an Electoral Union

Since 1999, all member states have employed some form of proportional representation to elect representatives in the European Parliament. The introduction of Spitzenkandidaten and greater policy and campaigning coordination within Europarties sought to create a common European ground of party competition. However, the actual election of MEPs remains a national endeavor, determined by a national electoral framework. The electoral success of an Europarty in a given member state is shaped by the conditions set by the national electoral model for European elections. Different national models also give rise to different electoral strategies employed by parties wishing to maximize their number of MEPs. While electoral alliances might reflect cooperation of ideologically similar parties (such as list combinations in the Netherlands) or presidential majorities, like the coalition list of parties that support the French President, sometimes, especially in newer member states (e.g., Croatia), they might represent a strategic choice of ideologically divergent parties to work together in order to pass the threshold and win seats in the European Parliament. This paper seeks to compare different national playing grounds in which Europarties, i.e., their member parties compete, as well as electoral strategies that accompany such different frameworks. Apart from Ireland and Malta, which use the single transferable vote system, all other (current) member states use list proportional representation. However, the choice of a national electoral district or division in regional electoral constituencies can have a significant impact on electoral outcomes. Further, the employment of an electoral threshold and its interaction with district magnitude can create different electoral conditions. Finally, methods of seat allocation have an impact on prospects of smaller parties. A comparison of national electoral models will uncover different conditions and levels of (dis)proportionality that all weave together into the ultimate makeup of the European Parliament.


Introduction
Four decades after the introduction of direct election of national representatives in the common European Parliament (EP) and two decades after the adoption of proportional representation as the lowest common denominator for the election of Members of the European Parliament (MEPs), the 2019 EP election demonstrated that member states still employ a range of different electoral rules, while political parties competing in the respective national electoral arenas subscribe to dissimilar electoral strategies. Within such a context, bold dreams of a creation of a common electoral space in the European Union, which would enable transnational policy advocacy and voter mobilization, remain hardly attainable. When discussing the idea of a European constitution, Habermas (2001Habermas ( , 2011 wrote about the need of a creation of a common public sphere in the EU. However, such a pan-European public sphere would presuppose a common European political sphere, i.e., a common electoral arena. The question remains, how far are the EP elections from the establishment of such a common electoral arena.

The Quest for a More Perfect Electoral Union
It is commonplace to observe that EP elections are second-order elections, which serve more as midterm elections to punish incumbent national governments, than to mobilize voters around European policy (Hix & Høyland, 2011, pp. 146-147). This is reflected both in lower turnout levels in virtually all member states, but also in the makeup of electoral supply, which is characterized by either young, upcoming candidates or older politicians close to retirement age, while the most prominent party members opt to run for national, rather than European offices. EP elections have a double, national and European role, whereby, despite increased policy coordination and grouping in European party federations (Europarties), the national dimension of this level of electoral competition still prevails. In his outline of the analytical concept of Europeanization of political parties, Ladrech (2002, p. 396) presented five dimensions: "1) policy/programmatic content; (2) organizational; (3) patterns of party competition; (4) party-government relations; and(5) relations beyond the national party system." EP elections pertain to the third dimension. Here, Ladrech (2002, pp. 397-398) points to increased politicization of the European Union and the creation of pro-EU/Eurosceptic cleavages in national party systems. Yet, Europeanization of party competition at EP elections in the sense of competition around common, European issues, is intrinsically linked to the national nature of the very process of election of MEPs. Parties will debate EU-related issues, yet within a specific national context, without truly reaching across member state borders.
Although common rules have been a long-term goal of architects of European unification and an explicit call for the adoption of harmonized electoral rules for the election of MEPs was made by the 1992 Maastricht Treaty, only in 2002 did a Council of the European Union decision establish provisions for EU-wide adoption of proportional representation for European elections (Sokolska, 2020, p. 1). The 2002 decision was somewhat of a post-festum act, since the United Kingdom, at that time the last member state to employ a majoritarian system on European level, switched from a first-past-the-post to a closed-list proportional representation system for the 1999 EP election.
The most recent, 2018 decision, which will affect elections from 2024 onward, stipulates not only a common usage of either list proportional representation (list PR) or single-transferable vote proportional representation (STV-PR), but also a framework for the application of electoral thresholds. The amendments of Article 3 of the Electoral Act thus prescribe a maximum (legal) electoral threshold of 5%, whereas member states that employ electoral districts with a magnitude of 35 or more seats will henceforth have to apply a threshold which is between 2% (minimum) and 5% (maximum) (Council of the European Union, 2018, p. 2). This is expected to create an interesting political and legal dynamic in Germany, the Union's largest member state, since in 2014, its Constitutional Court has declared the application of a threshold for the election of German MEPs unconstitutional (Bundeszentrale für politische Bildung, 2014). In addition, as one will be able to observe later in this analysis, the Council decision only deals with formal, legal thresholds, disregarding the fact that unequal conditions for electoral competition in individual member states are created by effective thresholds that can often be considerably higher than the legally determined electoral threshold.
In 2014, Europarties launched the process of electing common list leaders (Spitzenkandidaten), as an attempt to combat the perceived democratic deficit of the EU, as well as to boost the visibility and future accountability of MEPs. Further reform of this process was initiated for the 2019 EP election yet lack of intra-party consensus in Europarties prevented further steps that would bring EP elections closer to citizens (Wolfs, Put, & Van Hecke, 2021). Moreover, the idea of transnational lists, as an addition to existing national electoral constituencies, was put forward by EU officials and scholars, and received traction during the 2019 EP election (Van Hecke, et al., 2018). Transnational lists would add a new, supranational (and truly European) level of electoral competition to the EP election, but they would not tackle the underlying dissimilarities on the national level, i.e., in national electoral constituencies which would be used parallel to this addition. Thus, the current state of affairs still warrants a comparison and assessment of national "sub-elections" that make up the EP election.

Research Framework
The underlying approach followed in this research is a combination of historical and rational choice institutionalism (Peters, 1999), which views political actors, in this case national political parties competing in a nationally fragmented European political arena, as at least partially rational actors, that adapt to the institutional framework which is a product of long-term national and European institution-building and policy-making. This framework creates a maneuvering space for political parties for their achievement of national and European political goals.
The aim of this paper is to examine whether EP elections are conducted in a common, European political arena or are there still significant differences in electoral rules, strategies, and effects in individual member states. Farrell and Scully (2005) undertook a similar analysis regarding the 2004 EP election. This paper takes a fresh look at the same problem by employing 2019 data and introducing new factors to consider.
When discussing electoral rules, the first thing to consider is district magnitude and average district magnitude (in cases where a member state employs regional electoral districts, instead of a national, at-large district). District magnitude, especially in its interaction with the electoral threshold, remains the most important element of any electoral system, when one considers the effects electoral rules have on electoral outcomes (Gallagher & Mitchell, 2018, pp. 25-26). In addition, a comparison of national electoral rules has to include an assessment of effective thresholds, as well as a discussion of different seat allocation methods used in individual member states. Although Farrell and Scully's 2005 study also included ballot structure, i.e., the presence or absence of preferential voting as a criterion of comparison, this paper follows the assumption that ballot structure is an important element of any electoral system yet is less consequential than district magnitude and the threshold and can thus safely be excluded from such an EU-wide analysis. However, this paper introduces electoral strategies, assessed through the share of coalition lists, especially coalition lists than span several Europarties, as a new point of comparison regarding national variations at EP elections. In addition, a differentiation between old and new democracies is added to the equation in order to capture possible East-West differences at the 2019 EP election. New democracies are understood to be member states that were democratized after the Fall of the Berlin Wall. Thus, this group includes 11 post-communist EU member states. Regarding electoral effects, following in the footsteps of Farrell and Scully, this paper considers the effective number of (seat-winning parties) (ENP (S)) and the level of disproportionality (Gallagher's index). Contrary to the aforementioned paper, it does not consider the percentage of elected female MEPs as one of the electoral effects to be included in an EU-wide comparison. Namely, women's representation is generally higher in proportional representation systems than in majoritarian systems and female candidates have better chances in larger electoral districts than in smaller ones (Krook, 2018, p. 176). However, in the context of the EU, all member states employ some version of proportional representation and multi-member districts. Differences in women's representation stem from both intra-party and legally mandated gender quotas, but also from variations in political culture and general levels of gender equality in individual member states. Thus, the percentage of elected female MEPs cannot be adequately explained by variations in central electoral system elements such as (average) district magnitude, (effective) electoral threshold, and method of seat allocation. This paper will try to examine the assumption that EP elections, when viewed through the lens of electoral rules, strategies, and effects, are still a fragmented endeavor, steeped in national political considerations, and not following a common, EU-wide overarching theme. This assumption is, in turn, linked to the idea that Europeanization of party competition, in Ladrech's sense, does not mean just infusion of national campaigning with EU-related issues, but ultimately a restructuring of party competition as to reflect a common European partypolitical space. While the expected differences in electoral rules and their effects on electoral outcomes are understood as evidence of a fragmented, and not a unified electoral arena, the different electoral strategies used at EP elections are expected to show that we are still dealing with national political parties paying more attention to national party-political concerns, and less to common campaigning along Europarty lines. Thus, this party rests on the assumption that East-West differences will manifest themselves not just in different electoral rules and subsequently effects, but also in different electoral strategies, whereby parties in older member states are expected to more frequently run alone or in coalitions that reflect their Europarty membership, while parties in new member states are expected to more frequently form electoral coalitions in general and coalitions across Europarty lines in particular, thus confirming the primacy of the logic of national, rather than EU-related party-political concerns. Finally, this paper rests on the assumption that the comparison of pre-Brexit and post-Brexit data will show a slight increase in uniformity of the electoral framework, evidenced through electoral rules and their effects. In addition, one should expect to see post-Brexit differences in the overall effects electoral rules have on outcomes (measured in the level of disproportionality).

Data and Methods
This paper relies on data from the ParlGov database (Döring & Manow, 2021) for national electoral results, as well as the effective number of seat-winning parties (Shugart & Taagepera, 2018, pp. 42-43) and the level of disproportionality (Gallagher, 1991). Regarding thresholds, two separate measures are reported (see Appendix), the legal threshold (Sokolska, 2020), 1 and the effective threshold. The effective threshold is calculated following the 75%/(M+1) 1 The process of collecting data on legal thresholds employed by individual member states for the 2019 EP election also included official European Parliament information sources (European Parliament, 2019b). formula (Taagepera, 1998, p. 394). One should draw attention to the fact that this approach is a useful provider of a single statistic, yet an approximation, since in concrete electoral systems, in practice, we can find two natural thresholds, a lower one (threshold of representation or inclusion, the lowest percentage a party needs to win a seat), and an upper one (threshold of exclusion, the highest percentage a party may reach without winning a seat), as described by Lijphart (1994, pp. 25-27).
Data on seat allocation methods follows Pukelsheim's (2014, pp. 1-29) comprehensive book on proportional representation. Information on electoral coalitions was collected both from the ParlGov database as a starting point, as well as the official European Parliament website (European Parliament, 2019a).
The analysis presented in this paper was done in R, in R Studio (version 4.0.2). The first part of the analysis discusses differences in electoral rules in member states, including average district magnitude, effective thresholds, and methods of seat allocation. This comparison also looks at differences between old and new democracies. The second part is devoted to electoral strategies, measured through the presence or absence of electoral coalitions, particularly those than cross Europarty lines. Again, new and old democracies are contrasted. In the third part, electoral effects are considered, i.e., the level of disproportionality and the effective number of seat-winning political parties, with a particular attention to differences between electoral effects in old and new democracies. Further, electoral effects are assessed for interaction through the application of a series of Pearson's correlations. Finally, an OLS multivariate regression model is tested, with the level of disproportionality as dependent variable. For all steps of the analysis, the paper first considers original results of the 2019 EP election and then the amended data, after the post-Brexit reapportionment.

Electoral Rules
From the 29 analyzed cases, 26 employed list PR for the election of MEPs at the 2019 EP election, while only Ireland, Northern Ireland, and Malta used STV-PR, in line with their tradition of employing this electoral system for their respective parliaments and assemblies. The mean average district magnitude was 19.3, which is lower than the mean number of MEPs (25.9) elected in each member state. This is due to the fact that Ireland, Great Britain, Belgium, Italy, and Poland employed regional electoral districts. While the decision to employ regional, and not national constituencies can be easily explained by Belgium's linguistic federalism, Great Britain's composite polity, Ireland's usage of STV-PR, and Italy's stark regional differences (and animosities), Poland's employment of regional electoral districts has no immediate explanation other than the legislator's wish to suppress fragmented electoral results. The 2019 EP election was the first one where the French voters elected MEPs in an at-large district, instead of regional ones. One should also note that the average district magnitude for old democracies was 22.7, while in new democracies it was only 13.8. This large gap can be explained with demographic differences, i.e., the fact that Eastern Europe is less populated than Western Europe, which leads to smaller numbers of MEPs and finally lower average district magnitudes. However, one should consider that in 2019, Poland's (average district magnitude 3.9) national delegation (51 MEPs) amounted to a quarter of the total number of MEPs from new democracies (199), which means that electoral rules in that country have a large quantitative impact on comparative statistics for the whole group.
After Brexit, the mean average district magnitude rose to 21.2, while the mean number of MEPs was 26.1. Average district magnitude in new democracies increased to 14.2 and 26.1 in old democracies.
In 2019, ten member states that employed list PR did not have a legal electoral threshold. The mean effective threshold for all cases was 9.6, which is substantially higher than the 5% limit mandated by the 2018 Council decision. Among old democracies, the mean effective threshold was 12.3, while it was 5.24 for new democracies. Given the fact that effective thresholds are a product of interactions of average district magnitudes and natural thresholds, one would expect higher effective thresholds in new democracies, given their lower average district magnitudes. However, all cases, except Estonia, that were in the upper 25% of the distribution of effective threshold levels, were old democracies, leading to an overall steeper mean effective threshold in the Western part of the Union.
After Brexit, the mean effective threshold in old democracies fell drastically, from 12.3 to 5.61, bringing it very close to levels in new democracies (5.24), as well as to the 5% mark.
At the most recent EP election, 16 out of 29 cases employed the D'Hondt method for seat allocation, while four used the Hare quota, another four used the Droop quota, three opted for Sainte-Laguë, while two employed the Hagenbach-Bischoff formula. Since the majority opted for D'Hondt (both among old and new democracies), a divisor method deemed somewhat less proportional than other divisor methods and more beneficial to larger parties/party lists, a comparison was done regarding the usage of this method and two electoral effects -ENP (S) and the level of disproportionality. Cases using the D'Hondt method had, on average, a somewhat lower level of disproportionality (7.5) than those using some other seat allocation formula (9.6). Again, given the fact that disproportionality is more affected by the interaction of the threshold and district magnitude, than by the seat allocation method, one can explain this difference by pointing out that the mean average district magnitude for member states using the D'Hondt method was 21.0, while member states using all other aforementioned formulae had a mean average district magnitude of 17.0. Regarding the effective number of seatwinning parties, one could find very slight differences considering the seat allocation formula. In cases using the D'Hondt method, the mean ENP (S) was 4.7, while in the other group it was 4.4. Thus, we can state that the main and most important differences in national electoral rules at EP elections stem from variations in average district magnitude and effective thresholds.
After Brexit, average disproportionality among cases using the D'Hondt method was 7.0, while cases using other seat allocation methods had an average disproportionality of 8.5. In addition, the effective number of seat-winning parties rose to 4.8 for D'Hondt-using cases, while it stayed at 4.4 for cases using other formulae.

Electoral Strategies
Among new democracies, almost all cases (9 out of 11) had a presence of electoral coalitions at the 2019 EP election, whereas among old democracies, in a third of cases, parties employed such a strategy. However, almost all of the cases in old democracies included electoral coalitions of national parties that belonged to different Europarties. Cross-Europarty electoral coalitions can imply several things. First, they may be understood as evidence that national party politics does not take Europarty membership seriously and enters electoral coalitions solely to boost chances for winning seats. Second, this practice may be also understood as a reflection of national patterns of cooperation between various political parties belonging to different Europarties (e.g., green and social democratic parties working together, or EPP member parties collaborating with liberal, ALDE/Renew Europe members), that finds its way on the European level. In other words, it could either mean that party competition is not Europeanized at all, and that electoral strategies are just success maximization strategies or that party competition in a given case is very Europeanized in the sense that cross-Europarty cooperation is regarded as desirable and beneficial both on national (EP election arena) and European (European Parliament arena) levels.
In the case of Bulgaria, the Democrats for Strong Bulgaria (DSB) party, an EPP member, decided not to run together with prime minister's party GERB (also EPP member), but instead formed a coalition with a smaller unaffiliated party and another non-parliamentary party which is affiliated with the European Greens (Spirova, 2020, p. 57). The EP election in Croatia saw the emergence of the so-called Amsterdam coalition, 2 which included two ALDE/Renew Europe members, one former EPP member, currently unaffiliated (Croatian Peasant Party, HSS), and several smaller, unaffiliated parties (Nikić Čakar & Raos, 2020, p. 74). In the Cypriot case, the Civic Alliance, a European Democratic Party member and thus aligned to Renew Europe, entered a coalition with the Movement of Ecologists -Citizens' Cooperation (affiliated with the European Green Party), yet the alliance failed to win any seats (Sözen & Faustmann, 2020, p. 84). In Czechia, EPP member parties TOP09 and STAN ran on the same ticket with the Green Party (Kudrnáč & Petrúšek, 2020, p. 99). In France, the presidential majority list included several parties, all affiliated with Renew Europe. Indeed, the success of this alliance was a major contributor to the creation of Renew Europe group as an extension of ALDE and an accommodation of negative connotations of liberalism in French political culture (Houard-Vial & Sauger, 2020, p. 143). In Hungary, the Hungarian Socialist Party (S&D) and the Dialogue for Hungary party (Greens/EFA) formed an alliance to boost joint opposition prospects against the ruling Fidesz (suspended from EPP membership) party (Várnagy, 2020, pp. 175-176). Italy has an established tradition of electoral coalitions in national politics. At the 2019 EP election, the +Europa party (ALDE), a very pro-EU party, led by long-time representative of Italian radicalism Emma Bonino, entered an alliance with a variety of smaller parties, including former MEPs that were members of the Eurosceptic Europe of Freedom and Direct Democracy group (Castelli Gattinara & Froio, 2020, p. 204). In Lithuania, Christian Families Alliance, an EPP member catering to Polish minority interests, entered a coalition with the unaffiliated Lithuanian Russian Union (Jastramskis & Ramonaitė, 2020, p. 236), showcasing the paramount importance of ethnic representation over ideological congruence. Dutch elections have a tradition of list combinations (lijstcombinatie or lijstverbinding) that serve smaller parties in their efforts to boost chances for winning of seats. Thus, Christian Union (EPP) joined forces with the Reformed Political Party, an ECR member (Otjes & Voerman, 2020, p. 263). Similar to Hungary, in Poland the joint opposition against the ruling party (PiS) created a cross-Europarty alliance called simply 'European Coalition', including the Civic Platform (EPP), the Democratic Left Alliance (S&D), and the .Modern party (ALDE/Renew Europe) (Jasiewicz & Jasiewicz-Betkiewicz, 2020, p. 286). In Portugal, the Unitary Democratic Coalition included the Communist Party (GUE/NGL) and the Ecologist Party (Greens/EFA) (Magone, 2020, pp. 298-299). In the case of Romania, two ALDE/Renew Europe-affiliated parties, Save Romania Union and the Freedom, Unity and Solidarity Party joined forces for the 2019 election (Stan & Zaharia, 2020, p. 310). In Slovakia, the relative winner was a coalition the Progressive Slovakia party (ALDE) and the SPOLU party (EPP), two newly established that are not represented in the national legislature, but are affiliated to the state president, Zuzana Čaputová (Láštic, 2020, pp. 324-325). In Slovenia, the ruling Slovenian Democratic Party (EPP) chose to run together with the non-parliamentary Slovenian People's Party (EPP), while New Slovenia -Christian Democrats, also EPP member, declined to join a common ticket for the EP election, due to the ruling party leader's closeness to Hungarian prime minister Orbán (Krašovec, 2020, pp. 332-333). In Spain, the United We Can Change Europe coalition included both members of the GUE/NGL and Greens/EFA groups (Delgado, López, Redondo, & Fernández, 2020, p. 344). Thus, one can see that cross-Europarty coalitions, especially in new democracies, such as Poland, Hungary, and Slovakia, reflect national party competition dynamics and efforts by the opposition to use EP elections to gather momentum against the ruling parties, thereby confirming the notion of European elections as second-order elections.

Electoral Effects
New democracies had a mean disproportionality level of 9.3, as opposed to 7.9 in old democracies, whereas they had somewhat fewer effective numbers of seat-winning parties, with mean values at 4.2 in new democracies and 4.8 in old democracies. Since party systems in old democracies are generally more consolidated than in new ones, one would expect lower effective numbers of seat-winning parties in those cases. After Brexit, the mean disproportionality fell to 8.9 in new democracies, and 6.8 in old democracies. The effective number of seat-winning parties did not change in new democracies, yet slightly rose (4.9) in old democracies. Similar to Farrell and Scully (2005, p. 980), the assessment of correlation between the number of MEPs elected (delegation size), the average district magnitude, and the effective threshold, showed statistically significant relations to Gallagher's index levels, while the interaction between the effective number of parties and disproportionality was not significant. The effective threshold showed a high positive correlation (0.5), which is lower than the results for the 2004 EP election obtained by Farrell and Scully (0.73). When the correlations were run with post-Brexit data, this positive correlation surged to 0.64.
Further, the regression model that tested which electoral system elements and contextual factors (old vs. new democracies) could serve as predictors of levels of disproportionality confirmed the importance of the effective threshold. In the second model, post-Brexit data was used, which resulted in a very small drop in the percentage of the variance explained by the model. Delegation size, for data before Brexit, was found to have a negative effect on disproportionality, as expected. In other words, larger delegations also meant, in most cases, larger district magnitudes, and thus, lower scores on the Gallagher index. Besides that, other predictors were not found to be statistically significant, pointing to a need for further research and assessment of potential other factors that might contribute to variation in levels of disproportionality among EU member states.

Conclusion
The analysis presented in this paper has shown that EP elections still show significant national variations. In terms of electoral rules, variations in average district magnitude and effective thresholds create very different conditions for party competition, and give rise to different levels of disproportionality, failing to unify national electoral arenas into a single, European electoral arena. Variations in the application of seat allocation methods are not that pronounced and have a minor effect on electoral outcomes. Regarding electoral strategies, in many cases, particularly in new democracies, parties will enter electoral coalitions that cross Europarty membership as a response to national electoral needs, and not European concerns. Thus, the assumption that parties in new member states will be inclined to enter coalitions in general, and cross-Europarty coalitions in particular could be confirmed. East-West differences in both rules and effect have been confirmed as well, with differences less pronounced when post-Brexit data is considered.
The different electoral effects produced higher levels of disproportionality and lower effective numbers of seat-winning parties in new democracies. Apart from the effective threshold, which had a strong effect on increased levels on the Gallagher's index, other predictors, besides delegation size (number of MEPs), measured before Brexit, were not found to be statistically significant. Finally, at the present state, national electoral rules, as well as strategies, and effects, cannot be understood as supportive of the notion of an ever-closer electoral Union. A future electoral reform that would try to create a harmonized electoral framework for EP elections would have to specifically design a model with low variation in district magnitude, as a key component in any electoral system. Thus, the national rules, strategies, and effects of EP elections point to incomplete Europeanization of party competition.