Modernizing Fair Lending (2025)

Fair lending’s disparate impact doctrine aims to address lending disparities. But which disparities? Traditional fair lending has narrowly focused on equal outcomes—examining differences in loan approval rates or interest rates. However, this singular focus overlooks other critical dimensions of disparities that are essential for fair credit access.

We challenge the conventional focus on equal outcomes, demonstrating how it has failed to address some of the most pernicious harms of traditional credit allocation and has stifled necessary machine-learning and alternative data innovations. We argue that disparities in validity of creditworthiness predictions —the accuracy with which a model identifies creditworthy applicants—severely impact equal access to credit and fail to equally extend credit to the creditworthy. Despite the mounting empirical evidence of the harm of validity disparities, traditional fair lending enforcement inadequately recognizes this disparity dimension, a gap that will become increasingly harmful as lending decisions rely on advanced statistical methods.

Our updated, holistic account of disparities engages with how the competing fairness notions of equal outcomes and equal validity may be fundamentally in tension. Reducing differential validity by improving data quality and models for protected groups could increase outcome inequality. On the other hand, decreasing outcome inequality could exacerbate differential validity when noisy and inaccurate lending decisions are made for the most vulnerable borrowers.

Using a lender simulation exercise, we demonstrate that these tensions can be addressed by balancing equal outcomes and validity, providing regulators the flexibility to weigh each dimension appropriately. We discuss how this approach can be practically implemented by lenders and regulators; we chart a path forward for considering other tensions within fair lending, such as balancing business benefit with discriminatory impact; and we address how fair-lending policies impact different protected groups. Our framework has direct implications for other domains—such as housing and employment discrimination—where competing disparity notions should be considered under discrimination doctrines.