Abstract
Following the 2016 referendum, the UK notified its intention to withdraw from the European Union pursuant to Article 50 TEU. Given the political and legal consequences of a much-questioned referendum and the strong opinion of many parts of British society that the UK’s membership should not be terminated, the question arose whether such a notification could be revoked unilaterally. In the absence of any mention in Article 50, expert opinion was divided. International law – that is, the law of treaties and the law of international organizations – does not appear to provide a definite answer, while state practice is rather scarce. The constituent instruments of international and regional organizations containing withdrawal clauses are also silent, except for African organizations and development/investment organizations, which invariably allow Member States to rescind withdrawal notices. As regards the EU Treaties, before the Lisbon Treaty they did not contain a withdrawal clause. In the preliminary ruling given in Wightman v. Secretary of State for Exiting the European Union, which concerned whether an EU Member has the sovereign power under Article 50 to revoke unilaterally a withdrawal notice, the Court of Justice helped to clarify a critical question of EU Law but also of international law.
Keywords
1. Introduction
In its preliminary ruling of 10 December 2018 in Wightman v. Secretary of State for Exiting the European Union that a Member State retains the sovereign power to revoke unilaterally a notice of withdrawal from the European Union (EU), the Court of Justice of the European Union (CJEU) has filled in a significant omission in the withdrawal clause of the Treaty on the European Union (TEU), namely Article 50. 1 But more broadly, the CJEU has additionally helped to clarify a critical question of international law relating to the law of treaties and the law of international organizations (IO). 2 Wightman is therefore important because it is a rare case in which an international court has considered the question of a Member State’s withdrawal from an IO and the first in which the withdrawal’s revocability has featured as the principal issue. 3 Although the case centred on Article 50 TEU, the authority of the Vienna Convention on the Law of Treaties 1969 (VCLT), which has a residual role in EU Law, also featured in the deliberations. 4
This article considers the complex question of a Member State’s withdrawal from an IO in light of the problems and challenges that have arisen as a result of the United Kingdom’s (UK) withdrawal process from the EU, what became known as ‘Brexit’. In particular, after offering a detailed overview of the constitutional and EU Law ramifications of Brexit, the article examines the possibility of a withdrawal notice being revoked, retracted or cancelled, focusing on the CJEU’s pronouncements on the law of treaties regarding termination of membership in IOs.
The substantive section of this article begins with a discussion of the international law on the revocation of notices of withdrawal. As the TEU shares the inclusion of a withdrawal clause with many constituent instruments of IOs, a brief overview will be undertaken, especially of those provisions permitting revocation, to place Article 50 TEU in context, followed by questioning whether a general right to cancel notification of withdrawal exists in international law. A state of uncertainty is revealed. A short historical account on the issue of withdrawal from the European Economic Community/European Community, which was not envisaged in the then Founding Treaty, leads to a brief consideration of Article 50 TEU, and, thereafter, to an analysis of Wightman’s contribution to interpreting the relevant provisions of the VCLT.
2. Pursuit of Brexit leads to a constitutional crisis in the UK
In June 2016, an ill-conceived and flawed referendum on EU membership was held in the UK. 5 The referendum raised a number of serious concerns, including who was entitled to vote and who was excluded, 6 as well as a regrettable campaign partially conducted on the basis of mendacity, wilful misrepresentations and xenophobia. 7 And elements of illegality may have tainted it. In Wilson v. Prime Minister, the Court of Appeal acknowledged this fact, although, with no hint of irony, it expressed the view that the application for judicial review was ‘an attack on the democratic process’. 8 In fact, the Electoral Commission did impose a number of fines on campaign groups for exceeding spending limits (although some were reversed on appeal). 9 Again, events in the UK appear to compare unfavourably in view of similar developments in other European 10 and non-European jurisdictions. 11 It could be counter-argued that, even if there might have been irregularities and illegalities, these were insufficiently serious to justify the referendum’s annulment and to cancel the process of withdrawal. However, it is submitted that the matter should have been investigated, if for no other reason than to uphold faith in democratic processes and to dispel any doubts whether the standards of legitimacy on the holding of referenda, as laid down by the Council of Europe’s Venice Commission, were fully met. 12
Described by one political commentator as ‘a shallow act of direct democracy’, 13 nevertheless the referendum did result in a close win for the Leave campaign. 14 On 29 March 2017, the UK Government gave formal written notice, pursuant to the EU (Notification of Withdrawal) Act 2017, of its intention to terminate membership under Article 50 TEU. 15 But the ineptly managed process of withdrawal gave rise to multiple crises in the British body politic which, until the General Election of December 2019, veered on a day-by-day basis from Greek tragedy to bedroom farce. Brexit claimed the premierships of Prime Ministers David Cameron and Theresa May, resulted in a record number of resignations by Cabinet and junior ministers, 16 pitted the executive versus Parliament 17 and undermined the foundational pillars of the British Constitution, including the sovereignty of Parliament, 18 the rule of law, 19 constitutional conventions 20 and the role of the Head of State. 21 The minority Johnson Government in particular demonstrated a scandalous contempt for the UK’s democratic norms and institutions, deplorable in a supposedly mature and stable democracy. 22 Due to various reasons, partly to do with the Government’s internal workings, the UK Parliament was unable to approve the Withdrawal Agreement and Political Declaration, which had been negotiated between the UK Prime Minister and the EU and endorsed by the European Council in November 2018 in accordance with Article 50(2) TEU. 23 It was signed by the EU in January 2019. 24 These events eventually led to the removal of Prime Minister May in late May 2019.
The UK’s withdrawal was to come automatically into effect on 28 March 2019. When it transpired that the House of Commons would not have ratified the Withdrawal Agreement on time, 25 on 20 March 2019 the UK submitted a written request to extend the withdrawal period until 30 June 2019. As permitted under Article 50(3) TEU, the European Council granted the request but with a very different formulation: the period was extended until 22 May 2019, if the House of Commons had approved the Agreement by 29 March 2019, otherwise the UK had to indicate before 12 April 2019 ‘a way forward for [the European Council’s] consideration’. 26 The Withdrawal Agreement again failed to secure the House of Common’s approval. Thus, on 5 April 2019 the UK submitted a written request for a further extension, until 30 June 2019, to finalize ratification. The European Council complied with the request but prolonged the period considerably. Thus, withdrawal would have taken place on the first day of the month following the completion of the domestic ratification procedures or on 1 November 2019, whichever was the earliest. Effectively, this resulted in postponing withdrawal until no later than 31 October 2019. 27 The European Council attached several conditions, the following two standing out: (a) the re-opening of the Withdrawal Agreement was excluded, and it could not have been invoked to start negotiations on the future EU-UK relationship; and (b) any unilateral commitment, statement and act to be made by the UK vis-à-vis other Member States should have been compatible with the Agreement’s letter and spirit and not hamper its implementation. 28
Despite his bombastic assertions to the contrary, in October 2019 Prime Minister Johnson was compelled to seek yet another extension, which the European Council granted until 31 January 2020. 29 Following the General Election of December 2019, which resulted in a majority Conservative Government, on 24 January 2020 the Agreement on the UK’s withdrawal was concluded in Brussels and in London. 30 This was a renegotiated version of the text approved by the UK and the EU on 17 October 2019 and accepted by the two sides in November 201\ 31 The Withdrawal Agreement was approved by the Council on behalf of the EU on 30 January 2020. 32 The following day the UK finally withdrew, 33 the legal basis being the European Union (Withdrawal Agreement) Act 2020, 34 which provided for the Withdrawal Agreement’s ratification and incorporation into domestic law. The Act received Royal Assent on 23 January 2020 – the day before the Withdrawal Agreement was actually concluded.
3. Could the process leading to Brexit have been reversed or cancelled?
As explained, Decision 2019/584 extended the UK’s withdrawal to 31 October 2019. It also contained an intriguing introductory paragraph, which referred to a point of considerable importance unaddressed by Article 50 TEU. In particular, paragraph 10 thereof read, ‘This further extension cannot be allowed to undermine the regular functioning of the Union and its institutions…the United Kingdom will remain a Member State until the new withdrawal date, with full rights and obligations in accordance with Article 50 TEU…the United Kingdom has a right to revoke its notification at any time’ (emphasis added). Throughout the duration of the EU-UK negotiations, whether the notification could have been cancelled was not only raised as a theoretical proposition but became a recurring matter, indeed an explosive political and legal issue. And it acquired increased prominence on account of Article 50’s silence on the matter giving rise to claim and counter-claim. In Miller No. 1, the UK Supreme Court held that it was ‘common ground that notice…once given…cannot be withdrawn’ 35 (although it did not actually rule on this point), whereas, in a Report issued before the referendum, the House of Lords European Union Committee had found that there was no formal impediment to a notification being rescinded. 36 Indeed, that was the view of the expert witnesses, including Professor Emeritus Sir David Edward, former judge of the CJEU, before the said Committee, 37 and was a conclusion shared by an eminent body of opinion, including that of Lord Kerr of Kinlochard, an architect of Article 50, 38 as well as that of senior lawyers 39 and leading academics. 40 Such opinion extended to political parties, including the Liberal Democrats 41 and the Scottish National Party, 42 as well as by elements of civil society.
It quickly became evident that a definitive answer could be provided only by the CJEU. In September 2018, Scotland’s Court of Session requested a preliminary ruling from the CJEU in proceedings initiated by a number of politicians, including members of the Scottish Parliament and of the European Parliament, on whether the notification of the intention to withdraw under Article 50 TEU may be retracted unilaterally. 43 The existence of a right to revoke was therefore the crux in Wightman. Since Article 50 TEU is silent, the probability was high that the CJEU would apply international law and the VCLT to address it. This in turn raised the question as to what the governing international law was. The answer, as discussed below, is that its state is uncertain.
4. Constituent treaties of international organizations permitting retraction of a notice of withdrawal
The TEU shares with many constituent instruments of international and regional organizations, as well as with UN Specialized Agencies, withdrawal clauses with accompanying procedural conditions, including a period of time, before denunciation becomes effective. 44 Suffice to mention the Universal Postal Union, 45 the International Labour Organization, 46 UNESCO, 47 the World Trade Organization, 48 the Organization of American States, 49 the League of Arab States, 50 the Council of Europe, 51 the International Centre for the Settlement of Investment Disputes, 52 the Organization of Islamic Cooperation 53 and so on. 54 Article 54(a) VCLT requires that the state concerned must comply with the formal conditions laid down in the relevant treaty where applicable. Any procedural irregularity will render the withdrawal notice null and void, 55 presumably because, by analogy, any such irregularity in the instrument of ratification will lead the state in question not becoming a member of the IO concerned.
However, there are a number of examples where the IO concerned and its membership have exercised sound judgment and have afforded the withdrawing state considerable latitude. 56 One such example concerns Poland’s relationship with the ILO in the 1980s. Specifically, on 17 November 1984 Poland submitted its notice of withdrawal pursuant to Article 1(5) of the ILO Constitution, which stipulates that the notice shall take effect two years following the date of receipt by the Director-General and the withdrawing Member having, by that time, fulfilled all its financial obligations. 57 Therefore, Poland ought to have ceased being a Member State on 17 November 1986. However, four days earlier, it informed the ILO that it was unilaterally extending the notice for a further 12 months. Even though the ILO Constitution does not envisage this eventuality, there was no official reaction by the other Member States, which, by necessary implication, accepted a three-year withdrawal period. On 17 November 1987, Poland announced that it had withdrawn its notice. The ILO accepted its retraction even though, strictly legally speaking, it had already come into effect. 58 However, even if Poland were considered to have permanently left the Organization, it could have literally returned the following day. And this because of the re-admission clause in Article 1(6), which in conjunction with Article 1(3) thereof, stipulate that any UN Member State (and Poland was one of them) may join the ILO by merely communicating to the Director-General its formal acceptance of the obligations emanating from the ILO Constitution.
None of these constitutive instruments expressly refer to revoking a withdrawal notice. However, there are two groups/categories of IOs, which envisage that a notice may be retracted or cancelled without any consequences for the withdrawing state. The first group are African IOs, the second are international and regional development/investment/trade finance IOs usually operating as multilateral banks. Each of these categories will now be examined.
In the African Union (AU), a Member State’s renunciation of membership may be validly withdrawn, 59 which was also true for the predecessor Organization of African Unity (OAU). 60 The African Economic Community contains a comparable provision. 61 Similar clauses are found in most of the African sub-regional organizations, the so-called ‘African Regional Economic Communities’ – RECs. Thus, under Article 34(1)-(2) of the Southern African Development Community (SADC) Treaty 12 months’ written notice must be given but that notice may be revoked before its expiry. 62 This is equally the case with the Economic Community for West African States (ECOWAS), 63 the East African Community (EAC), 64 the Economic Community of Central African States (ECCAS), 65 the Inter-Governmental Authority on Development (IGAD) 66 and the Common Market for Eastern and Southern Africa (COMESA). 67 They all openly stipulate that states reserve the right to reconsider their decision to depart, and invariably use the term ‘unless withdrawn’. 68 The critical point is that none of them make the revocation subject to the consent of the respective IO’s organs or its Chairperson, or on the other Member States’ approval: withdrawal notices are never final. The decision to retract is clearly considered a sovereign choice and this is consistent with the somewhat absolutist approach that traditionally African states have taken in relation to sovereignty.
The second category comprises international and regional organizations which are active in the area of development, investment, trade finance and so on, and which have been established since the 1960s in all parts of the world. The relevant clause reads along these lines: ‘However, at any time before the withdrawal becomes finally effective, the Member may notify the [international organization in question] in writing of the cancellation of its notice of intention to withdraw’. This wording would suggest that, until and unless withdrawal has materialized, the state in question has only given an intention to do so. Suffice to mention the following IOs: the Council of Europe Development Bank, 69 the European Bank for Reconstruction and Development, 70 the Black Sea Trade and Development Bank, 71 the Inter-American Development Bank, 72 the Inter–American Investment Corporation, 73 the Caribbean Development Bank, 74 the Asian Development Bank, 75 the Asian Infrastructure Investment Bank, 76 the Islamic Investment Bank 77 and the International Islamic Trade Finance Corporation. 78
Apart from these African organizations and the aforementioned development/investment IOs, constituent instruments rather fail to stipulate whether formal notification may be subsequently retracted. But absent such express specification, does a general or implied right to rescind a notification exist in international law?
5. Does a general right to cancel a withdrawal notification exist?
One of the explanations given for the procedural requirement of time is ‘to allow for reconsideration’, 79 which logically implies that minds can change leading to a reversal of policy. This change can derive from a democratic mandate as expressed in a referendum or from a change in government, or because of new or altered circumstances, which make redundant the reasons behind withdrawal. A Member State, or its government (the executive branch), may thus resolve to withdraw as a protest at the way it believes the IO may have deviated from its stated goals or at the manner it functions (or fails to function) or at the fact that it favours one particular Member State over another or that it has allowed itself to be manipulated by one or more Member States, and so on. If, subsequently, the IO mended its operation and addressed the concerns of the complainant Member State, the reason for withdrawal would presumably lapse.
How the strained relationship between a Member State and an IO may lead the former to an abrupt withdrawal from the latter was recently manifested in the attitude of the USA towards the World Health Organization (WHO). On 18 May 2020, the US President gave WHO a period of 30 days to improve its performance significantly as he was dissatisfied with its handling of the COVID-19 pandemic. 80 The following day, the WHO governing body, the World Health Assembly (WHA), decided by consensus to promptly initiate an impartial, independent and comprehensive evaluation of the health response to the pandemic. 81 However, before the 30-day ultimatum expired, on 29 May 2020 the US President announced that the USA would withdraw: ‘We have detailed the reforms that [the WHO] must make and engage with them directly, but they have refused to act. Because they have failed to make the requested and greatly needed reforms, we will be today terminating our relationship with the World Health Organization.’ 82 That the US President considered that a Member State could withdraw on the very day the announcement was made is remarkable but it may have to do with the fact that the WHO Constitution does not contain a withdrawal clause, echoing the paradigm of the UN Charter which had been signed one month earlier. 83 But the President of Brazil also appears to have been sided with the argument that there is a right akin to automatic withdrawal. 84 In the event, the official US Letter of Withdrawal was sent on 6 July 2020 to the UN Secretary General stipulating that it will become effective on 6 July 2021. 85
Both the USA and Brazil are original WHO Member States; indeed, it was Brazil which in April 1945, together with China, proposed that an international health organization be established and to convene a conference to prepare its charter. But although Brazil’s ratification instrument of the WHO Constitution did not include any references to the issue of withdrawal, the US instrument did: according to a joint resolution of the House of Representatives and the Senate reached on 14 June 1948, the USA reserves its right to withdraw on a one-year notice, provided that all financial obligations are met. 86 A few weeks later, the First Session of the WHA ‘recognized the validity of the ratification of the Constitution by the United States of America’ and resolved that the UN Secretary-General be so advised. 87 The events of 1948 show that even when a contracting party ‘invents’ a right to withdrawal, it allows ample time to have outstanding issues sorted out with the IO in question and presumably to re-think the decision to leave it permanently. In relation to the current US withdrawal, one should wait for the outcome of the November 2020 presidential elections to ascertain whether a new President will cancel President Trump’s Letter of Withdrawal.
In short, the decision to withdraw can reflect that the relationship between an IO and its membership is never static but ever evolving. There is a body of opinion that notice may be rescinded as a matter of general treaty law, particularly as it upholds the stability of treaties, a fundamental principle of the international legal order. 88 Significantly, it is also the view of the ILC, which produced the draft text of the VCLT. 89 This conclusion also draws support from state practice.
A. Practice regarding withdrawal from international organizations
Given that many IOs, regional organizations and Specialized Agencies permit withdrawal but are silent on whether notice, once given, may be cancelled, it would be instructive to take account of historical precedents. 90 Little usage exists but there are some noteworthy instances, for example the League of Nations, the Covenant of which expressly provided for withdrawal subject to two years’ notice and the fulfilment of all obligations. 91 Sixteen Member States withdrew over the years. 92 For present purposes, important is the example of Brazil and Spain: in 1926 they notified their intention to withdraw in protest at Germany having been made a League Council permanent member while their own claims had been unsuccessful. The other Council Members expressed their hope that these notifications would be retracted, which turned out to be the case with Spain. 93 In 1932 Mexico also cancelled its notice. 94 Thus, nearly a century ago, it was accepted that formal notifications could be rescinded while the remaining Member States actively expressed the wish that they be revoked.
Further examples of situations where an IO believed that a withdrawal notice could be cancelled concerns UNESCO, at a time when its Constitution contained no clause on withdrawal. In response to Poland’s notification in 1952 of its intention to withdraw, UNESCO’s General Conference invited Poland to reconsider its decision. 95 This was similarly the case when South Africa submitted its notice of withdrawal in 1955. 96
More recent examples, recounted by Advocate General Campos Sánchez-Bordonain in his Opinion in Wightman, concern the purported withdrawals from the Rome Statute of the International Criminal Court (ICC) 97 by The Gambia and South Africa. 98 In both situations the withdrawal notices were cancelled, in the former case because of a change in government 99 and in the latter case because of a domestic court ruling that the withdrawal notice was unconstitutional and therefore invalid. 100 In March 2017, the South African government officially cancelled its notice. 101
There exist examples of revoked withdrawals from IOs the constitutions of which do not contain a withdrawal clause. 102 Indonesia’s purported departure from the UN in 1965 only to return the following year provides a possible example, although the situation is somewhat uncertain. One assessment is that Indonesia’s action should properly be considered as a temporary measure of non-cooperation rather than an actual withdrawal followed by a return. 103 Proceeding on this basis, it transpires that Indonesia was allowed to reconsider its decision. In 1965 Indonesia reversed its decision to withdraw from UNESCO in a letter of 30 July 1966 ‘superseding the notice of withdrawal of 12 February 1965 which has not yet taken effect’. 104
A more recent example, also referred to by the Advocate General in Wightman, is that of Panama’s abortive withdrawal from PARLACEN in 2009, one of the organs of the Organization of Central American States (ODECA). 105 The CACJ decided on two separate occasions, the first an advisory opinion, the second a contentious case, that a Member State cannot rely upon Article 56 VCLT to denounce or withdraw unilaterally from an IO the constitution of which does not mention withdrawal. 106 The CACJ placed particular emphasis on the principles of pacta sunt servanda and good faith. Panama was thus ordered to refrain from unilateral actions that breach the principles and purposes of the Central American Integration System. Subsequently, following the ruling of Panama’s Supreme Court of Justice that the government had acted unconstitutionally, 107 Panama’s withdrawal notification was rescinded, and Panama resumed its place in PARLACEN. 108
At no time has it been suggested in relation to any of the above examples that it was unlawful or at variance with international law to revoke withdrawal notices, quite the contrary. Practice, which may be considered as consistent, therefore indicates that a notice of withdrawal can be retracted.
B. The Vienna Convention on the Law of Treaties
The law of treaties in relation to withdrawal from an IO has been criticized as uncertain. 109 Nevertheless, the VCLT contains general rules regarding withdrawal from a treaty, which extends to the foundational treaty of an IO. Article 42(2) thereof reads, inter alia, that withdrawal may take place only as a result of the application of the provisions of the treaty in question or of the terms of the VCLT, which are exhaustive. 110 The effect of the former condition is that in casu, in conjunction with Article 5 VCLT, the terms of Article 50 TEU must, in principle, be followed to the exclusion of those of the VCLT under the lex specialis rule. 111 The latter condition is a reference to grounds of termination contained in the VCLT such as material breach or rebus sic stantibus. This is followed by Article 54 VCLT, which affirms that a party may withdraw from a treaty either in conformity with the treaty provisions, or with the consent of all the parties after consultation with them. 112 But alternatively, it could be said that the VCLT only sanctions withdrawal in the limited circumstances set out therein. 113 The solution appears to lie elsewhere in the VCLT. Article 65(1) VCLT, headed ‘Procedure to be followed with respect to invalidity, termination, withdrawal from or suspension of the operation of a treaty’, states, inter alia, ‘A party which, under the provisions of the present Convention, invokes either a defect in its consent to be bound by a treaty or a ground for impeaching the validity of a treaty, terminating it, withdrawing from it or suspending its operation, must notify the other parties of its claim.’ But, in the case of Brexit, an apparent difficulty with relying on Article 65 is that there does not appear to be an intention in the VCLT to recognize or grant a general unilateral right of withdrawal but only a right to invoke in relation to the preceding specified grounds, such as material breach or supervening impossibility of performance. This conclusion is based on an interpretation of the wording of Article 65, which specifies a defect in a state’s consent to be bound by a treaty or a ground for impeaching the validity of a treaty or withdrawing from it (emphasis added). Withdrawal is therefore predicated on grounds of invalidity and not as a general right. 114 However, Article 67(2) VCLT seems to provide a more solid basis in that it refers to any act declaring invalid, terminating, withdrawing from or suspending the operation of a treaty, inter alia, pursuant to the provisions of the treaty (emphasis added). This provision therefore appears to confer a right under the terms of a particular treaty that is distinct and separate to grounds of invalidity under the VCLT. 115 For present purposes, the significant provision appears to be the procedural rule in Article 68 VCLT which states, inter alia, that a notification provided for in Article 67 may be revoked at any time before it takes effect (emphasis added), a clear statement that a notification may be rescinded. 116
These provisions of the VCLT were to play a part in Wightman. The goal of Article 68 VCLT has been described as providing ‘an opportunity to a state intending to set the procedure in motion to stop it in its tracks’. 117 Given that Article 50 TEU is silent on whether the notice of withdrawal may be revoked, it appears prima facie that this specific issue should be read in light of Article 68 VCLT where appropriate. 118 However, as the VCLT is in principle inapplicable between non-parties and given that neither the EU nor all its Member States are contracting parties (France and Romania), it could be argued that the pertinent international law should be customary international law and not the VCLT. 119
C. Customary international law
Expert opinion is divided on whether the revocation of a notification of withdrawal enshrined in Article 68 VCLT has the status of a rule of customary international law. As has been seen above, there are examples of international practice to that effect, one of the key elements of customary law, even if few in number. This fact should not in itself prevent a rule of custom emerging should the necessary opinio juris exist. 120 There are abundant instances of such a right in the constituent treaties of regional, African, organizations, which can either be considered a codification of existing custom or as a contributing factor in crystallizing a customary norm. Indeed, the pervasive use of such a condition among African states could be suggestive of a regional procedural norm. 121 But given the dearth of actual instances of revocation, it is permissible to take account of relevant statements by states and international bodies as evidence of the formative elements of customary law. 122 Significantly, the ILC adopted Article 68 VCLT without dissent and neither did it attract specific objections at the UN Conference on the Law of Treaties. 123 Indeed, the ILC has steadfastly taken the view that withdrawal notifications may be revoked. Its position on what became Article 68 VCLT was that it had concluded that ‘the considerations militating in favour of encouraging the revocation of notices and instruments of denunciation…are so strong that the general rule should admit a general freedom to do so prior to the taking effect of the notice or instrument.’ 124 In relation to the VCLT-IO, the ILC was of the opinion that, ‘As long as these instruments [of withdrawal] have not taken effect, they can be revoked.’ 125 South Africa has also expressed the view recently that a notice of withdrawal may be revoked. 126 The ICJ’s statement in the Gabčíkovo-Nagymaros Project Case relating to Articles 65 and 67 VCLT may be relevant by extension, that ‘if not codifying customary law, at least generally reflect customary international law and contain certain procedural principles which are based on an obligation to act in good faith’(emphasis added). 127 There is thus academic opinion supportive of the view that revocation is a rule of customary law. 128
But the weight of academic opinion seems to be against such a conclusion. 129 Reliance is placed on the contention that the procedural rules enshrined in the VCLT were considered by the ILC as progressive development, and consequently not yet part of international law, and not declaratory of existing custom. 130 A counterargument would be that even if this line of reasoning were accepted, some 50 years have now elapsed during which time a customary norm may have crystallized, prompted by its codification, or been generated. 131 It has additionally been argued that procedural rules cannot have a customary nature 132 but there appears to be no reason in principle why this should be so if they have a norm-creating character. The rules on state immunity provide one conspicuous example. 133 But this would be to ignore the fact that existing judicial opinion, albeit very scarce, is also negative. AG Jacobs suggested before the European Court of Justice (ECJ) that the procedural rules, including Article 68 VCLT, did not enshrine a rule of customary law. 134 The ECJ agreed insofar as Article 65 VCLT was concerned. 135 But the required opinio jurisis is always difficult to discern; there seems little sense of legal obligation in the practice described above which may amount to mere usage. 136 In light of this difference of opinion and the lack of authoritative evidence it would be a stretch to suppose that Article 68 VCLT codifies or reflects a norm of customary international law.
Whilst revocation of a withdrawal notice may not constitute a rule of customary international law, it is an established fact that there are instances of such occurrences notwithstanding the absence of express authorization. The answer to this apparent puzzle seemingly lies in the sovereignty of states and specifically their treaty-making capacity. It rests on the residual principle of state freedom in the absence of any applicable law. The so-called ‘Lotus presumption’ asserts that ‘restrictions upon the independence of States cannot…be presumed’, that what is not prohibited is consequently allowed. 137 Judge El-Erian described withdrawal from an IO as a feature of sovereignty. 138 Barring any treaty limitation to the contrary or other self-imposed restriction, states must be considered free to revoke a notice of withdrawal.
There do not appear to be any known cases of: (a) constitutive instruments of IOs expressly prohibiting or limiting the revocation of a withdrawal notice; or (b) Member States willingly restricting their capacity to revoke a withdrawal notice. But the question arises whether states are free to revoke if the submission of the notice has given rise to a legitimate expectation for all or some of the other Member States that the state in question will withdraw? Could it be the case that the notice creates a situation of legal certainty, in the sense that the proper functioning of an IO should not be disrupted when a state unilaterally revokes the notice, even though there is no such stipulation in the withdrawal clause and the other Members and/or the IO organs have not asked the withdrawing state to reconsider. It is possible to distinguish withdrawal from an IO from denunciation of a treaty because the very conception of an IO lies in the close and perhaps multifaceted institutional cooperation which is established among the Member States, while the links among contracting parties to a multilateral treaty are invariably much looser. It is possible to imagine a scenario where the procedure for withdrawal has been initiated but where the withdrawing Member State suddenly cancels the notice a few days before the expiration of the period of the notice, putting the IO in a difficult situation. Theoretically, a hostile Member State could submit notices only to retract them near the time of withdrawal simply as a means of deliberately disrupting the IO’s proper functioning. 139 Of course, such scenarios would not arise if a state was acting in good faith 140 but this cannot be presumed. It is arguable that the VCLT and/or customary international law should take account of not only the different nature and purpose of ‘ordinary’ treaties versus the constitutions of IOs but also that participation in IOs creates complex relations between states and with the increasingly important role of IOs in the global community these relations have to observe certain fundamental principles of the law of IOs, for example legitimate expectations, legal certainty, and so on.
6. Withdrawal from the European Economic Community/European Community
As has been observed, withdrawal from the EU is governed by Article 50 TEU but prior to the Lisbon Treaty, which entered into force on 1 December 2009, the original Treaty of Rome, even though amended on several occasions, did not contain a withdrawal clause. 141 The EEC/EC was established for an indefinite period. 142 Therefore, the generally held view was that one-sided denunciation was excluded and membership could not be terminated because Member States had consciously ruled out their right to withdraw unilaterally. 143 Nevertheless, some supportive of the right of withdrawal have sought to rely on the general rules set out in the aforementioned Article 54 VCLT, 144 and Article 56(1) VCLT which states that the treaty is not subject to denunciation or withdrawal unless (a) it could be deduced that the parties intended to allow the possibility of denunciation or withdrawal, or (b) a right to do so is implied from the nature of the treaty. 145 An analogy could also be drawn with the UN: its Charter contains no withdrawal clause, which seems to accept the possibility of withdrawal in ‘exceptional circumstances’. 146 With regard to the first possibility, notwithstanding the lack of travaux preparatoires, it appears that the parties’ intention was not to permit withdrawal, 147 whereas the second ground can be discounted given the supranational character of the treaties. This conclusion finds support from the ECJ, which stated that power could not be withdrawn from the EEC except as a result of an express provision of the EEC Treaty, which of course did not exist. 148 The ECJ’s stance receives backing from the CACJ which has found on two separate occasions that a Member State cannot rely upon Article 56 VCLT to withdraw unilaterally from a supranational organization the constitution of which is silent on withdrawal. 149 In contrast, the Bundesverfassungsgericht (German Federal Constitutional Court) has held that the Member States are the ‘Masters of the Treaties’ (Herren der Vertraege) and that Germany, as indeed any other Member State, has retained its sovereign right to terminate its membership and cannot be prevented from withdrawing by the other Member States and/or by the EU’s autonomous authority. 150 In the final analysis, it is not in anyone’s interest, least of all the organization’s, to compel an unwilling state to remain a member. The legal arguments notwithstanding, in 1975 a referendum on membership of the then EEC was held in the UK which resulted in a convincing victory for continued membership. 151 It is not clear what would have occurred had the British public voted to leave. However, in 1985, Greenland, an autonomous region of Denmark, did withdraw from the EC but this was considered an exceptional episode without wider implications, either on the issue of withdrawal from the EC or on Denmark’s continued membership of the EC. 152 Finally, an explicit right of withdrawal was inserted into the abortive Treaty establishing a Constitution for Europe, 153 mainly at British instigation. 154 It was retained by the Lisbon Treaty. 155
7. Article 50 TEU
Under Article 50(1) TEU, a Member State may withdraw from the EU ‘in accordance with its own constitutional requirements’.
156
In the words of the CJEU, the subsequent procedures consist of: first, notification to the European Council of the intention to withdraw, second, negotiation and conclusion of an agreement setting out the arrangements for withdrawal, taking into account the future relationship between the state concerned and the [EU] [Article 50(2) TEU] and, third, the actual withdrawal from the [EU] on the date of entry into force of that agreement or failing that, two years after the notification given to the European Council, unless the latter, in agreement with the Member State concerned, unanimously decides to extend that period [Article 50(3) TEU].
157
Article 50 TEU, which it is claimed was inspired by Articles 65–68 VCLT, 158 is considerably more detailed than comparable provisions in the constitutive instruments of other IOs as it is designed to ensure an orderly exit from the EU, a highly complex regional organization. 159 It has been described as being about procedure rather than substance given that the withdrawing state is not required to give reasons. 160 However, this is common practice in relation to the withdrawal clauses of IOs; reasons are required usually in relation to treaties which permit withdrawal in certain circumstances only. 161 Academic opinion was divided on whether withdrawal from the EU is governed exclusively by Article 50 on the basis that it is a self-contained provision or whether recourse could additionally be had to the law of treaties to cover any possible lacunae. 162 But it must be borne in mind that the procedural specifications in Article 50 are not comprehensive. 163
8. The Judgment of the CJEU in Wightman and its repercussions on the question of the right to revoke a notification for withdrawal
On the substantive question the CJEU proceeded first and foremost on the unsurprising basis that it had to be examined in light of the EU Treaties given the special and autonomous nature of EU Law. 164 This is in keeping with the Advocate General’s view that Article 50 TEU constitutes a lex specialis. 165 The CJEU found that a Member State’s decision to invoke the right of withdrawal in Article 50(1) TEU was wholly unilateral in nature and was not one dependent on the other Member States or EU institutions. 166 It therefore considered that Article 50 TEU pursued two objectives, first, it upheld the sovereign right of a Member State to withdraw and, secondly, it laid down the procedure to be followed. 167 These conclusions would suggest that: (a) the sovereign right of a Member State to withdraw had always existed and could have been exercised from the day the EEC became operative; and (b) the procedure to be followed is compulsory and may not be altered by the EU, the withdrawing state or the other states. The CJEU observed that Article 50 TEU did not specifically mention revocation which it neither explicitly prohibited nor allowed. 168 Nevertheless, Article 50(2)TEU states that a Member State which decides to withdraw must notify the European Council of its ‘intention’. Arguably, if withdrawal is not automatic, that is to say it does not take place with immediate effect when the notice is submitted, 169 it could only be understood as an ‘intention’, in the sense that any act which has yet to materialize remains an intention. The CJEU understood this term ‘by its nature, [to be] neither definitive nor irrevocable.’ 170 It followed from the sovereign right of withdrawal that a Member State likewise possessed the right to revoke the notification of withdrawal provided that the withdrawal agreement had not entered into force or the period specified in Article 50(3) TEU had not expired. This was similarly a sovereign decision. 171 The CJEU, relying on a policy argument, considered the context of Article 50. It identified the EU aims and purposes and its shared values. 172 Given the negative impact that withdrawal would have on the rights of all EU citizens, a Member State could not be forced to leave the EU against its will. 173 The protection of the rights of EU citizens could be considered the ‘categorical imperative’. For that reason, it would be contrary to the EU’s democratic values if a Member State’s decision to withdraw was deemed irreversible notwithstanding a change of opinion. 174 The origins of Article 50 in the Constitutional Treaty supported this reading. 175 Accordingly, the CJEU observed that a notification of withdrawal does not inevitably lead to exit; a Member State retains the freedom to revoke that notification should it so wish. 176 The CJEU found support for its view in Article 68 VCLT which it found stated in ‘clear and unconditional terms’ that a notification of withdrawal may be revoked at any time before it takes effect. 177 The practical problem that this could lead to is that it may undermine legal certainty. It is possible to envisage a situation of ‘tactical revocations’ in a polarized country. However, in the Advocate General’s view such a scenario was unlikely since the fact that a revocation must be in accordance with the Member State’s constitutional requirements acts as a deterrent. 178
In finding that withdrawal was a unilateral sovereign right 179 the CJEU has taken a traditional, voluntarist approach which seems to be echoing the views of the German Bundesverfassungsgericht that as ‘Masters of the Treaties’ the Member States remain sovereign states with the autonomous right to terminate membership. 180 The Bundesverfassungsgericht considered that Article 50 TEU underscores the continued existence of state sovereignty, and that membership is predicated on the principle of ‘reversible self-commitment’. 181 The CJEU’s observations refute the Brexiters’ and other Eurosceptics’ misinformed claims about a European ‘super-state’ 182 and misconception and distortion, seemingly wilful, of the legal nature of sovereignty. While membership of the EU does entail a limitation of sovereignty in certain fields, 183 or delegation of sovereign powers if preferred, that reality is based ultimately on the free expression of state consent which lasts for the duration of membership, however long that may be. 184 There is no question of subordination, ‘alienation’, 185 ‘imposition’ 186 or ‘Diktat von Brüssel’ as a state accepts the rules of the IO when it becomes a member, 187 and in the case of the evolution of the EC/EU the UK expressly approved those developments in accordance with its constitutional requirements. 188 As the CJEU observed, a state cannot be forced to join the EU. 189 But the Brexiter viewpoint runs contrary to one of the most fundamental precepts of international law, that the ability to enter into treaties is a feature of state sovereignty. A state’s sovereignty is not violated by constraints thereon as a result of entering into a treaty. As the World Court observed in one of its most celebrated judgments, ‘any convention…places a restriction on the exercise of the sovereign rights of the State’ but it ‘declines to see in the conclusion of any treaty by which a State undertakes to perform or refrain from performing a particular act an abandonment of its sovereignty.’ 190 Can it seriously be suggested that France, Italy and so on are not sovereign states? The Constitutional Court of Poland has made it abundantly clear that the Member States remain sovereign entities. 191 But membership of the EU means that Member States ‘exercise their sovereignty in common’. 192 And as for the supremacy of EU Law which they find so intolerable, 193 the CJEU is simply applying another core principle of international law which provides that in its sphere international law, or in casu EU Law, prevails over domestic law. 194 As a matter of simple logic the system would not function if domestic law took precedence over international law or EU Law. The outcome of Brexiter logic should lead the UK to withdraw from the UN and all its specialized agencies, the WTO, NATO and so on, and denounce all its treaty obligations given that they consider such obligations as constraints incompatible with their absolutist notions of sovereignty. 195 Article 50 TEU preserves state sovereignty and dispels any misconceptions about the EU evolving into a super-state.
The CJEU was surely correct in its conclusion that the right of revocation was similarly a unilateral sovereign right. The CJEU relied on Article 68 VCLT to this end, which it found was emphatic and required no explanation. This is an important finding which, mutatis mutandis, has implications more broadly for the law of international organizations. But as has been seen, Article 68 VCLT has given rise to many questions which remain unanswered. The CJEU’s blunt conclusion suggests that Article 68 has general application and is not limited to the procedures in the preceding provisions as some have argued. No light was shed on the meaning of the term ‘takes effect’. It was particularly disappointing that the CJEU did not elaborate on other possible sources of the right of revocation, in particular whether such a norm existed in customary law which may have been codified in Article 68 VCLT. The Advocate General had cast doubt on the existence of such custom 196 but interestingly the European Council and the European Commission accepted that a Member State could revoke its notification of withdrawal. 197 But given that it is settled case-law that the rules of customary international law are binding on the EU, 198 may it be inferred from the fact that the CJEU had nothing to say about the relationship of Article 68 VCLT with custom that the CJEU did not believe it to be customary law? This question will therefore continue to divide international lawyers. A couple of further points arise. First, the VCLT is strict sensu inapplicable to the EU as neither it nor all the Member States are parties (France and Romania). Nonetheless, as per its past practice that the VCLT, although not binding per se on the EU, applies to the EU, 199 the CJEU once again avoided the question on the precise relationship between the VCLT and the EU Treaties. Secondly, the CJEU appears to have applied the rules of treaty interpretation provided for in Articles 31–32 VCLT, which reflect customary law. 200 Especially significant seems to have been the application of the principle of effectiveness. 201 Thirdly, the CJEU had recourse to travaux préparatoires, of the Constitutional Treaty, 202 which was rather unusual. 203
It follows from the above that the CJEU did not consider Article 50 TEU to be a self-contained provision. It reflects the Advocate General’s opinion that it is ‘not a self-contained provision which exhaustively governs each and every detail of that withdrawal process’ but that to fill the gaps in it recourse could be had to Article 68 VCLT. 204
The CJEU made various references to the procedural requirements, one of which, Article 50(1) TEU, specifies that the withdrawal decision must comply with a Member State’s constitutional requirements. 205 For the Advocate General this was a self-judging provision, a matter solely for the departing state. 206 But what if, as seemed possible, likely even as the Brexit crisis continued, that faced with a parliamentary stalemate, the British Government had proceeded with withdrawal by unconstitutional means, without obtaining parliamentary approval as constitutionally required? 207 This would have given rise to the peculiar situation of a state acting unconstitutionally in order to fulfil its international obligations. 208 The conditions of Article 50(1) TEU would nevertheless not have been met. Would the CJEU arrogate to itself the authority to decide this question if asked? Would the CJEU follow the precedent of the Inter-American Court of Human Rights (IACtHR) that flouting the procedural requirements would invalidate the withdrawal? 209 Had the British courts found the Government to have acted unconstitutionally the general rule is that an international court would be bound by those decisions. 210 But given the supranational nature of the CJEU, it is submitted that the CJEU would determine whether a state had complied with its constitutional relationship with EU Law on the basis that it is the final arbiter on questions of EU Law. 211
The European Council and the European Commission had argued before the CJEU that they accepted the right of a Member State to revoke its notification of withdrawal but that in order to avoid abuse and the circumvention of the requirements of Article 50(2)–(3)TEU that right could be exercised only with the European Council’s consent, the notion of ‘agreed revocation’. 212 The CJEU rejected that contention since it would ‘transform a unilateral sovereign right into a conditional right subject to an approval procedure.’ 213 The CJEU was thereby reinforcing state sovereignty. The survey of withdrawal clauses undertaken above has uncovered no such examples. Such a circumstance, continued the CJEU, would be incompatible with the principle that a Member State could not be forced to leave the EU against its will. 214 The CJEU was thus addressing the possibility that the European institutions could compel a Member State to leave through the simple expedient of refusing to accept the retraction, a form of ‘indirect’ expulsion. 215 Yet the institutions’ concerns about abuse are not without foundation but this problem was left unaddressed by the CJEU. However, the AG was of the view that the ‘protracted and complex’ procedural requirements of Article 50 made abuse unlikely. 216 He also referred to the principle of good faith and further suggested that the general principle of EU Law that abusive practices are prohibited be applied to Article 50. 217 Unfortunately, the Brexit negotiations make evident that good faith is patently lacking on the part of one of the parties. 218 But the CJEU’s statements give rise to further speculation. Since the EU Treaties do not contain an expulsion clause, is the CJEU signalling that a Member State cannot be expelled? The answer seems to be in the affirmative. 219
The CJEU therefore concluded that Article 50 TEU permits a Member State to retract its notification of withdrawal unilaterally, ‘in an unequivocal and unconditional manner’, by a notice to the European Council in writing after the Member State has taken the revocation decision in accordance with its constitutional requirements until such time as withdrawal occurs. The state remains a member of the EU with its status unchanged. 220 The right of revocation is therefore not unfettered but subject to the conditions set out in Article 50 TEU. This logical pronouncement is one that has been highlighted by the IACtHR which has held that procedural requirements must be met. 221 But the CJEU imposed further conditions, some additional to those required by international law or the VCLT: first, the revocation must be unambiguous; secondly, it must be in writing; 222 and thirdly, in accordance with a state’s constitutional requirements. The first condition seems a sensible precaution, whereas the latter provides for democratic legitimacy. Moreover, it should inhibit a state from submitting specious revocations. 223
The CJEU also made the important point that a departing Member State remains bound by all its obligations under EU Law until the period of notice expires in accordance with Article 50(3) TEU. 224 Naturally, the EU would itself be bound by its obligations towards the withdrawing State. This reminder is not unexpected; many withdrawal clauses contain such a condition 225 and reflects the principle pacta sunt servanda.
9. Conclusion
The CJEU’s judgment raises many points of interest; it answers several questions regarding Article 50 TEU and its relationship to international law but leaves others tantalizingly ambiguous. Where the constituent instruments of IOs permit withdrawal, a Member State is entitled by virtue of its sovereignty to invoke that right subject to satisfying any associated conditions. At the same time, the CJEU has made it clear that it is permissible under EU and international law for the notice of withdrawal to be retracted. Its finding implies that Article 68 VCLT has general application. It is absurd to suggest as a matter of law and logic that such notice is irreversible. Unless, since Article 50 is subject to domestic constitutional procedures, the intention to withdraw is made irreversible; this could be decided during a referendum or by the legislature or by whatever other domestic means. This could only be the case if a provision to that effect was found in the withdrawal clause, and none seems to exist. And for the evident good reason that states, or their people, retain the discretion to change their minds. Otherwise, the ridiculous situation would arise of a state immediately applying for readmission upon withdrawal. 226 The CJEU has brought clarity to Article 50 TEU and filled in some of the existing gaps. However, the question whether Article 68 VCLT reflects a rule of customary international law remains unanswered. And there is more than a hint of irony about this judgment. In its pronouncements on sovereignty the CJEU has disproved the Brexiters’ disingenuous claims on the matter.
