New Fed Data Reveals Growing Strain on Subprime Borrowers

A big batch of new data in the Federal Reserve Bank of Philadelphia’s latest release reveals subprime account delinquencies remain elevated and more consumers are making only minimum payments.

The 67 new variables added to the Philadelphia Fed’s Large Bank Credit Card and Mortgage Data series will allow users to better understand how consumers use and access credit cards and mortgages from the country’s largest banking institutions.

This information is noteworthy for lenders reliant on subprime customers because a more stringent underwriting process is likely propping metrics, but shifting who receives special treatment in the process.

The revised Philly Fed dataset places more effective decision-making resources in the hands of mortgage and credit decision-makers.

While rates are still high, and household budgets are strained, lenders must strike a balance between staying in the black and serving the borrowers most in need of credit.

Minimum Payments, Maximum Risk

The minimum-payment-only proportion of credit card accounts just hit a 12-year peak. That’s a red flag, particularly for subprime lenders. Minimum payment debtors are often one misstep away from defaulting.

While other indexes have fewer late payment rates, the number of 90-day past-due accounts tells a different tale: deep trouble is becoming more routine.

Lenders use these trends to reassess just how much risk they are taking on. Even a modest rise in long-term delinquencies, for example, can signal that they are about to crest a broader wave of default. Such signals push lenders to reassess loan-loss reserves and consumer-retention models, not just interest-rate spreads.

Access Shrinks for the Middle

Whereas high-balance lines at the top have risen, median credit limits have remained constant at $5,000. That plateau indicates that access has tightened to near-prime and subprime borrowers.

Meanwhile, big bank credit card accounts channeled to consumers whose score falls below 660 have dropped from 23.3% during Q1 2022 to just 16.4% during Q1 2025.

Tighter underwriting has improved credit quality, but it may also be steering moderate-risk borrowers to pricier alternatives or discouraging them from borrowing at all. For volume-driven card issuers, that migration creates a pricing and sales dilemma. How do you compete without bringing too much risk?

Some are responding with hybrid models — offering cards with partial security deposits or graduated rewards to attract at-risk customers, while others are using alternative data, including income stability and cash flow analysis, to fill out thin credit files.

Dashboards that Look Beyond the Surface

The Fed’s new dashboard, which includes the broader data set, can also be sorted by credit tier, geography, and loan type. This helps lenders keep tabs on performance trends and spot budding issues before they surface in charge-off rates.

For example, issuers can observe how adjustable-rate mortgage performance varies by region or how credit utilization changes at a score band level. Such granularity improves underwriting tools, most notably in assessing thin-file applicants.

Thin-file applicants have minimal credit history, making them harder to assess for risk.

Thin-file borrowers are an ongoing challenge for subprime lenders. With minimal history, they are harder to assess — but the expanded Fed dataset provides more proxies for measuring behavior, like minimum payment frequency or time since last prior delinquency.

Used cautiously, they can strengthen models without taking too much risk.

Mortgage Indicators in the Data

Conventional adjustable-rate mortgages (ARMs) made up over 25% of big-bank originations in Q1 2025, up from just under 8% in early 2021.

ARM balances have climbed for 12 consecutive quarters, reaching a series high of $344.3 billion — a 34.5% increase since Q1 2022. That spike implies ongoing affordability challenges, especially for low-income and first-time buyers.

During high interest rate periods, consumers are increasingly opting for the lower teaser rates ARMs are offering — but at a price of future risk. Shocks to payments can occur when those introductory periods expire. If inflation or job insecurity returns, subprime-specialty mortgage lenders could face both opportunity and threat.

Information on property type, category delinquency, and geographic dispersion can enable lenders to prepare for those circumstances. As one illustration, high ARM concentration regions and rising delinquencies may prompt more cautious loan-to-value limits or modified servicing processes.

Looking Ahead

The expanded Philadelphia Fed dataset helps banks and fintechs stay one step ahead of the curve. By segmenting borrower behavior by score range and time horizon, they are better positioned to tailor offerings, price risk, and channel capital where it’s most productive.

But they are only good if there’s a good strategy. As fewer consumers qualify for traditional credit, thoughtful underwriting isn’t about just saying “no” to risk. It’s about learning the right way of saying “yes.”