The 2014 European Parliament Elections in Southern Europe: Second-Order or Critical Elections?

Abstract

The article provides a comparative analysis of the 2014 European Parliament elections across the six countries of Southern Europe by gauging the performance of the second-order election model. Both the aggregate hypotheses of this model and evidence of micro-level foundations are analysed. The findings reveal that the political and electoral consequences of the economic crisis have not drastically challenged the second-order character of these elections in Southern Europe. However, electoral behaviour also exhibits some indications of a more critical contest.

Keywords::

• European Parliament Elections 2014
• Southern Europe
• Second-Order Election Model
• Economic Crisis
• Abstention
• Protest Vote
• Critical Elections

Since 1979 when the first European Parliament (EP) elections were held, national politics have dominated the campaigns and election results. In most member states of the European Communities, which later became the European Union (EU), European issues have been relatively unimportant or insignificant. Consequently, these elections have been described as ‘second-order’ national contests (Reif & Schmitt 1980). However, what was at stake at the European level during in the earlier years of the EP is certainly not the same as what is at stake 35 years later in the enlarged 28-member EU.

Over the past few years, the process of European integration is said to have entered its ‘post-functionalist’ phase (Hooghe & Marks 2009; Taggart & Szczerbiak 2004). Political conflicts increasingly cross national boundaries within the ‘higher level’ of the multi-level system of European governance (Hooghe & Marks 2009). Nevertheless, many studies prior to the 2009 EP elections have almost invariably concluded that voting decisions in these elections have been based on national politics (see Hix & Marsh 2007, 2011; Schmitt 2005, 2009; Van der Eijk & Franklin 1996) and that the parties have failed to campaign on EU issues (see Reif 1985; Weber 2007).

The elections of 22–25 May 2014 may have represented a new direction in this respect. This is due to two features, both of which had the potential to alter the second-order character of the contest and modify the configuration of electoral competition at the European level. The first factor is related to the financial and economic crisis that has affected Europe (and its southern periphery in particular) and the important role the EU has played in the regulation of this crisis. Suddenly, as a result of these EU policies, European governance became a polarising issue in some electoral arenas. A central question is whether in these elections EU politics (that is, a political dimension ranging from outright opposition to full support of integration) shaped political constestation within and among at least some of the member states (Bakker, Jolly & Polk 2012; Gabel & Hix 2002; Hix, Noury & Roland 2006; Hix 1999; McElroy & Benoit 2007). The second factor refers to the EU decision-making process, and in particular the efforts towards strengthening the role of the EP.

In 2014, for the first time in the history of EP elections, an explicit attempt was made to base the appointment of the president of the European Commission on the EU-wide results of the election. The Lisbon Treaty states that the president of the European Commission is to be elected by the EP, based on a proposal by the European Council that takes the results of the EP elections into account (Article 17[7] TEU). The campaign motto of the EP for these elections – ‘This time is different’ – was meant to capture this change. The nomination of leading candidates or Spitzenkandidaten (the German term that has often been used) for the post by the major European political groups was meant to strengthen the connection between voter preferences expressed in EP elections and the selection of the head of the EU executive,1 as well as to mobilise citizens’ interest and participation in these elections. Moreover, it was hoped that the nomination of lead candidates would strengthen the political role of both European political parties as well as the candidates competing in these elections. The question to be addressed here is whether these two features have actually changed the character of the campaign towards a ‘genuine European contest’ or whether the dominant role of the domestic political arena has been reaffirmed.

Initial empirical evidence on the electoral effects of this institutional and procedural innovation is not too impressive. It seems the campaign of the Spitzenkandidaten has had a limited mobilising effect (those who recognised them were found to be somewhat likelier to turn out; Schmitt, Hobolt & Popa 2015), while citizen awareness of them did not improve their evaluation of the performance of EU democracy, but rather polarised these evaluations among segments of society that were positive and critical towards the EU (Rohrschneider, Schmitt & Popa 2015). We are thus already sceptical about whether these new elements in the electoral process have indeed challenged the second-order nature of this ‘less important election’ (Marsh & Mikhaylov 2010).

Another aspect of the 2014 EP elections is that they represent a crucial test of EU electoral democracy during a time of unprecedented attacks on EU institutions – in terms of their intensity – for their lack of accountability and popular input as well as their policy directions, which appear to be dictated exclusively by elite concerns. This is particularly relevant for Southern European EU member states. These countries – Italy, Greece, Spain, Portugal and Cyprus (but not Malta) – have been hard hit by the economic crisis, and their governments have been forced by Eurozone authorities to implement harsh austerity policies.2 These policies have to varying degrees damaged the credibility of and support for governing parties across the region. They have simultaneously uncovered the increasing lack of manoeuvring room for national governments within the Eurozone, which has become an increasingly important dimension of EU politics. The result has been a surge of Euroscepticism in the countries of the southern periphery (and far beyond; see Treib 2014), countries that were hitherto some of the EU’s most Europhile members (Andreadis et al. 2014; Clements, Nanou & Verney 2014; Freire, Teperoglou & Moury 2014; Verney 2011).

The sovereign debt crisis and its consequences for Southern European electoral politics, and the ensuing debate about Europe, highlight the importance and complexities of the 2014 EP elections in these countries. This special issue is a result of a joint effort to provide an in-depth analysis of voting behaviour in these elections across the six countries by gauging the performance of the second-order election (SOE) model (Reif & Schmitt 1980). We also pay attention to the specific background of the electoral contests and electoral campaigns, as well as to the trends in political attitudes leading up to the EP election.

The Context of the 2014 EP Elections in Southern Europe

The 2014 EP elections in Southern Europe have received attention in the media, but much less in scholarly publications. This special issue is the first major collaborative effort to analyse these elections in Southern Europe. Their particular background was largely framed by the sovereign debt crisis and the economic recession. The starting points of the crisis were different in each country, as were their trajectories during the past few years. Three countries (Greece, Portugal and Cyprus) had signed bailout agreements – the so-called Memorandums of Understanding (MoUs) – with the Troika, made up of the European Central Bank (ECB), the European Commission (EC) and the International Monetary Fund (IMF).3 Spain and Italy did not have to go through formal bailouts, but they also experienced serious economic problems. Malta was not hit by the crisis. And while it is beyond the scope of this article to present a detailed account of the economic downturn in each country, some selected economic indicators are emblematic of the exceptional circumstances under which the EP election took place.4 According to 2013 Eurostat figures the unemployment rate reached 27.5 per cent in Greece, 26.1 per cent in Spain, 16.4 per cent in Portugal, 15.9 per cent in Cyprus, 12.2 per cent in Italy and 6.4 per cent in Malta (the 28-member-state EU average was 10.8 per cent).5 Real gross domestic product (GDP) growth (percentage of GDP) is another indicator of the harsh economic realities in the region: in 2013 all economies – except that of Malta – shrank: by  − 5.4 for Cyprus,  − 3.9 for Greece,  − 1.9 for Italy,  − 1.4 for Portugal and  − 1.2 for Spain. The economic crisis that had swept the world six years before clearly lingered on in Europe’s southern periphery.

It is known that ‘economic conditions interact with political events, campaign strategies, and economic policies in very consequential ways’ (Magalhães 2014, p. 8). However, the political consequences of the economic crisis were far from uniform across the countries of our study. The Portuguese ‘post-bailout election’ of June 2011 has not significantly changed the party system relative to the 2009 election (Magalhães 2012). In Spain, party system fragmentation remained largely unchanged after the national election of November 2011, even though the incumbent PSOE (Partido Socialista Obrero Español; Spanish Socialist Worker's Party) suffered massive losses. The disappointment of left and centre-left voters was expressed mainly through switching their allegiance to smaller parties (Torcal 2014). The Cypriot national election of May 2011 was held when Cyprus had lost access to international capital markets, but this led neither to an open banking crisis nor to discussions on a bailout programme. The result of the election has not changed the political landscape of the country either, but has rather continued the ongoing process of disengagement and loss of political confidence among Cypriot citizens (Christophorou 2012). By way of contrast, the three national elections held when the crisis intensified – in Greece (May and June 2012) and Italy (February 2013) – have changed the political universe of these countries significantly. Both contests were termed ‘earthquake’ elections (Bellucci 2014; Teperoglou & Tsatsanis 2014). Therefore, when analysing and comparing the results of the 2014 EP elections with previous national elections in these Southern European countries, one must remember that old certainties about the electoral allegiances of citizens have already been challenged in some countries, while in others the main features of party competition have been retained (Magalhães 2014, p. 8).

Another feature of the 2014 EP elections in Southern Europe is that they took place at a time of growing anti-EU sentiment. This reversal of pro-European stances is clearly connected to the onset of the economic crisis (Freire, Teperoglou & Moury 2014). In Italy, Greece, Spain, Portugal and Cyprus, there has been a sharp fall in positive evaluations of EU membership and in judgements that the country has benefited from EU membership (see Figures A1a and A1b, online Appendix). In Malta, on the other hand, which experienced a decline of EU support in the period 2009–11, high levels of positive evaluations for both indicators are recorded thereafter.

One of the themes of this special issue is therefore related to this increase in Euroscepticism. Eurozone policy choices have become a polarising issue in the political arena and seem to affect overall political attitudes towards EU integration. This suggests the ‘sleeping giant’ (Van der Eijk & Franklin 2004) could be stirring. In previous EP elections, traditional dimensions of political competition (related to the general left–right framework of orientation) proved far more significant in determining voter choices (De Vries 2007). It could have been foreseen that the EU dimension would be more significant this time, especially in the countries of the Southern European periphery that have been hit hard by the economic crisis. These countries have been portrayed as weak links in the Eurozone and have generally been forced to endure higher levels of austerity than the rest of the continent. However, as the articles in this special issue show, the European debate was not as influential as had been expected and the election campaigns have again been dominated mainly by domestic issues.

Comparing the Electoral Campaigns

Nevertheless, there are some noteworthy variations between the different countries covered here. For example, in their article Freire and Santana-Pereira show that in the Portuguese election campaign criticisms about the European dimension of Portuguese politics were raised mainly by the communists and the radical left and not by either of the major parties, the centre-left PS (Partido Socialista; Socialist Party) and the centre-right PSD (Partido Social Democrata; Social Democratic Party). The EP election was the first in the ‘post-bailout’ period and the debate focused on whether the end of the MoU was a ‘clean exit’ (without additional financial assistance).

The Italian electoral campaign was the most European of all. As Segatti, Poletti and Vezzoni argue, ‘the 2014 EP election took place in a context that might have altered the usual calculus of voters how they vote (or abstain from the vote) at European elections’. The fact that the Italian party system encompasses some openly Eurosceptic parties certainly contributes to this. Moreover, the governing centre-left PD (Partito Democratico; Democratic Party) and its Prime Minister, Matteo Renzi, mitigated the party’s pro-European stance.

In Greece, according to the article by Teperoglou, Tsatsanis and Nicolacopoulos, the electoral campaign lacked any true European dimension, despite the fact the leader of the main opposition, Alexis Tsipras, was the leading candidate of the Party of the European Left (EL). Greece is the only Southern European country in which the EP election coincided with another contest, in this case the second round of the municipal and prefectural elections, which undermined the European character of the EP election.

The debate around Europe was particularly absent in Spain. According to Cordero and Montero, none of the parties adopted a clear Eurosceptic discourse. The campaign focused on domestic issues, particularly Catalonian independence. Moreover, the public’s profound disillusion with the political parties and the discussion about the end of the two-party system dominated a shallow campaign.

Charalambous, Papageorgiou and Pegasiou note that the Cypriot electoral campaign concentrated on the bailout agreement and austerity. Efforts by the Cypriot government to transform the country into an important player in energy networks between the Middle East and the EU also played a role. Overall, domestic issues, such as the reduction of energy prices for households and the construction of a new power plant, dominated the Maltese election campaign. Nevertheless, as Carammia and Pace note, the visits to Malta of Jean-Claude Juncker (European People’s Party – EPP) and Martin Schultz (Party of European Socialists – PES), candidates for the presidency of the EC, made European issues more relevant, although the candidates emphasised domestic issues, such as the need for an EU approach to immigration.

Data from the European Election Study (EES) 2014 (Schmitt et al. 2015) reveal that the Maltese were those most likely to follow news about the EP elections on TV (37 per cent answered ‘often’, while the EU average is around 20 per cent). Greeks were next with 27 per cent, while propping up the table were the Portuguese (ten per cent). Regarding the recognition of the candidates for the EC, the visibility of Jean-Claude Juncker, Martin Schultz and Guy Verhofstadt (Alliance of Liberals and Democrats for Europe – ALDE) in Southern Europe was lower than the EU mean, but with some exceptions. Juncker’s recognition rates were higher among the Maltese and Greeks, while Schultz was more visible among the Maltese, Italians and Greeks.6

The context of the 2014 EP election in Southern Europe was also shaped by the collapse of positive sentiments towards national and European political institutions, especially after the onset of the economic crisis (see Figures A2a, A2b and A2c, online Appendix). The greatest expression of distrust was from the citizens of Italy, Greece, Spain and Portugal. The growing resentment of the political class in each country, the profound distrust of political parties, national governments and parliaments, is accompanied by a remarkable decline in the levels of satisfaction with national and EU democracy (Roth, Nowak-Lehmann & Otter 2013).

Still Second-Order or Critical Elections?

One of the goals of this special issue is to assess the continued utility of the SOE model (Reif & Schmitt 1980) under changing socio-political circumstances. The model has received considerable empirical support (Hix & Marsh 2011; Marsh 1998; Reif 1985, 1997; Schmitt 2005, 2009; Van der Brug & Van der Eijk 2007) and the conceptual framework of the articles of this special issue is built upon its assumptions.7 However, periods of economic turmoil, such as that experienced across Southern Europe, are the most likely triggers for lasting political change and durable political realignments (Abramowitz & Saunders 1998; MacDonald & Rabinowitz 1987).

The loss by national governments of the ability to steer economic policy in the Eurozone, the loss of national sovereignty and the parallel rise in the reach of European institutions, together with the recent growth in importance of the EU dimension in the EP (Hix, Noury & Roland 2006; McElroy & Benoit 2007 with contrary evidence), motivate our interest in whether these elections were ‘critical elections’ in the sense described by V.O. Key (1955). According to Key, elections are critical when the traditional coalitions between social groups and their political agents are subject to profound and lasting realignments. These coalitions are based on the perception of effective representation of the interests of social groups by their alleged political agents. Once these interests, or problem perceptions and issue concerns, are no longer represented in parliament and government, partisan alternatives – old or new – are likely to move in and present a more credible electoral choice option. In terms of EU policy issues, in an EP election or at the level of national electoral politics, such a process may be capable of significantly affecting the politics of European integration.8 Therefore, this special issue aims to answer the question of whether these elections can still be classified as typical second-order contests, or whether we can detect signs of a more ‘critical’ character as would be caused by profound and lasting socio-political realignments.

The articles of the special issue incorporate country-specific research questions. The authors of the Italian article focus on the success of the PD. A central question of the Greek article is whether the 2014 EP election has provided signs of party system stabilisation following the double national ‘earthquake’ elections in 2012, or whether the extreme fluidity and volatility of the previous few years continue. The Spanish article deals with the question of the different electoral ‘battles’ around these elections, while the Portuguese article analyses the new conditions under which the ‘sleeping giant’ of EU scepticism might emerge. In the case of Cyprus, particular attention is given to the analysis of abstention, which was exceptionally high in this small Mediterranean country. Finally, the Maltese article emphasises the sources of what is perceived as ‘Maltese exceptionalism’.

Properties of Second-Order Elections

As we have said before, the distinction between first-order elections (FOEs) and SOEs dates back to the study by Reif and Schmitt (1980), in which the first EP election of 1979 was showcased as an example of a SOE. Not all of this was new at the time of course. Carrying out a systematic study of the interdependence of political behaviour in different types of elections was probably first proposed in the literature on United States (US) mid-term election results. The two main approaches of that literature are the ‘surge and decline’ theory (Campbell 1960) and the ‘referendum’ theory (Tufte 1975), with Stimson’s (1976) cyclical model of presidential popularity somewhere in between. Contemporary German literature on the systematic connection between federal and state election results was also a fertiliser of the SOE model (Dinkel 1977). According to the model, FOEs are those which determine national governments. These are national parliamentary or presidential elections, depending on the political regime.

Second-order national elections, by contrast, are less important elections. In addition to EP elections, other examples of SOEs are local, municipal and regional elections, by-elections in Britain and mid-term elections in the US and in many presidential systems of Latin America. The main hypothesis, and to some degree the novelty and innovation, of the Reif and Schmitt model is its reference to different political arenas. In order to understand the results of EP elections, one first has to appreciate the decisive role of the political situation in the first-order political arena at the time when SOEs are being held (Reif & Schmitt 1980, p. 8).

The central assumption of the SOE model is that in all these less important elections, EP elections among them, there is less at stake because the results will not determine the composition of the executive (Reif & Schmitt 1980, p. 9).9 However, at least for the dynamic multi-level system of governance of the EU, the ‘less at stake’ dimension should not be conceptualised as a static element of the model, considering the (so far indirect) linkage of the 2014 EP election results with the selection of the president of the EC. To the degree that there is less at stake, one can derive some aggregate hypotheses for the 2014 EP election results. A first set concerns electoral participation, while a second focuses on party choice.

We will turn to these aggregate hypotheses below, but first let us note here that these have overshadowed the (somewhat less prominently exposed) theoretical micro-foundations of the model. These are the hypotheses about the motivations and intentions of individual voters which drive their behaviour in an SOE. Only recently have scholars started to uncover some of these micro-foundations and to confront them with micro-level empirical evidence (Hobolt & Wittrock 2011; Schmitt, Sanz & Braun 2008, 2009; Weber 2009). Some articles in this special issue analyse aspects of these micro-level indicators, making them an important contribution to the most significant attempt to develop the SOE model over the last thirty years. The articles rely on fresh and innovative data. The empirical analyses of the Italian, Greek and Spanish EP elections are mainly built upon data from online surveys conducted within the framework of the EES 2014 (Schmitt et al. 2015), while in the Portuguese article the post-electoral survey of EES 2014 is used. Pre-electoral surveys are used for the analysis of the Cypriot and Maltese EP elections.

Participation

A first hypothesis of the SOE model is that turnout is expected to be lower in an EP election (as well as other SOEs) than in both the preceding and the subsequent first-order contest (Reif 1985, p. 15). However, the levels of participation are also affected by the timing of the EP election in relation to the first-order electoral cycle. Turnout is highest when European and national elections are held concurrently (Franklin, Van der Eijk & Oppenhuis 1996). However, when EP elections are held immediately after an FOE, ‘electoral fatigue’ prevents higher levels of participation. Higher turnouts are typically observed towards the end of the first-order electoral cycle (Schmitt & Van der Eijk 2007). Another prediction of the aggregate model is that a higher percentage of invalid votes will be recorded in an EP election than in the previous or subsequent FOE (Reif & Schmitt 1980, p. 9). This, however, has not received much attention in subsequent studies.

Turning to the 2014 EP election, again we observe that participation was lower in all Southern European countries than in each previous first-order national contest. The lowest level of participation was recorded in Portugal. If we compare the figures with participation levels in the previous FOE, the greatest declines are observed in Cyprus, Spain and Portugal (Table 1). Greece represents something of an outlier, since the decline in participation reached only 2.5 per cent compared with the national elections of June 2012 (Table 1 and Figure 3Aa, online Appendix). It is also the only country in which participation has increased (+7.34) compared with the 2009 EP election. As noted previously, one possible reason for the high participation in Greece could be the fact that the second round of the local elections took place on the same day. Moreover, as is argued in the Greek article, we have also to consider the ‘referendum’ character of the election and the polarised first-order political arena.

Table 1 Summary of Testing the Aggregate Hypotheses of the SOE Model in the 2014 EP Elections in Southern Europe

SOE aggregate hypotheses	Italy	Greece	Spain	Portugal	Malta	Cyprus
1. Lower turnout compared with previous FOE	Yes	Yes	Yes	Yes	Yes	Yes
	( − 17.8%)	( − 2.5%)	( − 25.1%)	( − 24.4%)	( − 18.1%)	( − 34.7%)
2. Higher blank/invalid votes compared with previous FOE (y = yes/n = no)	5.3% (y)	3.8% (y)	4.1% (y)	7.5% (y)	2.2% (y)	2.9% (y)
3. Government party losses compared with previous FOE	No (PD: +15.4%; SE:  − 7.58%; Nuovo Centro Destra: +2.6%)	Yes (ND:  − 6.94%; Olive Tree/PASOK:  − 4.26%)	Yes (PP:  − 18.5%)	Yes (Aliança Portugal [PSD and CDS–PP]):  − 22.7%)	Yes (PL:  − 1.4%)	No (DISY: +3.5%)
3.1 Government party losses related to the electoral cycle	Honeymoon effects	Yes	Yes	Yes	Honeymoon effects	No
4. Opposition party win	No	Yes	No	Yes	No	No
4.1 Opposition big parties losses compared with previous FOE	Yes (Forza Italia:  − 4.8%; M5S:  − 4.4%)	Slight losses (SYRIZA:  − 0.3%)	Yes (PSOE:  − 5.8%)	No (PS: +3.4%)	Yes (PN:  − 3.3%)	Yes (AKEL:  − 5.7%)
4.2 Total losses for big parties	Only opposition – see above	Yes (ND and SYRIZA:  − 7.2%)	Yes (PP and PSOE:  − 24.3%)	Only governmental – see above	Yes (PL and PN:  − 4.7%)	Only opposition – see above
5. Small parties gains (higher ENEP in the 2014 EP election compared with previous FOE)	No (3.75 compared with 4.94)	Yes (6.8 compared with 5.2)	Yes (6.8 compared with 2.6)	Yes (5.7 compared with 3.9)	No (2.24 compared with 2.05)	No (4.17 compared with 3.86)
5.1 Anti-EU parties and extreme-right parties better performance	Yes	Yes	Yes	No	Yes	No
5.2 Emergence of new parties that gain seats in the EP	Yes	Yes	Yes	No	No	No
Electoral cycle	Mid-term election (end of honeymoon period)	Mid-term election	Mid-term election	Mid-term election	Mid-term election (end of honeymoon period)	Mid-term election
Previous FOE	24–25 February 2013	17 June 2012	20 November 2011	5 June 2011	9 March 2013	22 May 2011
Concurrent election with EP		Local elections

Average turnout across Southern Europe was 52.2 per cent – around ten points higher than the all-EU average (42.5 per cent). In the history of all EP elections in Southern Europe, there have been lower levels of participation in them than in the preceding and subsequent national elections. Nevertheless, participation in Southern Europe over the period 1979/81–2009 was always higher than in the EU as a whole (for more details, see Teperoglou 2010, pp. 260–261), thus there is some continuity in terms of regional patterns in participation.

The Notion of the First-Order Electoral Cycle and Types of Voting in EP Elections

A central assumption of the SOE model is that vote switching between national and EP elections can be largely attributed to two contextual factors. The first is related to the low importance attached to EP elections because no national government will be formed; the second is the timing of the EP election within the national election cycle (Marsh & Mikhaylov 2010, p. 11). Given the fact that EP election outcomes in the past could not be expected to have severe political consequences, one form of vote switching was to vote for the party the voter preferred most (‘sincere defection’; see Schmitt, Sanz & Braun 2008, 2009). In this case, voters do not act strategically as they may do in an FOE, e.g. in order not to waste their vote on a party that is unlikely to be a viable contender for government office. In this sense voters tended to ‘vote with their heart’ (or sincerely; see Oppenhuis, Van der Eijk & Franklin 1996, pp. 301–304). This type of sincere voting is an ‘expressive’ act (Reif 1985, pp. 8–9), while the reverse holds in national elections in which voters in many cases vote ‘with the head’ (tactically or strategically; see Marsh & Franklin 1996, pp. 16–21; Oppenhuis, Van der Eijk & Franklin 1996, pp. 301–304). Based on these mechanisms, three main hypotheses of the SOE model are formulated: government parties lose in the EP elections, big parties perform worse and small parties profit from this increase of sincere voting in EP elections (Reif & Schmitt 1980, pp. 9–10; Reif 1997, p. 118).

The magnitude of the losses for government parties is related to the timing of the SOE in the first-order electoral cycle. The losses will be greater the closer an SOE is to a first-order mid-term.10 This linkage of the different electoral performance of the incumbent party (or parties) in the 2014 EP election as a function of the time distance of the SOE from the previous FOE is analysed in all the articles here, taking into account some theoretical considerations. For the mid-term period (12–36 months of a four-year government), there is a consensus that losses will be greater (e.g. Reif 1985; Van der Eijk & Franklin 1996). During this period, disappointment with the performance of the government is greater (Marsh 1998, p. 606). For the electoral ‘honeymoon’ period (0–12 months), some scholars (Reif & Schmitt 1980) suggest the governmental parties will receive greater or near identical support in the EP election (post-electoral euphoria). Others (Van der Eijk & Franklin 1996) argue more voters will tend to vote with their heart.

Finally, for the later term (37–48 months) some argue (Van der Eijk & Franklin 1996) that the closer they come to the next FOE, the more likely it is voters will cast protest votes (‘strategic defection’; see Schmitt, Sanz & Braun 2008, 2009) or will ‘vote with the boot’ (Oppenhuis, Van der Eijk & Franklin 1996, pp. 301–304). In contrast, Reif and Schmitt (1980) claim the later-term period is characterised by some recovery in national government popularity, so parties will tend to lose fewer votes than during mid-term elections. Instrumental voting against the government is also evident in mid-term EP elections, where the motivation to express possible discontent is even stronger (‘cyclical signalling’; see Schmitt, Sanz & Braun 2008, 2009). This replicates Tufte’s (1975) referendum theory for the US mid-term losses of the party of the president.

Government Party Losses

The last EP election for all Southern European countries took place in the middle of the national electoral cycle, although in the case of the Italian and Maltese election it could be better defined as having taken place at the end of the honeymoon period (almost 15 months after the last national election in both cases). The possibility of government losses is clearly confirmed in Greece, Spain and Portugal. The greatest anti-government swing compared with the previous FOE was recorded in Portugal for the incumbent coalition of PSD and the conservative CDS-PP (Centro Democrático e Social - Partido Popular; Democratic Social Centre - People's Party) ( − 22.7), followed by Spain ( − 18.5) for the centre-right PP (Partido Popular; People's Party) and Greece ( − 11.2) for the incumbent coalition of the centre-right ND (Nϵ´α Δημoκρατι´α; New Democracy) and the centre-left PASOK (Παvϵλλη´vιo Σoσιαλιστικó Kι´vημα; Panhellenic Social Movement). The fact that in these three countries the main governing parties, as well as the main Italian opposition party, which also lost ground, are affiliated to the EPP contributed to the under-representation in the EP of MEPs from the ‘Old Southern Europe’ in this group compared with the 2009 EP (Table 2). In Malta the losses for the Labour Party (PL; Partit Laburista) were marginal and are better explained as the honeymoon effect of the FOE cycle. As Carammia and Pace note, these were lower losses than in the two previous EP elections (see also Table 3).

Table 2 Southern European Representation in the 8th European Parliament (2014 EP elections) Compared with the 7th Legislature (2009 EP elections)

	EPP	S&D	ECR	ALDE	GUE/NGL	Greens–EFA	EFDD^*	NI	Total†/% of new MEPs‡
Italy	13 (FI and La Destra); 3 (Nuovo Centro Destra – UDC); 1 (SVP)	31 (PD)		0	3 (L'Altra Europa con Tsipras)		17 (M5S)	5 (LN)	73 (+1)/75%
	[ ↓ 18]	[ ↑ 10]		[ ↓ 7]	[ ↑ 3]		[ ↑ 8]	[ ↑ 5]
Greece	5 (ND)	2 (PASOK under Olive Tree); 2 (The River)	1 (ANEL)		6 (SYRIZA)	0	0	3 (GD); 2(KKE)	21( − 1)/100%
	[ ↓ 3]	[ ↓ 4]	[ ↑ 1]		[ ↑ 3]	[ ↓ 1]	[ ↓ 2]	[ ↑ 5]
Spain	16 (PP); 1 (UDC under. CEU)	14 (PSOE)		4 (UPyD); 2 (C's); 1 (CDC under. CEU); 1 (EAJ–PNV under CEU)	4 (IU under. IP); 1 (AGE under. IP); 5 (Podemos); 1 (EH Bildu under. LPD)	1 (ICV under. IP); 1 (Compromis under. Primavera); 1 (ERC under Coal. EPDD); 1 (NECat under. EPDD)		0	54(+4)/57%
	[ ↓ 6]	[ ↓ 7]		[ ↑ 6]	[ ↑ 10]	[ ↑ 2]		[ ↓ 1]
Portugal	6 (PSD); 1 (CDS–PP), both under. Aliança Portugal	8 (PS)		2 (MPT)	3 (PCP under CDU);1 (BE)				21 ( − 1)/57%
	[ ↓ 3]	[ ↑ 1]		[ ↑ 2]	[ ↓ 1]
Cyprus	2 (DISY)	1 (DIKO); 1 (EDEK)			2 (AKEL)				6 ( = )/67%
	=	=			=
Malta	3 (PN)	3 (PL)							6 (+1)/50%
	[ ↑ 1]	=
Total Southern Europe 2014/ EU 28	51/219	62/191	1/71	10/68	26/52	4/50	17/48	10/52	181/751
Total Southern Europe 2009/EU 27	80/265	62/184	0/54	9/84	11/35	3/55	11/32	1/27	177/736

Notes: S&D: Group of the Progressive Alliance of Socialists and Democrats in the EP; ECR: European Conservatives and Reformists; Greens–EFA: The Greens/European Free Alliance; NI: non-attached members. SVP: Südtiroler Volkspartei -South Tyrolean People's Party; ANEL: Avϵξ´αρτητoι 'Eλληvϵς - Independent Greeks; KKE: Koμμouvιστικo˙ Ko˙μμα Eλλα˙δας - Communist Party of Greece; UDC: Unió Democràtica de Catalunya - Democratic Union of Catalonia; CEU: Coalición por Europa - Coalition for Europe; UPyD: Unión Progreso y Democracia - Union, Progress and Democracy party; C's: Ciudadanos, Partido por la Ciudadanía - Citizens, Party of the Citizenry; CDC: Convergència Democràtica de Catalunya - Democratic Convergence of Catalonia; EAJ-PNV: Eusko Alderdi Jeltzalea - Partido Nacionalista Vasco - Nationalist Basque Party; IU: Izquierda Unida - United Left; IP: Coalition Izquierda Plural - Plural Left; AGE: Alternativa Galega de Esquerda en Europa - Galician Alternative of the Left; EH-Bildu: Euskal Herria Bildu - Basque Country United; LPD: Los Pueblos Deciden - The People Decide; ICV: Iniciativa per Catalunya Verds - Initiative for Catalonia Greens; ERC: Esquerra Republicana de Catalunya - Republican Left of Catalonia; EPDD: L'Esquerra pel Dret a Decidir - Left for the Right to Decide; NeCat: Nova Esquerra Catalana - Catalan Left; PCP: Partido Comunista Português - Portuguese Communist Party; CDU: Coligação Democrática Unitária - Unitary Democratic Coalition, BE: Bloco de Esquerda - Left Bloc; DIKO: Δημoκρατικo˙ Ko˙μμα – Democratic Party; EDEK: Evιαı˙α Δημoκρατικη´ 'Evωση Kϵ˙vτρoυ – United Democratic Centre Union.

The table presents the changes in terms of allocation of seats per political group compared with the 2009 EP elections based on the allocation announced on 1 July 2009 (see: www.europarl.europa.eu/aboutparliament/en/00082fcd21/Results-by-country-%282009%29.html) and the allocation as of 25 November 2014. European Parliament. (2014) Review European and National Elections Figured Out, Special Edition, 2014 European Elections, Directorate General for Communication, Public Opinion Monitoring Unit, available online at: http://www.europarl.europa.eu/pdf/elections_results/review.pdf

In bold are the governmental parties at the time of the 2014 EP elections.

^*The comparison is made with EFD (Europe of Freedom and Democracy Group).

†European Council Decision of 28 June 2013 establishing the composition of the European Parliament (2013/312/EU); see: eur-lex.europa.eu/legal-content/EN/TXT/?uri = CELEX:32013D0312. In parentheses is the change in the allocation of seats between the 2009 EP election and the 2011 apportionment of members in the EP based on the amendment to the Lisbon Treaty which came into force on 1 December 2011 (in total 18 MEPs extra).

‡The proportion of the new MEPs in the EP of 2014–19 – compared with the previous legislation of the EP – is taken from the European Parliament (2014) (see above), p. 36.

Table 3 European Elections and Electoral Cycles in Southern Europe: Change in National Government Vote Support in EP Elections

European election	Change in vote percentages of the main incumbent party:^* EP elections compared with the preceding (or concurrent) national election
	Honeymoon: 0–12 months after national election	Mid-term: 13–36 months after national election	Later term: 37–48 months after national election
Italy
1979	− 1.9
1984	− 0.1
1989		− 1.4
1994	+9.6
1999		− 3.8
2004		− 8.4
2009		− 2.1
2014		+15.4
Greece
1981†	− 4.5
1984		− 6.5
1989†	− 3.2
1994	− 9.2
1999		− 8.6
2004	− 2.3
2009		− 9.5
2014		− 6.9
Spain
1987	− 5.04
1989			− 4.5
1994	− 8.01
1999			+0.9
2004	+0.9
2009		− 5.1
2014		− 18.5
Portugal
1987†	− 12.8
1989		− 17.5
1994		− 16.2
1999			− 0.6
2004		− 15.6
2009			− 18.4
2014		− 22.7
Cyprus
2004		− 6.8
2009		+3.8
2014		+3.5
Malta
2004		− 12.0
2009		− 8.8
2014		− 1.4

Notes: ^*In cases of a coalition government we consider only the change in vote percentages of the main coalition party. †Concurrent national election.

The two deviant cases are Italy and Cyprus. The centre-right government in Cyprus led by Democratic Rally (DISY – Δημoκρατικo´ς Συναγϵρμo´ς) increased its vote share (+3.5) compared with the 2013 national election. ‘Voting with the boot’ was expressed mainly towards the left-wing opposition and former government party, the Progressive Party of Working People (AKEL; Aνoρθωτικo´ Ko´μμα Eργαζo´μϵνoυ Λαoυ´). The authors argue that ‘the particularities of the Cypriot case in terms of government–opposition dynamics during the crisis appear to be an important reason why the “incumbent punishment” part of the SOE thesis is not confirmed’.

The 2014 Italian EP elections are notable mainly because of the victory of the main incumbent, PD (+15.4 compared with the previous FOE). The result is also remarkable in terms of the size of its electoral appeal (40.8 per cent of the vote). The SOE hypothesis is not verified either for the New Centre Right (NCD; Nuovo Centro Destra), which participates in the Renzi cabinet (+2.6). The third party in the Italian coalition government, Civic Choice (SC; Scelta Civica), founded by Mario Monti and which ran in the EP election as part of the European Choice coalition (SE; Scelta Europea), suffered extensive losses. Compared with the 2009 EP election, the configuration of power for the Southern European MEPs in the Group of the Progressive Alliance of Socialists and Democrats in the European Parliament (S&D) is stable mainly because of the electoral success of the PD (see Table 2).

The cyclical element in the evolution of support for the incumbent parties is verified in all our countries except Italy (Figure A3b, online Appendix). Segatti, Poletti and Vezzoni approach the PD’s electoral performance in terms of the honeymoon effects that followed the non-electoral change of government in February 2014, when Matteo Renzi won the PD primary and formed a new government: ‘similar effects may also follow a non-electoral government’s change, if this change represents a turning point in the national political context’. The result for the PD in the EP election is linked to the party’s strategy following the change of leadership, while the size of its electoral success is associated with the party’s poor performance in the 2013 national election.

One major conclusion from their individual-level analysis is that European issues were important in the voting choice of PD voters. For the first time, European matters appear as a dividing issue. The analysis of voting behaviour in the Greek EP election reveals that for the first time since the early 1980s more abstract stances towards European unification are beginning to become relevant to party choice. In both these Southern European countries we detect signs of a more ‘critical’ contest. The conclusion that electoral behaviour in the last Italian EP election, and to some extent in the Greek election, is not exclusively determined by national politics is in line with studies that consider the potential influence of European issues upon voter stances in EP elections (Carrubba & Timpone 2005; Ferrara & Weishaupt 2004; Hobolt, Spoon & Tilley 2009).

Big Parties’ Losses

A distinct but related hypothesis of the SOE model is that big parties will also perform poorly in EP elections. However, while governing parties lose votes regardless of their size (or party family; see Hix & Marsh 2007, p. 503), big opposition parties also tend to see their electoral fortunes fade (Marsh 2005).

This hypothesis does not hold true in Portugal and in Greece, but it is confirmed elsewhere in Southern Europe. In both Greece and Portugal, the main opposition party won the elections. In Portugal, the centre-left PS increased its share of the vote compared with the national elections of 2011, but, as Freire and Santana-Pereira argue, the ‘strong discontent with the incumbent government did not translate into a substantial victory for the PS’, and it was mainly the smaller parties that profited from this swing against the government. As Teperoglou, Tsatsanis and Nicolacopoulos note, one of the reasons the Greek EP elections could be described as a historic electoral contest is the victory of the radical left party of SYRIZA (Συvασπισμo˙ς Pιζoσπαστικη˙ς Aριστϵρα˙ς; Coalition of the Radical Left). Nevertheless, compared with the previous FOE, the proportional strength of SYRIZA remained stable (with a marginal shift of  − 0.3) and only increased again in the FOE in January 2015.

As for those countries in which incumbent party losses were not translated into an increase in the percentage received by other big parties, the most worthy of note is perhaps Spain, where the PSOE was apparently punished ( − 5.8). Cordero and Montero note that the electorate seemed to blame both the incumbent PP and PSOE for the economic crisis, letting the smaller parties win this EP election against the so-called bipartidismo.

In Italy, Silvio Berlusconi’s Forza Italia (FI) lost almost three million votes compared with those obtained by its predecessor, The People of Freedom (PdL; Il Popolo della Libertà), in the previous FOE. The Eurosceptic M5S (Movimento Cinque Stelle; Five Star Movement) lost the same number of voters compared with 2013. The elected MEPs of this party are the only ones from Southern Europe in the Eurosceptic European Group of Freedom and Direct Democracy (EFDD) (Table 2). One of the striking features of the results is the increased number of MEPs in the European United Left/Nordic Green Left (GUE/NGL) group.

Small Parties’ Gains

As noted above, the SOE model postulates that smaller parties (and in particular the new ‘flash’ and ideologically extreme parties) are expected to perform better in EP elections than the previous and subsequent FOEs (Reif 1985). Schmitt (2005, p. 662) suggested using the index of the effective number of electoral parties (ENEP, introduced by Laakso & Taagepera 1979) to test this hypothesis. If small parties perform better in the EP election, the ENEP will be higher than that for the preceding FOE. This is thought to originate in vote switches from a strategic FOE to a sincere SOE vote choice based on the desire of voters to express their true first preference, even if the chosen party will not gain representation (Schmitt, Sanz & Braun 2008, 2009). However, there is no uniform pattern regarding small parties’ gains in the 2014 EP election in Southern Europe.

Smaller parties performed better, particularly in Spain, but also in Greece and Portugal. Overall, and again with the exception of Italy, the gains for the smaller parties followed the first-order national electoral cycle (see Figure A3c, online Appendix). Furthermore, in Spain and Greece some of the new parties that emerged in the political arena prior to the elections managed to gain representation in the EP (the Spanish populist left party, Podemos [We Can], and the Greek Potami [The River]), confirming the hypothesis that new parties can perform well in EP elections (Reif 1985, p. 9). In Portugal, the Earth Party (MPT; Partido da Terra) entered the political arena.

While the overall result in Malta confirmed yet again the duopoly of Maltese politics, which creates a very special framework for party competition within the set of countries being studied here, Carammia and Pace note one striking exception: the ‘largely unnoticed result of the growth of the extreme-right, neo-fascist Imperium Europa (European Empire), which jumped to 2.7 per cent’.

In the south of Europe, the extreme right is strongest in Italy and Greece. The Italian Northern League (LN; Lega Nord) performed better than it had in the 2013 FOE, while in Greece the neo-fascist Golden Dawn (GD; Xρυση´ Aυγη´) continued to increase its share of the vote. In Cyprus, the extreme right-wing ELAM (Eθνικo˙ Λαι¨κo˙ Mϵ˙τωπo; National Popular Front) also increased its relative share of votes compared with the previous FOE.

Another indicator of party fragmentation is the concentration of the vote on the two main parties within a party system (see Figure A4, online Appendix). Here the result differs from country to country. In Italy and Cyprus, we see the share of the vote increasing – most significantly in Italy (+10.6). In Malta, the concentration of the vote in the two major parties [the conservative PN (Partit Nazzjonalista; Nationalist Party) and PL] was maintained, albeit at a lower level in the EP election, while in Portugal the party system is much more fragmented in the EP election, confirming a similar trend noted in the 2009 EP election. The combined vote share in the Greek EP election signifies the consolidation of the end of the old two-party system comprising PASOK and ND, which dates to the double earthquake elections of 2012. The most outstanding example, however, is that of Spain. The vote for the two main parties, PSOE and PP, fell by 24.3 per cent compared with the previous FOE, signifying a profound transformation of the party system. It remains to be seen whether the EP election in Spain was a prelude to another earthquake election in Southern Europe.

In Lieu of Conclusion

A major lesson from the six articles in this special issue is that the political and electoral consequences of the economic crisis have not drastically affected the second-order character of the 2014 EP elections in Southern Europe. However, the detailed analyses presented here reveal interesting cross-country variations. Reif and Schmitt’s model is empirically confirmed in the case of Spain. As the article shows, the last EP election is the most typical ‘low-stimulus’ contest in the history of EP elections in the country. A similar conclusion also emerges from the article on Greece, but only in terms of voting choice. The results of the Portuguese EP election confirm a classical second-order contest. Here, the characteristics of the SOE model are more accentuated than ever. The article for the Maltese EP election shows that, while the propositions of the SOE model are confirmed, a combination of domestic institutional, cultural-historical and contextual factors should not be dismissed when interpreting the results. The study of the Italian EP election presented in this special issue confirms – among other findings from individual-level analyses – the importance of the first-order national electoral cycle for its classification as a SOE. In other respects it is a deviant case in relation to the SOE model. Finally, some of the results of the Cypriot EP election are puzzling in terms of the SOE propositions, especially the success of the incumbent party.

The individual-level analyses of voting behaviour in the 2014 EP election in the contributions contained in this volume indicate greater differences between the Southern European electorates than the findings at the aggregate-level of analysis reveal. This underlines the importance of further analyses of the micro-foundations of the SOE model.

Overall, it appears Southern Europeans did not use the EP elections as an opportunity to send a clear message to Brussels. Nevertheless, the articles here demonstrate that electoral choice in the 2014 EP elections in Southern Europe has the characteristics of a classic SOE along with indications of a more critical contest in which the EU divide has become more significant, polarised and important in determining voting choices.

There is one final consideration we would like to share with our readers. It concerns the nature of critical elections in the EU’s multi-level electoral system, and in particular the possibility of a concurrence of such critical elections in a number of subsequent elections at different levels. In other words, we believe that sixty years after Key’s pioneering article, in the context of a more complex multi-level electoral system and a more competitive multi-party systems environment, ‘critical elections’ may extend over more than one level of the electoral system and over more than one particular election. We have assembled considerable evidence for such ‘stepwise’ critical elections on EU issues for Greece, where major realignments on the left of the political spectrum were observed during the FOEs, realignments that seem to have been consolidated in subsequent EP elections of a second-order nature. The final verdict about any critical realignment in European party systems on issues originating from the EU level of policymaking has not yet been announced.

Acknowledgements

We would like to thank all the authors who contributed to this volume, Susannah Verney the co-editor of South European Society and Politics who was responsible for this issue, and the anonymous referees. Last, but certainly not least, we are grateful to the main reviewer of the special issue.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Notes on contributors

Hermann Schmitt

Hermann Schmitt is Professor of Political Science at the Universities of Manchester (UK) and Mannheim (Germany). He has been involved in the series of European Election Studies from the very beginning in 1979 (www.europeanelectionstudies.net). He is also a co-director of the British Election Study 2015 and has contributed, in various roles, to additional comparative studies like the Comparative Study of Electoral Systems (CSES), the Comparative Candidate Survey (CCS) and the True European Voter (TEV). His main research interest is in comparative political behaviour, the multi-level nature of political representation and electoral democracy in the EU, and European integration more broadly conceived. He has published widely on all of these topics.

Eftichia Teperoglou

Eftichia Teperoglou is Lecturer at the Aristotle University of Thessaloniki and a researcher at the Centre for Research and Studies in Sociology, Instituto Universitário de Lisboa (CIES–IUL). Her main research interests are in the fields of political and electoral behaviour with a focus on European elections, comparative politics and public opinion. She has published work in international journals and in edited volumes. She is the author of the book Oι α˙λλϵς «ϵθνικϵ˙ς» ϵκλoγϵ˙ς: Aναλυ˙oντας τις Eυρωϵκλoγϵ˙ς στην Eλλα˙δα 1981–2014 [The Other ‘National’ Elections. Analysing the European Elections in Greece 1981–2014], Papazissis, 2015 in press.

Notes

^1. The European political parties that designated candidates were: EPP – Jean-Claude Juncker; PES – Martin Schulz; ALDE – Guy Verhofstadt; the European Green Party (EGP) – Ska Keller and José Bové; EL – Alexis Tsipras.

^2. While we concentrate here on Southern Europe, it should be noted that a major decline in support for the EU is also recorded for Ireland (www.pop.pt/en/graph2/politics/trust-in-the-european-parliament/politics/trust-in-the-european-commission/ie/?dg = ie), Latvia, Hungary and Romania, which are under IMF programmes too.

^3. Greece signed the first MoU in May 2010, while the second became active in March 2012. The Portuguese MoU was signed in June 2011, while the agreement in Cyprus took place in March 2013.

^4. For a detailed overview of the economic indicators of crisis before 2011, see Bosco and Verney (2012); for Italy, Greece and Portugal before 2013 see Verney and Bosco (2013) and Freire et al. (2014).

^5. Comparing with the 2009 figures for all countries, there is an increase in unemployment, ranging from 17.9 points for Greece to 4.4 points for Italy (Eurostat).

^6. The question was asked for the EPP, PES and ALDE candidates. The unweighted figures are as follows: Junker (Italy 13.2 per cent; Greece 21.84; Spain 10.94; Portugal 12.58; Cyprus 17.17; Malta 34.37; EU average 18.91). Schultz (Italy 20.26; Greece 18.25; Spain 10.32; Portugal 9.2; Cyprus 12.08; Malta 49.82; EU average 16.87). The visibility of Verhofstadt was below the EU average (8.78 per cent) in all Southern European countries, ranging from a low of 1.76 in Greece to a high of 8.71 in Italy (EES 2014; see Schmitt et al. 2015).

^7. Testing the SOE model during the 2004 EP election in the eight Eastern European member states demonstrates that government losses did not follow the cyclical pattern and that many small parties did not benefit. These deviations can be explained by the fact the SOE model is built upon consolidated electoral and party systems, while in most eastern EU member states stable and consolidated party systems have yet to develop (Schmitt 2005, p. 666).

^8. The victory of Franklin D. Roosevelt and his New Deal in the 1932 US presidential election, at a time of deep economic recession, constitutes an archetypical critical election.

^9. The original study by Reif and Schmitt (1980, pp. 10–15) included some other analytical dimensions: the specific-arena, the institutional-procedural, the campaign, the main-arena political change and, finally, social and cultural change. However, the less-at-stake dimension is the one we have almost exclusively analysed.

^10. This assumption of the SOE model reflects US electoral cycle literature (Goodhart & Bhansali 1970) and originates from Dinkel's (1977) study of the German Landtag elections.

Appendix

An online Appendix is available for this article which can be accessed via the online version of this journal at: http://dx.doi.org/10.1080/13608746.2015.1078271