Models of Irish Unification Processes

WORKING PAPER – NOT FOR CITATION Models of Irish Unification Processes Oran Doyle* and David Kenny** I Introduction The purpose of this paper is to identify and evaluate pathways to Irish unification that would be (a) constitutionally permissible in Ireland—i.e. the State currently governed by the Constitution of Ireland 1937—(b) consistent with the terms of the Belfast / Good Friday Agreement 1998 (GFA), and (c) consistent with the Northern Ireland Act 1998. Starting with the southern Irish perspective, 1 we argue that bare unification does not require constitutional amendment, and could be achieved by a plebiscite or (probably) legislation. There are, however, many related changes frequently suggested in the context of unification that would require either a constitutional amendment or constitutional replacement in the South. Without evaluating the merits of any of these changes, we make an assumption—which we believe to be reasonable although contestable—that unification would involve at least some changes of this type. It follows that unification would involve a referendum, either to amend or replace the Constitution of Ireland 1937. We argue that the GFA requires that this referendum be concurrent with a referendum in Northern Ireland, although there is some ambiguity as to the meaning of ‘concurrence’. If both referendums favour unification, there is then an obligation on the Irish and British Governments to introduce and support in their respective parliaments legislation to give effect to that wish. Irish unification must therefore involve, if it is to meet the three criteria identified above, concurrent referendums North and South followed by legislation approved by Westminster and the Oireachtas (Irish Parliament). We go further, however, and specify three desiderata against which unification processes can be assessed. First, that those voting in referendums should be able to have a reasonably clear idea of what the implications of their vote. Second, that the processes should reduce the risk of the UK and Ireland diverging over the final constitutional status of Northern Ireland. Third, that the processes should allow for a wide range of consultation and negotiation over the * Professor, Trinity College Dublin School of Law; oran.doyle@tcd.ie ** Assistant Professor, Trinity College Dublin School of Law; david.kenny@tcd.ie. We are grateful to Brendan O’Leary for comments on an earlier draft. 1 We attempt in this paper to use the terminology for various entities that is used in the Belfast/Good Friday Agreement, i.e. “Northern Ireland”, “Ireland”, “united Ireland”, “North and South” etc. However, at times, to avoid ambiguity as to the entity under discussion, it is necessary to use terms such as “southern Ireland” and “the South”. 1 WORKING PAPER – NOT FOR CITATION constitutional and political structure of a unified Ireland, including extensive attempts to engage with the views of northern unionists. We identify several models that fit the legal parameters. We suggest, however, that the model that best meets the desiderata is one in which the form of a united Ireland is settled through negotiations prior to the referendums. The required legislation in Westminster and the Oireachtas should then follow as a matter of course. Section II provides a general account of mechanisms of legal change in Ireland. Section III explores how bare unification—that is, unification without any other changes to the structure of the Irish state—might be constitutionally achieved. Section IV considers, beyond this, a range of changes to the structure of the Irish state, dividing them into those that could be achieved by legislation and those that would require constitutional amendment or replacement. Section V analyses the GFA to establish whether public international law complements or qualifies the Irish constitutional position in any way. Section VI then undertakes the same exercise with respect to the Northern Ireland Act. This allows the identification of legal parameters for a unification process that would cohere with Irish, UK, and pubic international law. Section VII restates those parameters, before specifying and defending the three desiderata listed above. Section VIII identifies models that comply with the legal parameters and evaluates them in light of the desiderata. Before we proceed with the analysis, we want to emphasise that this paper does not advocate for a united Ireland, still less for any particular version of a united Ireland. We have necessarily engaged with different substantive models of Irish unification, i.e. different constitutional arrangements for a united Ireland. But nothing in this paper should be interpreted as suggesting a preference for a united Ireland or for any particular form of constitutional arrangement (e.g. the maintenance in whatever form of the Irish Constitution of 1937). II Mechanisms of legal change II.A Legislation Unification would require legal change. This change could occur by legislation, provided such legislation is consistent with the terms of the Constitution. If a particular change required legislation that was inconsistent with the terms of the Constitution, that change would require constitutional amendment or replacement. 2 WORKING PAPER – NOT FOR CITATION II.B Constitutional amendment A constitutional amendment must be introduced in the Dáil and receive a majority in both Houses of the Oireachtas before being put to a referendum. A simple majority of those voting on the day is necessary to approve the constitutional amendment. 2 The Constitution does not impose any substantive limits on the power of constitutional amendment. Article 46 provides that any provision of the Constitution may be amended by way of variation, addition, or repeal. Moreover, the Irish courts have not followed the global trend of implying limits on the power of constitutional amendment. 3 A constitutional amendment must make a specific textual change to the Constitution. It is possible that the power of amendment does not extend to the Preamble. The textual argument to this effect is that the Preamble is not a ‘provision’ of the Constitution. The theoretical argument is that the Preamble was a historically situated provision that recorded the enactment of the Constitution; as such, it cannot be retrospectively altered. Nevertheless, the balance of expert opinion is that the Preamble to the Irish Constitution in particular can be subject to constitutional amendment.4 In this paper, we proceed on the assumption that there is no substantive limit to the power of the people to amend any provision of the Constitution, including the Preamble. II.C Constitutional replacement The third mechanism of legal change is to replace the Constitution. The Constitution does not envisage its own replacement. Accordingly, any replacement of the 1937 Constitution would contravene the 1937 Constitution and be, in those terms, unconstitutional. That said, constitutions are typically self-authorising and breach the terms of whatever constitution they replaced. Article 62 of the 1937 Constitution stipulated the method for the enactment of the 1937 Constitution: approval by the people signified by a majority of 2 For general analysis of the amendment process, see Oran Doyle, The Constitution of Ireland: A Contextual Analysis (Oxford: Hart, 2018), chapter 10. 3 See Re Article 26 and the Regulation of Information (Services Outside the State for Termination of Pregnancies) Bill 1995 and Riordan v An Taoiseach (No 2) [1999] 4 IR 343; Aileen Kavanagh, ‘Unconstitutional Constitutional Amendments from Irish Free State to Irish Republic’ in Eoin Carolan (ed), The Constitution of Ireland: Perspectives and Prospects (Dublin: Bloomsbury Professional, 2012) 331; Gerard Hogan, Gerry Whyte, David Kenny and Rachael Walsh, Kelly: the Irish Constitution (Bloomsbury Professional, 2018) [7.1.20]-[7.1.26] (hereinafter Kelly). 4 See the Report of the Constitution Review Group (1996) 3; the Report of the Attorney General’s Committee on the Constitution (August 1968) 3–4. The editors of Kelly suggest that even if there were some barrier, Article 46 itself could be changed to allow such an amendment if needed. Kelly (n 3) [2.1.48]. 3 WORKING PAPER – NOT FOR CITATION the votes cast at a plebiscite thereon held in accordance with law.5 This was an act of selfauthorisation since Article 62 had no legal force prior to the enactment of the Constitution. Moreover, this method of enactment was different from the mechanism for constitutional amendment stipulated by the Constitution of the Irish Free State. Nevertheless, the 1937 Constitution seamlessly and without legal dispute replaced the 1922 Constitution. One of us has argued elsewhere that this occurred because of a general official and popular acceptance that the electors of the 26 counties of Ireland were a legitimate constitutional author for that geographical space, and possibly also a legitimate proxy as a constitutional author for the entire island of Ireland.6 In this paper, we proceed on the assumption that it would be legitimate for the electorate of the 26 counties of Ireland to approve by plebiscite a new Constitution for that geographical space. Correspondingly, the electorate of the entire island of Ireland could approve by plebiscite a new Constitution for that geographical space. Finally, the electorate of the 26 counties of Ireland could approve by plebiscite a new Constitution for the entire island of Ireland, provided approval to that new Constitution was also given by the electorate of the 6 counties of Northern Ireland. All of these avenues would be constitutionally permissible even though they, at least arguably, breach the terms of the current Constitution. II.D Plebiscites Although not generally a mechanism of legal change, it is permissible under the current Constitution for legislation to provide for non-binding plebiscites.7 The Local Government Act 2019 provided for plebiscites to be held on whether there should be a directly elected mayor in the city of Cork, the city and county of Limerick, and the city and county of Waterford. If a majority of votes cast at one of these plebiscites was in favour of the proposal, the Minister for the Environment was under an obligation within two years to prepare and submit to each House of the Oireachtas a report containing proposals for the 5 The law in question was the Plebiscite (Draft Constitution) Act 1937, a statute that was clearly a breach of the Irish Free State Constitution. 6 See Oran Doyle, ‘Populist Constitutionalism and Constituent Power’ (2019) 20(2) German Law Journal 161; Oran Doyle, ‘Constitutional Transitions, Abusive Constitutionalism and Conventional Constraint’ (2017) 35 National Journal of Constitutional Law 67. 7 There is no clear-cut distinction between referendums and plebiscites. Different jurisdictions will distinguish between the two terms in different ways. In this paper, we follow standard Irish legal usage where a ‘referendum’ is the constitutionally prescribed process for constitutional amendment, and one other unusual legislative route that has never been applied in practice. But nothing conceptually turns on our use of ‘referendum’ or ‘plebiscite’. 4 WORKING PAPER – NOT FOR CITATION enactment of a law providing for a directly elected mayor in the area concerned. In this paper, we proceed on the assumption that (though it is not currently provided for) legislation could enable a consultative plebiscite in the 26 counties of Ireland that would provide a political mandate to the Government to explore and propose legal changes. Those legal changes, however, would have to be implemented in one of the three ways discussed above: legislation, constitutional amendment, constitutional replacement. Ordinarily, it would not be permissible for the Oireachtas to provide for a legally binding plebiscite.8 Article 15.2 provides that the Oireachtas is the sole legislative authority for the state. The courts have interpreted this as prohibiting the Oireachtas from delegating the legislative power to deal with matters of principle and policy, as distinct from matters of detail. 9 However, we will suggest below that Article 3 may, exceptionally, provide a constitutional basis for a binding plebiscite on Irish unification. II.E Dualism and the role of the Court Before we proceed to our analysis of the relevant Irish constitutional provisions, two other preliminary points should be made. Ireland, like the United Kingdom, is a dualist state. This means that public international law does not have legal status in Ireland or the UK unless the Oireachtas or Westminster respectively legislates to that effect. International agreements, such as the British-Irish Agreement (part of the GFA), have force only insofar as they are implemented in domestic law. Article 34 of the Constitution grants to the courts the power to review the constitutionality of all legislation of the Oireachtas. There is some judicial support for a doctrine of non-justiciability of political matters;10 moreover, the courts have been markedly deferential to Government decision-making in respect of Northern Ireland. 11 Nevertheless, the purpose of this paper is not to anticipate what approach the courts might take to any constitutional judicial review of Irish unification but rather to establish what is the correct constitutional position. Of course, this involves considerable reference to the previous judgments of the courts as authoritative interpreters of the Constitution. But we shall not speculate whether the courts might under-enforce the Constitution in this politically fraught context. The Government and 8 S 2 of The Water Services Act 2014 is an exception to this, but this provision is almost certainly unconstitutional in purporting to limit the legislative power in the absence of a plebiscite. 9 For discussion, see Oran Doyle and Tom Hickey, Constitutional Law: Text, Cases and Materials (Dublin: Clarus Press, 2019), chapter 8. See Kelly (n 3) [4.2.23]-[4.2.51]. 10 See Kelly (n 3) [3.2.117]-[3.2.121]. 11 Boland v An Taoiseach [1974] IR 338; McGimpsey v Ireland [1990] 1 IR 110. 5 WORKING PAPER – NOT FOR CITATION Oireachtas must seek to act constitutionally, irrespective of the prospects of judicial enforcement. III. Bare unification III.A Constitutional parameters In this section, we will consider what form of legal change would be necessary, as a matter of Irish law, to allow for bare unification, i.e. the extension of the territory of the current Irish state to include what is currently Northern Ireland, without any other changes. As originally enacted, Article 2 provided that the national territory of Ireland consisted of the whole island of Ireland, its islands and the territorial seas. Article 3 provided as follows: Pending the re-integration of the national territory, and without prejudice to the right of the Parliament and Government established by this Constitution to exercise jurisdiction over the whole of that territory, the laws enacted by that Parliament shall have the like area and extent of application as the laws of Saorstát Éireann and the like extra-territorial effect. A constitutional amendment would not have been required to achieve bare unification under these provisions. The question is whether the revised Articles 2 and 3, amended subsequent to and pursuant to the British-Irish Agreement in 1998, have changed this position. Articles 2 and 3 now provide: ARTICLE 2 It is the entitlement and birthright of every person born in the island of Ireland, which includes its islands and seas, to be part of the Irish Nation. That is also the entitlement of all persons otherwise qualified in accordance with law to be citizens of Ireland. Furthermore, the Irish nation cherishes its special affinity with people of Irish ancestry living abroad who share its cultural identity and heritage. ARTICLE 3 1 It is the firm will of the Irish Nation, in harmony and friendship, to unite all the people who share the territory of the island of Ireland, in all the diversity of their identities and traditions, recognising that a united Ireland shall be brought about only by peaceful means with the consent of a majority of the people, 6 WORKING PAPER – NOT FOR CITATION democratically expressed, in both jurisdictions in the island. Until then, the laws enacted by the Parliament established by this Constitution shall have the like area and extent of application as the laws enacted by the Parliament that existed immediately before the coming into operation of this Constitution. 2 Institutions with executive powers and functions that are shared between those jurisdictions may be established by their respective responsible authorities for stated purposes and may exercise powers and functions in respect of all or any part of the island. Article 3 establishes that the Irish Nation is committed to unification. Unification is the constitutional telos, borne out by the way in which the second sentence of Article 3.1 commences ‘until then’ rather than ‘unless or until then’. Very importantly, unification can only occur through peaceful means and with the consent of a majority of people, democratically expressed, in both jurisdictions in the island. But Article 3 does not specify how the consent of the people in the 26 counties of Ireland is to be democratically expressed. III.B The democratic expression of consent In our view, it would probably be constitutionally permissible—provided that the consent of a majority of people in Northern Ireland was democratically expressed—for the Oireachtas to legislate so as to extend the national territory to include the six counties of Northern Ireland. The passage of this Act would represent the ‘consent’ of the people in the South.12 However, there are credible arguments that can be made against this: that this is not sufficiently ‘majoritarian’ or ‘popular’ to satisfy Article 3, and that the term ‘the consent of a majority of the people’ requires more popular participation than a legislative consent mechanism would provide; or that Article 3 implies a similar means of democratic consent in Ireland and Northern Ireland, and so something similar to the referendum required by the British-Irish Agreement for Northern Ireland must be done on the other side of the border.13 In short, even though we consider that bare unification by legislation 12 One argument that might be made is that ‘consent’ to the British-Irish agreement was authorised by a constitutional change to Article 29 (‘The State may consent to be bound by the British-Irish Agreement’) but actually executed by statutory measures; see the British-Irish Agreement Act 1999. 13 Again, the British-Irish Agreement is constitutionally authorised by Article 29. 7 WORKING PAPER – NOT FOR CITATION is probably constitutionally permissible, the constitutional doubts over this position and other factors may militate against this route.14 A binding plebiscite in the South would avoid many of these counter-arguments. Ordinarily, binding plebiscites would represent an unconstitutional delegation of legislative power. However, if we are correct in our reading of Article 3 as an existing constitutional commitment to unification—subject of course to the dual consent requirements—it would not be unconstitutional for the Oireachtas to legislate to provide for a binding plebiscite through which the consent of a majority of the people in Ireland could be democratically expressed. It would be possible to provide in law for a consultative plebiscite to attain consent in Ireland, and legislation could follow this. However, as the legislation would provide the formal consent, it would be open to the same objections identified above. It would be possible to achieve bare unification by constitutional amendment, perhaps inserting something like the original wording of Article 2 to extend/clarify the extent of the national territory. It would be possible to achieve bare unification by enacting a new constitution through a plebiscite. Any legal change mechanism that uses a popular vote—whether a plebiscite or a referendum—may raise issues about the symmetry of the electorates between Ireland and Northern Ireland. This is particularly the case in respect of constitutional amendment referendums, where the electorate is limited to Irish citizens. Any change to this electorate would require a prior constitutional amendment. In contrast, the Oireachtas would have some latitude to prescribe the electorate for a binding plebiscite under Article 3. 15 If a new constitution were to be enacted by plebiscite, it could determine its own electorate. III.C Consequences of the democratic expression of consent 14 As Girvan J put it, in the UK legal context, in In re McCord [2018] NIQB 106, [5]: “In effect if not de jure there would have to be an agreement between the UK and the Republic to have parallel polls in each jurisdiction.” 15 However, there might still be some constitutional constraints. In particular, re Article 26 and the Electoral (Amendment) Bill 1983 [1984] IR 268 casts some doubt on whether British citizens resident in Ireland could be made eligible to vote. On the other hand, the opposite conclusion might be appropriate in the particular context of unification, and a parallel consent procedure in Northern Ireland in which Irish citizens would presumably be entitled to vote. 8 WORKING PAPER – NOT FOR CITATION If southern Ireland adopts a measure that approves unification, what would be the precise legal effect of such a measure? If the measure were adopted as part of a number of constitutional amendments or a constitutional replacement, those amendments or replacement could specify the legal effect of the measure. The only question that can sensibly be asked at this point, therefore, is what effect such a measure would have in the light of Article 3 as currently worded. In other words, if a majority of people expressed consent to a united Ireland—whether through legislation, a plebiscite, or constitutional amendment—without doing anything else, what legal effect would Article 3 accord to such a decision? The first and most obvious point is that such a decision would have no legal effect unless the consent of a majority was also democratically expressed in Northern Ireland. But if consent were democratically expressed in both jurisdictions, what would be the legal significance of this under Irish constitutional law? Are the two decisions sufficient in and of themselves to bring about unification or must further steps be taken? The argument that the two decisions are sufficient to bring about unification rests on the mandatory language, ‘a united Ireland shall be brought about’, reinforced by the implication from the second sentence that the laws enacted by the Oireachtas will then have effect on the entire island. Against this reading, however, the juxtaposition of ‘only by peaceful means’ and ‘with the consent of the majority of the people…’ suggests that the consent of the majority of the people is a necessary but perhaps not sufficient condition. Article 3 does not provide that a united Ireland occurs by the consent of the majority; rather unification occurs with the consent of the majority. The language of Article 3 therefore appears to leave open two possibilities: the referendum constitutes both ‘the peaceful means’ and the ‘consent of the majority of the people’ sufficient to bring about unification, assuming Northern Ireland has also voted in favour. Alternatively, some further official step—such as legislation or a Government declaration—is necessary to bring about unification, subsequent to the consent of the majority having been democratically expressed in both jurisdictions. The goal of legal certainty favours the latter approach. Under any analysis, the significance of the decision in the South depends on the result of the vote in the North. Rather than allow any potential for competing interpretations of that vote, it would be preferable to require a formal act to implement the result of the two decisions. As we shall later see in section V, this latter reading is more consistent with the GFA which implies that legislation is required. 9 WORKING PAPER – NOT FOR CITATION Although the legal effect of the decision in the South is dependent on the decision in the North, it is not—as a matter of Irish constitutional law—dependent on any decisions that may subsequently be taken at Westminster. We shall see in sections V and VI that both the GFA and the Northern Ireland Act contemplate legislation at Westminster (as well as the Oireachtas) to give effect to the unification decisions. From the perspective of Irish constitutional law, however, activities at Westminster are irrelevant. Once the democratic consent to unification is concurrently expressed North and South, either unification has occurred (the first reading above); or the Oireachtas or Government in the South has the legal competence to effect unification (the second reading above). Our tentative view, therefore, is that unification requires not only affirmative votes in the North and South but also some subsequent official act in the South to give effect to those referendum votes. It does not follow from this, however, that it would be constitutionally permissible for relevant actors in the South to delay indefinitely that official action. As noted above, the language of Article 3 is mandatory: a united Ireland shall be brought about. Moreover, in several contexts, the Irish courts have asserted that the Constitution reflects a doctrine of popular sovereignty; the Constitution is subject to the people.16 (We shall suggest at the end of section IV that the approval for unification in the South is almost certainly going to come in the form of a referendum.) It would therefore offend some of the most basic constitutional presuppositions for either the Government or the Oireachtas to ignore the will of the people on unification. There might be reasonable grounds not to give effect to unification straightaway; this would largely depend on the extent to which the terms of unification had been agreed in advance of the approvals, North and South. But the Oireachtas and Government do not have a veto power; and after a time—exceptionally hard to specify—delay would become tantamount to veto. It is likely that the courts would show considerable deference to the judgment of the Government and Oireachtas as to when to give effect to unification, particularly if the delay were due to difficulties in negotiations with the British Government. It is almost inconceivable that the courts would direct the Oireachtas to legislate to give effect to unification.17 But in 16 See e.g. “The people have a sovereign right to grant or withhold approval to an amendment to the Constitution.” Riordan v An Taoiseach (No 2) [1999] 4 IR 343, 354; “The Constitution reflected a fundamental truth as to the source of the sovereignty of the State, namely the People.” Pringle v Government of Ireland [2012] IESC 47, [2013] 3 IR 1 at [318], 109 per O’Donnell J. “The sole temporal source of power acknowledged by the Constitution is the people themselves. In that sense the separation of powers in the Irish Constitution while tripartite, has a fourth important, indeed overarching, component.” Jordan v Minister for Children and Youth Affairs [2015] IESC 33 at [152], [2015] 4 IR 232 at 271. See generally Kelly (n 3) [3.2.91]-[3.2.103]; Hickey, ‘Popular Sovereignty in Irish Constitutional Law’ (2017) 40(2) DULJ 147. 17 The legislative power is said in Article 15.2.1 to be the solely and exclusively that of the Oireachtas; the most basic instantiation of this is that the courts, while empowered by Article 15 to invalidate legislation, 10 WORKING PAPER – NOT FOR CITATION principle there is a constitutional obligation on the Government and/or Oireachtas to give effect to votes in favour of unification, North and South. If bare unification occurred—with or without the approval of Westminster—the territory of Ireland would just expand to include Northern Ireland. Irish and British citizens resident in what is currently Northern Ireland would acquire a right to vote in elections to the Dáil, but legislation would be required to establish new Dáil constituencies. Beyond that, however, the territorial expansion would be extremely disruptive unless accompanied by other legal measures. To take one simple example: the legal system of Ireland would replace the legal system in Northern Ireland. What would this mean for legal relationships entered into in Northern Ireland prior to unification, such as contracts and marriages, property ownership, the validity of official acts, the prosecution of criminal offences? What would be the status of public office-holders in Northern Ireland: administrative officials, judges, police officers? At least some legal measures would be required to manage unification in a non-disruptive way. In the next section, we shall identify some such legal changes. IV Changes accompanying unification In this section, we explore changes that might accompany unification. The changes that we discuss have featured in public and academic debate. However, we again emphasise that we do not express any position on whether any of these changes would be desirable; our sole purpose is to explore what form of legal mechanism would be required, as a matter of Irish constitutional law, to allow for these changes if desired. We divide these changes into two types: those that can be enacted by the Oireachtas; and those that would require a constitutional amendment or constitutional replacement. IV.A Changes achievable by legislation It would be constitutionally permissible for Ireland to rejoin the Commonwealth of Nations. Ireland continued to be a member of the Commonwealth for 12 years after the enactment of the 1937 Constitution. Article 29.4.2 rather elliptically referred to ‘any group or league of nations with which the State is or becomes associated for the purpose of will not write or rewrite legislation, or order that legislation be written or amended the Oireachtas. See Kelly (n 3) [3.2.139]-[3.2.142]; [4.2.71]-[4.2.89]. 11 WORKING PAPER – NOT FOR CITATION international co-operation in matters of common concern’. The context makes clear that ‘group of nations’ referred at the time to the Commonwealth. It would be constitutionally permissible for the Oireachtas to extend automatic Irish citizenship to people born or resident in Northern Ireland. Article 9.1.2 provides that the future acquisition and loss of Irish nationality and citizenship shall be determined by law.18 It would be constitutionally permissible for the Oireachtas to extend the franchise for Dáil elections to all British and Irish citizens living in Northern Ireland, respecting the current requirement that there be 1 TD for every 20,000-30,000 people. Article 16.1. 2(ii)—inserted by referendum in 1983—allows the Oireachtas to extend the franchise to other persons in the State, apart from citizens. The Electoral (Amendment) Act 1985 extended the franchise to British citizens ordinarily resident in Ireland. As unification would imply an extension of the State to include the six counties of Northern Ireland, British citizens ordinarily resident in Northern Ireland would automatically become entitled to vote in Dáil elections. Legislation would be required to establish new Dáil constituencies. It would be constitutional for the Oireachtas to devolve legislative power to Northern Ireland and to other regions. Article 15.2.2 allows for provision to be made by law for the creation or recognition of subordinate legislatures and for the powers and functions of those legislatures. However, it is not clear if there might be some limitation on this: there might be a limit to the powers that could be devolved or delegated.19 It would probably be constitutional for the Oireachtas to give legal status to the laws that formerly applied in Northern Ireland, although such laws would have to be consistent with the Constitution. In Leontjava and Chang v DPP, the Supreme Court recognised a wide discretion as to the form of legislation. In that case, the Court upheld the constitutionality of a statutory provision that purported to confer full statutory status on a series of ministerial orders. 20 The same principle would seem to apply in the context of incorporation by reference of the laws prevailing in Northern Ireland. However, it would 18 This would, however, raise the same question (in reverse) that has been ventilated in the da Souza litigation: is automatic citizenship in Northern Ireland consistent with the right of people living there to identify as Irish or British or both. Secretary of State for the Home Department v Parker De Souza EA/06667/2016, 14th October 2019. 19 There is a general limit on the delegation of powers to other actors that is seen to flow from Article 15.2’s provision that the sole and exclusive legislative power belongs to the Oireachtas (see Kelly (n Error! Bookmark not defined.), [4.2.23]-[4.2.51]) but the provision for subordinate legislatures is an express qualification to this provision. There may be some issues with delegation of financial functions given the Constitution’s requirements for money bills (see Articles 20-23). 20 [2004] 1 IR 591. For discussion, see Doyle and Hickey (n 9), [8-45]-[8-49]; Kelly (n 3) [4.5.35]-[4.5.36]. 12 WORKING PAPER – NOT FOR CITATION be a far broader application of the principle and it might not be prudent to allow the entire corpus of law in Northern Ireland rest on such a provision. It would be constitutional for the Oireachtas to provide for rights protection on the same model as the UK Human Rights Act 1998. A broadly similar approach was taken in the European Convention on Human Rights Act 2003. Such an Act, however, would have to be interpreted in a manner consistent with the Constitution. This might in some circumstances lead to a level of rights protection lower than currently pertains under the UK Human Rights Act.21 IV.B Changes requiring constitutional amendment or replacement22 In the event of unification, constitutional amendment or replacement would be required if it were desired to amend any of the following provisions: § Irish people defined with reference to those whom they struggled against (Preamble). § Implicitly Catholic or Catholic-influenced provisions in the Constitution (Preamble, Articles 41-45). § Irish as national language and first official language (A8). § Irish language titles for Taoiseach (Article 13), Tánaiste (Article 28), Oireachtas, Dáil and Seanad (Article 15). § National flag as tricolour of green, white and orange (Article 7). § Give status (beyond that associated with being Head of the British Commonwealth, if the Commonwealth were joined) to British Monarch in the Constitution (Article 29).23 § Granting full political rights (vote in referendums and presidential elections, in addition to parliamentary, European, and local elections) to non-citizens (re Article 26 and the Electoral (Amendment) Bill 1983 [1984] IR 268).24 § Any change to electoral system for Dáil (Article 16). § Consociational arrangements in the Dáil (Article 15.11). 21 See for instance McD v L [2010] 2 IR 199. For discussion, see Kelly (n Error! Bookmark not defined.), [5.3.141][5.3.143]. 22 For a consideration of some of these issues, see David Kenny, ‘The Irish Constitution, a United Ireland, and the Ship of Theseus: Radical Constitutional Change as Constitutional Replacement’ Workign paper, April 30th 2019, available at <https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3399054>. 23 The British Monarch is not automatically the Head of the British Commonwealth. Queen Elizabeth II currently holds that role. Prince Charles was appointed her designated successor at the 2018 Commonwealth Heads of Government meeting. 24 This would facilitate full political participation in the new constitutional order by residents of Northern Ireland who wished to retain British citizenship and not acquire Irish citizenship. 13 WORKING PAPER – NOT FOR CITATION § Consociational arrangements for election of national Government (Article 13.1 and Article 13.2). § Adaptation of Seanad as a territorial upper house (Article 18). § Expansion or alteration of constitutionally-protected rights. § Alterations to the amendment process, whether to change the thresholds for amendment or entrench provisions against change (Article 46). § Any separate court system (even interim or transitional) from the court system currently authorised (Article 34). § Fully federal or confederal model government arrangements going beyond devolution/creation of ‘subordinate legislatures’ in Article 15. The changes identified in the bullet points above would all require a constitutional amendment. A much more difficult question arises in relation to the devolution of executive powers to Northern Ireland. As we noted in section IV.A, Article 15.2 authorises the establishment of subordinate legislatures; no equivalent provision authorises either the establishment of subordinate executives or the devolution of executive power to be exercised on a regional basis. Article 3.2 provides for institutions with shared executive powers between North and South, but does not seem to provide for delegation of executive power.25 Article 28A, which provides some recognition for local government, does not provide a constitutional basis for a devolved regional government with broad executive powers. 26 The starting point for this analysis is that the Government cannot cede its own executive power.27 Moreover, there is a general constitutional principle that one organ of government cannot take over the powers of another organ of government.28 That said, there is a well-established practice of legislation regulating and controlling the way in which the Government exercises executive power.29 The courts have accepted that the Oireachtas may legislate to control areas of implicit executive power, most notably the power to control immigration.30 But the establishment of devolved government and 25 “Institutions with executive powers and functions that are shared between those jurisdictions may be established by their respective responsible authorities for stated purposes and may exercise powers and functions in respect of all or any part of the island.” 26 First, devolved regional government was not within the contemplation of the amendment when passed in 1999. Second, Article 28A. 2 provides that the powers and functions of local authorities shall be subject to the provisions of the Constitution. 27 Crotty v An Taoiseach [1987] IR 713. 28 Buckley v Attorney General [1950] IR 67; TD v Minister for Education [2001] 4 IR 247. 29 For instance, since the foundation of the State, legislation has prescribed criteria for eligibility for judicial appointment, notwithstanding that the nomination of judges is a power vested in the Government. 30 See Laurentiu v Minister for Justice [1999] 4 IR 26. For general discussion, see Conor Casey, ‘Under-explored Corners: Inherent Executive Power in the Irish Constitutional Order’ (2017) 40(1) Dublin University Law Journal 1. 14 WORKING PAPER – NOT FOR CITATION the assignment of executive powers to such government would be a much bigger step. Particular difficulties would arise in respect of budgetary powers, where the Constitution stipulates a close nexus between the Government and the Dáil. In our view, although there are some counter-arguments, it would not be constitutionally permissible for the Oireachtas to establish devolved government. The continuation of the current devolved institutions in Northern Ireland would probably require a constitutional amendment. IV.C Implications of the British-Irish Agreement For the sake of completeness, we consider an argument that the constitutional references to the British-Irish Agreement could provide a constitutional basis for a wider range of changes than suggested in the previous two sections. Article 29.7.2° provides as follows: Any institution established by or under the Agreement may exercise the powers and functions thereby conferred on it in respect of all or any part of the island of Ireland notwithstanding any other provision of this Constitution conferring a like power or function on any person or any organ of State appointed under or created or established by or under this Constitution. Any power or function conferred on such an institution in relation to the settlement or resolution of disputes or controversies may be in addition to or in substitution for any like power or function conferred by this Constitution on any such person or organ of State as aforesaid. This provision provides a constitutional immunity to institutions established by or under the British-Irish Agreement insofar as they exercise powers or functions conferred thereby. This provided a constitutional basis for all-Ireland bodies envisaged in the GFA, such as Waterways Ireland that manages rivers and canals across the island of Ireland. For present purposes, the question is whether it could provide a constitutional basis for institutions established in the event of unification. If unification proceeded entirely on the Belfast-Irish Agreement template and required new institutions, it could be said that those institutions were established ‘under’ the Agreement. But are they exercising powers and functions conferred ‘thereby’? This is questionable. Also, general principles of legal certainty would militate against reading the provision as a constitutional authorisation for the establishment of new institutions with significant constitutional powers to which no particular thought had been given at the time the provision was inserted into the 15 WORKING PAPER – NOT FOR CITATION Constitution.31 For these reasons, we do not consider that Article 29.7.2° displaces the constitutional analysis in sections IV.A and IV.B. IV.D The likely mode of unification We concluded in section III that unification could be achieved by legislation (probably); a plebiscite; a referendum to amend the 1937 Constitution; or a referendum to replace the 1937 Constitution. In this section, drawing on our knowledge of the public debate, we have explored changes that might accompany unification. In our view, it is exceedingly unlikely that unification would be contemplated without any of the changes identified in section IV.C. For the avoidance of any doubt, we are not advocating for any—still less, for all—of the changes identified in section IV.C. If we are correct in our prognosis, however, it follows that a referendum—whether to amend or replace the 1937 Constitution—would be constitutionally required. Such an approach would have the additional benefit of removing areas of constitutional uncertainty identified in section III and section IV.B. In the rest of this paper, we will proceed on the assumption that a referendum would be required in Ireland to give effect to unification. This assumption is contestable but, in our view, reasonable. The work of the next two sections is to identify the equivalent requirements of the GFA and the Northern Ireland Act 1998, before designing models that integrate the requirements in the three legal orders. The assumption significantly reduces the number of permissible models; it thereby helps to focus academic and public debate on a relatively small number of plausible approaches. We have suggested that the power to amend the Irish Constitution is unlimited. It follows that there is no legal reason to prefer constitutional replacement to constitutional amendment. The political arguments that favour replacement, we suggest, involve the extensive number of the likely changes and their fundamental nature: they would effectively create a new state with a new people and new territory. The political arguments that favour amendment, we suggest, are that the changes are only giving effect to what has been contemplated since 1937 and 1998. Moreover, the changes would only be directly authorised by the electorate in the South. It would be preferable for a new Constitution to be a joint act of the people of the entire island of Ireland. For the rest of 31 One might analogise the famous case of Crotty v An Taoiseach [1987] IR 713 where the Supreme Court held that the constitutional authorisation to do whatever was necessitated for European Union membership did not extend to ratification of the Single European Act, as that instrument ceded new governmental powers to the European Union. See Kelly (n 1) at [5.3.51]-[5.3.55]. 16 WORKING PAPER – NOT FOR CITATION the paper, we prescind from this debate. When we speak of a referendum in the South, we are disinterested as to whether this is a referendum to enact a new Constitution or a referendum to amend the existing Constitution. V Requirements of the GFA32 V.A GFA text: three implications In this section, we explore the requirements of the GFA that pertain to Irish unification. The GFA also required some provisions to be included in the Northern Ireland Act 1998. For the most part, we defer analysis of those provisions to the next section but some must be considered here. The ‘Constitutional Issues’ section of the GFA includes the following provisions: it is for the people of the island of Ireland alone, by agreement between the two parts respectively and without external impediment, to exercise their right of selfdetermination on the basis of consent, freely and concurrently given, North and South, to bring about a united Ireland, if that is their wish, accepting that this right must be achieved and exercised with and subject to the agreement and consent of a majority of the people of Northern Ireland…. if, in the future, the people of the island of Ireland exercise their right of selfdetermination on the basis set out … above to bring about a united Ireland, it will be a binding obligation on both Governments to introduce and support in their respective Parliaments legislation to give effect to that wish. Annex A to this section of the Agreement also includes provisions to be included in British legislation. The following is relevant to the understanding of the core-GFA requirements: 1. (1) It is hereby declared that Northern Ireland in its entirety remains part of the United Kingdom and shall not cease to be so without the consent of a majority of the people of Northern Ireland voting in a poll held for the purposes of this section in accordance with Schedule 1. We have italicised the words that identify three key requirements. First, the consent to unification in Northern Ireland requires a referendum (a majority of the people of 32 For a similar analysis of these issues, see Colin Murray & Aoife O'Donoghue, “Life after Brexit: Operationalising the Belfast/Good Friday Agreement’s Principle of Consent” (2019) Dublin University Law Journal (forthcoming). 17 WORKING PAPER – NOT FOR CITATION Northern Ireland voting in a poll). Second, the consent of North and South must be concurrently given. Third, if North and South give their consent, there is a binding obligation on both the Irish and British Government to introduce and support legislation in their respective parliaments to give effect to that wish. Each of these merits more detailed analysis. V.B A referendum in Northern Ireland The first point makes clear that democratic consent in the North requires a vote of the people. This factor might support a parallel obligation of a public vote in the South. Conversely, it might be said that the failure to make an equivalent specification of Irish legislation—the wording for the new Articles 2 and 3 was included in Annex B to the Agreement—should lead to the opposite conclusion. In any event, for the reasons advanced in section IV.D above, we are assuming that a referendum would be required in the South. Integrating the Irish constitutional and GFA requirements, therefore, there must be referendums, North and South. V.C Concurrent referendums The second point makes clear the referendums must be concurrent. One can debate what might be meant by ‘concurrently.’ It is important that the word ‘simultaneous’ is not used; it cannot be an absolute requirement that the referendums be held precisely at the same time. ‘Concurrently’ suggests that the two referendums must occur under the same set of political conditions, such that there is no reason to suspect that the consent given under one would have lapsed before the other consent is given. This could occur due to passage of time. By way of comparison, ratification of amendments to the US Constitution requires the approval of three quarters of the States. Ratification of the 27th Amendment began in 1789 and concluded over 200 years later in 1992. 33 This could scarcely count as ‘concurrent’. Political changes could affect concurrence much more quickly. For instance, between 2008 and 2010, Ireland suffered a calamitous economic crisis, the consequences of which are still being felt. It is arguable that a referendum held before the start of the crisis would not have been concurrent with a referendum held after the crisis commenced. If referendums are not to be held on the same day, the safest way 33 See Ruth Ann Strickland, ‘The Twenty-Seventh Amendment and Constitutional Change by Stealth’ (1993) 26(4) Political Science and Politics 716. 18 WORKING PAPER – NOT FOR CITATION to ensure concurrence is that the two are held very close together. No more time should elapse than whatever is required to organise the second referendum after the first jurisdiction has voted in favour of unification. Several prudential considerations apply in this context: § If the North votes ‘yes’ but the South votes ‘no’, this would likely destabilise Northern Ireland. It would have expressed a desire to leave the United Kingdom but there would be nowhere to go. § If the South votes ‘yes’ but the North votes ‘no’, this would not displace the GFA requirement of dual-consent, but could aggravate North-South tension. § If the island as a whole voted ‘yes’ but one or other jurisdiction (more likely the North) voted ‘no’, this could undermine support for the GFA. § If North and South vote on different days, this might undermine the claim of the new or amended constitution to be a shared framework for the entire island. § If the referendums are sequenced, there is a risk that public attention in the second jurisdiction will only focus on the issues once the result in the first jurisdiction is known. This would be problematic if negotiations were to precede the referendums, but less problematic if negotiations were to occur after the referendums. V.D The requirement for legislation The third point entails that legislation is required for unification. The UK and the Irish Government is each individually under an obligation to introduce and support such legislation. This has two implications for the interpretation of Article 3 of the Irish Constitution. First, it suggests that legislation might not be an adequate expression of democratic consent to unification. It would be odd to require the Irish Government to introduce legislation to give effect to legislation. However, this is not the only interpretation; consent legislation is conceptually separate to legislation putting the change into practice. One could clearly have the former without the latter, and the former might be conditional upon a vote in Northern Ireland. In this context, it would make sense to have a clause in the British-Irish agreement requiring legislation effecting unification to follow this, and the inclusion of such a clause would not necessarily suggest that legislation cannot provide consent. Second, it favours the view that the unification 19 WORKING PAPER – NOT FOR CITATION referendums, North and South, are not sufficient of themselves to bring about unification; each jurisdiction is required to enact legislation to give effect to that wish. If the UK and Irish Government agree on the form of unification, there should be few difficulties. It cannot be assumed, however, that agreement will be easy. As well as the North-South issues alluded to in section III.C above, there would also be significant EastWest issues in terms of sovereign debts, assets, and liabilities. Furthermore, the British Government would have a role in speaking for the interests of the prospective minority unionist community in the unified Irish state. The Brexit process is an illustration of how difficult it can be for two parties to agree on the terms of a separation. Article 50 of the Treaty of European Union—for all its flaws—at least specified clearly the legal consequence of a failure to reach agreement. The GFA makes no such provision. The GFA does not impose any obligation on Ireland or the United Kingdom to implement referendum votes in favour of unification. The obligation is rather on the Governments to introduce and support legislation in their respective parliaments that would implement those votes. This obligation applies separately and individually to each Government. That is, it does not depend on the Governments reaching an agreement with each other as to the form of unification. If agreement is not reached between the Governments, it is virtually certain that the Irish Government would meet its obligation under the GFA by introducing legislation that would be passed by the Oireachtas. At that point, unification would have occurred as a matter of Irish constitutional law. In these circumstances of disagreement, events at Westminster could play out in one of two different ways. Westminster could enact legislation giving effect to unification. In that context, residual disputes between the newly unified Ireland and the rump United Kingdom (assuming Scottish independence has not occurred) would have to be resolved through international negotiation and litigation. Alternatively, Westminster could refuse to enact legislation to give effect to Irish unification. In that scenario, Northern Ireland would remain in the United Kingdom as a matter of UK law but become part of Ireland as a matter of Irish law. There would be no breach of the GFA, as there is no obligation on the UK to respect the vote for unification. But the newly unified Ireland would presumably cite the lawful referendums North and South as the basis in public international law for a legitimate claim to self-determination that should be recognised by other states. VI. The requirements of the Northern Ireland Act 1998 20 WORKING PAPER – NOT FOR CITATION VI.A UK constitutional background Before analysing the relevant sections of the Northern Ireland Act 1998, we shall make a few observations about the UK constitutional system. The UK does not have a single mastertext constitution, in the same way as Ireland. Rather there is a patchwork of statutes, common law, royal prerogatives, principles, and conventions that collectively provide a constitutional framework. The core constitutional provision is parliamentary sovereignty. This means that Parliament has the power to enact any legislation that it sees fit; unlike Ireland, there is no legal document interpreted by the courts that constrains the legislative power of Parliament. Like Ireland, the UK takes a dualist approach to international law. Obligations in international law have no legal effect within the UK unless they are also enacted by Parliament. When combined with the doctrine of parliamentary sovereignty, this raises some difficulties. It probably explains why the GFA imposes no obligation on the United Kingdom—or indeed Ireland—to give effect to the result of the referendums. The UK courts have considered the Northern Ireland Act 1998 in a number of cases. In Robinson v Secretary of State for Northern Ireland, Lord Bingham commented (at para 11) that: ‘The 1998 Act does not set out all the constitutional provisions applicable to Northern Ireland, but it is in effect a constitution’. Accordingly, as with other constitutional statutes, it should be ‘interpreted generously and purposively, bearing in mind the values which the constitutional provisions are intended to embody’. 34 Lord Hoffman commented that the courts in interpreting the Northern Ireland Act should have regard to the background to its enactment, including the GFA. The requirements of the Northern Ireland Act 1998 considered in this section were required by the GFA itself. VI.B Status of Northern Ireland Section 1 of the Northern Ireland Act 1998 provides as follows: (1) It is hereby declared that Northern Ireland in its entirety remains part of the United Kingdom and shall not cease to be so without the consent of a majority of the people of Northern Ireland voting in a poll held for the purposes of this section in accordance with Schedule 1. 34 [2002] UKHL 32; [2002] NI 390. See generally Christopher McCrudden, ‘Northern Ireland, the Belfast Agreement, and the British Constitution’ in Jeffrey Jowel and Dawn Oliver (eds.) The Changing Constitution (6th ed, OUP, 2007) 227. 21 WORKING PAPER – NOT FOR CITATION (2) But if the wish expressed by a majority in such a poll is that Northern Ireland should cease to be part of the United Kingdom and form part of a united Ireland, the Secretary of State shall lay before Parliament such proposals to give effect to that wish as may be agreed between Her Majesty’s Government in the United Kingdom and the Government of Ireland. Section 1(1) essentially states a negative proposition: Northern Ireland shall not cease to be part of the UK, unless a majority in Northern Ireland votes for unification. There is no statutory obligation on the UK to respect a vote in favour of unification. Rather, section 1(2) imposes an obligation on the Secretary of State to lay before Parliament such proposals to give effect to that wish as may be agreed between the UK and Irish Governments. Interestingly, this obligation applies irrespective of whether the South has also voted in favour of unification. In this regard, it is inconsistent with the prior provisions of the GFA that require consent both North and South for unification. Presumably if the South had voted against unification, however, there would be no agreement forthcoming from the Irish Government to proposals that could be put before the Westminster Parliament. As noted above, the UK courts have treated the Northern Ireland Act 1998 as a constitutional statute and given it a broad, purposive reading in light of the GFA. It is unlikely, therefore, that the UK courts would enforce an obligation on the UK Government to lay proposals for Irish unification before Parliament in circumstances in which the South had voted against unification. Section 1(2) envisages negotiations between the UK and Irish Governments, but it is not clear whether the negotiations should take place before or after the referendum authorised by section 1(1). If section 1(2) read ‘such provisions .. as may have been agreed’, it would be clear that negotiations should occur before the referendum. The failure to use this phrase perhaps implies the opposite reading: negotiations should occur after the referendum. However, the better view—we suggest—is that both readings remain open: negotiations could take place before or after the referendum. Section 1(2) clearly presupposes that the UK and Irish Government will agree proposals to give effect to a vote in favour of a united Ireland. While such agreement is politically preferable, we must return to the question of what would occur in default of agreement. In section V.D, we suggested that the obligation on each Government to introduce and support legislation to give effect to votes in favour of unification exists independently of how the other Government acts. That is, if there is no agreement between the 22 WORKING PAPER – NOT FOR CITATION Governments, each Government should independently seek to legislate for unification. But recall that these provisions of the Northern Ireland Act 1998 are not just the UK interpretation of the GFA but are specifically required by Annex I of the ‘Constitutional Issues’ section of the GFA. If we read section 1 of the NIA as glossing the core obligations in the GFA, it could signify that unification should not proceed in the absence of agreement between the UK and Irish Governments. Such a reading would, however, give each Government a veto on unification, overriding what on any analysis is the core constitutional provision of the GFA, namely that it is for the people of Ireland, North and South, jointly to exercise their right to self-determination. We have already seen one respect in which section 1 appears to contradict the core provisions of the GFA, namely its implication that the UK Government should present to Parliament a form of unification agreed with the Irish Government, even if the South voted against unification. In that case, we favoured a reading that supported the core provisions of the GFA and we suggest that the same approach should be followed here. Agreement between the UK and Irish Governments is not a pre-requisite for unification; the GFA obligation on each Government to introduce and support legislation to give effect to a dual vote for unification applies irrespectively of the actions of the other Government. Section 1(2) of the NIA should therefore be taken as specifying a strong preference for the UK and Irish Governments agreeing the form of unification, but it does not make such agreement a precondition for unification. VI.C Calling the ‘border poll’ Schedule 1 of the NIA contains provisions on the holding of polls for the purposes of section 1. Paragraphs 1 and 2 provide as follows: 1 The Secretary of State may by order direct the holding of a poll for the purposes of section 1 on a date specified in the order. 2 Subject to paragraph 3, the Secretary of State shall exercise the power under paragraph 1 if at any time it appears likely to him that a majority of those voting would express a wish that Northern Ireland should cease to be part of the United Kingdom and form part of a united Ireland. Paragraph 1 does not impose any limitations on the date that the Secretary of State can set for a poll. Obviously, there are some dates—for instance 200 years hence—that would strain credulity. But, particularly given that the NIA is a constitutional statute, it would 23 WORKING PAPER – NOT FOR CITATION be permissible for the Secretary of State to set a date several years hence to facilitate negotiation and consultation prior to the holding of the poll. There has been a wealth of commentary and some litigation related to the interpretation of the trigger provision in paragraph 2.35 We do not propose to add to this legal analysis but will confine ourselves to three related points. Given the requirement for ‘concurrent’ referendums, it is highly unlikely that the Secretary of State would direct the holding of a border poll without the agreement of the Irish Government to conduct a concurrent referendum. If the Irish Government refused to hold a referendum, the referendum in the North could have no legal effect. Second, while the interpretation of paragraph 2 is important, it is likely that any decision to hold a poll would also or exclusively invoke the statutory discretion in paragraph 1, thereby reducing the risk of any successful legal challenge to the holding of the referendum. Third, it is possible that a court could direct the Secretary of State to hold a poll, if it considered that the criteria in paragraph 2 had been satisfied. This would not of itself impose any legal obligation on the Irish Government to hold a concurrent referendum, although it might increase the political pressure. Such court intervention seems a long way off at present, but the fact that it is possible means that the UK and Irish Governments should not assume that they fully control the decision whether to hold unification referendums. VII. Parameters and desiderata The analysis in sections III-VI sought to identify and integrate the legal requirements that apply under Irish constitutional law, the GFA and the Northern Ireland Act 1998 to Irish unification. There are essentially two legal parameters: § Concurrent referendums, North and South, to approve unification. § Legislation in Westminster and the Oireachtas to give effect to the referendum votes, if unification is approved. It is possible to design a wide range of models that comply with these parameters. In the remainder of this section, however, we suggest three desiderata against which different models can be evaluated. These desiderata reflect normative preferences; that is their point. They may of course be contested by those who hold different normative 35 For a recent review of this issue, see Colin Murray, ‘A Referendum on a United Ireland: Perspectives from UK Constitutional Law’ IACL-AIDC Blog (18 February 2020) available at <https://blog-iacl-aidc.org/irishunification/2020/2/18/a-referendum-on-a-united-ireland-perspectives-from-uk-constitutional-law>. 24 WORKING PAPER – NOT FOR CITATION preferences. Precisely for this reason, however, it is helpful to articulate and defend the desiderata as clearly as possible. The first desideratum, we suggest, is that the electorates voting in the referendums should, insofar as is possible, know the implications of what they are voting for. The second desideratum, we suggest, is that a process of unification should aim to minimise the risk that the UK and Irish Governments diverge in their approaches after referendums have been held. In sections III-VI, we have endeavoured to explore the complicated legal and political situation that would transpire if there were referendums that authorised unification but the UK and Irish Governments could not then agree on the form of unification. This situation is, in our view, highly undesirable. The third desideratum, we suggest, is that there should be as wide as possible political and public consultation on the proposed form of a unified Ireland. In particular, nationalist and unionist communities in Northern Ireland should be extensively consulted. What is at issue in such referendums is the constitutional framework for the newly unified Ireland. The legitimacy of that constitutional framework—both normative and perceived—will be enhanced if all the citizens of that new state had significant opportunities to shape its structure. VIII. Models of unification processes In the following diagrams, there is a temporal progression from left to right. The simplest way of respecting the two legal parameters is as follows: Ireland Referendum Legislation NI / UK Referendum (NI) Legislation (UK) Model A However, this does not provide for any negotiation. It is scarcely conceivable that unification could occur without any negotiations. Negotiations would necessarily involve the British and Irish Governments. It would be preferable for them to be far broader, however, and involve the institutions in Northern Ireland. It might also be appropriate, we suggest, to involve political parties and the general public North and South, whether through citizens assemblies or some other means. In the models that follow, we use ‘Negotiation’ to connote all those activities. 25 WORKING PAPER – NOT FOR CITATION We saw above that the meaning of ‘concurrence’ in the GFA is not entirely clear. We suggested (a) that it does not require that the referendums must be held simultaneously but (b) that there cannot be a significant change in political conditions between the two referendums. For this reason, we suggest that it would not satisfy the requirement of concurrence for the definitive referendum to be held in one jurisdiction, followed by negotiations, followed by the definitive referendum in the other jurisdiction. The electorate in the other jurisdiction would be voting under very different conditions from the electorate in the first jurisdiction, as only at the later time would the detailed proposals be known. Negotiation could take place after the referendums, which would look like this: Ireland Referendum NI / UK Referendum (NI) Negotiation Legislation Legislation (UK) Model B Model B has several difficulties. It produces exactly the situation that the first two desiderata seek to avoid: the implications of people’s votes would be unknowable because the form of a united Ireland had not yet been determined; it would increase the risk that negotiations would break down leading either to conflicting legislation being passed in Ireland and the UK to give effect to the referendums, or no legislation being passed in the UK—leading to an international dispute over the status of Northern Ireland. In this Model, as there is only one referendum in Ireland and it predates the negotiation, it follows that Ireland would have to make all the constitutional changes for the newly unified state without any discussions with the UK Government or political parties in Northern Ireland, let alone ordinary citizens. This would make it much more difficult to negotiate any agreement, increasing the risks of divergence between the UK and Ireland identified above. The Irish Government could postpone the attendant constitutional changes until after the negotiations like this: Ireland Ref - bare NI / UK Ref (NI) Negotiation Ref - details Legislation N/A Legislation (UK) Model C Model C is unattractive in several respects, however. It would mean that the electorate in the South would get to vote on the proposed form of a united Ireland, but the electorate in the North would be denied that information at the time of their referendum. Given the 26 WORKING PAPER – NOT FOR CITATION requirement of concurrence, a united Ireland would be mandated by the two simultaneous referendums that precede the negotiations. The Irish Government (and the UK Government) would then be under an obligation—under the GFA—to introduce and support legislation giving effect to unification, irrespective of whether the referendum on details were ever passed. This could require unification to go ahead in a form that nobody thinks is desirable. A better approach, we suggest, would be for the negotiation to occur before the referendums: Ireland Negotiation NI / UK Referendum Legislation Referendum (NI) Legislation (UK) Model D If the negotiations precede the referendums, the electorate in the South could vote on unification and the full package of constitutional changes. The electorate in the North would in theory only be voting on the principle of unification, but they would do so equipped with the knowledge of the constitutional structure of the unified state. The constitutional changes in the South would only take effect in the event of the North voting Yes—the same model as was applied for the referendum approving the GFA and related constitutional changes in 1998. To enhance the democratic mandate further, the Irish Government could adopt the model followed in relation to the abortion referendum in 2018. That is, it could publish in advance the Scheme of the Bill that it proposed to introduce in the Oireachtas if both North and South voted for unification. Model D significantly reduces the risk of divergent approaches in the UK and Ireland after the referendums. As both Governments would have approved the proposal to be put to the electorates, it is less likely that either Parliament would decline to enact the subsequent legislation. Of course, there could equally be disagreement between the governments before the referendums, which might prejudice the ability to hold the referendums. Our view, however, is that such disagreement is less problematic if there has not been a dual referendum vote in favour of Irish unification. The third desideratum is that there be broad public involvement in designing and approving the structure of a united Ireland. The time allowed for negotiation should be reasonably lengthy. We would therefore further specify Model D as follows: Ireland Negotiation 27 Ref Leg WORKING PAPER – NOT FOR CITATION NI / UK SoS - call poll (3-5 years) Ref (NI) Leg (UK) Model D1 For the reasons given in section VI.C above, we see no difficulty with this delay in holding the referendums. Model D1 is less than satisfactory in one respect, however. Unionists in Northern Ireland would have no incentive to participate in any negotiations about the shape of a united Ireland prior to the referendums being held. It is quite legitimate for Unionists not to participate in a process to which they are fundamentally opposed. After the referendums, however, it would be too late for Unionists to have any significant input into the shape of a united Ireland. The negotiation phase should still be designed in a way that is open to the involvement of Northern Unionists; but there should be realism about the likelihood of participation. It is this concern, we believe, that has prompted some to consider what are sometimes called ‘two-step’ approaches. Essentially, these involve referendums at two points in the process. If the public approval is to be disaggregated into two steps like this, we need to be clear about the precise role being played by each. Model E is similar to Model C above, except that it would apply the dual referendum requirement in both jurisdictions: Ireland Ref - bare NI / UK Ref – bare (NI) Negotiation Ref - details Legislation Ref – details (NI) Legislation (UK) Model E Model E reflects some suggestions in the context of the Brexit debate that there should have been a second referendum to ratify the terms of the withdrawal agreement. Our comments in respect of Model D make clear our support for the democratic principle that the electorate should get to vote on the final deal. But having the electorate also vote earlier in the process raises several problems. The precise problems depend on the legal role of the first referendums. The two-referendum approach only solves the problem of unionist participation if the first referendums counted as the definitive approval for Irish unification. If unification could be avoided at the ‘details’ stage, then unionists would retain an incentive either not to participate, or to negotiate the details in a way designed to make a united Ireland even less attractive, thereby increasing the chances of a ‘no’ vote. (Indeed, the incentives for unionists not to participate continue after the final referendum until after Westminster passes legislation to give effect to the unification referendums.) 28 WORKING PAPER – NOT FOR CITATION However, if the first referendums make Irish unification final, then there is a significantly increased risk that the form of unification cannot be agreed between Ireland and the UK, leading to a divergence in approaches in the two jurisdictions. There is also the risk that the referendums on details would never be passed, but that unification would have to proceed anyway (the problem with Model C above). Conversely, if the ‘details’ referendum is all that counts legally, unionists would still have strong incentives not to participate in the negotiations. Moreover, it is difficult to see what purpose the initial referendums would serve in that context. There is possibly a halfway house situation where the first referendum establishes a united Ireland under a set of transitional arrangements, which might thereby encourage Unionists—once the first referendum has been lost—to negotiate new constitutional arrangements for the newly unified state. But this halfway house raises the full panoply of problems that we have been identifying. Unionists would not have engaged in negotiations on the transitional arrangements before the first referendum, raising the very problem that the model is supposed to avoid. The two electorates would be voting on unification without knowing what the final constitutional arrangements would be, introducing a lot of uncertainty in the process and increasing the risk that the UK and Irish Governments would diverge on the status of Northern Ireland. The Irish Government would be simultaneously negotiating the form of withdrawal with the United Kingdom, for presentation at Westminster and the Oireachtas, and negotiating replacement constitutional arrangements within Ireland. Unionists would have incentives not to participate in the Irish discussions unless or until legislation was passed at Westminster. If there were to be referendums North and South on the replacement arrangements, Unionists would have an incentive to oppose the proposals in the hope of at least complicating the passage of legislation at Westminster. If the replacement measures were not approved at the subsequent referendums, the transitional arrangements would continue indefinitely, even if no-one thought this was appropriate. Once unification was formalised at Westminster and the Oireachtas, however, the Irish Government could simply pursue constitutional amendments in the normal way (50%+1 of those voting at an all-island referendum). This would be in no way protective of unionist interests. In short, the halfway house approach of Model E has the disadvantages of all the other models, with few of the advantages. A variation of the two-referendums approach is to hold an initial referendum in Northern Ireland only in order to provide the democratic mandate to commence negotiations on a 29 WORKING PAPER – NOT FOR CITATION united Ireland. Such a mandate already exists in Article 3 of the Irish Constitution, so no corresponding referendum would be required in the South: Ireland N/A NI / UK Ref to negotiate (NI) Negotiation Full ref Legislation Full ref Legislation (UK) Model F The argument in favour of such an approach is that it would incontrovertibly legitimise negotiations on the form of a united Ireland. This in turn might increase participation by northern Unionists. This is far from certain, however. The initial referendum could provide a basis for centre-ground parties (i.e. those without a committed position on the preferred long-term constitutional status of Northern Ireland) to engage in negotiations about the form of a united Ireland. But unionists would still—perfectly legitimately—retain the incentives not to participate in any negotiations before a united Ireland is a fait accompli. This seems to us to be an unavoidable political reality that prejudices any full achievement of the third desideratum. Until the final legislation is enacted in Westminster to give effect to the unification referendums, there remain incentives for unionists to seek to stop unification rather than engage in negotiations about its terms. Against that background, would a referendum in Northern Ireland to mandate negotiations be advantageous? There seems a real risk that the general public would misperceive such a referendum as being the ‘border poll’ for the purposes of the GFA. It could not play that role because it is not concurrent with a democratic expression of consent in the South. There is thus a real risk both of confusion for all members of the community and disenchantment for some. If it transpired that the details of a united Ireland could not be negotiated, such that the actual unification referendums were never held, the initial referendum might have the lingering effect of keeping the constitutional question indefinitely on the table for political discussion, making it more difficult to operate the institutions of the GFA. For all these reasons, our view is that Model D1 provides the best process for unification referendums. It complies with the legal parameters, realises the first two desiderata and partially realises the third desideratum: Ireland NI / UK Negotiation SoS - call poll (3-5 years) Ref Leg Ref (NI) Leg (UK) Model D1 30 WORKING PAPER – NOT FOR CITATION We should conclude with a few words about concurrence. As noted above, ‘concurrent’ does not mean simultaneous. It would be possible to separate the referendums in the North and South like this: Ireland NI / UK Negotiation SoS - call poll (3-5 years) N/A Ref Leg Ref (NI) N/A Leg (UK) Model D2 Under Model D2, the referendum in the South would only proceed if the referendum in the North were carried. The advantage of this approach is that it removes the risk of the North voting against unification but the island as a whole voting in favour of unification. Such an outcome, while formally quite consistent with the GFA, might undermine support for the GFA principle of dual consent, North and South. It leaves in place, however, the possibility that the North could vote in favour of unification but the South reject unification—likewise a destabilising outcome. It has two discrete disadvantages. First, it seems likely that public attention in the South would only properly focus on the issues after a vote in favour of unification in the North. As all negotiation and consultation would by that point have completed, this would be problematic. Second, it would mean that the foundational act for the new state was staggered over a period of time; this might undermine the legitimacy of the new state. For these reasons, we think Model D1 involving simultaneous referendums remains preferable. 31