The European Union: Consociational Past, Centripetal Future?

ABSTRACT

Consociational interpretations of the European Union (EU) are well established and help to explain the political stability of the 27-member state system. In contrast, the increasingly common centripetal elements have not yet received systematic attention. Using a framework originally designed to map the choices for divided societies, this article highlights centripetalism in EU party regulation and proposals for electoral reform. Going beyond the spatial distribution requirements that play such a central role in aggregative institutions in the EU, the article suggests that cross-national districts rather than a supra-national district provide the strongest incentive for European parties to organise EU-wide campaigns on European issues, fielding candidates with cross-national appeal. The article concludes with a reflection on the relationship between consociational and centripetal elements in the EU. It shows that consociationalism and centripetalism in the EU can continue to co-exist, though the balance is likely to change.

KEYWORDS:

• European Union
• consociationalism
• centripetalism
• elections
• parties

Introduction

Inspired by Lijphart’s (1977) theory of democracy in plural societies, scholars have highlighted how a culture of compromise in European decision making has helped to provide political stability (Bogaards, 2002). In the debate about the EU’s democratic deficit (Neuhold, 2020), consociational accounts stand out by their focus on state elites, intergovernmental bodies, and a general wariness of politicisation and democratisation.¹ As a consequence, the consociational literature has little constructive to say about the European Parliament (EP) and European elections and offers no thoughts on what many see as the key to democracy at the European level: a European party system.

Broadly speaking there are three scenarios for the development of European parties (Andeweg, 1995). First, a continuation of the present situation of increasingly strong party groups in the EP and weak party federations that contest EP elections with a mostly national character. This scenario implies that EP elections remain second-order: elections perceived by voters to be less important because there is less at stake (De Sio, Russo, & Franklin, 2019; Reif & Schmitt, 1980; Schmitt & Toygür, 2016). As a consequence, participation is lower, government parties tend to lose, and protest parties win as voters register their displeasure with national politicians without the constraint of strategic voting for government formation. The 2019 EP elections were again ‘essentially twenty-eight separate national elections for representatives to the same supranational institution’ (Mudde, 2019, p. 20).²

The second scenario is a so-called ‘split-level’ party system in which some national parties only contest EP elections, not domestic elections. These EP-only parties are usually Eurosceptic. They increase voter choice but do so in the wrong arena, contributing to what Mair (2000) calls a ‘displacement’ of issues. National elections should be about the process of European integration because it is member-state governments who decide on Treaty revision and take the most important decisions in the European Council. European elections, in contrast, should be about the substance of European policy making in accordance with the increasing legislative competencies of the EP.³

Most promising is the third scenario with ‘supranational parties presenting the voters with a relevant European choice’ (Andeweg, 1995, p. 69), resulting in a transnational party system (Bardi et al., 2010). Such a development is not expected to come about by itself but is linked to institutional reform. It is these reforms that are the focus here.

This article speaks to two audiences. First, participants, observers, and analysists of EU politics, who have produced a steady stream of proposals on how to reform EP elections. Their discussions have focused on proportional representation as the common principle for electing Members of European Parliament (MEPs), the adoption of a uniform electoral law, and the introduction of an EU-wide electoral district.⁴ The second audience of this article are scholars of institutional design in diverse, divided, and post-conflict societies. Their exchanges have focused on the (de)merits of proportional representation, which allows for the political organisation and inclusion of socio-cultural diversity, versus centripetal electoral systems that promote aggregation and broad-based parties.

The article’s message to the first audience is: current reform proposals for EP elections are probably not enough to produce the desired outcome of a European party system. Concretely, cross-national districts rather than a supra-national district provide the strongest incentive for European parties to organise EU-wide campaigns on European issues, fielding candidates with cross-national appeal. The article’s message to the second audience is: consociational interpretations of the EU have ignored the centripetal logic in EU party regulation and proposals for electoral reform. The EU’s combination of consociational and centripetal elements requires a rethink of its functioning and future. Overall, and in line with Bieber and Bieber (2021, p.11), ‘the argument is about establishing a dialogue between two schools of scholarship, namely scholars of the EU and those studying diversity and divided societies’.⁵

The article begins with two brief sections on elections to the EP and parties at the European level. This is followed by a discussion of consociational interpretations of the European Union, highlighting both their conservatism and possibilities for change. The next two sections document the central role of centripetal elements in existing rules and regulations as well as reform proposals. The article then suggests a novel electoral system for the EP, constituency pooling in cross-national districts, and examines its promises and pitfalls. The article concludes with a reflection on the relationship between consociational and centripetal elements in the EU. Building on Bogaards (2019), it shows that consociationalism and centripetalism in the EU can continue to co-exist, though the balance is likely to change.

Electing the European Parliament

Article 138(3) of the Treaty of Rome required the European Parliament to ‘draw up proposals for elections by direct universal suffrage in accordance with a uniform electoral procedure in all Member States’. The Amsterdam Treaty of 1997 also allowed for elections ‘in accordance with principles common to all Member States’. In practice, ever since the first direct elections in 1979, EP elections have been conducted under provisions internal to each member state. Broadly speaking, the reform efforts of the 1980s and 1990s had two objectives. First, to establish proportional representation as the only electoral system for European elections. This finally happened in 2002. Second, to combine overall proportionality with some form of constituency representation. In order to bring MEPs closer to the electorate, the EP proposed to make territorial constituencies compulsory in countries with more than 20 million inhabitants (Anastassopoulos, 2002). However, the Council, which ultimately decides how the EP is elected, did not adopt this proposal.

No major changes to the Electoral Act, by now a ‘rather messy document’ (Wouters, 2020, p. 12) and an ‘anachronism’ (De León, 2017, p. 353), have been agreed to since. The 2019 elections to the EP were conducted under 28 different electoral systems. The 28 national electoral laws differ in every conceivable respect except for their membership of the broad family of proportional representation. They differed in list proportional representation (PR) versus the single-transferable vote (STV), the precise electoral formula used, the use of preferential voting, the ballot structure, national versus regional constituencies, the size of electoral districts, the entitlement to vote, eligibility to stand, and the day of voting within the four-day window. To say that the electoral systems are ‘somewhat less than uniform’ (Farrell & Scully, 2005, p. 982) would seem to be an understatement, also because the systems differ in the degree to which they approximate the ideal of proportionality (Suojanen, 2006). If further harmonization is more than a goal in it self, it should be remembered that nothing in a uniform electoral law prevents national parties from fighting European elections as national contests (Bardi, 1990, p. 522; Hix, 1998, p. 35) and indeed nothing in a uniform electoral law encourages national parties to become transnational in their organisation and electoral appeal.

Parties at the European Level

In the Maastricht Treaty of 1992, an article on European political parties was inserted: ‘Political parties at the European level are important as a factor for integration within the Union. They contribute to forming a European awareness and to expressing the political will of the citizens of the Union’ (article 191 of the EC Treaty). The phrase ‘political parties at the European level’, while awkward, is more appropriate than the term ‘Europarties’ (as used by, for example, Bressanelli, 2014; Switek, 2016), because there are no European parties, only national parties that work together within a European, mostly EU, context. Two forms of cooperation can be distinguished. First, European party groups, which operate inside the EP. Second, European party federations, which are involved in elections to the EP. There is no one-to-one correspondence between a party group and a party federation: a group may consist of MEPs from various party federations and the members elected from the same party federation may sit in the EP for different party groups (Calossi, 2016).

Party groups resemble party caucuses in the national parliament and they emerged for the same reason: to improve the efficiency of the proceedings and facilitate decision making (Lindberg, Rasmussen, & Warntjen, 2008). From the beginning, ‘there was a general agreement amongst Members that nationality should not be the primary basis of internal organisation’ (Kreppel, 2002, p. 186). Instead, political groups formed that reflected the main European political families. These ‘party groups in the EP constitute a highly developed, relatively stable and reasonably competitive party system’ (Hix, Kreppel, & Noury, 2003, p. 327). The party groups have shown remarkable cohesion in voting (Mühlböck, 2012) and have adapted well to the influx of new national parties after enlargement (Bressanelli, 2014). After the 2019 EP elections seven political groups were formed.

Compared to the functions performed by classic parties, the role of party federations ‘is rather limited’ (Van Hecke, 2017, p. 12). They draft the election manifestos, but the campaigns are run by national parties, who also nominate the candidates. After the elections, the MEPs sit in parliament for their national party. In-between EP elections, the party federations are mostly active through party congresses and summits that bring together members of the party federation from the various EU institutions, especially the Commission and Council. Overall, the electoral connection is judged to be somewhere between ‘extremely weak’ (Ripoll Servent, 2018, p. 212) and ‘almost nonexistent’ (Hix & Høyland, 2013, p. 184).

Consociational Interpretations of the European Union

The literature on democracy in divided societies is dominated by consociationalism and centripetalism (Bogaards, 2019). Consociational accounts of the EU go back to the mid-1970s (Bogaards, 2002). They see member states as the European-level equivalent of the segments in divided societies and they look for evidence of the four consociational principles of a grand coalition, proportionality, mutual veto, and segmental autonomy in the EU. These are easy to find (see, for example, O’Leary, 2020, p. 35). The two most powerful decision-making bodies in the EU consist of representatives of the member states: heads of government in the European Council, ministers in the Council of Ministers. The allocation of positions and resources tends to be proportional and any deviation normally benefits smaller and less well-off member states. Segmental autonomy comes in the form of state sovereignty and the principle of subsidiarity. Finally, treaty revisions require unanimity, the European Council strives for consensus, and Qualified Majority Voting in the Council of Ministers requires the support of at least 55% of the member states representing at least 65% of the population.

There is no agreement on what kind of consociation the EU is: an inter-state consociation (Costa & Magnette, 2003), a federal consociation – at least in the making (Piattoni & Verzichelli, 2019), a confederation with consociational decision-making (O’Leary, 2020), an ‘extreme variant’ of Swiss-style politics (Blondel, Sinnott, & Svensson, 1998, p. 252), or even ‘a rather extreme case of a consociational democracy’ as such (Schmitter, 2003, p. 71).⁶ For analytical purposes, it is most fruitful to think of the EU as a consociation of states, as this allows a simultaneous appreciation of similarities and differences with a classic consociation of (subnational) segments (Bogaards, 2002). Take the view of the EU’s origin as a ‘consociative pact’ between states (Costa & Magnette, 2003, p. 9). On the one hand, this evokes the self-negating prediction of consociational theory, in which segmental elites opt for collaboration rather than competition in the realisation that the latter would tear the system apart (Bogaards, 1998). On the other hand, there was no system before the member states formed the European Coal and Steel Community. The member states therefore not only decided on the rules of the game (consociational) but on the very constitution of a common game. This is crucial because it suggests that unless the game is played by consociational rules, there will not be a game at all. In other words, the EU is consociational or it is not. Unless, of course, ‘European society’ would undergo the kind of desegmentation that made consociationalism functionally obsolete in a country like the Netherlands (See Bogaards, 2020). But even in a ‘post-national Europe’, the EU would resemble a consensus democracy (Lijphart, 2012).⁷

Because they take states and state interests as their starting point, consociational interpretations of the EU have an affinity with intergovernmentalism (Bogaards, 2002). This has two consequences. First, it introduces conservative bias: any signs of majoritarianism will be perceived as not just challenges to the consociational nature of the EU but its very existence (See Costa & Magnette, 2003). The most extreme example comes from Lord (2004), who makes ‘modified’ consociationalism the benchmark for his democratic audit of the EU. The four consociational principles of grand coalition (inclusion), proportionality, segmental autonomy, and mutual veto are elevated to the status of ‘tests’.⁸ In other words, democracy in the EU is equated with consociationalism, which in turn is equated with the protection of the interests of member states.

Second, the intergovernmental core of consociational interpretations of the EU opens up a possibility for change.⁹ If the EU’s main supranational institutions, the European Commission and the EP, are not normally seen as part of the interstate consociation, then changes within and among these institutions would not directly affect the consociational character and functioning of the EU at large. Politicisation is ‘both possible and desirable’, argue Papadopoulos and Magnette (2010, p. 714; p.717), because ‘even if politicised, the EU would remain a negotiation democracy’.¹⁰ Similarly, while Thomassen and Schmitt (1999, p. 266) caution against majoritarianism, they see consensus democracy in the EU as compatible with European parties competing for votes of a European electorate.

Centripetal Practice in the EP and EP Elections

In contrast to the consociational literature, centripetal scholars have ignored the EU. Reilly’s (2018) recent overview of centripetal institutions in established democracies does not include the supranational level. Lacey’s (2017) book on centripetal democracy in Belgium, Switzerland, and the European Union uses a notion of centripetalism as demos formation that is much more ambitious than moderation and aggregation, the usual goals of centripetalism. As a result, so far, the central role of centripetal elements in the EP and in reform proposals for EP elections has not been noticed.

Table 1 presents an overview of three centripetal practices and three centripetal proposals, or proposals with centripetal elements.¹¹ The EP’s Rules of Procedure regulate the formation of parliamentary groups. These rules have been tightened over time, from at least 19 members elected in at least one-fifth of the member states to at least 25 members from one-quarter of the member states.¹²

Table 1. Centripetal practices and proposals in (elections to) the European Parliament.

	Centripetal mechanism	EU institution
Centripetal practice	Spatial distribution requirement (seats)	European Parliament Party Group Composition
	Spatial distribution requirement (votes)	European Party Statute for Party Federations
	Spatial distribution requirement (parties)	‘Spitzenkandidaten’ nomination within European Party Federations
Centripetal proposal	Spatial distribution requirement (‘threshold cascade’)	European Parliament elections
	Spatial distribution requirement (votes)	European Parliament elections
	Supra-national district with spatial distribution requirement (candidates)	European Parliament elections

Source: Own compilation, see text.

To strengthen European party federations and above all make them financially independent from the party groups in the EP, in 2004 a Regulation on European parties entered into force (Day, 2014). Since 2007 the Regulation also funds foundations linked to party federations and in 2014 an Independent Authority was established that handles party registration and monitors compliance with the rules (Van Hecke, 2017, pp. 18–19). The budget has continued to increase and has recently reached 30 million Euro (Wolfs & Smulders, 2018).

The Regulation stipulates that a ‘political party at the European level’ must ‘have received at least three per cent of the votes in one quarter of Member States at the most recent European Parliament elections’.¹³ These criteria were controversial (Day & Shaw, 2004) and continue to be so. Lightfoot (2006, p. 310) calls them ‘the main stumbling block for the creation of new parties’ and warns that smaller party federations will be ‘more vulnerable to the vagaries of the European electorate’. However, so far, no parties have been denied registration or removed from the register because they failed to satisfy the geographic distribution requirement.¹⁴

At present, the President of the Commission is nominated by the European Council and elected by the EP. Since the 2014 EP elections, most party federations have nominated their candidate(s) for the position of president of the Commission. In 2014, the internal nomination of three of the six ‘Spitzenkandidaten’, or simply presidential hopefulls, was affected by distribution requirements. Five years later, the number was down to two (Wolfs, Put, & Van Hecke, 2020). Interestingly, the actual voting procedure did not take geographical spread of support in account.¹⁵

Centripetal Proposals for EP Elections

The first centripetal proposal listed in Table 1 is Pukelsheim and Oelbermann’s (2014) idea of a ‘threshold cascade’, whereby the national threshold that national parties need to pass in order to win national seats in the EP is lowered depending on how well the same party does in other member states. So, 5% of the valid vote in one member state, 4% of the valid vote in each of two member states, etc. This design is explicitly meant to ‘encourage the formation of unionwide parties’ and is implicitly centripetal (Pukelsheim & Oelbermann, 2014, p. 557).¹⁶

Borrowing from the literature on electoral system design in divided societies, Schleicher (2011, p. 116) proposes an electoral threshold with a geographical distribution requirement: at least three percent of the votes cast in a quarter of the member states.¹⁷ Electoral alliances among national parties are not allowed: it must be the same party that is on the ballot in the various member states. Schleicher sees this as a deliberately centripetal incentive that will make EP elections more European and thereby more meaningful.¹⁸

The most far-reaching proposal currently on the table calls for partial EU-wide electoral lists. Duff (1996), on behalf of the European Movement, advocated an additional quota of five percent of MEPs elected on a European Union basis. In the EP, the idea was introduced by Anastassopoulos (See Anastassopoulos, 2002). In 2000, the Committee on Constitutional Affairs of the EP proposed ‘that the Treaty should provide for the possibility that, as from 2009, an additional number of Members corresponding to ten per cent of the Members of Parliament could be elected in a single European constituency, by giving each voter two votes – one for the national lists and one for the European lists; the European lists shall comprise at least one citizen of each Member state’.¹⁹ The Commission was supportive, hoping that ‘organising the European elections in this way would encourage the development of Europe-wide political parties and produce members who could claim to represent a European constituency rather than a purely national one’ (See also Raunio, 2002).²⁰ However, the Council did not go along.

Andrew Duff, by then an MEP, produced two reports between 2009 and 2011 that contained plans for an EU-wide constituency with 25 seats, but these were not adopted by the European Parliament, mainly due to opposition from the European People’s Party (EPP), the largest party group in the EP (De León, 2017; Hbrek, 2019). Brexit briefly revived the discussion. French president Macron suggested the 73 British seats should be used to create an EU-wide constituency and Commission President Juncker supported the idea of transnational lists in his State of the Union in 2017. However, in 2018 the EP did not approve the Hübner-Perreira report adopted in its Constitutional Committee, thereby ending the third wave of electoral reform in the EU (Charvát, 2019, p. 31): the introduction of a pan-European constituency with transnational lists.²¹

Its proponents expect a lot from electoral reform. For Duff (2017, p. 3), ‘the introduction of European lists would at a stroke Europeanise the European elections and re-invent the Parliament’. For Bartl (2020, p. 60), ‘there are few single measures that can bring about such a political transformation as transnational lists’. However, whether these advantages materialise depends on several factors. First, how much of the EP is elected in an EU-wide district.²² The use of an additional contingent of EP seats elected in an additional contest would create two types of MEP: those elected nationally and those elected EU-wide. Because ‘national’ MEPs and ‘European’ MEPs would have different masters, this is likely to complicate and confuse accountability relationships, both inside the EP and with the voters. Moreover, the objection of Horowitz (2009) to the idea of a small federal constituency on top of Belgium’s linguistically separated party systems would also apply here: the number of deputies is too low to change the overall dynamics. Finally, in order to have a centripetal impact, an EU-wide district would need to be accompanied by a geographic distribution requirement (Stojanović & Bonotti, 2020, p. 608). The French position on the latest EP proposal for an EU-wide constituency from 2018 acknowledges this, requiring candidates to be registered ‘on the electoral roll in at least one third of EU Member States, with the percentage of candidates from a single country not exceeding 25%, the seven top candidates coming from different countries’ (Verger, 2018, p. 8).²³

Even if one single EU-wide district with spatial distribution requirements for mandatory transnational lists were agreed to, three problems remain. First, the size of such a continent-wide district would easily surpass South Africa as the largest PR-district worldwide. This would magnify problems of geographical linkage, problems South Africa has been unable to solve (Bogaards, 2018). Second, experimental research suggests that vote choice in an EU-wide district ‘would be substantially affected by the presence of national candidates on the lists’ (Bol et al., 2016, p. 539).²⁴ Third, an EU-wide district does not necessarily strengthen European party federations or transnationalize elections. This is most easily seen in the 2018 Hübner-Perreira report, which would have allowed lists submitted by national political parties or movements not affiliated with a European political party, effectively weakening the role of European party federations (Hoffmeister, 2020, p. 134).²⁵

The most radical solution is to ban national parties from contesting EU-wide elections, as implied by Van Hecke (2017, p. 46, p.48) and only allow European party federations on the ballot (Jouvenat, 2018).²⁶ Even such extreme measures, however, only deal with the supply side and not with the demand side, leaving open the possibility that parties and candidates get voted into the EP with the support of a narrow geographical constituency, not EU-wide endorsement. Is there an electoral system that can do better?

Constituency Pooling in Cross-national Districts

The favourite centripetal electoral systems are the alternative vote (AV) and the single-transferable vote (STV). Both ask voters to rank candidates in the order of preference. Winners are decided by counting first and, if necessary, lower preferences. AV and STV mainly differ in the treshold for election: an absolute majority for AV, a majoritarian electoral system, versus a much lower threshold (the higher the number of seats at stake in an electoral district, the lower the number of votes needed to get elected) in the case of STV, a proportional electoral system. AV and STV only promote vote pooling in heterogeneous electoral districts, where no group has a majority and candidates have to reach out beyond their own group to win elections (Bogaards, 2003; Reilly, 2001). However, in the European Union, due to its sheer size and the huge differences in population, it would be impossible to carve out geographically contiguous electoral districts with voters from multiple member states, except for border regions.

One way to integrate the European electorate(s) would be through cross-national districts. The idea of ‘constituency pooling’ goes back to the Ugandan electoral law of 1971, designed to overcome regional, ethnic, and religious differences (Bogaards, 2003). The basic idea is that candidates compete simultaneously in constituencies in different parts of the country/political system. To decide the winner, the votes from all constituencies in which the candidates competed are pooled. In this way, no candidate can win just with the vote of their own part of the country/political system. Instead, each candidate needs to appeal to voters beyond their home constituency. To appreciate what constituency pooling might look like in the EU, is it necessary to have information on two decisions. First, how are the cross-national districts carved out? Second, how are the winners decided?

With around 345 million eligible voters and 705 seats in the current EP, each MEP represents about half a million voters. On that basis, the EU could be divided into 705 constituencies with half a million voters each. Akin to the administrative bodies in charge of (re)districting in countries like the UK and USA, an EU electoral boundary commission could supervise this process. To elect MEPs, these 705 constituencies would then be pooled from across Europe. There are various ways to do this. One option is to divide the EU into regions or zones. The number of zones should be higher than two so as not to create a bifurcation and probably should not exceed six, as there is a practical limit to how many constituencies a candidate can cover and voters can place on their ‘mental electoral map’. The precise boundaries of these zones are not important, but they should be of roughly equal size in terms of eligible voters. Now let’s suppose the EU is divided into four zones: North, West, South and East. For the purpose of elections then, constituencies from each of these four zones will be pooled, creating a cross-national district. There are different ways to pool constituencies: by chance or according to criteria, but what matters is that the MEP(s) will represent a cross-national district.

As can be seen in Table 2, constituency pooling is compatible with different electoral formulas and can be used to elect a single person as well as multiple seats.²⁷ Depending on the particular combination of district magnitude (one or multiple MEPs) and electoral formula (FPTP, AV, STV, list PR), cross-national districts will have different advantages and disadvantages. Non-proportional electoral formulas, which are not allowed any longer for use in EP elections, are mentioned for the sake of illustration only. It is easy to imagine how constituency pooling works when voters from four different parts of Europe directly elect their MEP using plurality (FPTP). AV in combination with cross-national districts would provide the strongest centripetal incentive, but its disproportional outcomes and unusual format hinder its adoption.

Table 2. A comparison of the features of supra – and cross-national districts in EP elections.

Type of district	Scope	District size	Formula	Advantage(s)	Disadvantage(s)
Supra-national	Partial EP	Multiple seats	PR	(Depends on details)	- Creates two types of MEP - Limited impact - Geographic distribution requirement needed
	Whole EP	Multiple seats	PR	(Depends on details)	- No geographical linkage - Voters can keep voting for national candidates - Role of national parties does not necessarily change - Geographic distribution requirement needed
Cross-national	Whole EP	Single seat	FPTP	- Familiar formula - Direct geographical linkage - Strengthens European parties - Requires EU-wide campaign	- Disproportional outcomes - Winning candidate does not need a majority
			AV	- Strongest centripetal incentive - Direct geographical linkage - Strengthens European parties - Requires EU-wide campaign	- Disproportional outcomes - Unfamiliar formula
	Whole EP	Multiple seats	STV	- Doubly centripetal - Proportional - Geographical linkage - Strengthens European parties - Requires EU-wide campaign	- Allows voters to support national candidates - More complex
			Open list PR	- Proportional - Geographical linkage - Strengthens European Parties - Requires EU-wide campaign	- Allows voters to support national candidates
			Closed list PR	- Proportional - Geographical linkage - Strengthens European parties - Requires EU-wide campaign

Legend: PR = proportional representation, FPTP = first-past-the-post, AV = alternative vote, STV = single transferable vote.

A more realistic scenario sees MEPs elected in multi-seat districts that are pooled across Europe, with the winners decided by some type of proportionality. For example, constituency pooling could be combined with Hix and Hagemann’s (2009) proposal for relatively small electoral districts, by which they mean four to ten seats.²⁸ The the EP could be elected in 175 multi-member districts, each consisting of four members This cross-national district would pool constituencies from the four zones in which the EU is divided. Together, four times half a million voters from four constituencies from four different parts of Europe would elect four MEPs in one cross-national district. If a larger district size is preferred to increase proportionality, the EP could be elected in 88 multi-member districts with eight members each.²⁹ Within each cross-national district, party competition will follow the usual lines of division, except that only European parties will have the capacity to nominate candidates and run campaigns across Europe. One important decision is whether to allow European voters a choice of candidate and not merely party. The advantage is a stronger link between voter and MEP, the disadvantage is that it allows for national favouritism, though this is more of a liability in open-list PR than STV, with its doubly centripetal incentives.

As can be seen in Table 2, cross-national districts have several advantages. First, they promote the development of genuine European parties as these are the only organisations with the capacity to conduct multiple campaigns in geographically non-contiguous electoral districts across the EU. Second, constituency pooling promotes a focus on cross-national issues, that is, issues that by reason of interdependencies or parallel developments (for example, unemployment) or because of their non-bounded nature (for example, the environment) occur Europe-wide. To the extent that there is already a European electorate in the sense of shared preoccupations among European citizens that transcend borders (Van der Eijk & Franklin, 1996), constituency pooling strengthens it and stimulates the formulation of policies and programmes that address European issues. Third, constituency pooling is the electoral version of the well-known practice of partner cities. Electorates across Europe will become aware of their similarities, and their differences, when they are grouped together to elect common members in the EP. Fourth, constituency pooling strengthens linkage not only between European electorates but also between candidate and voters. To what extent MEPs will develop ties to and establish a presence in their constituencies will most likely vary, depending on resources and interests. But only constituency pooling offers the kind of linkage that is seen as desirable but difficult to organise in the EU (Cf. Farrell & Scully, 2010). Or, more precisely, only constituency pooling offers linkage that is transnational in nature. Fifth, constituency pooling requires the kind of transnational linkages in political communication that currently are found largely missing (Stier, Froio, & Schünemann, forthcoming). Finally, constituency pooling is an effective antidote against fears that any one country can dominate the EP.

Some potential problems should be flagged. First, to facilitate learning on the part of both candidates and voters, the pooling of constituencies should change as little as possible from one election to the next. Enlargement and population change, however, mean that some change in the combinations of constituencies is inevitable. Second, if one interprets article 190 as a right to national representation in the EP, constituency pooling violates this principle. However, the change from ‘representatives of the peoples of the States brought together in the Community’ to ‘representatives of the Union’s citizens’ in the Lisbon Treaty opens the door to a different interpretation. Finally, constituency pooling asks a lot of MEPs. To campaign across Europe and represent constituencies from across Europe, MEPs would have to rely heavily on party organisation, also to communicate with voters in their own language. It is precisely for this reason that cross-national districts provide such a strong incentive for the strengthening of EU-wide parties that effectively connect the European and the national/local level.

Conclusion

This article has revealed the central role of centripetalism in measures to promote European parties. Spatial distribution requirements, a centripetal instrument, apply to party groups in the EP, funding for European party federations, and intra-party rules on the nomination of ‘Spitzenkandidaten’. Spatial distribution requirements are also common in proposals for electoral reform, including a supranational electoral district. In all cases, the purpose is to promote a party system that transcends national boundaries and the mechanism is a requirement for parties to nominate candidates, compose lists, form parliamentary groups, establish party federations, and attract the support of an electorate that reflects the diversity of member states in the EU.

However, stronger incentives might be needed. A uniform electoral law or a (supplementary) supranational constituency will not by themselves bring a European party system closer. Cross-national districts, which Reilly (forthcoming) classifies as strong centripetalism, would.

If the EU’s past was consociational, is its future centripetal? For the time being, elements of both co-exist. This is not unusual. Bogaards (2019) shows how consociational and centripetal elements are often found together. The question then is: are they friends or foes? That is, do they weaken or do they reinforce each other’s main goals of representation and inclusion (consociationalism) versus aggregation and moderation (centripetalism).³⁰

Consociational interpretations of the EU have been static and conservative, in the case of O’Leary (2020) even regressive, but their focus on intergovernmental institutions (the Council and European Council, especially) also provides an opening: if supranational institutions are not central to the consociation of states, then centripetal tendencies within and between the EP and European Commission are compatible with continued consociation. In other words, within the EU consociational and centripetal elements are neither friends nor foes but ‘flatmates’ who share the same house, but occupy different spaces and have little interaction. This explains why many scholars who on the one hand insist on the inevitability of a consensual style of politics in the EU can on the other hand advocate politicisation and democratisation.³¹

Can flatmates turn into foes? Can the EU’s consociational core be shielded from the effects of politicisation and democratisation? Through co-decision, the Council is directly affected by changes in the composition and agenda of the EP. Any strengthening of the EP would likely come at the expense of the legislative role of the Council. Also, politicisation would threaten the unique character of the Commission as a politically neutral agency that promotes a culture of compromise (in the view of Costa & Magnette, 2003).³² The existence of a union requires consociationalism. Ever closer union requires ever stronger centripetalism. The question is whether the result will be a ‘vital democracy’ (Hendriks, 2010) that thrives on the tension between competing principles, or a ‘semi-parliamentary’ and ‘semi-consociational’ European democracy (Peters, 2003) that pleases no one.

Acknowledgments

A previous version of this paper was presented at the 2nd St. Petersburg Conference on Inequality and Diversity in November 2020. The author thanks the audience as well as the two anonymous reviewers of this journal for their helpful comments. The author also thanks the editors of Representation and the guest editor of the special issue on Quotas, Parties and Electoral Design: Mechanisms and Effects of Ethnic Representation in Diverse Societies, Netina Tan, for their support.

Disclosure statement

No potential conflict of interest was reported by the author.

Additional information

Notes on contributors

Matthijs Bogaards

Matthijs Bogaards is associate professor in the Department of Political Science at the Central European University in Vienna, Austria. Since obtaining his PhD from the European University Institute in Florence, Italy, he has been on the faculty of universities in the UK, Germany, and Hungary. He has published widely on Electoral Systems, Political Parties, Democracy in Divided Societies, Democratization, and African Politics.

Notes

1 On politicization, see Zürn (2019).

2 In Neunreither’s (2000, p. 147) analogy, “the situation resembles that of the German Bundestag or the French Assemblée nationale being composed solely of members from regional parties”.

3 Braun, Hutter, and Kerscher (2016) provide evidence that EP elections are increasingly about European policies.

4 Members of the European Parliament as well as of the national parliaments overwhelmingly favoured a uniform electoral law for EP elections (Wessels, 1999, p. 217).

5 Despite this, Bieber and Bieber (2021) neither mention nor engage with consociational interpretations of the EU.

6 Fabbrini (2017) regards the EU as neither a competitive nor consensual democracy but highlights its compound character.

7 Hale and Koenig-Archibugi (2016) miss this point. They claim that because survey data show no public "deeply divided on policy preferences along national or regional lines", consociational fears about the disruptive effects of greater contestation over political leadership and policy are unfounded. They expect and endorse a change from a consociational to a (more) majoritarian model of democracy. More likely, though, is a change among non-majoritarian types, from consociational to consensus democracy. On the distinction between consociational and consensus democracy, see Bogaards (2000).

8 Lord’s (2004) modification is that he ads accountability of national representatives in the EU to national publics and parliaments as a criterion.

9 In expert comments to an EP committee (recounted in Pukelsheim & Oelbermann, 2014, p. 551) and in a recent academic publication, O’Leary (2020, p. 37) recommends to go back, at least in part, to the days before direct elections when the EP was filled with representatives of the member state parliaments. It is not clear whether this recommendation follows from his confederal or his consociational reading of the EU, as O’Leary (2020) does not clarify the relationship between the two.

10 Similarly, Pelinka (2010, p. 90) and Scharpf (2015, p. 271) see no alternative to a consensus style democracy in the EU. The question remains, though, whether the interests of a consociation of states can be sufficiently protected in a consensus democracy.

11 Geographic distribution requirements also play a role in the practice and prospect of direct democracy in the EU. The European Citizens’ Initiative (ECI), introduced by the Lisbon Treaty, has a geographic distribution requirement. However, Weisskircher (2020, p. 799) shows that “the ECI’s institutional design does not incentivize the organization of truly Europe-wide campaigns” and that in fact “a selective focus on national mobilization is the most efficient approach”. Papadopoulos (2005) advocates a popular initiative on EU legislation with geographic distribution requirements for signatures, votes, and turn out.

12 In 2001, a Political Group in the European Parliament was dissolved, not because it failed to be broad enough geographically, but because it was too broad ideologically, lacking the “ideological affinities” required of the members of a Political Group (Settembri, 2004).

13 An alternative route is to have elected members in European, national, or regional parliaments in at least five member states.

14 Hix and Lord (1997, p. 216) suggest that EU subsidies for the EP election campaigns of national parties be made conditional on the use of European party manifestos in their election campaigns and cooperation with their federations in the selection of candidates.

15 For some (See, for example, Bogdanor, 1986; Hix, 1998), only the direct election of the Commission President will facilitate genuine European elections with cohesive European parties. To make sure the winning candidate has support across the EU, a geographic distribution requirement would be needed, following the examples of Kenya, Nigeria and Indonesia (Bogaards, 2008), where a successful presidential candidate has to draw a minimum level of support from a minimum number of geographical units.

16 So far, national parties only contest elections nationally and it is not clear what would count as “the same party” abroad.

17 Geographic distribution requirements in parliamentary elections are rare. Bogaards (2007, p. 183) discusses the example of the Comoros, where the 1992 constitution required parties to win at least two deputies from each of the three islands in order to take up their seats in the national parliament.

18 Bright, Garzia, Lacey, and Trechsel (2016) propose to let national parties compete EU-wide, what they call “transnationalising Europe’s voting space” (p.185). They do not explain how this would work in practice, for example how votes are added up across European member states or how seats are allocated.

19 This report can be found under number A5-0086/2000 at: https://www.europarl.europa.eu/sides/getDoc.do?pubRef=-//EP//NONSGML±REPORT±A5-2000-0086±0±DOC±PDF±V0//EN&language=EN. The quote is from p.5.

20 See: http://europa.eu.int/comm/archives/igc2000/offdoc/opin_igc_en.pdf.

21 The first two waves were the introduction of direct elections and the adoption of common electoral systems (Charvát, 2019; see also Costa, 2016).

22 The highest percentage is 50% of the seats in the EP, proposed by French president Macron for 2024 in his 2017 Sorbonne speech (Hbrek, 2019, p. 271).

23 Hoffmeister (2020, p. 144) adopts this geographical distribution requirement in his plea for an EU-wide EP district with 150 seats.

24 Likewise, Bright et al. (2016, p. 192) find that in EP elections, only eighteen per cent of their respondents would vote for a party from another EU member state, even if that party would be (much) closer to their preferred policy position.

25 Even the adoption of “pre-established European criteria”, presumably a geographical distribution requirement, would not be able to undo the damage to the overall goal of creating a European party system.

26 In fact, Jouvenat (2018) thinks an EU-wide district is not necessary when only European party federations can contest EP elections.

27 The options listed in table 2 are not exhaustive.

28 In fact, the current system is one of “degressive proportionality”, overrepresenting the electorates in small member states through a minimum of six EP seats per country and underrepresenting the electorates in large member states through a maximum of 96 in a way that is less than transparent (Pukelsheim & Grimmett, 2018).

29 In that case, two constituencies would be pooled from each of the four zones in which the EU is divided.

30 Trzciński’s (2018) term “hybrid power sharing” should not be adopted because centripetalism is not a form of power sharing but an alternative to it (Bogaards, 2000). In his empirical analysis, Trzciński (2018) overstates the importance of consociational elements in Nigeria (see Bogaards, 2010) and Indonesia.

31 It is not clear whether direct democracy in the EU is a consociational foe. Based on the Swiss experience, Papadopoulos (2005, p. 461) is hopeful that “referendums would be integrated into the overall consociational framework”. However, Plottka (2016) contrasts the conflict logic of the initiators of ECIs with the consensual logic typical of EU institutions.

32 Centripetalism and non-majoritarian models of democracy in the EU can also be friends. Decker (2002, p. 267) predicts that centripetal rules for a direct election of the Commission President would buttress consensus democracy in the EU.