but struggle with adverse selection (the ultra-poor are too high-risk for community pools) and premium affordability. Microinsurance products for ultra-poor populations have generally failed because premium collection costs exceed the small premiums the ultra-poor can pay.

What would unlock progress

The graduation model needs health shock protection as an integrated component, not an external referral. BRAC's own CHW network (Shasthya Shebika) operates in the same communities as TUP — but the two programs operate as separate verticals with different management structures, budgets, and reporting lines. Designing a joint TUP-CHW intervention — where the CHW assigned to a TUP household provides proactive health monitoring during the 24-month graduation window — could address the most common health failure modes (delayed care-seeking, untreated chronic conditions, maternal health emergencies) without building new infrastructure. The design challenge is integration across program silos within the same organization.

Entry points for student teams

A health systems team could work the failure question from the published replication archives rather than from BRAC's internal records, which no student team has standing to request: the six-country graduation RCT is posted in full on the Innovations for Poverty Action Dataverse with data and Stata code under CC0 (doi:10.7910/DVN/NHIXNT), and the Ghana graduation trial is posted alongside it (doi:10.7910/DVN/JROFZP). The first task there is reading the country codebooks to see which health measures each survey actually carries — illness episodes, health spending, days of lost work, care-seeking — and then testing whether those measures predict which households slipped back, which conditions carry the most weight, and how the timing lines up with the transfer. A design team could prototype an integrated CHW-TUP health monitoring protocol for the graduation window — what proactive health services, delivered by existing CHWs, would most reduce the health shock failure mode? — and specify the evaluation an implementer would run to test it, since the 24-month graduation window rules out running the trial in a semester. A policy team could compare health integration across the 50+ country graduation programs to identify which designs have reduced health-related attrition and why.