Can an Article 50 notice of withdrawal from the EU be unilaterally revoked?

There is no sign that the UK government (or indeed the opposition Labour party) is considering revoking the UK’s notice of withdrawal from the EU, or that either would be interested in asking the British public to vote again on the question of EU membership. Yet, sparked by the latest attempt of Nigel Farage to draw attention to himself, the prospect of a further Brexit referendum – which necessarily assumes that the Article 50 notice is revocable in some way – continues to be a subject of public debate. 

In that context, and in light of a new European Parliament report on this issue, here’s an argument from Professor Stephen Weatherill arguing that the Article 50 notice cannot be unilaterally revoked – with the contrary argument from Professor Steve Peers.

Why the withdrawal notification under Article 50 TEU is not unilaterally revocable

Professor Stephen Weatherill, Somerville College and Law Faculty, University of Oxford

The recently published European Parliament paper entitled ‘The (ir-)revocability of the withdrawal notification under Article 50 TEU’ addresses an issue that may come to have very sharp significance in determining the options and directions taken in the unfolding Brexit circus. As its title suggests, the paper inquires into the question whether the withdrawal notification under Article 50 TEU is revocable or irrevocable, and, most poignant of all, it asks whether that withdrawal notification is unilaterally revocable. This will matter a lot should the UK repent of its decision to leave the EU. I argue here that the notification is not unilaterally revocable.

The paper, authored by Ioannis Papageorgiou of the Parliament’s Policy Department for Citizens’ Rights and Constitutional Affairs, takes as its obvious starting-point that Article 50 TEU does not explicitly address the revocation of a withdrawal notification. It also notes that ultimately it is the Court of Justice that will provide an authoritative ruling, should it be provided with the opportunity (pages 6, 22). I agree. It observes that ‘Given the fact that Article 50 does not make any reference to revocation issues, both sides of the argument can be defended’ (pages 16, 21). I agree. It treats the Vienna Convention on the Law of Treaties as irrelevant, mainly because it is in the nature of the evolved EU legal order that answers are to be found within the EU system alone, as a self-contained system (pages 7-13, 24, 27). I agree. And the paper ably surveys both sides of the argument. At page 23 it is commented that ‘Among UK scholars and politicians there is a predominant position that a unilateral revocation is within the right of the UK’. I agree: there is such a predominant position. The purpose of this note is to argue against that predominant position.

Consider first matters of design from the perspective of costs and incentives. A State that notifies an intention to withdraw within the meaning of Article 50 TEU and then wishes to change its mind has generated large costs. It has caused resources to be allocated to the negotiations which it wishes to terminate which could have been used for other purposes. To treat Article 50 TEU as capable of unilateral revocation means that the costs incurred lie where they fall: the repentant State covers its costs and the EU-27 (a term I use to denote the 27 member States plus the EU institutions) covers its costs. That means that the State that wished to leave, but which changed its mind, is able to externalise the costs incurred by the EU-27 as a result of its decision. That in turn reduces its incentives to make a full and careful calculation of the consequences of its action before it issues the Article 50 notification.

By contrast an interpretation which denies that the notification made pursuant to Article 50 TEU may be revoked unilaterally, and which insists that instead revocation depends on the agreement of the EU-27, ensures that as part of that negotiation the repentant State may be required to meet the costs of its choices. It makes every sense to ensure that costs are imposed on the party that is most able to reduce those costs. That is the party that initiated the process in the first place by notifying an intention to withdraw which it then wished to revoke – here, the UK. As a general observation one may argue that it is a matter of simple justice that the party that has generated the costs should pay for those costs – more narrowly my argument here is that such an approach is desirable in order to preclude a State from externalising the costs of its actions, and thereby pushing it to consider fully in advance the costs involved. This militates in favour of denying a unilateral right of revocation, and instead ensures that revocation becomes part of a bargaining process within which costs can be allocated.

A separate strand in the debate concentrates on democracy. Why should a State not be allowed to change its mind? If – perhaps as a result of a fresh referendum, perhaps following an election leading to a change of government – the people of that state have clearly decided they wish to remain in the EU, why should Article 50 be interpreted to thwart that expressed popular will?

This is a good argument in favour of interpreting a notification pursuant to Article 50 to be revocable provided there is political agreement among the repentant state and the EU-27. It is a weak argument in favour of a unilateral right to revoke. A model that permits unilateral revocation entails that some of the costs incurred are borne by voters in the other Member States, albeit largely indirectly because they will be swallowed up by the overall EU budget. So voters in Germany, Italy and France and so on would be expected to pay for the consequences of decisions to leave and then, after a change of mind, to remain taken by the UK although those German, Italian and French voters have had no say at all, and can have no say at all, in influencing those decisions.

Much of EU law is animated by the concern that decisions taken at national level have cross-border implications and that it is accordingly necessary to impose some discipline on the capacity of states to inflict harm on each other: this is why free movement law and the law governing State aids set limits to national regulatory autonomy. The rules of the EU constrain State practices because those affected by such practices in other Member States have no voice in the (national) political process that has generated them. A unilateral right to revoke an Article 50 notification would contradict this model. It would allow the revoking State to export costs to other States. Treating the Article 50 notification as irrevocable by unilateral act forces the parties to negotiate on the terms that shall attach to revocation, and this in turn ensures that (directly or indirectly) all parties which are affected by the preference to revoke have a say. In short, a revocation does not affect only the UK and so should not lie within the power of the UK alone.

The argument so far is normative: based on concern to allocate costs in a way that maximises incentives to account in advance for choices made and on concern to ensure a voice for all affected parties, an Article 50 notification should be treated as incapable of unilateral revocation. But this normative argument cannot prevail if the terms of Article 50 contradict it. They don’t.

Article 50 is of course silent on the question of revocation. So it is necessary to look at its structure and purpose. A case in favour of unilateral revocation is a case which serves to protect the flexibility enjoyed by the State which has issued a notification of its intention to withdraw. But that is not in line with either the structure or purpose of Article 50. Once the notification is made pursuant to Article 50, the terms of that Treaty provision dictate that the pace and content of the process is entirely subject to management by the EU-27 – which, to repeat, covers the other 27 Member States and the EU institutions. After notification, guidelines are provided by the European Council; the Union negotiates an agreement with the withdrawing State; it is to be concluded on behalf of the Union by the Council, acting by a qualified majority, after obtaining the consent of the European Parliament; the Treaties cease to apply to the withdrawing State from the date of entry into force of the withdrawal agreement or, failing that, two years after the notification unless the European Council, in agreement with the withdrawing Member State, unanimously decides to extend this period; and Article 50(4) shuts the withdrawing Member State out of the relevant discussions of the European Council and Council.

This is all about the way in which the EU-27 shall act: it is not at all about the protection of the withdrawing State. The primary concern which underpins Article 50 is to ensure that, once a Member State has chosen to submit its notification of intention to withdraw, the interests of the 27 Member States and the EU institutions then come to the forefront and are protected. To find an extra unilateral right of revocation enjoyed by the withdrawing State contradicts the structure and purpose of Article 50.

A very concrete example serves to emphasise the point. Were a withdrawing State to enjoy a right of unilateral revocation, it would be in a position to give notice of withdrawal and then change its mind after 23 months, once it concludes that the deal is not likely to be good enough. In fact, on this approach, a State could even change its mind after 23 months if it realises it is not going to get a good deal agreed inside 24 months, revoke the notice of withdrawal, and then submit another notice of withdrawal soon afterwards and try again. And it would not be liable to account for the costs.

Hazy suggestions that this would be an abuse of law are one answer to this problem – and for sure there would be political costs for a State behaving in such a perfidious manner. But the better solution, the one which is fully in line with the structure and purpose of Article 50 itself, is simply to exclude the possibility of unilateral revocation under Article 50. This interpretation also has the merit of deterring a State from triggering Article 50 unless it is absolutely sure that it wants to run the risks of getting a bad deal or even no deal – which, to repeat, is in line with Article 50’s structure and purpose, which is to protect the 27 not the 1. The government of a State that sends a notification pursuant to Article 50 and then waits nine months before initiating internal discussions on what shall be the preferred shape of its future relationship with the EU is in a spectacularly bad position, but it is entirely its choice to have placed itself in that location. There is nothing in the structure and purpose of Article 50 which encourages the view it should be allowed a unilateral right of revocation, thereby to impose costs on the EU-27 and to subvert the protections for the EU-27 carefully spelled out by Article 50. Quite the reverse: Article 50 should not be treated as providing a unilateral right of revocation.

A final point in closing the door to unilateral revocation is needed. Article 50 provides only for notification of an ‘intention’ to withdraw. An intention, it may be argued, can change. So can a Member State set aside its notification of an intention to withdraw on the basis that subsequently it has changed its mind and so no longer has that intention?  This is an attractive argument for those who would wish to maintain maximum flexibility for the UK. But it is not convincing. The notification of an intention to withdraw has been made. The clock starts to tick: the EU-27 acts in reliance on the notification and begins to make preparations for the negotiation. The intention may change subsequently, but the fact of having notified it within the meaning of Article 50 does not and cannot change. This attempt to create a right of unilateral revocation based on subsequent change of intention places an undue strain on the wording of Article 50, as well as running contrary to its structure and purpose.

I do not here address the politics of the matter. It is generally assumed that if the UK changed its mind a way would be found to accommodate its preference to remain a Member State of the EU. That may be right. But it may be complacent. If by the time of the change of mind major employers have decided to quit the UK in favour of the EU-27 (which might be precisely the impulse most likely to trigger the UK to change its mind) the incentives to agree to a UK request to terminate the Article 50 process would be weakened. Moreover, if the UK’s change of mind occurred amid a febrile political environment within which an imminent fall of government and its replacement by another minded to re-issue an Article 50 notification seemed plausible, then the EU-27 might be tempted to let the withdrawal process run its course.

My argument here is simply one of legal interpretation, not political context. But the politics are in part determined by the legal position. These political obstacles to the scope for the UK to change its mind and remain a member of the EU vanish if it enjoys a legal entitlement to revoke the Article 50 notification unilaterally. The UK can resume its current position cost-free (save only for longer-term reputational damage). By contrast if, as I argue, there is no right to unilateral revocation of Article 50, the UK will need to overcome those obstacles, and as part of that process it needs to negotiate on the terms to be attached to the readiness of the EU-27 to acquiesce in the change of mind. That puts the UK in a very weak position. It is the UK that has placed the UK in a very weak position. The structure, purpose and wording of Article 50 show that the solace of a unilateral right of revocation of the notification of withdrawal is not available.

The case for unilateral revocability of the Article 50 notice

Professor Steve Peers, University of Essex

Any form of retreat from Brexit seems unlikely, and any attempt at revoking the Article 50 notice would likely be a political process first and foremost. Still, it’s useful to consider whether unilateral revocability of an Article 50 notice is possible, in the event that the political situation changes radically; and I agree with Professor Weatherill that the legal position would have some relevance even though the politics would take precedence. (Most importantly, the power to revoke unilaterally would mean that the UK would not have to offer any quid pro quo in order to stay in the EU, and so would change the dynamics of the process).

In my view, while there are legal arguments for non-revocability and unilateral revocability, it’s hard to see the legal (as distinct from political) argument for ‘revocability only with consent’. And ultimately, while it’s anyone’s guess how the ECJ might ultimately interpret Article 50 if the case arose, my view is that the stronger argument is for unilateral revocability of the withdrawal notice.

Let’s start with the case law of the ECJ. There’s not yet any case law on Article 50 as such, of course, and so any case law could only be loosely relevant by analogy. The only ECJ case in the last few years I could find about a Member State revoking anything was this asylum law judgment of 2015, which as it happens I discussed at the time here. This case in part concerns whether a Member State can revoke a refugee’s residence permit, even though the relevant EU legislation contains no express reference to such a possibility.

In the court’s view (see paragraphs 47-50 of the judgment), ‘[d]espite the lack of express provision authorising Member States…to revoke a residence permit issued to a refugee, a number of arguments support an interpretation whereby Member States are allowed to take such a measure’. First, the EU legislation ‘does not explicitly rule out the possibility of revoking a residence permit’. Second, revocation of a permit was ‘consistent with the aim of’ the legislative rule (because the law says expressly that Member States could refuse to issue or to renew a residence permit on grounds of national security, it made sense that they could revoke a permit on such grounds). Thirdly, revocation was consistent with other provisions of the law relating to potential national security concerns about refugees.

Refugee law is different from the EU withdrawal process, of course. But there are some general points here. First of all, the Court rejects the argument (commonly made by those who think no revocation of the Article 50 notice is possible) that ‘absence of an explicit clause on revocation must mean that no revocation is possible’. Rather, the judges rule the opposite: if revocation is not explicitly ruled out, then it must be possible. Secondly, the aim of the law is important. This is indeed the main thrust of Professor Weatherill’s argument – but as we’ll see, I take a different view of the aim of Article 50. Thirdly, consistency with other provisions dealing with the same issue is important. This reasoning is sensible in principle, but it’s hard to see how to apply it by analogy to withdrawal from the EU, since Article 50 is the only provision on the subject.

I can foresee a fundamental objection here: the revocation of a residence permit concerns a Member State’s power to regulate non-citizens on its territory, which is quite different from its relations with other Member States. Fair enough. So let’s look at a process, set out in the EU Treaties, which does concern the precise question of whether a Member State which has made a notification to the EU institutions can revoke that notice unilaterally.

I’m referring to the process of ‘enhanced cooperation’, as set out in Article 329 and Article 331 of the Treaty on the Functioning of the European Union. This process, rarely used, allows some Member States to go ahead without the others to adopt EU legislation. The procedure gets underway when a group of willing Member States makes a request to the Commission to start it. But what happens if one of those States becomes unwilling to participate in the planned enhanced cooperation, during the process of negotiations? Can it, in effect, withdraw the notification of its intention to participate?

The EU Council legal service has given a detailed opinion on this issue. You can find it here, heavily redacted. But a helpful little bird has given me the entire text. Let’s see what it says on the key issues. (Note that the UK government has already relied on earlier legal advice to the same effect).

At the outset, the Council legal service notes that ‘[t]he Treaties are silent’ on the possibility of a Member State withdrawing its wish to join in enhanced cooperation during the negotiation process. But ‘[i]n the absence of any express provision, recourse must be had to the interpretation of the relevant Treaty provisions on enhanced cooperation, bearing in mind its objectives and two fundamental attributes that govern its functioning: first, the voluntary character of participation in it; second, its openness and flexibility.’

On the first point:

The decision as to whether or not to take part in enhanced cooperation is for the Member States alone, on the basis of their own wishes. The basis of enhanced cooperation is thus a voluntary act of each Member State. The expression of the wish to participate in enhanced cooperation and the subsequent withdrawal of such a wish is an act of free will of the Member States, whose voluntary character only ceases to exist at the moment of adoption of the act establishing enhanced cooperation. (footnotes omitted)

The first step in the process – authorising enhanced cooperation – ‘determines that the conditions for having recourse to enhanced cooperation are fulfilled, thus enabling the willing Member States to proceed with it. However, that decision does not create an obligation to do so’. (emphasis added)

On the second point, the openness of enhanced cooperation is ‘characterised by [the] simplicity’ of the rules, needing no fresh decision for a Member State to join enhanced cooperation which is already underway or under negotiation.

While it is not possible for participating Member States to withdraw from enhanced cooperation once it’s established:

Before that point in time - during the period from the date of adoption of the authorising decision to the establishment of enhanced cooperation - and bearing in mind the voluntary, open and flexible character of enhanced cooperation, a Member State is entitled to withdraw its stated intention to participate.

The first step decision to authorise enhanced cooperation ‘does not definitively establish the content thereof’ and Member States ‘must…be able to stop their participation in the establishment of enhanced cooperation’. Furthermore:

Not allowing unilateral withdrawal at this stage could have the effect of discouraging Member States from proceeding with enhanced cooperation before knowing its precise and definitive content, running thus counter the objective (sic) of "promot[ing] participation by as many Member States as possible” (Article 328(1) second subparagraph TFEU).

The procedure for a Member State to revoke its intended participation in enhanced cooperation should be simple and purely unilateral:

In the view of the Council Legal Service, no modification of the authorising decision is required for the withdrawal of a Member State to take effect. In the absence of any formality required by the Treaties, a Member State wishing to withdraw from enhanced cooperation should simply notify its intention to the Council and to the Commission in written form or through a statement delivered at a Council meeting. No further act of the Council and/or of the Commission is necessary for the notification to take effect. As from the moment of such notification, the Member State concerned should no longer be considered as wishing to participate in the establishment of enhanced cooperation and therefore, should not take part in the vote for the adoption of the act.

….A more formalistic approach requiring, for instance, a modification of the authorising decision would ignore the voluntary, open and flexible character of enhanced cooperation, since the interested Member State would not be able to withdraw on its free will, but would depend on the will of the Commission to propose an amending act of the authorising decision, of the Council to adopt the amended authorising decision and of the European Parliament to grant its consent. (footnotes omitted)

While a legal service opinion is not binding – the Council wins some cases, but loses others – this is an interesting analysis which could apply by analogy to the Article 50 process. The absence of a provision on revocation of intention is not decisive: instead, the key point is the voluntary and flexible nature of the process, and the overall objectives of integration. A Member State can change its mind unilaterally, even after the first step in the procedure has passed, as long as the final decision has not been made. Article 50 is equally a voluntary process, and also has a first step (notification of withdrawal) followed by a negotiation phase before withdrawal becomes definitive. And the overall integration objectives of the EU are obviously served by facilitating a Member State’s wish (if it chooses) to reconsider its decision to leave the Union.

I’ll now turn to the detailed arguments of Professor Weatherill. First of all, he argues that, as a matter of policy, the costs of Article 50 negotiations are an issue which suggests unilateral revocation is not possible. But such an approach is not taken in analogous EU procedures. Some proposed EU laws can be blocked by the veto of a single Member State. Terminating the negotiation process in the event of such a veto imposes costs, but no one argues that a Member State therefore lacks a power to wield its veto unilaterally in such cases. Equally a proposed Treaty amendment can be frustrated by the failure to ratify by a single Member State. And the strongest analogy is the accession process, which can be scuppered by a would-be Member State ending negotiations or refusing to ratify an accession treaty, despite the enormous efforts expended on the ratification process.

In all these cases, the power to act unilaterally is unquestioned, despite the inconvenience caused to other States and EU institutions. That awkwardness stems necessarily from the exercise of national sovereignty – the very sovereignty which is embodied in the right to withdraw from the EU, to which Article 50 gives effect. Furthermore, in the particular context of the UK, it seems odd to be concerned about the costs of a revocation of the Article 50 notice which would be suffered by the EU27, given that a U-turn on Brexit would mean that a major net contributor to the EU budget keeps on contributing. In any event, holding detailed negotiations on revocation of withdrawal will impose more negotiation costs than simply accepting the revocation and returning to the status quo – with Adonis resuming skiing and Farage resuming screeching. (Yes, I know: they never really stopped).

On the issue of democracy, Professor Weatherill again repeats the argument as regards costs. But let me repeat the same rejoinder: the very possibility of withdrawal gives priority to the democratic choices made in the withdrawing Member State over the impact of that choice in the remaining Member States. Indeed, ‘a revocation does not affect only the UK’; but the same could be said, with much more force, of the power of withdrawal. Yet the Treaties accept that the power of withdrawal lies with the UK alone; vesting the power of revocation with the withdrawing Member State alone is a logical corollary.

As for the wording of Article 50, I draw the opposite conclusion from it. Indeed, there are detailed rules on the roles of the EU institutions and the EU27 Member States in negotiating and concluding a withdrawal agreement. But there are no such rules on the role of those institutions and States in the event that a Member State revokes its notification. One can reasonably deduce from this either that no revocation is possible, or that unilateral revocation is possible – although note that the ECJ case law and Council legal service opinion above prefer the latter interpretation in analogous cases. But the least plausible interpretation is that the Treaty drafters intended such rules but just forgot to mention them. To strengthen the argument, look at the detailed procedural rules relating to Treaty amendment, accession, and the negotiation of treaties between the EU and non-EU states, which were all amended in the same Treaty of Lisbon that created Article 50. How odd for the Treaty drafters not to set out rules on the procedure for agreeing revocation of an Article 50 notice at the same time.

What about the possibility of a Member State revoking its withdrawal notice and then issuing it again, to get more time to negotiate withdrawal? I agree with Professor Weatherill that the notion of ‘abuse of law’ is indeed ‘hazy’. But at least this notion already exists in EU law, and so could be clarified if necessary in the event of such alleged abuse in the Article 50 context. On the other hand, Professor Weatherill’s argument for negotiated revocability relies on conjuring up procedural rules which don’t explicitly exist in the Treaty, and are not recognised to exist in any case law to date. If ‘abuse of rights’ is a hazy notion, then ‘procedure for negotiating revocation’ is a Dickensian fog by comparison. 

Equally, if revoking a withdrawal notice places a strain on the wording of Article 50, then inventing a process to negotiate revocability places a bigger strain still. And while one purpose of Article 50 is undoubtedly to protect the remaining Member States’ interests, another purpose is to give priority to the sovereign decisions of the withdrawing Member State. The best way to reconcile the alleged conflict between these two purposes in the event of a unilateral revocation of a withdrawal notice is to conclude, as I noted above, that all the underlying objectives of the EU are best served by allowing a withdrawing Member State to decide to remain in the EU after all, should it wish to do so. This neatly balances the EU’s interest in its integration process with a Member State’s sovereign power to decide on whether it wishes to leave or stay in the EU.

Barnard & Peers: chapter 27

Photo credit: PA Images