AER Paper: Anti-manipulation ML improves credit forecasting in Kenya

Afinetheorem · x · 2026-08-28

Published in the American Economic Review, this paper addresses the vulnerability of ML models to strategic manipulation where agents 'fake their type'. The authors derive a more robust training method and apply it to cell-based credit scoring in Kenya, showing significant improvements in real-world performance.

Original post →

More from Research

Research channel →