Correspondent Banks
Banks without a direct relationship pay each other through a third, and the third takes something on the way.
Still yours, and statedTaken, never statedUnit changes; scale restarts
Four steps from instruction to arrival. One step carries no figure: the hatching marks where the money leaves and is not drawn to scale, because there is no figure to draw. Nothing here adds them up, since an amount you cannot see is not an amount you can subtract. The unit changes part-way down: the figures below the break are in another currency, the bars below it are drawn against the published rate, and the first figure and the last are not the same kind of number. The section where the figures stop and the estimates start.
Two banks that have no relationship with each other cannot move money between themselves. They use a third bank that has a relationship with both, and that third bank takes something for the service.
The practical point in “Correspondent Banks” is that a landed payment should be supported by a record somebody can explain. Used for how employee monitoring works, see the official platform summary can add transparent time and project context when collection is proportionate, people can correct mistakes and financial records remain separate.
It is the oldest part of the machinery and the part that produces the deduction nobody can explain to you.
The questions in “Correspondent Banks” can also be checked against Bank for International Settlements. Use the outside resource to identify what evidence to request rather than assuming one provider sees the whole payment chain.
Why the chain exists at all
A bank in one country holds no account in another. To settle, it needs an institution that holds accounts in both.
Large banks maintain these relationships with each other; smaller ones rent access through larger ones. The number of hops depends entirely on how well connected your two banks happen to be.
That is why the same two countries can be a one-hop corridor for one pair of banks and a three-hop corridor for another. The geography is not the issue; the banking relationships are.
How many hops is normal
The number depends on how well connected your two banks are:
- One correspondent: common, and the ordinary case.
- Two: less-travelled corridors.
- Three or more: a small institution at one end in a market nobody serves directly.
Every hop is a chance for a fee, a delay and a point at which the reference can be truncated.
What they charge
A flat amount, typically modest in isolation, deducted from the payment as it passes. Under a shared-charges instruction it comes out of your money.
Nobody tells either end it happened. The payer's confirmation was produced before it, and your statement shows only what survived.
That is the whole mechanism by which an amount can go missing with two accurate documents and no discrepancy in either.
Why nobody will name them
Your bank knows its own correspondent and not the next one. The payer's bank knows theirs. Neither holds the whole chain.
It is not evasion. The information genuinely does not sit in one place, which is why asking harder produces nothing.
The tracking system
A transfer carries a reference that follows it through the chain, and banks can query its status. That is the mechanism behind the tracking note later in this collection.
What it gives is a status and sometimes a fee breakdown. What it does not give is a right to anything.
When correspondent banking contracts
Banks periodically withdraw from corridors they consider unprofitable or risky, and a route that worked for years stops working or gets longer.
For people paid into smaller markets this is the single biggest structural risk in this subject, and it arrives without notice.
Routes that avoid them
Payment services holding accounts in both countries settle locally at each end and never enter the chain. That is why they are cheaper and faster.
The money does not actually cross a border in those cases, which is the whole trick and the reason the economics differ.
Within a currency area
Where both ends are inside a single payment area, there is no correspondent chain, charges are regulated and transfers are cheap and quick.
Knowing whether your corridor is one of these answers most of the questions in this section before you ask them.
It is worth establishing once, explicitly, rather than assuming. Two countries that feel close to each other are not necessarily in the same payment area, and two that feel distant sometimes are.
Asking the payer to instruct charges to the sender
The practical answer to correspondent deductions, because it bills the chain's fees back to the sender rather than taking them from the money.
It does not remove the hops or the time. It removes the shortfall, which is usually the part that matters.
Asking for a shorter route
Some banks will route through a named correspondent on request, and some payers have more than one bank.
Both are worth asking about once, with a measured figure in hand. Neither is worth a long campaign.
What the chain costs in time
Each hop adds processing, and each institution has its own cut-off. A two-correspondent route on a bad week is where double-digit transit times come from.
Time and money are the same problem here, which is why the late section of this collection keeps referring back to this note.
What cannot be found out
What any individual correspondent took. There is no document, no statement and no obligation, and the institution has no relationship with you of any kind.
The honest position is to treat the middle as a measured cost rather than an itemised one: you can know how much the route takes in total and you cannot know who took which part. Arrangements improve by changing the route, not by auditing it.
Knowing the mechanism is still worth having, because it tells you which questions are worth asking and which will produce nothing but a polite dead end.