themortgagellm^™

Written by The Mortgage LLM Team—a group of industry analysts leveraging our proprietary mortgage-domain language models to synthesize and decode housing data.

Across the full universe of FHA-insured single-family loans — 6.94 million active mortgages spanning all 50 states, the District of Columbia, and three US territories — the 60+ day delinquency rate climbed to 6.97% at March 2026, up 197 basis points from a year earlier and 295 bps from April 2024. The national aggregate, though, obscures more than it reveals. Across the 50 states, the Q1 2026 60+ rate ranges from 4.73% in Alaska and Montana to 9.72% in Louisiana — a spread of nearly five percentage points. Adjusting each state’s observed rate against the vintage-weighted historical baseline for an equivalent FHA book, ten states are running more than 100 basis points above their cycle-expected rate. The geographic distribution of where FHA credit stress is actually running above the historical norm, rather than tracking the national tide, is the more informative cut.

A note on the headline numbers

Two reporting realities frame the absolute 60+ figure cited above. First, in October 2025 the FHA implemented a mandatory Trial Payment Plan (TPP) requirement for home-retention loss mitigation: borrowers entering a workout must now successfully complete three consecutive trial payments before the loan is reclassified as current. During that three-month trial window the loan remains in a delinquent status, so the rule mechanically inflated headline FHA delinquency statistics across late 2025 and early 2026, independent of underlying credit stress. The Mortgage Bankers Association’s Q1 2026 National Delinquency Survey reported an overall seasonally-adjusted FHA delinquency rate of 11.88%, with a meaningful share of that level attributable to TPP-driven paperwork classification rather than fresh financial deterioration. Second, the 6.97% figure cited here is a stricter cut — 60+ DPD only, on a loan-level basis — than the MBA’s all-DQ survey universe, so the two numbers describe different things rather than contradicting each other.

What the TPP rule does not distort is the geographic decomposition. The rule applies uniformly across all FHA loans nationally, so the differential between high-stress and low-stress states reflects real credit dynamics, not a reporting artifact. The remainder of this piece focuses on that differential.

Why the geographic decomposition matters

A 295 bps national move in 24 months can be driven by broad-based deterioration — every market moving at similar pace — or by sharp concentration in particular regional credit environments. In the trade press, FHA delinquency is reported as a single national number, but that aggregate averages across a portfolio with enormous geographic variation in borrower demographics, labor markets, vintage mix, and home-price trajectories. A national 60+ rate of 7% can describe an industry where every state sits between 5% and 9%, or one where most states sit near 6% and a handful sit above 10%. Those are very different operating environments for servicers staffing loss-mit operations, MBS investors modeling Ginnie pool buyout risk and the prepayment-speed exposure that comes with it, and policy stakeholders trying to read FHA performance as a leading indicator. The state decomposition resolves the ambiguity.

The Q1 2026 state distribution

Half of all FHA-insured loans sit in jurisdictions with 60+ rates between 5.92% and 7.37% (the interquartile range), and the median state runs at 6.33% — slightly below the loan-weighted national 6.97%, indicating the larger-book states are running marginally hotter than the average state. Within the 50-state set, the widest spread is five percentage points, between Alaska and Montana (each at 4.73%) and Louisiana (9.72%). The District of Columbia at 11.28% sits further above but on a small FHA book of roughly 4,800 active loans — the lowest-volume jurisdiction in the dataset, making it highly sensitive to small shifts in loan count and subject to high volatility; it appears on the chart for completeness.

Above the 75th percentile (60+ above 7.37%): Louisiana (9.72%), Maryland (9.08%), Illinois (8.81%), Georgia (8.39%), New Jersey (8.02%), Delaware (7.87%), Massachusetts (7.70%), Rhode Island (7.70%), Mississippi (7.61%), Connecticut (7.56%), South Carolina (7.54%), Iowa (7.39%), Alabama (7.37%), and Michigan (7.35%); the District of Columbia also sits above this threshold but with the small-book caveat noted above. The cluster pattern is unmistakable: most of the top tier sit either in the Northeast/Mid-Atlantic corridor (Maryland, New Jersey, Delaware, Massachusetts, Rhode Island, Connecticut) or in the Deep South (Louisiana, Georgia, Mississippi, Alabama, South Carolina).

Below the 25th percentile (60+ below 5.92%): Puerto Rico (1.97%), Alaska (4.73%), Montana (4.73%), the US Virgin Islands (4.95%), Hawaii (5.39%), Wyoming (5.44%), New Mexico (5.61%), Idaho (5.70%), and others clustered in the Mountain West, Plains, and Pacific outliers. The geographic split between the top and bottom quartiles is structural enough to be visible in a regional map.

The baseline gap is the more informative cut

Cross-sectional comparisons can mislead on their own, because a state’s observed rate depends in part on the age and vintage composition of its FHA book. To isolate that effect, we compare each state’s observed Q1 2026 60+ rate to its vintage-weighted historical baseline: for each origination cohort in the state’s current book, we look up what the historical FHA 60+ rate was for that vintage at the same loan age the cohort currently has, then weight by the state’s vintage mix today. The gap between observed and baseline is the credit stress that’s not explained by where the book sits in its lifecycle.

By that measure, Louisiana is running 308 basis points above its vintage-weighted baseline on 112,000 active loans, the largest gap among states with statistically meaningful book sizes. Maryland sits 264 bps above baseline on 169,000 loans; Illinois 233 bps on 227,000; Georgia 199 bps on 297,000; New Jersey 148 bps on 148,000; Delaware 127 bps; Massachusetts 120 bps; Rhode Island 117 bps; Minnesota 103 bps. Twenty-three of the 54 jurisdictions in the dataset are running above their vintage-weighted baseline. (The District of Columbia’s gap of 625 bps is nominally the largest in the dataset, but it is computed on the 72% of its small FHA book that the baseline join could match to a vintage-age cohort — a measurement subject to materially more noise than the larger-book states.)

The Mortgage Bankers Association’s Q1 2026 National Delinquency Survey corroborates the geographic signal from an independent vantage point. The MBA flagged Mississippi (+131 bps YoY), Louisiana (+88 bps), Maryland (+84 bps), and Georgia (+78 bps) as the four states with the largest annual increases in overall FHA delinquency — the same Deep South and Mid-Atlantic concentration that emerges from the loan-level baseline-gap analysis. That two independent measurements — a tighter 60+ DPD loan-level cut against vintage-weighted baseline, and a broader all-DQ survey of YoY change — agree on which states are running hot is a meaningful cross-check on the regional finding.

The reverse pattern is just as informative. Puerto Rico runs 557 bps below baseline — its FHA book is performing markedly better than the historical norm for an equivalent vintage profile. Alaska sits 248 bps below; Montana 205 bps; New Mexico 168 bps; West Virginia 147 bps; Wyoming 140 bps; Hawaii 121 bps; Idaho 120 bps; Nevada 113 bps. The under-baseline jurisdictions cluster in the Mountain West, Plains, and Pacific outliers — the same regions that sit below the 25th percentile on absolute level.

Vintage adjustment alone — re-weighting each state’s portfolio to the national vintage distribution without anchoring to historical norms — moves the rankings only marginally. The largest single-state shift is Illinois at +24 bps after vintage adjustment; most states differ from their observed rate by less than 20 bps. That tells us the geographic pattern is not a vintage-mix artifact. It’s state-specific credit deterioration that survives the most obvious normalization.

The diagnostic role of the biggest FHA books — Texas (753,000 active loans), California (~750,000), and Florida — is also informative. All three sit within ±50 bps of their vintage-weighted baseline. The credit cycle is broad enough that those large books are participating in the national 295 bps move, but the acute stress signal is in the specific states running materially above norm.

What it implies

The above-baseline states — Louisiana, Maryland, Illinois, Georgia in the top tier of meaningful-book size, and the broader Northeast/Mid-Atlantic and Deep South clusters behind them — are not exhibiting broad cohort aging. They’re showing state-specific credit stress that may warrant additional loss-mit staffing review for servicers concentrated in those geographies.

For Ginnie Mae MBS investors, the same regional hot spots translate directly into localized buyout risk. When an FHA loan in a Ginnie pool reaches deep delinquency (typically 90+ DPD, sometimes triggered earlier by a loss-mit decision), the servicer is obligated to repurchase the loan out of the pool at par to execute foreclosure or modification. To the MBS investor that buyout looks identical to an early prepayment — a sudden return of principal at par. Elevated above-baseline stress in Louisiana, Mississippi, Georgia, and the broader Deep South therefore translates into elevated conditional prepayment rates (CPR) for pools heavily weighted with collateral from those states, and the speed differential has to be priced separately from any national CPR assumption. Pools with concentrated above-baseline exposure should be expected to run materially faster than national-average speed assumptions imply.

The mirror pattern in Puerto Rico, the Mountain West, and the Plains points to structurally lower default propensity in those FHA books — whether driven by lower house-price volatility, labor-market resilience, or borrower selection differences worth a separate analysis. The next monthly disclosure, covering April 2026 origination data, will tell us whether the above-baseline states are continuing to accelerate or beginning to stabilize. The national 60+ rate is unlikely to start falling first — and with the TPP rule still working through the data, headline FHA delinquency may continue to drift higher even after underlying credit dynamics have begun to stabilize.

---

Methodology. FHA performance data sourced from Ginnie Mae’s Single-Family Monthly Portfolio Disclosure (llmon) covering April 2024 through March 2026 monthly snapshots — the full available window. Analysis restricted to FHA-insured single-family loans. The 60+ delinquency rate is the count of active loans with current delinquency status of 60 days or worse (including loans in foreclosure, bankruptcy, REO, and the loss-mitigation trial-payment universe described in the headline-numbers note above) divided by total active loans. Vintage-mix-adjusted DQ rate re-weights each state’s portfolio to the national vintage distribution at the same loan age. Historical baseline rate is computed from point-in-time 60+ DPD rates per FHA vintage × loan-age cell across the full disclosure history, weighted to each state’s current vintage mix at Q1 2026. Coverage rate (share of each state’s active loans where the baseline join successfully returned a rate) averages 78–87% across states; coverage falls below 100% largely because of missing or reclassified fields in the legacy Ginnie Mae disclosure layout and early-vintage cohorts whose initial months of loan age sit before the disclosure series began. The lowest-coverage jurisdictions — the US Virgin Islands 67%, New York 70%, the District of Columbia 72%, Puerto Rico 74% — should have their gap measurements read with proportionally more caution. National all-DQ comparison sourced from the Mortgage Bankers Association’s Q1 2026 National Delinquency Survey. Informational, not advice.