A Step Closer to a Transnational Party System? Competition and Coherence in the 2009 and 2014 European Parliament

At this stage of European integration and given the high degree of Europe's politicization and salience caused by the recent global financial crisis, representative democracy in the EU can only function if parties mobilize beyond borders. We examine whether European Party Groups (EPG) in the European Parliament (EP) offer distinct policy alternatives and how coherent these are. We use party position data collected by two Voting Advice Applications designed for the 2009 and 2014 EP elections, respectively (EUProfiler and Euandi). We find evidence of competition between EPGs groups on both left right issues and European integration; on the latter issue, there is greater differentiation within the anti&#8208;EU camp. Coherence within EPG exists, though it varies across issues, EPGs and between election years examined: it is greater on European integration than on left&#8211;right issues and it is particularly high for right wing eurosceptics though for most parties it deteriorates between 2009 and 2014.


Party Democracy in Crisis 1 ?
The project of co-operation and integration devised to cure the continent from nationalism, racism and antagonism between European states is going through tough times. The European Union (EU) shows signs of disintegration and its future looks bleak. One consequence of the current economic crisis management is that citizens perceive politics as being 'irrelevant' within the EU and do not regard political parties as offering them alternative policy routes (see Alonso, 2014). This implies the death of representative democracy understood as: 'a competitive political system in which competing leaders and organizations define the alternatives of public policy in such a way that the public can participate in the decisionmaking process' (Schattschneider, 1960, p. 14). This is especially damaging for the reputation of political parties, the key actors in representative democracies. Parties are expected to voice citizens' policy preferences and to translate these into public policies. Given their centrality in representative democracies, if political parties fail to articulate clear alternatives and to give citizens real choices, democratic representation via parties will eventually also fail.
Political parties' ability to 'express' citizens' diverse opinions (Sartori, 1976) in such a way that they are represented in EU policy-making is important. It matters not only for representative democracy, but also for citizens' support of the European project: citizens who feel represented in the EU are more likely to continue supporting the EU, even when their perceptions of the state of the economy are poor (McEvoy, 2016). This issue is imminent because the crisis has politicized the issue of Europe to an unprecedented degree, has contributed to declining trust in national and EU institutions and to growing euroscepticism (Armingeon and Ceka, 2014;Serrichio et al., 2013).
National political parties are therefore now at a critical juncture: in order for representative democracy to function at this stage of European integration, political parties need to be able to mobilize across borders. Policy alternatives can only be formulated by transnational parties, which 'must become the integrating force in the EU because they share similar values and objectives' (Collignon, 2011). If the positions that transnational parties occupy are clearly differentiated from one another, then alternative EU-level proposals could be offered to Europeans.
The quest for political competition at the EU level and the development of transnational parties are at the heart of a long-standing debate about the legitimacy and democratic deficit of the EU (Follesdal and Hix, 2006;Hix, 2002Hix, , 2013Mair and Thomassen, 2010;Notre Europe, 2006;Scharpf, 1999;Schmitter, 2000). Political contestation at the EU level is advantageous for EU democracy not least because it has the potential to increase policy responsiveness to citizens' preferences (Follesdal and Hix, 2006).
While competition between transnational party policy proposals is a fundamental condition for representative democracy via parties at the EU level, policy coherence is a prerequisite for parties' capacity to operate beyond territorial borders and to pursue common, supranational policy goals: common European policies that aim at serving European citizens as a collective. Congruence among transnational party groups' members facilitates cohesive behaviour in policy-making (Hix et al., 2005) and strengthens the representative link between voters' choices (of national parties) at the ballot and those legislating on their behalf at the EU level (Lefkofridi and Katsanidou, 2013).
Competition between and coherence within transnational parties are two important indicators of the development of transnational partisanship 2 and of European democracy more broadly. They have been assumed to benefit from EP empowerment and especially from the connection of the European election's outcome with the Commission Presidency (Bardi et al., 2010;Follesdal and Hix 2006;Hix 2002Hix , 2013. The Treaty of Lisbon (2009) established the EP as an equal co-legislator to the Council, and linked EP election results to the Commission Presidency. Such developments have been expected to generate 'greater incentives for stronger party organizations and greater possibilities for parties 2 If partisanship refers to attachment to a 'community of shared commitments' and co-ordinated shared activities in pursuit of these commitments, it is 'transnational when these attachments overstep the boundaries of a nation-state' (White, 2014, p. 395). to shape EU policy outcomes in a particular ideological direction' (Hix et al., 2005, p. 211). Amidst a severe economic crisis, which enhanced intergovernmental dynamics and strengthened eurosceptic parties across Europe, the EP made an impressive power grab with some EU-level parties appointing lead candidates (Spitzenkandidaten) for the post of Commission President to embody the link with EP election results (Schimmelfennig, 2014). In theory, the latest European elections were to allow voters 'to give a mandate to a specific political platform for the EU's executive body, the Commission' (Hobolt, 2014(Hobolt, , p. 1529. Because voters can only choose among national parties (not European party groups), for voters to be able to give a specific democratic 'mandate', that is, the authority to carry out a policy or a course of actiontransnational party groups need to occupy distinct locations in the political space (offering clear alternatives to one another in different policy areas) and they have to exhibit internal coherence. What happened in practice, however?
Our study looks at the differentiation and coherence of European Party Groups (hereafter EPGs) resulting from the last two European elections (2009 and 2014). The time span between these two electoral events was a period of crisis during which European integration was becoming increasingly politicized and salient. Using party position data collected by the EU Profiler and Euandi (Garzia et al., 2015;Trechsel and Mair, 2011)  Similarly to previous studies using different data sources, we find that EPG positions are distinct from each other and that internal coherence exists, though it is not great. Moreover, we confirm previous findings on coherence of the Greens, which are considered the most 'Europeanized' parties (Bomberg, 2002) and have been shown to be internally coherent by previous works. Contrary to conventional wisdom, however, our study finds competition and coherence to be prominent on the issue of European integration, which was politicized and rendered salient through the crisis. This runs counter to studies of legislative behaviour and alliance formation in the EP (Hix et al., 2003Hix et al., 2009;McElroy and Benoit, 2007) that have established the left-right dimension as the major basis for citizen representation, co-operation and competition in the EP. Moreover, our study shows that internal coherence is high also for EPGs comprising nationalist right members, indicating eurosceptics' potential to function as a cohesive force in the years to come.

I. Policy-seeking in a Multilevel Polity
Political parties are purposeful organizations, which pursue multiple goals simultaneously: votes, office and policy (Müller and Strøm, 1999). The strategies they adopt in pursuing these goals are affected by the institutional framework in which they operate. Below we discuss how the EU institutional framework incentivizes parties, and how it affects their pursuit of policy. We embed this discussion in previous works, while drawing attention to recent developments that have the potential to affect competition between and policy coherence within transnational party formations: the politicization and salience of EU issues in the context of the economic crisis and the appointment of Spitzenkandidaten.
EU membership brings national parties into an international, multilevel policy arena with numerous other actors (such as EU institutions, interest groups) that pursue their own policy objectives. The agenda-setting capacity of individual parties is weakened by the EU structure. The optimization of public policy influence by individual parties comes progressively under pressure. However, when national parties join forces with like-minded parties from other Member States (MSs), their voice in the system grows stronger (Hix et al., 2003). This is important for the quality of representation as choosing among national political parties is the only way for citizens to be represented in the EU arena and the way parties behave affects whether (and to what extent) citizens' voices can be heard (Lefkofridi and Katsanidou, 2013).
Before the launch of the Single Market in the 1990s, policy-seeking parties may have not perceived European integration as impacting their policy goals to the extent that they would today. Since an increasing number of policy areas have been transferred to the European level, parties with serious aspirations to achieve specific policy goals now have greater incentives to be active at that level. In the EU-28 polity, no single party can unilaterally affect any policy outcome, irrespective of the size and power of the Member State. In sum, to shift EU policy outcomes in the desired direction, national parties have to co-ordinate their policy activity with parties that share their values and policy objectives.
The EP provides a unique opportunity in this regard. It is not just the only collective body whose composition can be determined directly by the European people, and the only supranational institution with a clear mandate of citizen representation; it is also the only supranational institution where citizens' representatives sit along party-ideological rather than national/territorial lines. Originally established as a weak institution, the Parliament increased its powers through successive treaty revisions (Hix and Hoyland, 2013;Rittberger, 2005) and the most recent reforms introduced by the Lisbon Treaty (in force 2009) established it as an equal partner to the Council in most policy areas. The EP's involvement in a wide range of policy areas enables it to shift EU policy-making in particular ideological directions and creates incentives to strengthen transnational party organizations and increase left-right competition between EPGs (Hix et al., 2005, p. 211). The EP can thus be utilized as an instrument for optimizing influence on public policy: being the medium for parliamentary representation at the EU level, the EP opens up unique opportunities to forge transnational co-operation among like-minded parties from different EU MSs.
After an EP election, the elected MEPs re-sort themselves into EPGs to make policy and (co-)decide on legislation that affects the EU citizenry as a whole. Individual national parties can only pursue policy more effectively in the EP by creating 'common fronts', that is by co-operating with parties in other MSs and by developing transnational parties. A non-affiliated party is much weaker in terms of influencing EP legislation. Even worse, the electoral message voters transmit through the EP ballot will be distorted if their chosen national representatives join EPGs that are not congruent with them on key policy issues. In other words, for policy representation to work in the multilevel structure of the EP, national parties need to get together with likeminded counterparts from other countries (Lefkofridi and Katsanidou, 2013).
Existing research on EPG affiliation has shown that national parties choose their EPGs based on policy congruence (Bressanelli, 2012;McElroy and Benoit, 2007, 2010Whitaker and Lynch, 2014). This research also shows that EPGs have identifiable and differentiated positions on multiple policy issues and that these group broadly into two orthogonal dimensions, one consisting of classic left-right issues and the other related to the empowerment of EU institutions Benoit, 2007, 2010). Notwithstanding change at the EU (enlargement, expansion of EP competences) and national levels (rise of euroscepticism and volatility), patterns of EPG policy positioning on the political space, competition between EPGS and coherence within them have been relatively stable (McElroy and Benoit, 2012).
Left-right issues and European integration constitute the key dimensions for political representation in the EU (for a discussion see Mair and Thomassen, 2010). On the one hand, EPGs can bring about change in left-right politics, especially after the Lisbon Treaty expanded EP competences to more policy areas. EPGs' positions have been found to span the entire left-right policy spectrum in the EP, and to occupy regions of the leftright policy space that are distinct from one another (McElroy and Benoit, 2010). Regarding specific (economic/social) policy dimensions, EPGs occupy two opposed camps, one on the left and another on the right, while few EPGs are located in between; the ALDE, which has been found to be located on the right on economic issues but on the left on sociocultural issues, constitutes the exception (McElroy and Benoit, 2012). The positions of EPGs have also been found to correspondalbeit with variationswith the central tendencies of their members Benoit, 2010, 2012) which implies internal coherence.
On the other hand, issues relating to European integration are not decided by the EP but by intergovernmental conferences. Thus, competition on these issues is problematic for citizen representation (Mair and Thomassen, 2010, p. 28). Moreover, because issues of European integration tend to divide EPGs along nationalrather than partisanlines, parties competing on this dimension risk undermining internal party cohesion (Hix et al., 2005). Competition between EPGs, coherence within them, but also their congruence with the citizens they are supposed to represent in the EP have been more pronounced on left-right issues compared to issues pertaining to European integration (Lefkofridi and Katsanidou, 2013). Based on the findings of existing research, EPGs are expected to occupy positions distinct to one another (H1) and to be internally coherent on policy issues (H2). Furthermore, competition and coherence are expected to be higher on leftright compared to EU issues (H3a on competition and H3b on coherence, respectively).
Despite this evidence of transnational competition and co-operation in the EP, the development of transnational parties and of a 'truly European' party system has been laggard. The source of this problem has been attributed to structural-institutional factors, that relate to parties' two other goals, namely their pursuit of votes and office.

National-level Dynamics and the Politicization of Europe by the Economic Crisis
Though parties' policy-seeking strategy depends upon their participation in EPGs, their vote-seeking strategy depends upon their appeal to domestic voters. Parties' pursuit of votes in order to gain representation in domestic and EU-level decision-making bodies is (still) strongly anchored at the national level. The EP election is nationally organized: voters can only vote for national parties, not for European ones. As EPGs only form after the EP election (if the necessary legal criteria 3 are met) the EPG system is not completely stable.
Although research on EPGs reveals continuity in the way they locate themselves in the political space (McElroy and Benoit, 2007), the EPG system is sensitive to national-level dynamics. National parties compete for the electoral support of domestic voters only. If Europe's voting space were (even partially) transnationalized, parties would compete for votes beyond national borders. For some observers, such an institutional framework would incentivize parties to campaign on transnational rather than national issues (Bright et al., 2016). However, this is not (yet) the case.
Under the existing structure, the behaviour of national parties largely determines the success of European elections in presenting competitive EU policy proposals. In the past, the European election has been run as an inconsequential beauty contest of domestic parties rather than a competitive battle between alternative EU-level policy proposals (Van der Brug et al., 2007). For the most part, national parties across EU MSs have done little to adapt to the EU environment 4 in terms of their organizational structures (Poguntke et al., 2007), their policy positions (Lefkofridi, 2014) or their style of competition in European elections (Lefkofridi and Kritzinger, 2008). That said, the weight of national political settings on party organization strategy and their style of competition in EP elections (such as Europe's de-politicization) has recently changed as party communication across borders has improved. This has been due to technological progress, to transnationalization policy issues and to the Europeanization of the public sphere (Bardi et al., 2010). Theongoingcrisis provided impetus to the latter processes by 'forcing' debates on thorny issues of European integration 'into the open'.
The economic crisis that broke out in 2009 rendered European issues all the more salient across the EU, and Europe and its policies were intensely politicized. However, the management of the crisis was dominated by intergovernmentalrather than supranational dynamics, with MSs retreating to state-centric calculations of interest that undermined cross-national partisan alliances that could enable a European party system (Lefkofridi and Schmitter, 2015). Eurosceptics made electoral gains in national elections all around Europe. In some countries, like Greece, domestic party politics was significantly reshaped as attitudes to European integration structured the political landscape, aligning with positions on the economy and creating a new cleavage (Katsanidou and Otjes, 2015).
EPGs are conglomerations of national parties whose electoral success is sought in domestic arenas, and parties have domestic reasons for distinguishing themselves from their competitors. How national parties (re-)position themselves in the political space depends upon (changes in) domestic party systems (McElroy and Benoit, 2012). Given the importance of national-level dynamics for the formation of EPGs, their patterns of competition and their coherence may experience change due to policy position changes in national party systems triggered by the crisis. First, these may result in shifts in the way EPGs are dispersed on the political space, thus affecting differentiation among groups. This is likely to be the case on the issue of European integration, which was politicized during the crisis. Following the 2014 election all pro-EU EPGs lost and all anti-EU ones gained seats (see online Appendix A1, Table 1). Between 2009 and 2014 the balance of power between the pro-and anti-EU camps in the EP changed, with the eurosceptic bloc becoming stronger than ever before.
Realizing that their opposition to European integration could benefit from cooperation, the Radical Right nationalist parties that emerged victorious from the 2014 election took pains to form an alliance, the Europe of Nations and Freedom (ENF). 5 However, the fact that the eurosceptic bloc is dominated by far right national parties, which are socially conservative, suggests that its strengthening may have had an impact on the axes of competition in the EP (McElroy and Benoit, 2012). Certainly, higher competition is likely to manifest in the EU dimension of conflict, which the crisis has politicized and rendered more salient. Eurosceptics comprise the ENF formed in 2014, but also the ECR 6 and the EFD/EFDD 7 that have also been specifically formed on the basis of EU issues.
Furthermore, the increasing weight of EU issues in national politics may come at the expense of EPG coherence on other issues. For instance, recent research shows that the issue of European integration aligns with different issues in Northern Europe than it does in Southern Europe: while in the North it aligns with the issue of immigration, in the South it connects to the economic left-right divide (Otjes and Katsanidou, 2017). The forces of the crisis may affect EPG coherence; for example, the crisis-imposed austerity in debtor countries was endorsed by Social Democrats in the North, but less so by those in the South.

The (Lack of) 'Government in Waiting' and Spitzenkandidaten
The inability of the European election to affect alternation of EU executive power, combined with the EP's weakness (in terms of policy competences) in the early decades of integration meant that Europeans have come to regard European elections as being 'not-so-important' events. 8 Though the EP was gradually empowered and its competences have been significantly expanded, the missing connection to some kind of 'EUlevel government' has persisted. In contrast to national systems, where elected parties in parliament compete for control of the executive, the European election has lacked a government 'in waiting' (Follesdal and Hix, 2006). This had been regarded as the key obstacle towards the development of a transnational party system and the strengthening of transnational parties (Bardi et al., 2010, p. 7). Hence, proponents of political contestation emphasized that for an institutional structure that provides parties with incentives to 5 The ENF was launched on 15 June 2015; key actors, such as Marine LePen, the leader of Front National, are former members of the European Alliance for Freedom, which had been created after the 2009 election (7 th EP) but fell short of the seven member requirement in 2014. 6 David Cameron was pivotal in the formation of this anti-federalist group in the 7 th EP (2009 election). According to its official website, this group was formed 'with a common cause to reform the EU on the basis of eurorealism, respecting national sovereignty and focusing on economic recovery, growth and competitiveness'. Following the 2014 election, this group gradually increased its membership; by May 2017, it was the third largest EPG. 7 The EFDD formed in 2014 (8 th EP) is a sequel of EFD that had been founded in 2009 (7 th EP), which itself was an alliance between eurosceptic national-conservative parties that had been present in previous EP terms. Following the 2014 election, the group lost members to other EPGs, especially other eurosceptic groups: the ECR and the ENF. The latter is the smallest group in the 8 th EP. Half of the EFDD is currently composed of British MEPs elected through the United Kingdom Independence Party (UKIP), whose leader is presiding over the EPG. 8 They are referred to as 'second-order' elections because voter turnout is low, government parties lose and small and/or extreme parties win votes (Reif and Schmitt, 1980;van der Eijk et al., 1996). compete, such control of the executive is necessary. Opening up contestation for the Commission Presidency has been proposed as a possible remedy to the democratic deficit (Follesdal and Hix, 2006;Hix, 2002) and a major prerequisite for the development of a European party system (Bardi et al., 2010, p. 7).
This specific proposal for fixing the EU democratic deficit articulated over a decade ago became realityeven if in a very minimalist fashionduring the crisis. The EP exploited an opportunity inherent in the Lisbon treaty (2009) that linked the Head of the Commission to the results of European election. Based on a generous interpretation of the new Treaty article, the EP conducted the 'Spitzenkandidaten' experiment in 2014 (see Hobolt, 2014;Schmimmelfennig, 2014). Manybut not all -EU-level parties nominated lead candidates for the Commission Presidency prior to the election and thus put forward a transnational form of accountability and control (Bardi et al., 2010, pp. 8-10).
The Spitzenkandidat idea originated in discussions within the German Christian Democratic Union and it was initially promoted by the European People's Party (EPP) (European Parliament, 2015). But it was the European Socialists that started off with the nomination of Martin Schulz as the 'party's front man'. They were followed by the Liberals, who nominated Guy Verhofstadt; the Greens with Ska Keller and José Bové; the European Left with Alexis Tsipras, and the EPP with Jean Claude Juncker, who eventually became the Commission President. Eurosceptics on the right saw this development as too 'federalist' and refused to appoint Spitzenkandidaten.
Though much less ambitious than what proponents of this plan might have hoped for, the Spitzenkandidaten innovation tried to address the no-government-in-waitingproblem. The Spitzenkandidaten procedure purportedly made EP elections similar to parliamentary elections in the Member States, where voters make a choice among parties (or candidates) 'in the knowledge that this is also a vote for a specific prime ministerial candidate and government' (Hobolt, 2014(Hobolt, , p. 1529. For the EPP's candidate Juncker, the Spitzenkandidaten allowed voters to see 'who would be appointed to the presidency of the Commission and how' and for Green candidate Keller, they constituted 'the face of the European dimension of this election' (European Parliament, 2015). The novel connection to the Commission Presidency sought to raise the stakes of the EP election and make it more competitive. The competition for influence over the Commission's policy direction was assumed to motivate EPGs' constituent members to focus on what unites them with their allies and what distinguishes them from their enemies: while it could help them differentiate themselves from one another, this innovation could also act as the necessary 'glue' uniting EPGs' constituent parties behind a common EU-level candidate against other EPGs, in an effort to countervail the disintegrating forces of the crisis. If the Spitzenkandidaten procedure exerts a positive effect on EPGs, then we should observe differences between those EPGs (S & D, Greens, GUE-NGL, ALDE, EPP) that sought a specific mandate by nominating a Spitzenkandidat for the Commission Presidency and those that did not.
Our final thoughts concern variation in levels of coherence across EPGs. First, while no ideology is completely uniform, some party ideologies are more coherent than others. Research on party families has found the Greens to be the most ideologically homogeneous family, followed by the Social Democrats. The Christian Democrats and Conservatives are less homogeneous, and so are Radical Right parties (Ennser, 2010). The most diverse family is the Liberal. Liberal parties have historically been split between conservative liberalism and radicalism, whereas nowadays they oscillate between right, centre-right but also centre-left (Marks and Wilson, 2000). Previous work on EPGs also shows that the liberal group (ALDE) exhibits the highest diversity of positions among its members (McElroy and Benoit, 2012). However, it should be underlined that EPGs are not exactly a reproduction of 'party families'. Party family is a more static concept than the dynamic EPG formation 9 (McElroy and Benoit, 2012).
Second, one might expect EPG size to matter for internal coherence: the bigger the group, the higher the likelihood of internal divergence. In the EP, however, the constituent members of the two largest groups (EPP and S & D) have a longer tradition of cooperation. Moreover, they include most governing parties, which are typically catch-all, have presidential organizational structures, and are thus much more flexible in adapting to transnationalization (Bardi et al., 2010, p. 5). That said, policy position change may have been motivated by the crisis-driven electoral decline of governing parties all around Europe during the period 2009-14. Also, the aforementioned differences between the South and the North of Europe may have impacted negatively on the coherence of large EPGs that contain governing parties.

II. Methodology and Data
We use data on party positions from the 2009 EU Profiler (Trechsel and Mair, 2011) and the Euandi 2014 (Garzia et al., 2015). The EU Profiler and the Euandi are Voting Advice Applications (VAAs) designed for the 2009 and 2014 EP elections, respectively. As the design of these VAAs is compatible with both spatial/policy proximity and directional models of party competition and voting behaviour, the data are ideal for our investigation of competition between and coherence within EPGs. Furthermore, these data allow investigation of eight EPGs (for distribution of seats, see online Appendix A1) composed of parties from all EU Member States (EU-27 in 2009, EU-28 in 2104) based on comparable measurements.
We use VAA data to capture parties' (not voters') positions on policy issues that are based on experts' coding of a wide variety of party documents (see Garzia et al., 2015, p. 3 , Table I). Importantly, national parties have confirmed these positions. As such, these VAA data are slightly superior to expert surveys of national party positions. We avoid using elite studies such as candidate surveys, not only because of their low response rates but also because, to capture party positions, we would have to average positions of individuals that are affiliated with both national and EU-level parties ('agents with two principals', Hix, 2002), which would be problematic for our analysis.

Issue Selection
We selected (quasi-) identical items from the 2009 and 2014 VAAs, which are coded on the basis of a five-point Likert-scale ranging from completely disagree (1) to completely agree (5). One side of an issue is captured by positions 1 and 2 and the other by positions 9 As EPG membership is determined by electoral success or loss at the national level, some EPG members may elect more/fewer MEPs or fail to get elected; also, old members may withdraw from one group, or new members may join. Such movements have the potential to impact each EPG's internal consistency depending on the policy congruence of those joining/leaving with the rest of the group. For example, the EPP, which won the 2014 election and appointed a Commission President, was the biggest EPG loser in terms of decrease in the number of seats (-57). 4 and 5. While differences between 1 and 2 or between 4 and 5 concern varied intensity, position 3 is the neutral position ('neither agree nor disagree'). Among all (quasi-) identical items from 2009 and 2014, here we focus on four items: redistribution, immigration, same-sex union and European integration. The exact wording of the items we use here is: • Redistribution: Social programmes should not be maintained at the cost of higher taxes; • Immigration: Immigration [into your country] should be made more restrictive; • Gay rights: The legalization of same sex unions is a bad thing; • EU: European integration is a good thing.
Our selected items concern the issue of European integration and three left-right issues that constitute major lines of conflict across Europe and are issues that European citizens recognize and understand. Crucially, the crisis even connected issues such as redistributionthat until now concerned national-level policymakingto the EU level (Katsanidou and Otjes, 2015). Lack of restrictions on immigration is a direct implication of the single market, and hence there is a connection between this issue and EU membership in the minds of citizens (Otjes and Katsanidou, 2017). Parties' opposition to or promotion of immigration may have economic and/or sociocultural underpinnings. Gay rights, which are connected to EU citizenship rights, 10 closely relate to parties' adherence to traditionalistconservative or liberal sociocultural attitudes; as such, they are contentious across the EU, which is reflected in the EP and on the ground in domestic politics (Ayoub, 2016). Finally, the question of European integration is an important topic in EU politics, and the crisis made it very salient in domestic arenas.

A Three Step Analysis
Our methodology entails three steps: First, we assess the degree of differentiation in policy supply at the EU level by examining the positions of groups formed after the 2009 and2014 EP elections. 11 To estimate the position of each EPG on the four aforementioned issues, we calculate the mean position 12 of the national member parties belonging to each transnational party weighted by the number of MEPs of each national party. Competition is manifested when transnational parties position themselves on both sides of the policy issues (pro/contra). The more parties spread in the political space, the more variety of policy choices there is.
Second, we assess the internal coherence of EPGs formed after the 2009 and 2014 EP elections. To this aim, we use two measures: first, we calculate the standard deviation (SD) of the positions of all member parties within an EPG; second, we calculate the percentage of constituent party members on each of the two sides of every issue under study, 10 According to a recently published opinion of the EU Court's Advocate General: 'Though Member States are free to authorize marriage between persons of the same sex or not, they may not impede the freedom of residence of an EU citizen by refusing to grant his or her spouse of the same sex, a national of a non-EU country, a right of permanent residence in their territory' (Court of Justice of the European Union, 2018). 11 Katsanidou and Lefkofridi (2018) "A Step Closer to a Transnational Party System" Replication Files https://dx.doi.org/ 10.7802/1616 12 A study using expert placements of EPGs' positions shows that the position of GUE-NGL and of the two largest groups reflects the central tendency of its members; the position of other groups is more on the left (Greens and ALDE) or more on the right (ECR and EFD) compared to their central tendency (McElroy and Benoit, 2012). and the percentage of member parties that adopt a neutral position. These two simple measures capture coherence: the lower the SD, the more coherent the EPG. For a satisfactory level of coherence on one issue the SD has to be below 1.5. Because the scale has only five points, there is only one point difference between the extreme and the moderate position on the same side of an issue. The one-point difference of SD allows some parties to have a neutral position. The percentages enable us to identify whether there is a 'dominant side' of each issue for each EPG; or, if there is no dominant side, how dense is the concentration on the neutral position. We report only issues with concentration of 60 per cent or more of member parties on one side of the issue and ignore concentration on the neutral position. Besides the SD, this indicator signals which issues are more or less divisive (left-right issues versus European integration).
Third, we assess how important positions on these issues are with regard to national party membership in an EPG (as opposed to all other groups). We conduct a logistic regression analysis, with the four issues as independent variables and membership in EPGs as a dependent variable. This part of the analysis focuses on differences between issues (not election years). The importance of an issue for EPG affiliation relates to the internal coherence within EPGs; it also relates to the divisions between EPGs, given that this is a test assessing the importance of a specific issue in joining an EPG as opposed to all others.

III. Empirical Results
In this section, we report results on party differentiation between and coherence within transnational parties in 2009 (in the beginning of the crisis, no Spitzenkandidaten) and in 2014 (mid-crisis and after the introduction of the Spitzenkandidaten procedure) on the three left-right issues and the EU issue.

Transnational Party Competition
How differentiated is the policy supply at the EP level? Figures 1 and 2  Second, while we generally observe stability, there are notable exceptions: on the issues of welfare and immigration, ALDE is located more on the left in 2014 (Figure 2) compared to 2009 (Figure 1). In contrast, EFDD (2014) is more opposed to welfare, and less opposed to immigration compared to EFD (2009). Though the ECR is consistently opposed to welfare (2009,2014), its opposition to same-sex union and immigration becomes more pronounced in 2014. In 2014, the strongest opposition on all three leftright issues is represented by ENF.
Third, there is a clear split between the pro-EU (EPP, S & D, ALDE, Greens) and anti-EU camps (GUE-NGL, EFD/EFDD and ENF). When comparing the two election years (Figure 1 versus Figure 2), we see that EPGs take a clearer stance (avoiding the middle grounds of the 'neither agree nor disagree' option). Looking at the pro-EU bloc, we see continuity from one election to the other: europhile EPGs are crowded around the positions 4 and 5 of the scale. The mean position of the eurosceptic ECR's constituent members is closer to the pro-EU than the anti-EU EPGs in both election years. National party movements resulted in changes in how anti-EU EPGs spread on the EU issue space: compared to the pro-EU bloc there is more differentiation within the anti-EU bloc, and this differentiation is greater in 2014 (Figure 2, graph on the bottom right). The most eurosceptic EPG formed following the 2009 election was the EFD; after the 2014 EP election this group experienced a lot of membership changes (see footnote 7), which resulted in a more eurosceptic position. However, EFDD now faces competition at the extreme pole of the pro/anti-scale: the most eurosceptic EPG is now ENF.
Given these patterns of competition on left-right issues and European integration (against H3a), the European party system seems able to express alternative proposals for EU policy-making. For these to translate into cohesive legislative behaviour EPGs need to be internally coherent. Along with the mean positions of EPGs' constituent members, the SDs plotted in Figures 1 and 2 give some first indications of EPGs' varied levels of internal coherence, to which we now turn. Transnational Party Coherence Table 1 shows the percentages of EPGs' constituent members' concentration on the same side of each policy issue. 14 Empty cells signify that concentration of national parties was lower than 60 per cent on the same side of the issue (and hence not reported). This table helps us to assess: whether/which EPGs are coherent (H2) and whether/how coherence evolved between 2009 and 2014; and which issues are the most/least divisive (H3b).
First, the more empty cells per EPG, the less coherent the EPG; a lot of empty cells across the board suggest a system of parties that suffer from internal divisions. Since half of the cells are full, there is some evidence in support of H2 (EPG coherence). Second, even a cursory look at the distribution of empty cells shows that coherence deteriorates between election years, which may be due to the (temporary) effect of the crisis. Based on Table 1, there is no EPG (with or without Spitzenkandidat) whose coherence increased on all four issues. On a few issues, however, a few EPGs do get more consistent: in 2014 ALDE and the Greens have a clearly more coherent pro-immigration position and ALDE becomes more coherent also on welfare. EFDD is perfectly coherent on the EU and samesex union issues. Most coherent are the most eurosceptic, EFDD and ENF (scoring three out of four cells in 2014). Striking are two cases of lack of coherence on issues which the EPGs' constituent members typically own: ENF on immigration (2014) Table 1 fails to support H3b since coherence is not higher on left-right issues compared to the EU, which, in fact, constitutes the least divisive issue for all EPGs except for: Greens (2014); GUE-NGL and ECR (both years). In 2009, GUE-NGL and all parties on the right appear coherent on welfare; in 2014, however, coherence exists only for EPP and ENF. Immigration seems divisive for parties on the right (except for EFD/EFDD), but not for parties on the left (except S & D in 2014). Coherence on same sex unions is enjoyed only by the Greens and EFD/EFDD.     To better grasp EPG's internal coherence, we use the example of S & D. Figures 3 and  4 illustrate the positions of its constituent parties on four different issues 15 (weighted by number of elected MEPs). Figure 3 plots S & D members on a two-issue space: European integration and welfare. Each national party is presented with a bubble, whose size depicts the number of MEPs elected by this party. A bigger bubble represents a bigger member party, while a small bubble stands for a smaller party. Coherence on integration is high, as all S & D members find themselves on the pro-EU side of the spectrum. However, on the issue of 'welfare vs. taxes', a classic Social Democratic concern, four S & D members diverge from the majority: three occupy a neutral position (the Polish, Hungarian and the Danish Socialist parties) and one (Cypriot EDEK) is even placed on the opposite side of the spectrum. Figure 4 that plots S & D members on a two-issue space using immigration and same-sex union, shows a worse picture. On both issues, S & D members locate themselves across the spectrum, though a higher concentration exists on pro-same sex union and pro-immigration sides; this implies that while coherence on these issues is not high, there is a preferred policy direction for these issues.
Finally, we probe the predictive power of four issues for EPG affiliation to complement the above discussion on divisions between EPGs and unity within them (H3a and H3b). Table 2 reports the strength of the relationship between congruence on each of the four issues and membership in each of the EPGs (as opposed to any other group). As the number of observations is low, Table 2 reports up to 0.10 levels of significance. Agreement with/opposition to the maintenance of social programmes at the cost of higher taxes is significant for affiliating with S & D, EPP and ALDE in both election years and GUE-NGL only in 2014. In both election years, the issue of European integration is a significant predictor of affiliation with both pro-and anti-EU EPGs located on the left (S & D and GUE-NGL) and right (EPP, and marginally also EFDD and ENF in 2014). European integration scores the highest coefficients, thus manifesting a higher impact than other issues.

Conclusion
Literature on the EP provides ample evidence of competition between EPGs and coherence within them, and especially on left-right issues, the kind of issues where the EP has policy competence. Our study of competition and coherence during a period of crisis (2009 and2014 EP elections) contributes to this literature in three ways: first, similarly to previous works, EPGs occupy both sides of four key political issues investigated: welfare, immigration, same-sex unions and European integration. Clear differences in EPG policy positions manifest their potential to function as a party system 16 that enables clear political conflict and, consequently, democratic representation of citizens in the European political arena. Second, European integration, which was politicized and rendered salient through the crisis, appears as the least divisive issue within Europarties, and the most divisive between them: while it unites europhiles against eurosceptics, it differentiates eurosceptics from one another. This challenges the conventional picture of competition and coherence in the EP being greater on left-right issues. Third, the Greens, but also 15 All figures of all EPGs in both election years are provided in the online Appendix A3. 16 The differentiation between EPGs (whose position we analyzed as aggregated positions of national parties) does not automatically imply that individual parties in one country differ. eurosceptic EFD/EFDD and ENF exhibit the highest levels of coherence. Although eurosceptic parties are by nature hostile to the EU, this specific finding advances the idea that eurosceptic party activity at the supranational level is a form of adaptation to a changing EU polity and policy environment (Lefkofridi, 2009).
Future academic work could place more focus on the specific issues that constitute the core of each party's ideology, for example, by utilizing issue ownership theory. Now that we know how congruent party members are with each other, the next step is to investigate the extent to which EPGs are congruent with their supporters, or the median European voter.
Last but not least, connecting the European election to the Commission presidency (via Spitzenkandidaten) does not seem to have produced the desired results, at least not in the short run. Our data provide no indication that this innovation impacted EPG differentiation or coherence. At the time of writing, the EP is trying to make radical changes to its electoral law, including the institutionalization and reinforcement of Spitzenkandidaten; we hope our study will prove a helpful guide for political parties at national and EU levels to prepare better for the 2019 EP election.