How to Set Up a Cost Allocation

Intra-group services are activities that one member of a multinational group performs for the benefit of one or more other members. They range from routine administrative support through to technical, financial, and management services, and the resulting charges appear under many names: cost allocations, management fees, charge-outs, headquarter charges, and overhead allocations. Whatever the label, the pricing of these charges directly determines how taxable profit is split across jurisdictions, which is why they attract close examination. Charges tend to flow from operating subsidiaries toward regional or central hubs, and a tax authority reviewing the recipient’s return will ask whether the deduction is warranted at all before it asks whether the amount is arm’s length.

The Two Gates: Benefit and Pricing

Every intra-group service charge must clear two separate gates.

The first is the benefit test. A charge is only supportable where the activity provides the recipient with economic or commercial value that enhances or maintains its business position, such that an independent enterprise in comparable circumstances would have been willing to pay for the activity, or would have performed it in-house. The test is applied at the level of the recipient, not the provider. It is possible for the same activity to satisfy the test for one group member and fail it for another.

The second is arm’s length pricing. Once benefit is established, the amount charged must reflect what independent parties would have agreed. For most support services this resolves into a cost-based charge with an appropriate return, but the method must still be the most appropriate one for the service in question.

Four categories of activity commonly fail the benefit test and should be screened out before any pool is built:

Shareholder activities deserve particular attention, because they are the most frequently disputed of the four and the easiest to bundle into a charge by accident. A shareholder activity is one the parent performs because it owns the group, in its own interest as investor, rather than to provide a service the subsidiary would pay for. The classic examples are the costs of the parent meeting its own legal, reporting, and compliance obligations, such as preparing consolidated group accounts and parent-entity filings; the costs of the parent raising funds to acquire or hold its participations; the costs of the parent’s own governance, including its board, shareholder meetings, and investor relations; and the costs of group-wide oversight and control exercised to protect the parent’s investment. The distinguishing question is whether an independent party in the subsidiary’s position would have paid for the activity. It would not pay for the parent to consolidate its accounts or to police its own investment, so those costs stay with the parent and cannot be charged out.

On-call services require care of a different kind. Where a central team stands ready to provide assistance that the recipient may or may not draw on, a charge for the availability itself is supportable only where an independent party would pay a standby fee, and only where the potential need is real rather than remote.

Direct Versus Indirect Charging

There are two mechanisms for getting a service cost from provider to recipient, and the choice is not stylistic. It follows from whether the service can be traced to identifiable beneficiaries.

A direct charge applies where a specific service is provided to a specific recipient and the cost can be identified and billed on its own. Direct charging is the preferred mechanism wherever it is workable, because the service performed and the basis for the amount are transparent and easy to test. It is especially expected where the provider renders similar services to independent parties, since that situation supplies a natural comparison point.

An indirect charge applies where a service benefits several members and the cost cannot practicably be traced to each on a direct basis. Here the provider assembles a cost pool and allocates it across beneficiaries using one or more allocation keys. Indirect methods carry an estimation element, so they demand discipline: the allocation must reasonably reflect the benefit each recipient receives, and the total allocated across the group must equal the total cost the group incurred in providing the service. Over-allocation, where the sum of the charges exceeds the underlying cost, is a frequent and avoidable error.

The practical takeaway is that indirect allocation is a fallback justified by impracticability, not a default. Where a service can be charged directly, an indirect method invites the question of why tracing was not attempted.

With the concepts in place, the five steps below build a defensible charge, from delineating the service through the cost pool, the allocation keys, and the mark-up, to the documentation.

Before any number is assembled, the service itself must be accurately identified. Delineation means describing what was actually done, for whom, and why, based on the conduct of the parties rather than on labels in an intercompany agreement. A charge described only as a “management fee” with no underlying detail is the single most common trigger for full disallowance, because it gives an examiner nothing to test.

For each service category, the analysis should establish:

• What the activity is, in concrete terms (for example, running the group payroll platform, not “HR support”).
• Who benefits, identified by entity, and on what basis the benefit is expected.
• Why the recipient would pay, framed against the independent-party standard: would it have bought this externally, or performed it itself, at a price?
• Whether any part is a shareholder, duplicative, incidental, or passive-association activity, which must be removed.

Where the group elects the OECD simplified approach for low value-adding services, the benefit test is applied on a simplified basis. Services qualify as low value-adding where they are supportive in nature, are not part of the group’s core profit-earning activity, do not use or create unique and valuable intangibles, and do not involve significant risk for the provider. Routine accounting, payroll and HR administration, general IT support, and routine legal and compliance work commonly qualify. Senior management functions, research and development, manufacturing, and core selling activity do not.

Once the qualifying services are identified, the next step is to assemble the pool of costs to be charged. The pool is the foundation of the whole charge: an error here propagates through the allocation and the mark-up, so it repays careful construction. Building it well means capturing the right costs, in the right amount, from the right records, and then separating out the costs that should not be allocated across the group at large.

The three cost categories

The starting point is the fully-loaded cost of running the activity for the group. Under the OECD simplified approach and under most cost-based analyses, that base is built from three categories of cost, and the distinction between them matters when deciding what to bring in and on what basis.

The difference between overhead and indirect cost is worth stating plainly. Indirect costs belong to the service function itself but are not captured at the individual-service level, so they are brought in by a reasonable internal allocation. Overhead costs are general costs of running the business that sit above any single function, so they are apportioned to the function first before any part reaches the pool. Conflating the two tends to over-load the pool with general overhead that an independent customer would not accept in the price.

The pool should be assembled from the provider’s own accounting records, reflect the actual cost of running the activity rather than a budgeted or notional figure (unless a budgeted base is used deliberately and reconciled to actuals), and reconcile to the provider’s books, so the exercise does not quietly create cost the group never bore.

Separate single-beneficiary costs

The pool is then split. Costs relating to a service performed for only one group member are pulled out of the multi-beneficiary pool and set aside, to be charged directly to that beneficiary. Only the truly shared costs remain in the pool for allocation by key. This split prevents one recipient’s dedicated costs from being spread across members that never used the service, and it is a point examiners test, because bundling a single member’s costs into a shared pool overcharges every other recipient.

Pass-through costs

Where the provider incurs a third-party cost purely as an intermediary, and adds no value in passing it on, that cost is treated as a pass-through. It is recharged to the recipient at cost, without a mark-up. The mark-up, where one applies, attaches to the provider’s own value-adding costs, not to amounts it merely paid on to a third party. Applying a mark-up to pass-through costs inflates the charge and is a recurring dispute; the discipline is to isolate pass-throughs in the pool build so they can be recharged separately from the marked-up base.

Stock-based compensation

Where the employees delivering the service receive stock-based compensation (SBC), a question arises over whether that cost belongs in the base. It can be a material component for technology and similar groups, and the difficulty is that SBC is an accounting expense booked by the entity whose employees receive the award, while the economic cost of the dilution is borne by the issuing parent, so the entity recording the expense is often not the one that bears the real cost.

There is no single settled treatment. US rules lean toward inclusion, requiring SBC in the cost base for cost sharing and, in practice, extending that to service cost bases; the OECD leaves it to a comparability analysis of the facts; most jurisdictions have no specific guidance; and recent case law has excluded notional SBC from a service subsidiary’s base where that subsidiary bore no real economic cost. The practical rule is to follow where the economic burden actually sits, align the base with that substance, and set the position out in the documentation and the intercompany agreement before it is challenged.

For the shared pool that remains after single-beneficiary costs are removed, cost is spread across recipients using an allocation key. The key is the proxy for benefit. It must reasonably reflect the level of benefit each recipient is expected to receive, and it must be applied consistently across all recipients of the same category of service.

The appropriate key depends on the nature of the service. The guiding principle is to match the key to the underlying driver of the service need.

Time spent is the most direct key where reliable records exist, because it ties cost to actual consumption rather than to a proxy. Total revenue is a common fallback where no narrower driver fits and the benefit plausibly scales with each recipient’s revenue, but it should not be used reflexively; a key that maps more closely to the actual benefit is more defensible. Whatever key is chosen, two comparability disciplines apply. The same key must be used across all recipients of a given service category, so that the members are measured on a common basis. And the same key should be carried from year to year unless a substantive change in facts justifies switching, since changing a key introduces distortions and invites the argument that the switch was driven by tax outcomes rather than by benefit.

A cost-based charge normally carries a profit element, because an independent provider would expect a return on the service it renders, not merely recovery of cost. How that return is determined depends on whether the service qualifies as low value-adding.

Qualifying low value-adding services

Under the OECD simplified approach, a standard mark-up of five percent applies to the relevant cost, and the same five percent is used across all categories of qualifying service. The mark-up does not need to be supported by a benchmarking study, which is the central administrative benefit of electing the approach. It applies to the value-adding cost base and not to pass-through costs.

Services that do not qualify

Where a service does not qualify as low value-adding, or where the group does not elect the simplified approach, the five percent floor does not apply. The mark-up must be determined under the ordinary method rules, typically by benchmarking the net cost-plus margin against comparable independent service providers. Higher-value technical, management, and specialised services generally command a higher return than routine support, and that return has to be supported by comparables rather than assumed.

The base the mark-up applies to

The base to which the mark-up applies matters as much as the rate. The mark-up attaches to the provider’s own costs of adding value. Third-party costs passed through without added value are recharged at cost. Applying a mark-up to the full base including pass-through amounts is a common way charges are inflated, and it is a point examiners test directly.

A worked pooled allocation

Bringing the steps together, consider a group that runs a shared IT support function from a central hub, serving subsidiaries in Portugal, Norway, and Chile. The total IT support cost pool for the year is EUR 2,400,000. Of this, EUR 180,000 relates to a system migration performed solely for the Portuguese subsidiary; that cost is removed from the shared pool and charged directly to Portugal. The remaining EUR 2,220,000 is a shared pool.

The service scales with the number of users, so the allocation key is each recipient’s share of total IT users, and the service qualifies for the five percent simplified mark-up.

The Portuguese subsidiary’s total charge is the sum of its share of the pooled costs and its directly charged migration cost, each carrying the mark-up:

Two checks confirm the allocation is sound. The allocated bases sum to EUR 2,220,000, exactly the shared pool, so nothing is over- or under-allocated. And Portugal’s dedicated cost is charged only to Portugal, not spread across the three recipients.

A cost allocation is only as strong as the file supporting it. Because indirect charging rests on estimation, the documentation carries the burden of showing that the estimation is reasonable and that the benefit was real. A charge that is correct in principle but thinly documented is still exposed, because the burden of demonstrating that services were rendered and benefited the recipient sits with the taxpayer, and failure on that point tends to produce full disallowance rather than a modest adjustment.

The supporting package should include:

• A description of each service category, the identity of the beneficiaries, and the reasons each category qualifies for the treatment applied.
• The commercial rationale for providing the services within the group, and a description of the benefit each category delivers to each recipient.
• The allocation keys selected, the reasons they reasonably reflect benefit, and confirmation of the mark-up applied.
• Written intercompany agreements covering the services, consistent with the conduct actually observed.
• Calculations showing how the cost pool was assembled, how single-beneficiary costs were separated, how the keys were applied, and how the mark-up was computed.

Contemporaneous evidence that the service was actually delivered carries particular weight: deliverables, correspondence, records of time spent, and similar material distinguish a real service from a label. Generic contract language with no underlying evidence of delivery is what examiners disallow.

Common Failure Points

Most cost allocation disputes trace to a short list of recurring errors.

A consolidated worklist for the full charge. Each item should be completed and evidenced in the documentation file.