Groups that operate across the Benelux usually end up with three entities before anyone has decided that they should. A Dutch company because the first European customer was Dutch. A Belgian company because the sales team is in Antwerp. A Luxembourg company because an investor asked for it. Each was created for a reason. None was created as part of a plan for where the group’s profit should end up.
The question presents itself the first time the group is profitable enough to matter: which entity should show the margin. It is usually asked as a rate question and it is not one. It is a question about what each entity does, what it owns and what can go wrong for it, and the three administrations apply the same test to that question and read each other’s answers.
The rates are closer than the argument suggests
Start by removing the incentive to shop. The Belastingdienst gives the Dutch rates in force in 2026 as 19 per cent on taxable profit up to 200,000 euro and 25,8 per cent above that. The Administration des contributions directes publishes the aggregate nominal burden of a resident company seated in the commune of Luxembourg at 23,87 per cent from the 2025 tax year, being corporate income tax at 16 per cent under article 174, the employment fund surcharge of 7 per cent on that tax, and municipal business tax of 6,75 per cent. Article 215 of the Belgian income tax code sets 25 per cent, with 20 per cent on the first tranche of 0 to 100,000 euro for companies treated as small; the crisis contribution formerly in article 463bis was repealed with effect from assessment year 2021, so 25 per cent is the whole rate.
The spread is therefore about two points between the three at the top of the scale. A group that moves a function across a border to capture that has taken on a permanent documentation burden, a permanent governance burden and a real risk of double taxation, in exchange for a margin that a single bad year erases.
The Belgian reduced rate carries a condition worth knowing before relying on it. Article 215 withdraws the 20 per cent bracket from, among others, a company that does not pay at least one company director a remuneration charged to the result of the taxable period of at least the indexed threshold, which for income year 2026 is 51,000 euro. A Belgian entity left without a properly remunerated director loses the reduced rate the group was counting on.
The rate is therefore a tiebreaker, not a driver. Where two allocations are both defensible on the facts, it can decide between them. Where only one is defensible, it is irrelevant.
Functions, assets and risks, not the organization chart
The allocation is settled by what each entity actually does. An entity that develops the product, decides which markets to enter and carries the cost of failure has a claim on the residual profit. An entity that sells into a defined territory using a brand and a product it did not create has a claim on a routine return for selling and no more.
The most common error is to let the legal chart drive the answer. The Luxembourg company is the parent, therefore it keeps the profit. That reasoning does not survive contact with any of the three administrations. Ownership of shares is not a function, and a holding company that neither develops nor sells nor bears risk has no basis for a residual return however many subsidiaries it owns.
The second most common error is the reverse. The Dutch company employs most of the people, therefore it keeps everything. That fails too, if the Belgian entity has built the customer relationships and bears the credit risk on them. Headcount is evidence of function; it is not the function.
What settles the question is a sober description of who decides what. Who signs off on pricing. Who decides to enter a market. Who carries the inventory and the receivables. Who would bear the loss if a major customer left. Write those answers down before drawing any conclusion about margins, and the allocation usually becomes obvious.
The three statutes say the same thing
There is no arbitrage in the drafting. Article 8b of the Wet op de vennootschapsbelasting 1969 provides that where bodies participate directly or indirectly in the management, control or capital of one another and agree conditions in their mutual relations differing from those independent parties would have agreed, the profits of those bodies are determined as if the arm’s length conditions had been agreed. Paragraph 3 requires them to keep in their records the data showing how those prices came about.
Article 56 of the Luxembourg income tax law is drafted in almost identical terms, following the treaty language: where enterprises are linked in their commercial or financial relations by conditions differing from those independent enterprises would agree, the profits are determined on the conditions prevailing between independent enterprises and taxed accordingly. Article 56bis sets out the definitions and the comparability analysis in statutory form.
Article 185, paragraph 2, of the Belgian income tax code says the same for two companies of a multinational group in their reciprocal cross-border relations, and adds the corresponding adjustment: where the other state has taxed the same profits on arm’s length terms, the Belgian profits are adjusted appropriately.
The consequence is that a position that fails in one of the three fails in all three. One analysis serves everywhere, and there is nowhere to hide a weak one.
The residual profit is where groups get it wrong
Routine functions are rarely the problem. A distribution entity earning a modest operating margin on third party sales, or a service entity charging cost plus a mark up, is accepted almost everywhere if the comparables are decent and the agreements match the conduct.
The dispute is always about the residual, the profit left after each routine function has been paid. It belongs to whoever controls the significant risks and owns or controls the assets that generate the excess return. If that is the Dutch operating company, the Luxembourg parent cannot take the residual by contract. Writing an intellectual property licence does not move economic ownership where the licensor has no capacity to develop, enhance, maintain, protect or exploit the asset.
This is where Benelux structures most often fail examination. The residual sits in an entity with two directors, a service agreement and no operational staff, and the file explains that the entity bears the entrepreneurial risk. The administration asks who inside it has the seniority and the information to take the decisions that risk implies. If the answer is nobody, the allocation is rewritten.
Intragroup interest is priced by the same rule
Intragroup debt is often used to move margin without moving function, and it is subject to the same provisions as everything else. The rate has to be one a third party would have charged that borrower on those terms. A rate set by reference to the group’s cost of funds rather than the borrower’s own credit standing is the first thing an examination tests.
The lender also has to be able to bear the risk it is paid for. An entity funded by its own parent, with no capital of its own and no capacity to assess or manage credit risk, is not entitled to a lender’s return. It is entitled to a fee for arranging, a different and much smaller number.
The documentation that has to exist
The Dutch obligation is specific. Article 29g of the Wet op de vennootschapsbelasting 1969 requires a Dutch group entity of a multinational group to include a master file and a local file in its records within the deadline for its corporate income tax return, in Dutch or English, where the group achieved at least 50,000,000 euro of consolidated group revenue in the preceding reporting year. The country by country report under article 29c is separate and applies above 750,000,000 euro.
Belgium sets its own thresholds. Under articles 321/4 and 321/5 a Belgian constituent entity files a master file within twelve months of the reporting period and a local file with its return, where it exceeds any one of 50 million euro of operating and financial income, a balance sheet total of 1 billion euro, or 100 full-time equivalents.
Below those thresholds the documentation obligation is lighter but the substantive obligation is unchanged. Article 8b, paragraph 3, requires the data showing how prices came about to be in the records regardless of size. A group with 20 million euro of revenue and three Benelux entities has no master file obligation and still has to explain its pricing.
In practice the file that matters is not the formal one. It is the set of intercompany agreements, board minutes and management reporting showing the same allocation of decisions the transfer pricing report asserts. Where those three sources disagree, the report loses.
Small countries, long memories
The Benelux administrations exchange information routinely and know each other’s practice well. A position taken in one is visible in the others, and an adjustment in one produces a claim for a corresponding adjustment in another. That mechanism is supposed to prevent double taxation, and it works, slowly.
The practical point is that an aggressive allocation does not produce a saving. It produces an adjustment in the country that loses profit, a claim for relief in the country that gained it, and a mutual agreement procedure that takes years while the tax sits paid. Groups underestimate the working capital cost of that outcome, which is usually larger than the rate difference they were pursuing.
The corollary is that a well documented allocation is worth more than a clever one. Where the file is consistent and the conduct matches it, an examination ends without an adjustment and the group carries no provision. That is the actual objective, and it is not the same as minimizing the current year charge.
Deciding it once, properly
The exercise is not complicated and it is rarely done. Describe each entity’s functions in plain language. Identify who controls each significant risk and who has the financial capacity to bear it. Determine which entity owns or controls the assets that generate the excess return, and test that against who actually manages them. Price the routine functions against comparables. Leave the residual where the analysis puts it.
Then align everything else to that conclusion: the agreements, the board composition, the reporting lines, the location of the people who decide. Where the conclusion is uncomfortable, because it puts the profit in the country with the higher rate, change the facts deliberately if the business case supports it, and accept the conclusion if it does not.
Groups that do this once, when the third entity is created, spend a fraction of what groups spend correcting it later. The correction is expensive not because the analysis is hard but because by then there are filed returns in three countries saying something different, and each has to be explained.