The Left–Right Dimension, Europe and Voting in Bailout Europe

Abstract

This chapter investigates several questions concerning the relative importance of EU issue voting in Greece, Ireland, Portugal and Spain. First, the relationship between EU issue proximity and left–right proximity for voting behaviour. Second, the importance of both these issues contrasting mainstream and challenger parties. Third, whether EU parliamentary debates, measured in terms of tone, moderate the importance of EU issue voting. We use post-election online panel surveys collected in 2019–2020 in each of the bailout countries to analyse these questions. Our results show that, in all countries surveyed, EU issue voting does cross-cut the left–right dimension. Yet, left–right proximity has a higher impact on likelihood to vote for a party than EU issue proximity in all countries. Ireland stands out due to the relative irrelevance of both issues in explaining vote choice. European issue proximity to parties’ positions is a significant predictor of voting behaviour, but not a more important determinant of likelihood to vote than left–right for challenger parties. Finally, the tone employed by parties in EU parliamentary debates matters: For parties which have a more negative tone, EU proximity determines vote choice more strongly than for parties with a positive or neutral EU tone in parliament.

Introduction

In the previous chapter, an experimental analysis of EU issue voting in the six countries included in this volume, namely Belgium, Germany, Greece, Ireland, Portugal and Spain was implemented. Now, we revisit the topic, but using observational data collected in countries which underwent bailouts during the Eurozone crisis: Greece, Ireland, Portugal and Spain.

Despite the fact that there were evident consequences from the onset of the crisis in terms of government incumbency and party system change (Hutter & Kriesi, 2019), less is known about the importance of the EU in relation to the left–right dimension of competition in these countries (for exceptions, see Hobolt & Tilley, 2016; Santana & Rama, 2018), which is the focus of this chapter. Our goals are the following: to establish the importance of the left–right dimension of competition for the vote in Greece, Ireland, Portugal and Spain, and to understand whether the EU issue reinforces or undermines this left–right divide in bailed-out Europe. Further, we test whether the relative importance of this issue is significantly different comparing mainstream and challenger parties. In the final section, we consider the importance of EU integration in the context of parliamentary debates: namely, we integrate the parliamentary debates data analysed in Chapter 3, to test whether EU issue voting impact on likelihood to vote for a party increases when the party is more negative about the EU in parliamentary debates.

The questions we pose will help us understand how the EU relates to the left–right dimension of competition in each of these countries, allowing us to make some generalizations about the contexts which shape voting in post-bailout Europe in a post-crisis scenario. In particular, we aim to contribute to two ongoing debates in the literature.

Firstly, extant research has shown that challenger parties tend to politicize novel issues, and this is then reflected in their importance in relation to voting for mainstream parties (Hobolt & Tilley, 2016; Santana & Rama, 2018). Yet, there are interesting changes in bailed-out European countries which are worth considering. In the case of Greece, where the main challenger party, namely SYRIZA, became incumbent since 2015. In Portugal, the PCP and Bloco de Esquerda, while remaining outside government, had a coalition of parliamentary incidence with the minority PS government between 2015 and 2019.

Second, we wish to test whether parliamentary debates can shape EU issue voting. Politicization of public debates around the EU has been extensively studied in the media, but less so in parliamentary debates (See Chapter 3). Analysis of EU parliamentary debates has focused on level (Auel et al., 2015; De Wilde, 2011) and causes (Auel & Raunio, 2014; Rauh, 2015). Here we consider its consequences for voting behaviour.

Our analysis is innovative for several reasons. First, by focusing on electoral behaviour, we look at what drives electors, rather than making assumptions about them, by looking at parties and their manifestos. Secondly, we measure the position of parties through the tone employed during parliamentary debates, just before the elections, to understand whether debates may influence voting behaviour.

In relation to data availability, each of the four countries considered—Greece, Ireland, Portugal and Spain—held legislative elections in 2019 and 2020. We were able to run online panels in each of these elections and, therefore, can measure the importance of EU issue voting, in a “real” legislative setting that is almost simultaneous across the four countries.¹ 2019/2020 constituted a post-bailout election for all countries involved, including the first for Greece, and thus an opportunity to measure EU issue voting when the relative salience of Europe due to the Eurozone crisis has subsided. In that sense, 2019–2020 is a good test to measure the lasting consequences of the crisis, and thus constitutes a “hard test” for EU issue voting, especially since it focuses on legislative rather than the 2019 EP elections.

The chapter will proceed as follows: first, we present the research which has focused on EU issue voting and how it interacts with the left–right dimension of competition, both conceptually, and in particular, in the countries we are concerned with. Next, we explain how challenger parties are supposed to contribute to the politicization of the EU issue. Our literature review leads us to a number of expectations, followed by a presentation of the data, which have hitherto not been presented in the book. We use country by country as well as pooled regression models to understand how different issues matter for voters in post-Eurozone bailout countries.

European Integration Issue Voting Before and After the Crisis

The process of European integration has been a focus of analysis as a driver of party system change (Marks & Steenbergen, 2004), and a factor for vote influence (De Vries, 2007, 2010). The way in which Europe relates to the left–right dimension of competition was particularly developed by, Hooghe et al. (2002), when they introduced the concept of GAL-TAN, “a new politics dimension […] conceive(d) as ranging from Green/alternative/libertarian (GAL) to traditional/authoritarian/nationalist (TAN)”, and found this dimension is the most general and powerful predictor of party positioning on the issues that arise from European integration, including immigration. GAL parties are in favour of European integration and immigration and TAN parties tendentially are against both issues. TAN parties tend to be radical right parties, and this pole of the GAL-TAN axis was responsible for the politicization of EU integration and anti-immigration, much more than GAL parties which did not initially tend to be so homogenously pro-EU.

Ideological and strategic considerations combine to form a u-shaped curve in terms of support for the EU if we consider parties on the left–right axis (Hooghe et al., 2002). Ideologically, the most left-wing parties, namely communist parties, opposed the EU for its alleged neoliberal policies. At the other end of the ideological spectrum parties of the populist right, the extreme right, or the radical right, also opposed the EU. In contrast to the communist parties, these parties’ opposition to Europe is driven by concerns about national identity, as well as alleged political sovereignty erosion. It has also been shown that the fact that the EU project changed over time, means that attitudes towards the EU also evolved. A longitudinal analysis showed that whereas Euroscepticism was located on the left originally, and thus had a linear relationship with left–right position, from Maastricht onwards it assumes a curvilinear relationship with left–right, as right-wing Euroscepticism grows (Van Elsas & Brug, 2015).

According to Kriesi et al. (2008) globalization altered both the left–right and the EU issues. Whereas economic globalization reinforced the importance of the left–right continuum, the cultural issues arising from globalization revolve around national identity, and reinforce the GAL-TAN continuum. On one side of this continuum, conservatives believe that in order to preserve national identity, it is necessary to reduce immigration, and reverse the process of EU integration, whereas liberals are in favour of a multicultural and European identity (Otjes & Katsanidou, 2017). Thus, as the nature of Euroscepticism changed to become more about cultural issues, it was increasingly associated with attitudes towards immigration too (Boomgaarden et al., 2011). Boomgaarden et al. (2011), go further and find anti-immigration attitudes are strongly correlated with opposition to the EU (for example see Boomgaarden et al., 2011), while Toshkov and Kortenska (2015) show that real-world levels of immigration from Central and Eastern Europe also lead to higher levels of euroscepticism in European countries.

Thus, even before the multiple crises from 2009, European integration as a political issue had been the focus of extensive studies on its nature, its relationship to the left–right axis, and its consequences for party systems. Yet, the Great Recession brought even more attention to these issues. Some authors saw the period as a “critical juncture” since the economic crisis not only led voters to punish incumbents, but to withdraw support from all mainstream parties (Hernández & Kriesi, 2016). Hobolt and Tilley (2016) present challenger parties as alternatives, to which many voters turned in the fallout from the crisis. This connects well to the idea that party system change originates from new parties, as existing parties tend to have difficulty adjusting to changes in voters’ preferences. According to Hobolt and Tilley (2016), the choice of specific challenger parties across Europe was determined by voters’ preferences on three issues that directly flow from the euro crisis: EU integration, austerity and immigration.

Southern Europe, the Left–Right Dimension and EU Issue Voting

Focusing on Southern Europe, and more precisely bailed-out Europe, studies have shown that challenger parties emerged first and foremost on the left (Hooghe & Marks, 2018). Concurrently, Santana and Rama (2018), show that EU issue voting in bailed-out Europe depended mostly on radical left parties. As the authors have shown, using different measures from the EES 2014 data, comparing voters from left radical parties to left mainstream parties in Europe, the former tend to be more Eurosceptic.

Yet, this view is perhaps not completely consensual. Jurado and Navarrete (2021), using EES data from 2014 to 2019, find that following the crisis EU issue voting did not increase in bailed-out countries. On the contrary, according to these authors, it is in countries with more influence in the EU, measured in terms of the number of MEPs or budget contributions to the EU, where this did occur. They posit that this may be due to a downplaying of the role of Europe by governments which underwent bailouts (Turnbull-Dugarte, 2020). In a similar vein, Hooghe and Marks explain that in Southern Europe “largely as a consequence of austerity, the euro crisis reinforced rather than challenged economic left–right conflict centered on distribution and welfare” (2018, p. 125). Also, Otjes and Katsanidou (2017) find strong evidence that “in the Southern European debtor states economic and European issues are merging as a result of strong European interference in their economic policy”.

Thus, on the one hand, some studies suggest that the EU integration issue is mobilized by parties on the extreme of the left–right spectrum, while others suggest that, at least in bailed-out Europe, austerity reinforced the left–right dimension of competition, and indeed EU issue voting does not seem to have increased.

EU Issue Voting in Context: The Tone of Parliamentary Debates

In the final part of this chapter, we examine the way in which the tone of parliamentary debates shapes the importance of EU issue voting in bailed-out Europe. Thus, our chapter goes further than just capturing the relative importance of left–right vis-à-vis EU issue voting, among challenger and mainstream parties using observational data. We focus on a specific context—the Parliaments’ plenary debates—to understand the effects of voting behaviour of parties’ EU politicization. As explained in Chapter 3, the plenary debates are both an important and unique setting to assess the magnitude of EU politicization since they constitute an unmediated discussion of societal issues, and thus can be assumed to capture parties’ “pure” positioning on these issues. Extant literature on the politicization of the EU at the institutional level has focused on the plenary debates (e.g., Auel & Raunio, 2014; Rauh, 2015; Rauh & De Wilde, 2018).

While there seems to be a rather negligible impact of institutional factors on the levels of EU politicization (Auel & Raunio, 2014), actors—and especially the presence of Eurosceptic parties in parliament—seem determinant to explain the levels of EU politicization in parliaments (Wendler, 2016). Further, in Chapter 3, while considering the trends in the politicization of the EU in the bailout countries Kartalis and Silva concluded that “the overall salience of the EU in the parliaments has been relatively stable and low. When it comes to the contestation dimension, while its variations seem more pronounced, the general tone towards the EU has only swung from very positive to less positive”.

Having reviewed the literature, there are a number of findings which have been put forward concerning bailed-out Europe, or Southern Europe, which we will analyse in this chapter. First, it is important to note that since the multiple EU crises occurred, there has also been much more research on consequences for party or party system change than individual-level studies. It is important to revisit the issue of the importance of the left–right dimension for voting in bailed-out Europe, in relation to EU integration issue voting, especially given the post-bailout scenario, where the EU is less salient. The focus on the importance of parliamentary debates in determining the strength of EU issue voting will also contribute to our understanding of the ways in which national institutions contribute to the legitimacy of EU integration.

We saw that the literature which focused on party system change explained how existing parties had difficulty in representing new issues. Instead, it is challenger parties that find it easier to voice these new concerns and indeed do so strategically in order to win more votes. This has also been confirmed by studies which consider EU issue voting. In addition, there is also evidence that this happened, especially during the crisis, by left challenger parties in bailed-out Europe. Thus, following a first broad evaluation of the importance of EU issue voting, we focus on its relative importance for mainstream vs. challenger parties.

Next, we consider the relationship between left–right and EU issue voting in each country. Existing research about party systems, and their alignment on left–right and Europe has found that the parties’ positions on Europe are curvilinear in relation to their left–right position. Namely, parties on the extremes of the left–right spectrum tend to have more Eurosceptic positions, vis-à-vis the centrist parties. We will show the degree to which this occurs in each country considered.

Finally, following the rising importance that national parliaments have gained in the EU we investigate whether the tone of EU parliamentary debates, positive neutral or negative, moderate the importance which EU issue voting has on vote. Indeed, we expect that the party’s EU tone, measured through parliamentary debates data may magnify the impact of congruence between respondents’ and parties’ perceived position on the EU in explaining likelihood to vote for that party.

Data and Methods

We make use of a unique dataset to investigate several of these claims. Fortuitously, Spain, Greece, Portugal and Ireland all held legislative elections in a twelve-month period, between 2019 and 2020. We were able to field panel surveys in all of these elections and thus are well positioned to analyse if the changes which were detected in the aftermath of the Eurozone crisis remain structural to electoral behaviour and political attitudes, ten years after the onset of the crisis. Our choice of countries is relevant since we did not opt to perform an analysis of Southern Europe, per se. A lot of the research which has been undertaken on the effects of the Eurozone crisis on political support, or parties, is aggregated. Analysis of individual data is lacking, and that is what we attempt in this chapter.

What unites our case studies is the fact that all of them faced bailouts in the years following 2009. Thus, beyond austerity, which was a common recipe applied in many countries in the EU, implementing a bailout had political connotations which are not strictly present in austerity policies. Namely, they involve the agreement between the government of the day and external authorities for a program which may be accepted by the government but is externally produced. This meant the salience of the Eurozone crisis was very high in each of these countries.

The survey is a representative two-wave panel online survey with a sample of 800–3000 respondents per country (see Table 6.1 for details) and fulfils a crossed quota of gender (2 categories), age (3 categories) and education (3 categories). While nonprobability online surveys are less established than probabilistic face-to-face surveys and tend to differ in their marginal distributions, they have been shown to yield very reliable results especially when it comes to causal inferences and explanatory models such as vote choice (Dassonneville et al., 2020), which is what we do in this chapter.

Table 6.1 Survey characteristics

For our analysis, we use data from the first survey wave in each country. Our main dependent variable is the probability to vote (PTV). This item asks respondents to indicate the probability that they would vote for a party on a scale from 0 (definitely would not vote for this party) to 10 (definitely would vote for this party). We then created a stacked dataset in which each respondent is turned into several observations, one observation for each party that is included in the PTV item (10 parties in Portugal, 9 parties in Ireland and Greece, and 6 parties in Spain). This stacked dataset is used to predict probabilities to vote in linear regression models, separately for each country and with robust standard errors clustered by the respondent. Our main independent variables are issue proximities. These issue proximities are calculated by using the respondents’ position r on an issue j (measured from 0–10) as well as the party’s position p on the same issue (equally measured from 0–10) in two issue dimensions (left–right and Euroscepticism), as follows:

This measure results in values from −10 (maximum issue distance from party) to 0 (same issue position as party). On the respondents’ side, we use the standard question on left–right self-placement (“In politics people sometimes talk of left and right. Where would you place yourself on a scale from 0 to 10 where 0 means the left and 10 means the right?”), as well as an item on European integration (“Some people believe that the process of European integration should move forward to the creation of the United States of Europe. Others believe that the European Union should be dissolved in order to return to a situation in which states are fully sovereign. At which point of the following scale would you place yourself? 0—The EU should be dissolved; 10—The EU should move towards the United States of Europe”). These items on respondents’ policy positions are also used as dependent and independent variables in the second part of our analysis. On the party side, we use Chapel Hill Expert Survey (CHES), (Bakker et al., 2020) data from 2019, which estimates parties’ positions on a wide range of issues, including left–right (lrgen), and EU integration (eu_position).

As a further independent variable, we use an item to distinguish types of parties according to their left–right position, as well as whether they are challenger or mainstream. Table 6.2 shows the coding of all parties.

Table 6.2 Party groups

In addition, we also use new data from the Maple project on the EU tone of parties in parliamentary debates before the elections. This data is the result of an automated content analysis of parliamentary debates in the year previous to the national elections of 2019/2020 in Greece, Ireland, Portugal and Spain. As described in Chapter 3 of this book, we used automated techniques to identify EU speeches and to measure their valence towards the EU. The tone of the speeches was measured the following way. Firstly, by translating all of the sentences mentioning the EU (using Google Translate API) in each speech. Afterwards, we attributed sentiment scores to each EU sentence using the R package “sentiment”. We use those sentiment scores to measure the levels of EU contestation/negativity in the parliamentary debates. In this analysis, we chose two different variables: the strength of the tone (hence, the absolute difference from 0, without distinguishing between positive and negative), and the direction of the tone (negative vs. positive tones). Both measures are on the party level (See Table 6.3 in the Appendix). To increase the variance in the sample, we use pooled models across all four countries for this part of the analysis, given that otherwise, the number of parties is very low. To ensure that we control for the country-level differences, we centre the tone variables by country mean before using them.

Lastly, all our models control for socio-demographic characteristics (gender, age and education), and the models on vote choice further control for trade union membership, religiosity, and party ID (0 “No party ID”, −1 “Left party ID”, 1 “Right party ID”).

Results

First, we want to understand if the left–right dimension of competition still matters most for citizens’ vote choice, or if European integration has become more relevant. To do that, we regress issue proximity to the two dimensions of competition (left–right, European integration) on probabilities to vote (PTV) and interact this effect with party type. Figure 6.1 shows the marginal effects plots for each country. We can see that left–right proximity remains the most important driver of likelihood to vote in all bailout countries except Ireland, which clearly is a special case when it comes to dimensions of competition. EU proximity matters significantly in determining vote choice as well, but to a lesser degree than left–right. This confirms our assumptions. Moreover, there are no clear differences between challenger and mainstream parties when it comes to the relevance of the Euroscepticism. Contrary to what we initially assumed, left–right proximity is the most important driver of the vote not just for mainstream, but also for new and challenger parties.

Fig. 6.1

Marginal effects of issue proximity on PTV across different party types. Marginal effects of interaction terms in country-wise linear regression, with 95% confidence intervals. Full models in Tables 6.4 and 6.5 in the Appendix

Next, we look at the relationship between the left–right dimension of competition, and EU attitudes, to better understand how they interact in their effect on the vote. Figure 6.2 shows predicted effects from regressing respondents left–right self-placement—direct and in interaction with itself—on their EU position. As expected, we can see that Euroscepticism crosscuts the left–right dimension in all four countries, showing a curvilinear relationship—although in Ireland, the relationship is least strong and inverted.

Fig. 6.2

Relationship between left–right self-placement and EU position. Predicted effects of interaction terms in country-wise linear regression, with 95% confidence intervals. Full models in Table 6.6 in the Appendix

Lastly, we introduce data on parties’ tone in parliament about the EU, to understand how discourses about the EU affect the strength of EU proximity on respondents’ likelihood to vote (see Table 6.3 in the Appendix). We assumed that a party’s EU tone, measured through parliamentary debates data, may magnify the impact of congruence between respondents’ and parties perceived position on the EU in explaining likelihood to vote for that party. Figure 6.3 shows the results of interacting with each parties EU tone in parliament with party-citizen EU proximity to predict PTV. As we can see, there is no significant effect for the strength of the EU tone (neutral vs positive or negative), but a significantly negative effect of the direction of the EU tone (positive vs negative). This means that for parties that show a more negative tone on the EU (hence, have a more Eurosceptic discourse), the EU proximity matters more in determining the voters’ choice than for parties with a positive or neutral EU tone in parliament.

Fig. 6.3

Marginal effects of issue proximity on PTV according to the EU tone of parties in parliamentary debates. Marginal Effects of Interaction Terms in Pooled Linear Regression, with 95% Confidence Intervals. EU Tone is Centred by Country Means. Full Models in Table 6.7 in the Appendix

Conclusions

In this chapter we assessed the relative importance of left–right vis-à-vis attitudes towards the Europe Union for vote choices in bailed-out Europe. Whereas there has been a lot of research devoted to party system changes following Europe’s multiple crises, less is known about the consequences of the crisis for voting behaviour. Our goals were the following: to establish the importance of left–right for the vote in Greece, Ireland, Portugal and Spain, and its relation to the EU issue. Then, we analysed the (relative) importance of these issues contrasting mainstream and challenger parties. Finally, we consider whether EU parliamentary debates, and in particular their tone, moderate the importance of EU issue voting.

This contribution is important for at least two reasons: firstly, due to the fact that in bailed-out European countries key challenger parties have become full or partial incumbents. In January 2015, SYRIZA which was hitherto a challenger party became incumbent in Greece, while the PCP and Bloco de Esquerda, while remaining outside of government, entered a coalition of parliamentary incidence with the Socialist party, which formed a minority government following the 2015 elections in Portugal. We are able to test to which degree in 2019, these parties continue to politicize the EU. Second, we wished to understand EU issue voting in in relation to the tone employed in national debates.

Our results show that in Greece, Portugal and Spain, for each of the categories we tested, left–right proximity has a higher impact on likelihood to vote for a party than proximity on EU issues. Ireland is a partial exception to this trend since neither of these issues seems to matter very much for the challenger left, and the mainstream left is quite different from the mainstream right. Whereas in the former the EU issue is more important than left–right, the situation is reversed for the mainstream right parties.

When we consider the difference between challenger and mainstream parties, we do not see that European issue proximity to parties’ positions are more important determinants of likelihood to vote than left–right for challenger parties. There does not seem to be any relevant difference between the importance of European issue proximities across party types. The reason for this similarity between mainstream and challenger parties, and the resilience of left–right, at least in Greece and Portugal, maybe due to the fact that in these countries, existing challenger parties on the left became either incumbent (Greece) or formally supported the government from parliament (Portugal), thus contributing to downplay the EU proximity issue, vis-à-vis mainstream parties.

Third, we explored the relationship between left–right placement and the EU proximity, by regressing respondents left–right self-placement—direct and in interaction with itself—on EU position. As expected, we find that the EU issue proximity has a curvilinear relationship with left–right. Thus, indeed, EU issue voting does cross-cut left–right in all countries concerned.

Thus, in this chapter, we have been able to show the resilience of the left–right self-placement, both overall, and also for challenger parties across bailed-out Europe. EU issue voting matters for voters across party types, and it does not reinforce left–right placement, in bailed-out Europe. Also, it is interesting to note that Portugal is not an exception in terms of voters while for a long time it has been considered as an exception for parties and party system perspective as an ultra-stable system. Even though these trends are unequivocal and identified, across the chapter it became clear that there are important country differences. For instance, neither of these issues seems particularly relevant in 2019, in Ireland. The next chapters which consider each of these countries as case studies, combining media, parliamentary and voter data will be able to explain some of these distinctive features.

Finally, we showed how politicization, in this case measured in automated fashion and exogenously in parliamentary debates is associated with an increase in EU issue voting, at least for Eurosceptic parties. Contrary to media, which has its own agenda, we can say that in plenary debates, parties have control of the message. Hence, EU politicization which is developed by parties in the plenary debates has an impact on voting behaviour. This is an important signal for the relevance of national institutions, and in particular parliaments, in contributing to legitimizing the EU.

Appendix

See Tables 6.3, 6.4, 6.5, 6.6 and 6.7.

Table 6.3 EU Tone (strength and direction) in parties parliamentary debates, one calendar year before the 2019 (Greece, Portugal, Spain) and 2020 (Irish) legislative elections

Table 6.4 The effects of left-right proximity on PTV across four party groups

Table 6.5 The effects of EU proximity on PTV across four party groups

Table 6.6 The effect of left-right self-placement on EU position

Table 6.7 The effects of EU proximity on PTV according to parties’ EU tone in parliament