Identity, Agency and Networks in the Ultra-Poor Graduation Model in Bangladesh: Experimental Assessment of Varying Coaching Intensity and Village Assistance Committees (VAC)

BRAC’s ultra-poor graduation model has been effective in lifting poor people out of poverty. This research aims to uncover the role of social and psychological factors in its success and further optimize the model by accounting for these factors. It builds on an ongoing experimental study to isolate the causal role played by forms of coaching and psychological support and by village-level support structures on explaining the overall outcomes and the extent to which they are mediated through changes in people’s sense of agency, efficacy and identity, and whether and how community level networks reinforce this.

Researchers: Stefan Dercon; Dr Imran Matin; Dr Narayan C. Das; Dr Munshi Sulaiman; Atiya Rahman

Partners: University of Oxford

Timeline: January 2022August 2022

Status: Ongoing

Method: Quantitative

Contact: Dr Narayan C. Das;


BRAC’s graduation model has shown great promise across different contexts in moving households out of extreme poverty in Bangladesh as well as in a number of African countries. Research such as Bandiera et al. (2017) has shown that the support package offered, including an asset transfer, skills training and other components, has substantial benefits for the poorest households. This study aims to build on this and ask whether it can be designed better, by unpacking its contributing factors and boosting certain features, and highlighting the parts that focus on the psychological and social framing of this work.


BRAC’s graduation model has been built with substantial (and relatively resource-intensive) coaching of the recipients. The study aims to uncover the role of social and psychological factors in its success and further optimize the model by accounting for these factors. It will also try to find out the role Village Assistance Committees (VAC) play as part of the graduation model in the program’s overall impact and how the community networks are reinforcing or changing their position, and examine the networks the beneficiaries have within the community. 

This study is relevant to SDG 1 (no poverty), particularly to ending poverty in all its forms everywhere.


In its fourth phase of the Ultra-Poor Graduation (UPG) programme, BRAC conducted a baseline survey in 2016 before rolling out the randomized interventions. This research will conduct a follow-up survey on the beneficiaries and an additional sample of non-beneficiaries. The study will cover 88 branch offices (the smallest unit in BRAC’s administrative structure) from 11 districts. These 88 branches will be randomly assigned into three treatments and a control group of 22 branches each (stratified by district). The baseline survey drew a sample of 100 eligible ultra-poor households from each branch. This study will measure the impact by dividing the sample into three coaching variations: the weekly visit (T1) is the default and has the highest intensity and operational cost. The second treatment arm (T2) is designed to test whether reducing the coaching intensity by half (from weekly to bi-weekly home visits) results in a reduction in impact (on productive and financial assets, income and consumption) that is higher than the reduction in operational costs. Finally, the third treatment (T3) includes a bi-weekly group meeting and bi-weekly household visits, resulting in a coaching intensity variation that costs between T1 and T2. Besides the cost-effectiveness aspect, the group meeting will also work to create social networks among the beneficiaries.

Findings and Recommendations

Study ongoing.