A. Membership referendums

Membership referendums account for the majority of EU referendums and those which most often pass. They take two distinct forms, an accession referendum or a withdrawal referendum, and we explore each in turn.

The accession referendum is usually undertaken by candidate states on the negotiated accession treaty, and accounts for the vast majority (19 out of 22) of the membership referendums;Footnote ¹⁰ all but two have passed (as Table 1 illustrates). No founding Member State put accession to the people.Footnote ¹¹ This is unsurprising for three reasons. Firstly, the constitutional significance of the organisation was not yet fully apparent. Secondly, the founding members, Belgium aside, had new constitutional clauses to legitimise the delegating of powers to international organisations. Thirdly, referendum practice was then largely non-existent. However, 16 of 22 candidate states have since put accession to popular approval. The accession referendum emerged with the first round of enlargement when three of the four candidate states held referendums on accession. The UK was the exception, but popular approval was forthcoming for continued membership only two years later (see further below). Norway is the only candidate state to have rejected accession, having done so in 1972 and again in 1994 when four other accession referendums delivered yes votes including the only sub-national accession referendum to have been held (by the Finnish Åland Islands).

Table 1 Membership referendums (22).

A potential peril of accession referendums is that they reject the patiently negotiated accession agreement. But this is also part of their promise because it reduces the likelihood of a State joining the EU without popular support, something the EU can clearly do without. The referendum cannot be rivalled as a device for legitimising accession given the vast ramifications that now flow from EU membership, such as the obligation to join the Euro. It is thus unsurprising that constitutions have been amended to require an accession referendum (Hungary) or to provide for the possibility of such a referendum (Czech Republic, Poland, Slovakia, Slovenia and Sweden), and in some States (Austria and Finland) this has been the first national level referendum in generations. Of the five candidate States not to have held an accession referendum since the first enlargement, three were recent military dictatorships (Greece, Portugal and Spain) joining prior to the integrationist steps of the Single European Act (SEA). The outliers are the recent entrants of Cyprus (2004) and Bulgaria (2007), though until that point neither had held a referendum under their existing constitutional system.Footnote ¹² It has been suggested that the European Council could impose a referendum requirement as a condition of EU accession.Footnote ¹³ This would seem an imposition too far and certainly could not be required for the actual accession agreement itself precisely because Article 49 TEU itself stipulates that the ‘agreement shall be submitted for ratification by all the contracting states in accordance with their respective constitutional requirements.’ In any event, with rare exception since the 1990s, the accession referendum has become the norm for legitimising accession.

The withdrawal referendum accounts for 3 of 22 membership referendums. The UK has held two. First in 1975 when following renegotiations it held a referendum that delivered strong support for continued membership. Second, in 2016 when a Conservative government fulfilled its manifesto commitment to hold an in–out EU referendum on renegotiated terms of membership with 51.9% voting to leave. The UK’s second withdrawal referendum took place following the Lisbon Treaty’s inclusion of Article 50 TEU specifying that ‘[a]ny Member State may decide to withdraw from the Union in accordance with its own constitutional requirements.’ Normally, States very rarely constitutionally require a referendum for treaty withdrawal, but the EU is no ordinary treaty. Moreover, as so many States have held an accession referendum, and others have used a referendum to approve a treaty revision (eg France, Spain and Luxembourg), it is highly unlikely that a different route would be used for the constitutionally momentous step of withdrawal.Footnote ¹⁴

Given growing euroscepticism and populist parties seeking to exploit this, it may be that the UK’s 2016 referendum will not be the last Member State withdrawal referendum. There is an obvious peril dimension to Member State withdrawal referendums because they have destabilising ramifications for the EU and all the more so when it is a State of the size and significance of the UK. A British ‘Remain’ vote, by contrast, might have brought some promise in terms of making the UK a less reluctant partner. A Brexit silver lining for the EU may be that it dampens demand for withdrawal referendums, and the willingness of politicians to call them unless they really do wish to leave the EU;Footnote ¹⁵ as Prime Minister Cameron found out, there are no guaranteed outcomes when one calls such referendums and a comfortable poll lead can rapidly dissipate.

The other express withdrawal referendum was held by Greenland, a constituent part of Denmark, and the nature of the peril for the EU here is incomparable to that of the Member State withdrawal referendum, precisely because the EU does not risk becoming one Member State smaller as a result. Greenland voted overwhelmingly against joining in the 1972 Danish accession referendum and was granted a form of home rule by the late 1970s. In 1982 it held a consultative referendum in which a small majority approved of withdrawal. Following Greenland’s parliamentary approval, Denmark negotiated Greenland’s withdrawal, which took place in 1985.Footnote ¹⁶

An important challenge to the membership dynamic comes from independence referendums within Member States. The 2014 Scottish independence referendum is clearly distinct from a constituent part of a Member State, as with Greenland, voting expressly and exclusively on withdrawing from the EU.Footnote ¹⁷ The Scottish-style referendum is not on an EU question as such. However, it has potentially significant ramifications for the EU as the would-be seceding entity wishes to remain within the EU. It was much debated whether a seamless internal EU enlargement was possible via the treaty revision procedure, or arguably the more persuasive view, whether a seceding entity would need to apply to join.Footnote ¹⁸ The ‘No’ vote on Scottish independence leaves this contentious question unanswered, but Catalonia’s on-going pursuit of secession from Spain ensures that it remains firmly on the table. The UK withdrawal referendum also raised the intriguing reverse question of whether a constituent part of a Member State, Scotland, could remain in the EU while the larger political entity of which it formed a part departed.Footnote ¹⁹

B. Treaty revision referendums

These are held by Member States as a precursor to ratifying a revision to the EU treaties and have been the second most common type of EU referendum. Treaty revision referendums, and the complications to which they give rise, stem from the fact that for a treaty revision to enter into force, the EU’s rules (currently Article 48 TEU) require ratification ‘by all the Member States in accordance with their respective constitutional requirements.’ If those domestic constitutional requirements are, or become, a referendum, then this becomes a prerequisite to treaty revision entering into force for the EU.

As Table 2 illustrates, all six main rounds of treaty revision, including the attempted revision via the Constitutional Treaty, have generated these referendums. Only six Member States have actually held them. Three (Luxembourg, the Netherlands and Spain) did so only once, precisely for the only revision framed in bold constitutional language (the Constitutional Treaty).Footnote ²⁰ Most are accounted for by Denmark and Ireland. The Danish Constitution’s transfer of powers clause (s20) stipulates a five-sixths parliamentary approval requirement, with popular approval required if only an ordinary parliamentary majority is obtained. However, the SEA did not fall within that clause. The first ever treaty revision referendum was actually Denmark’s non-constitutionally required consultative referendum on the SEA. The only other revising treaties to be subjected to Danish popular approval, Maastricht and Amsterdam, fell within the transfer of powers clause and referendums became constitutionally required as five-sixths majority approval was not forthcoming.Footnote ²¹

Table 2 Treaty revision referendums (16).

The Danish ‘No’ to Maastricht ratification in 1992 gave the EU its first experience of a negative popular vote bringing the treaty revision process to a halt and thus clearly demonstrating the peril for the EU of such referendums. European leaders agreed on a range of opt-outs so that Danish ratification could take place, as it did following popular approval in Denmark. The Danish ‘No’ also contributed to the French President calling a referendum on the Maastricht Treaty. This delivered the narrowest of margins in favour (51%) and is often viewed as signalling the end of the ‘permissive consensus’ on European integration.Footnote ²² This vote sent a powerful message to political actors that the discretionary treaty revision referendum – ie one that is not constitutionally required – should be deployed with utmost caution, for a yes vote could not even be guaranteed in a founding Member State with a generally favourable attitude towards European integration. This was made painfully clear when the Constitutional Treaty failed to surmount the referendums that political actors in France and the Netherlands decided to impose. Although discretionary referendums comfortably passed in Spain and Luxembourg, it seemed most unlikely that the Constitutional Treaty would have fared well in other States where referendums had been promised, such as the UK. The popular votes on the Constitutional Treaty underlined the peril of treaty revision referendums in the post-permissive consensus era and gave rise to its demise and repackaging (the Lisbon Treaty).

This brings us to Ireland, which, uniquely, has held a referendum on each major treaty revision that has come into force. This is owed to a constitutional challenge to the government’s attempt to ratify the SEA in which the Supreme Court ruled that EU revising treaties going beyond the scope or objectives of the existing treaties, as the SEA was considered to, would require a constitutional amendment and therefore a referendum.Footnote ²³ Since that ruling, Irish governments have put all main treaty revisions to a referendum.Footnote ²⁴ Negative votes on two occasions (Nice and Lisbon) brought the treaty revision process to a halt and led to European leaders giving Ireland various assurances that allowed for second referendums to take place. This ‘second referendum phenomenon’, whereby the people vote again following a negative vote illustrates another peril of referendums for the EU: it fuels euroscepticism.Footnote ²⁵ It has given rise to the damaging myth that Brussels will not take ‘no’ for an answer. This is most obviously incorrect in not distinguishing between different types of EU referendum. Thus, for example, Norway has not been forced to join the EU, nor was Greenland – or the UK – forced to remain. Moreover, even when focusing on the treaty revision referendums it should be underscored that certain relevant concerns were addressed to justify second referendums, and, on that basis, the Danish and Irish people turned out in larger numbers and voted comfortably in favour of the relevant treaties.

Treaty revision referendums can certainly have redeeming features, particularly so where they are constitutionally mandatory or politically obligatory rather than discretionary devices.Footnote ²⁶ They are a powerful and salutary reminder to political elites that the treaty revision process must pay meaningful adherence to popular opinion. Indeed, it is part of their promise that they contribute to legitimising integrationist steps, particularly in those countries with stronger currents of euroscepticism. One could contrast Denmark, which has held referendums on three treaty revisions, and the UK, where none were held, and where such a referendum might have contributed to blunting the demands for the leave/remain referendum with the many negative consequences for the EU and UK that will flow from that leave vote.Footnote ²⁷ However treaty revision referendums can be called for partisan gain, as was transparently the case with the French Maastricht referendum, and to bolster the negotiating positions of Member States during the treaty revision process. And because every State has a ratification veto point, there are serious questions about the legitimacy of the treaty revision process of an organisation with over 25 Members being held hostage to the outcome of a popular vote in a single State.Footnote ²⁸ This is accentuated by long-standing concerns about whether voters are even voting on the basis of the referendum issue (ie the treaty revision) or basing their vote choice on unrelated concerns, such as using the referendum to punish an unpopular government (a phenomenon known as second-order voting).

Given extant experience with direct popular veto points, and the well-documented growth in Eurosceptic sentiment since the Lisbon Treaty was drafted, it is unsurprising that elites have avoided larger scale treaty revisions.Footnote ²⁹ Tellingly, the EU has deployed an array of tools to deal with the Eurozone crisis, such that the constitutional architecture of European economic governance has been profoundly changed, but with only the most minor of treaty revisions.Footnote ³⁰ The treaty revision referendums may thus contain the seeds of their own destruction as their taxing and unpredictable nature leads to constitutional adaptation being pursued in ways that evade their application. This is another peril of the treaty revision referendum because it leads to the greater transparency and inclusiveness of the treaty revision rules being eschewed in favour of other routes to constitutional adaptation.

C. Policy referendums

Policy referendums are held by Member States on a specific EU policy field, such as monetary policy, fiscal policy or foreign policy, and have been the fastest growing category of EU referendums in recent years (Table 3). In terms of peril for the EU, the most troubling is the enlargement referendum. The EU’s enlargement rules (currently Article 49 TEU) require accession treaties to be ratified by all the Member States ‘in accordance with their respective constitutional requirements’. There was no such constitutional requirement in France, but the President called a non-binding vote on the first wave of enlargement in 1972 and if the outcome had been unfavourable it is unlikely that France would have simply ratified the accession agreement. In short, a ‘No’ vote would have imperilled the EU’s first enlargement. More recently (2005) France introduced its first ever constitutionally obligatory referendum to require enlargement referendums following any accessions after Romania, Bulgaria and Croatia, although following a 2008 amendment this can be avoided via a particularly high super-majority in parliament.Footnote ³¹ The French enlargement referendum lock threatens the future of the EU’s most successful foreign policy tool, especially if other Member States follow suit (an EU referendums domino logic was previously seen with the Constitutional Treaty).Footnote ³² The new Dutch citizen-initiated referendum, considered further below, will surely see Eurosceptic groups seek its activation vis-à-vis future enlargements. In sum, future accessions will most likely need to navigate multiple enlargement referendums, and the peril for the EU, and the would-be entrant, is thus obvious. If a negative vote occurred, recourse to the strategy employed with treaty revision – a second referendum – would seem difficult to justify. What could the European Council offer? The repackaging solution to the French and Dutch referendums on the Constitutional Treaty would also not appear to be feasible, for how does one repackage the accession of one or more countries in a manner more palatable to voters?Footnote ³³

Table 3 Policy referendums (10).

Whilst the perils of enlargement referendums are easy to identify, their promise is much less so for there is something intuitively unattractive about holding enlargement hostage to the direct popular veto point of Member States. These referendums have been referred to by one direct democracy scholar as a ‘parody of democracy’ as ‘[t]he 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.’Footnote ³⁴ To contest the legitimacy of such a referendum is not to defend the approach to recent enlargements which, as Shaw emphasised, saw no effective public debate taking place until candidate states had already received firm commitments.Footnote ³⁵ In terms of neutralising the peril of enlargement referendums, Hillion drew on the duty of loyal cooperation (Article 4(3) TEU) to suggest that as enlargement is one of the EU’s objectives, enforcement proceedings could be feasible where a State introduces constitutional requirements to make it virtually impossible to ratify an Accession Treaty.Footnote ³⁶ It is not clear whether a referendum requirement on Turkish accession counts as making it virtually impossible or whether even more rigorous requirements were intended such as a referendum and quorums, and parliamentary supermajorities. But either way, Article 49 TEU expressly stipulates ratification ‘by all the contracting states in accordance with their respective constitutional requirements’ thus according virtually complete discretion as to those requirements. Article 4(2) TEU on the EU respecting Member States’ national identities inherent in their fundamental political and constitutional structures presents a further obstacle. Therefore, contesting enlargement referendums appears legally untenable to say nothing of it being politically damaging for the EU if a Commission-led challenge were to be seriously contemplated.

Referendums on adopting the Euro account for two policy referendums. This is relatively few given that 19 States have adopted the Euro. With the exception of Denmark and the UK, Member States are legally obliged to join once they fulfil the entry conditions which helps explain why many States – including Austria, Finland, Malta, Slovakia and Slovenia – that adopted the Euro relatively soon after EU accession referendums did not hold separate Euro referendums. It also helps to explain Latvian and Lithuanian adoption of the Euro, without a popular vote, during the Eurozone crisis. However, this legal obligation, and an accession referendum less than a decade earlier, did not prevent Sweden from holding a consultative Euro referendum in 2003. The negative vote makes it politically impossible for Sweden to join without popular approval. That Swedish referendum was likely to have been influenced by its Danish neighbour holding the first Euro referendum in 2000, with popular approval not forthcoming.Footnote ³⁷

There is a clear promise dimension to Euro accession referendums as they supply popular legitimacy for the momentous decision of replacing a national currency and accepting Eurozone obligations. The main peril is that a popular veto prevents States from meeting their obligation to join. This is a very small price for the EU to pay to ensure that Eurozone accession has popular legitimacy. Unlike with a popular vote against a treaty revision or enlargement, a negative vote on Euro accession is not meaningfully extraterritorial: the Euro already exists and continues to exist regardless of referendums against joining.

The policy referendums category also includes two rather idiosyncratic policy referendums. The first was an Italian consultative referendum in 1989 in which the people voted for the European Parliament to be accorded a mandate to draw up a constitution for Europe. This referendum was distinctive in that nothing directly flows from a yes vote because it is not up to the Italian electorate to accord such a mandate. The second is the Greek bailout referendum in mid-2015, following protracted negotiations between Greece and its Eurozone creditors. Its distinctive features included the shortest time between a referendum announcement and the vote (just over one week), and that the terms of the bailout package on which the Greek people voted had already been withdrawn by the creditors – not to mention that the referendum decision was ignored.Footnote ³⁸

Finally, three recent additions to the EU’s referendum landscape have emerged. The first is referendums on ‘extra-EU treaties’. These treaties are intimately connected to EU law, and indeed can even make use of the EU’s institutions, but have hitherto only been concluded between a range of EU Member States and have also not required unanimous ratification. Accordingly, the promise of these referendums is to supply popular endorsement for adhering to such treaties, while not generating the same level of peril for the EU as the treaty revision or enlargement referendum.Footnote ³⁹ The Fiscal Compact Treaty, primarily aimed at enhancing fiscal discipline by Eurozone States, was the first extra-EU treaty to be put to a referendum. The Irish people supported this treaty in a 2012 vote following the Attorney General’s recommendation that a referendum was required. In 2014 Denmark held the second extra-EU treaty referendum on the Unified Patent Court Agreement. The Danish Ministry of Justice concluded that the Agreement transferred sovereign powers and a referendum was successfully held once it became clear that a five-sixths parliamentary majority could not be obtained.Footnote ⁴⁰

Unlike the extra-EU treaty referendums, the two latest additions to the EU’s referendum landscape create considerable peril for the EU. The first flowed from the Dutch Advisory Referendum Act of 2015 requiring a referendum to be held on laws or treaties approved by Parliament where at least 300,000 support it. It was immediately deployed vis-à-vis the EU Association Agreements with Ukraine, Georgia and Moldova. Only the Ukraine Agreement met the initial threshold of 10,000 signatures within four weeks, and then easily satisfied the 300,000 signatures requirement. Popular approval was not forthcoming on the lowest turnout (32%) of any EU referendum held to date.Footnote ⁴¹ Without Dutch approval the Agreement could not enter fully into force for the entirety of the EU.Footnote ⁴² The Heads of State or Government of the EU’s Member States adopted a decision on the interpretation of the EU–Ukraine Agreement at the European Council in December 2016 with a view to assisting the Dutch government in pursuing Parliamentary approval for ratification. Dutch ratification of this important EU Agreement is no small matter. The broader issue is that the new Dutch direct democracy instrument signals the birth of the citizen-initiated referendum on EU matters. As well as mixed agreements, the remit of this new instrument includes accession agreements and treaty revision and is becoming a focal point of activity for Eurosceptic groups. The peril from an EU perspective will be multiplied if the citizen-initiated referendum on EU matters spreads to other Member States.

The latest addition to the EU policy referendum landscape emerged in Hungary following the EU’s response to the refugee crisis. Hungary was one of four Member States outvoted on the mandatory refugee relocation quota.Footnote ⁴³ Prime Minister Orbán responded by calling a referendum on whether the EU should have the power to impose compulsory settlement of non-Hungarian citizens in Hungary without the Hungarian Parliament’s approval.Footnote ⁴⁴ This 2016 referendum resulted in a predictably overwhelming majority against such resettlement, though failing to meet the turnout quorum. There is a case for using infringement proceedings against such referendums for violating the Article 4(3) TEU duty of sincere cooperation. It is a discretionary referendum that can be viewed as doing the opposite of ‘ensuring fulfillment of the obligations arising out of the Treaties’ and refraining ‘from any measure which could jeopardise the attainment of the Union’s objectives.’ The political circumstances in Hungary were such that Commission intervention would never have halted the referendum.Footnote ⁴⁵ It is certainly advisable for domestic constitutional challenges to run their course, and ideally to prevent such a referendum from occurring, prior to Commission intervention.Footnote ⁴⁶ Nevertheless, there is a case for establishing that a Member State would breach Article 4(3) TEU where it holds, or seeks to hold, a referendum of this nature on an already adopted EU measure. This could deter Member States from embarking on the collision course that such referendums pose for the EU legal order. Recourse to EU level enforcement tools would also have damaging ramifications for the EU, but it may be a price worth paying to try to halt ex post referendums on binding EU decision-making from re-emerging and spreading.

D. Assessing EU-wide referendum proposals

There have long been calls for EU-wide referendums from eminent scholars in a wide array of fields.Footnote ⁴⁷ Our focus in this section is on EU-wide treaty revision referendums, but before doing so we comment briefly on proposals for an EU-wide referendum on a founding moment. De Gaulle suggested the use of a Europe-wide referendum for a founding document as early as 1949, as did Spinelli for a new federal constitution in the early-1960s.Footnote ⁴⁸ Such proposals proliferated following the Maastricht Treaty, most significantly via a proposal supported by some 97 members of the Convention.Footnote ⁴⁹ Although substantial legitimacy gains would flow from a popularly approved re-founding of the EU, where such proposals require popular approval in every Member State, as the convention proposal did, they impose the unrealistic ratification hurdle of unanimity. Many EU-wide referendum proposals for a new founding emerged when the EU had nearly half its current membership, and prior to popular rejections of the Constitutional Treaty. Those advocates might not be so favourably disposed to an EU-wide referendum for a much enlarged EU, and thus additional veto points, in an increasingly eurosceptic climate.

Cheneval has called for using EU-wide treaty revision referendums in the much enlarged EU.Footnote ⁵⁰ He advanced a powerful critique of the dysfunctional nature in which referendums have been haphazardly deployed in the treaty revision process, later accentuated by the Lisbon ratification saga where less than one per cent of the EU’s population – in a single Member State – was popularly consulted. Cheneval articulates prima facie compelling legitimacy and democracy enhancing qualities for an EU-wide treaty revision referendum. However, this proposal faces the same criticism applicable to an EU-wide referendum for a founding moment wedded to unanimous ratification, namely it is an unrealistic ratification hurdle. The EU has arguably the most rigid rules of amendment in existence.Footnote ⁵¹ To think in federal terms, no federation, even those with as few states as Australia – six – accord all constituent units a veto point over the general constitutional revision procedure. To convert the existing veto point into a popular veto point only exacerbates the rigidity of the EU’s rules of change. Even the last major treaty revision, Lisbon, was unable in the first instance to surmount the only Member State referendum held (Ireland).

Cheneval’s proposal emerged prior to the onset of the financial crisis and the Eurozone crisis, which has led to a considerable deterioration in the image of the EU as shown in Figure 2. This figure shows a marked downward trend in positive views of the EU and a commensurate rise in negative views of the EU since the crisis. However, this data is at EU-level and masks important differences among EU Member States. Hobolt and de Vries have rightly taken account of variation in national settings to provide a more nuanced and upbeat assessment of support for the EU.Footnote ⁵² Even if there is not such a clear downward trend in support for the EU, these are inauspicious times to be proposing EU-wide referendums wedded to unanimous ratification as the standard treaty revision model. Making future treaty change theoretically democratic, becomes, paradoxically, an undemocratic proposal if treaty change is made practically unattainable. It would shield the treaty text from meaningful democratic contestation and heighten the existing pressure on other less democratic means of constitutional adaptation. In short, the promise of such proposals seems unattainable while the peril seems all too real.

Fig. 2 Public image of the EU over time. Data source: Standard Eurobarometer 86 (2016) Public Opinion in the European Union. Question asked: In general, does the EU conjure up for you a very positive, fairly positive, neutral, fairly negative or very negative image?

Most EU-wide treaty revision referendum proposals have however argued for high double majorities (of both voters and Member States).Footnote ⁵³ They aspire to unleash the purported promise for the EU of popular voting, while avoiding the peril of increased rigidity via a single state popular veto point. Some have drawn on popular support for EU treaty revisions to take place via referendums to bolster the case for EU-wide treaty revisions.Footnote ⁵⁴ Proposing to surmount the unanimous ratification requirement is often challenged as a federalist step too far, even though other large international organisations do not remain wedded to unanimity.Footnote ⁵⁵ But adding a harmonised ratification requirement is considerably more intrusive in a way that federal regimes, unlike international organisations, often are. It also requires a device to be deployed that is either currently impermissible at the national level (eg Germany and Belgium) or impermissible in relation to treaties (eg Italy).Footnote ⁵⁶

EU-wide referendum proposals also often ignore the nature of the EU’s treaty text. Its overinclusiveness in particular, should lead one to caution against EU-wide referendums as the standard amendment procedure.Footnote ⁵⁷ For example the recent revision to transitionally accommodate a greater number of MEPs, because the Lisbon Treaty provisions providing for this only entered into force after the 2009 elections, would hardly be of sufficient constitutional import to warrant an EU-wide referendum. Determining which revisions warrant an EU-wide referendum could be modelled on the new treaty revision rules (Article 48(2) TEU). Under these the EU Convention route need not occur if the European Council determines, with European Parliament consent, as it did for the aforementioned MEP revision, that it is not justified by the extent of the proposed amendments.

An additional problem with EU-wide treaty revision referendum proposals is the absence of consideration to campaign regulation. Even within EU States that use referendums there is considerable regulatory divergence on questions that include the extent of public and private financing, constraints on broadcast and print media, regulation of information campaigns and civil society groups, as well as the role of independent electoral bodies.Footnote ⁵⁸ EU-wide referendum proposals have not engaged with the sheer heterogeneity of referendum campaign practices across the EU. Nor has a regulatory model been suggested. One could propose the least intrusive model whereby, with perhaps the exception of the day to hold the referendum, matters are left to the domestic constitutional system, through to the most intrusive model whereby some of the aforementioned issues are regulated at EU level.

We conclude by highlighting the most obvious problem with these EU-wide referendum proposals. It is difficult to envisage circumstances in which the political will to overcome the unanimous ratification lock could be found, much less doing so while not only stipulating, but potentially even regulating, the ratification procedure. Thus promise there may well be, but political feasibility there is not. And there is a peril that contributes to this political infeasibility, namely what happens to one or more Member States unable to deliver popular approval for a treaty revision? The federal answer, subject to any opting out possibility, is that the amendments would still apply to all; the non-federal answer would permit withdrawal, and both create obvious peril for the EU.

A. Origins and treaty-enshrined design choices

The origins of the ECI are usually attributed to a proposal by a German Bundestag delegate included in the draft Constitutional Treaty at the culmination of the convention process.Footnote ⁵⁹ This proposal underscored that its effect was ‘to bring Europe closer to the people, as Laeken recommended’ and represented ‘a large step in the democratisation of the Union.’Footnote ⁶⁰ The Commission’s right of legislative initiative was left intact as the request to submit proposals would be to the Commission. In effect it would equalise the position of EU citizens with the Council and the European Parliament’s treaty-derived power to request Commission action.Footnote ⁶¹ This made it a more palatable option than more adventurous direct democratic innovations put forth during the Convention process.

Crucially, the language proposed by the Convention expressly borrowed from the language of the Parliament’s right to request a proposal from the Commission by stipulating a requirement for the ‘citizens’ to consider that such an ‘act is required for the purpose of implementing this Constitution’ (‘treaty’ under then Article 192(2) EC). The Parliament’s right was never thought to extend to requesting proposals for treaty revision, that not being consistent with the language of being ‘required for the purpose of implementing this treaty’, and express reliance on the provision only ever involved requesting legislative proposals.Footnote ⁶² Direct borrowing of this language for the ECI, retained in the Lisbon Treaty, renders it unreasonable to conclude that citizens were empowered to request treaty revision proposals. This restriction is of considerable consequence because the policy density of the EU’s treaty text is such that much that might not be off limits in a national agenda initiative, because it falls within legislative competence, would be off limits in the EU. For citizens, this curtails the ECI’s potential promise and could generate discontent with the EU, thus constituting part of the peril of the ECI.

The Convention ECI proposal left much to be determined by an implementing law. However, the end product in the draft Constitutional Treaty, retained in the Constitutional Treaty and the Lisbon repackaging, added into the primary text the minimum required threshold of signatures, namely no ‘less than one million’, and the additional constraint that signatures must come from ‘a significant number of Member States’.Footnote ⁶³ A territorial based constraint, the specifics of which were to be determined by the implementing regulation, is understandable given that we are dealing with a political system composed of sovereign states and the very premise behind the instrument is to generate a transnational dimension. The signature threshold is comparatively low in that the Constitutional Treaty was expected to come into force for an EU of around 480 million, thus barely over 0.2% of the EU’s population and less than 0.2% by the time the Lisbon Treaty entered into force for an EU of then over 500 million.Footnote ⁶⁴ To use examples of agenda initiatives within the EU’s Member States, many have thresholds of over 1%, less than 0.5% is rare indeed, and only Italy has a requirement lower than the ECI at under 0.1%.Footnote ⁶⁵ The ECI’s low threshold therefore points to a willingness to see its democratic promise unleashed. Comparative borrowing could have been used to justify a threshold two or more times the one million figure, that paradoxically could have resulted in a direct democratic instrument that, because of high triggering thresholds, exacerbates popular discontent with the democratic workings of the EU.

B. Design choices in the implementing regulation

The potential promise of the ECI depended crucially on design choices made via the implementing regulation. While the Lisbon ratification saga was ongoing, the European Parliament called for the Commission to bring forth a legislative proposal without delay after entry into force of the Lisbon Treaty.Footnote ⁶⁶ The Commission Green Paper actually appeared before the Lisbon Treaty entered into force.Footnote ⁶⁷ Within months the legislative proposal emerged, political agreement was reached by the end of the year, and the Regulation was adopted in February 2011, and became applicable on 1 April 2012.Footnote ⁶⁸ This unusually fast journey through the legislative process attested to a desire to tap the potential of the Lisbon Treaty’s most important democratic innovation. In what follows we focus on some central design choices taken by the implementing Regulation; some enhance the democratic promise of this instrument, while others make it less accessible and may accordingly generate citizen discontent that could bolster the ECI’s peril for the EU.

The Treaty provides that the one million signatures must come from ‘a significant number of Member States’. The Regulation stipulated that at least one-quarter of the Member States would be required (Article 7(1)), in line with the Parliament’s initial call for a proposal and in contrast to the Commission’s proposal for one-third. For ordinary language usage, and with a view to facilitating the threshold being surmounted, one could argue for an even lower number, in that, for example, one-fifth is arguably a significant number of Member States. For the Commission, the one-quarter threshold was thought too low to guarantee that the Union interest was adequately reflected. Fortunately, the Parliament’s proposal prevailed which better serves the ECI’s democratic promise.

A second crucial choice was a signature threshold from each of the minimum number of Member States required. No such constraint was outlined in the Treaty. However, given the requirement of a minimum number of Member States and the related logic of creating a meaningful level of European-wide support, it would be anomalous for wholly nominal support in one or more of the requisite minimum number of States to suffice. The Parliament’s resolution proposed a 0.2% threshold, but the choice ultimately adopted (Article 7(2)) corresponds to the number of MEPs elected in each Member State multiplied by 750. The consequence of this degressive proportionality is that a number of the smaller Member States would have to obtain over 0.2% (considerably more in Luxembourg and Malta) to count as one of the seven Member States required, but with the major advantage that the vast majority of States would need 0.2% or less, and the four biggest would not even need to obtain 0.1%.

Unsurprisingly for a twenty-first century agenda initiative, the possibility to collect statements electronically was expressly provided for (Articles 5(2) and 6) which considerably facilitates attaining the requisite statements in support. The minimum age adopted for supporting initiatives was the age at which citizens are entitled to vote in European Parliament elections (Article 3(4)), which is 18 in all Member States bar Austria, where it is 16. The consultation exercise saw some support for a threshold of 16 on the basis that this would encourage youth civic participation on European issues and that this is not an election but merely an agenda-initiative. Tying the age requirement for initiatives to that of elections is the comparative norm, but there remains much to be said for a lower age requirement for the previously mentioned reasons and to facilitate reaching the signature thresholds.Footnote ⁶⁹ However, ultimately it does not appear to have been seriously contemplated, which is unsurprising insofar as it would require Member States (Austria aside) to produce new electoral lists.

The Commission Green Paper considered that requiring a citizens’ committee to organise an initiative might be too burdensome and preferred not to impose any restrictions on who could present an initiative. The legislative proposal accordingly permitted both EU citizens and legal persons and other organisations to organise initiatives. At the behest of the European Parliament, a requirement for a citizens’ committee composed of at least seven EU citizens resident in at least seven different Member States was introduced (Article 3(2)). Although additional hurdles make it harder to get an initiative up and running, the redeeming virtue here was to insert a transnational dimension that helps foster political debate across borders and hopefully creates a network for pursuing statements across at least seven Member States.

The citizens’ committee would be responsible for preparing the proposed initiative and submitting it to the Commission. In terms of the form and content, the Green Paper rightly saw the draft legal act requirement in some legal systems as unnecessarily restrictive and burdensome as well as not being required by the treaty text.Footnote ⁷⁰ The Regulation merely requires the title, the subject matter, the description of objectives (in respectively no more than 100, 200 and 500 characters) and the treaty provisions considered relevant.Footnote ⁷¹ These are prima facie not especially taxing requirements, though identifying relevant treaty provisions in the complex legal system of the EU is no small feat. But this alone will not get an ECI registered so that the signature collecting process can begin. An ex ante registration test, applicable prior to commencing the signature gathering exercise, was introduced. The European Parliament resolution proposed such a test, but the Green Paper was against this. The consultation process saw many support an admissibility test prior to all signatures being collected. The legislative proposal thus provided for an admissibility test following the collection of 300,000 signatures (Article 8), but during the legislative process this gave way to an ex ante registration test. The main logic is to avoid frustrating citizens who have supported an initiative that is later found to be inadmissible. On the other hand, an ex ante test could have democratically pernicious consequences for the EU by foreclosing the emergence of transnational debate. The extent to which debate will be foreclosed ultimately depends on the substance of the admissibility test and how it is applied.

The Regulation lays out four conditions before the Commission can register an initiative (Article 4(2)). The first concerns the organising committee being appropriately formed (Article 4(2)(a)). The second to fourth require respectively that it must ‘not manifestly fall outside the framework of the Commission’s powers to submit a proposal for a legal act of the Union for the purpose of implementing the Treaties’ (Article 4(2)(b)), nor be ‘manifestly abusive, frivolous or vexatious’ (Article 4(2)(c)), nor ‘manifestly contrary to the values of the Union as set out in Article 2 TEU’ (Article 4(2)(d)). The second of these conditions textually replicates the Treaty text itself, albeit now adding the rider of ‘manifestly’ which suggests an exacting standard for refusing registration on this ground. The third and fourth conditions appear to leave more discretion to the Commission.Footnote ⁷² Oversight is expressly stipulated in the Regulation, the Commission being required to inform organisers of its reasons for refusal and of judicial and extra-judicial remedies (Article 4(3)).

The Regulation adopted the 12 month period for signature collection of the Commission proposal (Article 5(5)), which reflected the earlier Parliament resolution. In comparative terms this seems defensible in that the collection period for agenda initiatives in EU Member States was generally six months or less.Footnote ⁷³ The Commission emphasised the need to ensure that citizens’ initiatives remain relevant and that the collection period be sufficiently long in light of the complexity of working throughout the EU. However precisely the latter rationale justified a longer period and the 18 months which some consultation contributions suggested would have been preferable.Footnote ⁷⁴ Whilst 12 months is more than any directly comparable instrument, the ECI is a novel transnational instrument for a complex political system to which citizens show great apathy.Footnote ⁷⁵ If an 18 month period for the related instrument of a citizen-initiated referendum was thought appropriate for the most experienced user of direct democracy, Switzerland, then arguably the EU should not be opting for less with its fledging instrument of direct democracy.Footnote ⁷⁶

Further crucial design issues concerned data. No uniformity was imposed as to what data should be collected from signatories. The main distinction is between those States requiring personal ID numbers (whether from a passport or national identity card), a requirement that the European Data Protection Supervisor recommended dropping, and those that do not.Footnote ⁷⁷ In principle the more data required, the more reluctant citizens will be to sign. When it comes to this personal data the organisers are, along with competent national authorities, the ‘data controllers’ for Data Protection Directive purposes (Article 12) and have obligations imposed upon them. Indeed, the Regulation specifically underscores their personal liability for any damage they cause in organising a citizens’ initiative, which will discourage the formation of citizens’ committees and therefore reduce the potential promise of the ECI.Footnote ⁷⁸

No meaningful consideration was given to public funding towards the costs of preparing an initiative and sustaining the signature gathering campaign. The Parliament’s resolution was silent and the Green Paper asserted that no specific public funding was foreseen and that this was ‘in the interest of preserving the independence and citizen-driven nature of initiatives.’ The ECI Regulation thus offers no public funding for organisers. Indeed, it provides for the Commission to maintain an online register with proposed citizens’ initiatives, and stipulates that organisers can provide the initiative in other official EU languages for this register and that such translation shall be their responsibility.Footnote ⁷⁹ It is especially striking that no translation support was provided given the translating resources at the EU’s disposal and how obviously critical this would be to gathering signatures in a transnational union with 24 official languages. Public funding for agenda initiatives is relatively rare, but even within the EU Spain provides this.Footnote ⁸⁰ This offered a valuable comparative lesson for the EU to secure a more hospitable environment for launching its new instrument of direct democracy.

A final crucial design choice concerned the obligations that flow from an initiative meeting the relevant thresholds. The legislative proposal required the Commission to examine the initiative within a set time frame and to set out its conclusions in a communication including what, if any, actions it would take and the reasons. This in itself is more than is formally required by, for example, the Italian agenda initiative.Footnote ⁸¹ Two crucial obligations were added during the legislative process. Firstly, that the Commission receive the organisers at an appropriate level so that they can explain the matters raised by the initiative and, secondly, that the organisers be given the opportunity to present their initiative at a public hearing at the European Parliament with the Commission present and other EU institutions and bodies that wish to participate (Articles 10 & 11). These valuable additions enhance the democratic promise for the ECI and generate a greater incentive to pursue initiatives as there is a tangible outcome flowing from meeting the required thresholds.

C. Assessing outcomes

Table 4 illustrates ECI activity levels over the first five years. This section is divided into two sections exploring respectively the outcomes in relation to registration refusals and registered initiatives.Footnote ⁸²

Table 4 ECI outcomes (April 2012 – April 2017).

D. The future of the ECI

Calls for ECI reform rapidly emerged and were effectively encouraged by Article 22 of the Regulation, entitled ‘Review’, which required the Commission to present a report on the operation of the Regulation to the European Parliament and the Council within three years.Footnote ¹⁰⁶ But no revision proposals were forthcoming in the eagerly anticipated review.Footnote ¹⁰⁷ The calls thus became louder and more critical, most significantly via a Parliamentary resolution in 2015.Footnote ¹⁰⁸ The punch-line of the Commission’s detailed response was that it was ‘too early to launch a legislative revision’.Footnote ¹⁰⁹ Finally, in April 2017 the Commission announced plans to revise the ECI to make it more accessible and citizen friendly with proposals expected in late 2017.Footnote ¹¹⁰

Many of the long-circulating proposals have much to commend them.Footnote ¹¹¹ Chief among these include: increasing the period of time for collecting signatures, whether that be to 18 or 24 months, given that we have seen that reaching the signature thresholds in 12 months is no small feat; reducing the age for signing initiatives to 16, as is currently the case in Austria, thus encouraging greater youth involvement with the EU and facilitating meeting signature thresholds; harmonising and reducing personal data requirements which deter citizens from signing initiatives; providing financial support for registered initiatives, and greater translation support, given the considerable expenditure involved and with a view to making the ECI more accessible;Footnote ¹¹² stipulating that the personal liability of organisers is not unlimited as the current phrasing in the Regulation discourages committee formation.

Considerably more problematic are proposals for including treaty revision within the ECI’s remit or placing more duties on the Commission with respect to follow-up of initiatives.Footnote ¹¹³ Amending the ECI Regulation to include treaty revision would, for reasons suggested above, be in tension with the Treaty text. But even if it were possible, it could have negative unforeseen consequences. Launching the treaty revision process has become a treacherous endeavour, not least because of Member State referendums. To expand the ECI to this terrain would generate unrealistic expectations as to its viability as a route to treaty change. If achieving legislative change is a perceived measure of success and has proved both elusive thus far and a cause for discontent with the ECI, for its scope to be expanded to treaty revision creates the potential for more frustration and discontent. An alternative proposal to help manage expectations would be to expressly rule out treaty revision in the ECI Regulation itself. In relation to imposing obligations upon the Commission to actually produce legislative proposals, this would be in clear tension with the Commission’s right of legislative initiative. The limited experience from the mere three to meet the signature thresholds provides insufficient basis for reconsidering the obligations on the Commission.

It is inevitable that, with a novel and complex instrument, there would be teething problems and scope for improvement. But we should not accept the increasingly standard line that portrays the first years of the ECI in a negative light with the Commission as the villain of the piece. There is an alternative more optimistic reading that is also more realistic in so far as it is shaped by awareness of agenda initiatives comparatively. The ECI is a wholly novel transnational instrument that has generated considerable activity in its first five years with over 60 proposed initiatives. This has created interactions between EU citizens in diverse Member States who have formed citizens committees, and others who may have tried unsuccessfully to do so. Registration refusals have been frequent, but this is not comparatively anomalous and was to be expected given the great complexity of the EU’s system of competences. Two significant exceptions aside – ‘Minority SafePack’ and ‘STOP TTIP’ – they have been unsurprising rejections and at least two were subsequently registered. From those registered, signed statements have exceeded seven million (if one includes the ‘Stop Glyphosate’ initiative) which strengthens awareness of the ECI.

The negative narrative flows partially from an impoverished definition of a ‘successful initiative’ as either being confined to those meeting the signature thresholds, in line with the Commission’s own unfortunate labeling in the official register, or, worse still, those which have resulted in legislative action which would mean that none have been successful. It is too simplistic to see achieving the signature threshold as the measure of success; this is a tall order even with the backing of civil society organisations. Even those not meeting the signature threshold can be part of the promise of the ECI in that they foster its cross-national deliberative and participatory aspirations. Bouza Garcia and Greenwood have rightly emphasised the ECI’s value in generating proposals from activists well outside the ‘Brussels bubble’ and contributing to creating new pan-European networks and even new organisations.Footnote ¹¹⁴ The European Ombudsman has also emphasised that a legislative proposal should not be seen as the only measure of success and that the process itself is of major significance.Footnote ¹¹⁵ Indeed the importance of views being aired in the arena of the European Parliament, even absent legislative initiatives, should not be downplayed, and even initiatives not meeting the signature thresholds can be examined by the Parliament’s petitions committee if it considers follow-up to be appropriate.Footnote ¹¹⁶

Ultimately, five years is very early days in the life of the first EU level direct democracy instrument. We can expect the revision process to result in an even more fit for purpose agenda initiative for the EU. But the ECI is in any event likely to become a powerful agenda initiative because of the political setting in which it is located. The EU’s political malaise and democratic disconnect vis-à-vis its citizens is such that there is enormous pressure to exploit all opportunities to combat this disconnect. And in an era where major treaty revision is no longer the order of the day, the existing treaty-enshrined instrument that is the ECI can make at least a modest contribution to addressing this disconnect. This promise of the ECI is dependent on political practice. If one type of political practice is seen to jeopardise the promise of the ECI, Commission reluctance to follow-up initiatives, there is scope for another type to contribute to that promise, namely, Parliament adding its weight, as it did with the ‘Right2Water’ initiative, to calls for action and if necessary backing initiatives via its own right to call for a legislative proposal. Footnote ¹¹⁷ In conclusion, the future of the ECI thus looks bright.