For as long as there have been cross-border structures, there have been entities that exist mainly on paper: a company in a favourable jurisdiction, with a registered address, a nominal director and little else, interposed to capture a treaty benefit or a directive. The European Union’s Unshell proposal, commonly known as ATAD 3, was the most direct attack yet on exactly that kind of entity. It was never adopted: it required unanimity in the Council, and in June 2025 the Council concluded that work on it should not continue, many delegations taking the view that its aims could be reached by clarifying or amending the reporting hallmarks of the administrative cooperation directive rather than by a standalone directive. Nothing in the proposal itself is in force. It is still worth understanding, because the questions it asked are the questions treaty anti-abuse provisions, the principal purpose test and Dutch ruling policy already ask, and a structure built to answer them is unaffected by whether Unshell is ever revived.
What Unshell is trying to do
The purpose was narrow and pointed: to deny the tax advantages of European entities that have no genuine economic substance and exist principally to obtain those advantages. It did not attack structures with real activity, real people and real decision-making. It attacked the shell, the entity whose only function is to hold and to route, staffed by nobody, deciding nothing. The distinction the proposal drew was precisely the distinction between a structure and a pretence.
Where the proposal actually stands
The document to read is COM(2021) 565 final, tabled on 22 December 2021 as a proposal for a Council Directive laying down rules to prevent the misuse of shell entities for tax purposes and amending Directive 2011/16/EU, under interinstitutional procedure 2021/0434(CNS). Its legal basis is Article 115 of the Treaty on the Functioning of the European Union, which is why unanimity in the Council was required and why the European Parliament was only consulted. Its transposition clause carried the placeholder dates of 30 June 2023 for publication and 1 January 2024 for application, in square brackets, as drafts do. Neither date was ever triggered.
The Council then stopped work on it. The ECOFIN report on tax issues approved at the Council meeting of 20 June 2025, document 10611/25, records that potential overlaps had been recognised between the hallmarks in the latest Unshell compromise text and those of DAC 6, with the risk that similar information would be reported twice and probably through separate systems; that many delegations were of the view that the aims of the Unshell proposal could be achieved with clarifications or amendments of hallmarks in DAC 6; and that, as a consequence, it emerged that the analysis of the Unshell proposal should not be continued in the Council. The report adds that this approach was broadly welcomed by delegations.
The Commission has followed. Its work programme for 2026, adopted on 21 October 2025, lists COM(2021) 565 final in the annex of pending proposals it intends to withdraw within six months, an annex whose stated criteria are the adoption date, the lack of progress in the legislative process, potential burden and non-alignment with the Union’s priorities. On 24 June 2026 the Commission tabled a recast of the administrative cooperation directive, COM(2026) 308 final. The practical consequence for a reader is plain: there is no Unshell reporting obligation, no deadline and no presumption to rebut, and any text that describes one in the present tense is describing a draft.
The gateways and the substance indicators
The proposed mechanism worked in stages. First, gateways would identify entities at risk: broadly, those with mostly passive income, significant cross-border activity, and outsourced administration. An entity that passed through those gateways would then report against substance indicators, essentially whether it has its own premises, its own bank account in the Union, and at least one qualified, resident director or a majority of resident, qualified employees genuinely running its activities. An entity that failed these indicators would be presumed to be a shell.
The thresholds were specific. More than 75 per cent of revenues over the preceding two tax years had to be relevant income as defined in Article 4 of the proposal; the cross-border test was met where more than 60 per cent of the book value of the relevant immovable and movable property sat outside the Member State, or at least 60 per cent of the relevant income was earned or paid out through cross-border transactions; and the third gateway asked whether, over those same two years, the undertaking had outsourced the administration of day-to-day operations and the decision-making on significant functions. A holding entity with no income at all was not thereby safe: assets capable of generating dividend, property or qualifying movable income were deemed to satisfy the income gateway where their book value exceeded 75 per cent of total book value.
The proposal did not ask whether an entity was clever. It asked whether anyone actually worked there, decided there, and could be found there. That is a question a mailbox cannot answer, and it is the question every serious examiner still asks.
The consequences of being a shell
Under the proposal, an entity presumed to be a shell and unable to rebut the presumption would have faced real consequences: it might have been denied the benefits of the tax treaties and the European directives it was interposed to access, other member states might have disregarded it and taxed the income as though it flowed directly to the shareholders, and it might have been unable to obtain the certificate of tax residence on which cross-border relief depends. In short, the structure stops doing the one thing it was built to do.
There was also a sanction, and it was not nominal. Member States were to provide penalties that include an administrative pecuniary sanction of at least 5 per cent of the undertaking’s turnover in the relevant tax year where a reporting entity misses the deadline or makes a false declaration. The information reported would in addition have been exchanged automatically with the competent authorities of all other Member States within thirty days of receipt, along with the identification of the entity’s shareholders and beneficial owners, which is the detail that turns a domestic filing into an international one.
Rebutting the presumption, and the exemptions
The presumption could be rebutted by demonstrating genuine commercial rationale and real substance, and certain entities were carved out, regulated financial undertakings and genuinely active companies among them. The burden would have sat on the taxpayer, and rebutting a presumption with evidence assembled after the fact is always harder than never triggering it. The entities that would have come through Unshell comfortably are the ones that were always real; the ones that would have struggled are the ones that were always thin.
The rules that are already in force
The ground the proposal covered is not empty. Directive (EU) 2016/1164, the Anti-Tax Avoidance Directive, applies to all taxpayers that are subject to corporate tax in one or more Member States, including permanent establishments there of third-country residents, and its Article 6 requires a Member State to ignore, for the purposes of calculating corporate tax liability, an arrangement or series of arrangements 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 applicable tax law and which are not genuine having regard to all relevant facts and circumstances. Not genuine is defined in the next paragraph as not put into place for valid commercial reasons which reflect economic reality. Directive (EU) 2017/952 later amended that directive as regards hybrid mismatches with third countries. Member States had to apply Article 6 of the 2016 directive from 1 January 2019, so this is law in force and not a draft.
The directive benefits themselves carry the same rule. Article 5 of the Parent-Subsidiary Directive exempts from withholding tax the profits a subsidiary distributes to its parent, and Directive (EU) 2015/121 of 27 January 2015 replaced its Article 1(2) with a minimum anti-abuse rule in the same words: Member States shall not grant the benefits of the directive to an arrangement 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 and which is not genuine. The same amendment preserves domestic and agreement-based provisions for the prevention of tax evasion, fraud or abuse. Member States had to bring that rule into force by 31 December 2015, so it has been operative for a decade.
The Court of Justice then removed the last argument a thin entity had. In the Danish beneficial ownership cases of 26 February 2019, T Danmark for dividends and N Luxembourg 1 for interest, the Grand Chamber held that national authorities and courts are to refuse the directive exemption where there is a fraudulent or abusive practice even if there are no domestic or agreement-based provisions providing for such a refusal, and that for interest the exemption is restricted to beneficial owners, meaning the entities which actually benefit from the payment economically and accordingly have the power freely to determine the use to which it is put. Proof of abuse requires objective circumstances in which the purpose of the rules has not been achieved together with an intention to obtain the advantage by artificially creating the conditions for it, and the Court named indications that can demonstrate it, among them the existence of conduit companies without economic justification and the purely formal nature of the group structure, the financial arrangements and the loans. Those are the Unshell questions in judicial form, and they were binding almost three years before the proposal was tabled.
Why this vindicates the substance approach
For a firm whose entire method is building structures with genuine substance, Unshell is not a threat but a confirmation. Everything the proposal asked for, resident directors who genuinely decide, real premises, a real bank account, real activity, is what we describe in our note on Dutch substance requirements and what we build as standard. The proposal would simply have made explicit, and enforceable across the Union, the standard that a serious structure met anyway.
What to do now
The sensible response is not to wait for a directive the Commission has listed for withdrawal, but to look at every existing entity and ask the Unshell questions of it today: does it have its own substance, its own decision-making, its own presence, or is it a shell that has simply not yet been challenged. Structures assembled years ago for a purpose that made sense then frequently fail that test now, and the time to fix a thin entity is before a tax authority tests it, not after. This connects directly to treaty access, which we cover in our note on treaty access and beneficial ownership: the same fragility that fails Unshell also fails the principal purpose test.
Montclare builds and operates European structures with the substance and documentation that these rules require. Our services are set out on our services page.
This article is informational and does not constitute tax or legal advice. Rules of this kind evolve and their application depends on the facts of each structure. Each engagement is subject to scope and applicable regulation.