Cash Pooling in 2026: ZBA, Notional, and What Changes Before January 2027

Most finance teams describe cash pooling as a bank product. It is not. A pool is a set of contracts that change who owes what to whom, and every structure carries three costs that never appear on the facility letter: a capital cost absorbed by the bank, a tax cost absorbed by the group, and an accounting consequence that lands in your covenants.

The capital cost is the one about to move. The Prudential Regulation Authority pushed UK implementation of the Basel 3.1 standards to 1 January 2027. In the EU, CRR3 has applied since 1 January 2025, with the output floor starting at 50% and phasing to 72.5% by 2030. For any group running a notional pool, that schedule matters more than any change in bank pricing you will negotiate this year.

This is a working guide to choosing a structure, priced properly and documented before the deadline rather than after it.

The Three Structures, Described By What Actually Moves

The industry vocabulary hides a simple distinction: physical pooling moves money, notional pooling moves only interest.

Zero balance accounts (ZBA). At a fixed cut-off each day, every participating account is swept to nil and the balance lands in a master account held by a treasury header company. The subsidiary ends the day with no cash. Because money genuinely moved, the header now has a receivable from, or payable to, each participant, and those intercompany loan balances must be documented and priced.

Target balance accounts (TBA). The same sweep, but each operating account retains a buffer, typically one to three days of disbursements. Only the excess is swept up and only the shortfall is topped up. TBA is what most operating subsidiaries actually need, because a ZBA leaves a local finance team unable to pay a same-day supplier without drawing on the pool.

Notional pooling. No money moves. Each entity keeps its own balance and its own legal relationship with the bank, and the bank calculates interest on the aggregate position. A debit balance in one account is offset against a credit balance in another for interest purposes only.

The mechanics are easy to illustrate. Take a three-entity sterling group:

EntityEnd-of-day positionStandalone annual interestPooled
A+£6.0m credit at 3.2%+£192,000—
B−£4.5m debit at 6.7%−£301,500—
C−£3.0m debit at 6.7%−£201,000—
Group−£1.5m net−£310,500 net cost−£100,500 net cost

The saving is £210,000 a year before the bank's pool management fee. At a realistic £15,000 to £20,000 fee, the group keeps around £190,000. That is not a rounding error on a mid-market balance sheet, and it is why pooling survives every change in the capital rules.

Why Notional Pooling Is Scarce, and Getting More Expensive

Notional pooling is the structure treasurers prefer and the structure banks least want to provide. The reason is capital. Without a legally enforceable cross-guarantee between participants, the bank cannot treat the credit and debit balances as offsetting for prudential purposes. It must hold capital against gross exposure while earning interest on the net. Under Basel III leverage and liquidity requirements, that is an expensive trade for the bank, and it is why availability has narrowed for a decade: fewer institutions offer it, they offer it selectively, and they offer it at a price that reflects the capital consumed.

As a former EACT chairman put it candidly, notional pooling is not forbidden by Basel III, but banks must allocate more capital to it, so it will be offered less often, to fewer clients, at worse terms. In practice that has meant three things: notional pools are quoted for operational deposits rather than investment balances, quarter-end positions are often required to be physically squared off, and cross-guarantee documentation is not optional if the pricing is to be tolerable.

The 2027 UK date is the point at which the group's own documentation quality becomes a pricing variable. A pool with weak cross-guarantees is not merely a legal loose end; it is capital the bank has to fund, and it will appear in your fee.

Physical pooling has no equivalent problem. Once funds have moved, there is no offsetting question for the bank to answer. That asymmetry, not fashion, explains why ZBA structures remain the default in almost every market where they are permitted.

The Tax Layer: Where Pools Quietly Fail Audit

A pool is a set of intercompany balances, and intercompany balances are priced. Two regimes decide whether the structure survives scrutiny.

Arm's length interest. Under the OECD Transfer Pricing Guidelines, cash pooling is a controlled transaction. Members are remunerated at arm's length rates on their debit and credit positions, and a pool header or leader that performs genuine coordination functions is entitled to remuneration for them, but not to the whole spread by default. HMRC's own guidance points the same way: what must be tested is the arm's length interest on the net intercompany balances that accrue over time, not the individual daily sweeps. Groups that reconcile the pool at month end and charge interest on the closing balance rather than on the balance history are the ones that get adjusted.

Interest deductibility. In the UK, the Corporate Interest Restriction caps net tax deductions at 30% of tax-EBITDA under the fixed ratio method, with a £2 million de minimis for the group. Most mid-market pools sit comfortably below that threshold, which is a reason to check the threshold rather than assume the rules do not apply. A pool can push a group over it by concentrating interest in one entity.

Withholding tax. UK domestic law requires 20% withholding on payments of UK-source yearly interest, rising to 22% from 6 April 2027. The same rate applies to the Property Income Distribution regime. There are exemptions, including the UK-to-UK exemption for payments to a company within the charge to corporation tax, and treaty relief may reduce the rate on cross-border interest, but treaty relief is not automatic and must be claimed. In a cross-border pool, interest flowing from a local entity to a foreign header is a real cash cost unless the treaty position is documented. Dividends are the exception: the UK generally does not withhold on dividends, so repatriating cash as a dividend rather than as pool interest can change the economics entirely.

This is general information rather than tax advice. Pool documentation and pricing should be reviewed by an adviser who can see the group's treaty position and functional profile.

The Accounting Consequence Nobody Costs

A pool changes interest. It does not necessarily change the balance sheet.

IAS 32 permits offsetting a financial asset against a financial liability only where the entity currently has a legally enforceable right of set-off and intends either to settle net or to realise the asset and settle the liability simultaneously. In a notional pool, each participant's balance is with the bank, in its own name, and the interest netting is a pricing convention. The IFRS Interpretations Committee looked at exactly this structure and the principle held: an interest calculation on an aggregate balance does not by itself create a right of set-off.

So a group with £6 million of cash and a £4.5 million overdraft can report £210,000 of interest saving and still present £6 million of cash and £4.5 million of debt. Gross. If the group has a leverage or net-debt covenant, the interest saving improves the income statement while the ratio barely moves. Physical pooling is different: once the sweep has run, the subsidiary genuinely holds no cash and the header genuinely holds a receivable, so the group presents a single net position.

This is the reason some groups choose a slightly more administration-heavy structure. They are not optimising for interest. They are optimising for a covenant.

Where Pooling Is Not Available

The structure is jurisdiction-specific, and assumptions built in one market fail badly in another.

China. Cross-border pooling has been liberalised rather than abandoned. The integrated renminbi and foreign currency regime expanded through 2025 and 2026, with nationwide availability and simplified registration for eligible multinationals, but the core controls remain: quota-based limits on net cross-border flows and registration requirements. A China pool is a capital-flow instrument first and a liquidity instrument second, and it needs a bank that operates the regime routinely.

India. Cross-border pooling is not available as a general structure, and intercompany lending is governed by exchange control rules rather than by treasury preference. Local groups concentrate domestically and fund cross-border needs through permitted borrowing routes.

United States. Domestic zero-balance and target-balance pooling is routine. Cross-border notional pooling into a US structure is much less so, and state-level deposit and set-off law plus the tax treatment of intercompany interest make it a specialist exercise rather than a standard product.

Everywhere else. Resident and non-resident account restrictions, minimum shareholding requirements, mandatory local borrowing and thin capitalisation rules each change what a pool can legally do. Every structure should be tested locally before it is designed globally.

Net First, Pool Second

Pooling addresses the interest cost of fragmented balances. It does not address the cost of moving money between your own entities, and that is usually the larger number in a mid-market group.

Before designing a pool, count the intercompany legs. A five-entity group generates up to twenty bilateral pairs, and each pair settling through an external rail carries a fee, a conversion and a settlement delay. Multilateral netting compresses those flows to a handful of net settlements per cycle. Where entities hold balances inside one account structure rather than separate banking relationships, an intercompany payment becomes an internal book transfer rather than a rail transaction, which removes the fee entirely.

The sequencing matters because it changes the size of the problem the pool has to solve. Reducing the number of external legs reduces the balances that end up stranded in transit, and it reduces the number of accounts the pool must include.

A Decision Framework by Group Shape

Group profileRecommended structureRationale
Single currency, two to five entities, one countryDomestic ZBA with TBA buffersCheapest to run, no tax or transfer pricing exposure beyond arm's length interest
Multi-currency, five to twenty entities, EU plus UKZBA per currency plus a multi-currency overlay headerPhysical sweeps avoid the capital-pricing problem entirely
Large net surplus, strong bank relationship, sophisticated tax functionNotional pool with enforced cross-guaranteesPreserves local autonomy and avoids physical transfers, but only justifies the fee at scale
Cross-border with China, India or Brazil exposureLocal concentration plus a permitted overlayThe structure is set by local regulation, not by preference
Group under a net-debt covenantPhysical poolingPresentation nets down; notional pooling will not move the ratio

Two rules of thumb are worth holding onto. If the combined interest saving is less than three times the pool management fee plus the internal cost of administering it, the structure is not worth building. If the group cannot produce a monthly arm's length interest calculation per participant without manual work, the structure is not ready to scale.

A Ninety-Day Implementation Sequence

Days 1 to 30: measure. Pull twelve months of balances by entity, currency and account. Calculate the saving a pool would have produced, using actual historical balances rather than a forecast. Identify which pairs of entities were structurally long and short at the same time, because that is the only saving that is real and recurring.

Days 31 to 60: design and document. Choose ZBA or notional on the evidence from the measurement, not on preference. Draft the pool agreement, the cross-guarantee if a notional structure is chosen, and the intercompany loan agreements with an explicit interest rate basis. Confirm the treaty and withholding position on interest flows before signature, not after.

Days 61 to 90: operate and govern. Set sweep cut-offs that match local payment deadlines. Agree a monthly cycle that produces the arm's length interest calculation per participant, the balance history behind it, and a reconciliation to the general ledger. Define who can override a sweep and under what authority, because the first exception request is the moment a pool stops being a control and becomes a convenience.

What to Do Before January 2027

The PRA deadline is not a cliff edge for most groups, but it is a deadline for a conversation. Any group running a notional pool should ask its bank, in writing, three questions: how the structure is priced after the new rules apply, what documentation the bank requires to treat the balances as offsetting, and whether the facility will be renewed on the same basis at all.

The answers will differ by institution, and some will be unwelcome. A group that discovers in December 2026 that its notional pool is being repriced or withdrawn has no time to build a ZBA structure, document intercompany loans and reset its tax position. A group that asks in the first quarter of the year has two clear quarters to migrate.

Pooling is not a feature of an account. It is a legal architecture, and the groups that get value from it are the ones that designed it deliberately, priced it honestly, and documented it well enough to survive both an audit and a capital rule change.