American Express Co

Strong business at a demanding price: the numbers show quality, but the shares trade above our base value.

American Express earns most of its money when Card Members spend on its cards, taking a fee from merchants called discount revenue and interest on revolving balances. Our tests found no loss years across 15 years of filings, owner earnings of $10.6B in the latest year, and a share count that fell by -3.2% a year. Return on equity averaged 29% but dropped to 14% in one year, and the little-debt test failed at $57.8B of borrowing. At $316.41, the shares sit above our base value of $303.98, with a bear value of $169.77 and a bull value of $493.02.

American Express earns most of its money when Card Members spend on its cards. It takes a fee from the merchant on each purchase, called discount revenue, and earns interest on revolving balances and loans. It also charges annual fees on many cards and runs its own card network and merchant acquiring business, so it sees each transaction from both ends.

• Discount revenue from merchant fees on card spending
• Interest income on revolving Card Member loans
• Net card fees from premium annual fees
• Partner payments and cobrand arrangements, led by Delta
• Spending growth from Card Members and merchant acceptance

Inside the circle? Fifteen years of filings show no loss years and a lowest return on equity of 14%. The main unknowns are partner renewals, card fee and rate rules, and whether AI-led payments move customers away, none of which the filings let us size. Debt ratios such as 1.7× say little for a bank that funds loans with deposits and debt, so a careful owner could sketch a decade-long range but not a precise figure.

The main protection is the brand and the Membership Model, which lets American Express earn fees from merchants and cardholders and still attract high-spending customers. The filing says the brand is ranked among the most valuable in the world and that card spend per card is higher than at network competitors, though it gives no figure for the gap. The network is a narrower asset, because the filing ranks American Express fourth by purchase volume. Gross margin and operating income are missing from our data, so return on capital cannot be checked and is marked unknown; return on equity averaged 29% but fell to 14% in one year, and bank borrowing lifts it. We rate the moat narrow because merchant discount rates are eroding and partners can renegotiate; at $316.41 the price sits above the base value of $303.98, so the margin of safety is -4.1%.

Management returns most of the cash the business generates to owners through share buybacks and dividends. The share count has fallen by -3.2% a year, which is the direction a Buffett-style owner looks for. Dividends take a modest share of profit, at 26% on average, so most of the return comes through buybacks. The filing says repurchases happen at prices management deems appropriate, but it gives no view of intrinsic value, and at $316.41 our base value of $303.98 sits below the share price. The company also funds its growth with debt and deposits, with total debt at $57.8B.

• Shares outstanding have fallen by -3.2% a year, so buybacks have more than offset new shares issued to staff.
• On our measure, each dollar kept has created 3.0× of market value. That result depends on our share price and value estimates, so treat it as indicative.
• Partner payments are rising. The filing says the amount paid to cobrand partners has increased, which can reduce the cash that stays with owners.
• The excerpts disclose the early 2026 buyout of the Swisscard joint venture but not its price, so we cannot test whether that use of capital was sensible.

• FailsLittle or no debtFails

  Total debt of $57.8B would take 5.7 yrs of normalised owner earnings to repay; operating income covers interest 1.7×. Interest cover decided this: under 4x fails (payback alone would be mixed).

  Why this test
• FailsMargin of safetyFails

  At $316.41, the price is above the base-case value of $303.98 (the bear case is $169.77), so there is no margin of safety (-4.1%). The standard lens asks for 25%.

Buffett's test of a durable competitive advantage looks strongest on brand. The filing says: 'The American Express brand is ranked among the most valuable brands in the world.' [0000004962-26-000080:business:12] The network is a weaker asset, because the filing says: 'We are the fourth largest general-purpose card network globally based on purchase volume, behind Visa, China UnionPay and Mastercard.' [0000004962-26-000080:business:22] On pricing power, the filing admits: 'We have also experienced erosion of our merchant discount rates as we increase merchant acceptance.' [0000004962-26-000080:risk:24] That weakens the case that the company can raise prices without losing business. Consistent earning power and owner earnings both clear our tests, and the company's capital needs are modest, with capital spending at 1.1× of depreciation. Buffett's little-debt principle is where the case weakens. The test failed, with debt of $57.8B and interest cover of 1.7×. For a bank, that comparison is not like-for-like, because deposits and borrowing fund loans by design, but it still shows how much of the return is borrowed. Buffett's rates principle matters too. The owner earnings yield of 4.6% sits below the 10-year Treasury yield of 5.2%, so the business has to grow to justify the price. Buyback activity fits Buffett's view of rational capital allocation, since repurchases make sense only below intrinsic value, and our base value is below the current price.

• Merchant discount rates and surcharging language in future annual reports

  Falling discount rates would show that pricing power is eroding, which is the central question for the moat.

• Terms of the Delta cobrand agreement as it approaches its end in 2029

  Delta drives a large share of billed business and loans, so weaker terms would hit revenue and partner payments.

• Growth of Card Member rewards and services expenses compared with revenue growth

  The filing says these costs grew faster than revenues in the latest year, which would squeeze returns if the pattern continues.

• Return on equity and funding mix between deposits and borrowing

  Our debt test is hard to read for a bank, so the funding mix shows how much of the return depends on borrowing.

• Capital rules and the company's cross-jurisdictional activity against the Category II threshold

  A move to Category II would bring tighter capital and liquidity requirements, which could limit buybacks.

Questions the filings couldn't answer

• Our data has no operating income or invested capital figures, so we cannot test return on capital, and it is unknown whether the business earns high returns on the capital it uses.
• The filings do not split funding between deposits and wholesale borrowing in a way that shows how much of the debt test reflects the bank model rather than risk.