Herstatt Risk: What PvP Settlement Does and Doesn't Fix

Two promises, one heartbeat apart

Imagine you and a friend agree to swap borrowed books. You hand over yours, wait, and then watch your friend get distracted by something shiny and walk off with both. That, in miniature, is what currency traders call settlement risk: the chance that one party pays a currency and never receives the one owed in return. In foreign exchange, that gap can be huge, because the full value of a trade—not just a slice of it—is at stake between the moment one side pays and the other is due to pay.

The risk has a name with a history. In 1974, a German bank called Bankhaus Herstatt was closed on a June afternoon in Frankfurt. Its counterparties had already paid out Deutsche marks earlier that day, but the dollars they were owed got suspended in New York. They were left exposed for the whole value of the trade. That episode is why settlement risk in currency trades is often called Herstatt risk. It was not just a bad day for a few banks; it shook trust, stalled payments, and prompted the creation of international committees that still watch over how the plumbing of finance works.

What payment-versus-payment actually does

Think of a payment-versus-payment (PvP) system as a strict doorman. The final payment of one currency goes through only if the final payment of the other currency goes through too. Neither leg happens first. In its pure form, that eliminates the Herstatt scenario entirely, because one side cannot pay with finality and then watch its counterparty disappear on the other leg.

The biggest and best-known PvP system is CLSSettlement, which handles payments in 18 major currencies. Others have emerged over time, some focused on currency pairs CLS doesn't cover—such as B3 in Brazil, CCIL in India, CHATS in Hong Kong, and Buna in the UAE.

One useful detail: PvP is not a central counterparty (CCP). A CCP inserts itself between the two sides to the trade and takes on their obligations. PvP doesn't do that. It simply synchronises the two payment legs. So while PvP eliminates the risk of one currency being paid without the other, it does not remove replacement cost risk—that is, the chance that replacing a trade after a default costs more at market prices—or the liquidity risk of not having a currency when you expected it.

The 2025 numbers: progress with a long tail

The April 2025 Triennial Survey from the Bank for International Settlements is a useful snapshot. Of the roughly $14 trillion of gross financial obligations settled on an average day, just over $5 trillion—36%—went through PvP systems. That's a meaningful chunk. About $7.6 trillion, or 54%, was settled using methods that reduce but do not eliminate settlement risk: pre-settlement netting, intragroup settlement, or settlement across bank accounts with timing controls. The remaining $1.4 trillion, roughly 10%, was settled on a gross bilateral basis—fully exposed, with no mitigation.

Compare that with 2006, when the comparable survey put gross bilateral settlement at 32% of average daily settlement. So things have improved substantially. But PvP usage itself has only moved modestly, from around 54% to roughly 56%—a reminder that "progress" and "problem solved" are different places.

Why doesn't everything just go through PvP?

The survey dug into why some trades still settle in the open. Among trades settled on a gross bilateral basis, about three-quarters were not eligible for PvP. Three overlapping reasons showed up. First, the counterparty lacked direct or indirect access to a PvP system—the most common reason, at 57%. Second, the currency pair was ineligible, often because it wasn't covered by an available system. Third, the trade type was ineligible; PvP systems generally operate on set cut-off times for T+1 or T+2 settlement, and same-day trades don't always fit.

The remaining 25% is the frustrating bit: these trades were eligible for PvP but settled bilaterally anyway. Operational snags, missed cut-off times, and credit-risk management complications with correspondent banks can all get in the way. It's a bit like having a direct flight available and booking a three-stop itinerary because the booking system crashed mid-purchase. Not a crisis—but a fixable gap.

What settlement finality doesn't fix

Here's the important clarity: the word "finality" in payments means a transfer is irrevocable. In a PvP system, finality on one leg is tied to finality on the other. That is genuinely powerful. But it does not make a counterparty solvent, it does not guarantee an institution has liquidity in the right currency on the right day, and it does not resolve replacement cost risk.

It also doesn't cover every trade. Non-bank financial institutions and smaller counterparties often lack access; emerging-market currencies are less frequently supported; and as the structure of FX intermediation shifts—with a rising footprint of non-bank players—the share of trades outside PvP systems may nudge upward without deliberate effort to broaden access. Jurisdiction matters too: legal frameworks, FX convertibility rules, and local payment-system reliability all shape whether a new currency can realistically be added to a PvP system.

The survey also published settlement fails data for the first time: trades due in April 2025 that remained unsettled at month end. They accounted for 0.01% of total gross financial obligations—a small number, but one worth tracking, especially as some jurisdictions shorten securities settlement cycles and firms have less time to fix confirmation and settlement issues.

The road ahead

Public and private efforts continue. The updated FX Global Code encourages a settlement hierarchy and regular reviews of settlement choices. And new technologies—including distributed ledger approaches where smart contracts execute settlement only if both sides pay—are being explored. Project Agorá and earlier projects like Jura and Mariana are part of that longer-term work.

Fifty years after Herstatt, the architecture is much stronger. But $1.4 trillion a day still moves without a safety net, and not every risk disappears the moment payments are locked together. That's not a reason for alarm—it's a reason to keep watching the numbers and asking which trades are still outside the doorman's view.