Managing FX and multi-currency flows. 7 chapters and a final quiz.

Chapter 1. Mapping your exposure, currency by currency.

A multi-currency payments manager does not manage transactions. They manage positions. A position is a balance in one currency, on one date, with a lag before it becomes usable. Until those positions are mapped, conversion and hedging decisions rest on aggregated bank balances, which have already lost the original currency and the collection date. The map takes a day to build and an hour a month to keep current.

Four exposures not to confuse

• Transaction: receivables and payables already booked, with a known amount and a fixed currency. It is the only one that can be hedged without assumptions.
• Forecast: revenue budgeted in a currency but not yet invoiced. The forecasting error becomes an exposure in its own right.
• Translation: the accounts of a subsidiary kept in another currency, translated at period-end. It affects the financial statements, not cash.
• Economic: price competitiveness against a rival that invoices from elsewhere. It cannot be hedged. It is handled in the price list.

The column that changes the most decisions is the net position, because many companies hedge their dollar receipts while already paying suppliers in dollars. The spending offsets the receipts. Hedging the gross amount then creates a position where none existed, pays for an instrument to protect it, and carries it until the contract matures. The natural hedge shows up in this column, before any instrument is bought.

The amount trap: ISO 4217 decimals

Chapter 2. Rebuilding the effective rate and pricing the hidden markup.

No invoice has a line called “FX markup.” The cost sits in the rate. Measuring it requires four data points per transaction, and providers do not always deliver them. A report that gives the amount credited without the applied rate or its quote time makes an audit impossible, however detailed the rest of the reporting. Negotiate these four fields in the contract: asking for them during acceptance testing is too late.

Where to find the rate, document by document

Batch measured: one month of USD collections credited in EUR
  transactions              = 4,812
  total_source_amount       = 12,000,000.00 USD
  total_amount_credited     = 10,930,800.00 EUR
  fixed_fees_deducted       =      1,200.00 EUR

Step 1 — effective rate for the batch
  effective_rate = (amount_credited + fixed_fees) / source_amount
                 = (10,930,800.00 + 1,200.00) / 12,000,000.00
                 = 0.911000

Step 2 — weighted benchmark
  For EACH transaction, rebuild the market rate at its timestamp.
  Weight by source amount, never by transaction count.
  weighted_benchmark_rate = 0.916500

Step 3 — spread in basis points
  spread_bp = 10,000 x (0.916500 - 0.911000) / 0.916500
            = 60.0 bp

Step 4 — cost of the batch, then annualized
  cost_month = 12,000,000 x 0.00600 =  72,000 USD
  cost_year  = 72,000 x 12          = 864,000 USD

Step 5 — dispersion
  Redo the calculation by amount band.
  A presentable average spread often hides heavily charged small tickets.

Takeaway: the contract advertises zero conversion fees.
It charges 60 bp in the rate, or 864,000 USD over the fiscal year.

• Get a transaction-level export: source amount, amount credited, applied rate, quote timestamp.
• Lock in the benchmark source and the time it is read, then archive the series and never edit it.
• Measure a full month. A hand-picked sample measures the selection, not the provider.
• Weight by amount. A per-transaction average gives a €40 ticket the same weight as a €400,000 payout.
• Break the data down by amount band, currency, and channel, and look at the dispersion before the average.
• Repeat the measurement every month. A spread creeps up slowly, never all at once.

This measurement opens a negotiation, not a legal dispute. A provider faced with a timestamped, reproducible figure either narrows its spread or loses the volume, and either outcome works. Without the measurement, the discussion stays on the price list, and the price list almost never includes the largest cost item.

Chapter 3. Dynamic currency conversion: deciding, governing, testing.

Dynamic currency conversion (DCC) offers a foreign cardholder the option to pay in the currency of their card. The rate comes from the acquiring side. Part of the markup goes back to the merchant or the ATM operator, so the product is sold to the merchant, not to the customer. The decision to enable it belongs to the payments manager, and it rests on measured figures, not on a sales pitch.

• Option accepted by default, or a decline button less visible than the accept button.
• Cashier makes the choice for the cardholder, even just to save time.
• Markup missing from the screen and shown only on the receipt, or vice versa.
• Local currency amount shown smaller than the converted amount, or in light gray.
• Mixed terminal fleet: one terminal model is compliant, another is not, and no one keeps the inventory up to date.

The rebate does not arrive with sales settlement. It comes in a separate flow, often monthly, that is hard to match to the transactions that generated it. This has two practical consequences. Accounting must book it as revenue, location by location, or the margin per store will be wrong. Financial planning and analysis (FP&A) must weigh it against the cost of handling disputes over the same period. Neither figure alone decides anything.

Chapter 4. Collecting locally: opening local collection accounts.

Collecting from abroad costs you twice. The transaction becomes cross-border, so it costs more, and issuers decline it more often because they apply stricter risk rules to foreign transactions. The funds then go through a conversion the merchant does not control, at a rate it discovers on its statement. A local setup fixes both problems but creates new ones, which must be handled in the contract rather than discovered in operation.

Four setups, four risk profiles

• Ownership and safeguarding: who legally holds the funds while they sit in the local account, and what happens if the provider fails.
• Conversion on instruction: the contract must prohibit automatic conversion on receipt. Without this clause, the merchant loses all control over timing.
• Four conversion fields per transaction in the reports: source amount, amount credited, applied rate, quote timestamp.
• Name matching: euro credit transfers have been subject to Verification of Payee since October 2025. A virtual account in the provider's name triggers a warning for every payer.
• Exit: notice period, cost of repatriating balances, and what happens to the funds if the provider loses a local license.

Account ownership drives every other question. An account opened in the provider's name makes the merchant a creditor, not an account holder. The provider's failure then becomes a balance sheet risk, not an operational incident. Some jurisdictions require client funds to be safeguarded in segregated accounts. Others rely solely on the institution's own capital. The difference between the two regimes shows up only in the license, so ask for the answer in writing, country by country, before opening the first account.

Chapter 5. Hedging or not: policy, instruments, accounting.

Hedging is not protection. Hedging trades uncertainty for a known price and gives up the gain an opposite move would have produced. So the question for the treasurer is not “should we hedge?” but “how much, over what horizon, with which instrument?” A written policy answers that once and for all. Without one, every decision is relitigated under market pressure, exactly when perspective is hardest to find.

A hedging policy in six lines

• What gets hedged: the firm transaction exposure, and what share of the forecast exposure.
• Horizon: the number of months hedged, and how the ratio tapers further out.
• Authorized instruments, and those a board decision excludes on principle.
• Delegation of authority: the amount above which a second level must approve the trade.
• Counterparties: number of banks, limit per counterparty, margin calls accepted.
• Review: frequency, metrics tracked, conditions for exiting a program already underway.

The forward rate is not a forecast. It is derived from the interest rate differential between the two currencies. A high-yielding currency trades forward below its spot rate. Many teams read that discount as a signal the currency will fall, reject the hedge because it looks too expensive, and leave the position open until collection. The forward rate predicts nothing; it reflects an interest rate gap.

Chapter 6. Repatriating funds under exchange controls.

Collecting is not the same as getting the money out. In some markets, local currency can leave the country only on terms set by the central bank. You do not discover these rules after the first sale. You address them in the market entry plan, with the same rigor as a tax clause. A blocked collection is a receivable, not cash you can use.

Five types of measures, five responses

The repatriation file starts at the sale

• Commercial contract and invoice, denominated in the currency actually collected.
• Proof of performance: delivery, service rendered, signed acceptance report.
• The transaction classification code, where local rules require one.
• Tax certificates the local bank requires before executing the transfer.
• A timestamped record of the request and the response: it documents the real delay, as opposed to the announced one.

A single metric is enough to manage the issue: the age of the local balance not yet repatriated, in days, currency by currency. It rises as soon as an allocation queue lengthens, well before the local bank announces anything and well before the finance department sees it on a consolidated report. There are three responses, and they are prepared in advance: spend locally, borrow locally against the balance, or net within the group where local rules allow it.

Chapter 7. Managing the real cost, the dashboard, and the RFP.

The cost of a currency breaks down into several items: the spread to the market rate, the stated fee, the cross-border transaction fee, fixed charges, the carrying cost, and the accounting translation difference. Only one of them appears on an invoice. The others have to be calculated, and only that calculation makes two providers comparable. The price list opens the discussion; it does not settle it.

Five metrics, five formulas

None of these metrics has a universal threshold, so the threshold is set at the first measurement and then tightened. A dashboard that compares the company with an outside benchmark never triggers anything. A dashboard that compares the company with itself, month after month, triggers decisions. That is the whole difference between reporting and managing.

The RFP: eight questions, and the proof to demand

• Which rate source, read at what time? Require the name of the source, not a vague “today's rate.”
• What spread is applied over that source, by currency and by amount band?
• Does the transaction-level export include the four conversion fields? Ask for a real file, not a screenshot.
• Can the currency received be held without conversion, and for how long?
• Who holds the local accounts, and are the funds safeguarded?
• Which licenses does the provider hold directly, in which countries, and which are borrowed from a partner?
• What is the contractual delay between collection and availability, country by country?
• What are the exit terms: notice period, cost, time to return balances?