CLS made FX settle both legs at once, and Herstatt risk finally died in 2002

On June 26, 1974, German regulators closed Bankhaus Herstatt at 3:30 pm Frankfurt time — after the day's Deutsche mark legs of dollar-mark trades had already been paid. Herstatt's New York correspondent then suspended all outgoing dollar payments. Counterparties had delivered their marks with finality and would never receive their dollars: the full principal of each trade was gone. The episode named 'Herstatt risk' and exposed a structural flaw — FX settled gross and bilaterally, each leg final in its own time zone, so every trade carried a window where one side could lose everything.

For two decades the market simply priced that window. By the mid-1990s, foreign exchange turnover reached about $1.25 trillion a day and central banks confronted what a single settlement failure could do to confidence. The Committee on Payment and Settlement Systems set a three-part strategy in 1996: individual banks should measure and cap their settlement exposures, industry groups should build risk-reducing multicurrency services, and central banks should induce private-sector progress. The dealer banks took the middle assignment and built what became CLS.

When CLS Bank went live in September 2002, its single design choice — payment versus payment — removed the fatal state of the world. A currency payment finalizes inside CLS if and only if the counter-currency payment does; members pay in each currency through accounts at the relevant central banks during that currency's own business hours, so no time-zone seam remains where principal can vanish one-sided. Multilateral netting cut the cash members must pre-fund from the gross value of their trades to a small net amount, making the safe route also the cheap route.

That combination is why it won. Every participant had an independent reason to route volume through CLS — not regulation, but arithmetic — and central-bank involvement gave the private utility the finality and credibility that only settlement in central bank money provides.

Settle both legs together or not at all

The risk was clock skew, not misconduct: each settlement leg was final in its own time zone, so only a shared architecture could fix the seam — more diligence at each bank could not.

If-and-only-if settlement made the catastrophic outcome (one leg paid, the other defaulted) structurally impossible, which bilateral credit checks never could.

Multilateral netting made participation cheaper than bilateral settlement in both liquidity and operations, so adoption followed self-interest rather than mandate.

Settling in central bank money under Federal Reserve Bank of New York-led cross-border oversight gave a private platform the finality banks need for trillion-dollar flows.

When a failure mode lives in the seams between systems, rebuild the sequence so the bad event cannot exist. Making one payment conditional on its counterpart turns risk management into structure.

CLSSettlement now clears 18 major currencies and the majority of volume in eligible pairs. The BIS Triennial Survey of April 2025 found $5.2 trillion a day — about 36 percent of daily FX settlement — moving through PvP, which eliminates settlement risk outright, with another $7.6 trillion (54 percent) in partially mitigating methods and $1.4 trillion (10 percent) still fully exposed on a gross bilateral basis, mostly where PvP access or currency eligibility is missing. The pattern has spread: PvP-style clearing now operates in Hong Kong (CHATS), Brazil (B3), India (CCIL) and the UAE (Buna).