In October 2025, Tim Spence had to tell Fifth Third's earnings call that his bank was eating a $170 to $200 million writedown on a single warehouse loan to a Texas subprime auto dealer called Tricolor Holdings. A week later, Jamie Dimon called JPMorgan's $170 million Tricolor charge "not our finest moment." Federal prosecutors had already started building the case alleging that Tricolor's executives double-pledged the same auto loan collateral to multiple lenders — roughly $800 million worth — and manipulated loan tapes to make delinquent paper look current.
Then, on May 8, the Federal Reserve published a FEDS Note dissecting the Buy Here Pay Here lending model that made the Tricolor blowup possible. It is, in effect, a postmortem of a business that credit union auto loan portfolios have spent years walking away from.
The contrast with what credit unions just reported is the actual story.
The Numbers Wall Street Didn't See Coming
NCUA's Q4 2025 Quarterly Data Summary shows the federally insured credit union system holding $480.1 billion in auto loans — by far the largest single asset category outside first mortgages. The auto loan delinquency rate at the end of 2025: 96 basis points. NCUA's exact phrasing was "essentially unchanged from a year earlier."
The system-wide net charge-off ratio came in at 78 bps — actually two basis points lower than Q4 2024.
This is not what the trade press was bracing for. Fitch reported subprime auto delinquency hit 6.65% at 60+ days past due in October 2025 — the worst reading since the index started in 1994. Industry analysts had been telegraphing 2026 as the year credit unions paid for their pandemic-era growth spree.
It didn't happen in 2025. The question is whether it gets pushed into 2026.
How Credit Union Auto Loans Held the Line
The auto loan portfolio shrank by $1.3 billion last year. New auto fell $3.8 billion (-2.3%). Used auto held up — up $2.6 billion. That is not the shape of a credit union system riding a trend down. It's the shape of one that started running off the worst 2022–23 vintages a year ago and is taking the runoff hit on net interest income to do it.
There's a reason most credit unions never had any direct exposure to Tricolor or its peers: the Buy Here Pay Here model is structurally adversarial to a member-owned lender. BHPH dealers underwrite to the title and the tow truck. Their borrowers are, per the Fed note, more than 50% deep subprime — credit scores under 580. The lender's job is to get the car back faster than the borrower's situation deteriorates. Nothing about that loop fits how a credit union examiner thinks about consumer auto.
What banks were actually buying when they extended warehouse lines to Tricolor was the structural protection: 60 to 80 percent advance rates against the underlying receivables, guarantors on 81% of facilities, special purpose vehicles to legally isolate the collateral. The Fed's own analysis is blunt: those protections "proved vulnerable to misrepresentation." Over-collateralization doesn't help when the collateral isn't real.
The Vintage Question Nobody Answered Cleanly
This is not a victory lap. The 96-basis-point auto delinquency number is steady, but it is steady at an elevated level that took two years of charge-offs to plateau at. The 2022–23 vintage problem hasn't been solved — it's been managed.
Loan modifications across federally insured credit unions hit $10.9 billion in December — the highest level NCUA has reported since 2016. That number is doing real work to keep the delinquency line flat. Historically, when banks tighten and absorb credit losses first, workout activity at member-owned institutions tends to lag a quarter or two behind, then convert to charge-offs as modifications run their course. The Federal Reserve's own credit card delinquency model uses a four-quarter lag on bank tightening signals for exactly this reason.
And the runoff is its own risk. A $480 billion portfolio that's shrinking 0.3% a year is not generating the spread it used to. Net interest margins are recovering — NCUA reported 3.39% for 2025, up from 3.12% — but a lot of that comes from deposit repricing, not asset yields. Even if nominal charge-offs decline in 2026, a shrinking denominator keeps the ratio elevated. TransUnion's December forecast has industry auto delinquency leveling at 1.54% by end of 2026 — meaning whatever stabilization happens, it happens at a high plateau.
What the Fed Note Is Really For
The May 8 FEDS Note ends with a section on what bank examiners should be looking for in warehouse facilities to non-prime auto: more granular loan tape verification, independent collateral audits, tighter covenants on advance rates. The implicit message — that the existing playbook missed it — is doing the heavy lifting.
For credit unions, the document reads less like a regulatory warning and more like a marker for where the cycle is. Banks took the visible loss in one quarter; the credit union system is absorbing a slower version through modifications and runoff. The institutions that didn't touch warehouse lines to BHPH dealers, that kept indirect dealer relationships tightly bounded, that took the runoff hit instead of underwriting through the cycle — they are the ones whose Q4 2025 reports look like NCUA's did.
The harder question for boards looking at 2026 isn't whether to chase auto growth back. It's how much of what looks like stability today is actually deferred — and how much of it converts to credit union auto loans charge-offs in the next two quarters.