The cases

The first Swiss case started in May 1992 when the seven European Free Trade Association countries and the 12 members of the European Communities decided to create the European Economic Area, where the four fundamental freedoms and enhanced cooperation in research and education would apply on a non-discriminatory basis. Sixty-one Swiss laws had to be amended to meet EC standards. The Agreement was submitted to a vote of the people and the cantons. Government, Parliament and all major parties except one, the Schweizerische Volkspartei, strongly recommended a ‘Yes’ vote, denying any threat of ‘mass immigration’ and alleging that, for Switzerland, there was literally no alternative to joining the European Economic Area. On 6 December 1992, the people, however, decided differently: 50.3% of the voters and 16 out of 23 cantons voted against, with an extremely high turnout by Swiss standards of more than 79%. Thus the European Economic Area Agreement, partly tailored for Switzerland, entered into force in January 1994 without Switzerland. There was, however, an alternative: after more than six years of tough negotiations between Switzerland, the EU and its member states, the Bilateral I Agreements were concluded in 1999, consisting of seven compacts on free movement of persons, technical obstacles to trade, public procurement market, agriculture, research, civil aviation and overland transport. On five occasions (referendums of 21 May 2000, 5 June 2005, 25 September 2005, 26 November 2006 and 8 February 2009), the Swiss people voted in favour of the Bilateral I Agreements.

The second Swiss case started in July 2011 when the Schweizerische Volkspartei launched an initiative against ‘mass immigration’, requiring an amendment to the Federal Constitution introducing quotas and ceilings for foreign workers and priority for Swiss citizens. The provision is to be implemented by statute in a three-year period, the 1999 bilateral Agreement on Free Movement of Persons is to be renegotiated and no new treaty violating the new provision can be validly concluded. As the Swiss people had repeatedly accepted the Bilateral I Agreements, the Federal Government as well as all major parties except the Schweizerische Volkspartei strongly advocated a ‘No’ vote. However, on 9 February 2014, a tiny majority (50.3% of the voters and 14½ out of 23 cantons) voted in favour of the initiative that became Article 121a of the Federal Constitution.Footnote ¹ On 26 November 2015, the Swiss Federal Tribunal handed down an important decision announcing that this provision was not self-executing and that it would not apply any statute implementing it in a manner not consistent with the Agreement on Free Movement of Persons, as long as that agreement was not renegotiated or renounced.Footnote ²

The British case started in January 2013, when Prime Minister David Cameron announced that a Conservative government would hold an ‘in-out’ referendum on EU membership before the end of 2017, on a renegotiated package, if re-elected in 2015. The Conservative Party won the 2015 general election with a majority. The European Referendum Act 2015 was enacted by Parliament to enable the referendum on 17 December 2015. On 19 February 2016, EU leaders agreed to a package of changes to EU rules in response to Cameron’s proposal. In a speech to the House of Commons on 22 February 2016, Cameron announced a referendum date of 23 June 2016. The question appearing on ballot papers was ‘Should the United Kingdom remain a member of the European Union or leave the European Union?’ On 23 June 2016, a majority of 51.9% voted in favour of the Leave option.

Constitutional setting

The constitutional setting of the first Swiss case is the mandatory referendum, an instrument of direct democracy created with the first Federal Constitution of 1848, which requires a popular vote with a double majority requirement (people and cantons) on any constitutional amendment, accession to organisations of collective security and supranational communities.Footnote ³ The vote is part of the amendment or accession process, meaning that it is to be organised ex officio, requiring no gathering of signatures, no approval by parliament and no governmental permit.Footnote ⁴

The constitutional setting of the second Swiss case is the popular initiative, an instrument of direct democracy introduced in the Federal Constitution in 1891 that gives the people (by collecting 100,000 signatures in an 18-month period) the right to propose amendments to the Federal Constitution, either in the form of a general proposal or of a specific draft. An initiative in the form of a specific draft shall be submitted without any formal change to a mandatory referendum, i.e. a vote of the people and the cantons, unless it violates the rule of consistency of form, the single-subject rule or mandatory provisions of international law (Article 139-3 of the Federal Constitution).Footnote ⁵ Parliament shall recommend whether the initiative should be accepted or rejected, but it has no power to block the proposal. It may, however, formulate a counter-proposal that is submitted to the people as an alternative to the initiative if the latter is not withdrawn. Thus, in Switzerland, there are no autonomous substantial limits to the initiative process: the people may propose any change of the Constitution, be it a threat to a sacred cow like the abolition of the Swiss army (rejected in 1986), a visionary proposal like an unconditional basic income for every inhabitant (rejected in June 2016) or an insignificant detail like prohibiting the cutting of (actual) cows’ horns (pending).

The constitutional setting of the British case is quite uncertain. Though rare, referendums are not unknown to the British constitution. Before Brexit, there were only two on the national level, the 1975 vote on EC membership and the 2011 vote on the electoral system; nine others were regional, dealing with Northern Ireland (1973, 1998), devolution for Scotland and Wales (1979, 1997) and the North-East (2004), direct election of the mayor and council of London (1997) and independence for Scotland (2014). Virtually every time, the decision to hold a referendum was taken by the government of the day and confirmed by Parliament, mainly for tactical reasons. But as Bogdanor has recalled, the British referendums have created precedents ‘which it would be difficult for any future government to ignore’. There might even be, he ventures, a case for a convention of the constitution: any secession from the United Kingdom, any significant devolution of powers away from Westminster to regional assemblies or supranational bodies, any ‘wholly novel constitutional arrangement’Footnote ⁶ , and any ‘in-out’ referendum on EU membership would need a referendum.

One of the most significant differences between the Swiss and the British popular votes is that the former are intrinsically ‘bottom-up’ while the latter are typically ‘top-down’.Footnote ⁷ ‘Bottom-up’ referendums are triggered by law or by the people themselves, through a signature-gathering process, without giving parliament or government any veto or amendment power. This means that the people decide themselves if, when and on what there is going to be a referendum. Even the mandatory referendum is ‘bottom-up’, as it is anchored in the constitution by will of the people (Article 195 of the Federal Constitution). The Swiss referendum is neither a tactical weapon in the hands of the political majority of the moment nor a plebiscitary device designed to provoke popular applause for a governmental initiative in need of some additional legitimacy. Rather, it establishes the people, i.e. the electorate, as the highest organ of the state, not only nominally in the sense that governments act or pretend to act in their name, but legally because the most basic and the most controversial norms and acts need express popular approval. ‘Government of the people, for the people and by the people’ was a brilliant formula by which Lincoln implored the survival of America's representative democracy during the American civil war. The Swiss have made it real by patiently shaping their brand of direct democracy in a process that roughly lasted from 1830 to the end of that century, officially relying on the longstanding people’s assembly (Landsgemeinde) tradition, secretly inspired by devices invented during the French Revolution (Condorcet)Footnote ⁸ and, in the short run, designed to fight (then-widespread) political corruption. Swiss referendums are binding by law, meaning that parliament, government and administration have to abide by their result, as long as this result is not overturned by a new referendum.

‘Top-down’ referendums are triggered by legislatures or office holders. They are tools by which representative governments tend to compensate their fading legitimacy by carefully choosing specific issues and moments where the people could help them to sustain their policy or position. Far from being a central and unavoidable partner of the democratic game, the people are called in by either one or both of the traditional players that are governments and parliaments. Most of the time, the vote is merely advisory, even though the ideological strength of people’s advice tends to outweigh its legal weakness. The hopes of the vote’s initiators are not always fulfilled, as de Gaulle in 1969 and Cameron in 2016 have bitterly learned.

The people as a state organ

‘The people’ as an organ of the state is an utterly strange body. It takes its decisions in an incomplete and ever-changing composition: never ever do all entitled voters take part in an election and the ones that voted yesterday are not the ones that vote today or tomorrow. The majority of the valid votes decide, but nobody knows of whom it is composed, as the secrecy of the vote is protected. The people, if you think about it, cannot think, cannot talk, cannot discuss; it can only say yes or no to an act or proposition taken by another body, usually parliament, or, as in Switzerland, the promoters of a popular initiative. The people might well be, as in Switzerland, the most powerful state organ, ultimately deciding on any topic it wishes. Yet it is the most helpless and dependent of them all, altogether unable to do anything by its own will and means.

While governments, parliaments and courts exist, ‘the people’ does not. You can meet and talk to the former, not the latter. ‘The people’ is a mathematical construct, a necessary fiction inherent to (direct) democracy, as the set of active voters who form a majority on a given day are not an organic whole. They can neither motivate nor justify their decisions, as each voter may cast a yes or no for any good or bad reason, and even without any reason. Thus the people is the only state organ that is not accountable for its acts and omissions. Its decision stands and binds, yet nobody can be held responsible for it.Footnote ⁹ Because the most powerful organ of the state is brainless, deaf and speechless, populist leaders of all kinds rather shamelessly pretend to speak in the people’s name and try to confiscate its inherent legitimacy.

The foregoing observations should by no means be understood as a sneaky critique of direct democracy in general and referendums in particular.Footnote ¹⁰ My aim is not to denigrate, but to demystify the referendum experience by showing that the way it works within civil society, on account of the state, can do without ideologically relying on any transcendental element like the sovereign’s will, the objective spirit or god’s voice. Direct democracy does not come from above, but can only be created and live on its own down here among us. ‘The people’ is the highest authority of the state, the legitimacy of its decisions is unequalled, yet it is but an abstract, albeit extremely useful, construct of the human mind.

The value, the price and the limits of direct democracy

Laws, rules and regulations need legitimacy in order to be more or less accepted by those who are obliged to follow them, or to be more or less followed by those who are obliged to accept them. An act adopted by parliament expresses the will of the political majority of the moment, to which its legitimacy is more or less directly tied. A decision taken by the people grows out of the political process, too, as its content is usually defined by parliament, but it outshines this process because the anonymous voters who took it stand above or beside political parties. The voters who decide on the outcome of a referendum are not a political majority, but a popular majority that escapes and outweighs any partisan consideration. The people’s decisions stand on their own and exclusively on behalf of the people, and enjoy a high degree of legitimacy because the people, as the symbolic pouvoir constituant, ‘are’ the highest authority of the state. These decisions are neither better nor worse than parliament’s, but they have more weight. The referendum, in its essence, is nothing else than a legitimacy-building device, by far one of the most powerful and effective ones in state affairs.

If direct democracy considerably strengthens the legitimacy, and therefore the efficiency, of state actions taken according to its canons, it has its price and exigencies. Besides time and money, the most stringent claim is that the governing bodies, particularly government and parliament, have to accept the fact that the people, sometimes though not always, may make binding decisions they disapprove and dislike. Their politics still run the country, but they must learn how to pursue their policies by taking into account the possibility of the people’s occasional disagreement with one or the other issue. The learning process of direct democracy, in other words, is tougher for governments than for citizens, as the latter can freely exercise the powers given to them without feeling any pressure of accountability, while the former continue to be held responsible for their acts and omissions even if they are approved – or rejected – by the people.

An existential condition for democracy is the rule of law: fundamental rights and liberties must be effectively guaranteed, legality preserved, separation of powers upheld and protected. The power of the people, occasionally limiting that of parliament, government and administration, must be thoroughly protected by the judiciary against any attempt of wilful or accidental abridgment. As a specifically democracy-oriented fundamental right, the freedom to vote (Wahl- und Abstimmungsfreiheit) requires that the ballot results truly reflect the freely expressed will of the majority of the validly voting citizens. In a long line of cases going back to the beginning of the 20^th century, the Swiss Federal Tribunal has patiently shaped and developed the many faces of this fundamental right, which was finally anchored in the 1999 Federal Constitution (Article 34).Footnote ¹¹

Yet the normative power of the people must also be contained to the limits set forth by the constitutional order, namely respect of internationally and nationally protected fundamental rights and liberties. As a creation of the constitution, the people must abide by the formal and material rules prescribed by the constitution and higher law. Like parliament, it may lawfully limit fundamental rights and liberties in the name of an overwhelming public interest while respecting the principle of proportionality; it has no right, however, even as the highest state organ, to unduly abridge them. In this respect, direct democracy in Switzerland, in spite of its long tradition and solid experience, undergoes a difficult learning process as a series of recently successful popular initiatives more or less openly challenge basic rights guaranteed by the European Convention of Human RightsFootnote ¹² and threaten the credibility of the country as a negotiating partner.Footnote ¹³

National referendums on European integration

The process of European integration has been by far the leading factor in promoting the use of referendums during the last four decades on the continent. To date there have been some 60 referendums tied to EC/EU issues in as many as 23 member statesFootnote ¹⁴ and four non-memberFootnote ¹⁵ states.Footnote ¹⁶ The common denominator of the ‘EU referendums’ is that they were both national and European: national because they were organised under state law; European because they were related in some way to the European integration process. Five main types of national EU-related referendums can be distinguished.Footnote ¹⁷

Membership referendums decide on joining or leaving the EC, EU, European Economic Area or the European Free Trade Association, i.e. accession for non-members or withdrawal for members. With 26 referendums, it is the largest category. Accession referendums have become a common feature of European constitutional law, even in states without any previous referendum experience. Of the 28 EU member states, only nine did not join by referendum,Footnote ¹⁸ two states only declining.Footnote ¹⁹ There have been (to date) three withdrawal referendums, two of which were positive.Footnote ²⁰

Integration referendums are held by member states in the treaty ratification process in accordance with what is now provided for in Article 48 TEU. The six main rounds of treaty revision (Single European Act, Maastricht, Amsterdam, Nice, Constitutional Treaty and Lisbon) have led to 16 such referendums in six member states.Footnote ²¹ Of the five negative results three were overcome by a second positive vote,Footnote ²² while two ‘Noes’ ended the treaty amendment process.Footnote ²³

Enlargement referendums held or announced by member states tend to threaten accession of a new state to the EU. There is only one example: in April 1972, the French voters were asked if Great Britain (sic), Denmark, Ireland and Norway should become members of the EC. Since 2005, the French Constitution submits accession of new states to a mandatory referendum,Footnote ²⁴ targeting Turkey; Austrian leaders pledged the same.

Policy referendums are held by member states on specific EU-related issues like monetary union, fiscal policy or EU’s foreign policy. Italy voted on mandating the European Parliament to draft a constitution for Europe (1989), Denmark and Sweden on the Euro (2000, 2003), Ireland on the Fiscal Compact Treaty (2012), Denmark on the Unified Patent Court (2014) and on opting-in to Justice and Home Affairs measures (2015), Greece on the bailout package (2015), the Netherlands on the Ukraine Association Agreement (2016) and Hungary on the EU’s migrant resettlement plans (2016). Negative results generally prevail.

Finally, non-member states may hold referendums on EU-related issues like national accession procedures and bilateral agreements. Switzerland has held eight such referendums,Footnote ²⁵ being, aside from France in 1972, the only country that has asked the people to vote on EU enlargement steps.Footnote ²⁶

The shaky legitimacy of integration and enlargement referendums

The democratic value and significance of integration referendums are highly doubtful. They allow the voters of one member state to stop a treaty amendment process that directly affects all member states and every single European citizen. National voters can thus veto a supranational issue according to national law and for mainly national motives. Referendums on purely national (or bi-national) issues are traditionally and rightly viewed as core elements of (direct) democracy because the strong legitimacy of the people as the highest organ within that state compensates, in a way, their intrinsic unaccountability and because the consequences of the people’s decision have to and can be borne by the people and the other state organs themselves. On the EU level, however, the vote of the people in one member state has no higher authority than ratification by parliament in another. The consequences of a negative vote in a single member state will have to be dealt with by not only its people and authorities, but also by all other member states and citizens of the EU. People are thus exercising a power that exceeds the limits within which their basic unaccountability can be compensated and justified. This is hardly compatible with the basic principle of democracy enshrined in Article 2 TEU.

Moreover, by deciding to hold a referendum on treaty revision, governments often tend to avoid political responsibility by shifting it to a body which by definition has none. By Swiss standards such votes would probably violate the fundamental right of freedom to vote, as the result does not truly and faithfully express the will of the people on the treaty revision issue, being dominated and hijacked by specific national considerations.Footnote ²⁷

Integration referendums are the result of three basic elements of EU’s constitutional order:Footnote ²⁸ the double unanimity lock for treaty revisions (intergovernmental conference and ratification), the sovereignty of the member states with respect to the ratification procedure (‘in accordance with their respective constitutional requirements’) and the absence in EU law of a truly European referendum.Footnote ²⁹ Here lies what has been called ‘the EU’s direct democratic dilemma’Footnote ³⁰ which in reality is more of a tragedy, because there is no escape.

Enlargement referendums are even worse; a parody of democracy. The voters of one state have simply no legitimacy to decide on the accession of another state willing to join, especially if the people of that state have confirmed this will by way of a referendum.

Challenging people’s decisions

People’s decisions are powerful and must be implemented (except in Greece, apparently),Footnote ³¹ but they can be challenged. Referendum outcomes are not eternally binding. Because of its heavy legitimacy the challenge of a people’s decision is, however, a delicate issue.

‘Born on 6 December’ was a movement of young Swiss voters who wanted to challenge the anti-European Economic Area vote of 1992 through a popular initiative calling for a second referendum. It failed for a number of reasons, and the European Economic Area was later replaced in Switzerland by the bilateral agreements.

The initiative against ‘mass immigration’ accepted on 9 February 2014 directly challenged the five votes on which the Swiss people have approved the bilateral agreements,Footnote ³² although its author, the Schweizerische Volkspartei, deliberately denied it during the campaign. RASA for ‘Raus aus der Sackgasse!’ (‘Let’s get out of the blind alley!’) is the name of a grassroots organisation that has successfully launched an initiative calling to delete the constitutional provision accepted on 9 February 2014.Footnote ³³ A new vote will have to be organised on the issue, unless the RASA initiative is withdrawn in favour of a parliamentary counter-proposal avoiding quotas and safeguarding the bilateral agreements. The lesson from Switzerland is that, politically speaking, only the people, through the initiative process, can ask to reverse the people’s decision through a new referendum, while government and parliament cannot, being bound to implement it.Footnote ³⁴

Four million people signed an online petition for a new EU membership referendum in the United Kingdom.Footnote ³⁵ While it is doubtful that Parliament could simply ignore the outcome of the Brexit vote, even after new elections, what remains of Westminster’s sovereignty certainly includes the possibility of calling a second referendum on that issue. If the British government, as seems likely, unduly delays the formal opening of the withdrawal procedure (Article 50(2) TEU) and if the negotiation on the future relationship with the EU does not go the way it wishes, the pressure for a second referendum might well grow. The British people, however, cannot oblige Parliament to hold a new referendum. The lesson from the UK, although still in the making, seems to go only one step in the Swiss direction, in that only the people can undo what the people have done, but the people can do so only if Parliament wants to.

The EU’s suicidal referendum policy

The EU’s biggest mistake is probably to have deliberately omitted to build and to use the referendum as a legitimacy-building machine for promoting its own goals and endeavours. By precluding referendums of any kind in its legal order, yet tolerating them in the member states on even the most vital EU issues like treaty revision and enlargement, it has fostered centrifugal anti-integration effects that have blocked both of its major endeavours – deepening and widening – and launched what may well become a wave of Leave referendums. Herein lies the main reason why the relationship between direct democracy and the EU is heavily loaded with irritations and rejections.

23 June 2016 is the price the EU pays for having systematically excluded the European citizens from the integration process and giving each member state a veto power over every step of that process. A legal order that puts the legitimising power of referendums in the hands of its foes without using it for its own benefit commits a serious error that might well prove lethal.

A continent-wide referendum under EU law on treaty revision and other constitutional issues has thus become something like a survival kit for the EU that is badly missing. However the EU hopes to overcome the Brexit shock – more powers to the EU, more powers to the states or more differentiated integration – it will need an overhaul of the basic treaties and thus a unanimity that will be difficult, if not impossible, to get. Only the European people – as much of a mathematical construct but not less of a legally irreplaceable concept as the German or Polish or Dutch people – would have the power and legitimacy to break the unanimity lock. A European referendum, however, would need a solid legal basis, and a legal basis would require unanimity. Under the existing constitutional setting there seems to be no solution to the EU’s direct democracy dilemma.