A group can build a transfer pricing position with genuine care, document it properly, benchmark it against real comparables, and still not know whether the tax authority agrees. It will find out years later, during an audit, when the transactions in question have already been repeated four or five times. That asymmetry is uncomfortable, and Dutch law offers a way out of it: an agreement with the authority reached before the transactions happen rather than after.
What an advance pricing agreement actually is
An advance pricing agreement, or APA, is a binding arrangement between a taxpayer and one or more tax authorities that fixes, in advance, the transfer pricing method to be applied to defined related-party transactions over a defined period. It does not fix the profit. It fixes the method, the critical assumptions on which the method depends, and the conditions under which the agreement remains valid, which is exactly what the Besluit vooroverleg rulings met een internationaal karakter requires the settlement agreement to record.
That distinction matters more than it first appears. The authority is not guaranteeing an outcome; it is agreeing that a particular way of arriving at the outcome is acceptable. If the business changes materially, the critical assumptions fail and the agreement can lapse.
The Netherlands also issues advance tax rulings, which address the tax treatment of a structure or transaction more broadly. In practice the two instruments are administered together and requested from the same international certainty team at the Belastingdienst.
The Dutch ruling practice, and how it changed
The Netherlands has issued rulings for decades, and that practice was for a long time one of the reasons international groups chose the country. It is also the reason the practice attracted political attention, and the policy that applies today is considerably stricter than the one that built its reputation.
Four conditions now shape what is obtainable. There must be genuine economic nexus with the Netherlands, meaning real activity, real people and real decisions, not a registered address. No ruling is given where saving Dutch or foreign tax is the sole or the decisive motive for the arrangement. No ruling is given where the certainty sought concerns direct transactions with entities established in a state named in the Dutch regulation designating low-taxed states and non-cooperative jurisdictions for tax purposes. And no ruling is given where the applicant, a director of the applicant, or an ultimate beneficial owner or intermediate holding company with a direct or indirect interest of five per cent or more appears on the EU sanctions list, the one bar of the four to which the decree allows no exception at all.
The nexus condition is more specific than its name suggests. The decree requires that the requesting entity belongs to a group carrying on real operating activity in the Netherlands, that such activity is carried on for the account and risk of the requesting party, that sufficient relevant personnel for it is present in the Netherlands at group level, and that the activity fits the entity’s function within the group. Those elements are cumulative, and the last of them catches the company whose staff are real but whose functions do not match the income it reports. The condition falls away entirely where the certainty sought is that there is no Dutch taxable presence at all under Articles 17 or 17a of the corporate income tax act.
Rulings are granted for a limited term: at most five book years in principle, and a maximum of ten where the facts justify the exception, long-running contracts being the example the decree gives, with an evaluation halfway through. Anonymized summaries are published, and the Belastingdienst posts them case by case, for requests refused, withdrawn or set aside as well as for agreements reached. A group requesting certainty should assume that the substance of what it agreed will be visible, in general terms, to anyone who looks.
Certainty is available, but it is no longer available to structures that need it most in order to survive scrutiny.
This is not a weakening of the instrument. For a group with real operations it is an improvement, because an agreement obtained under a credible regime is worth considerably more than one obtained under a permissive one.
Unilateral, bilateral, multilateral
A unilateral APA involves only the Dutch authority. It removes Dutch uncertainty and nothing else. The counterparty jurisdiction is not bound by it and remains free to take a different view of the same transaction.
A bilateral APA is concluded between two authorities under the mutual agreement procedure of the applicable tax treaty, a provision that the ruling decree notes usually follows Article 25 of the OECD Model Tax Convention. It is slower and more demanding, and it is the only version that genuinely eliminates the risk of the same profit being taxed twice, because both sides have agreed to the same method in advance.
A multilateral APA extends the same logic across three or more jurisdictions. It is rare and reserved for structures where the value chain genuinely runs through several countries.
Neither cross-border form is simply available on request. The decree states that the Belastingdienst directs its effort in principle at a bilateral agreement but cannot impose one, that a treaty containing a provision comparable to Article 25 must exist with the other state, and that the other state must be willing to hold the discussion in advance. A multilateral request that meets objections is treated as a request for several bilateral agreements, and the MAP team notifies the applicant of the split. Where the Dutch entity has both inbound and outbound related transactions, a critical assumption can take account of transfer pricing adjustments from countries outside the agreement.
The choice between them is a question of where the exposure sits. If the Dutch entity transacts principally with one foreign affiliate and the amounts are significant, unilateral certainty is half a solution.
What a request actually requires
Groups often assume an APA request is an exceptional exercise requiring exceptional material. It is not. The core of the file is the same functional analysis, the same benchmarking study and the same intercompany agreements that Article 8b already requires every group to hold. What changes is the standard of presentation and the fact that the authority reads it before forming a view rather than after.
Beyond that, the request sets out the transactions to be covered, the proposed method and why it fits, the critical assumptions, and the term sought. Where an intangible sits at the centre of the arrangement, the DEMPE analysis will carry most of the weight of the discussion.
The route matters as much as the file. A pre-filing meeting with the first handler is available before the request goes in, and no substantive position is taken during it. For an APA the Behandelteam IFZ is first handler rather than the local inspector, and no international ruling issues until the College IFZ has approved it, which is why the settlement agreement carries at least two Belastingdienst signatures. Bilateral and multilateral requests go to the MAP team instead. The decree handles an APA request partly on paragraphs 4.134 to 4.176 of the OECD Guidelines and always starts from the method the taxpayer proposes, under paragraph 4.9, and those Guidelines carry their weight through a Council Recommendation asking adhering jurisdictions to follow them as a whole rather than chapter by chapter.
The documentary list is specific. Alongside the facts and an unambiguous position on their consequences, the request carries the entities and permanent establishments involved, the other countries concerned, the worldwide organisational and shareholder structure with its NACE Rev. 2 code and every beneficial owner or intermediate holding company at five per cent or more, and a declaration matching that against the Dutch trade register. An APA adds the master file where one is required, the financial data, functions, assets and risks of the entities involved, the financial stake over the term, the comparability analysis, the assumptions and the effect of changes in them, the book years covered, a declaration that none of those persons is on the EU sanctions list, and a draft of the standard form used to exchange the ruling abroad. The third paragraph of Article 8b already obliges associated entities to hold in their records the data showing how their transfer prices came about.
A group that already maintains its documentation properly is most of the way there. A group that does not will discover during the process how much it was missing.
Duration, rollback, and the day the facts change
The applicant states in the first instance the term for which it wants certainty, and the decree answers with five book years as the principle and ten as the exception where the facts justify it, with an evaluation halfway through the longer term. It sets out no renewal mechanism, so certainty beyond the term means a further request, measured against the law, the policy and the case law standing at that later moment. The Belastingdienst reports that new or amended international tax legislation, and relevant case law, reliably produce fresh waves of requests in the years just after they take effect, which is another way of saying the ground under an expiring agreement has usually moved.
Rollback, the application of the agreed treatment to transactions that have already happened, is not the same instrument in each form. In a bilateral or multilateral agreement the competent authorities can agree that the treatment also covers transactions already concluded. In a unilateral APA it reaches them only where the relevant facts and circumstances in the elapsed period are comparable to those underlying the request, and only where the result is not that part of the profit ends up taxed in no country at all. As a matter of the decree’s own scope rather than of rollback, it applies as well to pricing arrangements made before the return is filed for the period in which the transactions took place.
Facts change during a term, and the agreement is written to absorb that. It must record how changes in facts and circumstances are handled, the conditions the taxpayer must keep meeting and the procedures ensuring it does, the critical assumptions on which it rests, and a clause under which it loses validity immediately on a relevant change in legislation or in rules based on legislation, later changes in policy or case law also being capable of ending it. One further clause is the one groups underestimate: the agreement lapses if the agreed pricing or method is not actually recorded in the contracts with the related entity, or is not actually paid or received, unless otherwise agreed.
A worked example: when the arithmetic favours certainty
Consider a group whose Dutch entity licenses technology to affiliates across Europe, generating royalty income of twelve million euro a year. The pricing rests on a benchmarked rate, and the position is defensible but not agreed.
Five years later the arrangement is reviewed and the authority takes the view that the rate should have been lower, reallocating two million euro a year of profit. Across the five years in which the inspector can still raise an additional assessment, that is ten million euro of additional Dutch taxable profit, generating roughly 2.6 million euro of corporate income tax at the 25.8% headline rate that applies above two hundred thousand euro of profit, plus interest running from each original due date.
That window is not five years in every case. The same provision lengthens it by any filing extension granted, and substitutes twelve years where the tax was underpaid on a component of the taxable object held or arising abroad. For intangible income with foreign elements, five years is a floor rather than a ceiling.
The corresponding downward adjustment abroad is not automatic. Obtaining it means opening a mutual agreement procedure, which between member states is available under Council Directive (EU) 2017/1852 on tax dispute resolution mechanisms in the European Union. It takes years even against the average of twenty four months that the BEPS Action 14 minimum standard sets as the target, and it consumes advisory fees throughout. Until it concludes, the group has paid tax twice on the same ten million euro.
Against that, an APA covering the same royalty flow costs professional fees and, on the gross lead times the Belastingdienst published for 2025, eleven months of process for a unilateral agreement and thirty nine for the bilateral and multilateral ones the table reports together, and it produces a position both sides accept before the money moves. The comparison is not between the cost of an APA and zero. It is between the cost of an APA and the discounted cost of a dispute the group cannot control the timing of.
When it is not worth it
An APA is a poor instrument for small, routine, low-value transactions. The process consumes management time and professional cost that a modest service recharge will never justify, and a well-documented position under ordinary rules is entirely adequate for most intercompany flows.
It is also a poor instrument for a group unwilling to be transparent. The request requires disclosure of the commercial reality, not a curated version of it. A group whose pricing depends on the authority not looking too closely should not invite the authority to look closely.
The instrument earns its cost in four situations: recurring transactions of significant value, intangibles at the centre of the model, a business restructuring that moves functions or risks between jurisdictions, and a group that has already been through an adjustment and does not intend to repeat the experience.
The trade the group is making
An advance pricing agreement is best understood as a purchase. What is bought is certainty, and specifically the removal of a contingent liability that would otherwise sit on the balance sheet for the length of the statute of limitations. What is paid is professional cost, a process measured in months for a unilateral agreement and in years for a bilateral one, and full visibility of the commercial arrangement to the authority.
That visibility has a defined shape. A summary is published of every international ruling and of every case where the discussion was held but produced no ruling, the latter with an explanation, each anonymised so that third parties cannot trace it to an individual taxpayer. In 2025 every summary went up within three weeks of the agreement being formalised or the request being refused, withdrawn or set aside, and the core information also reaches foreign tax administrations on an internationally prescribed template. The agreement itself records the taxpayer’s undertaking not to invoke the exception clauses of Article 14(2)(e) of the international assistance act for the information it contains.
For groups whose structure reflects genuine activity, that is usually a good trade, and the transparency costs them little because there is nothing in the arrangement that depends on obscurity. For groups whose structure does not, the price is unpayable, which is precisely the outcome the current policy intends.
The question worth asking is not whether an APA is available. It is whether the group would be comfortable showing the authority exactly how its intercompany pricing works, today, without preparation. If the answer is yes, certainty is purchasable. If the answer is no, that discomfort is the finding, and it will not improve by waiting for the audit.
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.