Despite growing recognition of disability as a driver of extreme poverty, evidence on effective livelihood interventions for persons with disabilities remains scarce, particularly across varying contextual settings. The Disability-Inclusive Ultra-Poor Graduation (DIUPG) programme adapts the established Ultra-Poor Graduation (UPG) framework to address extreme poverty among persons with disabilities by combining standard UPG components — such as asset transfers — with disability-specific support including rehabilitation and psychosocial therapy. Previously evaluated in Uganda, where it demonstrated improvements in household consumption despite the compounding challenges of COVID-19 and funding cuts, the programme is now being evaluated in Bangladesh to generate comparative evidence on its transferability and robustness across different contexts. The evaluation employs a cluster-randomised superiority trial with two parallel groups, where clusters represent branch office catchment areas of the implementing organization. Eligible households are identified through standardized screening tools targeting ultra-poor households with members with disabilities, with 48 clusters assigned to the intervention arm and 24 to the control arm. Primary outcomes include per capita household income, income and employment of persons with disabilities, per capita food consumption and expenditures, and per capita non-food expenditures. Baseline data were collected between May and June 2023, prior to programme implementation, with endline data collection scheduled for October to November 2025 following the 20-month intervention period. This evaluation will produce critical evidence on whether the benefits observed in Uganda can be replicated in a distinct socioeconomic and cultural context, thereby strengthening the global evidence base for disability-inclusive livelihood programming.
Authors: Carew, Mark T.; Rahman, Semab; Banks, Lena Morgon; Gomes, Mohima; Kamruzzaman, Md.; Das, Narayan; and Kuper, Hannah
Type: Journal Article
Publisher: Springer Nature
Year: 2026
