AMERICAN EXPRESS CO (AXP) stock intrinsic value fair value estimate & owner earnings analysis

American Express runs a closed-loop card network: it issues the cards, signs the merchants and runs the network, so it earns annual card fees from members, discount revenue from merchants and interest on card loans. In 2025 revenue net of interest expense was $72.2 billion, net income $10.8 billion, and the dividend was raised about 16% from the first quarter of 2026. In the first half of 2026 revenue was $38.5 billion (up 10.7%) and net income $6.08 billion (up 11%). How we measured earnings. We start from operating cash flow, strip out swings in working capital, and deduct stock compensation and the capital spending needed to maintain the business. Because American Express is a lender, we also deduct the provision for credit losses: losses on card loans are a recurring cost of the business, not a one-off write-down. On that basis owner earnings per share were $9.06 in 2022, $10.95 in 2023, $11.96 in 2024 and $13.71 in 2025. The filing's own cash-flow lines add up exactly to reported operating cash flow in each year. The earnings level. We use $11.46 per share, the median of 2022 to 2025. We leave out 2020 ($3.65, pandemic credit losses) and 2021 ($10.05, flattered by a $1.4 billion release of loan-loss reserves), because each reflects a one-off credit condition; the credit cycle is carried in the bear case instead. Owner earnings grew 14.6% in 2025, and the base case assumes only 6% in its first year, so it does not depend on that pace continuing. Debt. For a card lender, deposits ($157 billion at June 2026) and long-term debt ($57.0 billion) are the raw material of the loan book and interest is already an operating cost inside these earnings, so we do not subtract them again. The only claim ahead of common shareholders is the $1.6 billion Series D preferred stock (carried at $1,584 million), about $2.32 per share, deducted in every case. Growth. Growth needs regulatory capital, since the company manages its common equity tier 1 ratio within a 10 to 11 percent target (10.5% at year-end 2025). So cash to owners in each year is owner earnings times one minus growth divided by the return on capital. The growth we credit is the continuation of a demonstrated, self-funded process: revenue compounded about 8% a year from 2016 to 2025 through the 2020 downturn, card fees are paid up front, and in 2025 the company returned $7.6 billion to shareholders while still growing. We credit nothing for new products, share gains, or the next card refresh. Base case, $269 per share. Owner earnings start at $11.46 and grow 6% a year for five years, 5% for three and 4% for two, funded at a 30% return on capital (2025 return on average equity was about 34%), then stay flat after year 10 with no further growth assumed. The business is worth $271.42 per share, it owes $2.32 of preferred, and the difference is $269.10. About two-thirds of this value comes from the years after year 10, resting on a franchise that has kept earning through 2008-2010 and 2020. Bear case, $161 per share. Two named risks. First, a credit cycle of the kind this business has actually had: in 2020 owner earnings per share fell 56% from 2019, and the 2008-2010 cycle saw card write-off rates several times today's 2.3%. We assume earnings fall to 44% of the normal level for two years, recover to 70% in year four and 85% in year five. Second, the Delta cobrand agreement, which accounts for about 13% of billed business and 21% of card loans, runs only through the end of 2029. We assume it lapses with nothing replacing it, taking 8% off earnings permanently from year four (an estimate, since the filings do not disclose Delta's profit contribution). No growth is credited. The business is worth $163.38 per share, it owes $2.32 of preferred, and the difference is $161.06. Even at the trough, earnings stay positive and the company would not need outside capital. This exceeds tangible book value of about $41 per share because the franchise earns its returns from fees and merchant discounts on a closed network rather than from its capital alone, and that earning power survived both prior downturns without dilution. Bull case, $328 per share. Earnings hold the 2025 level of $13.71, already achieved, and grow 7% for three years, 6% for two, 5% for two, 4% for two and 3% in year ten, funded at a 32% return on capital, then flat. This is still below the roughly 8% revenue growth actually delivered over the past decade. The business is worth $330.09, less $2.32 of preferred, $327.77. What we tested. Capital spending in the first half of 2026 was $2.0 billion against $1.0 billion a year earlier and we could not tie it to a disclosed project; if it recurs, maintenance spending is understated and value is lower. The annual report names potential credit card interest-rate caps and pricing or network regulation as risks; none has been enacted, so none is modeled, but a cap on card interest would hit the loan income that makes up a growing share of revenue. Change since our previous estimate. The base case rises from $248 to $269 because the first half of 2026 confirmed growth of about 11% in revenue and net income, which supports a modestly higher near-term path; the bear case is essentially unchanged because a slightly deeper credit trough offsets the lower preferred deduction. Margin of safety price, $160. We are fairly confident in the earnings level: it is the middle of four ordinary years, below the latest year, and credit losses are already charged. The worst plausible outcome is a severe credit downturn arriving alongside the loss of Delta or an interest-rate cap; the business would survive it, with no maturity wall that matters and capital well above minimums, but earnings could be cut in half for two or three years. If we are wrong about the credit cycle depth and the durability of the premium fee model, the shares would be worth about the bear case, roughly $160. During the holding period, the company pays a dividend of about $3.80 per share a year and retires shares steadily. The margin-of-safety price of $160 is close to the bear-case value, and if the business simply continues on the base path, the gap to value plus dividends implies about 20% a year over three to four years. Any higher price leaves the 20% target dependent on the base case arriving quickly.