Events

CDS Seminar: The Value of Prediction in Allocation

Speaker: Unai Fischer Abaigar (NYU)

Location: 60 Fifth Avenue, Room 7th Floor Open Space

Date: Friday, October 2, 2026

Abstract: Public institutions increasingly use predictive algorithms to guide the allocation of scarce societal resources. For instance, algorithms are used to target resources to jobseekers at risk of long-term unemployment, allocate cash transfers, and target tutoring services to students at risk of dropping out. From a design perspective, these risk predictors are challenging to evaluate because their value cannot be assessed without reference to the broader institutional context. Predictions are not an end in themselves, but a means to improve downstream welfare and advance institutional objectives. Once we accept that a predictive system is embedded in a broader institutional setting, that setting also implies a design space of alternatives a planner might pursue to raise downstream welfare, such as expanding the capacity of a program or improving service quality. This raises a more fundamental question: how valuable is improving prediction in solving allocation problems relative to the other investments a planner could make?

In this talk, I will discuss a line of work that develops mathematical and empirical tools to study these tradeoffs, drawing on real-world case studies in German employment services and targeted cash transfer programs in Ethiopia.

Joint work with Juan Carlos Perdomo, Emily Aiken and Christoph Kern.

Bio: Unai Fischer Abaigar is a Faculty Fellow in the Center for Data Science at New York University. His research focuses on algorithmic systems that inform decisions about individuals, particularly in public institutions. He examines not only predictive algorithms themselves, but also how they are designed, deployed, and integrated into institutional practices and processes. He earned his PhD in Statistics from LMU Munich and his BSc and MSc in Physics from Ruprecht-Karls-University of Heidelberg. During his PhD, he was a visiting fellow at the Center for Research on Computation and Society at Harvard University and a visiting student at the Laboratory for Information and Decision Systems at MIT.