The shrinking correspondent network, in numbers

The infrastructure most cross-border payments depend on has been contracting for as long as anyone has measured it. BIS CPMI data, built from payment message activity across more than 200 jurisdictions, shows the number of active correspondent banking relationships fell about 22% between 2011 and 2019 and roughly 30% by end-2022, with a 4% drop in 2022 alone. There is no year in the series where the network grew.

This page collects the numbers. All figures are from the sources listed below, as of the CPMI’s May 2023 chartpack (data to end-2022) and subsequent regional reports; we will refresh it as new data is published.

What the data actually measures

The CPMI series is worth understanding before quoting, because it is the only global measurement of the network’s size. It is built from monthly SWIFT payment message data. An “active correspondent” is a bank that sent or received at least one message over a corridor in the year; an “active corridor” is a country pair with at least one such message. The measure therefore captures the network’s breadth, who can still pay whom directly, rather than its capacity.

That definition makes the decline more striking. A relationship drops out of the data only when not a single payment message crosses it in a year. The roughly 30% of relationships that disappeared did not slow down; they went to zero.

The headline numbers

• Active correspondent relationships, 2011-2019: down about 22% (BIS CPMI commentary, August 2020).
• 2011-2022: down roughly 30%, including a 4% decline in 2022 (BIS CPMI chartpack, May 2023).
• Payment traffic: up. The value and volume of messages flowing over the network grew across the same period. Fewer pipes, more water.
• Corridors: fewer. Active country-to-country corridors declined alongside relationships, meaning some routes now have no direct path at all.

Where the decline is worst

The global average conceals the real story, which is regional. Declines in active correspondents, 2011 to 2022:

Region	Decline
Melanesia	~63%
Polynesia	~54%
Caribbean	~52%
Small island developing states (country average)	~41%
Advanced economies (country average)	~23%
World	~30%

The Pacific Islands Forum’s 2024 report puts the Pacific’s loss at roughly 60% since 2011, about double the world average, and documents banks in some island states operating with a single remaining relationship. The places losing coverage fastest are the places with the fewest alternatives.

Fewer banks, more traffic

Because traffic grew while relationships fell, the surviving correspondents each carry a larger share of the world’s payments. That concentration has three consequences.

First, resilience: coverage for entire regions now depends on a short list of institutions, and each further exit removes a larger fraction of what remains. The Pacific report documents jurisdictions down to a single USD clearing relationship, which turns one bank’s risk decision into a country’s payment infrastructure question.

Second, pricing: corridors with one or two remaining channels price like monopolies, which is part of why remittance costs stay furthest above target exactly where de-risking bit deepest. The World Bank’s global average masks corridor costs several points higher in thinly served routes.

Third, visibility: banks that lost direct relationships reach the system through nested arrangements, adding intermediaries, fees, and opacity to every payment from those corridors. Each extra hop also adds a cut-off time and a compliance stop, so the corridors with the least coverage also get the slowest, most expensive service the system offers, a compounding of the hidden costs that better-served corridors never see.

Why the exits happen at all is an economics story, told in full in our explainer on de-risking: fixed compliance costs against thin corridor revenue, rational for each bank, corrosive for the system.

The official response

The decline is one reason improving cross-border payments became a G20 priority in 2020. Five years in, the FSB’s October 2025 progress report concluded the end-2027 targets are unlikely to be met at the global level, and the BIS’s own researchers have sketched what they call next generation correspondent banking, an upgrade agenda for the model rather than a replacement. The scorecard on those targets is tracked in our G20 targets page, and the full map of what is being built alongside the correspondent system, from interlinked instant payment systems to consortium ledgers and stablecoin rails, is in alternatives to correspondent banking.

The through-line in the data: the correspondent network is not collapsing, it is concentrating, and it is quietly withdrawing from the edges of the map. The corridors it leaves behind are the clearest market for settlement that does not depend on a chain of willing intermediaries.

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.

A shrinking network changes the question an institution has to answer: from “which correspondent do we use” to “which rail still reaches this counterparty, and on what terms.” Frame is rail-neutral, routing each payment across whichever rail fits the corridor, the counterparty, and the policy that governs it, a fiat network, a regulated stablecoin, or a tokenized deposit. Compliance is enforced inside settlement by Frame’s Rules Engine, and every settled transaction produces verifiable evidence that its conditions were met. For banks, payment providers, exchanges, platforms, and enterprises in corridors the correspondent map no longer serves well, that is a path that does not depend on the network growing back.

See how a rail-neutral settlement layer works in the Frame Blueprint.