CREDIT ACCEPTANCE CORP (CACC) stock intrinsic value fair value estimate & owner earnings analysis

Credit Acceptance lends to used-car buyers with poor or no credit through about 15,700 active dealers (a record in 2025). Its distinctive structure is the Portfolio Program: rather than buying the loan outright, it advances the dealer a portion of the car's financed amount and then shares collections, paying the dealer a 'holdback' only after it has recovered its advance and fees. That puts the dealer's own money behind the quality of the loan and has let the company stay profitable in every year of the past decade, including the 2020 shutdown and the 2023-2024 trough. Because borrowing is the raw material of a lender, we measure owner earnings as net income to common shareholders, after interest and after loss provisions, and do not subtract the $6.4 billion of asset-backed financings and senior notes again; stock compensation (about $51 million in 2025) is already charged in that figure. Earnings over the last ten years: $333 million (2016), $470 million (2017, flattered by a one-time tax reform benefit), $574 million, $656 million, $421 million (2020, a reserve build at the onset of the pandemic and a new accounting standard), $958 million (2021, flattered by the release of those reserves), $536 million, $286 million, $248 million (2024) and $424 million (2025). The 2023-2024 dip was the 2021-2022 loan vintages, written at inflated used-car prices, collecting less than forecast. The 2025 figure absorbed a $74 million legal charge; the first half of 2026 earned $272 million. Across the full cycle from 2018 to 2025 the median is about $480 million and over ten years about $447 million; we use $450 million as the sustainable level. That is a level, not a growth assumption: 2025 earnings rose 71%, and we assume about $500 million for 2026 (an 18% rise, below the trailing rate and supported by the first-half result) before settling at $450 million. We credit no further growth in profit dollars, because loan volumes depend on competition from other subprime lenders and on the company's own choice of advance rates, neither of which is secured. Per-share growth has historically come from buying back shares (about 45% of the share count retired since 2015); we value total earnings on today's 10.7 million diluted shares and do not add a buyback bonus on top. Base case: $500 million in 2026, then $450 million a year with no growth and continuing indefinitely, on the evidence of a 30-year record of profitability through every credit cycle. The business is worth about $7.5 billion; because the debt is the lending book's funding, nothing is subtracted, giving about $703 a share. Bear case: a full subprime auto credit downturn of the kind the industry had in 2008-2010 and the company itself had in 2023-2024, when earnings fell about 74% from peak to trough. We assume $430 million in 2026 (including a further legal cost), then $300 million, $200 million and $200 million as loss forecasts are cut and reserves rebuilt, $300 million in the recovery year and $350 million thereafter, with lasting earnings of $300 million a year as competition and the multi-state and New York consumer-protection matters (on which the company recognised $82.6 million of cumulative contingent losses and reached preliminary alignment with the agencies on a potential settlement) squeeze returns. No dilution is assumed, because in both 2009 and 2024 the business stayed solidly profitable and kept its funding markets open. That is worth about $5.1 billion, or $477 a share, well above tangible book of about $148 a share; the excess is justified only because the dealer-holdback structure has kept the company profitable through every downturn on record. If funding markets shut while losses rise, this case would be too generous. Bull case: $540 million in 2026, consistent with the first-half run rate, then $520 million rising about 3% a year as record dealer enrolment is converted into volume, while about 12% of earnings is retained to fund the larger book. That is worth about $9.9 billion, or $924 a share. These figures are not comparable to our previous estimates, which were built on a different basis. Margin of safety price. We are fairly confident in the earnings level: ten years of audited profits and a complete cycle give a defensible range of roughly $250 to $650 million. The worst plausible outcome is a severe recession combined with a tightened regulatory regime for subprime auto lending and a closure of the asset-backed securities market at the same time. With debt of roughly four times equity, the business depends on rolling over its warehouse lines and securitisations; it survived 2008-2009 and 2020 doing exactly that, but a simultaneous funding and credit shock could force it to shrink sharply and halt buybacks. The three things we are most likely to be wrong about are the depth of the next credit downturn, the size of any future regulatory penalties or restrictions, and whether competition squeezes returns permanently; if all three went against us the shares might be worth around $350 to $400. The business returns cash to its owners almost entirely through repurchases, roughly $40 to $45 a share of earnings a year, which at a low share price compounds per-share value quickly over the holding period. The margin-of-safety price is $300 a share: at that price the annual earnings alone represent around 14% to 15% of the price when retired through repurchases, and if the value merely reaches our bear-case level within three years the combined return the bear case implies exceeds 20% a year, while a scenario worse than our bear case would imply a modest loss rather than a ruinous one.