From the Agent Counter to Wise: Thirty Years of Money Transfer

In the 1990s, the counter was the product

A cross-border transfer in the early 1990s often began with cash, a paper form and a trip to an agent. The customer paid for access to a physical network as much as for currency conversion. Jamaica provides a neat scale marker for Western Union: its 1990 partnership received money from four countries. Twenty-five years later, the network received transfers from 126.

Swift dates the beginning of its story to 1973. Four years on, its network gave banks a common language for payment instructions. It did not move customer money by itself; correspondent banks held accounts with each other and settled the instructions. The distinction matters because people often call any international bank payment a SWIFT transfer, rather as every vacuum cleaner briefly became a Hoover.

Retail customers rarely saw the plumbing or its price. A sending fee appeared on the receipt, while the exchange-rate margin sat inside the conversion. Intermediary banks could deduct more on the way. Several days could pass before the recipient made the trip to an agent, identification in hand. Physical cash delivery solved a real access problem, though shops, security and ready cash made it costly to run.

The web removed the trip into town

An email address becoming a payment identity was PayPal's small but consequential trick. Its early figures are almost quaint: 12,000 accounts and $235,000 of payment volume by the end of 1999. Six years later, the account count was above 100 million; payment volume had passed $27 billion, about 115,000 times the first-year total. It mostly served commerce, but taught customers to expect money movement on a screen.

Money transfer companies put web forms in front of older networks, which removed the journey to a counter for the sender. The deeper change arrived when providers started collecting and paying out through domestic bank systems. Instead of pushing every transfer along one chain of correspondent accounts, a firm could accept pounds in Britain and release locally held rupees in India. Matching flows and prefunding accounts cut time, although the balance-sheet work stayed backstage.

Once Faster Payments arrived in 2008, the first domestic leg could feel immediate to a UK customer. Wise followed in 2011 with a model built around local transfers and a separately disclosed fee at the mid-market rate. Its founders' original problem was wonderfully ordinary: one earned euros in London while the other needed euros for an Estonian mortgage. The private swap became a regulated payment business.

The phone made poor pricing harder to hide

Smartphones put onboarding, identity checks and tracking into one pocket. Customers could photograph a passport, fund by bank or card and see an estimated arrival time. Then came a convenience: no call to an agent, just a notification in your pocket. Digital firms also made the recipient amount visible before confirmation, which exposed the old zero-fee trick: a transfer can have no separate charge and still carry an expensive exchange rate.

Open banking improved the UK funding step after its 2018 launch. It replaced the familiar fiddle with a bank approval screen: no retyping card details and no copying bank information. Open Banking Limited counted more than 16.5 million live user connections by January 2026, alongside almost 33 million payments in November 2025. Put another way, one month of open-banking payments was equivalent to roughly half the UK's adult population making a payment.

Yet the phone does not erase the last mile. A bank deposit in France can run over connected account systems; cash collected in rural Sierra Leone needs an agent and local liquidity. Compliance rules differ, operating hours do not line up and some currencies are hard to source. The Financial Stability Board still described high cost, low speed and limited transparency as global problems in 2020, not relics from a fax-machine museum.

The transfer improved; the old costs did not vanish

The World Bank now records prices across 377 corridors, while public calculators can refresh in seconds. Our own rate monitor adds the missing receipt: a screenshot of the provider quote with amount, route and time. In the 1990s, a customer could keep the paper slip; today the evidence can be compared across firms and stored before the quote disappears.

Competition also blurred the categories. Banks offer app transfers, remittance brands fund digital wallets and currency brokers provide automated dealing beside a human desk. Some technology firms supply their payment network to banks instead of competing only for the end customer. The visible brand may own the relationship while another regulated company executes part of the payment, so reading the legal footer has become oddly useful.

Progress has limits. The FSB's 2025 review found that policy work had not yet produced clear global gains for end users, and average costs remained sticky even as remittance speed improved. The thinning of the bank network was substantial too. BIS counted about one quarter fewer correspondent-banking relationships in 2020 than in 2011. Fewer links can simplify networks in some places and reduce access in others.

I expect the next chapter to be quieter than the app revolution: payment systems linked directly, better data travelling with each transfer and currency conversion placed nearer settlement. The BIS has tested that shape in Project Rialto. How much arrives? Thirty-five years of technology has made that answer faster to obtain and harder to hide, which may be the most useful change of all.