Year: 2026
Type: preprint
Abstract: Expected-loss provisioning can affect more than when banks recognize credit losses; it may also be associated with how banks allocate credit. We examine whether bank-year provision pressure after implementation of the Current Expected Credit Loss (CECL) standard is associated with mortgage denial outcomes across institutions, loan purposes, and borrower groups. We link 2018–2024 Home Mortgage Disclosure Act (HMDA) applications to the Philadelphia Fed HMDA Lender File and bank Call Reports. Provision pressure is measured as loan-loss provision expense scaled by gross loans, and the main tests use application-weighted cell regressions. The evidence is concentrated among banks with at least $10 billion in assets and in nonpurchase mortgage channels, especially home improvement and refinancing. In these settings, higher post-CECL provision pressure is associated with larger Black-White denial differences. Common-sample debt-to-income tests compare identical DTI-defined cells with and without DTI controls, and the focal Black interaction remains positive and statistically significant in all three principal channels. The findings are associational because provision pressure is not randomly assigned and public HMDA does not provide borrower credit scores or banks' internal underwriting information. The study identifies a financial-services setting in which expected-loss provisioning is systematically associated with the distribution of mortgage credit, consistent with a role for formal bank decision systems.
Source: Research Square
Authors: David Olsen, Cindy Greenman
Institution: Utah Tech University
Language: English
Subfield: Finance
Domain: Social Sciences
SDG: Reduced inequalities
Open Access status: green
OpenAlex ID: W7204195576