Why Correspondent Banks Batch Transactions (And What That Costs You)

A payment leaving Frankfurt for Lagos does not spend three days in transit. Nothing is travelling. The money sits in a queue — waiting for a cut-off, a screening run, an FX window, and a settlement cycle to line up. The delay is not distance. It is batch processing, and it is a deliberate design choice made by every large bank in the world.

If you run treasury or payment operations at a PSP, EMI, or exchange, understanding why correspondent banks batch is more useful than complaining that they do. The batching logic explains almost every symptom you deal with: the unpredictable landed amount, the Friday afternoon that becomes Tuesday morning, the fees deducted by a bank you never chose, the tracking page that says "in progress" for 40 hours. Once you can see the mechanics, you can route around them.

Key Point Summary

• The plumbing beneath a cross border payment: The role of the correspondent bank
• Six reasons banks batch instead of processing in real time
• What batch processing actually costs you
• What is actually changing
• Practical steps for treasury and payment teams

The plumbing beneath a cross border payment: The role of the correspondent bank

No bank holds accounts everywhere. Correspondent banks are foreign banks that domestic banks use to access foreign financial markets: instead of moving money physically across borders, banks rely on relationships that let them settle international payments by updating account balances through a chain of institutions. For treasury and payment operations teams at PSPs, EMIs, exchanges, and banks, that model shapes how cross-border money actually moves — and why settlement can be slower, less predictable, and more expensive than the payment experience suggests.

That relationship runs on two accounts. The nostro account ("ours with you") is what the domestic bank holds at the foreign institution, denominated in the foreign currency. The vostro accounts ("yours with us") are the mirror image on the correspondent's books. The bank holding the vostro is the correspondent; the bank whose funds sit there is the respondent bank. Deposits in these accounts are the fuel for every settlement that follows, and they also create the operational realities this article examines: batching, pre-funding, fees, opacity, and reconciliation overhead.

Here is the part that trips people up. When your client sends international wire transfers, the originating bank does not send money abroad. It sends payment instructions over the swift network, which connects more than 11,000 financial institutions and is used for secure transaction messaging — an MT103, or its ISO 20022 successor — and then a chain of banks debits and credits accounts against each other. The intermediary bank in the middle may have no direct relationship with either end. Value moves by adjusting balances; messages move by SWIFT. The two are not the same thing.

This structure does real work. It is the reason smaller banks in new markets can serve importers and exporters without a licence in every jurisdiction. It is what supports trade finance, card transactions settling across borders, payroll for other financial institutions, and the ordinary business of converting currencies at scale. Correspondent banking, for all its friction, is the mechanism allowing domestic banks to reach customers and counterparties they could never reach alone — and the rest of this piece explains how that system works, why banks batch rather than settle in real time, what that costs operators, what has changed in cross-border payments, and what treasury teams can do to optimize routing, liquidity, and reconciliation.

Six reasons banks batch instead of processing in real time

1. The core systems were built for it

Most bank ledgers were designed in an era when computing time was scarce and expensive. Transactions accumulate through the day and post in cycles — end-of-day, sometimes two or three windows. Real time processing was never the design goal; end-of-day accuracy was. Replacing a core ledger is a multi-year, nine-figure programme, so the batch cadence survives even where the surrounding technology has modernised.

2. Netting saves enormous amounts of liquidity

If a correspondent processed each transaction individually, it would have to fund every one separately. Batching lets it net offsetting flows across the same account and move only the residual. Two hundred payments in one direction and one hundred and eighty back can settle as a single net movement. For a bank funding nostro balances in a dozen multiple currencies, that difference is capital that stays deployed instead of sitting idle. Netting is the strongest commercial argument for batching, and it is why it will not disappear quickly.

3. Compliance screening is a queue, not a switch

Every cross-border transaction is screened for sanctions, money laundering, and fraud typologies. Screening engines throw false positives constantly — common names, ambiguous references, incomplete beneficiary data. Each hit goes to a human analyst. Banks staff those teams in shifts, so exceptions clear in batch, not on demand. A payment that hits a review queue at 16:45 local time is a next-day payment, no matter how clean it turns out to be. AI powered screening is starting to compress this, but adoption across the correspondent chain is uneven.

4. FX is executed in windows

Retail-style "instant" currency exchanges are a front-end illusion in the institutional world. Banks aggregate client foreign exchange demand and hand it to a treasury desk that executes at defined times, applying a spread to the aggregate. Your payment does not get its own fill. It gets the batch rate, plus a margin, and you find out the number after the fact.

5. Message and processing costs are per-item

SWIFT messages, reconciliation effort, exception handling, and investigation cases all carry a marginal cost, and correspondent banks charge fees for their services, often in the $25 to $75 range. Batching amortises fixed operational overhead across many payments. This is why transaction fees on a $50,000 payment and a $500 payment look so similar in absolute terms, and why low-value international payments are economically unattractive for correspondents.

6. Settlement systems have opening hours

Underneath the correspondent chain sit domestic RTGS systems — TARGET2, Fedwire, CHAPS, and their equivalents. They operate on business days with defined cut-offs. Even a bank that wanted to process payments continuously would hit a closed system on the other side. Add time zones and you get corridors where the working windows overlap by two or three hours a day.

What batch processing actually costs you

Time, in an unpredictable shape. The problem is not the average settlement time in international transactions. It is the variance. A corridor that clears in 24 hours 70% of the time and 96 hours the rest of the time forces you to hold buffers for the tail, not the mean. Miss a cut-off by ten minutes and a local holiday two hops down the chain turns a Tuesday payment into a Friday one.

Money you cannot quote in advance. Lifting fees are deducted mid-chain by banks you did not select. Each intermediary takes a cut, and the FX applied is the batch rate plus an undisclosed spread. Your beneficiary receives an amount you could not have promised them. For any business invoicing clients in a foreign currency, that gap between quoted and landed amount is a reconciliation problem and a commercial one.

Trapped working capital. Pre-funding is the tax nobody puts on the invoice. To move value quickly, a payment processor must hold funds in nostro accounts across every corridor it serves — idle balances sized for peak volume, in currencies it does not otherwise need. Correspondent banking also underwrites supply chain payments and humanitarian aid flows, which makes those delays and funding costs matter beyond ordinary treasury planning. Batching makes that worse, because funding decisions are made against uncertain settlement timing.

Opacity. SWIFT gpi improved visibility, but a status update is not a guarantee. There are no real time updates on where a payment truly sits when it is parked in an overnight queue, and your clients are asking you, not the correspondent.

Reconciliation load. When receiving banks post netted credits, matching them back to individual wire transfers requires reference data that survived the whole chain intact. Often it did not. Your operations team spends hours on breaks, follow-up reporting, and payment confirmations that exist only because of how the batch posted.

Concentration and security risks. Every additional intermediary is another party handling counterparty data. More hops mean more systems, more staff, and more exposure to data breaches. It also means more parties whose own risk appetite can drop you: de-risking has removed correspondent relationships from entire regions, and the institutions that lose access to foreign banks are usually the ones serving the corridors that need them most, even as the transaction chain is expected to stay secure.

What is actually changing

Three things are moving at once.

Domestic real time payments rails — FedNow, SEPA Instant, UPI, PIX — have made instant settlement normal inside borders. That has reset expectations without solving the cross-border leg. ISO 20022 migration is improving data quality, which cuts false positives and reduces the exception queues that cause batch delays in the first place. And stablecoin settlement removes the intermediary chain entirely for the value transfer step, replacing nostro pre-funding with on-chain settlement that runs 24/7.

None of these makes correspondent banking obsolete. Banks will still batch, netting will still save liquidity, and compliance will still take time. What changes is that you no longer have to accept a single chain as the only route.

Practical steps for treasury and payment teams

• Map your corridors by variance, not average. Measure P90 settlement time per corridor and per correspondent, not the mean; better-instrumented routing saves time for treasury and operations teams by reducing manual follow-up.

• Instrument your fee leakage. Compare quoted versus landed amounts systematically. Most teams discover the real cost is 2–4x the visible fees.

• Know your cut-offs cold. Publish them internally for cross border payments, per currency, per country, including the correspondent's own deadline rather than your bank's marketing one.

• Reduce hops where possible. Fewer intermediaries means fewer deductions, less delay, and lower operational risk.

• Hold a second route. Corridor redundancy is cheaper than the working capital you tie up covering a single chain's tail.

• Push for structured remittance data. Clean beneficiary data is the highest-leverage fix for screening delays.

Conclusion

Batching is not going away, and there is little reason it should. Correspondent banking still does something no other rail does at the same depth: it reaches almost everywhere, in almost every currency, with finality a bank is willing to underwrite. The better question is narrower — which of your flows genuinely need that, and which are simply inheriting a cut-off they never chose.

That distinction is the whole strategy. Where reach and depth matter, keep the correspondent relationship. Where speed and predictability matter — customer payouts, corridors you pre-fund to cover the gap, weekend and holiday flow that currently has nowhere to go — route around the batch.

FinchTrade works with PSPs, EMIs, banks, and exchanges on exactly that second route: institutional OTC liquidity for the conversion, and settlement rails that do not observe a cut-off.

For requesting more information about how we can help reach out to us. We're here to help and answer any questions you may have.

Contact us!