Credit Acceptance's $710M settlement prices the cost of being the deepest subprime lender in America

Only $75.5M of the $710M leaves Credit Acceptance Corp's bank account. The other $634M is debt the company extinguishes on its own balance sheet — balances owed by subprime borrowers on loans originated between November 2015 and November 2025 that the 40-state coalition deemed \"destined to fail.\" That makes the headline number misleading on three fronts at once: it overstates the cash hit, understates the GAAP impact, and quietly embeds a permanent change to CACC's underwriting playbook.

The closest analog is the May 2020 multistate settlement with Santander Consumer USA: 34 attorneys general, $550M total, $65M cash restitution, $478M in deficiency-balance waivers. Credit Acceptance Corp's deal is 29% larger in headline terms and notably heavier on principal forgiveness relative to cash. The structural concessions also go further — Santander's 2020 deal required an ability-to-repay analysis but did not impose a 95% post-default waiver; CACC's does.

The two settlements together define the modern AG template for subprime auto: deficiency-balance forgiveness plus a structural underwriting concession, with the cash component reserved for restitution rather than deterrence. The 2020 Santander deal had a measurable chilling effect on originations in the following quarters; expect the same playbook to be filed against any lender whose repossession rates, APRs, or default metrics look like CACC's. By Capstone's January 2021 analysis, CACC's repossession rate was 35% versus \"mid-teens\" at Santander, Exeter, DriveTime, and American Credit Acceptance — and its borrowers defaulted more than twice as often as Santander's. That gap is now the bullseye on every other subprime lender's back.

CACC reports on a forecast-cash-flow accounting basis: every quarter, it re-projects expected collections on the entire loan book. Any change to those forecasts flows through revenue. In Q2 2026, the company reported a $39.1M negative forecast revision (-0.3% of portfolio) and still printed $135.9M of net income on $587.4M of revenue. A $634M debt forgiveness plus the new 95% waiver rule on in-flight vintages is many multiples larger than the typical quarterly revision. The market's muted1.5% stock drop (to $592.30 on Sep 17, from $600.11 on Sep 15) suggests investors expect the charge to be treated as a one-time item — but a 12-month waiver rule structurally raises future charge-offs and depresses the spread (currently 21.9% annualized on 2026 vintages) that CACC earns on its advance-plus-fee model.

The brief's \"private-credit BDC\" framing doesn't fit cleanly here: CACC doesn't tap BDC balance sheets. Its $6.29B of long-term debt is overwhelmingly non-recourse ABS notes and warehouse lines — meaning the real exposure sits with ABS investors (CLOs, money-market funds, separately managed accounts that buy the B-piece and senior tranches of CAALT 2026-1 and CAALT 2026-2). The contagion question runs through three lanes instead: captive-finance auto retailers, prime-to-nonprime crossover lenders, and the ABS shelf itself.

Carvana is the most direct read-through. Its nonprime share has been documented at44% of borrowers with prime delinquencies \"four times the industry average\" (per November 2025 third-party reporting), and its servicing arm is DriveTime-owned Bridgecrest — a captive that KBRA affirmed-and-upgraded in August 2026 on the back of strong originations. Q2 2026 results were record on every line (revenue $7.376B, net income $513M, retail units +38% YoY), and management guided Q3 2026 adjusted EBITDA of $2.7B-$3.0B with sequential unit growth — so the underlying franchise is firing. The risk is regulatory: a 40-state template that explicitly targets deep-subprime practices will draw AG attention to any captive whose book starts to look like CACC's. Carvana's stock closed at $65.43 on Sep 16 (down about 35% YTD in 2026), so the read-through risk is more about a regulatory headline than a fundamentals break.

CACC's August 20, 2026 ABS deal — CAALT 2026-2 — priced at an expected average annualized cost of 5.5%, transferred $750.2M of consumer loans into the trust, and left the company with roughly $1.8B of cash and credit capacity. Just over three weeks later, the same collateral pool faces a 40-state settlement forcing a 7.9% principal writedown. The April 2026 CAALT 2026-1 deal carried an S&P expected cumulative net loss (ECNL) of 20.50% — already the highest-tier assumption in subprime auto. Any new deal launched in Q4 2026 will price against (a) the just-announced forgiveness experience, (b) the 95% default-waiver rule on the most recent vintages, and (c) a competitive environment where Santander Consumer USA, Westlake, and Exeter are all still tapping the market.