The Court of Justice of the European Union (CJEU), in its judgment of 15 October 2026 (Case C-203/25, Neo Group), has held that the subsidiary’s State may deny the withholding tax exemption on dividends paid to a parent company resident in another Member State even where that parent carries on a genuine economic activity. For that purpose it is sufficient, in principle, that the distribution forms part of a non-genuine chain of transactions whose ultimate destination is the person who controls the group.

The judgment also contains two clarifications of practical significance. First, the examination of the abuse may extend to transactions carried out outside the subsidiary’s State and by persons not resident there. Second, it is not necessary to prove that the paying subsidiary was aware of the abuse; it is sufficient that one of the group companies involved in the chain, or the individual who obtains the advantage at its end, was or could have been aware of it.

For family groups and international investors who channel their investment in Spain through a holding company (a company holding shareholdings) located in another Member State, the judgment calls for a review of a common assumption: that the substance of the intermediate company is, in itself, a sufficient guarantee for applying the exemption.

The regulatory framework

Council Directive 2011/96/EU of 30 November 2011 on the common system of taxation applicable in the case of parent companies and subsidiaries of different Member States (the Parent-Subsidiary Directive) seeks to ensure that profits distributed by a subsidiary to its parent company located in another Member State do not bear an additional tax burden by reason of crossing a border.

To that end, its Article 5 exempts those dividends from withholding tax at source, that is, from the tax that the subsidiary’s State would levy on the dividend and that the subsidiary itself would pay over, on behalf of the parent, at the time of payment.

Since the amendment introduced by Council Directive (EU) 2015/121 of 27 January 2015, Article 1(2) and (3) of the Parent-Subsidiary Directive has contained a common anti-abuse clause. Member States shall not grant the benefits of the Directive to an arrangement, or a series of arrangements, which is not genuine and which has been put into place for the main purpose, or one of the main purposes, of obtaining a tax advantage that defeats the object or purpose of the Directive.

Two terms deserve attention. An “arrangement” (arreglo in the Spanish version of the Directive) refers to any transaction, agreement or structure, and the provision makes clear that it may comprise more than one step or part. An arrangement is “not genuine” to the extent that it has not been put into place for valid commercial reasons which reflect economic reality.

In Spain, the exemption in Article 5 of the Directive is laid down in article 14.1.h) of the consolidated text of the Non-Resident Income Tax Act, approved by Royal Legislative Decree 5/2004, of 5 March (TRLIRNR). The provision exempts profits distributed by subsidiaries resident in Spanish territory to their parent companies resident in other Member States. It requires, among other conditions, that the parent hold a direct or indirect shareholding of at least 5 per cent, held without interruption during the preceding year, and that both companies be subject to, and not exempt from, one of the taxes on profits listed in the Directive.

Note that the specific anti-abuse clause in article 14.1.h) TRLIRNR has a narrower scope than that of the Directive. The exemption does not apply where the majority of the voting rights in the parent company is held, directly or indirectly, by individuals or legal entities not resident in Member States of the European Union or of the European Economic Area with effective exchange of information, unless the incorporation and operation of the parent company are driven by valid economic reasons and substantive business reasons. Paragraph 2 of the same article also excludes the exemption where the parent company is resident in a country or territory classified as a tax haven.

This difference in scope will be relevant when we assess the reach of the judgment in Spain.

The facts

NEO Group is a Lithuanian company wholly owned, since 2008, by Retal Industries, a Cypriot company that carries on a genuine economic activity. The Lithuanian tax authority never disputed that activity.

In 2016 and 2017, NEO paid Retal dividends amounting to EUR 15,000,000, and another Lithuanian subsidiary of the group paid it EUR 13,485,032. NEO did not withhold tax, relying on the exemption provided for in the Lithuanian corporate income tax law, which transposes Article 5 of the Directive.

The Lithuanian tax inspectorate did not challenge Retal, but what was happening above it. A Belize company, the owner of Retal, sold all of its shares to PPH, a Cypriot company incorporated less than three months earlier, for EUR 310,400,000. Payment of the price was deferred until 2027 and the resulting receivable was assigned to PG, an individual resident in Cyprus who was already the beneficial owner of both the seller and the buyer.

From then on, PG received substantial amounts from PPH, in the form of loans and assignments of rights. According to the inspectorate, those funds came from the dividends of the Lithuanian subsidiaries and served gradually to repay PPH’s debt to PG himself.

The inspectorate concluded that all of this constituted a non-genuine arrangement aimed at avoiding the Lithuanian 15 per cent tax on dividends. It issued NEO with an assessment of EUR 2,250,000 —15 per cent of the EUR 15,000,000 distributed—, plus late-payment interest, and imposed a penalty on it.

The Lithuanian tax disputes body hearing the appeal stayed the proceedings and referred six questions to the CJEU for a preliminary ruling, that is, six questions on the interpretation of EU law needed to resolve the dispute. The Spanish Government submitted observations in the proceedings.

What the Court holds

The parent company’s genuine activity does not in itself exclude abuse

NEO’s defence rested on a widespread reading of the judgment in T Danmark and Y Denmark (Joined Cases C-116/16 and C-117/16), of 26 February 2019. In that judgment, the Grand Chamber of the CJEU identified as a typical artificial scenario that of a conduit company interposed between the paying subsidiary and the beneficial owner of the dividends, whose sole activity consists in receiving them and passing them on.

In this context, the beneficial owner is the person who receives the dividend for his or her own benefit and with the power to dispose of it, and not as a mere intermediary. From that judgment it was inferred that a parent company with a genuine activity, not being a conduit company, fell in practice outside the reach of the anti-abuse clause.

The Court rejects that inference. It recalls that in T Danmark the conduit company was mentioned “in particular”, that is, as an example and not as the only case of abuse. It also relies on its judgment of 3 April 2025, Nordcurrent group (Case C-228/24), in which it had already accepted that the anti-abuse clause could apply to a subsidiary that was not a conduit company. From this it infers that the clause is of a cross-cutting nature and applies to any non-genuine arrangement, not only to particular configurations.

The consequence is that, where an arrangement comprises several steps, the fact that the first of them is genuine is not enough to rule out abuse. In the case, that first step was the payment of dividends by the subsidiary to a parent company with a genuine activity. Analysing only the parent’s activity, without regard to the other circumstances (the sale of shares, the loans and the assignment of receivables), would be incompatible with the overall examination required by the Directive.

The Court also addresses the double taxation convention between Lithuania and Cyprus, which limits taxation in the subsidiary’s State to 0 per cent where the parent company is the beneficial owner of the dividends and holds at least 10 per cent of its capital. The interpretation of the convention is a matter for the national court. The Court recalls, however, that conventions between Member States cannot be relied on to achieve a result contrary to EU law, even where they follow the Model Convention of the Organisation for Economic Co-operation and Development (OECD). And it points out that the Commentaries on that Model themselves warn that beneficial owner status does not entail automatic entitlement to the reduced rate in cases of abusive use.

Indicia: matching amounts and transfers over time

The second and fifth questions concerned evidence. The Court accepts that a similar, or even identical, correspondence between the dividends received by the parent company and the amounts received at the end of the chain by the person who controls the group constitutes a relevant indication of abuse. That indication cannot, however, give rise to a general presumption of abuse.

In T Danmark, it had been noted as an indication that the dividends were passed on in full, or almost in full, within a very short period. The Court now clarifies that the shortness of the period may be an indication, but it is not a requirement. A non-genuine arrangement whose effects recur at regular intervals over a certain period is no less contrary to the Directive than a one-off one.

Consequently, the tax authority of the subsidiary’s State is not obliged to confine its examination to a short period. Where the arrangement involves changes in the group structure and lending transactions, as in this case, it must include in its investigation all the elements contributing to the possible abuse.

Transactions outside the subsidiary’s State and advantage obtained at any step

The third and fourth questions raised two doubts. The first, whether the Lithuanian tax authority could rely on transactions carried out in Cyprus between persons not resident in Lithuania. The second, whether a tax advantage could be found where the person obtaining it is not the paying subsidiary, but the individual at the end of the chain.

The Court answers both in the affirmative. Since the Directive by definition concerns cross-border situations, the examination must be an overall one and may cover transactions carried out outside the subsidiary’s State, before or after the payment of the dividends.

Nor does the anti-abuse clause require the tax advantage to be obtained by the subsidiary. It may be obtained by the individual who controls the group and is resident in another Member State, irrespective of the step in the chain at which it materialises. As the Court reasons, such undue advantages could not be eliminated if the State granting the exemption were unable to take into account the transactions carried out outside its territory with the amounts of the dividends.

The Court considers this interpretation consistent with the principle of proportionality. NEO argued that it amounted to an unjustified restriction on freedom of establishment (Article 49 of the Treaty on the Functioning of the European Union, TFEU) and on the free movement of capital (Article 63 TFEU). The Court reiterates, citing T Danmark, that those freedoms cannot be relied on to call into question national legislation on the taxation of dividends where it has been established that its application makes it possible to prevent an abuse.

Knowledge of the abuse need not be established in the subsidiary

The sixth question is, in our view, the one with the greatest practical reach. The Lithuanian tax inspectorate had not examined whether NEO, as a subsidiary dependent on the decisions of its shareholder, knew or could have known of the existence of non-genuine transactions within the group.

The Court starts from the premise that proof of abuse requires two elements. An objective one: that, despite formal compliance with the conditions laid down by the rule, its purpose has not been achieved. And a subjective one: the intention to obtain an advantage from EU law by artificially creating the conditions required for obtaining it.

However, a distribution of dividends may result from a decision of the parent company, itself part of a series of decisions taken within the group. If the subjective element had to be verified only in the subsidiary, the anti-abuse clause would lose much of its effectiveness.

It is therefore sufficient to establish that element in respect of at least one of the group companies taking part in the chain or of the individual who obtains the advantage at its end. It is not required that the subsidiary itself was aware of that intention.

Assessment

In our opinion, the judgment maintains the standard of proof of abuse and broadens the perimeter to which it applies. The burden of proving abuse still lies with the tax authority, which must carry out an objective analysis of all the relevant facts and circumstances, and matching amounts are not sufficient to presume it. What changes is the object of the analysis, which is no longer confined to the company receiving the dividends and now also encompasses what subsequently happens to them, within or outside the subsidiary’s State.

This broadening falls, to a large extent, on the paying subsidiary. It is the subsidiary that is obliged to withhold and, frequently, it takes no part in the group’s decisions on the destination of the funds and has no access to that information. The judgment accepts that the adjustment may be directed against it even if it is unaware of the abuse.

The Court rules on the denial of the exemption and does not address the penalty, which had also been imposed on NEO. In our view, in Spanish penalty proceedings the culpability of the subsidiary would remain a requirement in its own right, which the judgment does not alter.

There is, moreover, a certain tension in the formulation of the subjective element. In its reasoning, the judgment requires the intention to obtain the advantage to be demonstrated in one of the group companies or in the ultimate individual. The operative part, by contrast, is satisfied if one of them “had or could have had knowledge” of that advantage. Knowing, or being able to know, is not the same as pursuing the advantage, and the difference is likely to be argued before the national courts.

As regards Spain, the judgment raises a question of statutory fit. The scenario in Neo Group, a group ultimately controlled by an individual resident in another Member State, would fall outside the literal wording of the anti-abuse clause in article 14.1.h) TRLIRNR, which only operates where the majority of the voting rights in the parent company is in the hands of residents outside the European Union or the European Economic Area.

This does not mean that the Spanish tax authority lacks instruments. The T Danmark judgment held that national authorities must deny the exemption in the event of abusive practice even in the absence of domestic or convention-based provisions providing for such denial. Our legal system also has the general clause in article 15 of Act 58/2003, of 17 December, General Tax Act, which governs conflict in the application of tax rules and whose declaration requires a prior favourable report from an advisory committee. Which route the inspectorate will use in each case, and with what procedural safeguards, is a question the judgment does not resolve.

Practical consequences

For structures in which an EU holding company holds shares in Spanish companies, we consider that the judgment advises reviewing several aspects.

The substance analysis of the intermediate company remains necessary, but it is no longer sufficient. It must be supplemented by an examination of the destination of the dividends once received. The position is very different depending on whether they are reinvested in the holding company’s activity or in new investments, or whether they flow to the ultimate shareholder through loans, assignments of receivables or payments of deferred price.

Particular attention should be paid to reorganisations or acquisitions in which the purchase price of the holding company is deferred and gradually paid out of the subsidiaries’ dividends, especially where the seller and the buyer share the same beneficial owner. In our view, following this judgment the risk of the exemption being denied in these cases is high, since they reproduce the elements taken into account by the Lithuanian inspectorate.

It is also advisable to document the commercial reasons for each step of the structure, and not only for the incorporation of the holding company. The judgment recalls, relying on Nordcurrent group, that an arrangement that was initially justified may come to be regarded as not genuine if it is maintained after the circumstances that justified it have ceased to exist.

Lastly, the Spanish subsidiary has an interest of its own in knowing the use to which the dividends will be put, since it is the subsidiary that is liable for the withholding. In groups where treasury and debt are managed above the subsidiary, it may be prudent for the subsidiary to obtain information or representations from the group on that destination, and for indemnity clauses to be considered where the structure displays indicia such as those described.

In conclusion, what this judgment of the Court of Justice makes clear is that the withholding tax exemption under the Parent-Subsidiary Directive is assessed in the light of the structure as a whole and of the full path of the dividends, so that neither the substance of the parent company nor the lack of knowledge of the paying subsidiary is, in itself, sufficient to secure it.

Sources