De-risking: why correspondent banks cut whole regions off

De-risking is the withdrawal of banking services from customers, banks, or whole regions judged to carry more compliance risk than their business is worth. In correspondent banking it has a specific and severe form: a large international bank closes the accounts through which smaller and foreign banks reach the global payment system. When enough correspondents make the same decision about the same place, that place loses its financial connection to the rest of the world.

The IMF called this a case for policy action as early as 2016. Nearly a decade later the pattern has not reversed, and it is one of the clearest illustrations of the economics that govern the correspondent model.

The economics that drive it

A correspondent relationship carries a largely fixed compliance cost. The correspondent must perform due diligence on the respondent bank, monitor its transactions, screen its customers’ payments, and answer for failures under anti-money-laundering and sanctions rules that carry severe penalties. That cost does not scale down for a small relationship: knowing your customer’s customers in a small corridor costs roughly what it costs in a large one.

Revenue, meanwhile, does scale down. A corridor that produces a thin stream of low-value payments generates fees that can sit below the cost of monitoring it. The correspondent keeps all of the regulatory risk and little of the reward. Faced with that arithmetic, exit is the rational move for each individual bank. The IMF’s staff analysis identified exactly this combination, rising compliance costs and expectations against low profitability, as the core driver, compounded after major enforcement actions raised the perceived price of getting it wrong.

The result is a system-level failure produced by individually sensible decisions. No bank is the villain here; each is responding to the incentives the inherited structure hands it. The corridor still needs payments. The economics just stopped supporting the only mechanism the system offered.

What the data shows

The decline is measurable and uneven. BIS CPMI data covering 2011 to 2022 shows active correspondent relationships fell by roughly 30% globally, with the steepest losses concentrated where economies are smallest: about 63% in Melanesia, 54% in Polynesia, and 52% in the Caribbean. Country-level declines averaged 23% in advanced economies against 41% in small island developing states.

The Pacific is the extreme case. The Pacific Islands Forum’s 2024 report found the region’s correspondent relationships fell about 60% since 2011, roughly double the global average, and the World Bank has been running a dedicated program to keep what it calls the region’s financial lifelines open. In the Caribbean, an association survey cited by the IMF found 21 of 23 banks across 12 countries had lost at least one correspondent relationship, with Belize, Suriname, and parts of the Eastern Caribbean hit hardest.

The full numbers, including what happened to payment volumes while relationships fell, are mapped in the shrinking correspondent network, in numbers.

What happens to the payments that remain

De-risking does not stop payments; it reroutes them, at higher cost, through longer and less visible paths.

• Nested relationships. A bank that loses its direct correspondent borrows access through another bank that still has one. The payment now crosses an extra intermediary, with extra fees and an extra compliance hop, and the ultimate correspondent sees even less of who is behind the flow.
• Higher remittance costs. Fewer competing channels means worse pricing in exactly the corridors that depend on remittances most. The World Bank’s global average cost of 6.36% conceals corridor costs well above it where coverage has thinned.
• Concentration. The remaining correspondents carry more of the traffic, so the system’s coverage now depends on a shorter list of institutions and each further exit hurts more.

The FSB flagged the irony in its monitoring work: pushing flows out of monitored direct relationships into nested arrangements and informal channels can make the system harder to police, an outcome nobody was aiming at.

What would change the economics

De-risking is a cost problem, so durable fixes have to attack the cost side. Richer, structured payment data (ISO 20022) lowers the price of screening and investigating each payment. Shared due-diligence utilities spread fixed costs across banks. And the deepest version of the fix moves the checking itself: if the conditions a payment must satisfy were enforced in settlement, so that a transfer that cannot prove its conditions does not settle, then serving a thin corridor would no longer mean underwriting an open-ended monitoring obligation. The economics of where compliance happens in a payment are, in the end, what decides which corridors get served.

Where Frame fits

Frame is the settlement layer for global finance: one integration to orchestrate payments at scale across fiat rails, stablecoins, and tokenized deposits, with compliance enforced on every transaction and settlement in seconds instead of days.

De-risking is what happens when compliance is a cost bolted onto every relationship. Frame’s design starts from the opposite premise: its Rules Engine evaluates every transaction against its governing policies inside settlement, and a transfer that cannot satisfy them does not settle. Every settled transaction produces verifiable evidence that its conditions were met, without exposing the underlying business data. For banks, payment providers, exchanges, platforms, and enterprises, that changes the shape of the question a marginal corridor poses: from “can we afford to monitor this relationship” to “what policy should govern these payments.”

See how compliance enforced in settlement works in the Frame Blueprint.