Consumer & auto ABS & subprime — the household-credit tail, and the collateral-fraud failure mode

Built 2026-06-14. Structured data + edges: macro-consumer-abs-subprime.json. Companion to macro-uncovered-risk-pools and macro-cre-privatecredit.

Household debt hit ~$18.8T (Q1 2026) with a benign 4.8% aggregate delinquency that masks a stressed subprime tail: subprime auto 60+ delinquency at a 32-year high (6.74%), repossessions at 2009 levels, student-loan delinquency re-priced from ~0 to 7.7%+, and ~$400B of invisible BNPL "phantom debt." Tricolor and First Brands (both 2025) showed the failure mode is collateral fraud in the funding chain, not just borrower default.

1. Tricolor & First Brands — collateral double-pledged across financiers

• Tricolor: subprime auto lender (Chapter 7 ~Sept 2025) amid "systemic levels of fraud." DOJ (Dec 2025) charged execs with a "continuing financial crimes enterprise," allegedly double-pledging collateral (the trustee's review found ~$548M double-pledged across ~31,000 loans; the broader alleged fraud is larger) with duplicate VINs and falsified loan tapes. JPMorgan took a ~$170M writedown (Q3 2025); Fifth Third flagged ~$200M; Barclays exposed.
• First Brands: auto-parts supplier (Chapter 11, Sept 28, 2025): ~$5B sales, >$9B liabilities, $12M cash. ~$2.5B of invoices unmatched to real sales (fabricated/double-pledged receivables); DOJ-SDNY charges; ex-CFO pleaded guilty (Mar 2026).
• What they revealed: the ABS and factoring chains relied on trust in self-reported loan tapes/invoices; the same collateral pledged across multiple warehouse lines/factors went undetected — a verification failure at the funding layer, the same trust-the-self-report defect this project tracks in marks.

2. Auto stress — the canary

• Total auto debt ~$1.69T (Q1 2026). Subprime auto 60+ delinquency 6.74% (Dec 2025, Fitch) — a 32-year high. ~1.73M repossessions in 2024 (Cox) — highest since 2009; 2025 projected >3M (weak projection); 2024 default rate 3.13% (highest since 2011).

3. Cards, BNPL, student loans — the K-shaped consumer

• Cards ~$1.25T; flow into serious delinquency ~7.04% (Q1 2025), early-transition ~8.6% (Q1 2026) — elevated but plateauing; stress concentrated in subprime.
• BNPL: US GMV ~$122B (2025); ~$400B of balances largely invisible to bureaus (phantom debt); 47% of users reported a late payment (Mar 2026, up from 41%); ~8% used BNPL for groceries. Affirm 30+ delinquency ~2.8%; Klarna trades below its Sept-2025 IPO price with rising losses.
• Student loans ~$1.66T: the on-ramp ended Sept 2024, bureau reporting resumed Feb 2025 → 90+ delinquency jumped to ~7.74% (Q1 2025) from <1%; >2.2M borrowers' scores fell >100 points; the national average fell ~717→715, the sharpest drop since the Great Recession.

4. The ABS chain

• Total US ABS ~$358B YTD 2025 (+16.8%, auto-driven, record); auto ABS ~$127B. Subprime auto + unsecured consumer credit is originated, warehoused by banks (JPMorgan, Fifth Third, Barclays), then securitized off balance sheet into ABS bought by asset managers/insurers — banks retain warehouse-line tail risk (the Tricolor channel).

Synthesis

Household debt is at a record $18.8T with a placid 4.8% aggregate delinquency — but the tail is at multi-decade highs and ~$400B of leverage is invisible. Tricolor/First Brands show the acute failure mode is collateral fraud in the funding chain — the same "trust the self-reported number" defect that runs through the bank-HTM, private-credit-NAV, and pension-mark threads. The off-balance-sheet ABS chain disperses the loss while hiding collateral quality.

What is NOT asserted

• No claim of system-wide household-credit insolvency — aggregate delinquency is benign; the point is a stressed, partly-invisible tail + a collateral-fraud failure mode.
• BNPL "~$400B phantom debt" and "47% late" rest on survey/secondary sources (contested); ">3M 2025 repossessions" is a projection (weak).
• Tricolor's double-pledge (~$548M per the trustee) is distinct from JPMorgan's ~$170M writedown; Fifth Third's ~$200M is potential, not finalized.
• Overlay edges are excluded from the proofs.

Sources: NY Fed — Household Debt & Credit Q1 2026; CNBC — Tricolor charges; DOJ-SDNY — First Brands; Auto Finance News — subprime 60+ at 32-yr high; Carscoops — repossessions at 2009 levels; Affirm fiscal Q3 filing; CFPB — BNPL market report; NY Fed — Q1 2025 student-loan spike; Asset Securitization Report — unsecured consumer ABS records.

The 2026 subprime measurement break (read trends with caution)

On 10 Aug 2026 the NY Fed's Consumer Credit Panel switched credit-score methodology - from Equifax Risk Score 3.0 to VantageScore 4.0 - pulling ~33 million previously-unscored people into the scored population (the "subprime" metric = share scoring below 660). Because the denominator jumps, the subprime share drops sharply for reasons unrelated to credit quality: per the St. Louis Fed / FRED (10 Sep 2026), New York County's subprime share fell from 30.27% (old measure, Q4 2025) to 17.72% (new measure, same quarter) - a 12.55pp artifact, not improvement.

The Fed itself cautions that "differences may reflect the change in scoring model rather than changes in borrower credit quality." Honesty guard for this block: the real stress signals here (subprime auto 60+ delinquency 6.74% Dec 2025 = 32-yr high; repos at 2009 levels; Tricolor / First Brands failures) are pre-break. Do not let a post-Aug-2026 drop in the subprime share be read as the consumer-credit tail healing - that drop is definitional, and the new series needs its own history before trend claims are safe.

Source: St. Louis Fed FRED Blog - A new calculation for subprime credit risk (2026-09-10).

← Research index · structured data: macro-consumer-abs-subprime.json · macro-consumer-abs-subprime.md