A settlement exposes the tension inside a financing model
On September 17, 2026, New York’s attorney general announced a multistate settlement with Credit Acceptance, an auto-finance company serving borrowers with impaired or limited credit histories. The announced package combined more than $630 million of debt relief, $60 million for consumer restitution and $15.5 million in state payments. The central dispute was about a question that reaches beyond this company: can a loan be profitable for the business funding it while still leaving its borrower with a damaging outcome? [1]
Those amounts are different kinds of relief. Forgiving an outstanding balance is not the same as mailing that amount in cash, and the headline total is not an estimate of an additional accounting loss. Credit Acceptance said the monetary components required no charge beyond amounts it had already accrued. It denied wrongdoing and said the agreement did not fundamentally change its model. The settlement resolves a dispute; it does not establish that every borrower had the same experience. [2]
From a dealer’s sale to a finance company’s asset
Credit Acceptance began in 1972 collecting installment contracts from dealerships owned by founder Donald Foss. It expanded to outside dealers in the 1980s. Its business is indirect lending: the dealer originates the consumer’s vehicle-purchase contract, then assigns it to Credit Acceptance. The company’s financial statements distinguish the customer’s contract from its own financial relationship with the dealer. [3]
That distinction explains the unusual Portfolio Program. The dealer receives a customer down payment and an advance from Credit Acceptance. Collections first cover specified collection costs, then a servicing fee generally equal to 20% of collections, then the dealer’s advance balance and other amounts owed. Remaining eligible collections become dealer holdback. Dealer pools can be cross-collateralized, so one pool’s performance can affect holdback eligibility elsewhere. The Purchase Program instead pays the dealer once and gives Credit Acceptance the collections, without future holdback. In 2025, the Portfolio Program accounted for 74.2% of assigned contracts by number. [3]
Why the accounting label matters
Under the Portfolio Program, the company accounts for its advance as a loan to the dealer. That does not mean a household has borrowed from the dealer to operate a dealership. The household still owes under its vehicle contract; the accounting label describes the asset carried by Credit Acceptance. Its June 2026 quarterly filing identifies dealer loans and purchased loans as separate portfolio segments, reflecting their different exposure to changes in collections. [4]
The distinction changes who absorbs disappointment. Reduced collections can consume a dealer’s expected holdback before they reduce the finance company’s expected recovery. That is a cushion, not immunity. If actual cash receipts fall far enough or arrive much later, the company can still lose money. Under the Purchase Program, there is no corresponding future dealer payment to shrink. The company retains the upside from collections above its forecast but also bears the negative variance. These contractual incentives help explain why a car sale, a household obligation and an accounting receivable cannot be treated as three names for one number. [4]
A forecasted collection percentage is not a default rate
Credit Acceptance’s August 4, 2026 earnings release reported $135.9 million of GAAP net income for the three months ended June 30. It also reported $130.1 million of adjusted net income, a company-defined non-GAAP measure. Neither tells the reader what percentage of customers successfully repaid their loans. [5]
Its collection forecasts use a different denominator: the principal and interest contractually due when consumer contracts were assigned. For contracts assigned in the first quarter of 2026, the June forecast was 66.5% of that original contractual amount; for second-quarter assignments it was 67.7%. Those are estimates of lifetime dollar collections, not observed shares of borrowers who defaulted. Canceled contracts can reduce the displayed percentage because their original contractual amounts remain in the denominator. A recent cohort also has much of its repayment life ahead of it. [5]
A simplified hypothetical shows the distinction. Suppose 100 contracts initially call for $3 million of total payments, including interest. A 67% collection forecast means $2.01 million expected across those contracts. It does not mean 67 of the 100 borrowers will repay, nor that $990,000 of principal has already been lost. Some people may repay fully, others partly, and collections may include proceeds obtained after repossession. The cash paid to dealers would be another number. Subtracting that advance from forecast collections still would not produce profit: funding costs, operating costs, dealer payments and timing remain. These are illustrative amounts, not a reconstruction of the company’s portfolio.
The money to fund the next vehicle
The dealer can be paid long before a consumer finishes making installments. Credit Acceptance therefore needs financing of its own. A September 15, 2026 announcement offers a concrete example: a $500 million secured had its revolving period extended to September 15, 2028, with a rate of plus 175 . Only $180 million was outstanding on the announcement date. The commitment and the drawn amount were different figures. [6]
In the same announcement, a separate $500 million asset-backed financing had its revolving period extended to September 15, 2028, and its interest rate increased from 5.43% to 5.83%. One facility’s price fell while the other’s rose. This illustrates why a single headline about lower market rates cannot describe every funding contract. A warehouse can finance eligible assets while collections accumulate or other financing is arranged; the company still faces contractual limits and refinancing risk. Borrower payments support that chain, but a household does not negotiate with the company’s warehouse lenders. [6]
What the 2026 agreement does, and what it does not prove
The legal chronology matters. The Consumer Financial Protection Bureau (CFPB) and New York sued jointly on January 4, 2023, alleging deceptive and abusive practices. The CFPB requested withdrawal in April 2025, and the court granted that request on April 29. Withdrawal by one plaintiff was not a ruling clearing the company of the state’s allegations. New York continued its case, leading to the September 2026 settlement announcement. [7][1]
The publicly linked New York document is a proposed filed September 17. It records the company’s denial of the allegations and says the resolution is without a trial, adjudication or finding of liability. The reviewed copy has an unsigned judicial-entry line, so it supports the agreed terms rather than a separately verified date of court entry. This article does not treat the attorney general’s description of historical conduct as a verdict. [8]
The agreement’s protections are substantive. For contracts originated after December 1, 2025, the proposed order provides a 95% deficiency-balance waiver when specified credit-score and payment-to-net-income tests are met and the vehicle is involuntarily repossessed and sold within 12 or 18 months. It also restricts collection lawsuits and transfers for those eligible accounts. It also addresses disclosures and dealer oversight. Those are eligibility-bound provisions, not an automatic 95% reduction for every Credit Acceptance customer. [8]
New York’s announcement additionally describes outreach outside the showroom about purchased add-ons and a cancellation process that allows the consumer to keep the vehicle. [1]
Three outcomes that can move in different directions
For a household, the outcome is transportation, the cost of payments, the condition of the car and what happens if income falls. For the dealer, it is the cash received at sale plus any later holdback. For the finance company, it is the amount and timing of collections relative to advances, expenses and funding. The economic inference is that success for one participant does not by itself prove success for the others.
The model can expand access for borrowers who have few alternatives, as the company emphasizes. [2] Yet an approval is not evidence that a payment leaves enough room for repairs, insurance and other bills. Conversely, a missed payment does not on its own prove that the original underwriting was unlawful. The decisive evidence lies in contract terms, actual borrower treatment, cohort performance and compliance with the agreed protections. Credit Acceptance’s story is therefore about both the availability of vehicle credit and the distribution of its risks, with neither question settled by a profit figure alone.
Sources
- New York Attorney General, settlement announcement, September 17, 2026Official releaseBack to text: ↑1↑2↑3
- Credit Acceptance, company settlement statement, September 17, 2026SourceBack to text: ↑1↑2
- Credit Acceptance, 2025 Form 10-K, filed February 13, 2026Filing / reportBack to text: ↑1↑2
- Credit Acceptance, Form 10-Q for June 30, 2026, filed August 4, 2026Filing / reportBack to text: ↑1↑2↑3
- Credit Acceptance, second-quarter 2026 earnings release, August 4, 2026Filing / reportBack to text: ↑1↑2
- Credit Acceptance, warehouse and asset-backed financing extensions, September 15, 2026Filing / reportBack to text: ↑1↑2
- CFPB, Credit Acceptance enforcement-action chronology, including April 2025 withdrawalOfficial sourceBack to text: ↑
- New York v. Credit Acceptance, proposed consent order filed September 17, 2026Official source · PDFBack to text: ↑1↑2