Cash pooling: notional, physical or ZBA — which one for a multi-currency group
The three structures differ not only in their banking mechanics, but in their tax, legal and FX consequences. What each really changes when the accounts are not all in the same currency.
Treasury centralisation is presented as a technical optimisation. It is first a legal and tax decision and, in a multi-currency group, an FX decision. The three available structures have neither the same cost, nor the same constraints, nor the same level of risk.
The three structures
| Notional pooling | Physical pooling (ZBA) | Centralisation through advances | |
|---|---|---|---|
| Mechanics | Balances are merged for interest calculation, with no movement of funds | Balances physically sweep each evening to a header account | The parent grants contractual advances to subsidiaries |
| Movement of funds | None | Daily, automatic | Occasional, on request |
| Intercompany flows to document | Limited | Substantial — each sweep is a loan | Substantial, but controlled |
| Multi-currency | Rare: few banks accept it outside a single currency | Possible, but one pool per currency | Possible, but creates intercompany FX risk |
| Main constraint | Bank availability and cross-guarantees | Administrative volume and interest taxation | Contractual formality |
The rule that decides it in a multi-currency group
A cash pool must never mix currencies. Merging a credit balance in dollars with a debit balance in euros into a single figure creates a permanent, undocumented FX position carried by the header company.
The correct practice is one pool per currency: a euro header, a dollar header, a sterling header. Consolidation then happens at the analysis level, not at the account level. Any conversion between pools becomes an explicit FX decision — dated, quoted and traceable — instead of an automatic side effect.
The three preliminary questions
- 1Is the intercompany interest rate defensible?
Advances between related companies must bear a market rate. Too low a rate exposes the lending company to a tax challenge; too high a rate limits deductibility for the borrower. The benchmark is documented when the agreement is signed, not after the fact.
- 2Can foreign subsidiaries take part?
Some states restrict or regulate cross-border cash sweeps, and exchange controls remain a reality in several jurisdictions in Africa, Latin America and Asia. The question is examined country by country before any bank configuration.
- 3Who carries the counterparty risk?
In a physical pool, subsidiaries deposit their cash with the header company. If it defaults, their claim is unsecured. This point is rarely raised and becomes suddenly central if one entity in the group runs into difficulty.
What centralisation really delivers
Three gains, only one of which is usually quantified in the files we see.
- Saving on debit interest: offsetting one subsidiary's overdraft against another's surplus. This is the gain always put forward, and often the smallest in value.
- Better investment terms: consolidated cash of EUR 8 million gets terms no EUR 1 million pocket obtains. The gain is proportionate to the yield difference between fragmented and consolidated investment.
- Fewer FX transactions: this is the largest gain in a multi-currency group, and the one no comparison quantifies. Netting opposite flows before converting them simply removes part of the conversions, and therefore part of the bank margins paid.
In a decentralised group, a substantial share of the FX volume traded cancels out internally: two subsidiaries buy and sell the same currency in the same month, and the group pays a bank margin on both legs. That margin is capped by no tariff schedule and can exceed 2 % depending on the product used. Netting removes it outright, without negotiating anything.
The cheapest first step
Before any banking structure, a company can capture most of the FX gain through simple internal netting: list opposite flows between entities and currencies each month, offset them on paper, and only trade the net balance in the market. That requires no banking agreement, no change of account, and can be in place within weeks.
Formal cash pooling comes afterwards, when the volume of flows justifies the infrastructure — not the other way round.
Consolidate your FX positions across entities
ELEGG FX aggregates the positions and exposures of several entities into a single view, by currency and in euros.
See the portfolio viewNot ready to open the pricer yet?
Get the one-page FX policy template — hedge ratios by horizon, tolerance band, delegation table. The one your auditor will ask for.
- One field only
- One-click unsubscribe
- No data resold
Frequently asked questions
The questions finance departments ask most often on this topic
Is notional pooling still offered?
It remains available at several institutions, but its use has declined since prudential requirements on balance offsetting in banks' balance sheets were tightened. In practice it is most often reserved for sizeable groups, in a single currency, and comes with cross-guarantees between the participating entities.
Do you need a written agreement for treasury centralisation?
Yes, always. The treasury agreement sets out the participating entities, the sweep mechanism, the interest rate on advances, repayment terms and exit provisions. Its absence exposes the group on the grounds of improper management, and leaves entities with no legal basis if the header company runs into difficulty.
Can you centralise cash from subsidiaries outside the eurozone?
Technically often yes; legally it depends on the country: some states maintain exchange controls or restrictions on cross-border lending, and others regulate the deductibility of interest paid to a non-resident related party. Each jurisdiction is examined separately before configuration.
At what volume does cash pooling become worthwhile?
There is no universal threshold, but there is a simple test: compare the expected annual gain — interest saved, better investment terms, conversions avoided — with the full cost, which includes bank structure fees, management time and legal and tax set-up fees. Below three or four entities, manual netting generally captures most of the gain at no cost.