When a Payment Has Arrived but Has Not Been Delivered
Cross-border payments are not finished when the beneficiary bank receives the payment instruction. This series examines the last-mile execution of a cross-border payment—the interval between receipt at the beneficiary bank and credit to the end-customer—to explain why funds can remain unavailable and how to judge proposed fixes. The central question is simple to state but easy to miss: what still has to happen after the payment reaches the receiving bank?
The answer matters because “arrived” can describe two different events. A message may have travelled successfully through the international banking network, while the beneficiary still cannot use the money. That is not necessarily a contradiction, a failed network, or evidence of misconduct. It may mean that the receiving bank has entered a local process involving information checks, currency and account rules, reporting duties, operating hours, or manual review. Understanding that distinction is the starting point for understanding why cross-border payments can be slower than domestic ones.
Delivery has two meanings at the receiving end
Imagine a company waiting for payment against an overseas invoice, or a household expecting money sent by a family member working abroad. From the sender’s perspective, the transfer may look complete once their bank confirms that it has been sent. From the international network’s perspective, the payment instruction may be traceable and may have reached the bank that serves the recipient. From the recipient’s perspective, however, the only decisive event is usually that the funds are credited and available in the account.
Those events can occur at different times. A beneficiary bank may receive the message and the associated settlement information but still need to establish whether the payment can be credited under local rules. The bank may need to verify who the beneficiary is, what activity or purpose the payment represents, whether the account is eligible to receive foreign funds, and whether the supporting information is complete. A payment that is perfectly ordinary to the sender can therefore become a pending item on the receiving side.
This distinction is especially important because the international part of a payment is often discussed as though it were a single pipe. It is better understood as a chain with different responsibilities. One part initiates the transaction. Another transmits instructions and coordinates movement between institutions. Another completes the local checks and makes the funds usable. A fast first or middle part cannot, by itself, guarantee a fast final part.
Domestic payments often hide this division because the participants share a currency, a legal setting, operating conventions, and a common settlement infrastructure. A cross-border payment has to bridge those differences. The sender’s bank and the recipient’s bank may use different account identifiers, data formats, risk procedures, and business hours. Even when both institutions are technically capable of processing a payment quickly, they may not have a common way to express the information needed at the next stage. The result is not always a dramatic failure. More often, it is a small incompatibility or an extra confirmation that adds time at the point where the customer experiences the delay.
The distinction also prevents an unfair conclusion about regulation. Checks can delay a payment, but that does not mean the checks are pointless. Capital-flow restrictions may serve macroeconomic purposes in some circumstances. Anti-money-laundering and countering-the-financing-of-terrorism controls protect the integrity of the financial system. Tax and reporting rules may be legally required. The practical problem is more precise: when necessary checks rely on incomplete data, incompatible systems, repeated manual work, or narrow operating windows, a legitimate control can create an unnecessarily long wait.
Three legs, not one transfer
The IMF working paper supplied for this series separates a typical cross-border payment through correspondent banks into three legs. The originator leg runs from the payer’s initiation of the payment to the originating bank’s submission of it to the messaging network. The in-flight leg runs from that submission until the beneficiary bank receives the payment, possibly through one or more intermediary banks or a payment-market infrastructure. The beneficiary leg begins at receipt and ends when the funds are credited to the end-customer.
That last definition changes where we look for an explanation. If a payment is slow on the in-flight leg, the problem may involve routing, messaging, an intermediary, or settlement between institutions. If it is slow on the beneficiary leg, the international message has already reached the receiving bank; the remaining work belongs largely to the beneficiary-side process and the local environment in which that bank operates. The same payment can therefore be “fast” according to a network timestamp and “slow” according to the customer’s experience.
The receiving bank is not simply a mailbox at the end of the route. In the traditional correspondent model, banks maintain relationships that let one institution provide services in a jurisdiction where another has no physical presence. Accounts held between correspondent and respondent banks, several rounds of communication, and sometimes multiple intermediaries allow the payment to cross borders without physically moving every underlying asset. They also create more points at which currencies, records, operating schedules, and compliance responsibilities have to line up. Once the payment reaches the beneficiary bank, that institution still has to connect the incoming transaction to its own customer account and local obligations.
The IMF study measures this beneficiary-leg interval using the Unique End-to-End Transaction Reference associated with Swift’s Global Payments Innovation tracking. That makes it possible to distinguish the time before receipt from the time after receipt. The measure is not a general impression that cross-border payments feel slow. It is a specific elapsed-time concept: how long a completed payment takes to move from beneficiary-bank receipt to end-customer credit.
A simplified map looks like this:
The map is deliberately modest. It does not claim that every payment passes through the same number of intermediaries or the same checks. It shows why receipt is a boundary: before it, the payment is travelling through a cross-border chain; after it, the receiving institution must turn an incoming transaction into a locally acceptable account credit.
A tracking record can illuminate this boundary without resolving it. End-to-end timestamps can show that a payment spent little time in transit but a long time before final credit. They cannot, by themselves, say whether the delay came from a missing document, an account rule, a sanctions alert, a batch window, or an internal exception. That requires information about the receiving process and the rule being applied. In other words, better visibility is valuable because it locates the interval; it is not the same as a diagnosis of the interval.
What the receiving bank may still need to establish
For inflows, the IMF paper describes checks that can begin or be completed after the beneficiary bank has received the payment. They can include verifying the beneficiary’s information—for example, the type of beneficiary, their activity, or the accuracy of a tax identity number. The bank may need to verify the purpose of the payment, whether expressed as a code or in text, and examine supporting documents such as a contract, invoice, or export bill. Some jurisdictions add amount limits, restrict the account types eligible for foreign transfers, or require the beneficiary bank to report transfers to a local regulator.
These requirements create dependencies. If the payment data already contain the information the bank needs, and the customer and transaction are familiar, processing may take minutes. If a purpose is unclear, a document is missing, an account is not obviously eligible, or a human reviewer must resolve an exception, the same stage can take much longer. The paper notes that incomplete or missing information can extend processing to several days. The time is not necessarily spent moving money between countries; it may be spent establishing that the money can be released within the receiving country’s rules.
Other frictions can surround the checks. Banks and market infrastructures may operate on different schedules, use batch processing, or close during local hours and holidays. Currency conversion can introduce another institutional handoff. A correspondent chain may carry multiple messages and create more opportunities for reconciliation or repeated screening. These factors can coexist. A payment waiting for documentation may also miss a processing window, and a payment with complete documentation may still encounter a system that cannot exchange the relevant data automatically.
It is important not to collapse all of this into “capital controls.” The IMF paper studies the relationship between a country-level measure of financial-account restrictions and beneficiary elapsed time, but it cannot identify which individual payment was delayed by which individual rule. AML/CFT screening, sanctions checks, tax reporting, capital-flow management, ordinary fraud controls, and internal bank procedures may overlap in practice. The useful diagnosis is therefore not that every delay has one cause. It is that the beneficiary leg is a real operational stage where local requirements and system design can determine when an otherwise successful cross-border payment becomes usable.
The stakes are unevenly distributed. Wholesale payments between financial institutions are often large, lower in volume, and made between recurring counterparties. They have historically been served relatively well by correspondent banking and institutional settlement arrangements, even though they are not frictionless. Retail transfers and remittances are usually smaller and higher in volume. Retail payments may be occasional, while remittances tend to recur between the same people but can involve cash-in and cash-out through users with limited access to formal financial services. Fixed compliance work, foreign-exchange charges, beneficiary fees, and an expensive first or last leg weigh more heavily on these users.
The BIS review reports that the global average cost of sending $200 in remittances was $12, or 6 percent, in 2024—an improvement from $18, or 9 percent, roughly two decades earlier. But retail transfers can still take several days, and the cost remains meaningful for someone sending a modest amount. A delay is not merely an abstract performance metric when the recipient is waiting to pay a bill, receive wages, or meet a household need.
The broader lesson is that cross-border speed is not only a question of how quickly a message can travel. It is a question of whether the institutions on both sides can understand the same payment, apply their respective rules without unnecessary repetition, and coordinate the final credit. New technology may improve tracking, data quality, or automation. It cannot by itself decide whether a beneficiary, purpose, account, amount, or document satisfies local law. The problem this series follows is therefore narrower and more useful than “international payments are slow”: after the beneficiary bank receives the payment, what institutional and interoperability frictions stand between receipt and usable funds?