At a glance
Moving money across borders is not one operation. It is a stack of five: collect funds locally, hold them safely in custody, convert currency, clear across payment rails, and pay out locally in the destination market. Every platform that touches customer money in more than one country runs all five, whether or not it has ever named them.
Key distinction: Most platforms assemble the five layers from separate providers, a bank in one market, an electronic money institution in another, an FX broker, a payout partner. That patchwork is what fragments reconciliation, compliance, and settlement timing. The alternative is to source the whole stack from a single specialist correspondent.
This article names the stack, explains why it usually fragments, and walks through each of the five layers. It is written for founders, operations, and treasury leads who know they need to move money internationally but have not yet mapped the full surface they are responsible for.
What moving money across borders actually involves
When a platform collects a payment in one country and pays out in another, the money does not travel. What moves is a sequence of debits and credits across accounts held by different institutions, in different currencies, on different rails.
To make that sequence work, the platform has to solve five distinct problems: collect the incoming funds on a local rail, hold them while they are in flight, convert them if the payout currency differs, clear the transfer across the settlement infrastructure connecting the two ends, and pay out locally on a rail the recipient's bank recognises.
Each is a separate capability with its own licensing, counterparties, and failure modes. A platform can outsource any of them, but it cannot skip any. The five together are the money-movement stack, and understanding them as a stack is the first step to controlling it.
Why the stack fragments into a patchwork
Almost no single provider offers all five layers well, so platforms buy them piecemeal. Collection comes from one acquirer or virtual IBAN provider. Custody sits at a bank. FX runs through a broker. Clearing depends on whichever correspondent chain the banks happen to use. Payouts come from a local disbursement partner in each market.
The patchwork compounds with every new market, and the reasons it persists are structural, not accidental:
- Licensing is jurisdictional. The permission to hold or move funds is granted market by market, so no single bilateral relationship covers a platform's whole footprint.
- Banks retreat from cross-border. The number of active correspondent banking relationships fell about 20% between 2011 and 2018, according to the BIS, pushing platforms toward a wider spread of smaller providers.
- Each layer optimises for itself. An FX broker wants spread; a payout partner wants volume; a bank wants the deposit. None is accountable for the end-to-end journey.
The result is a chain of counterparties where no one owns the outcome. When a payment is late or a reconciliation breaks, the platform is the only party that can see the whole picture, and the only one on the hook for it.
Layer one: collect
Collect is how funds enter the platform. In practice this means a local receiving account, often a virtual IBAN, that lets customers pay in on a domestic rail rather than an expensive international wire. The quality of this layer is measured by how many markets it reaches and how cleanly the incoming funds are attributed to the right end user.
Layer two: hold
Hold is custody: where the money sits between collection and payout. This is the layer platforms most often get wrong, because the default option is a pooled account where every customer's balance is commingled under the platform's name. The distinction that matters is between named and pooled accounts. In a pooled model, the platform's internal ledger is the only record of who owns what, and if that ledger and the bank balance ever diverge, end users are exposed.
Custody is also where regulatory obligation concentrates: client-fund segregation, keeping customer money separate from the platform's own operating funds, lives here. For crypto and stablecoin businesses, the same problem shows up as the need for named on-ramp accounts rather than a shared float.
Layer three: convert currency
Convert is the FX layer. If a platform collects in euros and pays out in dirhams, someone has to price and execute the conversion, and that price is rarely the mid-market rate the platform sees on a screen.
FX is where margin quietly leaks. A broker's spread, a bank's markup, and a poor execution window each take a cut, and because the conversion is bundled into the payment, the true cost is hard to isolate. Platforms running a separate FX relationship also carry timing risk: the rate is struck at one moment, but settlement lands at another. The layer is not just about price, it is about certainty of execution at the moment the payment needs to move, which is why FX sits closer to clearing than platforms usually assume.
Layer four: clear
Clear is the settlement engine that connects the two ends of a payment. It is the least visible layer and the one platforms understand least, because it happens inside the correspondent banking system rather than on the platform's own screens.
Clearing is a chain of nostro and vostro accounts: a bank holds funds at a correspondent, which holds funds at another, until the value reaches a bank that can settle locally. Every hop adds a counterparty, a fee, and a delay, and the payment moves only as fast as the slowest balance sheet in the chain.
The core problem here is not speed but certainty. What platforms need is settlement certainty, not speed: the confidence that a payment initiated today lands, in full, on a known date. The correspondent model struggles to deliver that because its economics reward holding funds, an incentive mismatch between the intermediary and the platform relying on it.
Layer five: pay out
Pay out is the mirror of collection: releasing funds on a local rail the recipient's bank recognises, whether that is SEPA in Europe, Faster Payments in the UK, ACH in the US, or the IPP in the UAE. A wire that technically arrives but sits in a correspondent queue for three days is a failed payout in every way that matters to the recipient.
Payout quality is defined by local reach and by finality: funds that settle on a domestic rail, in the recipient's name, with a clear record that the obligation is discharged. For platforms handling disbursements on behalf of clients, this is also where COBO and POBO structures are implemented, and where segregation has to survive all the way to the final credit.
Build vs. partner vs. one counterparty
A platform has three ways to assemble the stack. It can build each layer itself, negotiating direct banking, FX, and payout relationships in every market. It can partner layer by layer, stitching together a patchwork of specialist providers. Or it can source the whole surface from a single specialist correspondent that operates all five layers on one integration.
The build route offers the most control and costs the most: every market is a new banking relationship, integration, and compliance review. The partner route is faster to stand up but is the patchwork, distributed fragility across many counterparties, none of whom owns the outcome. Consolidating onto a single clearing partner trades many thin relationships for one deep one. Provider comparisons for the account, payout, and USD clearing layers map each market in detail.
The table below sets the patchwork against the single-correspondent model across the dimensions that actually bite in operations.
| Dimension | Patchwork of providers | Single specialist correspondent |
|---|---|---|
| Relationships to manage | One per market, per layer | One |
| Reconciliation | Many formats, many ledgers, manual matching | One ledger, one reporting framework |
| Settlement certainty | Depends on the slowest correspondent in each chain | Direct access to local rails, known settlement date |
| Client-fund segregation | Varies by provider; often pooled | Named custody accounts, segregated per client |
| Adding a new market | New banking relationship and integration cycle | A configuration change on the existing integration |
| Compliance surface | Distributed across every counterparty | Consolidated under one regulated relationship |
The infrastructure decision
The five-layer stack does not go away. The only real choice is how many institutions a platform trusts to run it, and whether any of them is accountable for the whole journey. The patchwork spreads that accountability so thin that no counterparty owns the outcome, which is precisely when reconciliation breaks and settlement timing drifts. Consolidating the surface onto one specialist replaces a web of thin relationships with a single one spanning collect, hold, convert, clear, and pay out.
Lorum is a globally licensed specialist correspondent institution. It operates a non-lending, 100% reserve model focused on three functions: global multi-currency clearing, named custody accounts, and cash management including wholesale FX. For fintech and PSP platforms, that means the entire money-movement stack, from local collection to local payout, sits behind one integration, one compliance relationship, and one reporting framework, with client funds held in named custody rather than a pooled float.
Frequently asked questions
What are the five layers of moving money across borders?
Collect (receive funds on a local rail), hold (keep them in custody while in flight), convert (execute FX if the payout currency differs), clear (settle the transfer across correspondent infrastructure), and pay out (release funds on a local rail in the destination market). Every cross-border platform runs all five, whether or not it has named them.
Why do platforms end up with a patchwork of providers?
Because licensing is granted market by market and few providers cover more than one or two of the five layers well. Platforms buy collection from one partner, custody from a bank, FX from a broker, and payouts from local disbursement partners. Each layer optimises for itself, so no single counterparty is accountable for the end-to-end payment.
What is the difference between clearing and payout?
Clearing is the settlement engine that moves value between institutions, usually through a chain of correspondent accounts. Payout is the final step that credits the recipient on a domestic rail their bank recognises. A payment can clear across correspondents yet still fail as a payout if it sits in a queue instead of settling locally.
Why is settlement certainty more important than speed?
A fast payment that cannot be relied on to arrive in full, on a known date, just moves the risk rather than removing it. Platforms plan payouts, payroll runs, and supplier payments against expected settlement dates, so certainty of arrival matters more than raw transaction speed.
How does using one counterparty change adding a new market?
In a patchwork, each new market is a new banking relationship, a new integration, and a new compliance review. With a single specialist correspondent that already covers the stack, adding a market is a configuration change on the existing integration, which is why consolidation scales more predictably than a distributed network of providers.
