Interventions in remote, rural settings face high transaction costs. We develop a model of household decision-making to evaluate how non-governmental organizations (NGOs) address these implementation-related challenges and influence intervention effectiveness. To test our model's predictions, we create a sample of observationally similar Indian villages that differ in their prior engagement with a local development NGO. In partnership with this NGO, we then stratify a randomized technology promotion intervention on this institutional variable. We uncover a large, positive, and statistically significant "NGO effect": prior engagement with the NGO increases the effectiveness of our intervention by at least 30 per cent. Our results have implications for the generalizability of experimental research conducted jointly with NGOs. In particular, attempts to scale-up findings from such work may prove less successful than anticipated if the role of NGOs is insufficiently understood. Alternatively, policy makers looking to scale-up could achieve greater success by enlisting trusted local partners.
external validity field experiments NGOs non-governmental organizations