What 30 Million Applications Reveal about Mortgage Denial Thresholds

This is the second in a series of four posts that explore the challenges potential homebuyers face when seeking a mortgage loan.

When Congress passed the Dodd-Frank Act in 2010, it established a debt-to-income ratio of 43% as the key threshold for “qualified mortgage” status. Loans meeting this standard gave lenders a legal safe harbor from ability-to-repay lawsuits, making the 43% mark one of the most discussed numbers in housing finance. In 2021, the Consumer Financial Protection Bureau (CFPB) replaced this ratio-based definition with a price-based standard, partly because of concerns that the 43% cap was constraining credit to many households and thus preventing them from purchasing a house.

While 43% is no longer a hard limit, many lenders treat this debt-to-income ratio as a rule of thumb. Yet does this 43% threshold constrain mortgage lending? Using Home Mortgage Disclosure Act (HMDA) data on more than 30 million home purchase mortgage applications from 2018 to 2024, we find that it does not. The real underwriting cliff—the point where denial rates spike dramatically—has been sitting at 50%.

Flat Through 43%, a Wall at 50%

The figure below plots denial rates by debt-to-income category for 2024. The pattern is striking: Denial rates were essentially flat across the 20% to 50% ratio range, hovering between 8% and 10%. There was no visible jump at 43%. But once the ratio crossed 50%, denial rates rose sharply and surpassed 80% for applicants above 60%.

A note on the data: The public HMDA files report debt-to-income ratios as a mix of exact values and categorical bins (for example, “36%–<40%,” “40%–<43%,” “43%–<45%”). The bins are designed to bracket the 43% threshold for qualified mortgages, which means we can observe applicants on either side of it with reasonable precision. For our analysis, we converted binned values to midpoints. This coarseness limits fine-grained precision, but the patterns are so stark that the binning does not affect the conclusions: There is no detectable change at 43%, while the jump at 50% is massive.

This pattern reveals a large gap between the often-cited benchmark and actual lending practice. The 43% threshold, long considered a bright line in mortgage underwriting, appears to have little practical bite. Instead, lenders treat 50% as the functional boundary, and the consequences of crossing it are severe.

Beyond the Figure

While the figure is suggestive, we want to apply two formal statistical tests to confirm what the data show visually.

First, if the 43% threshold mattered, we would expect to see borrowers and lenders clustering applications just below it to stay on the safe side. We find no such clustering. If anything, there are slightly fewer applications just below 43% than just above—the opposite of what we would see if borrowers or lenders were treating it as a meaningful line.

Second, we directly estimate the jump in denial rates at each threshold. Rather than look at the raw bin values, we used the regression discontinuity (RD) method to isolate the jump in denial rates at each threshold from the underlying trend, comparing what happens just below and just above each cutoff. At the 43% debt-to-income ratio, the jump is less than half a percentage point—trivial against a baseline denial rate of about 15% for home-purchase mortgages in 2024. At 50%, the jump is 15 to 17 percentage points, according to the RD results. The 43% mark that has drawn enormous regulatory attention barely registers in actual lending decisions. The 50% mark, which has received no comparable scrutiny, is where applications hit a wall.

Why a Threshold at 50% and Not 43%?

The likely explanation is institutional. While 43% was the statutory qualified-mortgage benchmark, the government-sponsored enterprises (GSEs), like Fannie Mae and Freddie Mac, were granted a temporary exemption; this “GSE patch,” which ended in 2022, allowed them to purchase loans with debt-to-income ratios above 43%, provided those loans met other qualified-mortgage requirements. Because the vast majority of conforming mortgage originations are sold to or guaranteed by the GSEs, this patch effectively neutralized the 43% threshold for most of the market.

Meanwhile, the “ability-to-repay” standard under Dodd-Frank permits lenders to approve loans with debt-to-income ratios above 43% when compensating factors are present, such as strong credit scores, significant reserves, or low loan-to-value ratios. In practice, lenders appear willing to extend this flexibility up to about 50%, beyond which the risk profile becomes too unfavorable regardless of compensating factors.

Why This Matters Now

This finding has direct policy relevance. In this series’ first post, we documented how rising interest rates during 2022–23 pushed borrowers’ debt-to-income ratios higher, driving up denial rates. Because the cliff sits at 50% rather than 43%, many borrowers who would have been flagged by the old regulatory threshold were actually passing through it without difficulty, only to hit the real wall further up the distribution of debt-to-income ratios.

The implication is that policymakers and researchers who are focused on expanding credit access should direct their attention to the 50% to 60% debt-to income range, where underwriting constraints bind. The CFPB’s 2021 reform, which replaced the 43% ratio-based qualified-mortgage definition with a price-based standard, addressed a threshold that was already largely irrelevant in practice. The de facto 50% cutoff, where denial rates jump by 15 to 17 percentage points, remains without comparable regulatory attention.

For borrowers, the practical lesson is clear: A debt-to-income ratio of 45% is treated by lenders much like a ratio of 35%. But crossing 50% changes the game entirely.

Notes

1. This post is based on our St. Louis Fed working paper “The Determinants of Mortgage Denial Using Public Data,” April 29, 2026.
2. Our series also includes an upcoming post about the specific challenges facing first-time homebuyers and another on why cheaper houses are tougher to finance.