Inside one country, banks settle through a shared central bank, and a payment is just a rewriting of balances. Across borders there is no shared central bank and no common ledger — so moving money between countries is a genuinely harder problem, solved by a centuries-old patchwork of banks holding accounts with one another. This module explains correspondent banking and the nostro/vostro accounts at its heart, clears up the widespread confusion about what SWIFT actually does, traces exactly why cross-border payments are slow and expensive, and examines "de-risking" — the quiet process by which whole regions get cut off from the global financial system.
Return to the simplest insight of Module 03: within a single country, a payment usually moves no money at all. If you and the recipient bank at the same institution, paying is just lowering one balance and raising another. If you bank at different institutions, they settle the difference through their accounts at the central bank — the shared ledger every domestic bank connects to. The central bank is the common point of trust and settlement that makes domestic payments work.
Now send money to someone in another country, and that foundation vanishes. There is no global central bank, no shared ledger spanning nations, no single venue where a bank in Kenya and a bank in Germany can settle against each other directly. Each country has its own central bank, its own currency, its own rails — and they are not natively connected. The smooth domestic mechanism simply does not exist across the border.
So the cross-border problem is fundamental, not a mere matter of slow software: how do two banks that share no common settlement venue, and may have no relationship at all, move value between two different monetary systems? Everything in this module — correspondent accounts, SWIFT, the fees, the delays, the exclusion — is part of the workaround the world built because the clean domestic answer is unavailable internationally.
Domestic payments have a shared central-bank ledger to settle on; cross-border payments do not. There is no global central bank. Moving money between countries therefore means stitching together separate monetary systems through private arrangements between banks — and the cost, slowness, and gatekeeping of cross-border payments all trace back to that missing shared foundation.
The workaround the world settled on is correspondent banking: in the absence of a shared ledger, banks simply hold accounts with each other across borders. A bank in one country opens an account at a bank in another, and that pair of banks can now move value between their countries by adjusting the balance in that account — recreating, privately and bilaterally, the "just rewrite the balances" trick that the central bank provides domestically.
The jargon is worth learning because it is unavoidable and, once decoded, simple. From the perspective of a bank looking at the account it holds at a foreign bank, that account is its nostro (from the Italian for "ours" — "our money, held over there"). The same account, seen from the foreign bank that is hosting it, is a vostro ("yours" — "your money, held here by us"). One account, two names, depending on which side you stand on. To pay a beneficiary abroad, your bank instructs its foreign correspondent to pay out from your nostro balance there.
The complication is that no bank can hold an account at every other bank on earth. When your bank has no direct relationship with the recipient's bank, the payment must travel through a chain of correspondents — your bank pays a bank it does have a relationship with, which pays another, which finally reaches the recipient's bank. A single cross-border payment can hop through several institutions, each holding the next one's accounts, each passing the value along. That chain is the source of much of the cost and delay we examine shortly.
With no global ledger, banks hold accounts with each other (nostro/vostro) and move cross-border value by adjusting those balances — and when there is no direct relationship, the payment is relayed through a chain of correspondent banks until it reaches the destination.
Almost everyone has heard that international payments "go over SWIFT," and almost everyone misunderstands what that means. The single most useful correction in this module: SWIFT is a messaging network, not a settlement system. It moves instructions, not money.
SWIFT (the Society for Worldwide Interbank Financial Telecommunication) is, in essence, a secure, standardized global messaging service for banks. When your bank wants its correspondent to pay someone, it sends a SWIFT message — a structured instruction saying, in effect, "pay this beneficiary this amount, debit our account with you." The money never travels through SWIFT; it moves through the nostro/vostro balances of Section 02. SWIFT just carries the orders that tell banks how to adjust those balances. Think of it as the global postal service for payment instructions, sitting on top of the correspondent network that does the actual value transfer.
Why does this distinction matter so much? Because it explains both how the system works and how it can be wielded as a weapon. Since virtually every international bank relies on SWIFT to communicate, cutting a bank — or a whole country's banks — off from SWIFT severs their ability to instruct cross-border payments, even if money theoretically still exists to move. That is why disconnection from SWIFT has become a potent geopolitical sanction: it does not seize the money, it cuts the communication that lets the money move. We return to that power in Section 06.
Domestic payments in many countries are now instant and nearly free. Cross-border payments remain, by comparison, slow, costly, and frustratingly opaque — often taking days and losing a meaningful percentage of the amount along the way. The reasons follow directly from the correspondent structure:
| Source of friction | Why it adds cost or delay |
|---|---|
| Chain of correspondents | Each intermediary in the chain takes a fee and adds a processing hop; more hops, more cost and more time |
| Repeated compliance checks | Each bank in the chain re-screens the payment for money-laundering, sanctions, and fraud, adding manual review and delay |
| Pre-funded nostro accounts | Banks must park money in foreign accounts in advance to settle payments; that idle capital is a cost ultimately passed to customers |
| FX conversion | Crossing currencies means a conversion, and the spread on that conversion is often the largest hidden cost |
| Cut-off times & time zones | Payments wait for the next processing window; a chain spanning several zones can stall for a day or more |
| Opacity | The sender frequently cannot see the total fees or the arrival time in advance, so costs are discovered only after the fact |
Nowhere does this bite harder than in remittances — the money migrant workers send home, a lifeline for hundreds of millions of families and, for some economies, a major share of national income. Precisely the people who can least afford it pay some of the highest costs: the global average cost of sending a remittance has hovered stubbornly around 6% of the amount, far above the few-percent target international bodies have set, and well above what the underlying technology should require. A worker sending $200 home can lose $12 or more to fees and FX spread, repeated every month. The frictions in the table are not an abstraction; they are a tax falling hardest on the poorest cross-border users.
The correspondent system has a darker failure mode than mere expense. Maintaining a correspondent relationship is not just a technical link; it is a compliance liability. The host bank must monitor the payments flowing through the account for money-laundering, terrorist financing, and sanctions violations — and if it gets that wrong, it can face enormous fines and legal jeopardy. So each correspondent relationship carries an ongoing regulatory cost and risk.
Faced with that calculus, large global banks increasingly conclude that some relationships are simply not worth it. If a small bank in a poor or high-risk jurisdiction generates modest fees but substantial compliance risk, the rational move for the big bank is to drop the relationship entirely — a practice known as de-risking. The big bank protects itself; but the small bank, and often its entire country, loses its bridge to the global financial system. When a region's banks cannot find any global correspondent willing to deal with them, that region is effectively cut off — unable to receive remittances easily, settle trade, or connect to international finance at all.
Notice the cruel logic: the compliance rules exist for good reasons (stopping crime and terror financing), yet their unintended effect is to disconnect the poorest and most vulnerable economies, pushing their payments toward informal, less-monitored channels — the opposite of what the rules intended. It is a vivid example of financial innovation's recurring theme, which the companion track examines closely: a well-meant intervention producing a serious unintended consequence for those it was never aimed at.
Module 06's moat does not stop at the water's edge — it reappears, in a new form, in the cross-border system. To move money internationally you need correspondent relationships, and those are extended only to trusted, regulated, established banks. A newcomer cannot simply plug into the global payment network; it must, once again, route through the incumbents who hold the relationships. The cross-border layer reproduces the same gatekeeping as the domestic one: a club of connected institutions, and everyone else renting access through them.
Two features sharpen the concentration. First, a small number of large global banks sit at the hubs of the correspondent network, because building and maintaining worldwide relationships is something only the biggest can afford — so cross-border flows funnel through a handful of institutions. Second, a small number of currencies, above all the U.S. dollar, dominate international payments; an enormous share of cross-border value is settled in dollars even between countries that are not the United States, because the dollar is the common language of the correspondent system.
That concentration has a geopolitical edge, as Section 03 hinted. Because so much of the world's money moves through dollar correspondents and SWIFT messaging, the authorities who control those choke points can exclude a target from the global system — freezing its access to dollar settlement or cutting its banks off from SWIFT. This has been used as one of the most powerful sanctions available, capable of isolating entire national banking systems. It is the clearest demonstration that the cross-border "plumbing" is not a neutral utility but a concentrated structure whose control confers real power — and a major reason some countries are actively seeking alternatives that do not run through the incumbent network.
It would be easy to conclude that cross-border payments are doomed to be slow and costly by their nature. The comparative evidence says otherwise: where regions have chosen to build shared infrastructure, the friction largely disappears. That proves the difficulty is structural and addressable, not fundamental.
The currency-dominance question runs alongside this. Part of the motivation for regional networks like PAPSS, and for the multi-central-bank experiments the next track covers, is a desire to settle cross-border payments without routing everything through a single dominant currency and its correspondents — both to cut cost and to reduce dependence on infrastructure that can be switched off. Whether the world fragments into competing regional rails or converges on new shared ones is one of the open questions of contemporary finance.
Gather the module together. Because there is no global central bank, moving money across borders relies on a patchwork of bilateral correspondent relationships, coordinated by a messaging layer, dominated by a few currencies and a few global banks. The result is a system that is slow, expensive, opaque, gatekept like the domestic moat, exclusionary toward the regions de-risking abandons, and concentrated enough to be wielded as a weapon. And the SEPA example proves none of this is inevitable — it is the consequence of missing shared infrastructure, a problem that can be engineered away.
That is a large residual problem, and it sets up some of the most active innovation in finance. The companion track takes up the responses directly:
Two modules now remain in this classic track. Module 08 asks the question this one keeps brushing against — who does the whole banking apparatus, domestic and cross-border, leave out entirely? Then Module 09 gathers every residual problem the track has surfaced — the credit risk of deposits, the structural fragility, the charter moat, the cross-border cost and exclusion of this module, and the unbanked of the next — into the single brief that opens Innovation in Banking.
Six questions on the cross-border payment system, its frictions, and its consequences. The questions test the reasoning rather than memorization of any single figure.