Interest on related-party debt: Article 10a, the fiscal unity and the per-element approach

The judgments

The question

Article 10a of the Corporate Income Tax Act 1969 (CITA) is the oldest Dutch rule against profit stripping through intra-group debt. It refuses a deduction for interest on a loan from a related entity where the loan is connected with certain “tainted” transactions, such as a capital contribution into, or the acquisition of, a related company. The taxpayer escapes the rule by showing business reasons or sufficient taxation of the interest at the lender. For foreign-owned groups with Dutch holding or acquisition companies, and for Dutch groups investing abroad, the rule decides whether interest on acquisition debt reduces the Dutch tax base at all.

The judgments reviewed here answer two connected questions. First, does EU freedom of establishment allow the Netherlands to apply Article 10a where a purely domestic group could have avoided it by forming a fiscal unity, which only Dutch-resident companies can join? That is the per-element question decided in X BV and X NV (C-398/16) and the Supreme Court’s final judgments of 19 October 2018. Second, once Article 10a applies, what does EU law require of the test itself: may the deduction be refused in full where the interest rate is at arm’s length? That is the question in X BV (C-585/22), decided by the Supreme Court on 16 January 2026, followed on 1 May 2026 by a judgment on fraus legis where Article 10a did not reach the whole structure.

The legal framework

Article 10a. In its current form, Article 10a(1) refuses a deduction for interest, including costs and currency results, on debts owed directly or indirectly, in law or in fact, to a related entity or related natural person, to the extent the debt is connected with (a) a profit distribution or repayment of capital to a related party, (b) a capital contribution into a related entity, or (c) the acquisition or extension of an interest in an entity that is related after the acquisition or extension. Article 10a(3) contains the counter-proof: under point (a) the taxpayer makes it plausible that the debt and the connected legal act are “based predominantly on business considerations”; under point (b) the taxpayer makes it plausible that the interest bears, on balance, a tax on profit that is reasonable by Dutch standards (at least 10% on a profit computed under Dutch rules). Under point (b), the inspector may then still show that the debt or the legal act is not based predominantly on business considerations. Article 10a(4) defines related entities by a one-third interest.

The judgments apply older wordings, and the differences matter:

• 2004 (X BV, C-398/16). The rule was then Article 10a(2)(b): interest on loans from related entities connected with “an acquisition — including the paying up — of shares … in a related entity, except in so far as a change is made to the ultimate ownership or the ultimate control of that entity” (CJEU, para. 3). The counter-proof in Article 10a(3) had no 10% threshold and no inspector’s counter-proof.
• 2007 (X BV, C-585/22). From 1 January 2007 the tainted transactions were regrouped in Article 10a(1)(a) to (c), and point (c) for the first time covered external acquisitions of companies that become related only through the acquisition. The 10% benchmark was written into Article 10a(3)(b). The Supreme Court annexed the 2007 text to its referral judgment of 2 September 2022.
• From 1 January 2008 the inspector’s counter-proof in Article 10a(3)(b) applied, as HR 2026:736 explains (para. 4.2.2). Our current text contains that counter-proof, together with later additions not found in the 2004 or 2007 texts, such as the words in point (a) about debts in fact owed to an unrelated person, Article 10a(8) and the reference to Pillar Two top-up taxes in Article 10a(11).

The fiscal unity. Under Article 15(1) a parent and a 95% subsidiary may, at their joint request, be taxed “as if they were one taxpayer”: the activities and assets of the subsidiary form part of those of the parent. The judgments refer to Article 15(3)(c) (both companies established in the Netherlands), Article 15(5) (start no earlier than three months before the request) and Article 15(6)(d) (ending at joint request) in the 2004 and 2008 texts. In our current text those rules are in Article 15(4)(c), Article 15(9) and Article 15(10)(h), and Article 15(1) now also requires 95% of the voting rights, profits and assets. Consolidation eliminates loans, capital contributions and intra-group transactions between the members (HR 2016:1350, para. 2.8.2.6). That elimination is what made Article 10a invisible inside a domestic fiscal unity: a loan used to fund a contribution into a member of the unity no longer relates to any visible contribution.

The participation exemption. Article 13(1) leaves out of account benefits from a participation and the costs of acquiring or disposing of it. Currency gains and losses on a participation fall within the exemption (HR 2016:1351, para. 2.4.1). This matters for the companion case X NV.

X BV: the 2004 facts and the referral

X BV was a Dutch holding company in a Swedish group and the parent of a Dutch fiscal unity. On 26 April 2004 the group announced its plan to take its listed Italian subsidiary C off the Italian stock exchange; 28.81% (7,416,000 shares) was held by the public, which at €32 a share required €237,312,000. On 29 April 2004 the group’s adviser incorporated an Italian company with €10,000 capital; on 3 May 2004 it was converted into G SpA, which issued shares with a nominal value of €110,000 to X BV, on which X BV paid €237,302,000. The adviser transferred his shares in G to X BV on the same day for €10,000. G acquired most of the public float through bids in 2004 and 2005. X BV funded the payment with a loan of €237,302,000 from B AB, the group treasury company, which operated for the account and risk of the Swedish top holding A. Interest for 2004 was €6,503,261. A was subject to Swedish corporate tax at 28% but paid none for 2004 because group contributions shifted its profits to loss-making Swedish group companies (HR 2016:1350, paras 2.1.1 to 2.1.8). The inspector refused the deduction by an additional assessment.

The Court of Appeal of ’s-Hertogenbosch (4 December 2014) found that the adviser bore no risk on his €10,000 and that the economic interest in G lay with X BV from the start, so the contribution did not change the ultimate interest in G. It treated the contribution as a tainted transaction under Article 10a(2)(b), held that the involvement of X BV and the route chosen were almost entirely tax-driven, tested the compensating tax at A’s level and treated the Swedish group contributions as offsetting of losses that defeated it. It assumed a restriction of freedom of establishment but held it justified (paras 2.3.1 to 2.3.5).

On 8 July 2016 the Supreme Court rejected ground I (economic interest, a factual finding) and ground II (that Article 10a only reaches cases where fraus legis would otherwise apply; para. 2.5). It upheld ground III. The taxpayer has a free choice of how to finance a company, but Article 10a’s double business test can still catch the financing structure (para. 2.6.3). A group is, however, free to place its activities in a Dutch company, even for tax reasons, and parliamentary history does not show that Article 10a was meant to limit that freedom: the fact that a Dutch company was interposed for tax reasons is therefore irrelevant to the business test (paras 2.6.3 and 2.6.4). What mattered was whether there were business reasons to take C private through an Italian bidding company rather than by X BV directly. Ground IV also succeeded. Because the words “in law or in fact, directly or indirectly” point to economic reality, the compensating-tax test is applied not only at a creditor obliged to pass the interest on, but also at a related company that is the actual financier and lent the money through another party (para. 2.7.3).

The Court gave no final decision, because the outcome depended on ground V, the EU-law ground (see also AG Wattel’s further opinion of 8 June 2018, point 3.1). If G had been Dutch, it could have joined X BV’s fiscal unity from 3 May 2004, the contribution would have disappeared through consolidation, and Article 10a would not have applied (para. 2.8.3). In judgments of 24 June 2011 and 21 September 2012 the Supreme Court had read the CJEU’s X Holding judgment (C-337/08) as covering every consequence of consolidation, and so had rejected the per-element approach (para. 2.8.4). Groupe Steria (C-386/14) and Finanzamt Linz (C-66/14) cast doubt on that. The Supreme Court’s concern was cherry-picking:

If it can or must be assessed, for each element of the fiscal unity regime, whether there is unequal treatment of a company not established in the Netherlands that could be included in a fiscal unity if it were established here, the result will probably be that the companies concerned rely only on the consequences of a fiscal unity that favour them. Moreover, they can assess year by year, after the event, for which element there is such a favourable consequence.

Domestic groups, by contrast, must choose a fiscal unity in advance and accept all its consequences, including disadvantages such as a single use of the lower rate band (€22,689 in 2004), joint and several liability and pooling of investment allowances (para. 2.8.2.7). The Court referred the question to the CJEU.

X NV: the companion case on currency losses

X NV was the parent of a Dutch fiscal unity. Its member A Holdings BV held all shares in A Holdings (UK) Ltd and a hybrid receivable treated for tax purposes as equity. On 11 November 2008 A Holdings BV contributed the UK shares to another UK subsidiary, C Ltd; in February 2009 the shares in C Ltd were moved, through a Dutch member of the unity, to a Luxembourg subsidiary (HR 2016:1351, paras 2.1.1 to 2.1.9). X NV claimed a deduction for 2008 and 2009 for the negative currency result on the capital invested in the UK branch of the group. The District Court of The Hague dismissed its appeal; the Court of Appeal of The Hague (13 January 2015) held the whole currency loss deductible. The State Secretary appealed.

The Supreme Court held that currency results on a participation fall within the participation exemption, and that this is compatible with EU law in the light of X AB (C-686/13) (para. 2.4.2). It accepted, however, that if the UK subsidiary could have joined the fiscal unity, its profit would have been exempted under the permanent-establishment method, which leaves currency results on the capital invested in the Dutch tax base (paras 2.6.2.3 and 2.7.2). It referred three questions: whether Articles 49 and 54 TFEU require the loss to be taken into account, and, if so, which subsidiaries and which years would have to be assumed inside the hypothetical fiscal unity.

The Court of Justice: X BV and X NV (C-398/16 and C-399/16)

Advocate General Campos Sánchez-Bordona (opinion of 25 October 2017) proposed that Article 49 TFEU precludes the interest rule and does not preclude the currency rule (point 112). In C-398/16 he found the situations comparable and rejected the coherence and anti-avoidance justifications: the risk of tax avoidance was the same where the companies were all Dutch, yet only the domestic group could escape (points 81 to 83).

The Court, in a judgment of 22 February 2018, refused requests by both taxpayers to reopen the oral procedure (paras 14 to 17). It began from X Holding: reserving consolidation of profits and losses to resident groups is justified by the balanced allocation of taxing powers, but that does not mean every difference between integrated and non-integrated groups is compatible with Article 49 TFEU; advantages other than the transfer of losses need a separate assessment (paras 21 to 24, citing Groupe Steria).

Interest (C-398/16). The difference in treatment arose from the combination of Article 10a(2)(b), which does not itself distinguish between domestic and cross-border groups, and Article 15. Comparability was therefore assessed in the light of the purpose of Article 15 only, and on X Holding the situations were comparable (paras 34 to 37). None of the justifications held:

• Allocation of taxing powers. The advantage claimed is not the consolidation advantage itself. Dutch law allows interest deduction to every company and restricts it only in the cases in Article 10a(2)(b); a parent that escapes the restriction through a fiscal unity “does not, therefore, obtain an advantage specifically linked to the tax scheme of the single tax entity” (para. 40). The rule also does not depend on where the interest is taxed (para. 41).
• Coherence. The Netherlands showed no direct link between the advantage and a particular tax levy (paras 43 to 45).
• Abuse. Article 10a(2)(b) “unquestionably” pursues that aim (para. 48), but the difference in treatment is not specifically aimed at abuse, because it stems from Article 15, which has another purpose (para. 49). The Court then added the passage that later drew most criticism:

When a parent company finances the purchase of shares in a subsidiary by a loan taken out with another related company, the risk that that loan does not reflect a genuine economic transaction and is intended simply to create a deductible charge artificially is no less if the parent company and the subsidiary are both resident in the same Member State and together form a single tax entity than if the subsidiary is established in another Member State and is not, therefore, permitted to form a single tax entity with the parent company.

The answer was that Articles 49 and 54 TFEU preclude legislation refusing the deduction where a domestic parent could have obtained it by forming a fiscal unity with the subsidiary (para. 51).

Currency losses (C-399/16). The situations were not objectively comparable. A Dutch company cannot suffer currency losses on a resident subsidiary except where the holding is in another currency; inside a fiscal unity, a decline in value of a subsidiary is not deductible whatever its cause (paras 56 and 57); and a Member State is not required to allow losses asymmetrically on results it does not tax when positive (para. 58, citing X, C-686/13). The holding exemption is “a priori neither advantageous nor disadvantageous” (para. 59). There was no need to answer the second and third questions (para. 61).

Back in The Hague: the final judgments of 19 October 2018

The Advocate General’s objections. In his further opinion of 8 June 2018 in X BV, AG Wattel argued that the CJEU’s answer was “not readily usable” because it rested on a misunderstanding of the facts and of Dutch law (point 1.9). A capital contribution into a foreign subsidiary carries a greater risk of base erosion than one into a Dutch subsidiary, whose profits remain taxable in the Netherlands; and the case concerned a share premium contribution, not an acquisition of shares. He also argued that X BV would not have been better off domestically: on 3 May 2004 the adviser still held 10/110ths of G, so the 95% test was not met when the contribution was made, and a hypothetical inclusion of an Italian G would have turned it into an Italian branch whose results, and the interest, would have fallen within the exemption for foreign permanent establishments (point 1.14). He proposed that the Supreme Court uphold grounds III and IV and refer the case back to another court of appeal (point 1.25). He noted that the government had announced conditional retroactive legislation to remove five effects from the fiscal unity so that they could not be cherry-picked (point 4.3).

X BV (HR 2018:1968). The Supreme Court did not follow him. It held that G could have joined X BV’s fiscal unity from 3 May 2004, “if necessary by application of Article 4(1) of the Fiscal Unity Decree 2003” (Besluit fiscale eenheid 2003), and that Article 10a would then not have applied (para. 2.2). The cross-border and domestic situations are objectively comparable in the light of the aims of the fiscal unity (para. 2.3). It then changed its own case law:

It follows from the judgment of 22 February 2018 that it must be assessed for each element of the fiscal unity regime whether there is unequal treatment. In its judgments of 24 June 2011 … and of 21 September 2012 … the Supreme Court wrongly rejected the so-called “per-element approach”.

The Court set a limit at once. The comparison does not extend to features of the hypothetical situation in which the foreign subsidiary is included in the fiscal unity; in particular, how double taxation relief for the foreign subsidiary’s profit would have been given in that hypothetical case is irrelevant (para. 2.4.2). That disposes of the AG’s branch-exemption argument. Because the difference in treatment arises from Article 10a and Article 15 together, the aims of both provisions must be considered when looking for a justification, and the CJEU had found none (para. 2.6). Article 10a therefore had to be disapplied for X BV (para. 2.7).

The Supreme Court decided the case itself. It annulled the judgments of the Court of Appeal and the District Court and the inspector’s decision, and annulled the additional assessment for 2004 and the decision on interest charged. The State Secretary was ordered to pay the costs of cassation, including the preliminary reference, fixed at €7,140, and to reimburse the court fee of €497; the inspector was ordered to pay €1,503 and €1,879 for the appeal and first instance, €747 for the objection stage, and court fees of €478 and €302 (decision).

X NV (HR 2018:1969). In his further opinion of 12 June 2018, AG Wattel accepted that the CJEU seemed to have “missed the mark” on whether a fiscal unity with a foreign permanent establishment shows currency results, but considered it clear that the answer would have been the same (points 1.13 to 1.16). The Supreme Court held briefly that the CJEU’s ruling showed the Court of Appeal had erred in law (para. 2.1). It upheld the State Secretary’s appeal, annulled the Court of Appeal’s judgment and confirmed the District Court’s judgment dismissing X NV’s appeal, with no order for costs (paras 2.2 and 3).

The 2007 case: from fiscal unity to wholly artificial arrangements

Facts. X BV’s sole shareholder was A NV, a Belgian company listed in Brussels in which another Belgian company, B NV, held 39% (from 22 December 2002, 44.47%). The group included C, a Belgian coordination centre, which from 1999 to 2010 had a flat-rate cost-plus tax base and did not withhold tax on interest. On 14 March 2000 X BV bought 72% of the shares in a Dutch company F from third parties, and A bought the other 28%; X BV made an additional payment on 13 July 2000. X BV funded the purchase with loans from C, which C had funded each time shortly before with a capital contribution from A (HR 2022:1121, paras 2.1.1 to 2.1.4). The inspector refused the deduction of the interest for 2007.

Lower courts. According to AG Pauwels, the District Court of Gelderland accepted that C had a substantial function in the group and substance even in 2000, but not that this function mattered for the decision to give the funds to C as equity and then lend them to X BV; since the funds were converted one-to-one and X BV gave no insight into its reasons, C was no more than a conduit (opinion of 28 March 2025, point 8.20). The Court of Appeal of Arnhem-Leeuwarden (20 October 2020) assumed a restriction of the EU freedoms and held it justified by the aim of preventing abusive erosion of the Dutch base, with Article 10a(3)(a) keeping the rule proportionate (HR 2022:1121, para. 2.2.2).

The referral of 2 September 2022 (HR 2022:1121). Most grounds failed without reasons under Article 81(1) of the Judiciary Organisation Act (para. 3.1). On the remaining grounds the Supreme Court set out its reading of Article 10a(1)(c). An external acquisition, and related-party debt funding it, is in principle based predominantly on business considerations (para. 3.4.1). The exception is “rerouting”: where the funds the related lender provides came from the taxpayer itself or another related entity, the taxpayer must show business reasons for the rerouting; otherwise the rerouting is artificial and the deduction is refused, subject to Article 10a(3)(b) (para. 3.4.2). The rule targets base erosion “not by excessive interest expenses, but by contrived interest expenses”, so the test concerns the real financing function, not the arm’s-length nature of the terms (para. 3.5). Article 10a(1)(c) applies equally to resident and non-resident lenders, but can disadvantage cross-border cases because a resident lender will generally meet the compensating-tax test (para. 3.6.1). It is specifically aimed at wholly artificial arrangements and the counter-proof is not so burdensome as to lack real meaning (para. 3.6.2). The Court saw C-398/16 as confirming this and the step from internal reorganisations to external acquisitions as immaterial (para. 3.6.4).

The doubt came from Lexel (C-484/19), in which the CJEU had said that transactions at arm’s length could fall within the Swedish rule and therefore did not constitute wholly artificial arrangements (para. 3.7.3). The Court noted that part of the literature read that as meaning an arm’s-length loan can never be abusive (para. 3.7.5), and that Lexel, paragraph 51, and the EFTA Court’s judgment in PRA Group Europe (E-3/21) raised doubt whether a full refusal is proportionate (para. 3.7.8). It referred three questions: arm’s-length terms, full disallowance, and whether it matters if the target was already related before the acquisition (decision).

The Court of Justice: X BV (C-585/22)

The Court of Justice, after a hearing on 15 November 2023 and the opinion of AG Emiliou of 14 March 2024, refused X BV’s request of 10 April 2024 to reopen the oral procedure; the scope of Lexel had been debated and no new fact was put forward (paras 14 to 20). The EC Treaty applied ratione temporis to the 2007 assessment (para. 22). Because related entities under Article 10a(4) hold one third, which can give definite influence, only freedom of establishment applied (paras 26 to 31).

Difference in treatment. Article 10a(1)(c) and the counter-proof apply without distinction (paras 39 and 40). If, however, Dutch-resident related lenders could not in practice be taxed below 10%, the 10% condition in Article 10a(3)(b) would in fact disadvantage only cross-border situations; that is for the national court to verify (paras 42 to 46). The situations are comparable: a taxpayer is not in a different position because the lender is established elsewhere and taxed at not more than 10% (paras 50 to 52).

Justification. Preventing tax fraud and avoidance and combating wholly artificial arrangements are overriding reasons (paras 55 to 59). The Court recalled that in C-398/16 it had found that Article 10a, as in force in 2004, “undeniably” pursued that aim (para. 60), and held that extending the rule to external acquisitions does not change this, since in both cases the rule targets the redirection of own funds and their conversion into loan capital (para. 61).

Proportionality. The rule is suitable (para. 64). It creates a presumption, based on the related-party and acquisition criteria, that the debt is or is part of a wholly artificial arrangement; the counter-proof confines the refusal to loans dictated by tax reasons to the point that they are not needed for economically justified aims and would not have been contracted at all between unrelated parties (paras 69 to 71). On Lexel, the Court held that the arm’s-length test from Thin Cap Group Litigation (C-524/04) covers not only the terms of the loan but its economic sense and the related transactions (paras 74 to 76), that formal compliance is not enough under the general principle against abuse as restated in BMW Bank (paras 77 and 78), and that Lexel concerned a rule that was not aimed at wholly artificial arrangements and in which the economic validity of the loan was not examined (paras 79 to 83). It cannot be inferred from paragraph 56 of Lexel that a loan without economic justification escapes the abuse concept because its terms are at arm’s length (para. 84). On full disallowance the Court distinguished two cases (paras 87 and 88): where only the interest rate is excessive on a loan that reflects economic reality, only the excess may be refused; but

where the loan is, in itself, devoid of economic justification and, but for the relationship between the companies and the tax advantage sought, would never have been contracted, it is consistent with the principle of proportionality to refuse the deduction of the whole of the said interest, since such a wholly artificial arrangement must be ignored by the tax authorities when calculating the corporate tax due.

The rule also meets the requirement of legal certainty, since its application depends on criteria clearly set out in the legislation (paras 89 to 92). Article 49 TFEU therefore does not preclude a full refusal of interest on such a debt, even if it was incurred on an arm’s-length basis (para. 93).

The final judgment of 16 January 2026 (HR 2026:60)

The taxpayer’s two arguments. After the CJEU’s ruling, X BV argued that the EU test for a wholly artificial arrangement is lighter for the taxpayer than the Dutch “predominantly business considerations” test, requiring only some link with economic reality, and that point 88 of C-585/22 sets three cumulative conditions, one of which (that the loan would not have been made without the special relationship) an arm’s-length loan cannot meet.

The Advocate General. AG Pauwels advised dismissal. On point 88, he read it as one condition, not three: it presupposes that the debt is part of a wholly artificial arrangement and concerns only the consequence (full refusal), the clauses about the special relationship and the tax advantage merely describing a debt that lacks all economic reality (points 11.2 and 11.3). He also agreed with Marres that the CJEU’s standard is stricter, for the authorities, than the Supreme Court’s own test under its Mauritius case law, under which business reasons fail as soon as tax reasons weigh more (points 10.33 to 10.35); but he considered the Court of Appeal’s finding that tax saving was the “entirely decisive” reason consistent with the EU standard (point 11.6). On the taxpayer’s reliance on the Supreme Court’s judgment of 3 March 2023 (ECLI:NL:HR:2023:330) about group finance hubs, he accepted that C was such a hub but agreed with the lower courts that it acted as a mere conduit for these funds (points 8.10 and 8.19 to 8.21).

The Supreme Court. The Court summarised the CJEU’s answer (paras 2.3.2 and 2.3.3) and rejected the reading of Lexel. On the standard, it held:

Such a standard cannot, however — as is not reasonably open to doubt — have been intended by the Court of Justice. … Although the absence of economic reality is one of the elements of that qualification, the element of tax motives (the avoidance of profit tax) is the decisive element for the existence of a wholly artificial arrangement. It must therefore be assumed that the qualification of a transaction … as (part of) a wholly artificial arrangement concerns only cases in which tax considerations aimed at avoiding tax were decisive for entering into that transaction, regardless of whether other, weighty or weightier, business considerations also exist.

The Court of Appeal had applied the right test when it held that tax saving was the entirely decisive reason for redirecting the cash flows through C (para. 2.4.3). Paragraph 2.4.1 also states the Supreme Court’s reading of the CJEU: in Article 10a cases the artificiality lies in the redirection of group equity, presented to a Dutch company as debt, with the interest deducted in the Netherlands.

On point 88, the Supreme Court followed the AG: it does not say when a loan is wholly artificial and contains no criteria for that question; it addresses whether a full refusal is proportionate once the loan lacks all economic justification. The clause about the special relationship and the tax advantage only describes such a loan; “there is no question of a test with three conditions” (paras 2.5.3 and 2.5.4).

Finally, the Court of Appeal had found that C did not in fact perform a financial function in the group for this financing and was no more than a conduit through which group equity was converted into debt of X BV. Those findings did not misapply the concepts of finance hub and conduit in the 2023 judgment and were otherwise factual (para. 2.6.1). In the light of point 88, the loans therefore lacked all economic justification and a full refusal was not contrary to the EU principle of proportionality (para. 2.6.2). The appeal was dismissed, the other complaints under Article 81(1) of the Judiciary Organisation Act, with no order for costs (paras 2.8 to 4).

Fraus legis where Article 10a stops: HR 2026:736

Facts. The taxpayer was a Dutch joint-venture company of a Dutch bank (NL Bank, B shares) and a French bank, whose Luxembourg subsidiary LuxCo bought the taxpayer as a shelf company. In two tranches, on 17 August 2006 and 14 March 2007, NL Bank paid share premium (€424.5 million, later another €325 million) and LuxCo paid €10 million in capital and lent a €75 million Super Senior Loan and limited-recourse loans (LR Loans) of €415 million and later €325 million. The taxpayer invested the NL Bank money and the Super Senior Loan in a floating-rate bond portfolio swapped with NL Bank into a fixed rate of 4.176% (later 4.108%), and used the LR Loans and LuxCo’s capital to buy ordinary and all preference shares in LuxSub, LuxCo’s Luxembourg subsidiary in a Luxembourg tax consolidation. The preference dividend (4.176% and 4.108%) was deductible in Luxembourg and exempt in the Netherlands, and repayment and interest on the LR Loans were limited to what the taxpayer received on the preference shares. NL Bank’s dividend on its B shares was adjusted for the Dutch tax saving (HR 2026:736, paras 2.1 to 2.9.3). The taxpayer declared taxable amounts of €5,307 (2007/2008) and €126,980 (2008/2009), after deducting LR Loan interest matched by exempt participation income of €31.2 million and €31.1 million. The inspector corrected the full interest by additional assessments (para. 2.11).

Lower courts. It was not disputed that the LR Loans were tainted. The Court of Appeal of Amsterdam (30 September 2021) held that the taxpayer failed the business test, but showed a compensating tax at LuxCo. For 1 January to 30 September 2008 and the year 2008/2009 it applied the inspector’s counter-proof, in force from 1 January 2008 with immediate effect, and found the debt not predominantly business-driven; for 1 October to 31 December 2007, when the counter-proof did not exist, it refused the deduction under fraus legis (paras 3.1 to 3.5.6). It found Lexel irrelevant (paras 3.6.1 to 3.6.4).

The Supreme Court. Ground I, on whether the Court of Appeal could raise the 2008 date of its own motion, failed: the inspector’s original limitation of his reliance to 2008/2009 had been an evident mistake of law, and the taxpayer had been able to respond (para. 4.1.4). Ground II argued that the counter-proof should not apply where Luxembourg taxed the interest at 29.63% and 28.59%, more than the Dutch rate of 25.5%. The Court held that the counter-proof was meant to limit the absolute safe harbour introduced on 1 January 2007 and so applies whenever the tax on the interest is 10% or more (paras 4.2.1 and 4.2.2); reliance on a policy decision of 23 December 2005 and on legitimate expectations failed (para. 4.2.3).

Ground III, on fraus legis for late 2007, also failed. In 2007 a compensating tax guaranteed deductibility within the “10a structure”, but that does not exclude a refusal for artificiality that lies wholly or partly outside it, as abuse of the Act as a whole (para. 4.3.2). The 10a structure here comprised the French bank, LuxCo, the taxpayer and LuxSub, with the LR Loans as tainted loans and the purchase of the preference shares as the tainted act; the relations with NL Bank and the bond portfolio lay outside it (para. 4.3.3). Contrived interest set against the profit of an enterprise, in a way that frustrates tax arbitrarily and continuously through legal acts that are not needed for business aims, is contrary to the aim and purpose of the Act (para. 4.3.4), and the Court of Appeal could find that here (para. 4.3.5).

Ground IV invoked the EU abuse concept. It failed on the reasoning of HR 2026:60, taking into account the amendment of Article 10a(3)(b) on 1 January 2008, and the Court added:

As regards the doctrine of fraus legis, which was not at issue in the judgment of the Court of Justice of 4 October 2024 … that doctrine must in this context be interpreted and applied according to the same standards as the rule of Article 10a of the Act. … Contrary to what the ground argues, this doctrine therefore does not go further than combating abuse in the EU-law sense.

The principal appeal was dismissed; the State Secretary’s conditional cross-appeal lapsed under Article 8:112(2) of the General Administrative Law Act; there was no order for costs (paras 5 to 7). For the EU abuse concept in a neighbouring field, see our review of the dividend tax anti-abuse cases.

How the judgments fit together

Two different EU-law problems. C-398/16 did not hold that Article 10a breaches EU law. The Court accepted that the rule pursues an anti-abuse aim (para. 48). What failed was the combination with a fiscal unity reserved to residents, which let domestic groups switch the rule off. C-585/22 then examined Article 10a on its own and upheld it. The 2018 cases are about equal access to a domestic advantage; the 2024 to 2026 cases are about the content of the anti-abuse test.

Limits of the per-element approach. X NV shows that the per-element approach does not mean every effect of consolidation can be claimed cross-border. Where the domestic rule is symmetrical, as with currency results on participations, there is no comparable situation (C-399/16, paras 56 to 59). And under HR 2018:1968 (para. 2.4.2) the hypothetical comparison stops short of the double-tax relief that would have applied to a consolidated foreign subsidiary. The AG’s objections in 2018, that the 95% test was not met on the day of the contribution and that a consolidated Italian G would have been a branch with exempt results, were answered only by these short findings in paras 2.2 and 2.4.2.

The meaning of “wholly artificial”. C-585/22 frames the abuse test around loans that would not have been contracted between unrelated parties and lack economic justification (paras 71 and 88). HR 2026:60 reads this as turning on whether tax motives were decisive, regardless of other weighty business reasons (para. 2.4.2), and treats that reading as beyond reasonable doubt, so no further referral was made. AG Pauwels had observed that the CJEU’s standard was stricter for the authorities than the Supreme Court’s earlier “tax reasons weigh more” test (points 10.33 to 10.35). The Supreme Court’s wording in para. 2.4.2 does not adopt that test expressly; it speaks of tax considerations that were “decisive”. How far a loan with substantial, but not decisive, business reasons is now safe under Article 10a(3)(a) is the main open question left by these judgments.

Arm’s length is not the test. Both the Supreme Court (HR 2022:1121, para. 3.5) and the CJEU (C-585/22, paras 75 and 84) treat the arm’s-length nature of the interest rate as beside the point where the loan itself has no economic justification. Full disallowance follows (para. 88). Only where the loan is genuine but the rate excessive is the refusal limited to the excess (para. 87).

Fraus legis and EU law. HR 2026:736 ties the national doctrine to the EU abuse standard (para. 4.4.2). That keeps fraus legis available where Article 10a is satisfied, as in the 2007 period when a compensating tax of at least 10% guaranteed deduction within the 10a structure, while confining it to wholly artificial arrangements aimed at avoiding Dutch tax.

What it means in practice

• Acquisition debt from group companies. Under Article 10a(1)(c) an external acquisition funded by related-party debt is in principle business-driven, but not where the lender’s funds came from the group itself (HR 2022:1121, paras 3.4.1 and 3.4.2). Keep evidence of where the lender obtained the funds and why the group financed the purchase with debt rather than equity.
• Treasury or finance companies. A group finance company with real staff and capital is not enough if, for the loan in question, it acted as a conduit for group equity converted into debt (HR 2026:60, para. 2.6.1). Records should show the finance company’s own role in the specific loan, not only its general substance.
• Interposed Dutch companies. The reasons for using a Dutch company are not tested under Article 10a(3)(a); the reasons for the debt and for the connected transaction are (HR 2016:1350, paras 2.6.3 and 2.6.4).
• Compensating tax. The 10% test looks to the actual financier, including a related company that lent through another party (HR 2016:1350, para. 2.7.3). Since 1 January 2008, a tax of 10% or more, even above the Dutch rate, does not end the inquiry: the inspector may still show that the debt or the transaction is not predominantly business-driven (HR 2026:736, para. 4.2.2).
• Arm’s-length pricing. Transfer pricing documentation of the rate does not answer an Article 10a challenge. The question is whether the loan would have been contracted at all without the group relationship and the tax advantage (C-585/22, paras 75, 76 and 88).
• Arrangements that share the tax saving. Features such as a dividend adjusted for the Dutch tax saving, limited-recourse loans matched to exempt income and swaps aligning return with the interest charge were treated by the Court of Appeal as evidence of a tax-driven structure (HR 2026:736, paras 3.4.1 and 3.4.2).
• Per-element claims. For years to which the current Article 15(16) does not apply, a cross-border group may rely on the per-element approach for an advantage that a domestic fiscal unity would have given, but not for symmetrical rules or for the treatment of the hypothetical foreign member (HR 2018:1968, para. 2.4.2; C-399/16, paras 56 to 59).

Status

All the cases are closed. X BV (2004) ended with the annulment of the assessment; X NV ended with the District Court’s judgment confirmed; the 2007 case and HR 2026:736 ended with the taxpayers’ appeals dismissed.

The law has since moved, as our current text shows. Article 15(16) now provides that Articles 10a, 13(9) to (15) and (17), 13a, 15ba and 20a are applied “as if there were no fiscal unity”, so a domestic fiscal unity no longer switches off Article 10a. AG Wattel referred in 2018 to announced legislation that would take five effects out of the fiscal unity (point 4.3), but none of the judgments reviewed here applies Article 15(16) or gives its date of entry into force. The current Act also contains an earnings-stripping rule in Article 15b(1) (net interest deductible up to the higher of 24.5% of adjusted profit or €1,000,000) and a general anti-abuse rule in Article 29i. Neither is discussed in these judgments, and their relationship to Article 10a and to fraus legis as framed in HR 2026:736 is not addressed in the sources.

Provisions discussed

• Corporate Income Tax Act: Art. 10a, Art. 13, Art. 15

Related reviews

How to cite

Case law review by BCA EU, for information. Quotations from the judgments are our translations; only the original texts published by the courts are authoritative. Judgments checked on the official sources on 30 October 2026. This page is not legal or tax advice.