The True Cost of Paying International Payroll by Bank

Paying international payroll by wire looks like the same problem as paying an overseas supplier: a bank, a correspondent, a visible fee and an exchange rate nobody fully explains. It is not. We already broke down the cost of a SWIFT transfer from the supplier-payment angle in how much does a SWIFT transfer cost in 2026, and that layered-fee logic still holds here. What changes is the shape of the problem: payroll is not one large payment, it is dozens of small, recurring payments, on a date that cannot move, with a person, not a company, absorbing the difference.

At Soulbit Academy we look at this cost from the treasury of a US or European company paying contractors across Mexico, Colombia and Brazil, all owed on the same day each month. The goal is to show where the real cost hides when paid by bank, without naming specific providers, and why it does not behave like the cost of a single isolated wire.

What changes between paying a supplier and paying international payroll

Paying international payroll by bank carries three traits a supplier payment does not: many recurring recipients, each amount small relative to the wire's fixed cost, and a payment date already committed to a person who depends on that income. A supplier can usually accept a three day delay. A contractor waiting on a paycheck almost never can.

That difference changes where the cost bites. In a supplier payment, the company negotiates one large transfer directly and absorbs the margin inside the agreed price. In payroll, the amount per person is too small to negotiate, and the contractor rarely has any way to dispute a fee never itemized. The cost does not disappear, it changes pockets.

A company paying 15 contractors across Mexico, Colombia and Brazil is not sending 15 identical fees. It is sending 15 wires through different corridors and correspondents, each with different odds something goes wrong. The problem is not the payroll total, it is the multiplication of friction points per recipient.

Cost per recipient versus cost per batch

The real cost of international payroll is measured per recipient, not on the total transferred, because most bank charges are fixed and do not scale down when the amount is small. A sending fee reasonable on a 50,000 dollar supplier wire is disproportionate on a 1,200 dollar paycheck.

That makes payroll close to the worst use case for correspondent banking. The bank charges per transaction, not per relationship, so paying 30 people through individual wires costs, in fixed charges alone, a multiple of one large payment.

Why doesn't a payroll batch always lower the cost per person?

Because the savings from batching depend on correspondents willing to receive and redistribute the full batch in the destination country, and that infrastructure does not exist equally in every corridor. A batch sent to a country with few active correspondents can end up costing nearly as much per person as individual wires, just with one extra administrative step.

Correspondent fees and the FX margin: who actually pays them

Correspondent bank fees and the FX margin are, in practice, paid by the worker, not by the company sending payroll. When a correspondent deducts a charge from an amount in transit, that reduction shows up in the beneficiary's account, not on the company's statement, which already recorded the payment as sent in full.

That asymmetry is the most distinctive trait of payroll versus supplier payments. A supplier invoices for an amount and disputes it if less arrives, backed by the commercial weight to demand an adjustment. A contractor who receives less than expected almost never knows who to ask, because the deduction happened at a link neither party controls.

The FX margin compounds the problem because it never shows up as a separate line. It is built into the rate applied, and it hits the full paycheck every month, compounding over time. By World Bank data, the traditional banking channel averages close to 15 percent in cost on a 200 dollar remittance, against a 6.36 percent global average across all channels measured. That series measures personal remittances, not corporate payroll, but it travels the same correspondent corridor and carries the same kind of built-in margin, so it works as an honest reference for the channel's cost, not the exact price of a B2B payroll run.

Cost layer	Who ends up paying it	How it shows up in payroll
Sending fee at the originating bank	The company	Visible when the payroll batch is ordered
Correspondent deduction in transit	The worker	Deducted from the amount; net pay drops with no advance warning
FX conversion margin	The worker	Built into the applied rate, repeats every month on the full paycheck
Retry from incorrect bank details	The company	Administrative and time cost, not a bank fee
Tracing a delayed payment	The company	Extra charge if an investigation has to be requested from the bank

Value days versus a payment date that cannot move

An international wire settles anywhere from one to several business days depending on the corridor, while payroll carries a payment date the company has committed to that it cannot shift without creating a labor problem. That gap forces the company to build in a buffer it should not need.

What happens if the payroll batch does not settle on time?

The company has no one to negotiate a delay with the way it could with a supplier. A contractor who does not receive a paycheck on the expected date has an immediate problem, and in several countries a late payment can also trigger a labor claim or interest owed. That is why most finance teams send the batch two or three days ahead of the bank's estimated window, tying up cash systematically just to cover the rail's uncertainty.

That buffer carries a financing cost rarely booked as one: capital sitting in transit that earns nothing while it waits. For a company with recurring international payroll, that extra day becomes a structural cost, not an occasional surprise.

Retries from wrong bank details and the work of reconciling N payments

A payment retry happens when the bank rejects a wire because a beneficiary detail does not match, and reconciling N wires against N payroll records is the additional work that grows with the number of recipients, not with the payroll total. With one supplier, an error like this is fixed once. With a payroll of 20, 40 or 60 people across several countries, a small percentage of outdated details each month becomes a constant stream of cases.

Each retry requires reaching the contractor, confirming the correct detail, resending the instruction and waiting again for settlement, all before the committed date. No bank bills that as a line item, but it consumes finance team hours every cycle.

On top of that comes reconciling N separate wires against N payroll records, each with its own receipt, exchange rate and settlement date. Reconciling one large supplier payment takes minutes; reconciling 40 small payments with different details takes hours, and that administrative time is another layer of the real cost almost never measured alongside bank fees.

Aspect	Individual wire per recipient	Batch payment
Fixed cost per person	High relative to the paycheck amount	Spread out, but depends on the corridor's infrastructure
Settlement time	One to several business days, varies by corridor	Depends on the full batch, not the individual payment
Retries from wrong details	Handled one by one	Can block the entire batch if the process does not isolate the case
Reconciliation	N separate receipts to review	One batch record, if the provider delivers it that way
Net amount visible before sending	Rarely, known only once the payment arrives	Depends on whether a quote is visible before confirming

What the available data says about the cost of the banking channel

The most cited public figure on the cost of the banking channel comes from the World Bank: a 6.36 percent global average cost, rising to close to 15 percent when the channel is a bank, though that series measures personal remittances, not corporate payroll. The World Bank's Remittance Prices Worldwide monitor measures a 6.36 percent global average cost to send 200 dollars, and close to 15 percent when the channel is a bank, the most expensive provider type measured.

Does remittance data help estimate the cost of international payroll?

It works as a thermometer for the channel, not as an exact budget. A 200 dollar remittance and a 1,200 dollar paycheck do not share the same fixed cost structure, but both travel the same correspondent banking network and face the same kind of built-in FX margin.

The other public reference is the target the financial sector itself set to fix this. The cross-border payments program coordinated by the Financial Stability Board (FSB) sets, for the end of 2027, a global average cost cap of 1 percent for retail cross-border payments, with no corridor above 3 percent. The FSB's October 2025 progress report admits it is unlikely that timeline will be met. The correspondent banking network has also shrunk steadily over the last decade, according to the BIS Committee on Payments and Market Infrastructures: active correspondents fell close to 22 percent globally between 2011 and 2019, and 34 percent in Latin America, the steepest regional decline measured. Fewer correspondents means longer chains to reach the same destination, and more points where a payroll payment can lose value in transit.

What Soulbit solves today for international payroll, and what it does not

Soulbit solves recurring and batch payroll in stablecoins today, holding balances in USDC and USDT and in fiat USD, EUR and GBP, converting between them at a quote the company sees before confirming. That means the company knows the price before sending the batch, with no built-in FX margin and no correspondent deductions in transit, because a stablecoin transfer does not depend on that banking chain. It includes KYB verification and anti-money-laundering checks on every account.

Being honest about this means saying where it stops. For a contractor to receive payment this way, they need to be able to receive stablecoins directly, or have their own way to convert that balance into local currency, and not everyone on a distributed team has that option today. The local banking rail built into the V1 exists only in Colombia; everywhere else, the last step into local currency is left to the worker or the company. There are also no cards, no yield on balances, no native token and no native mobile app in this version.

Where this changes the problem described here is in three places: cost per recipient stops multiplying fixed correspondent charges, settlement no longer depends on a buffer built for banking uncertainty, and reconciling N payments gets simpler because every transfer is recorded on chain with its own verifiable identifier, as we explain in reconciling stablecoin payments in accounting. Retries from wrong details do not disappear entirely, but the layers of intermediary correspondents nobody sees until a payment arrives short are removed altogether.

For a company that also needs to classify staff correctly, it is worth reviewing contractor vs employee in Latin America first. For the recurring payment itself, we go deeper in pay international contractors in USDC and in scheduling recurring payroll in stablecoin, and compare the banking rail against the stablecoin rail in SWIFT vs stablecoin for international payments.