Expected-Loss Provisioning and the Distribution of Mortgage Credit

Every time a bank lends money, it faces a simple but critical question: how likely is it that the borrower will fail to pay it back? For decades, banks answered this by waiting until a loan actually went bad before setting aside money to cover the loss. This approach meant that during good economic times, banks often held too little cash in reserve, only to scramble for it when a crisis hit. In recent years, a new accounting rule called the Current Expected Credit Loss standard changed this rhythm. Instead of waiting for trouble to arrive, banks now must estimate future losses the moment they make a loan and set aside money for those potential losses immediately. This shift forces banks to look forward, treating the risk of default as a cost that exists right now, not just a possibility for tomorrow. While this change was designed to make bank balance sheets more honest and resilient, it raised a quiet question for researchers: does this new way of counting risk change not just how much money banks lend, but who gets to borrow it?

A team of researchers at Utah Tech University set out to find the answer by looking at the massive flow of mortgage applications across the United States. They wanted to see if the pressure banks feel to set aside more money for potential losses was influencing which mortgage applications got approved and which got denied. To do this, they built a unique picture of the lending landscape by connecting three massive public datasets. They took records of millions of mortgage applications, which include details about the borrower's race, income, and the type of loan they sought. They linked these applications to the specific banks that processed them. Finally, they attached financial reports from those banks that showed exactly how much money each institution was setting aside for bad loans in a given year. This allowed them to see if banks that felt more financial pressure to reserve funds for losses were treating different groups of borrowers differently.

The researchers focused their investigation on the largest banks in the country, those with at least ten billion dollars in assets. They reasoned that these massive institutions rely heavily on computer models and standardized rules to make lending decisions, rather than personal relationships with local borrowers. If a new accounting rule changed the way these banks thought about risk, the effect should be most visible in their automated systems. They also looked closely at different types of home loans. They distinguished between loans used to buy a new home and loans used for other purposes, such as refinancing an existing mortgage or funding home improvements. These latter loans are crucial because they help homeowners manage their finances and maintain their property value long after they have moved in.

The study examined data from 2018 through 2024, a period that captures the years before and after the new accounting rule took full effect. The researchers did not look for evidence that banks were intentionally discriminating against specific groups. Instead, they asked a more subtle question: when a bank feels a higher pressure to set aside money for potential losses, does that pressure lead to a larger gap in approval rates between Black applicants and White applicants? They controlled for many factors that could influence a loan decision, such as the size of the loan, the income of the borrower, and the value of the home. They also paid special attention to the debt-to-income ratio, a key measure of how much a borrower owes compared to how much they earn, to ensure the results were not simply driven by differences in borrower debt levels.

The findings revealed a clear pattern, but one that was concentrated in specific places. The researchers found that after the new accounting rule was implemented, banks with at least ten billion dollars in assets that felt higher pressure to set aside money for losses were more likely to deny mortgages to Black applicants than to White applicants. This gap in denial rates was not uniform across all types of loans. It was most pronounced in loans that were not for buying a new home. Specifically, the disparity grew larger for loans used to refinance existing mortgages and for loans used to fund home improvements. In these channels, higher pressure to reserve funds for losses was associated with a significantly wider gap in approval outcomes between Black and White borrowers. For example, in the refinancing market, a shift in the bank's provision pressure was linked to a difference in denial rates that could reach over two percentage points, a meaningful amount when applied to millions of applications.

The study also looked at whether this pattern held up when the researchers accounted for the specific financial health of the borrowers. They found that even after controlling for the debt-to-income ratio, the gap remained. In the case of home improvement loans, accounting for debt levels explained part of the difference, but a significant portion of the gap persisted. This suggests that the pressure to set aside money for losses was influencing decisions in ways that went beyond simple measures of a borrower's ability to pay. The results were less consistent for Hispanic and Asian applicants, indicating that the effect was not a broad trend affecting all minority groups in the same way, but rather a specific dynamic between Black and White applicants in the large-bank sector.

Crucially, the researchers were careful to state what their study did not prove. They did not find evidence that banks were intentionally discriminating or that the new accounting rule was designed to harm specific groups. The data showed an association, not a direct cause-and-effect relationship. The pressure to set aside money for losses is not randomly assigned to banks; it is tied to the bank's own portfolio and economic conditions. Furthermore, the public data used in the study did not include credit scores or the internal risk models banks use, so the researchers could not see every factor that went into a final decision. However, the consistency of the results across different tests, including those that compared borrowers within the same bank and the same year, suggests that the new accounting framework is interacting with bank risk systems in a way that alters the distribution of credit.

The implications of this discovery extend beyond the balance sheets of banks. It suggests that when accounting rules change how banks measure risk, those changes can ripple through the financial system and affect the lives of homeowners in unexpected ways. For a family trying to refinance a mortgage to lower their monthly payment, or a homeowner needing a loan to fix a leaky roof, the decision to approve or deny their application might be influenced by a bank's accounting calculations. The study highlights that the rules governing how banks count their money can shape who gets access to the financial tools needed to maintain and improve their homes. By linking the abstract world of accounting standards to the concrete reality of mortgage approvals, the research offers a new perspective on how financial regulations can have real-world consequences for the distribution of credit and the stability of homeownership.