Single-family loan performance: the year to March 2026

The picture in one paragraph

Fannie Mae's single-family book remains very clean by historical standards, but credit metrics turned during the winter of 2025-26. Serious delinquency (90+ days or REO) on the ~15.5 million-loan book was 0.55% in March 2026, up 2.9bp year over year and about 5bp off the Jul '25 trough of 0.49%. On a balance-weighted basis the move is bigger: seriously delinquent UPB rose from 0.57% to 0.62%. Early-stage delinquency is the louder signal: 60-89 day delinquency is up ~15% since mid-2025 and 30-59 day rolls spiked in November-December 2025. Realized losses remain small — roughly 11,474 loans exited to a credit event in the twelve months to March 2026 (~0.07% of the book) — but foreclosure/REO liquidations roughly quintupled over the winter. The deterioration is concentrated: the 2022-23 vintages, Gulf Coast and Illinois metros, purchase and cash-out loans, and the federal-workforce corridor (DC/MD) carry most of it. Unemployment rose only ~0.1-0.2pp nationally over the period, so this is mostly an affordability-and-vintage story layered on localized shocks, not a broad labor-market downturn — yet.

National trend: a quiet 2025, a colder winter

Delinquency drifted lower through most of 2025 — SDQ bottomed at 0.49% in Jul '25 — then turned up around November-December 2025, first in early-stage buckets (30-59 day rolls hit 1.11% in November) and then in the serious-delinquency stock, which reached 0.56% in February 2026 before easing slightly in March. The November-December inflection lines up with the late-2025 federal government shutdown and the softest labor prints of the cycle (national unemployment touched 4.5% in November). Seasonals argue for some of it — mortgage delinquency typically peaks in winter — but the year-over-year step-up is real and is now feeding the liquidation pipeline.

Where defaults are realized

Exits to credit events stayed small but the mix shifted hard toward terminal resolutions. Third-party sales ran a steady ~300/month all year and then disappeared from the reporting in Q1 2026 (likely reclassified rather than stopped). Short sales are negligible (~35/month — equity cushions are intact for most borrowers). The two forces to watch: REO / deed-in-lieu dispositions climbed from ~140/month in early 2025 to 600-740/month in Q1 2026 — the pipeline conversion of the winter delinquency bulge — and lump-sum non-performing / reperforming note sales (1,315 loans in April 2025, ~590/month mid-2025, 1,684 in October 2025), which is Fannie clearing seasoned legacy defaults rather than new distress. Defaulted loans average a ~715 FICO and ~38 DTI at origination — clearly weaker than the book — with original LTVs of 84% on REO dispositions and 91% on short sales.

Geography: distress is regional, not national

The highest serious-delinquency rates are still the usual suspects — Louisiana, Mississippi, Puerto Rico, Florida's Gulf/Atlantic corridor — a structural pattern tied to insurance costs, storm exposure, and weaker income cushions rather than anything new. What changed over the past year is the gradient: the biggest increases are in the Washington corridor (DC +0.21pp, Maryland +0.14pp — plausibly the federal workforce shock), Nevada (+0.11pp, tourism/hospitality), and Texas/Arizona/Georgia where rapid home-price gains left thin-equity recent vintages. Florida is the contrarian bright spot: SDQ fell 0.11pp YoY despite unemployment rising ~0.8pp, as the state laps hurricane-related distress and benefit from strong home equity. Notably, the state-by-state delinquency level correlates with unemployment (r≈0.4) but the year-over-year change does not (r≈−0.1) — the deterioration is concentrated where jobs are fine.

The metro list reads like three different problems: storm/insurance metros (New Orleans, Baton Rouge, Lake Charles, Houma, Punta Gorda, Cape Coral, Ocala, Homosassa Springs, Panama City), thin-margin Rust Belt (Rockford, Peoria — Peoria's +0.26pp is the biggest jump on the board), and border/agricultural (McAllen, Yuba City). Houston (0.82%, −0.06pp) is quietly improving while Orlando and Tampa actually improved YoY — inside Florida the stress is concentrated in the smaller coastal metros.

Vintage and loan-type: the 2022-23 pocket

The clearest fault line is vintage. Loans originated in 2022 carry a 1.01% SDQ — double the book average — and 2023 is close behind at 0.89%. These cohorts were written at peak home prices into the fastest rate shock in decades; many borrowers have thin or negative equity cushions, so a cash-flow shock converts to default rather than a workout or sale. The massive 2020-21 refi vintages (0.30% and 0.46%) remain bulletproof — sub-3% rates and large equity. By purpose, purchase loans (0.71%) and cash-out refis (0.62%) are where distress lives; rate/term refis (0.26%) are clean. Occupancy is overwhelmingly primary residence among defaults (93%), so this is borrower cash-flow stress, not investor walk-aways.

The labor market is not the driver — so far

LAUS unemployment (via FRED) tells a gentle story: national unemployment rose from ~4.0% to 4.5% in November 2025, then eased back to ~4.1-4.2% by mid-2026. State-level changes are small and, where they moved, often in the "wrong" direction for the delinquency map — New Jersey and Ohio saw unemployment fall ~1pp while their SDQ rose ~6bp. The exceptions that do show a labor fingerprint: DC (6.2% unemployment in March 2026, highest on the board) and Maryland, consistent with federal layoffs and the shutdown; Nevada hospitality softness; and New Mexico/Arizona border-economy drift. The broader implication: the past year's credit weakening reflects cumulative affordability strain and the bad-vintage effect more than job loss — which is why it is showing up in early-stage rolls and geographically random pockets rather than as a wave.

Inflation and energy: a late-breaking squeeze

The macro pressure point shifted in early 2026. Headline CPI re-accelerated from ~2.4% YoY in January to 4.2% in May 2026, driven by an energy shock: energy CPI ran +23% YoY in May and ~+20% in September, with retail gasoline pushing from ~$3.90 toward $4.35-4.50/gal by October. Meanwhile 30-year mortgage rates climbed from ~6.4% in May to ~7.4% in mid-October 2026. The timing matters: the delinquency turn (Nov-Dec 2025) preceded the energy spike, so energy is an amplifier of existing vulnerability — tight budgets in commuter-heavy Texas, inland California and Southeast metros — rather than the original cause. The mortgage-rate backup simultaneously closes the refi escape hatch and freezes financially stressed borrowers into loans they cannot restructure cheaply.

Outlook

• Upside risk: the 2022-23 vintage pocket continues to underperform; if home prices soften, those loans (already ~84-91% LTV at default) produce real loss severities rather than note-sale transfers. A further leg up in mortgage rates above ~7.5% would freeze the book longer and keep early rolls elevated.
• Labor: if unemployment re-tests 4.5%+ (the November level), the correlation that has so far been absent would assert itself quickly — watch DC/MD/VA (federal), NV (hospitality), and MI/OH/IL (manufacturing) where early rolls are already rising.
• Energy/inflation: gasoline above ~$4.30 sustained for another quarter historically shows up in entry-level and commuter-metro delinquency with a ~2-quarter lag; TX, GA, AZ and inland-CA metros are the transmission zones.
• Downside relief: Florida's improving trend shows how fast storm-related delinquency can cure when equity cushions exist; a Gulf quiet season plus stable oil would pull the worst-MSA list down by year-end. NPL note sales will keep absorbing the oldest distress regardless.
• Signals to watch next quarter: the 30-59 roll rate (still ~1.0% vs 0.82% in March 2025), whether the February SDQ print (0.563%) proves to be the peak, the REO disposition run-rate, and whether third-party sale reporting resumes — its disappearance in Q1-2026 looks like a reporting change worth confirming before reading the pipeline.