Political Constitutionalism and Referendums: The Case of Brexit

Referendums have increased in frequency and number in democratic states across the globe over the past 50 years (LeDuc, 2003: 29). Between 1975 and 2000, the 58 functioning democracies with populations above three million held some 39 referendums (Tierney, 2012: 1), and since then they have become 4 times more common (Thompson, 2022: 195). These referendums have typically been of four main types (Tierney, 2009: 360). The first deal with the very existence or shape of the polity, be it to approve secession from an existing state and the founding of a new state, as occurred in South Sudan in 2011, or the remodelling of the polity by the granting of greater sub-state autonomy, as in the failed Bolivian referendum of 2006. The second focus on the approval of reforms to the polity's regime, be it to endorse the creation and proposal for a new constitutional order, as in Venezuela in 1998 and 1999 or Chile in 2020 and the failed referendum of 2022; or to ratify constitutional amendments, as occurred with regard to the rights to abortion, divorce and blasphemy in Ireland in 2018. The third is a hybrid type and involves international treaties raising both polity and regime issues as a result of membership of an international organisation, such as NATO or the European Union (EU), whereby accession or reforms involve the transfer of certain state competences to supranational or intergovernmental bodies and/or changes to how those competences are exercised. Notable examples include both the failed French and Dutch referendums on the Treaty establishing a Constitution for Europe in May and June 2005, and the United Kingdom's ‘Brexit’ referendum on whether to leave or remain in the EU of June 2016 – the focus of this article. A final type – potentially of a non-constitutional nature, at least in the strict sense, relates to particular policies, as is often the case with various local Swiss referendums.

Referendums of the first three kinds are often praised or criticised as instances of popular sovereignty, through which the people possess the constituent power to determine both the locus of legal and political authority, and as such who, where and what will be subject to it, and the terms and conditions of its exercise, and consequently how and with what limitations such authority can be deployed. Supporters of such referendums see them as valuable instances of popular self-determination, that can provide legal and political systems with democratic legitimacy. They contend they offer a mechanism through which citizens can participate in shaping a polity and its regime and identify with them as ‘theirs’ (Tierney, 2009, 2012). Meanwhile, the fourth type provides an additional channel for citizens to enhance the responsiveness of politicians, either by challenging unpopular policies or proposing new ones that parties have kept off the electoral agenda. By contrast, opponents caution against the potentially populist and plebiscitary character of referendums (Offe, 2017). They believe referendums can diminish democracy and remove important liberal constraints on majority tyranny (Van Crombrugge, 2021: 111–113). Contrary to populist claims, these critics consider referendums more susceptible to elite manipulation and less deliberative than electoral, representative democracy, not least through reducing complex issues to a simple binary choice. Meanwhile, as alleged expressions of a putative singular popular will that lie outside the constitution, they may fail to attend to minority voices and rights – not least when ‘the people’ gets framed by rightist populists in native, ethnic terms. On this account, referendums only prove legitimate when they supplement rather than supplant representative democracy, and operate within rather than outside the extant constitutional framework (Trueblood, 2020; Van Crombrugge, 2021).

These critics see the Brexit referendum as ‘a textbook example of how not to conduct a referendum’ (Trueblood, 2022: 183), offering ‘a clear and unambiguous lesson on what democracies ought not to do’ (Offe, 2017: 22). Indeed, it has become common to assess quite different referendums – from the Columbian peace agreement referendum of October 2016 and the Chilean referendum on a new constitution of September 2022, to the above mentioned Irish referendums of 2018 – on the basis of how far they supposedly repeated or avoided the alleged mistakes of the Brexit referendum (e.g., Gargarella, 2022). That said, commentators often disagree as to how far and in what ways these different referendums paralleled or differed from Brexit, these disagreements reflecting differing views of the coherence and legitimacy of appeals to constituent power and popular sovereignty, notions that play a key role in many constitutional traditions (Verdugo, 2022a). Determining how far, and in what ways, the Brexit referendum can be regarded as flawed or not potentially has a wider significance, therefore

Much of the debate over the Brexit referendum has centred on whether it might have been better regulated under a legal constitution, that codifies and entrenches as superior law certain individual rights and the rules of the political process, as opposed to the United Kingdom's distinctively political constitution. Critics of the referendum typically see it as being both at odds with, and indicative of the weaknesses of, the UK’s political constitution. On the one hand, they regard what they see as the referendum's appeal to popular sovereignty to decide a complex issue as being at odds with the doctrine of parliamentary sovereignty (Offe, 2017: 18, 21–22), which forms the keystone of the political constitution (Ewing, 2013: 2118; Gordon, 2019: 133–37). The political view of constitutionalism locates the constitution in the character and design of the political system and the modus operandi of its component political processes (Bellamy, 2007; Gee and Webber, 2010). According to this political conception, there can be no higher laws other than the laws that emerge from a duly constitutive and constitutional political process. Moreover, it is the representative and parliamentary processes that gives the democratic system its constitutional qualities. These allow citizens to hold governments and the administration accountable either indirectly through elections or directly via elected parliaments and the need for the government to govern and legislate with the support of a plurality of the population and a majority of MPs (Griffith, 1979; Tomkins, 2005: 1–10). By contrast, these critics contend referendums produce a populist disfigurement of democracy (Urbinati, 2019), that deny the pluralism and tolerance of representative democratic decision-making through emphasising the existence of a mythical and singular people's will (Van Crombrugge, 2021: 115–116; Weale, 2018).

On the other hand, though, these critics also believe the referendum highlights the weaknesses of the political constitution as a curb on populism and a potential tyranny of a democratic majority (Trueblood, 2022). They claim that not only can any government possessing a stable parliamentary majority act as an ‘elective dictatorship’ and make and interpret the law as it wishes (Hailsham, 1978: 127; Pettit, 1999: 176), but also the executive can further overrule parliament by appealing directly to a popular majority in a plebiscite, as Prime Ministers Theresa May and Boris Johnson both attempted to do in order to avoid Parliamentary scrutiny of their respective proposals for leaving the EU. These critics claim that a codified constitution that acts as a higher law that frames the operation of politics would have ensured that both the referendum and the parliamentary process operated in an inclusive way, that protected minority voices (Bogdanor, 2019a, 2019b: ch. 3).

Needless to say, defenders of the referendum tend to dispute both these points. They argue that representative democracy can be prone to elitism and a lack of responsiveness to popular concerns and provide insufficient space for citizen participation through more truly democratic avenues. They consider referendums as a truer form of political constitutionalism, offering an opportunity for the direct expression of popular political sovereignty on which the legitimacy of parliament's legislative sovereignty ultimately rests (Tierney, 2013). Meanwhile, they note that many legal constitutions recognise popular sovereignty as the constituent power, with some constitutions explicitly authorising this power to operate outside constitutional limits, given that it may initiate and approve a thorough reform of the extant constitution (Colón-Riós, 2020). To that extent, legal constitutionalism per se cannot be viewed as offering a constraint on popular sovereignty.

This article challenges the views of both opponents and proponents of the Brexit referendum. Putting to one side the strictly legal issue of the compatibility of the referendum with prevailing UK public law, it explores the broader normative issue of the compatibility of referendums in general, and this one in particular, with the very notion of a political constitution based on parliamentary rather than popular sovereignty. Referendums are not foreign to the UK, which has held 11 of them since 1973. Indeed, A. V. Dicey, whose Introduction to the Study of the Law of the Constitution (1915 [1885]: Part II, ch. 13, 402, 405) offers the locus classicus of the doctrine of parliamentary sovereignty, was the first to advocate the referendum in Britain (Dicey, 1890). Although Dicey's position has been viewed as ‘paradoxical’ (Bogdanor, 1994: 34), I shall argue that referendums can play a valid political constitutional role as part of a system of checks and balances on executive authority so long as they are appropriately embedded within a system of representative democracy.

On this account, referendums need not displace parliamentary sovereignty with an unmediated popular sovereignty. They can allow an important role for the people while nevertheless relying upon parliament to structure the debate and oversee the decisions’ implementation. As a result, they can encourage popular participation without producing populism. So conceived, referendums supplement representative democracy on those constitutional issues where allowing the elected government to decide would make it ‘judge in its own cause’ – notably, changes to the electoral system or to the demos (Thompson, 2022: 196). I shall argue that the Brexit referendum was such a case, given it involved changes to the demos and hence to the scope of parliamentary sovereignty. However, while it can be regarded as meeting certain minimal standards of democratic legitimacy, the process could certainly have been improved – even if the result might well have remained the same.

The argument proceeds as follows. First, I briefly outline both the constitutional qualities political constitutionalists associate with a parliamentary representative democracy and some of the limitations that critics attribute to it. Second, I explore how far referendums can be said to be compatible with the former while potentially addressing the latter. I shall argue this can be possible to the extent referendums can be nested within, and made a component of, the standard picture of the political constitution so as to form part of a system of political checks and balances with constitutional qualities. Third, I turn to the Brexit referendum and consider whether, while far from perfect, it was so nested and can be regarded as democratically and constitutionally legitimate in terms of the political constitution, performing an important supplementary function appropriate to such constitutive issues. I conclude by noting that there are positive as well as negative lessons that can be drawn from the Brexit referendum concerning the ways referendums can enhance both representative democracy and constitutionalism as part of a political constitution.

Political Constitutionalism

Political constitutionalism proceeds from the argument that the laws determining the terms of social and political co-existence, including the basis and interpretation of fundamental rights, are matters of reasonable disagreement among those who are subject to them (Bellamy, 2007: 4–5; Waldron, 1999: 107–18). Political constitutionalists claim that the most appropriate way of recognising the contested nature of these terms and rights is to subject them to a collective decision-process as to their content and implementation that treats all concerned impartially and promotes reciprocity among them, so all are regarded with equal respect and concern (Bellamy, 2007: ch. 4; Waldron, 1999: ch. 11). Representative democracy linked to parliamentarism is held to satisfy these desiderata in a manner that deploys certain qualities standardly associated with constitutionalism (Bellamy, 2007: 12, 259, 260; Gordon, 2019: 131–3; Griffith, 1979; Tomkins, 2005: 1–10).

These constitutional qualities are achieved not through legal mechanisms that hold politicians and the administration to the norms of a codified and entrenched constitution that acts as a higher law, but through the operation of a political system that renders them subject to authorisation by and accountable to citizens (Bellamy, 2007: ch. 6). However, these political mechanisms incorporate two key constitutional devices – those of checks and balances (Bellamy, 2007: ch. 5), reflecting respectively negative and positive aspects of constitutionalism (Barber, 2018: 2–9). On the one hand, governments are checked through requiring authorisation by, and being held regularly accountable to, citizens through the electoral process. These mechanisms provide checks on arbitrary rule and the protection of individual rights. They can also be supplemented by weak form judicial review, that highlights the adverse impact of general legislation on the rights of particular individuals (Bellamy, 2011; Waldron, 2016: ch. 9). On the other hand, governments, and indirectly those citizens supporting them, are balanced by having to compete against opposition parties – both in elections and in the legislature, and hence must ‘hear the other side’. These balances promote the participation in decision-making of citizens so as to support the need for politicians to appeal to their commonly avowable reasoning and interests and enhance identification with the public good. Bicameralism, federalism and a proportional electoral system can all, where appropriate, provide further sources of balance that foster deliberation and the need to address all sections of the community (Waldron, 2016: chs 3–5).

All these mechanisms can be seen as sharing a lineage with the ancient notion of the mixed constitution, which aimed at achieving a degree of mutual checks and balances between the different classes of a society (Bellamy, 1996). Although in their modern incarnation they involve democracy, their virtues lie not as means for collective self-rule or popular will formation, as in certain theories of deliberative and direct democracy. Rather, as per the neo-Roman republican tradition, their core concern is with the promotion of freedom as non-domination and relatedly of equality (Bellamy, 2007: ch. 4; Tomkins, 2005: ch. 2). From this perspective, what matters is that public authorities cannot act arbitrarily: that is, simply as they will, without consulting the views and interests of those subject to their rule.

The capacity for arbitrary rule arises when there is an inequality of power and a relationship of dependence of the ruled on the rulers (Lovett, 2001). Political constitutionalists regard a democratic system as a way of overcoming this possibility by placing rulers under the equal influence and control of the ruled, and rendering all equal before the laws (Bellamy, 2007: ch. 6). Indeed, given citizens are both rulers and ruled in turn, with politicians elected by appealing to a majority among them, citizens too can be considered as mutually influencing and controlling each other. In such a situation, no politician or citizen can reliably will what the collective rules should be without consulting others and appealing to their commonly avowable judgements, while personal dependence on a particular master or patron is replaced by the mutual dependence of equals for their considered and voluntary support.

By advocating an institutional framework for a non-dominating system of democratic government, political constitutionalism seeks to achieve a form of rule that treats all with equal respect and concern (Bellamy, 2007: ch. 4). The securing of equal respect can be regarded as an intrinsic feature of political constitutionalism. Elections based on one person, one vote and majority rule treat all citizens as possessing equal status, and provide a mechanism that gives equal weight to their views and interests that is neutral and impartial with regard to their worth for any collective decision (Christiano, 2008). This proves instrumental to securing equal concern and the negative and positive constitutional purposes of protecting rights and fostering participation and the public interest respectively (McGann, 2004: 56, 71). The balancing features of the political constitution encourage reciprocity among citizens and a willingness to compromise to create viable party programmes with broad popular appeal. For example, competition between different parties incentivises politicians to fish for votes and construct programmes that build coalitions between different groups of citizens that appeal to the median voter – usually that set of preferences that represent the Condorcet winner across the electorate (Ordeshook, 1986: 245–57). ¹ As R.A. Dahl famously observed, within pluralist societies majority rule tends to be the rule of an alliance of various minorities (Dahl, 1989: 218). Given many minorities could swap allegiance this motivates both a regard for minority rights and the need to frame policies in terms of their benefit to the public interest rather than particular sectional interests.

Meanwhile, the prospect of future elections constrains arbitrariness by governments, which will be held accountable for their failings when they next seek authorisation to rule. That parties may alternate in power also creates a reason to wish the rules of the game to remain fair and for the judiciary to be independent and ensure all are equal under the law. As noted above, bicameralism – especially where the second chamber is selected by different electoral rules to the first – may likewise provide a spur to inclusive deliberation and a regard for the public interest. The precise nature of these institutional arrangements will tend to reflect the social characteristics of the polity concerned (Dahl, 1989: 251 ff).

Of course, there are numerous imperfections and flaws in any extant political constitution. Many of these problems are those associated with representative democracy more generally (Dunn, 2005: 19). Critics focus on the elite character of representative systems, the lack of responsiveness of parties and the ways collusion between them can keep key issues off the electoral agenda, and the resulting disempowered nature of citizen participation, limited as it is to a periodic vote on the, often overlapping, programmes offered by politicians (Tierney, 2012: 3–4; Tierney, 2013: 2187–8). These failings are held to largely depoliticise democracy and limit both its checking and balancing capacity, thereby weakening its negative and positive constitutional qualities. If parties fail to adequately contest each other and offer clear alternatives, then it becomes harder to check governments and hold them to account. Likewise, if the main parties ignore the preferences of certain groups of citizens and take them off the electoral agenda, then these preferences will carry no weight in the balance of political priorities within any of the party programmes (Mair, 2013). The danger in such cases is less that of the tyranny of the majority as of a tyranny of a minority – a feature exacerbated in plurality electoral systems such as that of the UK, which reduce the number of viable parties in play, and result in parliamentary majorities that rarely represent a majority of voters. This possibility has seemed especially perilous in the case of constitutional rules – especially those relating to the political process, where a government or politicians as a class can have an interest in enacting policies that entrench their own position and weaken competition, as with gerrymandering (Ginsburg and Aziz, 2018). Indeed, a number of authors sympathetic to the political constitutionalist position on matters of substance, such as abortion rights, have advocated judicial review on matters of democratic process on these grounds (Dahl, 1989: 188–192).

Given that the normative case for political constitutionalism has been linked, as we saw, with republican notions of citizenship and the avoidance of the domination resulting from arbitrary rule, the participatory deficits of representative democracy have appeared to many critics as in conflict with the aspiration of its advocates to offer an alternative to liberal accounts of democracy (Tierney, 2013: 2187). After all, a key criticism of legal constitutionalism by political constitutionalists relates to its alleged elitist and anti-democratic character. Against this criticism, some legal constitutionalist critics have countered that contestation through legal challenges, such as class actions, offer an avenue for more citizen participation, and an important supplementary legal source of constitutional checks and balances (Lafont, 2019: ch. 8). I leave these arguments to one side (for a discussion see Bellamy, 2013). However, other, more radical, republican and democratic critics argue that referendums – especially on constitutional and constitutive issues, but also with regard to certain policies – provide a way of both giving such legal constitutionalism a political endorsement of a democratic kind (Ackerman, 1991), and providing a more directly democratic form of political constitutionalism (Tierney, 2012, 2013). As they typically note, referendums would seem, at least prima facie, to provide a stronger realisation of the intrinsic quality of democracy than voting in elections, in that it gives each citizen a direct say on a salient issue – one that may even have been occluded by parties (Tierney, 2012: 19, quoting approvingly Bogdanor, 1981: 93). However, even if this obtains, worries persist that referendums may lack many of the instrumental constitutional qualities associated with a representative democratic system (Lord, 2021). Nevertheless, as was suggested above, suitably designed, they could possibly provide a potential check and balance to the legislature on matters where incentives are lacking for politicians not to put their own interests in holding on to power above those of the public at large (Thompson, 2022). It is to these issues that I now turn.

The Constitutional Qualities of Referendums

As was noted in the introduction, referendums typically fall into four different issue categories. All of these can take two broad forms, with two distinctive modalities. With regard to issues, we saw how they can be categorised as related to constitutive, constitutional, international or policy matters (Tierney, 2012: 11–15). The first involves decisions that redefine either the polity, as in referendums on secession - such as the recent Scottish referendum on independence, or the fundamental form of its regime, as in referendums on an entirely new constitution. The second relates to more limited amendments to an already existing written constitution or constitutional statute, which operate within the prevailing constitutional structures with regard to both the polity and the regime, such as the recent Irish referendum on abortion or the UK referendum on changing the electoral system. The third relates to the ratification of international treaties that raise constitutive, constitutional and possibly policy issues. Finally, the fourth concerns policies ranging from changes in local planning regulations, as is common in Swiss communal and some English Parish referendums, to major policy decisions, such as the California Proposition 218 that – while technically a constitutional amendment – was designed to give voters a direct say over levels of local government taxes. As this last example indicates, there can be some overlap between the issue types – a policy issue can have constitutional implications, as in this case, while most constitutive referendums will raise both constitutional and policy issues, as was the case with Scottish independence and is likely to be true of referendums on international treaties, such as Brexit. It will also be a matter of degree as to how far a constitutional amendment can be regarded as altering the nature of the regime or polity, as might have been the case with the UK referendums on electoral reform and the establishment of elected Mayors.

As to forms, referendums can be regarded as either top down, where the government calls and frames the referendum, or bottom up, where a process exists for citizens to request a referendum – for example, by gathering a prescribed number of signatures. Here too there are forms that combine elements of both, such as the Irish use of citizen juries to identify issues on which referendums might be called.

Finally, concerning modalities referendums may be negative, called to abrogate an existing law or arrangement; or positive, aimed at approving a proposal for a new measure or arrangement. Once again, this distinction may not be clear cut. In some cases, a successful negative referendum may entail the need for new positive measures or arrangements, though the precise nature of these may not be clearly specified. Likewise, if a positive proposal has failed to get approval it has by implication been abrogated.

The degree to which referendums can be regarded as either possessing themselves constitutional qualities or complementing those of representative and parliamentary forms of democracy will depend on the particular ways the issue, forms and modalities interact with each other and, where relevant, with the system of representative democracy (Lord, 2021). Broadly speaking, the intrinsic securing of equal respect appears strongest in referendums that are bottom up and allow citizens to propose new policies, and weakest in those that are top down and merely seek approval of a government measure (Cheneval and el-Wakil, 2018). Referendums of the former type, that belong to the policy issue category, take a bottom up form and involve a positive modality, approximate most closely to exercises of direct democracy that can be characterised as embodying popular sovereignty. As such, prima facie at least, they appear best placed to counter the criticism of representative democracy for its lack of participation and elitism. However, these types of referendum are also those where critics have most doubted their instrumental constitutional qualities. These worries involve the limitations of referendums with regard to collective policy making; citizen competence and elite manipulation; majority tyranny; and the lack of accountability or undermining of the role of representatives (Lord, 2021: 35; Tierney, 2012: 22–24; Trueblood, 2020: 435–441). I shall consider each of these worries in turn, although as we shall see they relate in various ways to each other.

The key problem for referendums with regard to collective policy making relates to the weighting of citizens’ preferences across the whole domain of policy options (Bellamy, 2018: 314–16). A referendum necessarily focuses on a single issue, whereas elections focus on party programmes that bundle together a range of policies and seek to weigh them against each other, including giving consideration for the knock-on effects of decisions in one policy area for decisions in others (Tierney, 2012: 37). A decision on a policy taken in isolation from others risks producing policy incoherence – as when voters vote for tax cuts that would undermine the government's ability to fund public services and infrastructure that they wish to see expanded (Haskell, 2001: 16). True, the Ostrogorski paradox suggests the possibility that bundling may involve putting together a number of policies, many of which taken on their own are not supported by a majority (Lord, 2021: 34; Setälä, 2006). In this scenario, a bottom-up, positive, policy referendum might be justified to challenge parties having taken off the agenda or ignored a majority view in a given area. However, that would depend on the policy being discrete, which in most cases is unlikely.

The policy implications of a popular vote may also be imprecise. This is especially true of negative, abrogative, referendums, where what is to be put in place of the rescinded measure – if anything – may be unclear. Yet, positive referendums may also be imprecise unless the policy proposal is clearly specified. A related issue stems from the binary nature of many referendums, whereby the two options on offer may be arbitrary and suppress alternatives that would have been the most preferred (i.e., the Condorcet winner), particularly if voters could rank their preferences among multiple possibilities (Bellamy, 2018: 317–8; Weale, 2018: 62–66). Of course, voters could be given multiple options to choose from and allowed to rank them, as occurs in some Swiss referendums (Lord, 2021: 36). Yet, these are likely to still be options relating to a single policy.

Worries about citizen competence enter here. Some advocates of policy referendums claim they benefit from the ‘wisdom of crowds’ (e.g., Landemore, 2013). However, that depends on citizens possessing sufficient knowledge about the relevant policy area to be able to make a reasonable guess as to which of the available options might be best. Meanwhile, the aforementioned problems regarding policy connectedness and the representativeness of the options on offer still apply. Even referendums that emerge from a popular initiative can be subject to elite manipulation (Honohan, 2002: 220–30), as they will generally arise as a result of a campaign led by a small group of activists, who will then seek to shape the referendum question so as to favour their preferred result. In the absence of personal expertise, most voters will also follow elite cues from media and other campaigns (Honohan, 2002: 218). In these respects, referendums may at best allow for a similar degree of misrepresentation and manipulation as elections, and at worst an even greater degree given party ideology tends to act as a relatively stable indicator of the overall direction of the policies advocated taken as a whole, whereas such guides may be lacking when it comes to a single issue. Of course, as advocates of these initiatives note, the issue concerned may be one that divides parties, with support and opposition cutting across standard party lines (Glencross, 2021: 57–8). A referendum in such circumstances may help overcome such divisions. Yet, it could also be seen as an abnegation of responsibility by politicians to take a difficult decision. After all, free votes in legislatures can often be among the most deliberative precisely because politicians do not simply follow the party lead. By the same token, though, the fact that politicians have no clear mandate for how they would vote could be seen as legitimising a popular debate and decision.

Nevertheless, so far as policy issues are concerned, the disadvantages of referendums may outweigh the advantages. As was noted in the last section, the need to fish for votes from a diverse electorate and build a coalition for a party programme helps support minority interests. But when it comes to a vote on a single issue the need for coalition building is likely to be less. That increases the danger of majority tyranny, since the need for compromise and accommodation is greatly reduced, as is the capacity, within the procedure, to develop a compromise (Bellamy, 2018; Haider-Markel et al., 2007). Meanwhile, policies that are so decided tend to become entrenched – politicians will be wary of reversing a measure possessing a popular mandate, even if they regard the policy as deleterious. Moreover, they will bear no personal responsibility for it and so cannot be held to account for its failings. Once again, the worry is that referendums may not only lack adequate instrumental constitutional qualities themselves but also undermine those associated with representative democracy.

These concerns arise when referendums are seen as exercises of direct democracy, involving the exercise of popular sovereignty. Some theorists have argued that while referendums with this quality may be inappropriate on policy issues, they are suitable for constitutive and constitutional issues (Tierney, 2012, 2013). As I noted above, referendums on these two issues concern respectively the definition of the polity and its demos; and the character of the regime – particularly the rules of the democratic game. Both these issues have been regarded as being suitably conceived as expressions of ‘we the people’ (Ackerman, 1991). In the first case, a referendum could be seen as providing a way for those concerned to contract to form a people who then subject themselves to a common authority under their mutual control. In the second case, a referendum allows the secondary rules that define the legal and political system to be authorised directly by the people to whom they will apply, encouraging them to identify with these rules as theirs.

Two additional claims are often made in regard to both issues (Tierney, 2012: 12–15; Tierney, 2013): first, that these are matters of collective and common interest, that transcend the particular interests individuals may have with regard to specific policies; and second, that referendums spark a more participatory and deliberative debate of a kind that promotes public reasoning on a topic that defines both the public or demos and their public interactions. On this account, constitutive and constitutional referendums form part of a dual democracy (Ackerman, 1991): they provide a democratic mechanism for bringing into being a democratic system that can address policy issues. Yet, it is disputable whether they can escape the problems identified above with regard to policy referendums. To do so, it must be assumed that the referendum does indeed take the form of a contract whereby all citizens agree on fair terms of cooperation and consent to them, with constitutional and constitutive politics differing in deliberative and epistemological quality to normal politics, such that citizens are moved by the better argument and avoid the conflicts of interest and ontological disagreements typical of policy debates (Habermas, 1996: 278–79). However, there is no compelling reason to believe this will be the case (Bellamy and Schönlau, 2004).

If, as political constitutionalists argue, rights and procedural rules can be matters of reasonable disagreement, then no consensus is likely to be reached on them through a public wide discussion and referendum any more than in a legislative debate. Nor are such discussions necessarily any more bottom up and free from elite manipulation, given that as with policy proposals they are likely to be prompted by political entrepreneurs and the debate heavily influenced by the media. Meanwhile, such a discussion may lack the incentives of checks and balances that lead to legislators ‘hearing the other side’. Indeed, as the 2008 Swiss referendum on minarets, recurrent Swiss referendums on immigration and residence of foreign criminals, and California Proposition 8 illustrate (Frey and Goette, 1998; Hainmueller and Hangartner, 2019; Moeckli, 2011), such exercises of popular democracy may allow minority rights to be abrogated by tyrannous majorities. Some such checks might be incorporated into the referendum process – for example, it might be necessary to have a threshold for the turnout of over 50% and/or to achieve a super majority if the measure is to pass. However, while these requirements may prevent certain unjust proposals carrying they can also inhibit reforms that remove injustices from the prevailing status quo (Schwartzberg, 2013). They can only be justified if it is reasonable to assume that most changes would be likely to be worse than the prevailing system – hardly an endorsement of popular sovereignty. Arguably, a comprehensive constitutional reform might allow for more compromise to arise of a balancing kind between the representatives of different groups, and – as with the Icelandic constitutive assembly – there could be a direct crowd sourcing of ideas by the public, with the whole package then put to a referendum. However, this proposal seems to suggest that constitutive and constitutional referendums become more acceptable the more they develop the characteristics of representative, competitive party politics.

Although there is some evidence that Dicey may have come to regard referendums in this light (Weill, 2003), the development of his argument seems to have been driven by the desire to stop Irish Home Rule at any price (Cosgrove, 1980: 105–10, 247). However, that need not mean that referendums could not be incorporated into a Diceyan version of political constitutionalism where political sovereignty lies with the people but legal sovereignty rests with parliament, which retains both constitutional and legislative authority (Weill, 2003: 474–5). Dicey's point – at least until 1911 – seems to have been that an executive required a significant level of popular political authorisation for major constitutional change. However, as with the Great Reform Act of 1832, he considered – at least initially – an election victory in which the change figured prominently in the campaign as sufficient (Weill, 2003: 478–9). He only turned to advocating a referendum as Home Rule appeared unlikely to be opposed. The logic of his argument was not to consider the people as possessing legislative sovereignty, though, as some have argued. Rather, he saw referendums as a means for exercising a popular political constraint on the legislature in the case of significant constitutive or constitutional laws. This constraint paralleled – and could even be exercised by – the withdrawal of popular authorisation of a government at an election (Weill, 2003: 480–88).

In a similar manner, Eoin Daly (2015) has argued referendums may be considered as forming a part of the mixed constitution, adding another political channel for checks and balances in areas that legislatures may be ill-suited to provide for themselves (Daly, 2015: 32). Rather than being bottom up and positive exercises of popular legislative sovereignty, and as such as rivals to parliamentary sovereignty, he contends that referendums are better seen as negative and top down, serving a contestatory function aimed at guarding against arbitrary rule and promoting public justification (Daly, 2015: 37). As such, they can contribute to the negative and positive instrumental constitutional functions identified above, usefully supplementing representative democracy in the process.

Daly considers referendums of this kind as particularly relevant for constitutive and constitutional issues. Both these issues could be regarded as relating to areas where politicians might have an interest at variance with those they represent. For example, a government might wish to introduce electoral reforms or constitutional amendments that make it more likely they remain in power or the policies they favour are entrenched. He contends that forcing governments to put proposals on these two sorts of issues to a referendum can promote a wider public discussion and promote civic participation. As he notes, in Europe Ireland, Switzerland and Denmark all have mandatory constitutional amendment referendums for issues of this kind (Daly, 2015: 38), as does Australia (Daly, 2015: 41). Even in Germany, which had been thought averse to referendums given their use by the Nazis, the Basic Law has been amended to allow for a referendum on changes to the Basic Law itself (A146) (Lord, 2021: 31). Their aim is to reinforce the likelihood such measures treat all with equal respect and concern by giving all citizens an equal vote on them while having the proposal itself and its implementation the product of coalition building and compromise by parties in the legislature, whose actions remain electorally authorised and accountable. He distinguishes referendums that still retain parliamentary legislative sovereignty from those, such as De Gaulle's 1962 referendum amending the rules on Presidential elections, that involve a direct populist appeal by a political leader to the ‘people's will’ in order to by-pass parliamentary amendment procedures (Daly, 2015: 36–37). Daly likens referendums of the former type to the role of a second chamber that adopts a different form of representation to the first and primary legislative chamber. As a form of multicameralism (Daly, 2015: 43), referendums both support the representation of all relevant interests on such important questions, and promotes their balancing through enhancing the scrutiny to which they are subjected.

Daly's argument presents an account of referendums that can be reconciled with political constitutionalism in so far as it distinguishes political and legal sovereignty, leaving the latter with Parliament. So conceived, referendums introduce a check and balance that possess the intrinsic and instrumental constitutional qualities of democracy that political constitutionalists seek to deploy, at least when they are embedded within a system of representative democracy and act as a supplement to it. However, his account gives referendums a largely negative, contestatory, role. They operate not to propose change so much as to approve or reject changes proposed by the legislature. As such, they only partly address the alleged failings of representative democracy. They operate against elites when they are acting against the public interest, but not when they are failing to act in the public interest. Yet, though any given individual act of omission may pose less of a danger than any single act of commission, taken cumulatively over time they may do as much, if not more, to undermine popular support for democracy.

Even in plurality systems such as the UK, though, electoral competition may work against parties successfully colluding to take certain constitutional issues completely off the agenda. For example, although the two main parties have generally opposed any move towards proportional representation for national elections, the hung parliament of 2010 gave the Liberal Democrat's the opportunity to push for electoral reform. Given the type of issue and the fact that no parliamentary majority probably existed for this measure, putting a proposal to a referendum could be regarded as a legitimate alternative to canvas the support of the broader public for such a fundamental reform. Likewise, the referendums on Northern Ireland remaining part of the UK (1973) and the Good Friday Agreement (1998) and the Scottish (1979, 1997) and Welsh (1979. 1997, 2011) referendums on devolution, and in the case of Scotland independence (2014), all deal with constitutional issues where it is conceivable no majority would exist at Westminster for the change, and those voters with a strong interest in the issue lacked sufficient representation. Even though not all of these detailed the precise form the proposal would take, they can be seen as a means for checking against potential inaction by politicians on a core constitutive or constitutional issue through canvassing the wider and more diverse chamber of the relevant public as a whole. It would be misleading to regard them as operating as expressions of popular sovereignty or consent.

As I have noted, international treaties can raise both constitutive and constitutional issues, and as such seem apt as an issue for referendums. However, Article 75 of the Italian constitution, that allows abrogative referendums, explicitly excludes referendums on ‘laws authorising the ratification of international treaties’ along with those on ‘tax, budget, amnesty or pardon laws’. This exclusion has been justified as necessary to ensure credible commitments to international agreements regardless of changes in the popular mood. However, many treaties contain provisions allowing one or more of the parties to it to ‘denounce’ it in specific circumstances and to terminate or suspend participation in it. Indeed, 216 of the 561 international organisations created between 1815 and 2006 have ceased to exist (Eilstrup-Sangiovanni, 2020). If domestic constitutive and constitutional questions are valid subjects for referendums, then international agreements that raise similar questions should be too. After all, international agreements no less than domestic arrangements may either be negotiated by politicians who fail to adequately represent the concerns of citizens or cease to match changing circumstances that politicians with a vested interest in their continuance collude in keeping off the electoral agenda. In that respect, these agreements differ from the other exclusions of Article 75, involving policy issues where the citizens themselves might be regarded as acting as ‘judge in their own cause’ and which are far more likely to be subject to electoral competition and change. The exception might be when these issues, most notably tax and budget matters, are themselves entrenched within international agreements. But then it is precisely the inappropriate constitutionalising of a policy issue that renders it justified to allow its challenge in a referendum (Bellamy and Weale, 2015). It remains to be seen whether the Brexit referendum can be justified on these grounds.

The Brexit Referendum and Political Constitutionalism

The Brexit referendum has been subjected to a high degree of criticism, especially, but not solely, by those favouring remaining in the EU. These criticisms have ranged from concerns about the misleading character of the Leave campaign, including financial irregularities, misinformation and the use of algorithms for media advertising that derived from dubiously obtained personal information (Hansson and Kröger, 2021), to dissatisfaction with the franchise, especially for its failure to include resident EU nationals or to reduce the voting age to 16 (Grayling, 2017: 189–97; Offe, 2017: 18–21). The evaluation developed here focuses on the second set of criticisms relating to the referendum's institutional design and constitutional credentials rather than the first set relating to its conduct. However, neither the referendum's conduct nor design needs to be perfect for it to be legitimate. Though both were flawed in various ways, it was arguably no worse in these respects than current general elections in the UK or most other established democracies, such as the United States. These too rightly attract criticism for their many shortcomings. For example, similar campaign techniques to those of the Brexit referendum feature in normal elections in these countries, while a case exists in both the UK and USA for allowing permanent residents to vote, lowering the age of the franchise, and changing the electoral system to some form of PR. Yet, though undoubtedly improvable, both systems are regarded as possessing sufficient democratic qualities to be legitimate processes. The Brexit referendum arguably met a similar threshold of legitimacy, despite its manifold faults (Bellamy, 2019a; Electoral Commission, 2016). Granting this to be the case, my concern is whether it could be regarded as compatible with political constitutionalism along the lines given in the last section.

Note, if it can, that would offer a different ground for the referendum's legitimacy than was given by not only many Leave supporters but also others, such as the former Prime Minister Theresa May, who supported Remain but felt obliged to honour the referendum's result: namely, that it was an expression of the people's will, with popular sovereignty trumping parliamentary sovereignty (Weale, 2018: ix-x). It also provides a response to those who considered the referendum as unconstitutional – either in principle, or due to the absence of its procedures forming part of a legal constitution (Grayling, 2017: 189–97). Instead, it could be seen not as a form of direct democracy requiring constitutional regulation but rather as a constitutional form of democratic politics as part of a mixed political constitutional system.

To fit this role, a referendum needs to be both embedded within and regulated via the system of representative democracy, and as such subject to the legal authority of parliamentary sovereignty, while offering a justified check and balance upon it issuing from the political authority of the people taken as a whole. The argument of the previous section suggested this is most likely to be the case in referendums on constitutive and constitutional issues, or those on international agreements that combine one or other or both, which are top down and related to affirming or contesting a proposal from the executive, and that can only take place when legislated for by a duly elected legislature, which retains the responsibility for acting on the result as its members think fit. How far was this true of the Brexit referendum?

The decision to call the referendum resulted from a manifesto pledge of the Conservative Party in the 2015 election. That origin has been viewed as somehow tainted – a hypocritical move to blunt the electoral threat posed by the UK Independence Party (UKIP) to the Conservatives. Yet, it could equally be seen as a mechanism for giving voice to an issue that the major parties had largely kept off the electoral agenda. True, the Conservatives had tried to make the EU an electoral issue in the 2001 and 2005 elections without success, a point to which I return below. However, to the extent that EU membership forms a constitutive issue as to the very nature of the polity, then it can be regarded as rightly a matter on which the people as a whole should be consulted. Indeed, across the EU that has steadily become the norm, including in Germany where the Constitutional Court (2009: paras 179 & 263) has recently argued that the development of a ‘federal state’ at the European level could reach a point that would require ‘a decision of the German people beyond the present application of the Basic Law’. If a referendum is increasingly seen as necessary for not only joining the EU but also for any significant deepening of European integration, then the same should hold for a decision to leave the EU altogether. Making that possibility a manifesto pledge within an election can be regarded as providing a democratic test of its significance for the electorate. Meanwhile, doing so to fend off a challenge from a small party with significant but diffuse support (UKIP has only won one parliamentary seat in an election, which was held by a popular MP who had defected from the Conservative Party), can be regarded as providing a form of balance within a plurality system for views that might otherwise go unheard.

I also noted how referendums are likely to most fit with political constitutionalism to the extent they are regulated by parliamentary legislation. Here too that was the case. Legislation governing referendums in general had been passed under the Political Parties, Elections and Referendums Act 2000, and this was specifically applied to the Brexit Referendum in the European Referendum Act 2015. This Act was passed with an overwhelming majority in the Commons of 544 to 53 in favour, with only the Scottish National Party voting against.

As I remarked above, the main complaints with this legislation related to the franchise, the lack of supermajority rules, the length and rules for the conduct of the campaign and the alleged vagueness of the question posed. The relevant question to ask here is whether the proposed changes could be justified as necessary for the referendum to perform its checking and balancing role. My claim will be that at best they were unnecessary for the constitutional legitimacy of the referendum and at worst they would have been illegitimate.

With regard to the franchise, the referendum adopted that used for national elections under the provisions of the Representation of the People Act 1983 and 2000. Although those provisions are contestable, they are largely consistent with those found in other established democracies. A related but distinct issue is whether a reasonable case can be made for keeping to those provisions, or whether not to depart from them in this case involved an unequivocal injustice. Here it is useful to distinguish between who can be legitimately included in making a decision and the side constraints that might arise with regard to those affected by a decision. In the case of a constitutive referendum, which concerns the very shape of the demos, a reasonable case can be made for restricting the decision to citizens. On the stakeholder account of political rights and obligations (Bauböck, 2015), the claim to being a citizen of a given sovereign political community belongs to those whose freedom and rights are inherently linked to the collective self-government and flourishing of this polity over time. This argument links citizenship rights to the performance of civic and social duties and a commitment to the political community and its members, including to future generations. On this view, rightful inclusion in the demos depends on assuming the obligations entailed by long-term membership of a community, not least by naturalising as a citizen of the host country – something that was relatively easy for long-term resident EU citizens to do. This view also seems the most consistent with political constitutionalism, which concerns the ability for citizens to decide the rules governing their social co-existence on an equitable basis. It would be inapposite to include in such a process those who are not committed to living under those rules over time. That said, EU citizens are certainly affected by the decision, and for any Brexit deal to be legitimate it should also be constrained in ways that acknowledge their legitimate expectations regarding their rights to remain – as the EU negotiators justifiably insisted.

What about the argument that the franchise ought to have included 16-year olds? Critics of this decision argue that it excluded those most directly affected by the proposal. They note that while those aged over 65 many of whom may never have to live with the consequences of their vote, divided 66–34 pro-Leave, the excluded younger generation, whose future will be profoundly shaped by Brexit, were 72–27 pro-Remain (Clarke et al., 2017: 155). Again, these are important objections. Cohort injustices are possible, and some have even suggested disenfranchising the elderly from voting on certain issues (Van Parijs, 1998). Still, the elderly tend to have children and grandchildren, so are not entirely without any concern for securing the interests of future generations, while any economic downturn as a result of Brexit would also have short-term and direct effects on their pensions and access to health care.

Super-majority thresholds are often required for constitutional and constitutive referendums. For example, the 1979 referendums on devolution in Scotland and Wales required the approval of at least 40% of the eligible electorate rather than just a majority of those who voted, a threshold they failed to meet. Neither did the Brexit referendum, where the majority of 51.89% on a participation rate of 72.21% represented just 37.4% of the electorate – although it was achieved by the 1975 referendum, which had a lower turn out (64.62%) but a higher majority in favour (67.23%). The argument favouring such criteria is that the (constitutional or quasi-constitutional) infrastructure of a polity should be relatively stable, particularly in those respects where it matters greatly for some members (perhaps a minority) in the society: for example, they have built their lives around the assumption that it will remain in place. Putting part of that infrastructure up for a 50–50 vote is like playing dice. This may be because there will always be some who vote against the status quo on the basis of frustration with the government on any front or because those affected will not realise, until it is too late, how badly they will be affected. For any such reason there will always be noise in the system, so to speak. This is an important consideration. However, a super-majority is an illegitimate way to address this point. For a super-majority violates a basic democratic and liberal norm that decision-making processes should be impartial and neutral between views in order to be fair (May, 1952). Super-majorities entrench the status quo and may consequently lock in inefficient or unequal measures that generate injustices (Schwartzberg, 2013). Instead, the issue was addressed in part by the referendum being advisory to Parliament, which subsequently had to debate and implement relevant legislation. In that sense, there was an opportunity for Parliament to deliberate the implications of the decision, and for it to be reconsidered by the electorate – a point I return to below.

A slightly different version of the supermajority argument is whether the vote should have obtained a majority among – or even in all of – the different ‘nations’ of the UK. The devolved regions argued that the UK included at least four demoi: those of Scotland, Wales and Northern Ireland, as well as England. Indeed, a case could even be made for suggesting that London, with a population almost as large as that of Scotland and Wales combined and devolved powers of its own, represented a fifth demos. Of these five, only two: Wales and England (minus London), voted to leave the EU. This requirement could be conceived as an additional political check and balance necessary in multinational political systems. However, it is not at present one that exists within the UK for non-devolved matters. Rather, power is devolved downwards from Westminster to these authorities for the purpose of managing domestic issues. The UK at present is not formally a federation and even if it has federal like features, federative powers with regard to the sovereignty of the UK state, such as the ability to contract obligations under international law, remain at Westminster, for which purpose the UK operates as a single nation – much as the EU represents all the member states in those areas where competences have been conferred upon it, notably the negotiation of trade issues within the WTO. Nor is this an unusual arrangement – it is true of federations such as the USA and Germany, for example. It reflects the fact that the federal parts ultimately form a unitary state, which in the area of external affairs has to be able to act as a single unit, for example on issues such as defence. The two parts of the UK most likely to leave – Scotland and Northern Ireland – had had respectively referendums on independence (2014) and joining the Republic of Ireland (1973), with the referendum on the Good Friday agreement (1998) also, in part, a referendum on the conditions under which Northern Ireland might leave the UK and join the Republic – and although Brexit may prompt them to be repeated sooner rather than later, at the time of the EU referendum a reasonable case existed for saying that for these purposes the UK formed a single demos.

With regard to the question posed, unlike the referendum on electoral reform of 2011 the electorate was not asked to accept or reject a specific constitutional proposal, which would then automatically become legislation. Formally speaking, as with the other UK referendums hitherto, it was a consultative referendum and the EU Referendum Act did not include any requirement that the government implement the decision. As such, it was consistent with parliamentary legislative sovereignty. Instead, the question was ‘Should the United Kingdom remain a member of the European Union or leave the European Union?’ However, this could be seen as a referendum on whether to accept or reject David Cameron's, the then Prime Minister's, renegotiation of the terms of Britain's EU membership. In other words, it offered a check on the prevailing view of the government and in the legislature more generally – and, indeed, was the first – and so far the only – time that view has been rejected in a referendum. The issue then turned to what terms of leaving the EU should follow from that decision, and should the electorate have been consulted via a second referendum on them?

The argument against a repeat referendum tended to take the form that these should be a once in a generation exercise. Adapting a parallel argument to Thomas Jefferson's for a periodical ratification of the terms of the US Constitution to ensure they still reflected ‘we the people’ (Jefferson, 1789), the argument goes that repeat referendums should only occur once the generation that decided the original referendum no longer form a majority. The reasoning is that with constitutive and constitutional referendums, which concern the basic political and legal framework, there is a need for continuity of a kind associated with the rule of law, whereby individuals can plan ahead. That criterion was met in 2016 with regard to the first EU referendum of 1975, but would require at least a decade for a repeat of the 2016 referendum to be legitimate. However, that objection does not really address two different arguments for a second referendum: one addressing the case for a super majority in a way more consistent with democratic norms, and that fits with the checks and balances rationale given for them here, the other raised by the need for a ratification of the terms of the withdrawal from the EU.

With regard to the first argument, the case for a second referendum might proceed as follows. In his novel The Life and Opinions of Tristram Shandy Gentleman (1762: Vol VI, Ch. XVII), Laurence Sterne has Shandy, the novel's narrator, report, misremembering Herodotus on Persian customs, that the ancient Goths of Germany took major decisions twice: first drunk and second sober, with only those drunken resolutions that met with sober approval being acted on. Shandy remarks that his father was much taken with this argument but being teetotal adapted it to involve discussing important domestic matters with his wife on the first Saturday night of the month and then on the following Sunday morning, referring to these two deliberative occasions as his ‘beds of justice’ (Waldron, 2016: 76, 327, n. 29). Adapting this argument, it can be maintained that, for the reasons given earlier, decisions involving changing the very rules of the political game deserve to be considered especially carefully. At the same time, such consideration ought to respect democratic norms and allow all views to be fairly weighed and expressed. Balancing passionate frustration with the possible inadequacies of the status quo against reflective consideration of the consequences of changing it, potentially for an even worse arrangement, reflects a well-established deliberative norm of ‘hearing the other side’ that informs adversarial debate in Parliament and the Courts as well as normal elections. Indeed, Sterne's ‘two bed’ argument has been employed to justify bicameralism, in which the second chamber acts as a scrutiny chamber, and could be equally employed to allow for judicial review (Waldron, 2016: 77). This argument responds to the common criticisms of the first referendum while reflecting general considerations that derive from the reasoning underlying political constitutionalism rather than being ad hoc arguments that reflect one's view of the issue the referendum sought to decide. Unfortunately, the early triggering of Article 50 rendered it impossible to hold a second referendum before time ran out. However, arguably doing so for this reason would have been appropriate from a political constitutionalist perspective, and not to do so was a missed opportunity.

The second argument, that a vote should have been held on the specific terms of withdrawal, rather than the general issue of whether to withdraw at all, is more problematic. This moves the decision from a strictly constitutive and constitutional issue to something more like a policy issue, where we saw referendums prove more problematic. Parliament – arguably like the public as a whole – was divided between a) remaining in the EU, b) leaving and staying in the single market (roughly the position of Norway and countries within the European Economic Area) or customs union, c) leaving and not staying in the single market and customs union (a version of which would have been Theresa May's amended Chequers deal or the withdrawal agreement ultimately negotiated by Boris Johnson), and d) leaving without a deal (and trading according to World Trade Organisation rules). One solution, proposed by the then Conservative MP Justine Greening, would have been to offer voters these choices and accepted either the plurality winner or, using a Borda or Condorcet count, the most preferred of these options. However, that would still not have necessarily weighed the impact of these options against the full range of other policies of importance to voters. For this purpose, a general election in which voters have an opportunity to see how these options fit into a programme of government offers a better solution. As it happens voters were given such a choice, with elections in both 2017 and 2019. The latter election, which the government fought on the slogan ‘Get Brexit Done’, gave Boris Johnson's Conservative Party a landslide victory, with the party winning 43.6% of the vote – the highest percentage obtained by a party since 1979, and a parliamentary majority of 80 seats. Arguably, therefore, the second argument was met.

Of course, criticisms continue about the conduct of both the election and referendum campaigns and especially the misinformation and failure to discuss certain key issues, notably Northern Ireland. Yet, these failings are arguably down as much to the failure of the opposition parties as the process itself. After all, the referendum and election were overseen by an independent regulator, the Electoral Commission, which in its report argued that no procedural failures were serious enough to invalidate the result. In other words, while flawed and improvable, the process cannot be regarded as so defective as to justify being annulled. Indeed, recent opinion polls suggest that even with the many problems attendant on leaving the EU already having become apparent, a re-run of the referendum at the time of writing would be unlikely to produce a different result.

Conclusion

Political constitutionalism offers a system of checks and balances for resolving disagreements in a fair manner, which forces citizens and politicians to ‘hear the other side’. This article has argued that referendums can form a justifiable part of such a system, especially with regard to constitutive and constitutional issues where representatives may act as ‘judges in their own cause’ and ‘international agreements that raise similar concerns. However, such referendums remain subject to the sovereignty of parliament in being regulated by parliamentary legislation and subject to approval by parliament – a check that also operates against their being used as executive led plebiscites. It was been contended that such referendums ought to be restricted to providing a potential negative check on government proposals on such questions (Daly, 2015; Trueblood, 2022). However, a failure to consider reforms may be as legitimate to address as a proposal for reform. Consider how the two main British parties have kept electoral reform off the agenda. An open referendum can be a way of forcing parliament to deliberate on proposals for constitutional or constitutive reform, and then to put either a consensual proposal or a range of proposals to a second referendum, as I suggested should have happened with regard to Brexit and the withdrawal agreement.

I have claimed that the 2016 Brexit referendum can be viewed as consistent with political constitutionalism and met a reasonable threshold of democratic legitimacy. That does not mean that we should accept Brexit as reflecting the will of the British people. Rather, we should accept that the plurality of citizens who voted for this result made their case in a fair way that was suitably checked and balanced. Those of us who supported Remain should also acknowledge the failures of our campaign, which relied less on the positive case for staying in the EU and emphasized instead the negatives of Leaving – justified as these may well have been. That weakness was compounded by a failure of those opposed to the ‘hard’ Brexit proposed by May and particularly Johnson to cooperate to defeat it either in parliament or the subsequent election. One reason for embedding referendums within representative democracy is to allow representatives to negotiate electorally attractive policy packages capable of obtaining the approval of the electorate. The Remain camp failed to do that – although, unlike the 2022 Chilean referendum where a similar failure resulted from the large number of independents within the constitutional convention (Couso, 2022; Verdugo, 2022b), this failure was less a matter of institutional design than of the politicians themselves.

All of which is to recognize that institutional design cannot – and arguably should not – guarantee those outcomes one thinks best (Van Crombrugge, 2021: 122–23). Indeed, liberals consider that individuals have the right to make mistakes, hoping they – or at least the rest of us – may learn from them. Those who consider leaving the EU to be such a mistake, as I do (Bellamy, 2019b: xiii–xv), can continue to try and convince others of their view, campaign for better terms with the EU (Weale, 2017), and ultimately – ten years or more, as per the Jeffersonian principle discussed earlier – seek a further referendum on rejoining.