Auto loan delinquency is at a series record while used-car values have stabilized well below their 2021 peak — lenders remain exposed, but mainly through legacy 2021-2023 vintages, not a fresh price collapse

Summary

The two series move together but with a multi-year lag, and the relationship runs mainly through loan underwriting at origination, not through real-time collateral marking. Used-vehicle wholesale values (Manheim Index) spiked to an all-time high of 257.7 in December 2021, then fell roughly 24% to a trough of 196.1 by June 2024, and have since stabilized in the low-to-mid 200s (208.2 in August 2026). Auto loan delinquency kept climbing through this whole period, including after prices started falling: the 90+ day serious delinquency rate reached a series record 5.5% of balances in Q2 2026, above even the 2010 Great Recession peak of 5.3%, and the 60+ day rate hit its highest level since 2008 in Q3 2025. The link is that record-high prices in 2021-2023 pushed borrowers into larger loans and higher payments, and the subsequent price decline then left many of those same borrowers underwater (negative equity, now 29.6% of trade-ins) years later — not a case of today's price level directly triggering today's missed payments. On exposure: yes, lenders (especially independent/non-bank subprime specialists) are exposed, but the Philadelphia Fed's April 2026 analysis finds the exposure is concentrated in a backlog of loans that entered delinquency in 2022-2023 and are taking longer to resolve, rather than a fresh wave of newly distressed borrowers — a nuance the headline delinquency rate does not show on its own.

What this corpus (askamerica) does and does not carry

search_catalog was checked directly for both 'delinquency' and a used-car/used-vehicle price series. Neither exists in this warehouse: 'delinquency' returned zero matching tables or columns across the whole catalog (banking, fiscal, housing, sec schemas), and the only vehicle-price series loaded is econ.new_vehicles_cpi (BLS series CUUR0000SETA01/SS45011/SS45021) — new vehicles only. The used cars and trucks CPI series (CUUR0000SETA02) and any wholesale/auction used-vehicle index (Manheim, Black Book) are not ingested. This is a genuine sourcing gap, not a query failure — confirmed by inspecting the actual loaded series list (SELECT DISTINCT series FROM econ.new_vehicles_cpi returns only the three new-vehicle codes). Everything below therefore comes from primary external sources rather than an askamerica query, and no correlation, regression, or diff-in-differences was computed on any series in this session — every quantitative relationship cited below is a figure taken directly from the named study's own published analysis.

How closely delinquency has tracked used-car prices

The pattern is a lagged, indirect relationship: elevated prices at origination (2021-2023) — not the level of prices today — are the main channel connecting used-car values to delinquency. High prices meant larger loan balances and higher monthly payments; a 2024 Federal Reserve Board FEDS Notes study (Adams, Bord & Sannoh, cited in the Philadelphia Fed report) reports that higher monthly payments, driven primarily by elevated vehicle prices rather than interest rates, accounted for roughly 40% of the rise in the two-year delinquency rate for 2019-2022 vintages — a figure from that study's own analysis, not computed in this session. Falling prices after 2021 then left many of those same borrowers underwater, making it harder to sell or refinance out of a loan that had become unaffordable, compounding delinquency even as the price index itself declined.

An important qualifier, from the Philadelphia Fed (April 2026): the headline delinquency rate is a stock measure. Their decomposition of NY Fed Consumer Credit Panel/Equifax auto tradelines finds that since Q3 2022, the rise in delinquency is driven mainly by loans that remain delinquent quarter after quarter (a growing backlog not resolving through cure, charge-off, or repossession) and by 'redefaulters,' rather than by a rising flow of newly distressed borrowers. Lenders have expanded use of loan extensions (deferring payments to the loan's end), which the report links to troubled loans staying in the delinquent pool longer rather than resolving. So part of the 'record' delinquency level reflects slower workout of older problem loans, not only a fresh acceleration of new borrower distress.

Are lenders currently exposed to falling collateral values?

Yes, in three concrete ways, though the exposure is concentrated rather than system-wide:

• Negative equity is at or near record levels. 29.6% of new-vehicle trade-ins carried negative equity in Q2 2026 (up from 26.6% a year earlier), with the average underwater amount around $7,000-7,200 — near an all-time high. Q1 2026's 30.9% underwater share was the highest for any quarter since Q1 2021. This directly means that on repossession, many lenders recover a vehicle worth less than the outstanding loan balance.
• Exposure is concentrated in subprime and non-bank lenders. Subprime borrowers hold only 17% of active auto loan accounts but account for nearly two-thirds of all delinquent loans; the subprime 60+ day rate has hovered near 6% since mid-2024, the highest in over 20 years of data. Independent, non-bank 'other finance' lenders — who skew heavily toward used-vehicle and subprime lending — hold about 40% of subprime accounts and show the longest unresolved-delinquency timelines, making them the segment most exposed to weak recovery values on repossessed collateral.
• Loan structure amplifies the risk. Widespread use of 72-96 month loan terms and reportedly aggressive loan-to-value lending (some programs cited above 140% LTV for subprime borrowers, per industry reporting) mean many loans were structurally underwater from origination, before any price decline — falling collateral values simply widen an already-existing gap for these borrowers.

Offsetting this: used-vehicle prices have not collapsed further recently — they stabilized in 2024-2026 and were running slightly above year-ago levels as of late 2024 and again through mid-2026 (Cox Automotive reported a modest +0.4% year-over-year gain at end of 2024). So the acute risk today is less 'prices are currently falling out from under lenders' and more 'a large stock of 2021-2023-vintage loans, originated against since-depreciated collateral, is still working through the system slowly.' If used-car prices were to resume falling meaningfully from here, that would compound losses on this already-strained back book, particularly for subprime specialty and buy-here-pay-here lenders and the asset-backed securities investors holding pools of these loans — industry reporting (Fitch data, via CarEdge) places the subprime 60+ day delinquency index at its highest level in over 32 years as of January 2026.

Caveats and what would sharpen this further

• This corpus carries neither an auto-loan delinquency series nor a used-vehicle price index, so the relationship above is described from primary published sources (NY Fed, Philadelphia Fed, Cox Automotive, Edmunds) rather than computed with a direct correlation or regression inside this connector. No diff-in-differences, correlation, or regression tool was run in this session; a proper lagged correlation between Manheim index changes and delinquency rate changes, run on the NY Fed CCP microdata plus Manheim's full monthly series, would be the natural follow-up if that data gap is closed.
• The Philadelphia Fed's stock/flow decomposition is the single most important qualifying finding here: it argues the 'record' headline delinquency rate somewhat overstates the pace of genuinely new borrower distress, because much of the increase is unresolved backlog rather than fresh defaults. That distinction matters for how alarmed to be about the trend going forward versus the level today.
• Figures cited (Manheim Index levels, NY Fed delinquency rates, Edmunds negative-equity shares, Philadelphia Fed decomposition) were read directly from the cited primary releases via fetch. One exception: the subprime 60+ day index's '32-year record' figure is from a CarEdge secondary write-up of a Fitch-sourced chart originally posted on LinkedIn, not a direct Fitch publication fetch — flagged here as a lower-confidence data point relative to the NY Fed and Philadelphia Fed figures, which were read from the agencies' own releases.

Every query behind this report

1 warehouse call ran in this session, in order. Each is reproducible against the same snapshot.

SELECT DISTINCT series FROM econ.new_vehicles_cpi