A supplier agreement signed in one currency and performed in another carries an exchange exposure that neither the sourcing team nor the counterparty usually prices. Over a twelve-month deal a few percentage points of rate movement is noise. Over a five-year manufacturing or services agreement, it can exceed the margin the deal was signed to earn.
What makes this hard to manage is not the economics, it is the visibility. Currency terms sit in the payment article of individual agreements, expressed in a sentence or two, and nothing aggregates them. Treasury models exposure from forecast cash flows; legal owns the clause language; procurement owns the supplier. None of the three holds a complete list of which contracts carry which currency terms.
This article covers the clause patterns that appear in cross-border agreements, what each one actually allocates, and the contract data a finance team needs in order to see portfolio exposure without reading every agreement. The aim is not to eliminate exchange risk, which is rarely possible or worth the cost, but to know where it sits and who agreed to carry it.
The clause patterns and what each allocates
Currency provisions cluster into a small number of recognizable forms. Identifying which one an agreement uses is the first step, because the monitoring each requires is different.
- Single currency, no adjustment, the simplest and most common form; the party transacting outside its functional currency carries the entire rate risk for the life of the contract.
- Fixed reference rate, conversion happens at a rate agreed on signature, which shifts risk in one direction and typically needs a review or reset date to stay tolerable.
- Floating rate with a named source, conversion uses a published rate on a defined date, spreading risk over time without eliminating it, and requiring both the source and the date to be unambiguous.
- Sharing band or collar, the parties absorb movement within an agreed band and share or reallocate movement beyond it, which is the fairest structure and the most operationally demanding.
- Indexed price adjustment, the price itself is restated periodically against a currency index, usually alongside other cost indices, which turns the currency term into a recurring calculation obligation.
Each of those patterns implies different downstream work. A single-currency clause needs nothing but awareness. A collar needs someone watching the rate against the band and initiating the adjustment when it breaches. An indexed clause needs a scheduled recalculation with an evidence trail. Knowing which pattern applies to which agreement is therefore a prerequisite for managing any of it. It also determines who should own the clause after signature, since a collar is a treasury responsibility and an indexed adjustment is a category management one.
A clause that says conversion happens at the prevailing exchange rate is an invitation to argue. Prevailing where, published by whom, at what time of day, on the invoice date or the payment date. The drafting cost of naming a specific published source and a specific determination date is close to zero, and it removes the most common category of currency dispute entirely.
What goes wrong in practice
The failure modes are rarely dramatic. They accumulate quietly across a portfolio and surface as unexplained variance in cost of goods or in receivables.
- The clause is never extracted. Currency terms stay inside the executed PDF, so the only way to answer a portfolio question is to reread agreements one at a time.
- Reset dates pass unnoticed. Fixed-rate and banded clauses usually carry a review date. Without an alert, the original rate simply continues and one party absorbs the drift.
- Invoices are converted inconsistently. When accounts payable applies its own month-end rate rather than the contractual rate, the contract and the ledger disagree and neither is obviously wrong.
- Amendments change the currency silently. A later order form or amendment restates pricing in a different currency, and the master record still shows the original term.
- Hedging ignores contractual protection. Treasury hedges exposure that a sharing band already covers, paying twice for the same protection because nobody read the clause.
Five steps to bring currency exposure under control
Step 1, Inventory the cross-border agreements
Start by identifying every active contract where the counterparty, the place of performance or the payment currency sits outside your functional currency. This is a smaller set than the full portfolio and it is where all of the exposure lives. Counterparty normalization matters here, because entity naming is what determines whether a foreign subsidiary is correctly identified.
Step 2, Extract the currency term into fields
Capture payment currency, rate source, determination date, any band or collar thresholds, reset dates and indexation formulas as structured data on the contract record. Prose in a clause library is not enough; the point is to be able to filter and total, which requires typed fields rather than text.
Step 3, Reconcile the contract term against the invoicing practice
Compare what the clause says with what accounts payable and accounts receivable actually do. Divergence here is common and it is usually accidental. Where the practice is wrong, fix the process; where the clause is unworkable, note it for the next amendment cycle.
Step 4, Put reset and review dates on the calendar
Every fixed rate and every banded clause needs a dated obligation with an owner. Treat a rate reset the same way you would treat a renewal notice: a deadline, a responsible person and an escalation if it passes. Missed resets are a pure transfer of value to whichever party the drift favors.
Step 5, Feed the portfolio view to treasury
Aggregate exposure by currency, counterparty and remaining term, and hand it to the people who hedge. Contractual protection should reduce the hedge notional; where a sharing band already caps downside, the treasury position should reflect it rather than duplicating it.
See currency terms as structured contract data
Watch how payment currency, rate references and adjustment bands become reportable fields rather than buried prose.
Why this is a contract data problem
Every step above depends on the currency term existing as a field rather than as a sentence. That is a data modelling question before it is a treasury question, and the approach is the same one used for any other commercially significant term, as set out in How to build a contract metadata model for AI search. Currency terms are also a classic case of value quietly leaving the business after signature, which is the wider pattern catalogued in Stopping Value Leakage: What Your CLM Must Track After Signature. Once the fields exist, the reporting question answers itself. Most organizations find that the number of agreements carrying active currency mechanics is small enough to manage by hand, once they know which ones they are.
Clause pattern and the data it requires
The monitoring burden varies sharply by clause type. This table maps each pattern to the fields it needs and the recurring action it creates.
| Clause pattern | Fields required | Recurring action | Owner |
|---|---|---|---|
| Single currency | Payment currency | None, awareness only | Contract owner |
| Fixed reference rate | Rate, reset date | Reset review at the dated interval | Finance |
| Floating with named source | Source, determination date | Rate lookup per invoice | Accounts payable |
| Sharing band or collar | Band thresholds, base rate | Monitor rate against band | Treasury |
| Indexed adjustment | Index, formula, cadence | Scheduled recalculation and notice | Category manager |
Currency terms are easy to overlook because they look administrative. They are not. In long-dated cross-border agreements the currency clause frequently moves more money than the discount that took three rounds to negotiate, and it does so without anyone making a decision. Getting control of it does not require sophisticated modelling. It requires knowing which clause pattern each agreement uses, holding that as data, and putting an owner and a date against the small number of clauses that create recurring work. That is a modest amount of structure for a term that compounds quietly across every invoice for the life of the agreement.
For related guidance, see What agentic AI changes in contract operations, The legal ops guide to contract data migration and How to build a clause library reviewers actually use.
