1Analysts have long viewed voting behaviour as dominantly structured by class conflict. In today’s post-industrialization era, Western societies are faced with a progressive melting down of the traditional working class and an increasing dualism between labour market insiders and outsiders, two groups with unequal levels of employment security and prospects (Emmenegger, Palier, Seeleib-Kaiser, 2012). Those on the “inside” occupy well-paid stable jobs with good access to social welfare entitlements, while those on the “outside” confront increasing levels of social and economic risks. Our paper addresses the political impact of this structural divide in the European elections of May 22-25, 2014. To sustain this assessment, we compare the impact of various indicators of socio-economic position on vote choice.
- 1 The survey has been by the Centre for European Studies of Sciences Po (CEE). The field work was don (...)
2To test to what extent this new divide predicts vote choice, we use data from a cross national post electoral survey conducted just after the 2014 European elections in seven countries1: France, Austria, Germany, Italy, Spain, Portugal, and Greece. These seven cases display contrasted economic, social and political characteristics, and different varieties and levels of “outsiderness”.
Table 1. Socio-economic indicators by country
|
gnp growth
|
% unemployed
|
% 15-24 years old unemployed
|
poverty risk (2013)
|
post transfer poverty risk
|
gini index
|
austria
|
0,3
|
4.8
|
8.7
|
18.8
|
14.4
|
27.0
|
germany
|
0,4
|
5.2
|
8.1
|
20.3
|
16.1
|
29.7
|
france
|
0,2
|
10.8
|
24.6
|
18.1
|
13.7
|
30.1
|
portugal
|
-1,4
|
15.5
|
37.7
|
27.4
|
18.7
|
34.2
|
italy
|
-1,9
|
12.7
|
35.3
|
28.4
|
19.1
|
32.5
|
spain
|
-1,2
|
26.7
|
53.2
|
27.3
|
20.4
|
33.7
|
greece
|
-3,9
|
27.4
|
55.3
|
35.7
|
23.1
|
34.4
|
eu 28
|
0.1
|
10.9
|
23 0
|
24.5
|
16.7
|
30.5
|
Sources: Eurostat data for 2013 except youth unemployment (2012), population at risk of poverty or social exclusion according to the SILC-EU definition before and after social transfers, Gini index (0-100) for household disposable income.
- 2 As defined by the EU statistics on income and living conditions (SILC) the AROPE indicator (At Risk (...)
3The main economic divide runs between the northern and the southern countries of our sample. The southern countries of our sample were more seriously affected by the 2008 Great Recession than the northern ones (Table 1). In 2013, the year before the election, only Austria, Germany and France display a positive growth rate, all the other countries are in recession, with a record negative rate of almost -4% in Greece. Unemployment follows a similar pattern, below the 28 European Union (EU) average in the three northern cases, and well above in the southern ones. But the rates are more dispersed, from some 5% in Austria and Germany, to the double in France and Italy, the triple in Portugal, and more than five times higher in Spain and Greece. If one looks at the unemployment rates among the young, always overrepresented among the outsiders (Häusermann and Schwander, 2012, p.34 and 2013), the comparative figures are even more impressive, ranging between some 8% in Austria and Germany and over 50% in Spain and Greece. The risk of poverty and social exclusion2, to which outsiders are primarily exposed by definition, is quite higher in the southern group, reaching a peak in Greece where over one third of the population is concerned, compared to less than 20% in Austria, Germany and France. And if the welfare state still acts as a cushion and softens the impact of poverty, even after social transfers, the same cleavage endures. On the whole, the southern societies studied are more unequal than the northern ones, as summed up by the Gini index on disposable household income (Table 1).
Table 2. Political context by country in the 2014 European Elections
|
government
|
turn out 2014 (2009)
(% registered voters)
|
ist party
|
main radical parties
(left/right)
|
main non-classified parties
|
austria
|
L/R coalition
|
45.4 (-0.6)
|
Right (ÖVP)
|
European Anders (2.1) / FPÖ (19.7)
|
-
|
germany
|
L/R coalition
|
48.1 (+4.8)
|
Right
(CDU/CSU)
|
Die Linke (7.4) / NPD (1.0)
|
-
|
france
|
Left (PS)
|
42.4 (+1.8)
|
Far Right (FN)
|
Front de Gauche (6.3) / FN (24.9)
|
-
|
portugal
|
L/R coalition
|
33.7 (-3.1)
|
Left (PS)
|
-
|
-
|
italy
|
L/R coalition
|
57.2 (-7.9)
|
Left (PD)
|
Altra Europa (4.0) / Northern League (6.1)
|
M5S (21.1)
|
spain
|
Right (PP)
|
43.8 (-1.8)
|
Right (PP)
|
Podemos (7.9) / -
|
-
|
greece
|
L/R coalition
|
59.9 (+7.3)
|
Far Left
(SYRIZA)
|
SYRIZA (26.6) / Golden Dawn (9.4)
|
-
|
4The political supply is also quite different from one group of country to another (Table 2). In Austria as in France, the most noticeable trend is the electoral dynamic of far right parties. In France the lists of the National Front drew almost a quarter of the valid votes, coming ahead of the mainstream left by 11 percentage points, of the conservative right by 4 points, and improving its score of 2009 by 18 points. There was no similar take-off at the other end of the political spectrum, where the Eurosceptic left led by Jean-Luc Mélenchon and the Front de Gauche drew hardly 6% of the votes (instead of 11.1% in the 2012 presidential election). In Austria, the Freedom Party (FPÖ) came in third position, but with a score of almost 20% of the valid votes, 7 points higher than in the previous European elections. As for Germany until now the progression of its far right movements such as the Republicans in 1989, was limited and short lived. But in 2014 a new Eurosceptic movement, Alternative for Germany (AfD), started by a Hamburg economics professor, made a surprisingly good score (7%). In the South, on the contrary, left-wing anti-austerity radicals are more dynamic. In Greece the extreme right, led by the neo-Nazi party Golden Dawn, drew more than 9% of the votes, but the dynamic of SYRIZA, the party led by Alexis Tsipras, was even more spectacular, its lists coming ahead of the socialist left and the conservative right, with a record score of 26.6%. In Spain, the conservative right came first, but the surprise came from the good score received by a four-month-old party, Podemos, to the left of the socialist left. Founded by a charismatic political scientist, Pablo Iglesias, in the wake of the indignados mobilizations against austerity and unemployment, it drew some 8% of the votes, sending 5 deputies to the European Parliament. In Portugal no new significant movement emerged despite equally dramatic economic conditions. But the electorate voted for the socialist opposition, whose lists came ahead all others, against the government’s austerity policies. The Italian case stands apart, where the Northern League made a very poor score, and where the rising newcomer is the populist and Eurosceptic Five Star Movement (M5S) founded in 2009 by the comedian Beppe Grillo (21.1%). It had rapidly become the second party of Italy, but it was not the case in the European elections, where it only got some 21% of the votes, beaten by the socialist Democratic Party and its new leader and Prime Minister, Matteo Renzi, who drew a record score of 40.8%.
5The last political feature to take into consideration is the level of mobilization. Participation in the European elections has been gradually declining, from 62% of registered voters in 1979 to 42.5% in 2014. But the situation diverges from one country to another. Turnout reached its minimum in Portugal, where hardly one third of the voters went to the polls, continuing a trend observed since the end of the 1990s. But it rocketed up in Greece to a record 60%, with the hopes aroused by the success of the radical left party, SYRIZA. Elsewhere, the trends were less spectacular. Participation fell sharply in Italy (-7.8%), hardly in Spain, and Austria. On the contrary, it went up in Germany (+4.8%) and a little in France (+1.8%).
6Taken together, the seven countries of the sample offer a good opportunity to test the political effects of class and outsiderness. After defining outsiderness more precisely, we outline our general research strategy, before finally testing our hypotheses.
7A new distinction is emerging between the old structural divide of social class, based on occupation, and a new social divide called outsiderness, based on labour market status. Here the relevant criteria for structuring social divides in a society are not limited to occupational strata based on a hierarchy of manual vs non-manual and skilled vs non-skilled occupations. Outsiderness considers whether a person has a weak vs a secure position in the labour market. Occupational structure—that is, class—and labor market status are thus interconnected in shaping a new social divide around insiders and outsiders.
8Several different conceptualizations of outsiderness have been put forth in the literature. The most straightforward one, and the one that this paper utilizes in its analyses, draws the distinction between insiders and outsiders as a function of labour market status—between those who have stable and secure employment, and those who do not (Rueda, 2005). In this conceptualization, insiders have protected jobs and are insulated from the threat of unemployment, whereas outsiders are either unemployed, or holding precarious jobs characterized by low pay and low levels of protection and benefits.
9A second conceptualization of outsiderness emphasizes the risk of precarious employment (i.e. of being unemployed or atypically employed—including temporary contracts and involuntary part-time work) at the occupational group level, compared to the average in the work force. Individuals are defined as outsiders according to the group they belong to, whatever their actual situation (Häusermann and Schwander, 2011). Research has established that exposure to labour market risks is a powerful predictor of political preference, and these risks can be categorized according to potential vs realized risk of being un- or under-employed (Cusack et al., 2006; Iversen & Soskice, 2001; Mughan, 2007; Rehm, 2009, 2011). While both conceptualizations seek to capture precariousness, the second one remains more class-based than the first.
10A third approach, going back to earlier work on subjective (vs objective) social class (Michelat, Simon, 1977), is to take into account, in a context of recession, the subjective dimension of outsiderness: one’s perceived socio-economic position and even more importantly, its trajectory. Outsiderness should be linked to the feeling of being at the lower levels of society and/or falling down, if not off, the social scale, experiencing downward mobility (Peugny, 2009; Goldthorpe and Jackman, 2007).
11Academic attention to outsiderness and its political impact comes in the wake of long debates about the fading of class voting (Clark, Lipset, 1971; Franklin, Mackie, Valen, 1992; Evans, 1999; Knutsen, 2006; Thomassen, 2005). However, rather than the decline of class voting leading to a structural void, the literature on outsiderness suggests that the dualization between insiders and outsiders forms a new structural divide in society.
- 3 Income not taken into account because of the high level of no answers (up to 19% in Portugal, 18% i (...)
12The concept of outsiderness has first been used in the research fields of political economy and more specifically the literature about the transformation of the welfare system and the way it contributes to a dualization process. More recent research has explored its political and electoral impact. It has been shown that outsiderness fosters non-voting (Bornschier, Kriesi, 2012; Mayer, 2014; Rovny and Rovny, 2015) and voting against the incumbent parties (Rueda, 2005; Lindvall and Rueda, 2012). With a few exceptions, there is still little work clearly articulating outsiderness with the sociology of class change in post-industrial societies. These exceptions include the pioneer work of Thomas Amossé and Olivier Chardon (2006) in France, that assumes the rise of a new “class” of manual and non manual unskilled workers, cutting across the traditional class lines; the work of Daniel Oesch (2006), retracing the evolution of class boundaries with new horizontal stratifications; and the works of Silja Häusermann and Hanna Schwander, which include a structural class dimension in their approach of outsiderness (Schwander, Häusermann, 2012; Häusermann, Kurer, Schwander 2015; Häusermann, Picot and Geering, 2013). In this paper we systematically check the impact of outsiderness, controlling for the other variables of social position, especially education3.
13Study after study (Verba and Nie, 1972; Verba, Schlozman and Brady, 1995; Gallego, 2007) shows that political participation is associated with socio-economic status (education, income, occupation) and the resources that go with it (time, money, and civic skills). Because most of them lack these resources, outsiders should participate in politics less, and display lower levels of turnout, especially in the case of European elections which breed lower interest than national ones, due to their second-order nature (see Sauger in this volume). A previous study based on the first five ESS waves, indeed shows that in the case of national elections in the 16 Western European countries covered, outsiderness does not translate into a political divide. Traditional socio-economic factors, such as age, gender, education and income, structure voters’ decision-making. The main effect of outsiderness is lower turnout (Rovny and Rovny, 2015). In addition, withdrawal from the labour market may entail a retreat into private life and a loss of social contact with co-workers, engendering a decreased sense of community, which may spill over into the realm of political and civic participation and engagement, leading to alienation and apathy (Jahoda, Lazarsfeld, and Zeizel, 1972).
14Political economy research, stemming from Downs’ pioneering work, links income and status loss with support for the left. The lower the median voter’s income falls below the mean income, the higher will be his/her support for redistribution (Meltzer and Richard, 1981). In this perspective, have-nots and outsiders should lean to the left, if not the extreme left, to defend their interests. Several studies have shown that outsiders have stronger preferences for state intervention and welfare guarantees (Häusermann and Schwander, 2011). Others stress the fact that social-democratic parties are more inclined to defend insiders, and that this pushes outsiders away from them and toward more radical left parties (Lindvall and Rueda, 2013).
15Furthermore, the “New left” is seen as representing outsider interests more so than the “Old left,” as traditional left parties seem to protect the core interests of insiders who are more mobilized in trade unions and who have traditionally been the key constituency of the old left parties. The new working class, or “precariat”, is characterized by job insecurity, which is the new demarcation of a division among working-class voters, between old and new left parties. Outsider preferences are thus seen as naturally divided between support for redistribution, on the one hand, and relaxed employment protection (liberalizing the hiring of outsiders) on the other. In this sense, neither the mainstream left (representing redistribution and protection of core worker, i.e. insider, jobs), nor the mainstream right (representing a more flexible labour market, but a reduced role of the state in the economy, i.e. less generous social policies) speaks to the needs and interests of the new outsider class. Therefore, outsiders have an incentive to look beyond the mainstream left and right, and seek representation at the extremes of the political spectrum. The new generation of radical left parties may provide a voice for demands of new social rights, whereas supporting radical right parties may either be a simple protest against a system from which outsiders feel excluded, or a demand for welfare retrenchments against immigrants, which is known as welfare chauvinism (de Koster, Achterberg, van der Waal, 2012).
16A third block of research focuses on the increasing appeal of the far right parties to the working class, in relation to the deterioration of its working and living condition, and the decline of the parties and unions that used to speak in its name. Research by Daniel Oesch (2006, 2008) on the transformation of class boundaries and the proletarianization of the social bases of the radical right; work by Hanspeter Kriesi and his colleagues on the way they appeal to “globalization losers” (Kriesi et al., 2008; Bornschier and Kriesi, 2012); Guy Standing’s (2011) work on the “new dangerous class”, the “precariat”, all go in the same direction. In that perspective, outsiders should be a potential reservoir of the populist radical right parties developing in Europe. Outsiders are disenchanted with the system, due to their declining status, and thus are more likely to vote for “anti-system” parties. Here, however, it becomes necessary to make a distinction between various types of outsiders. Whether one is better educated, therefore having increased earnings potential and having been exposed to different norms than someone who is less educated, despite having precarious employment, is likely to affect vote choice, with the less educated more likely to vote for the populist radical right. In the same line, recent work on the impact of social mobility stresses the appeal of the far right among downwardly mobile workers (Peugny, 2006 and 2009).
17From this literature review, we derive the four following hypotheses:
H1. Objective class position no longer has an impact on vote choice;
H2. Outsiders are less likely to go to the polls than insiders;
H3. Outsiders are less likely to vote for the mainstream parties of the left and of the right; they are more likely to vote for radical parties on either end of the political spectrum;
H4. Subjective social position and trajectory should have the same direction but more pronounced impact as objective indicators.
18To develop further and test the relationship between outsiderness, social position, and vote choice, we rely on a dataset collected after the European elections of 2014. A post-electoral survey has been conducted through the Internet, in the days following the elections, in seven different countries, with national representative samples of 4 000 people in Austria, France, Germany, Italy, and Spain, and 1 000 people in Greece and Portugal (see Sauger, Dehousse, Gougou, 2015).
19The key dependant variable of our analyses is vote choice. It is measured by a variable including five items. It contrasts abstention (and voting for unclassified parties) with voting for the radical left, the mainstream left, the mainstream right and the radical right (for the classification of parties, see Table 1). In most cases, the categorization of parties in these families does not trigger major problems, social democratic parties representing the mainstream left parties, conservative and major Christian Democratic parties the mainstream right. As for Green parties, the choice was to include them in the family of mainstream left parties. Parties without clear left-right position have been added to abstention; this is the case mainly for very minor parties, except for the M5S in Italy (for more details on this party and its voters, see Ivaldi as well as Bedock and Vasilopoulos in this volume). Some choices for the classification between mainstream right and radical right parties have been made as well. In particular, the AfD in Germany has been classified as a mainstream right party as priority has been given to greater homogeneity of the radical right family. In any case, changes in classifications of single parties and especially smaller ones do not change significantly the results as the sample considered cover seven countries.
20Our predictors of interest focus on four indicators. The two first indicators test “objective” social position, i.e. occupation and outsiderness. Occupation has been classified into five categories, along the lines of the new typology proposed by Oesch (2006). These categories are the higher-grade service class (1), the lower-grade service class (2), the small business owners (3), skilled workers (4), and unskilled workers (5). These classes are based on a recoding of ISCO-08 classification, aiming at capturing contemporary class cleavages. The second objective indicator is outsiderness measured along the lines proposed by David Rueda (2005). It includes three categories. The first is composed by insiders: people retired or having a job, with a non-fixed term contract, working full time (or close to), or people owning their business (independents). They represent 60% of our sample on average, but less than 50% in Greece and Spain. The second category is outsiders, including unemployed people looking for a job, and workers with a temporary contract or without a contract, as well as people working part time (less than 18 hours a week). This category represents close to 30% of our sample, up to 44% in Greece. The third category combines people outside the labour market: mainly housewives, unemployed people not looking for a job, and students. The two other indicators of social position are based on subjective assessments. Current social position is directly measured by the answer on an eleven-point scale, from “poor” (0) to “rich” (10), to the question “Taking everything into account, at about what level would you put your family’s standard of living”. The second indicator used measures one’s social trajectory. It is built as the difference, on the same scale, between the respondent’s current situation and her situation five years ago.
21To test our hypotheses, we opt for multinomial logit models predicting the five possible electoral choices previously delineated. We estimate our models on the entire sample of countries available in our data, and evaluate the predicted probabilities of each choice as our predictors of interest change, while all else is held constant (at its mean value). However, since three of the countries in our sample do not have any significant radical right parties (Germany, Portugal and Spain), we run a model omitting these three countries when assessing the predicted probabilities for radical right vote. Our estimates control for country fixed-effects, and correct for the clustering of answers by country.
22To provide robust estimates of our variable of interests, a number of controls have been added to the models we test, in order to avoid omitted variable bias. These controls are the gender of the respondent, her age and the square of her age, her marital status (whether she lives alone or in a couple), the size of the household where she lives, whether she or any of her parents was born outside of the country where she votes, her level of education (recoded into three categories, for primary, secondary and tertiary education), and her religious denomination (Catholic, other Christian, other religions, no religion). We now turn to present our results graphically, while all details are available in the appendix4.
23Our analysis aims to assess the effect of one’s relationship to the labour market on vote choice. We address this relationship in two ways. First, we consider the structural characteristics of individual voters in terms of their social class and their outsiderness. Second, we contrast these objective measurements with subjective self-placements.
24Figures 1 to 4 provide an overview of the results (detailed results are in the appendix). On the whole, the relationship to the labour market seems to have limited impact on political behaviour. As far as our main findings go, we firstly find that subjective placements clearly outperform objective measurements in terms of accurately predicting behaviour. Those respondents who view their situation positively are significantly more likely to vote, and when they do, they are significantly more likely to support major parties. On the contrary, individuals who evaluate their social situation negatively are much more likely to abstain, and when they do vote, they cast their ballots in favour of radical left or right parties. Secondly, supporting radical left parties is surprisingly less conditioned by the relation to the labour market than all other behaviours.
25Considering the objective measurements more closely, we find that outsiderness has two significant effects on political behaviour. First, being an outsider significantly reduces one’s propensity to turn out to vote. Indeed, outsiders are on average 5.1 percentage points less likely to vote than insiders. Second, outsiders are significantly less likely to support major right parties. The propensity to vote for this party family decreases by 4.6 percentage points for outsiders. The effect of outsiderness on the support for major or radical left parties is minor and statistically insignificant. Interestingly, and contrary to expectations in the literature, outsiders are less likely to vote for radical right parties than insiders. In substantive terms, outsiders are 1.4 percentage points less likely to cast a vote for the radical right than insiders. While outsiders are significantly deterred from voting for the major right, there is no impact on major left vote. There is also no effect of outsiderness on radical left vote. Interestingly, the effect of social class on voting is relatively weak compared to the effect of outsiderness. We only find decreased turnout propensity among the unskilled working class.
26Turning now to subjective measurements, the overall picture seems strikingly similar between subjective socio-economic position and subjective trajectory. A better self-perception is associated with higher turnout and more chances to support mainstream parties at the expense of radical parties, as is a subjective assessment of one’s positive trajectory. There is, however, an interesting difference between these indicators. Subjective position is associated with more variance of the chances for voting for the mainstream parties. Subjective trajectory is associated with more variance of the chances to support radical parties.
Figure 1. Vote probability across insiders and outsiders
Figure 2. Vote probability across social class

Figure. 3. Vote probability across subjective social position
Figure 4. Vote probability across subjective social trajectory
27“Old” social class, defined on the basis of occupation, does not seem to significantly structure vote choice in the 2014 European elections, with the exception of a slight effect on turnout, which declines as one moves down the class ladder, towards unskilled workers. Indeed, our findings seem to confirm the views of the now classical literature arguing for the decline of class voting (H1 confirmed). Our findings, however, also point to the possibility that labour market dualization, as expressed in the formation of insider and outsider groups, may start forming a new social divide that will come to structure politics.
28Our analyses suggest that the new structural divide of outsiderness has significant negative effects on some key vote choices. Outsiders are less likely to vote (H2 confirmed), and they are likely to eschew casting a ballot for the major right (H3 confirmed). This negative effect points to two conclusions about outsiders and politics. First, outsiders do not (yet) consider themselves a social group that may be able to perceive their particular social situation, organize, and pursue their interests in the political arena. Indeed, the increased likelihood that outsiders turn away from politics and abstain from voting suggests that they may be particularly limited in their ability to organize. Second, most political parties do not (yet) cater to the interests of outsiders, and remain relatively inept at exploiting the social grievances of these groups. However, given the increased insider-outsider dualism in the context of the recent economic crisis, this social divide may prove a useful mobilizing tool for various political entrepreneurs.
29Finally, our analyses suggest that the most substantively significant predictor of voting behaviour, as expected in Hypothesis 4, in a context of economic crisis, is the subjective assessment of one’s social position and trajectory. This survey was not specifically designed to study outsiderness, but it is clear that its subjective dimension, which until now has hardly been explored, offers a promising line of research (also Mayer, 2014 and Braconnier, Mayer, 2015).