Description
Non-governmental organizations (NGOs) are often responsible for implementing interventions in low-and middle-income countries, but little is known about how and why they are effective, for example, whether success (when there is success) is impacted by which NGO is implementing it and whether previous interaction with a community increases intervention uptake and effectiveness. To test the impact of one of these factors, NGO reputation, we worked with a locally-based NGO engaged in varied environmental and development activities to conduct a campaign to promote improved cookstoves (ICS) in rural India.
In this campaign, improved cookstoves were promoted to nearly 1,000 households across approximately 100 geographically distinct hamlets in Uttarakhand, India. Trained personnel visited homes to explain the benefits of ICS and identified themselves as being affiliated with the partner NGO. Half of those who received house visits lived in hamlets already familiar with the NGO, while the other half lived in hamlets that had not had any exposure to the NGO. Results suggest that ICS promotion increases adoption overall (nearly half of all households purchased an ICS) and that prior familiarity with the NGO further increased purchases: purchase rates were approximately 14 percentage points (31 percent) higher in communities with prior familiarity with the NGO. Additionally, those who were familiar with the NGO were 16 percentage points more likely to continue using intervention stoves than those who were not familiar, increasing the effect of the treatment by 50-80 percent.
These findings suggest there is a large, positive, and statistically significant “NGO reputation effect” that helps to also explain why successful implementation in one setting may not easily translate to success in another setting if a relationship with the NGO implementing the intervention is not already established.
Methods and Results
We developed a theoretical model of household decision-making in the presence of NGOs and transaction costs to measure the impact of the NGO on environmental health decisions. We then combined a stratified field experiment with a triple-differences estimation strategy to assess the impact of an NGO’s prior engagement on intervention success. To evaluate the relative strength of the evidence we uncovered and to see how it builds on and shapes the collective knowledge in this domain related to the “overall NGO effect,” we also interpreted our results within a simple Bayesian framework. To do so, we began by reviewing the implementer-identity literature to formally specify a distribution for our prior understanding of the differences in effectiveness between NGOs and other implementers (i.e., the “overall NGO effect”). Next, we asked how our study adds to this prior knowledge. To answer this question, we used this prior distribution as a critical component of a multilevel Bayesian analysis and showed that a synthesis of existing insights with our results in this way allows us to more precisely estimate the share of the overall NGO effect that is driven by an NGO implementer’s prior reputation among its beneficiaries.
Team
Members
Faraz Usmani (Mathematica), Marc Jeuland (Duke University), Subhrendu Pattanayak (Duke University)
Related Content
Related Links
Climate & Sustainability, Environment, Health, Economic Development, Program Implementation, Rural, Technology Diffusion, India