The Fastest Part Is Not the Whole Payment
Cross-border payments are the subject of this series: the movement of money between people, businesses, banks and payment providers in different jurisdictions. The opening question is simple but easily missed: when one domestic payment rail can move money almost immediately, what still has to happen before the payment is genuinely complete across a border? By the end of this series, readers should be able to tell the difference between a fast local transfer and a fast, reliable cross-border outcome—and to judge what a new payment connection or policy proposal actually changes.
That distinction matters because “instant” is often a description of one leg of a journey. A sender may see a quick confirmation in an app while another institution, payment system or jurisdiction is still waiting to process the other leg. The payment may need to cross a settlement window, use an intermediary, translate information into a domestic format, or satisfy rules that differ between the sending and receiving sides. The visible button is quick; the system behind it may not be.
A cross-border payment is a chain of responsibilities
A useful starting point is to stop picturing a cross-border payment as one network. It is better understood as a chain of responsibilities. The payer gives an instruction to a payment service provider (PSP), such as a bank or another authorised provider. A domestic payment system processes or settles that instruction. From there, the payment may rely on a correspondent bank, an interlink between payment systems, a shared platform, or another arrangement that carries it toward the receiving jurisdiction. A receiving PSP then makes the result available to the payee under the conditions of that market.
The exact path varies. Some systems are designed for fast retail payments, commonly called fast payment systems or FPS. Others are real-time gross settlement systems, or RTGS systems, which are important for settling obligations between financial institutions and can also support retail-payment settlement. These systems are not interchangeable, and neither one automatically provides a route into another jurisdiction. A domestic system can be modern, secure and fast while remaining primarily domestic in reach.
The distinction is more than technical. The G20 Roadmap for enhancing cross-border payments aims to make these payments cheaper, faster, more inclusive and more transparent. Those goals support international trade and e-commerce, make it easier for people to receive remittances, and can broaden access to financial services. They also reinforce one another. A payment that arrives quickly but carries an unexpected charge is not fully transparent. A low-cost service that is available only to large institutions is not inclusive. A system that settles quickly but cannot preserve the information needed for screening or recipient verification may create new operational and safety problems.
The stakes are therefore wider than convenience. The BIS Committee on Payments and Market Infrastructures describes inefficient cross-border payments as a source of fragmentation in the global financial system. The 2025 monitoring survey covered 82 jurisdictions and 197 operational payment systems. Its breadth is useful because it shows that the problem is not a single outdated product waiting to be replaced. It is a collection of domestic infrastructures, commercial relationships and public rules that must work together often enough for a particular corridor to be useful.
Why local speed does not automatically cross the border
Consider two statements that can both be true:
- A domestic FPS may accept and process a payment very quickly, sometimes in seconds.
- The cross-border payment involving that FPS is delayed, costly or unavailable.
There is no contradiction. The first statement concerns what happens inside one system. The second concerns whether the next system can receive the payment, whether the relevant institutions are permitted to participate, whether the systems are open at the same time, and whether the payment data can be understood without manual repair.
The survey makes this boundary visible. In 2025, only 47% of the FPS covered by the survey and 48% of its RTGS systems were reported as capable of processing cross-border payments. Even among those systems, restrictions were common. They could concern the initiation of outbound payments or the conversion of inbound messages into domestic formats. In other words, having a fast rail is not the same as having a cross-border service. The rail is a necessary component in many designs, but it is not the completed corridor.
The same point appears in the Federal Reserve's 2026 proposal concerning FedNow. The Board invited public comment on a proposal that would allow U.S. banks and credit unions to use intermediaries to transfer funds through the FedNow Service. The release gives a cross-border example: a U.S. bank could use FedNow to transact with a correspondent bank that facilitates the international portion of a cross-border payment. Under the arrangement described in the release, a FedNow transfer can currently include only two U.S. banks.
That proposal is significant precisely because it identifies a boundary rather than pretending the boundary does not exist. It would address how participants can use an intermediary around a domestic instant-payment service. It does not, on the evidence supplied in the release, establish a globally connected FedNow network, settle a foreign currency, harmonise every message standard, or remove the legal and operational work required on the international portion. It is a proposal open for comment, not a demonstrated end-to-end result.
The practical lesson is to ask where the word “instant” stops. Does it stop when the sending PSP accepts the instruction? When the domestic rail processes it? When an intermediary receives it? When settlement becomes final? Or when the payee can actually use the funds? Those moments may coincide in a well-designed corridor, but they should not be assumed to coincide everywhere.
The diagram is deliberately less tidy than a single arrow from sender to recipient. The border is a decision point. A payment can continue through a connection or intermediary only if the surrounding conditions allow it. If they do not, speed on the first domestic leg cannot compensate for the missing condition.
The five conditions behind whole-corridor speed
The rest of the subject can be organised around five connected conditions. They are not five competing technologies. They are five questions about whether the chain can operate.
Time. Are the systems that must settle or pass on the payment open at compatible times? FPS commonly operate around the clock, but RTGS operating hours are more restricted in many jurisdictions. A time-zone gap can force a payment to wait, require liquidity to be positioned in advance, or concentrate activity into a narrow window. Extending hours can improve overlap, but it also brings staffing, technology and liquidity costs. The first question is therefore not simply “How fast is the rail?” but “When can the relevant parts of the corridor actually settle?”
Access. Who is allowed to initiate, clear or settle a payment? Direct access lets an entity participate independently, typically with its own settlement relationship. Indirect access means relying on a direct participant. Technical access without a settlement account is another possibility. These choices affect the number of handoffs, funding dependence, competition and risk. A rail that is technically available but closed to a foreign bank or non-bank PSP may be fast for some participants and unreachable for others.
Connection. What arrangement joins the two jurisdictions? A correspondent-banking relationship, a bilateral FPS link, a single access point, a common platform and a hub-and-spoke model are different ways of creating a path. Each brings its own participants, responsibilities and governance. A connection is not just a cable between systems; it is an agreement about how instructions, settlement and accountability cross the boundary.
Data. Can every participant interpret the payment information and preserve what the next participant needs? ISO 20022 can carry richer and more structured information, and APIs can automate exchanges between systems and providers. But adopting a standard or an interface does not guarantee common interpretation. Data can still be truncated or translated inconsistently. A payment that moves quickly but arrives with incomplete or ambiguous information may be delayed for repair or rejected altogether.
Permission. Are the access, data-sharing and oversight arrangements legally and operationally acceptable in both jurisdictions? Cross-border payment systems involve rules for PSPs, privacy, anti-money-laundering controls, supervision and the responsibilities of linked systems. These rules do not sit outside the infrastructure; they determine which connections can operate and under what safeguards. Wider access and more data sharing may improve competition and transparency, but they require clear accountability and risk controls.
These conditions explain why progress tends to be incremental. The BIS survey reports movement toward longer RTGS hours, wider access, more FPS interlinking, broader use of ISO 20022 and greater use of APIs. Yet it also reports uneven regional progress, continuing restrictions and many plans that remain exploratory. The G20 targets are unlikely to be fully met by 2027, according to the brief, even though jurisdictions can improve services one step at a time.
That is not a reason to dismiss faster rails. It is a reason to evaluate them accurately. A domestic instant-payment system can be the foundation for a better cross-border service. A well-designed intermediary arrangement can connect institutions that could not previously use the rail in the same way. But the benefit appears only when the entire corridor—its clocks, participants, connection, data and rules—can carry the payment to its intended recipient. The next question is where time enters that chain, and why a closed settlement window can make an instant instruction wait.