The agreement is an all-stock deal and the document says what every all-stock deal document says, which is a ratio. Discover shareholders receive 1.0192 Capital One shares for each share they hold.
At Monday’s announcement that valued Discover Financial Services at about thirty-five point three billion dollars, a premium of roughly twenty-six percent, leaving Capital One holders with sixty percent of the combined company and Discover holders with forty.
Those are the headline entries. What follows is the rest of the ledger, because the price is the least informative number in this transaction and the asset being bought is not the one in the headlines.
What is actually on the other side of the trade
Capital One is an issuer. It lends to cardholders, carries the balances, takes the credit risk and earns the interest. It is one of the largest in the United States and it is very good at it.
Discover is two businesses stapled together, and the smaller one is the point. It is also an issuer, with its own cardholders and its own balances. It additionally owns and operates a payment network: the rails that carry a transaction from the terminal in a shop to the bank that authorizes it.
There are four such networks operating at scale in the United States. Two of them, Visa and Mastercard, are not owned by any bank; they are utilities that every issuer rents. The other two, American Express and Discover, are owned by companies that also issue cards on them.
A network is not for sale in the ordinary course. Building one requires merchant acceptance, which requires cardholders, which requires acceptance, and the circularity is why there are four and not fourteen. The only way to acquire one is to acquire the company around it.
The entry that makes the deal work
Here is the economics, stripped of the announcement language.
When a Capital One card is used today, the transaction runs over somebody else’s rails, and the fee for carrying it goes to the owner of those rails. Capital One earns the interchange as issuer and pays the network fee as a tenant. It does this on every transaction, on every card, every day.
Move that volume onto a network it owns and the fee stops being a payment to a third party and becomes an internal transfer. Nothing about the transaction changes for the cardholder or the merchant. The money simply stops leaving the building.
That is the ledger entry the thirty-five billion is against: not Discover’s loan book, which is a perfectly good but ordinary asset, but the difference between renting rails and owning them, compounded over every future transaction the issuer originates.
Who pays
Existing Capital One shareholders pay, and they pay in dilution rather than in cash. Holding sixty percent of a larger company is the price, and whether it was a good price depends entirely on how much volume can actually be migrated, which is the number nobody has published.
Migration is not automatic and it is not free. Merchant acceptance for Discover is real but narrower than Visa and Mastercard, particularly outside the United States. Moving a premium travel product onto a network a cardholder finds is not accepted in a shop abroad is a way of losing the cardholder, so the migration will be selective and slow, and its pace is the whole return on the transaction.
Who collects
Discover shareholders collect the premium, in stock, subject to the deal closing.
The incumbent networks collect nothing and lose something, though less than the coverage has suggested. They lose one tenant’s future volume, on a timetable that is measured in years and is at the acquirer’s discretion.
What merchants collect is the open question. The argument offered for the deal is that a better funded fourth network competes harder with the largest two, which would in principle bear on the fees merchants pay. That case is coherent. It is also the case every acquirer in a concentrated market makes, and it rests on a network’s owner choosing to compete on price rather than to capture the spread itself.
The structure of the payment is worth one paragraph, because all-stock is a choice and not a default. Paying in shares means no capital has to be raised and no cash leaves, which matters for a bank holding company whose regulators care about capital ratios. It also means the seller’s holders keep their exposure: if the integration goes badly, Discover’s former shareholders own forty percent of the disappointment. Cash would have transferred that risk entirely to the buyer, and the buyer declined to take it.
It follows that the thirty-five point three billion is not a fixed price. It is a ratio multiplied by a share price that moves every day between announcement and closing, so the figure in this week’s headlines is a snapshot of Monday and will be a different number on the day the deal completes, in either direction.
The entries that are not yet numbers
Two liabilities sit on this ledger without values attached.
The first is regulatory. A transaction that combines two large card issuers and hands one of them a payment network will be examined by banking regulators and by the antitrust agencies, in an environment that has been notably unfriendly to consolidation. The deal is priced as though it closes. Every month it does not close is a cost.
The second is integration, which is where acquisitions of this shape usually disappoint. Two issuers means two card platforms, two servicing operations, two sets of terms, two compliance histories and two workforces, and card platform migrations have a long record of overrunning.
Nothing on this ledger is visible to a cardholder. The card in the wallet keeps its number, its logo and its terms, and the transaction clears in the same second it always did. That is the normal condition of payments infrastructure: the parts that decide where several billion dollars a year of fees end up are the parts nobody standing at a till can see.
The balance, such as it is
Thirty-five point three billion in stock, for a loan book that is ordinary, a customer base that overlaps, and one of four sets of rails in the United States.
The rails are the reason. Everything else on the ledger is the cost of reaching them, and the return is a fee that stops being paid out, on volume that has not yet been moved, over a period nobody has committed to in writing. It is a plausible trade and it is not a proven one, and the number that will eventually settle it is not on any document filed this week.
