Schmitt, Dicey, and the power and limits of referendums in the United Kingdom

Abstract Carl Schmitt and AV Dicey are two of history's most influential constitutional theorists, and they offer two of history's most influential accounts of referendums. In most respects, their approaches to referendums are in direct opposition to each other. On Schmitt's view, the purpose of referendums is to acclaim executive actors. On Dicey's view, the role of referendums is to constrain them. Despite disagreeing about whether referendums should acclaim or constrain the executive, Schmitt and Dicey agree that an agenda-setting role for representatives in referendums is inevitable. This paper argues that, in the UK context, if Schmitt and Dicey are right about the necessary agenda-setting power of representatives in referendums, then the accounts of referendums they each offer must be two sides of the same coin. Given the dominance of the executive over the legislature in the UK and the uncodified nature of the constitution, referendums are processes that necessarily both acclaim and limit the executive.


Introduction
Carl Schmitt and AV Dicey are two of history's most influential constitutional theorists, and they offer two of history's most influential accounts of referendums. 1 In most respects, their approaches to referendums are in direct opposition to each other. On Schmitt's view, the purpose of referendums is to acclaim the executive. On Dicey's view, referendums constrain the executive. Despite disagreeing about whether referendums should acclaim or limit the executive, Schmitt and Dicey agree that an agendasetting role for representatives in referendums is inevitable. This paper argues that, in the UK context, if Schmitt and Dicey are right about the necessary agenda-setting power of representatives in referendums, then the accounts of referendums they each offer must be two sides of the same coin. Given the dominance of the executive over the legislature in the UK, referendums are processes that necessarily both acclaim and limit the executive. The current state of referendums in the UK creates opportunities for referendums to constrain the executive; but also runs the risk that even wellmeaning governments may use them as tools of power consolidation. This paper has four parts. It begins by explaining Dicey's negative account of referendums as limiting the executive. Second, it turns to Schmitt's positive account of referendums as processes of executive acclamation. The third part explains the surprising similarities between Schmitt and Dicey's accounts and why their theories each necessarily offer one side of the coin in thinking about referendums in the UK context, where (a) the executive dominates the legislature and (b) the use of referendums is (largely) uncodified. Suppose Schmitt and Dicey are right that representatives play an agenda-setting role in referendums, and the executive are the dominant representatives in the UK. In that case, referendums are necessarily processes that both acclaim and limit the executive. This dual relationship between acclamation and limitation holds for three further conception reasons, the nature of (i) voting, (ii) decision-making, and (iii) authority and power. The fourth part of the paper demonstrates the dangers of power consolidation through referendums and argues that the UK should codify the use of referendums. The aim of the codification proposals put forward here is to build on Dicey's approach. Their objective is to achieve a state where, so far as possible, referendums are constraining the executive. Codification also helps ensure that referendums capture broad public support for propositions. While codification cannot eliminate the abuses of referendums entirely, it can help ensure that referendums are used in a clear, consistent way. It is necessary to concede, though, that the circumstances Schmitt identifies are inescapable. There will always be a tactical element to the use of referendums, just as there are with other democratic processes such as elections. Political actors will inevitably use processes such as referendums to achieve their ends. Given that the executive already tends to dominate the legislature in the UK, and the propensity for referendums globally to be tools by which the executive circumvents legislatures, the challenges of codification must be confronted directly.

Dicey: referendums as limitation
As the House of Lords' power, the traditional check on the House of Commons, was losing its force, 2 Dicey was concerned about the consolidation of power in the executive through political parties. Dicey argued that the UK should introduce the referendum to prevent the government from instructing a majority in Parliament to push through changes 3 that '[did] not command the sanction of electors '. 4 Indeed, Dicey thought the referendum was the one available check on party leaders. The context in which Dicey was writing is, in many ways, recognisable today. He says that '[over] forty years parliamentary government has suffered an extraordinary decline or, as some would say, a temporary eclipse'. 5 The consolidation of power in parties was happening in the UK, Dicey thought, and worldwide. 6 Dicey attributed this global democratic crisis to the rise of party government. 7 Dicey's concern, which persists today, is Lord Hailsham's idea of 'elective dictatorship'. 8 Elective dictatorship means 2 With the rise of mass party democracy, the House of Lords was losing its legitimacy as a check on the House of Commons. This shift was institutionalised by the passing of the Parliament Acts 1911 and 1949, which limit the capacity of the House of Lords to block legislation from the House of Commons. 3 The constitutional issue that dominated Dicey's thinking about this was Home Rule for Ireland: M Qvortrup 'AV Dicey: the referendum as the people's veto ' (1999)  that the House of Commons ceases to check on an executive with a Parliamentary majority. Instead, the House of Commons has the purpose of giving effect to the prevailing party's policies. To address this problem, Dicey made a case for the referendum as a 'people's veto', 9 an 'alternative second chamber'. 10 On Dicey's view, the strength of the referendum lies not only in the fact that it was a veto, and a veto of a certain kind. The referendum, for Dicey, is both a democratic and conservative institution. 11 He explains what it means for referendums to be both democratic and conservative, saying it is democratic because of its popular appeal, but conservative since 'it ensures the maintenance of any law or institutions which the majority of electors effectively wish to preserve'. 12 Furthermore, it is a majoritarian 13 veto, but not a veto in the service of homogenisation in the way that, as will be seen shortly, Schmitt's view requires. The aim of Dicey's use of majoritarianism is the control of party government and maintenance of the status quo; it is not the elimination of heterogeneity or difference, as it is for Schmitt. 14 For this, and a range of other reasons, it is best to read Dicey as the exemplar of offering a case for referendums in the liberal democratic tradition. 15 That is all to say: on Dicey's view, the role of referendums is to protect the status quo from unbounded party majorities in Parliament. Dicey was not arguing for referendums' use on a participatory basis, and he was not arguing that voting in referendums is intrinsically educative or valuable. 16 Nor was he arguing that referendums are epistemically better ways of making decisions; he accepts that voters will make mistakes. 17 He concedes that the referendum has the weaknesses as well as the strengths of a veto. 18 He is not arguing that referendums are perfect. What he does think, however, is that the referendum's 'strongest recommendation is that it may keep in check the inordinate power now bestowed on the party machine'. 19 On Dicey's view, the referendum was a mechanism that could stop this and do so in a way that was consistent with the doctrine of popular sovereignty.
With the House of Lords' declining role, and the fusion of political parties with government, Dicey thought the UK required a change to maintain the integrity of Parliamentary democracy. Parliamentary democracy needed to be protected against the influence of political parties in a way that was consistent with the doctrine of popular sovereignty. For Dicey, that protective democratic process was the referendum. This paper will argue, though, that while Dicey is correct to say that referendums can play a role in constraining the executive, they alsonecessarilyplay a role in consolidating executive power. The use of referendums in the UK has exacerbated precisely the problems that Dicey sought to address. 9 AV Dicey 'Ought the referendum to be introduced in England?' (1866) 57 The Contemporary Review 489. 10 Qvortrup, above n 3, at 542. 11 Dicey, above n 4, p cvx. 14 However, it is essential to recognise that in advocating for a referendum, Dicey was trying to make the process of Home Rule impossible through a different form of homogenisation. A referendum across the entire UK would leave the Irish as a minority vastly outvoted by the majority.

15
'Dicey [deserves] a place among the great liberal thinkers': Qvortrup, above n 3, at 546. 16 'This argument forms an important supplement to the argument for the referendum as a veto, an entrenching device. For, as we have seen, the case for the veto is broadly a conservative one, based upon a distrust of the efficacy of representative institutions and, in particular, of political parties. But the case for the referendum based upon its educative nature and the need to encourage participation relies upon arguments of an entirely different type': V Bogdanor The People and the Party System (Cambridge: Cambridge University Press, 1981) p 85. 17 'It must, in short, be admitted that a veto on legislation, whether placed in the hands of the King, or in the hands of the Lords, or of the House of Commons, or of the 8,000,000 electors, would necessarily work sometimes well and sometimes ill': Dicey, above n 4, p cxiii.

Schmitt: referendums as acclamation
Schmitt's theory of plebiscitary democracy follows from his critique of liberalism. Liberals, Schmitt thinks, are trying to have their cake and eat it. 20 They try to base their theory of democracy on principles, such as the rule of law, which are universal and neutral. 21 Schmitt argues that these ideas are not neutral and that democracy is fundamentally about substance. 22 Liberalism's lack of willingness to engage with substance means that it will always postpone addressing issues by setting up 'commissions of inquiry and avoiding conflicts'. 23 As a consequence of this absence of clarity, decisions, compromises, and majorities in parliamentary democracies will only ever be transitory, and political commitment will be absent. In a time of increasing bureaucratisation, what Schmitt thought was required was explicit, decisive action on behalf of the people. Only a system of plebiscitary democracy acclaiming an authority can make this possible. 24 By contrast, Schmitt's system of plebiscitary democracy authorises 'the government or an authoritative organ' 25 to act clearly and authoritatively on the people's behalf. The contrast between parliamentary and plebiscitary democracy, Schmitt thinks, is between one that is 'anachronistic, transitory, dangerous and self-contradictory and one that is vibrant, democratic, efficient, and permanent'. 26 In addition to these structural failings, there is a more profound and irresolvable paradox between the legality required by parliamentary democracy and the legitimacy needed for democracy through expressions of popular sovereignty. 27 Schmitt argues that 'every democracy rests on the on the presupposition of the indivisibly similar, entire, unified people'. 28 Since this is so, Schmitt concludes, 'plebiscitary legitimacy is the single type of state justification that may be generally acknowledged as valid today… it is the last remaining accepted system of justification'. 29 Schmitt's theory of plebiscitary democracy follows logically from his understanding of the nation and constituent power. 30 For Schmitt, 'constitutional authority cannot escape its origins in constituent power'. 31 This means that constituent power cannot be constrained or controlled by the law. The constitution is legally valid because it is a decision of the people. 32 Schmitt's is a way of thinking that continues to have enormous resonance in the rhetoric of referendums, particularly the idea of 'vox populi vox dei', the voice of the people is the voice of God. 33 While Schmitt argues that referendums are expressions of legitimacy coming from the people, this does not mean that he thinks referendums are directly democratic. Schmitt does not believe that 'the people' themselves set the terms of referendums. He is clear that the people can only ever respond to questions asked of them. Schmitt says, 'the question can only be posed from above; the answer only comes from below'. 34  The question of who 'the people' are is, of course, a serious problem in legal and constitutional theory, not least because of the boundary problem: F Whelan 'Prologue: democratic theory and the boundary problem' (1983) 25 Nomos 13. This paper is not advocating for the use of the term 'people' in its analysis. In its positive suggestions in section 4, it explicitly uses the word 'voters' to recognise that the very definition of a, or the, people is contested. 34 Schmitt, above n 24, p 90. democracy. First, the dependence of the people in a referendum on the prior actions of authorities. It is also related to the trust that Schmitt places in authorities to use that power judiciously. Schmitt's view is based on the presupposition that executive organs will use this power properly. Schmitt says: 'Plebiscitary legitimacy requires a government or some other authoritarian organ in which one can have confidence that it will pose the correct question in the proper way and not misuse the great power that lies in the posing of the question'. 35 The abuse of the idea of a unified people expressing their will through referendums to acclaim dictator has become 'one of the chief hallmarks of fascism', 36 and one of the ways in which dictators of all stripes continue to abuse referendums today. 37 The relevance of Schmitt's work remains clear. Schmitt's theory of plebiscitary democracy helps make sense of the many contemporary cases of abuse of referendums, 38 particularly by those executive actors seeking to use them to consolidate power. 39 It is not only despots who use referendums to consolidate power, however. The following two sections demonstrate how referendums may be sources of power for the executive, even in healthy constitutional democracies.
3. Two defining features of referendums in the UK So far, this paper has contrasted Schmitt and Dicey's theories of referendums. It has shown how the two accounts of referendums appear to be opposites of each other: Schmitt makes a positive case for referendums as acclaiming the executive; Dicey, that referendums are negative ways of constraining the executive through vetoes. This section argues that, specifically in the UK context, Schmitt and Dicey both offer inseparable insights into the uses of referendums. Indeed, in the UK, the two accounts offered by Schmitt and Dicey are necessarily two sides of the same coin. These shared insights have salience in the UK because (a) the use of referendums is (largely) uncodified and the (b) dominance of the executive relative to the legislature. This paper will explain first the relevance of the UK context before outlining the three further conceptual reasons that acclamation and limitation must be two sides of the same coin: the nature of (a) decisions, (b) voting, and (c) the relationship between power and authority. The essential point is this: referendums will always provide a mandate to the executive in the UK. These mandates arise at once, both in the form of what the executive can do and what they cannot do.
(a) Referendums in the UK There are two defining features of the use of referendums in the UK context: the absence of codification and the dominance of the executive relative to the legislature. Take the matter of codification first. The use of referendums in the UK is governed mainly by the Political Parties, Elections and Referendums Act 2000 (PPERA). The PPERA is administrative, not substantive, legislation. The PPERA says very little about the role of referendums in the UK, although some actsspecifically the devolution actsstipulate circumstances in which referendums must be used. The Scotland Act 2016, Wales Act 2017, and the Northern Ireland Act 1998 all use referendums as methods of entrenchment for the devolution settlement. The existence of the Scottish and Welsh Parliaments is not to be abolished but for the consent of people voting in a referendum. However, the Acts provide no further procedural guidance about how these referendums are to work or how these referendums are to be 35 Ibid. 36 McCormick, above n 27, p xxxii. 37 Setälä says that referendums are 'handy tools for dictators in boosting their legitimacy, and were used to consolidate the powers of the Nazis, and among former communist countries in Eastern Europe': Setälä, above n 1, pp 1-2. initiated. The Northern Ireland Act is slightly different in that s 1(2) clarifies that if there is a majority vote in a referendum to no longer be a part of the UK, then there is a duty on the Secretary of State to bring forward proposals for a change in the relationship. While there would be extreme political consequences if the Westminster Parliament were to change these Acts unilaterally, it is legally very likely that they could lawfully do so. There was one other way it was clear by law that representatives must put a question to a referendum. The European Union Act 2011 previously required a referendum on any further European integration; however, this has now been repealed by the European Union Withdrawal Act 2018.
The PPERA is also-opened ended about the matters for which referendums must be used. While the House of Lords Constitution Committee concluded that referendums should be used for 'fundamental constitutional issues', the Committee itself was, understandably, reluctant to define a fundamental constitutional issue. 40 While it is doubtlessly right that the boundaries of a fundamental constitutional issue are contested, it does not follow that there are no clear-cut cases where referendums should be required. Additionally, the PPERA does not clarify how referendums are to be used, whether, for example, they are pre-or post-legislative. This was a core problem identified in the wake of the Brexit referendum. 41 Even the question of whether referendums are binding or advisory is a question left to the government of the day. The bottom line is this: the way a referendum works in the UK depends enormously on the specifics of the enabling legislation. For instance, for the Alternative Vote referendum, s 8 of the Parliamentary Constituencies and Voting Act 2011 outlined two possible paths for the Bill depending on the popular vote results. 42 If there was a majority of votes in the referendum for the proposed changes, then the Secretary of State was to introduce an order bringing the changes in legislation into force. If there was a majority of votes against the changes, then the Secretary of State was to bring an order repealing it. The enabling legislation for the 2016 Brexit referendum, by contrast (the European Union Referendum Act 2015), made no such provisions for the outcome of the vote. 43 40 The HLCC did put forward a list to provide examples of what could count as a fundamental constitutional issue while emphasising their qualification: 'This is not a definitive list of fundamental constitutional issues, nor is it intended to be'. The list included: 'to abolish the Monarchy; to leave the European Union; for any of the nations of the UK to secede from the Union; to abolish either House of Parliament; to change the electoral system for the House of Commons; to adopt a written constitution; and to change the UK's system of currency': House of Lords' Constitution Committee 12 th Report 'Referendums in the United Kingdom' (2009) p 49. 41 Recommendation 18 of the report holds that: 'Referendums should be held on proposals that are clear and immediately actionable. This means that, wherever possible, referendums should be held post-legislatively: the relevant Parliament or assembly should legislate in detail for the change, subject to the approval by voters in the referendum. Should the result favour the change, the provisions would then be implemented': 'Report of the Independent Commission on Referendums' (Constitution Unit, July 2018), https://www.ucl.ac.uk/constitution-unit/publications/tabs/unit-publications/182_-_Independent_Commission_on_Referendums, p 86.

42
'(1) The Minister must make an order bringing into force section 9, Schedule 10 and Part 1 of Schedule 12 ("the alternative vote provisions") if -(a) more votes are cast in the referendum in favour of the answer "Yes" than in favour of the answer "No", and (b) the draft of an Order in Council laid before Parliament under subsection (5A) of section 3 of the Parliamentary Constituencies Act 1986 (substituted by section 10(6) below has been submitted to Her Majesty in Council under section 4 of that Act.
(2) If more votes are not cast in the referendum in favour of the answer "Yes" than in favour of the answer "No", the Minister must make an order repealing the alternative vote provisions': Parliamentary Constituencies and Voting Act 2011, s 8. 43 The UK Supreme Court stressed this point in Miller where the majority held that: 'Both sides of the argument proceed on the basis that the referendum on membership of the EU was held under the European Union Referendum Act 2015 ("the 2015 Act"), which resulted in a vote to leave the EU, does not provide the answer. The Secretary of State's argument proceeds on the basis that the Crown has taken the decision under Article 50(1), accepting the result of the referendum. The Miller claimants argue that only Parliament can take that decision. Both the Secretary of State and the Miller claimants proceed on the basis that the referendum result was not itself a decision by the UK to withdraw from the EU, in accordance with the UK's constitutional requirements, and that the 2015 Act did not itself authorise notification under Article 50(2). In these circumstances, there is no issue before the court as to the legal effect of the referendum result. Nor is this an appropriate occasion on which to consider the implications for our constitutional law of the developing practice of holding referendums before embarking on major constitutional changes: a matter on which this court has heard no argument': R ( The absence of codification around referendums, particularly how and when they must be used, gives enormous power to the executive. The question of whether a referendum is required is essentially up to the government of the day. The government has enormous power over constitutional change processes in the UK, not just concerning referendums. 44 The question alone of what amounts to a fundamental constitutional issue gives enormous power to the government. This ambiguity alone would be risky, combined with the power that comes to dictate the process and outcome of referendums. This absence of codification runs directly into the risk of referendums as tools that the executive can manipulate. Indeed, the threat, or promise, to hold a referendum may be as politically useful as holding a referendum in the first place. 45 The threat to hold a referendum is by no means limited to the UK. 46 Nevertheless, the absence of much codification introduces an element of flexibility around referendums present in other constitutions. This changeability is not in itself necessarily problematic; the reason it is problematic is the dominance of the executive relative to the legislature. (b) The dominance of the executive in the UK One of the most common and compelling arguments against the use of referendums is that elites manipulate them. 47 Referendums may be abused by elites in all kinds of ways, for example: to bolster personal legitimacy; 48 for internal party management; 49 and to achieve political goals, broadly speaking. 50 Social scientific literature also shows that referendums tend, as a baseline, to lend power to the executive. 51 This is all to say: the worry about the elite-driven nature of referendums, and the potential for elite manipulation, are clear and well-understood. These risks are particularly acute in the absence of codification and where the executive is strong relative to the legislature.
In the UK, as explained above, the process for legislating on any issue is nearly identical to legislating on constitutional issues. 52 The difficulty is that the executive dominates the legislative process in the UK. As Norton argues, the legislature is fundamentally responsive. 53 This is not to overstate the case. Particularly in the first two decades of the twenty-first century, Norton argues, Parliament is 'now stronger than at any time since the development of mass-membership organised political parties in the 1860s'. 54 Parliament performs many essential functions; it is perhaps more effective in its role as a scrutineer of legislation. 55 There are plenty of counterexamples of instances where private members 44 House of Lords' Constitution Committee 15 th Report 'The Process of Constitutional Change' (July 2011) https://publications.parliament.uk/pa/ld201012/ldselect/ldconst/177/17702.htm (last accessed 11 November 2021). 45 Indeed, the absence of codification allows for referendums to be promised, for example, on the Lisbon Treaty, and then the promise retracted. This practice has given rise to legal challenges including R (on the Application of Wheeler) v Office of the Prime Minister (2008)  Norton argues that this has improved, particularly with the strengthening of the select committee system: above n 53, pp 169-171. bills have made an impact. 56 In Legislation in Westminster, Russell and Glover are effective in challenging the argument that Parliament is relatively weak. While the influences exercised by the different parts of Parliament are diffuse and piecemeal, they are nevertheless effective. 57 While Russell and Glover's study is persuasive and important, Dicey's core concernthe possibility of Lord Hailsham's elective dictatorshipremains valid. If a government is legislating on an issue of constitutional significance, it is likely essential for the government. If an issue is vital to the government, it is very likely, albeit not certain, that the Bill will be passed. 58 That is not to suggest that the chambers of Parliament and its select committees have no role whatsoever in shaping legislation with regard to referendums. The claim is only that the absence of codification gives governments a structural advantage in legislating for referendums on terms that suit them politically.
These two defining features of the use of referendums in the UKthe absence of codification, and the dominance of a government majority in the legislative processmean that referendums are exacerbating rather than addressing power consolidation. It is correct to say, though, that even in these circumstances, referendums are still acting as limits on the executive's actions. As the next section demonstrates, Schmitt and Dicey's insights show that the processes of limitation and acclamation are inseparable from each other.

The interconnectedness of acclamation and limitation
All referendums in the UK act as limits on the actions of the executive and as ways of acclaiming the executive. The duality of these two forces can be seen in Schmitt's discussion of Napoleon. Schmitt makes the compelling point that: 'For those seeking support, the appeal to the people will always lead to some loss of independence'. 59 This is the essential idea driving this paper. Referendums provide support for the executive to execute propositions, but this direction always involvesas Schmitt rightly sayssome loss of independence as well. Put differently: in instructing representatives in what to do through referendums, referendum outcomes also identify what the executive cannot do. In this way, each of Schmitt and Dicey's arguments is, at least in the UK context, one side of the coin. Referendums are always, at least to some extent, both negative and positive. 60 There are three other conceptual reasons that these negative and positive processes cannot be separated from each other in the context of the UK: (a) referendums as decisions; (b) the nature of voting; and (c) the relationship between authority and power.
(a) Referendums as decisions Whatever else referendums are, they are decisions. Referendums have outcomes. The purpose of holding a referendum is to conclude a particular debate (at least temporarily), even if the conclusion is to maintain the status quo. Further, decisions are not self-executing. Decisions require action by representatives, evenagainif the action is to maintain the status quo. The executory function of referendum outcomes may be directed to different actors. Referendums may task a legislature, executive, or ad hoc body with acting based on a particular outcome. Examples of such actions include instructing the legislature or executive to enter or leave a treaty or requiring a minister to bring a Bill before a legislature. The bottom line is that for the outcome of a referendum to have meaning, voters require representatives to take that action on their behalf. That is not an action they can take themselves. This is true even when the outcome is an affirmation of a status quo. Inevitably, these referendum outcomes constrain and empower elites. They empower elites and representatives of various kinds to act, and in empowering elites to take specific actions, by definition, these outcomes constrain what representatives can do. This happens in two ways, both directly and indirectly. First, referendums constrain and empower those tasked with undertaking particular functions resulting from the referendum. To return to the Parliamentary Constituencies and Voting Act 2011, the Act identified the referendum's consequences and automatically tasked a particular minister with bringing legislation into force. This both directly empowered the Secretary of State to take one of two actions and limited his options. In an indirect sense, the outcome also empowered and limited the executive in a political sense. The outcome offered support for those opposed to electoral reform, such as the Conservatives, and was a blow to the advocates for electoral reform, the Liberal Democrats. 61 The bottom line is this: whatever else they are, referendums are decisions that cannot be self-executing. Referendums are, among other things, instructions to representatives. In instructing representatives to take a particular action, they limit what representatives can do too.
A reader might worry that this paper's argument holds occasionally, but certainly not all the time. As was seen above, referendums are used in an enormous variety of ways. This variation occurs even in a single jurisdiction like the UK. 62 Little if anything holds about their use in general, still less their relationship concerning representatives such as the government. This is an essential and helpful worry. Referendums are indeed democratic processes that vary. The specifics of their use, and whether they are advisory or compulsory, for instance, matters a great deal. Despite this plurality, however, this paper maintains that the same dynamic holds between representatives and voters, albeit to a greater or lesser degree. Whatever else referendums are, and they are and can be many different things in different contexts, they are always decisions. That they are decisions is key to understanding how they both limit and empower the executive. It is right to say this will vary depending on the referendum in question. Take the Parliamentary Constituencies and Voting Act, for example, as one end of the spectrum. The AV referendum was essentially a veto device in the Diceyean tradition; there was a proposal on the table that voters could choose to veto. If a majority of voters in the referendum supported the legislation, then the Bill would become law. Nevertheless, it was still a positive source of legitimacy for the coalition government's Conservative members. It was both a way of limiting the government's actions and achieving the government's goals. The referendum outcome enhanced the power of the winning side, and undermined supporters of the failed proposition. 63 It both limited the government's actions on a particular proposition and allowed them to pursue other political objectives. This is all to say: the degree to which each referendum constrains and acclaims the executive varies depending on the enabling legislation. Nevertheless, because referendums necessarily require decisions to be taken by executive actors, they will always both empower and constrain them. This dual relationship between empowerment and constraint can also be seen in the nature of voting in referendums.

(b) Voting
Voting is not one phenomenon, but many. A vote has many kinds of significance. Whatever else it does, however, voting demonstrates support for a proposition. This is called the contributory or mandate view of voting. 64 The mandate view is not an exhaustive theory of voting, and this is not to suggest that voting is only about contributory mandates. In certain circumstances, it may well be that voting 'Survey research shows that voters do not believe they are pivotal but do believe their vote influences the (mandate) outcome. The contributory model of votingcontributing to advancing a greater public goodis also consistent with the fact that turnout is higher in more important elections. It is higher in countries whose constitutions are more responsive to voters' should count, for example, as an act of deliberation. On the minimalist, mandate view advocated here, however, votes are understood as contributing to a mandate for a proposition, whatever else they are. Guerro advances the mandate view of voting in the context of elections. He rightly argues that it does not only matter for purposes of legitimacy who gets elected but how they are elected. 65 To that end, he introduces the idea of elected representatives as having manifest normative mandates. 66 Guerro's account of manifest normative mandates is trying to solve a paradox. How can it be rational for voters to continue to cast their votes, even if their vote is not dispositive in putting their preferred candidate over the top? Voting remains rational in these circumstances, Guerro argues, because the greater the mandate the representative has, the better placed they are to achieve their goals. Guerro is right about this. This paper builds on his argument to distinguish between primary and secondary normative mandates.
A primary normative mandate, Guerro argues, is support for a representative. There is no reason why primary normative mandates cannot exist for propositions, too, in cases such as referendumsthe greater support for a proposition, the greater the mandate. The importance of clear mandates is seen, for example, in the debates in Quebec around the clarity act and the question of what amounts to a clear majority. 67 Referendums involve mandates in another interesting way, however: the idea that representatives might indirectly receive a mandate from an outcome. These secondary normative mandates may be for representatives or groups such as umbrella campaign organisations and political parties. These secondary normative mandates are mandates to execute or give effect to a particular outcome. They are not precisely support for a representative, but they are authorisation for a representative to Φ or to execute an outcome. Further, what a representative has a mandate for and against are inseparable from each other. If voters choose between two options: to Φ or not to Φ, and they elect to Φ, that outcome provides a primary normative mandate to Φ and not to not Φ, as well as secondary normative mandates to the same effect for representatives to execute those outcomes. Put more simply: if a government has a mandate to leave the European Union, they also have a mandate not to not leave the European Union, and the decision to leave the European Union provided indirect legitimacy to those who supported the winning proposition. This example captures Schmitt's point that referendums entail support for executive action and entail some loss of independence. These types of independence take different forms depending on the degree of authority and power offered by a particular referendum.

(c) Authority and power
The idea of secondary normative mandates is connected to the exercise of power and authority through referendums. First, consider the ways in which referendums can be powerful. They can be powerful insofar as they require representatives to act in a particular way as a political matter. Returning to the example of leaving the European Union, winning the 2016 Brexit referendum lent political power to those who led the campaign. Additionally, referendums can be authoritative. That is to say, they can be legitimate as a matter of domestic law, international law, or political morality. They can authoritatively trigger Bills to become laws, as in the AV referendum in 2011. Referendums can also be a combination of powerful and authoritative. That is to say: they can provide clear mandates to representatives and unambiguous legal outcomes. However, power and authority can also come apart. Results can be powerful, even when referendum outcomes maintain the status quo. The referendum in Scotland in 2014 did not result in majority support for independence, but it did result in changes to the devolution settlement. This is all to say: referendums, by definition, provide primary normative mandates for political propositions and secondary normative mandates for representatives. These mandates may be powerful, and they may also be authoritative. The point is this: referendums do not need to be abused by dictators to lend political power to political actors. The use of referendums, or at least the attempt to use referendums, to bolster a government's legitimacy, is a necessary feature of their use given: (a) the (largely) uncodified state of referendums in the UK; and (b) the degree of dominance the executive has over the legislature.
Finally, it is not an entirely problematic feature of referendums that they provide legitimacy to authorities; it is part of the basis of their appeal. Schmitt is right to argue that it is part of referendums' purpose is to come to clear decisions and then for those decisions to be executed. Referendums must authorise what the people want their representatives to do, and that clarity and finality canin certain circumstancesbe helpful in democracies. 68 However, Schmitt is only right about this in a democratic context when read alongside Dicey. Referendums must limit authorities' actions, too, constraining how and to what end they can use the power referendums necessarily provide. Reading Dicey and Schmitt together demonstrates that both of their approaches are correct, albeit only in part.

The case for codification
So far, this paper has argued thatat least in the UK context -Schmitt and Dicey each offer essential contributions to understanding referendums. Referendums limit what the executive can do but they also acclaim the executive, instructing the executive to execute a particular action. This final section argues that this duality in how referendums are used must be confronted directly, particularly that the use of the referendums should be codified to limit the degree of power they can lend to the executive. This is not to say that legislating for referendums is a silver bullet that will remove all of their failings or prevent the consolidation of power in the executive. Further, as noted above, it is not entirely undesirable that referendums instruct the executive to act in a particular way. There are also risks in legislating for referendums and the loss of constitutional flexibility that it entails. The point here is that it is not necessary that new primary legislation be perfect or address all the risk of referendums. Referendums have presented challenges in countries with clear legislative frameworks. 69 The claim is only this: new legislation will do more good than harm in preventing referendums from consolidating power in the executive in the UK. 70 Currently, the executive has far too much control over the process, outcome, and content of referendums, giving rise to Lord Hailsham and Dicey's concerns about elective dictatorship. It is suggested that a new referendum Act should clarify three matters. First, what issues must be put to a referendum? This would codify existing conventions, but the debates around these conventions have become increasingly fraught, and it is essential to clarify what a fundamental constitutional issue is. Secondly, the timing and process of referendums should be clarified. It should be agreed in the abstract when referendums will be held; for example, on the same dates as elections or distinct from elections, to prevent the manipulation of referendum processes to benefit various constituencies who are more and less likely to vote. Thirdly, a new Act should clarify that referendums must be used post-legislatively. This would allow Parliament to play the role thatas argued aboveit is more effective in than as a legislator: 68 Although note the possibility that this finality can undermine deliberation, as argued in S Chambers 'Constitutional referendums and democratic deliberation' in M Mendelsohn and A Parkin (eds) Referendum Democracy: Citizens, Elites, and Deliberation in Referendum Campaigns (Palgrave, 2001) p 232. 69 The challenges and debates around the Australian postal vote on same-sex marriage show that no framework on its own can guarantee the clear and consistent use of referendums. 70 This is not to overstate how distinctive the abuses of referendums are to the UK. Referendums have been used poorly in Canada, which has similarly little legislative guidance on their use. The referendum question on Quebec's secession in 1995 created confusion and nearly constitutional chaos. the role of scrutineer. Using referendums post-legislatively, ie to approve legislation rather than to initiate processes of constitutional change, clarifies the content of referendums rather than leaving the content inchoate and subject to manipulation. In this way, the vision of codification offered here attempts to build on Dicey's approach. The aim is to get to a place where, so far as is possible, referendums are constraining the executive; and to make sure that referendums are used in a clear, consistent way and capture broad public support for propositions. It is essential to concede, though, that the circumstances Schmitt identifies are inescapable. There will always be a tactical element to referendums, just as there are with other democratic processes such as elections. Political actors will inevitably use these processes to achieve their goals. Given that the executive already dominates the legislative process in the UK, the challenge isas referendums tend to privilege the aims of the executive vis-à-vis the legislaturehow to prevent referendums from being manipulated. Some tactical use of referendums is inevitable, but the aim should be to prevent, so far as is possible, referendums from being used to consolidate power in the executive, especially when that is already one of their natural tendencies.

Conclusion
At first glance, it seems that Schmitt and Dicey's account of referendums are opposites of each other. Schmitt is offering a plebiscitary theory of democracy whereby referendums acclaim authorities. At the other end of the spectrum, Dicey argues that referendums are negative expressions of popular sovereignty. On Dicey's view, referendums limit majorities in Parliament from making constitutional changes that do not have broad public support. Despite these differences, there are striking similarities between these two accounts, the most obvious of which is that both agree that role of authorities as agenda-setters in referendums is essential. This paper has argued that Schmitt and Dicey capture essential features of the use of referendums in the UK: they both emphasise the role of agenda-setting required of representatives in referendums. Schmitt is correct to say that referendums lend legitimacy to authorities, particularly the executive, and Dicey is also correct to say that referendums can and should limit authorities. These processes are inseparable from each other. Acclamations are limits; they tell authorities what they cannot do and what they can. Limits are also acclamations; they are instructions to authorities to takeor not takeactions. If referendums are instructing authorities on what steps to undertake, which Schmitt and Dicey show they necessarily do, they will always empower and constrain authorities. There is always the potential that referendums can constrain authorities andeven in democratic cultures such as the UKthe possibility that they can be abused and manipulated. Although codification cannot prevent acclamation altogether, legislating for referendums will help constrain the executive. Reading Schmitt and Dicey together helps address the distinct challenges presented by referendums in the UK and provides opportunities to improve their use.