Seminar: Mushfiq Mobarak (Yale University)

Ahmed Mushfiq Mobarak is a Professor of Economics at Yale University with concurrent appointments in the School of Management and in the Department of Economics. Mobarak is the founder and faculty director of the Yale Research Initiative on Innovation and Scale (Y-RISE).

He also co-chairs the Urban Services Initiative at the Jameel Poverty Action Lab (J-PAL) at MIT, and leads the Bangladesh Research Program for the International Growth Centre (IGC) at LSE and Oxford. He has previously worked at the World Bank and the International Monetary Fund. Mobarak has several ongoing research projects in Bangladesh, Nepal, and Sierra Leone. He conducts field experiments exploring ways to induce people in developing countries to adopt technologies or behaviors that are likely to be welfare-improving. He also examines the complexities of scaling up development interventions that are proven effective in such trials.

His research has been published in journals across disciplines, including Econometrica, Science, American Economic Review, Review of Economic Studies, the American Political Science Review, Nature, Proceedings of the National Academy of Sciences, and Demography, and covered by the New York Times, The Economist, NPR, BBC, NBC, The Washington Post, Wall Street Journal, Science, Nature, the Times of London, and other media outlets around the world. He received a Carnegie Fellowship in 2017.

Presenting --

Title: "Poverty Targeting at Scale: Algorithmic vs. Traditional Approaches"

Abstract: Innovations in data and algorithms are enabling new approaches to targeting policies and interventions at scale. We compare three paradigms for poverty targeting — proxy means-testing, community based targeting, and an algorithmic approach using machine learning and big data — and study how their effectiveness varies with the scale and scope of the program. We collect new data from Bangladesh, including mobile phone records from four phone companies, community-based wealth rankings from 180 communities, a census of 100,000 households, and detailed surveys of 5,000 households, to measure the accuracy of these three methods at identifying poor households. We combine these new data with existing mobile phone and survey data to show that, after accounting for large differences in costs, algorithmic targeting is most cost-effective for programs with limited budgets spread across large populations, while proxy-means testing is superior for programs with large budgets relative to the population screened.

Location: HE-114 in the Library