On 26 June 1974, regulators shut down Bankhaus Herstatt at half past three, Frankfurt time. It sounds like a footnote. It is, quietly, one of the most consequential afternoons in the history of money — because of what the timing did to everyone on the other side of its trades.
Herstatt had spent the morning receiving Deutschmarks from banks it had traded currency with. In return it owed those banks US dollars, payable in New York — where, thanks to the time difference, the business day had barely started. When German regulators pulled the plug mid-afternoon, the Deutschmarks were already in. The dollars never went out. Counterparties around the world had paid one leg of a trade and would never receive the other. They had, without meaning to, made an unsecured loan to a corpse.
The industry got a permanent vocabulary word out of it: Herstatt risk, the risk that in a two-currency trade you pay your side before your counterparty — who may not survive the gap — pays theirs. In a world where the two legs settle in two time zones, every FX trade quietly contains that window of faith.
A 28-year answer
The fix took nearly three decades to build. In 2002 the industry launched CLS Bank, whose entire reason to exist is a single principle: payment-versus-payment. Both legs of an FX trade settle at the same instant, or neither does. There is no window in which you have paid and not been paid. It now settles trillions of dollars a day across the major currencies, and it is one of the least famous critical pieces of infrastructure on the planet.
What I find worth teaching about Herstatt is not the history — it is the shape of the question, because that shape recurs everywhere. Every settlement system ever designed is an answer to "what if my counterparty dies mid-transaction?" Cards answer it with networks and rules and a clearing cycle. CLS answers it with payment-versus-payment. Instant rails answer it by settling in central-bank money in seconds, so the window shrinks to nothing. Blockchains answer it with atomicity — the trade executes as one indivisible operation or not at all, which is Herstatt's lesson expressed in code.
Same demon, different exorcisms. When you evaluate any new way to move money — a stablecoin corridor, a rail-to-rail link, a tokenized-deposit pilot — the sharpest first question is not how fast it is or what it costs. It is: where is the Herstatt window, and what closes it? If the answer is "there isn't one, because both sides move together," you are looking at real progress. If the answer is a shrug, someone is about to relearn a lesson a German bank taught the world one afternoon in 1974.