When an anti-poverty programme reaches a whole locality rather than a handful of households, it changes the wages, rents, and prices that everyone faces. This five-month scoping study builds the design and evidence base for a randomized saturation trial of the market-wide effects of BRAC’s Ultra-Poor Graduation programme in rural Bangladesh.
Researchers: Munshi Sulaiman, PhD; Imran Matin, PhD; Narayan Das, PhD; Oriana Bandiera, PhD; Robin Burgess, PhD; and Stefano Caria, PhD
Partners: BRAC Bangladesh (Ultra-Poor Graduation Programme); BRAC USA; STICERD, London School of Economics; and GiveWell
Timeline: July–November 2026
Status: Ongoing
Contact: Raisa Adiba; raisa.adiba@bracu.ac.bd
Context
BRAC’s Ultra-Poor Graduation (UPG) programme is among the world’s most rigorously evaluated poverty interventions, with sustained welfare gains documented for direct beneficiaries in Bangladesh and across sub-Saharan Africa. Yet almost all of this evidence stops at the household boundary. When a graduation programme operates at scale, it also moves the markets around it: agricultural day-labor wages, land rental and tenancy terms, and prices along local commodity value chains. Recent work on large-scale cash transfers in Kenya found a substantial local fiscal multiplier, as demand-side expansion raised incomes for non-recipients. Whether an asset-based intervention such as UPG produces similar spillovers in the densely populated, agrarian economy of rural Bangladesh remains unknown, leaving its welfare footprint and cost-effectiveness at scale incompletely understood.
Objectives
The scoping study aims to synthesize evidence on the market-level effects of graduation-type programmes; to characterize labor, land, and commodity market conditions across contrasting agro-ecological settings; to identify localities with sufficient ultra-poor density to support a saturation design; to develop an operationally feasible randomized saturation design for the next phase of work.
Methodology
The study will run three parallel workstreams. The first analyses 2022 population and housing census microdata matched to small area poverty estimates, alongside upazila land office records and BRAC’s administrative coverage data, to identify unions with sufficient ultra-poor density to sustain a saturation design. The second collects primary scoping data across roughly 400 unions in four agro-ecological zones, namely the drought-prone Barind tract, saline-affected coastal areas, riverine chars, and the Chittagong Hill Tracts, covering casual labor market structure and prevailing wage rates, savings and credit portfolios, commodity value chains, land tenancy and rental arrangements, and the availability of administrative records for tracking. The third translates these inputs into the trial design itself: a union-level randomization into high-saturation, low-saturation, and pure control arms, with villages and households randomized within, together with cluster matching protocols, power calculations, and a cost model.
Findings and Recommendations
Forthcoming.
