Of all the transactions a Dutch inspector may examine, none attracts attention faster than a royalty. The reason is structural: a royalty can be set at almost any level, is rarely benchmarked against anything a third party actually pays, and moves residual profit rather than a margin on costs. It is also where the gap between the paperwork and the facts is widest. A licence agreement takes an afternoon to sign; the substance that makes it defensible takes years.
legal ownership is where the analysis starts, not where it ends
Legal ownership determines who can license, who can sue, who can assign. What it does not determine, on its own, is who is entitled to the return the intangible generates.
The OECD Transfer Pricing Guidelines, which the Dutch authorities apply as the interpretive framework for article 8b of the Wet Vpb 1969, separate two questions groups habitually conflate: who owns the asset, and who performed and controlled the functions that created, sustained and monetised its value while bearing the associated risks with the capability to manage them. Only the second determines profit entitlement. An entity holding title but performing none of the relevant functions is a funder at best, and a funder is entitled to a funding return, not to the residual.
Legal ownership determines who can sign the licence. It does not determine who is entitled to keep what the licence produces.
what the DEMPE functions actually test
Development, enhancement, maintenance, protection and exploitation are not a checklist to be ticked. They are five questions about where decisions are taken.
- Development. Who sets the research agenda, approves the budget, decides which projects continue and which are abandoned?
- Enhancement. Who owns the roadmap, prioritises features, decides what is a new version rather than a patch?
- Maintenance. Who is accountable for keeping the asset operational and commercially current?
- Protection. Who instructs counsel on filings, renewals, oppositions and infringement actions, and who funds them?
- Exploitation. Who decides pricing, licensing terms, market entry and channels?
The distinction that matters most is between performing a function and controlling it. A Dutch entity may outsource development to a related party and still be the DEMPE owner, but only with genuine control: people competent to assess the work, with authority to stop or redirect it, who in practice do so. Approving invoices is not control, and nor is signing minutes drafted elsewhere. Control requires decision-makers who could have decided otherwise.
Risk assumption follows the same logic. Contractual allocation of development risk to a Dutch company is respected only where that company can manage the risk and bear its financial consequences. Where it can do neither, the risk is reallocated, and its return goes with it.
the Dutch entity that owns the IP while the developers sit elsewhere
This pattern is usually the product of history rather than design. A group incorporates a Dutch holding or licensing company, moves the IP into it, and continues to run engineering from the country where the team has always been. The Dutch entity signs licences with the operating subsidiaries, receives royalties, and pays the development entity a cost plus service fee.
On paper the structure is coherent. Under a DEMPE analysis it frequently is not. If engineering leadership, product decisions, release approvals and the roadmap all sit abroad, the Dutch entity controls almost nothing. Its entitlement reduces to whatever compensates the functions it genuinely carries out, typically funding and, where real, the management of financial risk. The residual belongs elsewhere.
The remedy is not cosmetic: a board resolution reciting decision-making does not create it. Where a group intends the Dutch entity to be the economic owner, that intention must be visible in who is employed there, what they are qualified to decide, and what evidence exists that they decided it. This is the substance question running through any Dutch holding structure, but the stakes are higher with intangibles because the profit at issue is residual rather than routine.
one royalty rate, five different transactions
A single percentage applied to licensee revenue is convenient and, in most technology groups, wrong. The licensee receives a bundle:
- a licence to use the software or patented technology;
- maintenance: updates, patches and version upgrades;
- technical support, from helpdesk to escalation engineering;
- access to a hosted platform, with the underlying cloud costs;
- use of the group trade mark.
These are economically different transactions. The licence element is a return on an intangible, tested by CUP where comparable licences exist or by profit split where value is genuinely co-created. Maintenance and support are services with an identifiable cost base, normally tested on cost plus or TNMM. Platform access is closer to a pass-through of infrastructure costs, and trade mark use warrants a separate, much smaller rate.
Bundling them into one rate makes benchmarking almost impossible, since no external comparable covers the same bundle, and invites the inspector to test the whole payment against its weakest component.
a worked example: an 8% royalty with two people behind it
Consider a software group with consolidated turnover of EUR 140 million, so within the Master File and Local File obligation, which begins at EUR 50 million of consolidated group turnover. A Dutch BV holds legal title to the platform and employs a managing director and a financial controller. Development is carried out by 55 engineers in a foreign group company, which invoices the BV on a cost plus 6% basis.
The operating subsidiaries pay a royalty of 8% on third-party revenue of EUR 100 million, so EUR 8 million a year. The BV’s costs are the development service fee of EUR 5.3 million (a cost base of EUR 5 million plus 6%) and EUR 0.2 million of overhead. Leaving aside amortisation of the capitalised platform, Dutch taxable profit is EUR 2.5 million; tax at 25.8% is approximately EUR 645,000.
Now apply DEMPE. The engineering director, the product owners and the release board sit abroad. The Dutch managing director approves budgets prepared elsewhere. On these facts the BV controls no development, enhancement or maintenance function. What it does provide is funding: EUR 25 million was injected to acquire and capitalise the platform, and that treasury decision was genuinely taken in Amsterdam.
A funder that controls financial risk but not development risk is entitled to a risk-adjusted return on the funding, not to the residual. If that return is benchmarked at 4%, the BV’s arm’s length profit is approximately EUR 1 million rather than EUR 2.5 million. Roughly EUR 1.5 million of annual profit is reallocated to the development jurisdiction. Over a five-year review period, some EUR 7.5 million of profit is in dispute, on which Dutch tax of about EUR 1.9 million has already been paid. The figures are illustrative; the funding return and the outcome turn on the facts of each case.
The 8% itself also fails to survive unbundling. Once the flows are separated, the analysis might attribute 3 points to the licence, 2 to maintenance, 1.5 to hosting, 1 to support and 0.5 to the trade mark. Only the first and last are royalties; EUR 4.5 million of the annual charge is consideration for services and infrastructure, priced on a cost base rather than as a share of licensee revenue, and characterised differently at source.
the adjustment risk runs in one direction
The uncomfortable part of that example is not the reallocation but the asymmetry of the correction: a group that identifies the problem cannot simply reduce its Dutch taxable base. Since 1 January 2022, the Dutch rules against transfer pricing mismatches restrict downward adjustments in the Netherlands where there is no corresponding upward inclusion abroad. The regime that ended informal capital applies here too: the deduction follows the pick-up in the other jurisdiction, or it does not happen.
So the foreign authority may assess profit that the Netherlands has already taxed, with relief depending on a correlative adjustment through mutual agreement procedure or arbitration, a process measured in years. The exposure is not the Dutch rate on the disputed amount; it is the prospect of paying tax twice while the two authorities discuss the file.
Article 8b compounds this. It is not only a pricing standard; it carries a standing obligation to hold documentation showing how the price was determined, with no turnover threshold at all. The EUR 50 million and EUR 750 million thresholds govern the Master File, Local File and country-by-country report, not the underlying duty, as we set out in the foundational article of this series. A group that cannot produce a contemporaneous DEMPE analysis is materially weaker when the burden of proof is discussed.
Characterisation matters at the other end too: royalties paid from the Netherlands to affiliated entities in low-taxed or listed jurisdictions, or in defined abusive situations, fall within the Dutch conditional withholding tax, levied at the highest corporate income tax rate of 25.8%.
what a defensible intangibles file contains
A file that survives scrutiny is rarely the longest. It answers the DEMPE questions with evidence: organisation charts showing where decision-makers sit; board and steering committee materials recording decisions taken rather than ratified; a functional analysis distinguishing performance from control; an allocation of the royalty into its component streams; and a documented view on why the chosen method, whether CUP, TNMM, resale price, cost plus or profit split, suits each.
It also states honestly what the Dutch entity does not do. Files claiming too much are easier to dismantle than files claiming the right amount. Where the entity is a funder, saying so and pricing the funding correctly is stronger than calling it an entrepreneur it cannot evidence. Groups weighing where intangibles should sit across European holding jurisdictions should begin from the same premise: the entity that holds the IP must be the entity that can staff and govern it, or it will be priced as though it cannot.
Montclare has published a short self-assessment, the Transfer Pricing Readiness Check, which sets out ten questions that identify where a group’s exposure sits. It can be downloaded from our transfer pricing page.
This article is informational and does not constitute tax advice. Each engagement is subject to scope and applicable regulation.