The Value of Prediction in Identifying the Worst-Off

Unai Fischer Abaigar, Christoph Kern, Juan Perdomo

International Conference on Machine Learning 2025 · Oral

This article delves into a crucial dilemma faced by public agencies: how to effectively allocate scarce resources to individuals most in need, especially when leveraging machine learning (ML) for decision-making. Presented by Unai Fischer Abaigar from the University of Munich, alongside co-authors Christoph Kern and Juan Perdomo, the talk "The Value of Prediction in Identifying the Worst-Off" investigates the complex tradeoffs between improving predictive model accuracy and expanding institutional capacity. It challenges the intuitive assumption that better predictions always lead to better outcomes, particularly in public sector contexts focused on welfare.

AI review

A careful, policy-relevant theoretical paper that formalizes the tradeoff between predictive accuracy and institutional capacity in public-sector screening problems. The central object — the Prediction Access Ratio — is clean and the core message is directionally correct and underappreciated in applied ML circles. The theoretical contribution is real but modest: linear-Gaussian assumptions do most of the heavy lifting, and the empirical validation, while honest, confirms rather than stress-tests the framework. A solid contribution for researchers working at the ML-policy interface; less…