Introduction
Although a growing body of literature focuses on populist radical right parties’ (PRRPs’) nativist interpretation of the welfare state – so-called welfare chauvinism (Andersen and Bjørklund, Reference Andersen and Bjørklund1990; Mudde, Reference Mudde2007: 130–132) – little is known about whether, and under which conditions, this cornerstone of their contemporary programmes matters to the voters. Several studies have focused on welfare chauvinistic opinions as predictors of PRRP support (e.g. Rydgren, Reference Rydgren2008; De Koster et al., Reference De Koster, Achterberg and van der Waal2013; Goerres et al., Reference Goerres, Spies and Kumlin2018; Im, Reference Im2021). Still, the impact of welfare chauvinism in PRRPs’ election campaigns has received much less attention (Harteveld, Reference Harteveld2016; Schumacher and van Kersbergen, Reference Schumacher and van Kersbergen2016). The present study addresses this gap in the research. It seeks to advance our understanding of the conditions when PRRPs’ supply of nativist welfare politics interacts with voters’ demand and leads to electoral gains for these parties.
Previous research shows that the popularity of xenophobic discourses differs between countries (Schmidt and Spies, Reference Schmidt and Spies2014), and this study argues that the welfare state structure helps explain this variation. This expectation builds on the literature claiming that the principles of redistributive justice embedded in different welfare regimes either depress or enhance welfare chauvinistic opinions (e.g. Larsen, Reference Larsen2006, Reference Larsen2008, Reference Larsen2019; Crepaz and Damron, Reference Crepaz and Damron2009; van der Waal et al., Reference van der Waal, de Koster and van Oorschot2013). Extending this proposition, I argue that welfare regimes should also moderate the electoral appeal of PRRPs’ nativist view of the welfare state. The present study tests this argument by scrutinizing the interplay between PRRPs’ electoral discourses and voters’ demands in the two most prominent welfare regimes in Western Europe – the conservative and social-democratic regimes – between 2002 and 2018.
While ‘welfare chauvinism’ may seem like a uniting ideological principle to PRRPs, these parties have incentives to adapt their positions to ‘public perceptions of normality’ (Larsen, Reference Larsen2019: 49) that differ between welfare regimes (Fenger, Reference Fenger2018: 204). The study argues that an essential difference in this regard concerns promises to expand or reduce social benefits, both of which are inherent features of a welfare chauvinistic discourse (Busemeyer et al., Reference Busemeyer, Rathgeb and Sahm2022: 6). More specifically, the claim advanced here is that electorates in the conservative and social-democratic regimes should vary in their support of nativist pledges that primarily seek to preserve or (partly) dismantle the existing welfare state. Rather than focusing on the rare occasions when PRRPs present specific policy proposals (Ennser-Jedenastik, Reference Ennser-Jedenastik2018; Fenger, Reference Fenger2018; Otjes et al., Reference Otjes, Ivaldi, Jupskås and Mazzoleni2018), I utilize manifesto data (Marpor) (Volkens et al., Reference Volkens, Lehmann, Matthieß, Merz, Regel and Weßels2019) to measure these opposing positions. To not confuse this measure with precise policy proposals, the study refers to it as ‘welfare-nativism’, denoting a nativist electoral discourse ranging between a predominant focus on welfare retrenchment and a primary emphasis on welfare expansion. Merging Marpor-data with nine rounds of the European Social Survey (ESS), the study demonstrates that the two core constituencies of PRRPs – economically exposed and immigration-sceptic voters – display very different reasons for supporting PRRPs’ welfare-nativist positions in the two welfare regimes.
First, economically exposed voters support PRRPs who pledge to expand social benefits in the conservative regime, while this group of voters is more prone to back cutdowns in the social-democratic regime. Second, the results show that voters who are primarily driven by immigration-scepticism respond differently. In the social-democratic regime, these voters support PRRPs who pledge to preserve the welfare state, and increased migration considerably boosts the probability that they do so. By contrast, immigration-sceptic voters in the conservative regime are more likely to support PRRPs who seek to partly dismantle the welfare state, and migration is unrelated to these choices.
These results have important implications as they indicate that welfare regimes moderate voting behaviour and public opinion differently in the contemporary age of populism compared to previous eras.
Welfare regimes and welfare chauvinism
Esping-Andersen (Reference Esping-Andersen1990a; Reference Esping-Andersen1990b) conceptualized welfare regimes as clusters of countries with similar institutions, producing similar public perceptions of solidarity. Building on his classification of the liberal, conservative, and social-democratic regimes, a range of studies have argued that the principles of rights and stratification embedded in these regimes also shape citizens’ degree of welfare chauvinism – that is, the desire to exclude immigrants from the welfare state (e.g. Larsen, Reference Larsen2006, Reference Larsen2008, Reference Larsen2019; van Oorschot, Reference van Oorschot2006; Crepaz and Damron, Reference Crepaz and Damron2009; van der Waal et al., Reference van der Waal, de Koster and van Oorschot2013). A key claim in this literature is that the principle of ‘inclusive solidarity’ embedded in the social-democratic regime inhibits welfare chauvinistic opinions from taking root and enable countries in this regime to ‘absorb immigrants with less groaning and moaning from […] native populations’ (Crepaz and Damron, Reference Crepaz and Damron2009: 456). Thus, van der Waal et al. (Reference van der Waal, de Koster and van Oorschot2013: 176) find two ‘worlds of welfare chauvinism’ in Europe – one where such opinions are widespread (the liberal/conservative regimes) and one where they are likely to remain marginalized (the social-democratic regime).
When studying if welfare regimes also moderate the electoral appeal of PRRPs’ welfare positions, the present study compares the conservative and social-democratic regimes while leaving out the liberal one. The reason is that the liberal regime in Western Europe only consists of Ireland and the UK, both of which lack successful PRRPs except for the brief success of the UK Independence Party in 2015. By contrast, PRRPs have experienced over a decade of electoral success in most countries of the conservative and social-democratic regimes.
Conceptualizing welfare-nativism
Research on PRRPs often conceptualizes welfare chauvinism as a combination of pro-welfare and nativist positions (Kitschelt and McGann, Reference Kitschelt and McGann1995: 22; Schumacher and van Kersbergen, Reference Schumacher and van Kersbergen2016; Harteveld, Reference Harteveld2016; Ennser-Jedenastik, Reference Ennser-Jedenastik2018: 295). Scholars in this literature disagree on the blurriness of PRRPs’ increasingly welfare-friendly positions (e.g. Rovny and Polk, Reference Rovny and Polk2020; Enggist and Pinggera, Reference Enggist and Pinggera2022; Busemeyer et al. Reference Busemeyer, Rathgeb and Sahm2022) but generally agree that these stances are firmly chauvinistic – that is, informed by nativist ideology (e.g. Afonso and Rennwald, Reference Afonso, Rennwald, Manow, Palier and Schwander2018: 177). Moreover, the few comparative studies on the topic show that PRRPs primarily base their welfare chauvinistic pledges on generic nativist statements, while concrete policy positions are rare and, to the extent that they exist, vary substantially between countries (Ennser-Jedenastik, Reference Ennser-Jedenastik2016, Reference Ennser-Jedenastik2018; Fenger Reference Fenger2018; Otjes et al., Reference Otjes, Ivaldi, Jupskås and Mazzoleni2018). The present study, therefore, avoids the term welfare chauvinism, which denotes precise policy positions (Ennser-Jedenastik, Reference Ennser-Jedenastik2018), and highlights the overall trend that today unites most West European PRRPs; an electoral discourse combining strong elements of nativism with more or less elusive positions on welfare (Afonso and Rennwald, Reference Afonso, Rennwald, Manow, Palier and Schwander2018). The study refers to this discourse as welfare-nativism and conceptualizes it as a nativist electoral appeal ranging between predominantly retrenchment- or expansion-oriented positions on the welfare state.
The retrenchment-oriented position builds on the ‘welfare populist’ claim that the welfare state is no longer capable of delivering help to those who deserve it and that it, therefore, should be profoundly restructured (De Koster et al., Reference De Koster, Achterberg and van der Waal2013). While this type of discourse stresses expanded welfare for the ‘common man’ (Ennser-Jedenastik, Reference Ennser-Jedenastik2016: 414), it puts an equal emphasis on rolling back benefits for immigrants, corrupt elites, and even vulnerable groups of natives (Rathgeb, Reference Rathgeb2021).
The expansion-oriented position is profoundly different and implies that PRRPs claim to cherish the welfare state as such (Meret and Siim, Reference Meret and Siim2013; Jungar and Jupskås, Reference Jungar and Jupskås2014; Harteveld, Reference Harteveld2016: 227). This position primarily emphasizes nostalgic pictures of an ethnically homogenous past (Fenger, Reference Fenger2018: 190) and builds on slogans that, for example, underscore the hard choice between welfare and immigration (Betz, Reference Betz2019: 120). While the overall message is xenophobic – immigrants milk the system – the main focus is not excluding any particular group but rather preserving the current welfare state.
While PRRPs, of course, mix retrenchment- and expansion-oriented positions in real-life campaigns, the main argument in this study is that a stronger emphasis on either of these positions should affect their popularity differently in the two welfare regimes (Fenger, Reference Fenger2018: 204). In the conservative regime, voters have divergent views on welfare state generosity (Enggist and Pinggera, Reference Enggist and Pinggera2022: 103), implying that the median voter has no clear stakes in preserving the welfare system in its current form (Reeskens and van Oorschot, Reference Reeskens and van Oorschot2012). Therefore, PRRPs should have incentives to challenge the welfare state structure and propose a mix of nativist appeals, seeking to expand benefits for some groups and reduce them for others (Ennser-Jedenastik, Reference Ennser-Jedenastik2016, Reference Ennser-Jedenastik2018; Rathgeb, Reference Rathgeb2021: 643).
By contrast, the median voter in the social-democratic regime has developed considerable stakes in preserving the existing system of generous welfare provisions, while support for universal benefits is widespread (Korpi and Palme, Reference Korpi and Palme2003: 441). Therefore, PRRPs in this regime should have incentives to avoid proposals that challenge existing social programs and instead present themselves as defenders of the welfare state.
While the aforementioned regime differences should be expected in general, the appeal of PRRPs’ electoral discourse is likely to differ among constituencies within regimes. Previous research identifies two primary explanations for why people vote for PRRPs – economic grievances (e.g. Stoetzer et al., Reference Stoetzer, Giesecke and Klüver2021) and anti-immigration attitudes (e.g. Mudde, Reference Mudde2007; Rovny and Polk, Reference Rovny and Polk2020). Although these are complementary explanations – economically exposed people are often immigration-sceptic – the following sections present hypotheses specifying the conditions when economic exposure and immigration-scepticism lead to different voting patterns in the two welfare regimes.
Support for welfare-nativism among economically exposed voters
A widespread assumption is that economic grievances correlate with welfare chauvinistic views and voting for PRRPs (e.g. Kitschelt and McGann, Reference Kitschelt and McGann1995: 22; Im, Reference Im2021). Scholars often hypothesize that the underlying mechanism is relative deprivation (Rooduijn and Burgoon, Reference Rooduijn and Burgoon2017: 1728), implying that economically exposed voters view immigrants as competitors over scarce welfare resources (Mols and Jetten, Reference Mols and Jetten2016) and, therefore, support welfare chauvinistic policies (Im, Reference Im2021). However, this logic is likely to differ between the two welfare regimes.
Crepaz and Damron (Reference Crepaz and Damron2009: 446) claim that the contribution-based schemes of the conservative regime create ‘a moral logic of exclusion’ that generates economic outsiders with difficulties of fulfilling what Larsen (Reference Larsen2006: 57) refers to as the ‘identity criterion’. Due to these voters’ relatively high degree of economic exposure (van Der Waal et al., Reference van der Waal, de Koster and van Oorschot2013: 166–167), we should expect them to make unfavourable comparisons with ethnic outgroups who find themselves in a similar precarious situation (Larsen, Reference Larsen2006, Reference Larsen2008). Furthermore, given that economically vulnerable voters are more exposed in the conservative regime than in the social-democratic one, they should be more prone to support parties who promise economic compensation only for them (Finseraas, Reference Finseraas2008: 414). Therefore, economically exposed voters in the conservative regime should primarily support PRRPs if their welfare-nativist position is expansion-oriented.
H1a: Economically exposed voters in the conservative regime are more likely to vote for PRRPs, the more expansion-oriented their welfare-nativist position is.
The logic differs in the social-democratic regime. Crepaz and Damron (Reference Crepaz and Damron2009: 446) maintain that the rights-based welfare schemes in this regime generate ‘a moral logic of inclusion’ that dampens the relevance of discussions on immigrants’ deservingness of welfare (Larsen, Reference Larsen2006: 57, Reference Larsen2008: 154, Reference Larsen2019; van Oorschot, Reference van Oorschot2006). The inclusive logic of the universal welfare state, thus, implies that economically exposed citizens are less likely to see immigrants as economic threats.
Still, several studies show that ethnic tensions also generate economic grievances among exposed voters in economically equal settings (Eger, Reference Eger2010; Rooduijn and Burgoon, Reference Rooduijn and Burgoon2017; Burgoon and Rooduijn, Reference Burgoon and Rooduijn2021). However, as these voters have a more privileged insider position in relation to immigrants in the social-democratic regime than in the conservative one (Gschwind, Reference Gschwind2021), the underlying mechanism explaining support for PRRPs’ welfare-nativist discourse is more likely to be relative gratification (Mols and Jetten, Reference Mols and Jetten2016: 278). This mechanism can, for example, manifest itself in the belief that one’s ingroup is more deserving of social benefits than outgroups, coupled with fears that outgroups threaten existing benefits. Therefore, economically exposed voters in the social-democratic regime should be more likely than their counterparts in the conservative regime to support retrenchment-oriented positions.
H1b: Economically exposed voters in the social-democratic regime are more likely to vote for PRRPs, the more retrenchment-oriented their welfare-nativist position is.
Support for welfare-nativism among immigration-sceptic voters
Although voters may support PRRPs due to economic grievances (Kitschelt and McGann, Reference Kitschelt and McGann1995: 22), the most common reason is that they oppose immigration (e.g. Mudde, Reference Mudde2007; Rovny and Polk, Reference Rovny and Polk2020). While immigrant-sceptic voters generally endorse PRRPs’ welfare-nativist discourse due to its xenophobic nature (Rydgren, Reference Rydgren2008), I present two arguments for why their support might differ between the conservative and social-democratic regimes.
The first argument is that the institutions of the two welfare regimes should affect voters’ willingness to take the risk of supporting inexperienced PRRPs, whose welfare-nativist positions may undermine the welfare system (Rooduijn and Burgoon, Reference Rooduijn and Burgoon2017: 1728). In the conservative regime, PRRPs face a welfare system designed to preserve the status of different professions (Esping-Andersen, Reference Esping-Andersen1990b: 108, 112). Due to the somewhat fragmented and unequal provision of social benefits, people are generally less likely than in the social-democratic regime to develop strong attachments to the welfare system (Korpi and Palme, Reference Korpi and Palme2003). When also considering that PRRPs’ constituencies have divided views on welfare state generosity in the conservative regime (Enggist and Pinggera, Reference Enggist and Pinggera2022: 103), immigration-sceptic voters should be more prone than in the social-democratic regime to support radical changes from the status quo. More specifically, these voters should generally be more willing to take the risk of voting for untested PRRPs (Rooduijn and Burgoon, Reference Rooduijn and Burgoon2017) who propose a mix of dismantled benefits (not only for non-natives) and expanded/preserved ones for specific groups of ‘deserving’ natives (Rathgeb, Reference Rathgeb2021).
H2a: Immigrant-sceptic voters in the conservative regime are likely to vote for PRRPs if they propose a mix of expansion- and retrenchment-oriented welfare-nativist positions.
In the social-democratic regime, PRRPs instead face a predominantly universal welfare system that, during its construction phase, played an integrated role in the formation of national identities (Kuisma, Reference Kuisma2007). A reoccurring argument in the literature on the Nordic welfare model is that its unparalleled generosity was possible due to collective feelings of solidarity in homogenous populations that are difficult to extend to non-natives (see Finseraas, Reference Finseraas2008; Eger, Reference Eger2010; Schmidt and Spies, Reference Schmidt and Spies2014: 529). Thus, in contrast to the conservative regime, ‘welfare state nationalist’ sentiments in the social-democratic regime imply a widespread desire to preserve and defend the system (Kuisma, Reference Kuisma2007). It follows from this that immigration-scepticism in the social-democratic regime is closely aligned with, or even driven by, welfare nationalist sentiments (Keskinen, Reference Keskinen2016). Therefore, PRRPs in the social-democratic regime should primarily appeal to immigrant-sceptic voters if they tone down retrenchment and present themselves as ‘nativist’ defenders of the social-democratic legacy (Meret and Siim, Reference Meret and Siim2013: 138; Jungar and Jupskås, Reference Jungar and Jupskås2014: 216, 225).
H2b: Immigrant-sceptic voters in the social-democratic regime vote for PRRPs if they combine nativism with pledges to preserve and expand welfare.
The second argument is that net migration moderates the associations above in different ways in the two welfare regimes. This expectation builds on the scholarship suggesting that universal welfare states are more economically vulnerable to the inflow of immigrants than contribution-based ones (Kitschelt and McGann, Reference Kitschelt and McGann1995: 269; Finseraas, Reference Finseraas2008; Eger, Reference Eger2010; Burgoon and Rooduijn, Reference Burgoon and Rooduijn2021). More specifically, the predominantly contribution-based welfare schemes of the conservative regime imply less redistribution from natives to non-natives than the universal social-democratic regime (Ennser-Jedenastik, Reference Ennser-Jedenastik2018: 297). Against this background, Reeskens and van Oorschot (Reference Reeskens and van Oorschot2012: 124) concluded that individualized insurance systems, as opposed to universal ones, provide no ‘appeal to in-out group conflict’. We can, therefore, expect that:
H3a: Net migration does not affect the probability that immigrant-sceptic citizens in the conservative regime will support PRRPs due to their welfare-nativist positions.
By contrast, the economic stress implied by high net migration in the universal social-democratic regime should raise public awareness of the trade-off between preserving the welfare system and permitting high levels of immigration (Larsen, Reference Larsen2019). In turn, higher levels of immigration should antagonize immigrant-sceptic voters and increase their support for the core welfare chauvinistic argument that immigrants ‘milk the system’ and threaten its survival (Betz, Reference Betz2019: 120). In other words, immigration should enhance the link between immigration-scepticism and the welfare nationalist desire to protect and preserve the welfare system. Therefore, the last hypothesis expects that:
H3b: Net migration increases the probability that immigrant-sceptic citizens in the social-democratic regime will vote for PRRPs who combine nativism with pledges to preserve the welfare state.
Data, methods, and measures
To test the hypotheses, this study models the individual voter’s probability of voting for PRRPs, utilizing the ESS data (https://www.europeansocialsurvey.org) from the conservative and social-democratic regimes between 2002 and 2018 (these data only include people 18-years-old and above who voted in elections). Table A1 (Online Appendix A) shows the sample and the included parties.
Following Esping-Andersen’s (Reference Esping-Andersen1990a: 74–75) categorization, the conservative regime consists of seven countries – Austria, Belgium, France, Germany, Italy, the Netherlands, and Switzerland. The social-democratic regime consists of four Nordic countries – Denmark, Finland, Norway, and Sweden.
The study matches ESS data from these countries with a new measure from the Manifesto Research on Political Representation (Marpor) (presented in the next section). It utilizes multilevel models with random effects for each survey and country – 97.901 respondents, nested in 77 country-years and 11 countries. This design implies that Marpor-measures sometimes are identical for 3–4 consecutive country-years. Still, this procedure is necessary due to data limitations and is common in previous research (e.g. Schmidt and Spies, Reference Schmidt and Spies2014; Rooduijn and Burgoon, Reference Rooduijn and Burgoon2017; Burgoon and Rooduijn, Reference Burgoon and Rooduijn2021; Harteveld et al., Reference Harteveld, Kokkonen, Linde and Dahlberg2021). However, to ensure that country-years do not drive the results, Table F1 (Online Appendix F) tests the same models using election periods (43) as the level−2 unit. While cross-level interactions are slightly underpowered in these analyses, the results are identical to the main models.
The study operationalizes the hypotheses as cross-level interactions. To avoid complicated higher-order interactions, I test them on two subsamples – the conservative regime (Level 1 = 61.912, Level 2 = 46, 7 countries) and the social-democratic regime (Level 1 = 35.989, Level 2 = 31, 4 countries) (Tables D1–D2 in Online Appendix D show the analyses of the total sample). The analyses include power tests because cross-level interaction effects with few contextual units are potentially insufficient regarding statistical power.Footnote 1 Moreover, to avoid overly optimistic inferences for the interaction terms, all models control for random slopes for the lower-level variable. Failure to control for random slopes for the level−1 variable is, as explained by Heisig and Schaeffer (Reference Heisig and Schaeffer2019: 261), ‘equivalent to assuming that the contextual variables perfectly determine the lower-level variable’.
It is essential to note that the combined analyses of aggregate supply-side trends and voting behaviour might be subject to various confounders. Therefore, a range of robustness and sensitivity tests were carried out – for example, different model specifications and level−2 units, the inclusion of country- and year-fixed effects, jackknife analyses, and re-definitions of welfare regimes (see Online Appendices F–H).
Operationalization of welfare-nativism
Utilizing manifesto data, Schumacher and van Kersbergen (Reference Schumacher and van Kersbergen2016: 303) operationalize welfare chauvinism as pro-welfare positions (adjusted for anti-welfare stances) multiplied with negative stances on multiculturalism. In this study, I present a new version of this measure. PRRPs’ welfare-nativist position is a measure that includes pro- and anti-welfare stances as opposite poles on a single nativist scale. It gauges (welfare expansion) per504 × (nativism) per608 minus (welfare limitation) per505 × (nativism) per608. Following the standard approach in the literature, the variable is a logit scale of position (Lowe et al., Reference Lowe, Benoit, Mikhaylov and Laver2011: 131) and varies from −1.29 (strongest position on nativism and retrenchment) to 5.67 (strongest position on nativism and expansion).
Studying the Marpor corpus, Horn et al. (Reference Horn, Kevins, Jensen and van Kersbergen2017: 409–412) show that the item for welfare expansion (per504) measures what it is supposed to measure – that is, pledges to expand or preserve the welfare state’s consumption-oriented pillars – such as pensions, elderly care and health care (Enggist and Pinggera, Reference Enggist and Pinggera2022: 107). The item welfare limitation (per505) – referred to as retrenchment – measures pledges to limit social security and services. I argue that the more parties emphasize retrenchment over expansion, the stronger is the signal that they seek to restructure (and partly dismantle) the welfare state. If parties instead emphasize expansion over retrenchment, they signal that they favour the status quo or improvements of existing programmes (Horn et al., Reference Horn, Kevins, Jensen and van Kersbergen2017: 409–412). Lastly, the item per608 is labelled ‘multiculturalism negative’ but measures positions on cultural homogeneity, assimilation, and the rejection of minority rights that lie at the heart of nativist ideology (Abou-Chadi, Reference Abou-Chadi2014: 425).
The supplementary analyses in Online Appendix C show that PRRPs’ positions on expansion, retrenchment, or nativism alone never drive the scores of the welfare-nativist scale. Hence, this scale always captures combined positions on welfare and nativism. Moreover, Online Appendix E (Tables E4–E6) shows that the impact in the main models depends on the combined positions of nativism and expansion/retrenchment and never stem from PRRPs’ positions on expansion, retrenchment, or nativism alone. Additionally, Appendix E includes separate analyses for expansion- and retrenchment-oriented welfare-nativism (Table E1), a reversed version of the scale (Table E2), and a logit scale measuring the overall saliency of welfare and nativism (per504 + per505 × per608, Table E3) (Lowe et al., Reference Lowe, Benoit, Mikhaylov and Laver2011: 134). These analyses confirm the main findings (see Appendix E for further detail).
In sum, while the measure employed here does not capture PRRPs’ somewhat elusive welfare chauvinistic positions per se, it is the only available longitudinal one that aligns with the aim of this study.
Independent variables and controls
The second contextual variable is net migration, which according to previous research, is the most relevant migration-related measure in studies of exposure to migration and the success of PRRPs (Rooduijn and Burgoon, Reference Rooduijn and Burgoon2017). The variable registers annual data on the crude rate of net migration from the Eurostat and covers all 77 country-years.
Turning to the individual-level variables, economic exposure is a dummy variable tapping into objective economic conditions; it separates between individuals who, during the last week, were unemployed and/or permanently sick or disabled and all others. Given that objective economic deprivation drives feelings of relative deprivation (e.g. Stoetzer et al., Reference Stoetzer, Giesecke and Klüver2021: 4), this variable is suitable for testing H1a and H1b. However, an alternative variable is the four-point ESS-measure of subjective economic well-being (Rooduijn and Burgoon, Reference Rooduijn and Burgoon2017). Yet, as this variable is skewed, I only utilize it in supplementary analyses.
Immigration-scepticism is a composite scale based on ESS items that have been widely used to measure nativism (Harteveld et al., Reference Harteveld, Kokkonen, Linde and Dahlberg2021) and anti-immigration attitudes (Schmidt and Spies, Reference Schmidt and Spies2014: 529–530; Burgoon and Rooduijn, Reference Burgoon and Rooduijn2021). I base it on respondents’ responses to the following two questions: ‘Would you say it is generally bad or good for [country] ‘s economy that people come to live here from other countries?’; ‘Would you say that [country]’s cultural life is generally undermined or enriched by people coming to live here from other countries?’ The items scale well (α = 0.74), and the variable was recoded to run from 0 to 10.
The analyses include socio-demographic control variables that are relevant predictors of voting for PRRPs, such as gender, minority status, age, working-class backgroundFootnote 2 , education (in years) and residence unit (city, town, suburb, village). Moreover, I control for respondents’ left-right placementsFootnote 3 and their attitudes on pro-redistribution.Footnote 4
Lastly, it is essential to note that party competition is inherently dynamic, implying that PRRPs’ electoral appeal depends on the position of other parties in their respective party system. Therefore, I utilize Marpor-data to construct variables controlling for all parties’ mean positions on the left-right and GAL-TAN dimensions. These variables are explained in online Appendix B together with summary statistics (Table B1).
Results
Before formally testing the hypotheses, we review variations in PRRPs’ positions in the two welfare regimes.
Panel A in Figure 1 shows variations in the logit scale measure of PRRPs’ welfare-nativist position – that is, manifestos that stress the combination of nativism and welfare expansion (+) rather than nativism and welfare retrenchment (-). Panels B-C account for the two interactive components of this measure, and panels D-F show the constitutive parts.
Figure 1. Trends in PRRPs’ logit scales of position in welfare regimes.
Source: Manifesto Research of Political Representation (Marpor).
The panels in Figure 1 illustrate the theoretical rationale of including pro- and anti-welfare stances as opposite poles on a nativist scale. While panels A-B confirm existing evidence and show that PRRPs in the social-democratic regime are the most outspoken proponents of both nativism and expansive welfare politics (Meret and Siim, Reference Meret and Siim2013; Harteveld, Reference Harteveld2016: 227; Jungar and Jupskås, Reference Jungar and Jupskås2014), the decreased score around 2010 is more surprising in light of previous research (e.g. Enggist and Pinggera, Reference Enggist and Pinggera2022: 115). Panels C-E shows that this change is due to PRRPs’ increased emphasis on retrenchment and not a decline in their nativist positions. Figures C1 and C2 in the Appendix illustrate that the True Finns and the Sweden Democrats largely drive this trend, reflecting their ambitions of becoming accepted partners in centre-right coalitions (Jungar and Jupskås, Reference Jungar and Jupskås2014: 223).
Panel A also corroborates previous studies by showing that PRRPs in the conservative regime increasingly emphasize welfare expansion and nativism while toning down retrenchment. In Appendix C, we see that this trend is driven primarily by the Dutch Freedom Party (De Koster et al., Reference De Koster, Achterberg and van der Waal2013; Fenger, Reference Fenger2018), the French FN (Rassemblement national) (Otjes et al., Reference Otjes, Ivaldi, Jupskås and Mazzoleni2018: 284), and the German newcomer, Alternative for Germany (Goerres et al., Reference Goerres, Spies and Kumlin2018). By contrast, the Austrian Freedom Party (Afonso, Reference Afonso2015; Rathgeb, Reference Rathgeb2021) and (to a lesser extent) the Flemish Bloc (Belgium) (Harteveld, Reference Harteveld2016: 227; Otjes et al., Reference Otjes, Ivaldi, Jupskås and Mazzoleni2018: 283) have more apparent backgrounds in market liberalism and today put an almost equal emphasis on expansion and retrenchment. Finally, the Swiss SVP (Schweizerische Volkspartei) is the only PRRP in the study that still places its primary focus on nativism and retrenchment (Afonso and Papadopoulos, Reference Afonso and Papadopoulos2015), while the Italian League tones down both nativism and welfare (Ivaldi et al., Reference Ivaldi, Lanzone and Woods2017).
In sum, the welfare-nativist scale reveals significant fluctuations in PRRPs’ manifestos. Thus, there are no clear regime trends, and descriptive analyses alone cannot detect, no less explain, the electoral implications of these variations.
Modelling cross-level interactions
Table 1 tests four empirical models on two subsamples – the conservative (models 1–4) and the social-democratic regimes (models 5–8). It is, therefore, essential to note that conclusions often concern effects between regimes. At the end of this section, I discuss the robustness of the reported results, referring to a range of supplementary studies in the Online Appendices.
Table 1. Estimating the probability of voting for PRRPs in the conservative and social-democratic regimes
+ < p .1. *p < .05. **p< .01. ***p< .001. Entries are regression coefficients with standard errors within parentheses. 1 = Austria, Belgium, France, Germany, Italy, Netherlands, and Switzerland, 2 = Denmark, Finland, Norway, and Sweden.
Note: Estimates of control variables are found in Table D3 in Online Appendix D.
The main effects in models 1 and 4 indicate that expansion-oriented welfare-nativist positions only generate votes for PRRPs in the conservative regime, but these results do not hold up to alternative model specifications (Online Appendix E, discussed further below). Still, we find a robust difference between the two welfare regimes when reviewing the first cross-level interactions (models 2 and 6) utilized to operationalize H1a and H1b.Footnote 5
The first hypothesis (H1a) expects expansion-oriented welfare-nativism to appeal to economically exposed voters in the conservative regime. H2b instead expects this group of voters to respond more favourably to retrenchment-oriented positions in the social-democratic regime. While Table 1 only reports a significantly positive interaction term in the conservative regime (model 2), Figure 2 shows a 5% difference between economically exposed people in the two regimes (cross-level interactions in the total sample in Appendix D confirm this result).
Figure 2. Marginal effects of PRRP welfare-nativist position conditioned by economic exposure (based on models 2 and 6 in Table 1). Whiskers are 95% Cis.
Note. A higher degree of exposure is people unemployed or permanently sick/disabled. A lower degree is all other voters.
Figure 2 shows that if a PRRP moves one unit towards an expansion-oriented welfare-nativist position (on a log-ratio scale between −1.3 and 5.67), economically exposed citizens in the conservative regime are 2% more likely to vote for them than other citizens. This result supports H1a. While the effect goes in the opposite direction in the social-democratic regime, as stated in H2b, we see that differences between groups are only barely significant. Thus, H2b receives relatively weak support but cannot be rejected outright.
Looking closer at the evidence supporting H1a, differences among voters in the conservative regime are substantively important. For example, suppose a PRRP in this regime moves two standard deviations from 0 (an equal emphasis on expansion and retrenchment) to 3.42 on the welfare-nativist scale.Footnote 6 In that case, we observe an increase of 10% in the probability that economically exposed voters will vote for them. Conversely, we see the opposite tendency in the social-democratic regime, albeit less clearly. While these effects differ within regimes, all interaction terms control for country-specific slopes (random slopes). Moreover, the robustness tests discussed at the end of this section provide additional evidence for these results.
In sum, the results support H1a and show that economically exposed people in the conservative regime tend to vote for PRRPs due to an expansion-oriented position. The results also align with H2b, although support is weaker; if economically exposed voters in the social-democratic regime care about PRRPs’ welfare politics, to begin with, they differ from their counterparts in the conservative regime and tend to support retrenchment-oriented positions.
The data studied here do not allow us to observe voters’ degree of relative deprivation or gratification directly. However, while economic exposure is an objective measure of an individual’s degree of economic deprivation, we find the same regime-difference (but slightly smaller) when utilizing the ESS-measure of subjective deprivation (Figure D1 in the Online Appendix). Additionally, we observe the same difference (also less substantial) when replacing measures of economic exposure with working-class background (Figure D2) – that is, a large group of citizens who, on average, are more economically exposed than others. Thus, three different measures of individual economic exposure point to the same conclusion. Given that higher degrees of economic inequality in a country (the conservative regime compared to the social-democratic one) intensify relative deprivation among economically exposed groups of people (e.g. Stoetzer et al., Reference Stoetzer, Giesecke and Klüver2021: 4), the findings strengthen the case that relative deprivation explains electoral responses among economically exposed voters in the conservative regime, while relative gratification is a more plausible mechanism in the social-democratic regime.
Hypotheses 2a-b state that the welfare state structure predicts different responses among immigration-sceptic voters. The coefficients in models 3 and 7 (Table 1) appear to align with both hypotheses.Footnote 7 The insignificant interaction term in model 3, together with supplementary analyses (discussed below), indicates that the effects of retrenchment- and expansion-oriented positions neutralize each other in the conservative regime (H2a). In contrast, the positive and significant interaction term in model 7 suggests that only expansion-oriented positions benefit PRRPs in the social-democratic regime.
However, marginal effects reveal no significant difference between the regimes. I, therefore, refer to the supplementary analyses in Online Appendix E (Table E1) to shed further light on the results. Utilizing separate models for expansion- and retrenchment-oriented welfare-nativism, these analyses show that immigration-sceptic voters in the conservative regime indeed vote for PRRPs due to both positions (Figure E1). Moreover, these analyses show that effects are small and uncertain in the social-democratic regime. Thus, the results in Table 1, together with the supplementary analysis (Appendix E), align with H2a but only provide weak support for H2b. In other words, the findings reported so far indicate that PRRPs’ welfare-nativist discourse primarily generates votes among immigration-sceptic voters in the conservative regime. However, the remaining findings point in a different direction.
H3a expects that net migration does not moderate the popularity of PRRPs’ welfare-nativist positions among immigration-sceptic voters in the conservative regime. By contrast, H3b expects increased net migration to considerably boost the appeal of an expansion-oriented position among these voters in the social-democratic regime. The findings in model 8 strongly support this hypothesis. Although results come from a three-way interaction, the coefficient is well-poweredFootnote 8 , and the findings hold up to a range of robustness tests presented below.
Figure 3 illustrates how the interaction between PRRPs’ welfare-nativist position and voters’ immigration-scepticism differs when net migration varies with one standard deviation (+/−2.81) around the mean (+3.93).
Figure 3. Marginal effects of PRRP welfare-nativist position conditioned by immigration-scepticism and net migration (based on Models 4 and 8 in Table 1). Shaded areas are 95% Cis.
Figure 3 shows that when net migration exceeds the mean, the electoral popularity of an expansion-oriented welfare-nativist position increases substantially among immigrant-sceptic voters in the social-democratic regime. When comparing panels D-F, the estimated probability of a vote for a PRRP increase by approximately 10% among the most immigration-sceptic voters. This effect occurs when the PRRP moves one unit towards a more expansion-oriented position, and net migration increases with one standard deviation from the mean. In contrast, panel C shows a 5% decrease in the conservative regime. While this effect is significantly negative, it does not imply that net migration boosts the popularity of a retrenchment-oriented position (separate effects are insignificant in Table E1 in Online Appendix E). Thus, the result also aligns with H3a.
Previous studies consistently show that opinions are much less ‘welfare chauvinistic’ in the social-democratic regime than in the conservative one (Crepaz and Damron, Reference Crepaz and Damron2009; van der Waal et al., Reference van der Waal, de Koster and van Oorschot2013). However, the cross-level interactions studied above allow us to move beyond mean differences in aggregated country-level attitudes and show rather great potential for PRRPs also in the ‘tolerant’ social-democratic regime (see also Burgoon and Rooduijn, Reference Burgoon and Rooduijn2021). While the hypotheses tested in this study are complementary – economically exposed voters are often immigration-sceptic – the results reveal distinct mechanisms explaining why voters who are primarily driven by economic grievances or anti-immigration attitudes vote differently in varying institutional contexts.
Robustness tests
The results in the previous section broadly hold up to a series of robustness checks (Online Appendices F-H).
In Online Appendix F, the results in Table 1 are robust when level−2 units are election periods instead of country-years (Table F1). We also find the same results as in Table 1 (Table F2) when utilizing the model specifications suggested by Giesselmann and Schmidt-Catran (Reference Giesselmann and Schmidt-Catran2019). These authors argue that analysts should control for country-means of individual-level variables to detect cross-level interaction effects accurately. The results in Table 1 are robust to this manipulation, implying that the cross-level interactions capture differences between countries and changes over time within countries. Lastly, Table F3 presents alternative specifications of the nesting structure (see Burgoon and Rooduijn, Reference Burgoon and Rooduijn2021, online app Table A1). Utilizing three models with a multilevel structure and one without, the results show that the baseline findings are virtually the same when country- and time-fixed effects eliminate cross-country and cross-survey variation.
Online Appendix G includes two jackknife analyses. Table G1 shows that all results are robust when dropping the country-years one at a time. However, Table G2 shows that the three-way interaction in the social-democratic regime is sensitive to dropping one country (all other interaction terms remain significant). Although one could interpret this as a sign of non-robustness, it is not surprising that significant effects disappear when dropping a fourth of the sample. Most importantly, the coefficients in model 6 (Table G2) are the same as in Table 1. Also, the marginal effects (displayed in Figure G1) remain significant when comparing the most immigration-sceptic voters in the two regimes.
Finally, the analyses in Appendix H (Table H1) test if the results are sensitive to the definition of welfare regimes. Researchers have classified Belgium, the Netherlands, and even Germany as mixes between the conservative and social-democratic regimes (Crepaz and Damron Reference Crepaz and Damron2009: 448). However, when including these countries one by one, the results in Table 1 are weakened. Additionally, significant effects disappear or weaken when including the Netherlands and Belgium and all three countries simultaneously.
Concluding remarks
This study explores how the electoral appeal of PRRPs’ nativist interpretation of welfare politics varies between constituencies in the two dominant welfare regimes in Western Europe. Previous research shows that welfare chauvinistic opinions are substantially less prominent in the inclusive social-democratic welfare regime than in the contribution-based conservative one (e.g. van der Waal et al., Reference van der Waal, de Koster and van Oorschot2013). However, when accounting for the interaction between supply- and demand-side factors (Mols and Jetten, Reference Mols and Jetten2020), this study finds that large electorates support PRRPs’ nativist discourse on welfare in both welfare regimes. Still, the logic behind this support varies. I have argued and showed that support depends on whether PRRPs combine their nativist ideology with pledges to defend or partly dismantle the welfare state. The study offers two new insights into why voters’ support for PRRPs’ welfare-nativist positions varies between welfare regimes and their constituencies.
The first insight is that economic grievances predict different voting patterns in the two welfare regimes. The findings demonstrate that economically exposed voters in the conservative regime support positions combining nativism and welfare expansion. However, this group of voters is more likely to support an emphasis on nativism and retrenchment in the social-democratic regime. The study interprets these differences in light of the higher degree of economic inequality in the conservative regime, generating more exposed citizens who experience relative deprivation (Stoetzer et al., Reference Stoetzer, Giesecke and Klüver2021: 4). I have also argued that the opposite tendency in the social-democratic regime most likely reflects these voters’ privileged position vis-à-vis immigrants (Gschwind, Reference Gschwind2021), resulting instead in feelings of relative gratification (Mols and Jetten, Reference Mols and Jetten2016: 278). While the data studied here do not allow us to observe these mechanisms directly, the results are robust and align with theoretical expectations. Most importantly, they highlight new avenues for future research, which should utilize more detailed data and explore how individual and aggregate economic conditions jointly shape different economic motivations for supporting PRRPs (Rooduijn and Burgoon, Reference Rooduijn and Burgoon2017).
The second insight is that anti-immigration attitudes also predict different voting patterns in the two welfare regimes. In the conservative regime, immigrant-sceptic voters seem more prone to support PRRPs proposing radical changes from the status quo (positions that combine retrenchment for some with expansion for others). By contrast, in the social-democratic regime immigration-sceptic voters support PRRPs, the more they pledge to expand and preserve the welfare state. Moreover, the results show that net migration is unrelated to these associations in the conservative regime but boosts them considerably in the social-democratic regime. The study relates this difference to the institutional setup of the welfare state. In short, as migration implies more economic stress in universal systems than in contribution-based ones (Reeskens and van Oorschot, Reference Reeskens and van Oorschot2012: 124), immigration-sceptic citizens in the social-democratic regime have better reasons than in the conservative regime to believe the core welfare chauvinistic argument that immigration threatens the survival of the welfare state (Ennser-Jedenastik, Reference Ennser-Jedenastik2018; Betz, Reference Betz2019: 120).
The patterns revealed in this study, thus, shed new light on the argument that the inclusive social-democratic regime reduces intolerance and welfare chauvinistic opinions (Larsen, Reference Larsen2008, Reference Larsen2019; Crepaz and Damron, Reference Crepaz and Damron2009; van der Waal et al. Reference van der Waal, de Koster and van Oorschot2013). While this claim builds on institutional theories, explaining countries’ aggregate-level support for welfare chauvinism, the present study accounts for attitudinal heterogeneity within countries and variations in net migration. Consequently, it moves beyond previously dominating aggregate-level explanations and demonstrates that PRRPs’ combination of nativist and welfare-friendly positions appeals to vast electorates also in the ‘tolerant’ social-democratic regime. Accordingly, this study takes a first step towards analysing how the legacies of the welfare state shape political behaviour and opinions in the current age of populism.
The study has limitations, notably due to data availability on parties’ positions. While we have seen that Marpor-measures still can provide new and previously overlooked information on variations in PRRPs’ overall electoral discourses, these parties’ exact positions remain ambiguous in this study. For example, we cannot be sure whether PRRPs link nativism and welfare politics in concrete welfare chauvinistic statements or independently take positions on the two issues. While these ambiguities do not weaken this study’s main argument that PRRPs’ positions on welfare and nativism still add up in distinct welfare-nativist signals to the voters, the lack of precise longitudinal data remains a problem in quantitative studies. I, therefore, suggest that future studies in the direction of this one more systematically should map PRRPs’ welfare policy proposals and whether these parties formulate their ambition to exclude immigrants differently if their primary aim is to defend or restructure the welfare state.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S175577392200011X.
Acknowledgements
I would like to thank Brigitte Pircher and Jan Teorell for important help. I also express my gratitude to the anonymous reviewers and the editors of EPSR for very helpful comments, enabling me to improve this paper.
Open access
