The FHA Gap Measures Concentration, Not Timing

The question underneath the FHA split is a good one, and it’s worth stating precisely before looking at whether the data answers it.

FHA-insured mortgages are the entry-level channel. A borrower can buy with 3.5% down. Thin equity, thin reserves, a payment that works while everything else does. If household distress shows up first where the cash cushion is smallest, it should show up here first.

Two claims are hiding in that sentence. One is about level. The other is about order. They need separate evidence, and only the first has strong evidence behind it.

The level claim holds

The FHA delinquency rate was 13.70% in June 2026, according to HUD’s monthly FHA report. The rate for single-family mortgages booked on commercial-bank balance sheets was 1.86% in the second quarter, a Board of Governors of the Federal Reserve System series retrieved via FRED as DRSFRMACBS.

That puts FHA at about 7.4 times the bank rate.

This isn’t one bad quarter. HUD’s monthly series, as we hold it, starts in July 2017, and FHA’s rate sits above the bank rate in all 36 quarters we can compare since then. Even at its closest, in Q1 2018, FHA ran about three times the bank rate: 10.29% against 3.48%.

The two have also moved apart. In Q2 2018, FHA was at 10.62% and the bank rate was at 3.22%. Eight years later, FHA was higher and the bank rate was lower. Whatever else is true, the entry-level channel carries a different rate of missed payments than the bank-held book.

The order claim can’t be settled on this record

Our record of HUD’s FHA rate runs 110 months with no gaps, from July 2017 to August 2026.

That record doesn’t reach the last mortgage crisis. The bank rate peaked at 11.48% in Q1 2010, more than seven years before our HUD series starts. That’s the stretch where the order would matter most, and we hold no HUD figure for it.

What the record does hold is two upturns. To take out HUD’s seasonal swings, this table compares each quarter with the same quarter a year earlier. It shows when each rate started rising against the year before and kept rising.

That’s the strongest case for the early-warning version: FHA went up first both times, by three quarters and then by four. Neither case is clean. In 2020, banking agencies said loans with COVID-19 payment deferrals weren’t expected to be reported past due because of the deferral, and that may have held the bank rate down. We haven’t tested how much. In 2024, the bank rate’s first rise was just 0.01 of a point, and the Fed re-estimates that history at almost every release, so a revision could shift the quarter.

The way down doesn’t help the thesis either. Both rates started falling against the year before in the same quarter, Q2 2021.

American Default’s leading-indicator research now tests the FHA series too. That scan covers 88 series and compares FHA with the bank rate in both directions. Neither direction passes any of its three crisis tests: the 2001 recession, the 2008 crisis and the pandemic. HUD’s record as we hold it doesn’t reach the first two.

What the gap does measure

Take the widest gap on HUD’s record. In Q1 2020, FHA was at 11.17% and the bank rate was at 2.35%. One quarter later, FHA was at 17.41% and the bank rate was at 2.54%.

So the gap shows how much of the trouble sits in the FHA channel and not in the bank-held book. It’s a concentration measure. A reader who treats a widening gap as a countdown gets no support here: the widest gap on record arrived when the rest of the mortgage book looked quiet.

Almost all of the gap’s movement comes from the FHA side. Across these 36 quarters, FHA ranged from 9.88% to 17.59%, while the bank rate stayed between 1.70% and 3.64%.

One plausible mechanism, offered as a hypothesis and not a finding: FHA’s population is held roughly constant by program design, with low down payments, thin reserves and a credit profile the conventional channel declined. Its rate runs high in good conditions and bad. In a broad mortgage crisis, the bank rate would climb toward it and the gap should narrow.

HUD’s record can’t test that, because it holds no such crisis. The bank rate’s 11.48% peak in Q1 2010 is the case we’d want, and our HUD series starts too late for it. The simpler alternative is that FHA’s rate just swings more than the bank rate in every kind of period. A bank rate that climbs a lot while the gap stays wide would prove the hypothesis wrong.

The comparison is not like for like

One caveat has to travel with every number above, because it limits what the level of the gap can mean.

The two series come from different institutions and cover different loans. HUD’s report counts FHA-insured single-family loans that are 30 or more days past due, including loans in foreclosure or bankruptcy. The comparison series is a Board of Governors measure of single-family mortgages on commercial banks’ balance sheets. It counts balances 30 or more days past due, plus loans the bank has stopped booking interest on. FHA-insured lending isn’t confined to that group.

HUD’s figures aren’t seasonally adjusted, and the Fed’s are. We pair the two by using HUD’s last month of each quarter. HUD’s first-quarter figures tend to run low and its fourth-quarter figures high, so the closeness of Q1 2018 is partly the season. That’s why this article measures change between like quarters.

The Mortgage Bankers Association’s quarterly survey publishes a separate FHA rate that’s seasonally adjusted and leaves out loans in foreclosure, so its figure won’t match HUD’s.

So part of the level difference comes from the definitions, and no amount of care in the arithmetic removes it. What survives is the movement: each series is consistent with itself over time, so the direction and size of changes in the gap stay useful even where the ratio isn’t apples to apples.

Where this leaves the thesis

The borrower split is real and it’s measurable. Households with the least cash cushion miss payments at a rate the bank-held mortgage book doesn’t approach, in every quarter we can compare, including quarters when the wider numbers looked calm. That’s worth publishing on its own.

The early-warning version isn’t supported. It needs a record of the quarters where the last crisis turned, and ours starts after them. What we have is two upturns, one tangled up with the pandemic and one resting on a bank move that small. The strongest honest statement is that the gap tells you how concentrated the distress in a quarter is, which is a real thing to know and a different thing from knowing when.

Two things would change that conclusion. One is earlier HUD figures showing FHA turning up well ahead of the bank rate before the last crisis. The other is a large rise in the bank rate that FHA clearly leads.

The thesis said the thin-buffer channel would show it first. The record says it shows it worse. Only the second has the quarters behind it.

Correction, November 4, 2026: An earlier version of this piece used Mortgage Bankers Association figures for the FHA rate and called that survey the only public source. HUD publishes a monthly FHA rate, so we rebuilt the article on HUD’s figures, which count delinquency differently, and every FHA figure changed. HUD’s record as we hold it starts in July 2017, so the 2007 and 2010 comparisons are gone. The earlier version also guessed that most of the gap’s movement came from the bank side. HUD’s record shows the opposite, and we removed that claim. The section on our research scan is updated because the scan now includes this series.