Moving money across borders has always been hard. Payments must travel across time zones, currencies, legal systems, regulatory regimes, and banking cultures that were built independently of each other over centuries.
For a long time, that complexity was a reasonable explanation for why the experience was as slow, expensive, and opaque as it was. It is becoming a less convincing one.
The electronics supplier in Shenzhen settling invoices with distributors across the Middle East, the platform in Jakarta paying creators from Manila to Mumbai, the trading house in Dubai moving treasury across six currencies before lunch. None of them think of themselves as doing something exotic. They're just doing business.
And yet the payments infrastructure underneath them still behaves as though crossing a border is an unusual event requiring unusual effort. Global commerce has outrun the system built to serve it, and customers are increasingly aware of the gap. Patience for explanations of why it hasn't been fixed are wearing thin.
In 2020, the world's major economies decided to do something about it. The G20 asked the Financial Stability Board to coordinate a global program to make cross-border payments faster, cheaper, more transparent, and more accessible, with the Bank for International Settlements' Committee on Payments and Market Infrastructures doing much of the technical heavy lifting alongside it. The following year, the G20 endorsed quantitative targets with a common deadline of end-2027: remittance costs below 3%, three-quarters of wholesale payments credited within an hour, and full tracking visibility for senders and receivers. Ambitious, and crucially, measurable. For the first time, the industry had a deadline.
Which makes the FSB's consolidated progress report, published in 2025, sobering reading. Five years in, with most of the policy development work now complete, improvements have "not yet translated into tangible improvements for end-users at the global level." The average cost of sending a $200 remittance is still 6.5%, more than twice the target. The share of wholesale payments credited within an hour sits at 54.6%. The FSB's own assessment is that hitting the 2027 targets is unlikely. The reason is not that the policy frameworks are wrong. It's that the real obstacles to cross-border payments modernization are structural, operational, and in many cases deeply human.
1. Correspondent banking is built on relationships, not APIs
The G20 roadmap envisions something resembling multilateral, frictionless connectivity.
For most banks, the reality is a web of bilateral correspondent relationships accumulated over decades. Each with its own legal agreements, compliance frameworks, credit lines, and internal stakeholders. Every hop in that chain also carries its own fees and FX spread, so a payment crossing several correspondents accumulates cost the originating bank can’t fully see or control, eroding margin before the money reaches the beneficiary.
These aren't just technical integrations. They're business relationships. And renegotiating or routing around them involves friction that doesn't show up on a technology roadmap.
Compounding the problem is banks' quiet retreat from correspondent relationships. CPMI and SWIFT data shows active relationships fell 20% between 2011 and 2019, and that trend has only continued to deepen. Cost and compliance are driving this shift.
Maintaining relationships in markets perceived as high-risk has simply become too expensive. Many banks decided it wasn't worth it, with severe knock-on economic impact on those markets.
To break out of this model, banks must make deliberate choices
Solving this challenge requires making explicit choices about which relationships to maintain and which capabilities to stop building in-house.
Most institutions have yet to confront these kinds of choices directly.
That's why a 2025 CPMI-FSB monitoring survey found that jurisdictional implementation of policy recommendations remains nascent across most markets. Not because solutions are unknown, but because the network itself is structurally resistant to the change the roadmap requires.
2. ISO 20022 is being adopted, just not the data inside it
Every major institution running a transformation program has heard or made the ISO 20022 pitch: richer structured data, better compliance screening, improved straight-through processing.
The CBPR+ migration has been a serious, sustained undertaking that has delivered real results. Swift's data confirms that payment speed improves measurably once both parties have completed the migration. And the FSB notes that many fast payment systems have already adopted ISO 20022 for payment messaging.
What gets less airtime is what happens to that data once it enters a multi-hop correspondent chain.
Information that was present when data entered the chain doesn’t survive the journey:
- Legacy processing systems at intermediate banks strip or truncate structured fields rather than rejecting messages they can't handle.
- A purpose code populated at origination arrives blank at the receiving end.
- Beneficiary address data structured to pacs.008 requirements emerges as a single free-text field with characters missing.
The result is a bank can be fully compliant with migration requirements while systematically degrading the data quality that gives the standard its worth. That discrepancy doesn't appear on most institutions' transformation scorecards.
Adopting the format and carrying it forward are two very different things
The industry is considerably further ahead on adoption than it is on carrying what the format is supposed to carry.
But a standard that’s implemented differently is just a shared vocabulary for miscommunication.
Banks have to invest beyond patching legacy systems for compliance and into the innovations needed to get the full value out of ISO 20022. That’s what turns the standard from a box to check into something the business can use. Cleaner transaction data, better inputs for the services customers actually want, and a sharper view of liquidity. The benefits are real, but they sit on the far side of that investment.
3. Visibility is only as good as the least-connected participant
Swift GPI has addressed one of the industry's most persistent frustrations: confirmation of credit to the beneficiary account is now standard on GPI-enabled transactions.
The FSB's 2025 KPI data shows that improvement in wholesale payment speed over the past year was driven almost entirely by faster processing at the beneficiary leg. That's progress worth noting.
The remaining problem is structural: visibility is only as good as the least-connected participant in the chain.
The remaining problem is structural: visibility is only as good as the least-connected participant in the chain.A bank with full GPI capability can still find itself unable to explain to a customer why their payment is delayed. It may be occurring at an intermediate correspondent that lacks GPI connectivity, provides incomplete status data, or simply hasn't responded.
“The payment is in progress” is no longer good enough
When a payment goes quiet mid-chain, operations teams can't determine whether the hold is compliance-related, a technical issue, or simple queue depth.
Without that visibility, the default is a holding message: "the payment is in progress." This is accurate in the narrowest sense, yet operationally useless and frustrating for end users.
The FSB notes it currently lacks a data source for calculating its wholesale transparency KPIs at all. This is itself a measure of how far that work has to go.
4. Compliance friction hits hardest in the corridors that can least absorb it
AML screening, sanctions checking, and beneficiary verification generate delays roughly proportionate to corridor risk. But those delays bear almost no relationship to the operational capacity of the institutions handling them.
Take, for example, capital control processing. This can delay crediting of funds by hours or even days, with verification of transaction purpose among the primary culprits.
The bottom line: the same processes absorbed efficiently by a large institution with automated workflows can bring a smaller bank's operations to a halt.
What normal looks like in the corridors the G20 roadmap cares about most
Remittance flows to markets with limited banking infrastructure. SME cross-border trade. Transactions involving counterparties with thinner documentation trails. This is the environment.
And while the FATF’s revision of Recommendation 16, finalized in June 2025, introduced clearer requirements for beneficiary institutions to run alignment checks, full implementation is not required until end-2030. What’s more, supporting guidance won't be complete until late 2026.
Smaller institutions lack the automated workflows to process compliance checks at scale. The result is corridors that matter most for economic development are where friction is most acute.
5. The people who understand both worlds are very busy right now
Every major bank running a payments modernization program is competing for the same rare skill set: professionals who understand legacy correspondent banking infrastructure deeply, who can also design on modern payment rails, ISO 20022 data architecture, and API-based connectivity.
That combination is genuinely scarce, and every institution running a transformation program is chasing it at the same time.
The talent shortage is only part of the story
In institutions where correspondent banking is a mature, revenue-generating business, a modernization program is not an uncontested good. It is a threat to existing revenue models and the internal influence of the teams who built the current system.
Cooperation requires more than executive sponsorship. It also means managing the commercial interests that point in a different direction.
That is a change management challenge as much as a technical one—and the challenge that most publicly visible progress reports have the least to say about.
Even the G20 says hitting the 2027 targets is unlikely
When setting the program’s goals, the FSB partnered with the Bank for International Settlements' Committee on Payments and Market Infrastructure in 2021. They established 11 targets, including:
- Remittance costs below 3%, with no corridors with costs higher than 5%.
- 75% of wholesale payments credited within an hour of payment initiation.
- Full tracking and transaction cost visibility for senders and receivers.
Five years later, none have come to pass
According to the FSB's own joint 2025 progress report, the average cost of sending a $200 remittance is still 6.5% — more than twice the target. Meanwhile, the share of wholesale payments credited within an hour sits at 54.6%.
In the report's own words: "It's unlikely that satisfactory improvements at the global level will be achieved in line with the 2027 roadmap timetable."
Years of policy work have produced comprehensive frameworks. They have not yet moved the headline numbers. That is worth sitting with.
The deadline is still useful, regardless if anyone meets it
What the deadline is doing, and what makes it valuable regardless, is forcing a specificity the industry previously lacked.
Banks now have a concrete accounting of which capabilities are blocked by technology, which by compliance architecture, which by correspondent network inertia, and which by the absence of people capable of executing the work.
The stakes in that accounting aren't only regulatory. While the roadmap's been running, a generation of faster-moving entrants, many now holding their own banking licenses, has been building on modern rails from the start and winning corridors and flows one at a time. For an incumbent, every quarter spent maintaining infrastructure it's outgrown is a quarter in which the real question shifts, from whether it'll hit the 2027 targets to whether it'll still own the customer relationships those flows represent.
The institutions making the most visible progress have already drawn one line clearly. They've decided which capabilities are genuinely differentiating and worth owning, and which are better accessed through partners who've already built them. That single decision, more than any technology choice, is what frees a modernization team to spend its energy on what it wants to build next rather than on maintaining infrastructure that no longer earns its keep.
At Nium, we work with banks and financial institutions around the world, making exactly these calls about what to build and what to source. You can see how we support that decision at nium.com/solutions/banks.
