Why Cross-Border Payments Are Losing Their Economic Advantage

Cross-border payment friction is becoming easier to measure and avoid. The World Bank’s third-quarter 2025 data put the average cost of sending a remittance at 6.36% of the amount sent. This figure covers retail remittances, not large bank transfers, but it shows how much value can be lost between a sender and a receiver.

Wholesale payments are changing through a different route. In March 2026, Swift said it was working with more than 40 financial firms on a blockchain-based ledger for 24/7 cross-border payments. In May 2026, the BIS said Project Agorá had shown atomic settlement across several currencies with tokenized central-bank money and bank deposits.

Banks are not leaving the system. But delay, repair work, prefunding, and long routing chains are losing value as sources of control.

Most payment systems serve one country and one currency. Correspondent banks link them by holding accounts for other banks. A payment message moves across the network, while the cash balance changes across a separate set of ledgers.

This model creates several forms of income. Banks can charge transfer fees, earn an FX spread, and bill for local clearing, checks, or account access. Some may also gain value from balances held in the network.

These income sources are not the same, yet they rest on one base: control of a broken path. The more steps a payment needs, the more firms can charge for access, trust, and repair.

The path also creates a cost for bank balance sheets. Banks often place cash in foreign accounts before clients send payments. The BIS says these idle buffers can cut delay and failure risk, but they also raise funding costs or create credit risk elsewhere.

Prefunding is not a payment fee. It is a cost built into the service.

Programmable settlement can join more of the process. The payment order, checks, FX terms, and asset move can share one system state. Code can confirm that all rules have been met, then atomic settlement moves both sides at once or neither side moves.

This cuts the gap between a message and the final cash move. It also reduces manual repair and ledger checks. The edge comes from tighter links between steps, not from removing all banks.

Legacy networks still have strong assets. They connect banks across many markets and support local cash access, legal claims, sanctions checks, and error handling. These tasks carry real cost because they solve real risks.

Their weak point is the handoff. Each bank keeps its own ledger, runs its own checks, and may use a different data format. Local payment hours may not match, so cash often has to sit in place before it is needed.

Programmable rails use code to line up the trade before funds move. Shared data can reduce repeat checks and back-office work. A system that runs all day can also free cash sooner, giving firms more use of the same working capital.

Autonomous rails do not remove risk. Stablecoins add reserve, issuer, legal, and cash-out risk. Public chains can have thin liquidity or uneven fees, while private ledgers can protect bank control but rebuild closed access.

The stronger model will mix both sides. It will use trusted money and clear law, along with shared data, code, and near-constant settlement. The key issue is who owns that combined stack.

Capital moves first where the cost gap is easy to prove. Large firms will favor paths that need less prefunding, show clear status, and release cash sooner. Payment firms will have to explain each fee.

The weakest link is the firm that only passes a message. That role is easier to replace than local clearing, credit, deep FX liquidity, or compliance. Regional processors may also lose margin when direct routes reduce the value of access rented from larger banks.

The fee pool will not disappear. It will move toward tasks that remain scarce: identity, custody, local currency access, sanctions controls, dispute work, liquidity, and links to central-bank money. Firms that own these points can keep pricing power.

The same split applies to retail and wholesale flows. Retail users care about the fee, speed, and final cash-out. Large firms care more about liquidity, proof of finality, and balance-sheet use. Both markets reward shorter paths, but they do not share the same risk model.

Capital will choose the rail that joins low friction with legal certainty. Speed alone is not enough. The system must move trusted money and reach the final local account.

Cross-border payments once earned margin from distance, delay, and closed access. Programmable settlement turns those traits into design choices, not fixed costs. Banks can still control the market, but they must own the trusted money, the local endpoint, the rules, or the new rail. The old edge was the length of the chain; the new edge is control of the few links that still matter.

Sources: World Bank, Swift, Bank for International Settlements