Charted: Instant Domestic Rails Still Carry Only ~9% of Retail Cross-Border Value

Domestic instant rails feel ubiquitous. PIX clears hundreds of millions of Brazilian payments a day. UPI dominates Indian P2P. FedNow, Faster Payments, TIPS, PayNow, and PromptPay keep expanding operating hours and participant lists. The marketing line writes itself: instant at home, therefore instant abroad.

The ledger does not cooperate. On a desk estimate of global retail cross-border value — remittances, person-to-person, and small merchant receipts, not wholesale FX or securities settlement — about 74% still clears on correspondent banking chains that message over SWIFT (or equivalent) and settle through nostro/vostro accounts. FPS interlinks and multi-jurisdictional fast payment systems account for roughly 9%. Closed-loop nonbanks and card-acquired retail paths take about 14%. A residual 3% covers crypto on/off-ramps and miscellaneous channels.

That is the core gap this dashboard is built to hold: system capability ≠ value share. The BIS CPMI 2025 monitoring survey finds that 47% of fast payment systems and 48% of RTGS systems can process a cross-border leg [BIS CPMI 2025]. Among FPS that already enable cross-border payments, 43% used interlinking arrangements in the 2024 edition (nine of twenty); by 2025 the interlinked count held at 13 of 27 capable FPS — roughly half [BIS CPMI 2024]. Those are infrastructure facts. They do not mean nearly half of retail FX value now rides PIX-to-PayNow style pipes.

Toggle Rail mix timeline, Corridor share, Regional stack, Speed vs cost, System arrangements, and Domestic FPS lens. Switch Value share / Cost / Speed and filter by region. The stacked areas show correspondent share grinding down from the mid-80s in 2019 toward a 66% 2027 carry — still a majority.

Scoreboard: rails vs capability

Read the left column as a value story and the capability rows as a plumbing story. Mixing them produces the false headline that “instant rails have already replaced SWIFT for retail.”

Correspondent banking still clears most of the money

Correspondent chains remain the workhorse because they are currency-complete and corridor-complete. A US dollar remittance into a thin African corridor, a euro invoice into a Balkan account, or a sterling payout into a jurisdiction without an FPS link still needs a bank that holds the foreign currency and knows the beneficiary’s bank. SWIFT messaging is the coordination layer for that network; the settlement is the nostro account, not the message.

Closed-loop providers (digital remittance apps, card-funded wallets) have taken a visible bite of customer experience — especially US→Mexico and US→India — without always exiting correspondent banking underneath. Many “instant” consumer receipts are funded by a nonbank that still settles net through banks. That is why the dashboard separates closed-loop (~14% of value) from FPS interlink (~9%). Instant to the user is not the same as instant on a public FPS-to-FPS rail.

The timeline panel’s slow decline in correspondent share — roughly 86% → 74% from 2019 to 2025 on the desk series — is real progress and still a majority story. Even the 2027 carry at 66% assumes continued Asia-Pacific link densification and European SCT Inst / TIPS reach. It does not assume FedNow becomes a cross-border hub.

Instant rails win corridors, not the globe

Where FPS–FPS links are live, value share can flip. Desk corridor cuts put India → Singapore and euro-area retail cross-border near or above 40% FPS share of corridor value, with correspondent compressed [BIS FPS Interlinking]. Singapore → Thailand, Australia ↔ NZ, and selected GCC→South Asia routes show double-digit FPS shares. Those corridors prove the product: seconds-to-minutes credit, lower disclosed fees, fewer intermediary layers.

They do not generalize. Sub-Saharan corridors still print ~72% correspondent on the desk ladder, with high remittance costs (~7.8%) and multi-day medians. US → Euro area and China → ASEAN remain correspondent-heavy because the bilateral FPS fabric is thin or absent. US → Mexico is a closed-loop story more than an FPS-interlink story: big nonbank share, limited public FPS-to-FPS path.

Asia-Pacific leads on interlinking in CPMI surveys; the Americas lag on average in planning many priority enhancements [BIS CPMI 2025]. That regional asymmetry shows up in the Regional stack panel: APAC FPS value share ~18%, Americas ~7%, Africa & Middle East ~5%, with correspondent still 55–78% everywhere (desk cuts).

Capability surveys measure pipes, not flows

CPMI’s monitoring surveys ask central banks what their payment systems can do and how cross-border legs are arranged. By 2025, nearly half of FPS and RTGS respondents can process a cross-border payment. For RTGS, correspondent banking remains the primary method for more than half of systems that handle inbound/outbound legs. For FPS, interlinking is rising — especially in Asia-Pacific — and a smaller slice is multi-jurisdictional by design.

That arrangement pie (dashboard System arrangements) is essential and easy to misuse. A jurisdiction can report “FPS capable of cross-border” while 25%+ of that capable set still routes the foreign leg solely through participants’ correspondent relationships. Capability includes “we accept an outbound payment that a bank will still send via SWIFT.” It is not a census of PIX-style foreign credits.

World Bank Remittance Prices Worldwide and the FSB's G20 KPI reports frame cost and access, not rail taxonomy. The global average cost of sending a $200 remittance sat at 6.5% in 2025 — above 6% every year since tracking began — against the G20 3% retail target [FSB Progress 2025; World Bank RPW]. Speed distributions still show a fat tail of multi-day correspondent paths even as linked FPS corridors deliver minutes. The Speed vs cost scatter puts live FPS paths in the lower-left (cheaper, faster) and correspondent paths up and right — but bubble sizes remind you which paths still carry the value.

Domestic giants are not automatic cross-border giants

PIX, UPI, FedNow, and peers are domestic success stories first. The Domestic FPS lens indexes 2025 volume against a rough outbound touch share — the fraction of that jurisdiction’s outbound retail remittances that actually touch the FPS as a cross-border leg.

UPI and PayNow show meaningful outbound touch where links exist. FedNow’s domestic index is still early and its cross-border status is effectively domestic-only for retail value. PIX’s international work is better read as pilot / expanding than as a global value shift. Faster Payments and SPEI often appear as correspondent overlays: domestic instant credit after a bank has already moved value across the border the old way.

Treating domestic TPS (transactions per second) fame as proof of cross-border rail share is the most common error in this file. Domestic instant adoption is necessary for interlinking; it is not sufficient for global value migration.

What would actually move the 74%

Three conditions have to stack before correspondent share falls much faster:

1. Dense bilateral or multilateral FPS links on high-value remittance corridorsnot demos, live with FX conversion and dispute rules.
2. Extended RTGS/FPS hours and nonbank access so the last mile is not forced back into weekday correspondent windows (CPMI’s operating-hours and access workstreams).
3. ISO 20022 + API harmonisation that makes data travel with the payment instead of repairing it in investigating messages.

Absent those, closed-loop apps will keep winning UX share while still leasing correspondent capacity underneath, and the headline rail share will move in single digits per year — exactly the slope on the timeline panel.

Caveats (read before you trade the 74%)

• Value shares are desk estimates. CPMI discloses system capability and arrangement type; it does not publish a unified global retail-FX value ledger by rail. The 74 / 9 / 14 / 3 split is an order-of-magnitude synthesis consistent with survey direction, public remittance volumes, and corridor case studiesnot a SWIFT or CPMI extract.
• Retail ≠ wholesale. This post excludes large-value RTGS wholesale, securities settlement, and bank-to-bank funding. Correspondent dominance is even stickier in wholesale.
• Closed-loop vs FPS. Consumer “instant” often means a nonbank credit promise. We allocate those flows to closed-loop unless a public FPS–FPS or multi-jurisdiction FPS path is the clearing rail.
• Survey sample drift. CPMI 2024 (68 jurisdictions) and 2025 (82) samples differ; percentages are not a perfect panel.
• 2026–2027 points are carried. They assume continued APAC link growth and European instant reach, not a US FPS cross-border breakthrough.