Ohio v. American Express

In 2003, Jean-Charles Rochet and Jean Tirole formalized what the credit card industry had long understood: some markets have two distinct groups of customers who need each other, connected by a platform.^[1] Credit cards need merchants to accept them and cardholders to carry them. Neither side has value without the other. This creates indirect network effects — more merchants accepting a card makes it more valuable to cardholders, and more cardholders makes merchants more willing to accept the card.

The critical economic insight: in two-sided markets, the price structure matters as much as the price level. A platform can raise fees on one side and use the revenue to subsidize the other, growing the network overall even if one side pays more. Rochet and Tirole called this "price structure non-neutrality" — a platform's optimal strategy is almost never to charge both sides equally.

This is the economic theory that American Express rode to the Supreme Court. If merchant fees and cardholder rewards are two sides of one coin, then you cannot evaluate one side in isolation. Raising merchant fees is not anticompetitive if the revenue funds rewards that grow the network.

American Express's non-discrimination provisions (NDPs) were the conduct at issue. Merchants who accepted Amex were contractually prohibited from steering customers toward cheaper card networks at the point of sale. A merchant could not say "We prefer Visa" or offer a discount for using a lower-fee card — even though Amex's interchange fees were substantially higher than Visa's or Mastercard's.^[2]

The anti-steering rules created what economists call a "competitive externality." Discover, which charged lower fees, could not gain market share because merchants were forbidden from communicating the price difference. The price signal — the mechanism through which competition normally works — was contractually suppressed. Visa and Mastercard settled with the DOJ and dropped their NDPs. Amex refused.

The DOJ filed suit in October 2010 alongside 17 state attorneys general. Visa and Mastercard settled almost immediately, agreeing to drop their anti-steering rules. American Express fought through eight years of litigation, from the Eastern District of New York to the Supreme Court.^[3]

The case took a decisive turn in June 2017 when the Trump DOJ switched sides and filed a brief supporting American Express. Ohio stepped in to lead the petitioners. The 5–4 decision, handed down in June 2018, split along ideological lines — with Justice Kennedy providing the decisive fifth vote in one of his final opinions.

Everything turned on market definition. Justice Thomas, writing for the majority, held that credit card networks are "two-sided transaction platforms" — a category he distinguished from two-sided non-transaction platforms like newspapers or shopping malls. Because every credit card transaction inherently requires both a merchant and a cardholder, the relevant market is the transaction itself, not either side independently.

The practical consequence: to prove antitrust harm, the government had to show that Amex's anti-steering rules produced net harm across both sides of the platform. Showing that merchants paid higher fees was insufficient unless the government also proved that cardholder benefits did not offset the harm. The government had not attempted this proof, so the case was over.

Breyer's dissent was unusually sharp. He argued the majority invented a new economic category with no basis in antitrust precedent, then used it to impose an evidentiary burden so high that no plaintiff could meet it. The table below shows the five key points of disagreement.^[4]

Every major platform antitrust case since 2018 has had to contend with Ohio v. Amex. The decision gave platform defendants a powerful shield: argue that you operate a two-sided market, then challenge plaintiffs to prove net harm across both sides. Some courts have accepted the framing. Others have pushed back.

The Amex framework has become the first move in every platform defense playbook. Its staying power depends on whether courts distinguish "transaction platforms" (credit cards) from "non-transaction platforms" (app stores, search engines, marketplaces) — a distinction the majority introduced but left undeveloped.

Khan published her paper in January 2017. The Supreme Court decided Ohio v. Amex in June 2018. Read together, they form a complete picture: Khan identified the problem (the consumer welfare standard cannot see platform power), and the Court demonstrated it (a platform raised merchant fees 20 times and the law said it was fine). The dissent essentially wrote Khan's argument into Supreme Court jurisprudence — but lost 5–4.^[5]