Introduction
The literature addressing attitudes on social policy and the welfare state has been telling us for decades that welfare state interventions are supported by individuals who directly benefit from a specific measure (Ferrera, Reference Ferrera1993; Forma, Reference Forma1997; Andress and Heien, Reference Andress and Heien2001). In other words, social risks and support for social policy coincide and, consequently, the welfare state becomes the battleground between vulnerable and less vulnerable groups over the extent and nature of these policies (e.g. Korpi, Reference Korpi1983).
This self-interest-based argument has been corroborated in many empirical studies and for different social policy domains. For instance, individuals from lower social classes favour redistribution, while individuals from higher social classes back market-based solutions (e.g. Hasenfeld and Rafferty, Reference Hasenfeld and Rafferty1989; Svallfors, Reference Svallfors1997). Jobseekers support policies ensuring them a decent standard of living, while employed and affluent workers are sceptical of such schemes, as these potentially increase their taxes (Baslevent and Kirmanoglu, Reference Baslevent and Kirmanoglu2011; Rehm, Reference Rehm2011). Women are more supportive of the welfare state, and particularly of family-related services, because they are likely to rely on such measures at some point in their working lives (Edlund, Reference Edlund1999; Svallfors, Reference Svallfors1997; Baslevent and Kirmanoglu, Reference Baslevent and Kirmanoglu2011). Finally, pension schemes are especially popular with elderly respondents but enjoy very broad support because everybody expects to benefit from them one day or another (e.g. Ferrera, Reference Ferrera1993; Blekesaune and Quadagno, Reference Blekesaune and Quadagno2003; Bonoli and Häusermann, Reference Bonoli and Häusermann2009).
However, during the last three decades, we have been confronted with the diffusion of a type of social policy that challenges the relationship between need/risk and support for the respective intervention. Our knowledge of the micro-level determinants of social policy attitudes is called into question by activation measures or, more precisely, by some types of active labour market policies (ALMPs). Conventional wisdom suggests that ALMPs are supported by individuals at risk of unemployment (i.e. the ‘outsiders’, e.g. Rueda, Reference Rueda2007), as these measures foster labour market access.
While self-interest-based support patterns are undisputed for so-called ‘enabling’Footnote 1 ALMPs (e.g. job-search assistance and training), these do not seem straightforward for so-called ‘demanding’ ALMPs. Eichhorst and Konle-Seidl (Reference Eichhorst and Konle-Seidl2008: 5) define demanding ALMPs as measures that put pressure on the unemployed to accelerate their labour market reintegration by i) tightening individual job search requirements, ii) curtailing the duration and generosity of passive benefits, and iii) introducing monitoring schemes supervising the job search process. Examples of such policies include sanctions for a lack of job search effort and broadening the definition of acceptable work to include occupations that do not match previous skill or revenue levels (Eichhorst and Konle-Seidl, Reference Eichhorst and Konle-Seidl2008: 5; Clasen and Clegg, Reference Clasen, Clegg, Clasen and Clegg2011; Bonoli and Natali, Reference Bonoli and Natali2012; Knotz, Reference Knotz2014).
Since these demanding measures have diffused so successfully and, as shown by Knotz (Reference Knotz2014), even prevail in Organisation of Economic Cooperation and Development (OECD) countries, it is also likely that the beneficiaries of these measures see their utility and support them. In fact, labour market participation is pivotal, particularly in modern societies. It not only ensures economic independence but also heavily impacts each individual's identity, prestige, social embeddedness, and psychological wellbeing (for an overview, see Brand, Reference Brand2015). Thus, unemployed individuals should support measures that help overcome the hardship and stigma attached to this status – even if it comes at a substantial cost. An alternative explanation for the diffusion of these measures is based on ideological preferences. Plausibly, demanding ALMPs are supported by a broad coalition of right-oriented partisans who, independent of their labour market position, attribute unemployment to moral hazard requiring conditionality and sanctions to be countervailed (e.g. Daguerre, Reference Daguerre2007; Bonoli, Reference Bonoli2013).
Inquiring into the micro foundations of demanding ALMPs is a further step to unpacking the multidimensionality of ALMP measures (Bonoli, Reference Bonoli2013; Nelson, Reference Nelson2013; Vlandas, Reference Vlandas2013). Moreover, it helps distinguish individual preferences from party or union strategies, advancing the understanding of ALMP politics (Knotz, Reference Knotz2014; Clasen et al., Reference Clasen, Clegg and Goerne2016: 33). I thus analyse the question: who supports demanding ALMPs, where and why? In more detail, I inquire whether and how labour market risk, ideology and their interaction influence attitudes towards these policies in six Western European countries (Denmark, France, Italy, Germany, Switzerland and the UK) at the height of the economic crisis in 2010.
To complete the picture, and following the insights from the literature on the contextual determinants of social policy preferences, I inquire whether the institutional settings – in this case, ALMP legacies – are correlated with the support for demanding ALMPs (Kumlin, Reference Kumlin2004; Larsen, Reference Larsen2007; Jaeger, Reference Jaeger2009). In fact, the institutional feedback hypothesis predicts that the aggregate public opinion favours measures that correspond to the ideological roots of the existing ALMP scheme the most. The reason behind this logic is that institutional features affect the way the unemployed are framed. In turn, these frames affect voters’ preferences and, by consequence, as suggested by Brooks and Manza (Reference Brooks and Manza2006), government policy-making (Larsen, Reference Larsen2007). I analyze this relationship by looking at Switzerland and Denmark, which share a tradition of human capital-based ALMPs whose aim is primarily to decrease the structural skills mismatch and avoid marginalisation of the unemployed (Nicaise et al., Reference Nicaise, Bollens and Dawes1995; Bonoli, Reference Bonoli2013) and at the UK whose model was instead heavily influenced by the US workfare approach that attributes individual responsibility to the jobless (King, Reference King1995; Lødemel and Trickey, Reference Lødemel and Trickey2001; Daguerre and Taylor-Gooby, Reference Daguerre and Taylor-Gooby2004). Finally, I include France, Italy and Germany that are situated between these extremes and focus on social integration through occupational measures (cf. Barbier and Fargion, Reference Barbier and Fargion2004; Daguerre, Reference Daguerre2007).
Illustrating the relationship between institutional legacies and aggregate support for policies involving demands and conditionality is particularly interesting in times of increasing concern with welfare state freeriding, as such policies may gain relevance in other welfare domains, such as the regulation of immigrants’ welfare access.
The remainder of this article proceeds as follows. First, I discuss the theoretical foundation and develop several hypotheses regarding how labour market risks and ideological preferences affect individual attitudes towards demanding ALMPs and how institutional legacies are related to aggregate levels of support. To test the hypotheses, I run several ordinary least square (OLS) models. Next, I discuss the results, and the final section concludes by discussing the main findings and their implications.
Theory
Origin and characteristics of demanding active labour market policies
In the 1980s, governments in OECD countries started adopting ALMPs because, in an austere economic context, expanding passive benefits was no longer a viable solution for the new social risks – especially the skyrocketing unemployment rates – that were increasingly putting new demands on the welfare state (Bonoli, Reference Bonoli2013). It is well-known that ALMPs include a set of heterogeneous measures ranging from training courses to job search monitoring schemes. The literature proposes different ALMP typologies. However, most of these distinguish between ‘enabling’ policies, i.e. those which are human capital investment-based, and ‘demanding’ policies (King, Reference King1995; Torfing, Reference Torfing1999; Eichhorst and Konle-Seidl, Reference Eichhorst and Konle-Seidl2008; Bonoli, Reference Bonoli2013; Nelson, Reference Nelson2013). This dichotomy originates from the ideological roots of the first ALMP schemes implemented in the 1950s in Sweden (enabling) in contrast with those initiated in the 1980s in liberal countries (demanding). While, in social-democratic countries, training-based ALMPs prevent the marginalisation of individuals with low or obsolete skills and optimise the match between labour demand and supply (Nicaise et al., Reference Nicaise, Bollens and Dawes1995; Lødemel and Trickey, Reference Lødemel and Trickey2001; Bonoli, Reference Bonoli2013), in the US and UK, demanding ALMPs try to accelerate the labour market reintegration of the unemployed, who are held individually responsible for lacking work (King, Reference King1995; Daguerre, Reference Daguerre2007; Bonoli, Reference Bonoli2010: 439). Currently, virtually all countries incorporate some demanding elements. However, substantial differences in the overall conditionality intensity of national ALMP schemes persist (Eichhorst and Konle-Seidl, Reference Eichhorst and Konle-Seidl2008; Bonoli, Reference Bonoli2013; Knotz, Reference Knotz2014).
If we concentrate on demanding ALMPs, their setup seems to imply that unemployment is a consequence of individuals’ behavioural shortcomings rather than the result of structural problems (Daguerre, Reference Daguerre2007). This assessment has a series of consequences. First, if it is assumed that the unemployed prefer to rely on benefits over working, negative incentives and sanctions become essential instruments to accelerate labour market reinsertion (Gilbert, Reference Gilbert2002; Hvinden and Johansson, Reference Hvinden, Johansson, Hvinden and Johansson2007). Second, demanding ALMPs increase the vulnerability of the unemployed vis-à-vis the demands of the labour market, especially regarding concessions in terms of fit and the quality of a new job, e.g. lower wages and longer commuting times (Knotz, Reference Knotz2014). Third, these measures stress self-reliance and personal initiative. Tellingly, activation policies have been compared to a ‘trampoline’ rather than a ‘safety net’ (Giddens, Reference Giddens2000). In short, the prevalence of demanding ALMPs has changed the understanding of social rights from being universally granted to being an entitlement to be ‘earned’ through individual effort and compliance with the system (Bonoli and Natali, Reference Bonoli and Natali2012; Gilbert, Reference Gilbert2002; Handler, Reference Handler2003). The expectation is that workers suffering from high levels of labour market risk are the most affected by the increased pressure entailed in this welfare re-orientation.
The effect of labour market risks on attitudes towards demanding ALMPs
After clarifying the characteristics of demanding ALMPs, let us now analyse the determinants of their support. The literature on welfare state attitudes has unveiled several mechanisms steering individual preferences for (particular) welfare programmesFootnote 2 . The main hypothesis, however, pits the preferences of a homo economicus against those of a homo sociologicus. Rational choice-based motives imply that individuals favour schemes that maximise their self-interest by addressing the needs/risks associated with a disadvantaged position in society (Svallfors, Reference Svallfors1997: 290; Kumlin, Reference Kumlin2004). The sociological literature, instead, stresses the importance of socialisation processes for preference building (Glass et al., Reference Glass, Bengtson and Dunham1986). Thereby, values (e.g. Feldman and Steenbergen, Reference Feldman and Steenbergen2001; Kulin and Meuleman, Reference Kulin and Meuleman2015), religion (Stegmueller et al., Reference Stegmueller, Scheepers, Rossteutscher and de Jong2012), and (political) ideology (Westholm, Reference Westholm1991; Margalit, Reference Margalit2013) gain importance. Thus, I analyse the effect and interaction of self-interest and ideology on preferences for demanding ALMPs.
In the labour market domain, the strongest determinant of self-interest is undoubtedly the current unemployment experience (Blekesaune and Quadagno, Reference Blekesaune and Quadagno2003; Rehm, Reference Rehm2011). Not only does unemployment lead to the loss of economic independence, it also has negative psychological effects, including the loss of social status and self-esteem (Price et al., Reference Price, Choi and Vinokur2002; Wanberg, Reference Wanberg2012). As a consequence, I expect that unemployed individuals should reject measures that put additional pressure on them, decreasing their support for demanding ALMPs compared to employed individuals (Hypothesis 1a).
This negative relationship should also apply to indirect sources of labour market risk. Plausibly, a previous unemployment experience has a comparable (Naumann et al., Reference Naumann, Buss and Bähr2015), though weaker, effect on demanding ALMP attitudes (Emmenegger et al., Reference Emmenegger, Marx and Schraff2015: 12). Moreover, first-hand experience of unemployment among family members and friends may counteract negative stereotypes about the behavioural shortcomings of the unemployed and thus decrease support for demanding ALMPs (Hypothesis 1b).
Finally, as we have known since Stryker (Reference Stryker1980), subjective perceptions may determine real action. Individuals who believe they are at high risk of unemployment should reject demanding ALMPs more decidedly than individuals with secure employment prospects (Hypothesis 1c). Instead, individuals suffering low levels of labour market risk should prioritise the reduction of welfare expenditures and consequently taxes, over suboptimal individual matches in terms of quality, skills and the pay of the new job (Svarer, Reference Svarer2011; Arni et al., Reference Arni, Lalive and van Ours2013) and thus favour demanding policies.
A wide range of situations increase labour market risk, and self-interest-based reasoning might thus be triggered by factors such as part-time work, low educational credentials, and/or low income. Analysing these indirect effects in detail is beyond the scope of this contribution; however, I will include controls for these alternative explanations.
The effect of political ideology on attitudes towards demanding ALMPs
Attitudes towards demanding ALMPs are likely to be influenced by values – particularly political ideology (Arts and Gelissen, Reference Arts and Gelissen2001; Feldman and Steenbergen, Reference Feldman and Steenbergen2001; Margalit, Reference Margalit2013). When applying the classical left-right distinction to ALMP attitudes, the respondents on the political right should support measures involving demands and conditionality more strongly than those on the left, due to attitudes towards both self-reliance vs. state responsibility and the prioritisation of economic performance vs. individual development.
In a context of high and ‘democratising’ unemployment, as was the case at the time of the survey, right-wing partisans in particular may be cross-pressured by self-interest and ideological preferences (Häusermann et al., Reference Häusermann, Kurer and Schwander2014). In the event of unemployment, however, right-orientated individuals should prioritise self-interest because this is a major shock and consequently may prevent value-based reasoning. In line with Margalit's (Reference Margalit2013) suggestion for passive welfare benefits, the effect of unemploymentFootnote 3 on support for demanding ALMPs should be more pronounced for right-wing partisans than for those on the left. Right-wing partisans, who normally favour conditionality, have more room to alter their attitudes once they experience unemployment compared with left-wing partisans, who already reject demanding ALMPs due to their ideological affiliation. In other words, a ceiling effect applies to left-orientated individuals because they reject demanding ALMPs even when employed. Additionally, right-leaning individuals might change ALMP preferences due to a learning effectFootnote 4 resulting from a personal experience with ALMPs. In sum, an unemployment event should affect partisans on the right more than those on the left (Hypothesis 2).
Differences in aggregate public support for demanding ALMPs
At the macro level, I analyse the correlation of ALMP legacies with aggregate public attitudes towards demanding measures (Hasenfeld and Rafferty, Reference Hasenfeld and Rafferty1989; Svallfors, Reference Svallfors1997; Blekesaune and Quadagno, Reference Blekesaune and Quadagno2003; Larsen, Reference Larsen2007). Similar to Larsen's (Reference Larsen2007) argument, ALMP legacies are likely to pre-structure the elites’ framing of the unemployed and thus affect public opinion by determining the ‘starting point’ from which individuals form their attitudes (Larsen, Reference Larsen2007).
Particularly relevant is the prominence of demanding policies in the original ALMP model, which likely influences the extent to which these policies resonate among the public (Bonoli, Reference Bonoli2013: 59ff; Schmidt, Reference Schmidt2002). In the literature, three ALMP models have been identified: the work-first, the human capital and the occupational approaches (King, Reference King1995; Løedmel and Trickey, Reference Lødemel and Trickey2001; Barbier and Ludwig-Mayerhofer, Reference Barbier and Ludwig-Mayerhofer2004; Daguerre, Reference Daguerre2007). These strategies vary conspicuously with respect to the centrality of demanding ALMPs and the negative behavioural evaluations associated with the unemployed.
In the Nordic countries, ALMPs have been based on human capital enhancement because unemployment was problematised as a structural problem (Daguerre, Reference Daguerre2007; Barbier and Ludwig-Mayerhofer, Reference Barbier and Ludwig-Mayerhofer2004). Thus, the focus rested on increasing the workers’ employability through human capital investment; unsurprisingly, in Nordic countries spending on training is especially high (Bonoli, Reference Bonoli2013: 110). Demands and conditionality were thus less central and tied, first and foremost, to the request to actively seek work (Bonoli, Reference Bonoli2013; 71ff.). In liberal welfare regimes, demands have instead always been a key aspect of ALMPs. Initially, in the US and later in the UK, governments adopted strict supervision and sanctions of the unemployed to countervail moral hazard whilst barely investing in human capital training (Daguerre and Taylor-Gooby, Reference Daguerre and Taylor-Gooby2004; Bonoli, Reference Bonoli2013: 110). This strategy was paralleled by a long-standing public discourse associating unemployment with idleness and with fostering a dependence culture (Schmidt, Reference Schmidt2002). Finally, the continental countries are generally classified as falling between these two approaches, applying strategies that focus on social integration relying on occupational programmes that work with moderate levels of human capital investment (Barbier and Fargion, Reference Barbier and Fargion2004; Daguerre, Reference Daguerre2007; Bonoli, Reference Bonoli2013). This focus is clearly recognisable in the prominent role taken by the concepts of social solidarity and exclusion avoidance in these countries’ welfare reform debates (Esping-Anderson, Reference Esping-Andersen1990; Schmidt, Reference Schmidt2002). The importance of demands and conditionality in the original ALMP model should influence the framing of unemployment, and thus affect public support for demanding ALMPs, with the expectation being that support should be highest in the UK and lowest in Denmark, with the other countries situated somewhere in between.
However, in recent decades, demands and conditionality have become increasingly important in most countries (Knotz, Reference Knotz2014; Nelson, Reference Nelson2013). In particular, Denmark experienced a dramatic departure from its low- to high-demanding approach in 1994. The definition of an acceptable job was thereby enlarged dramatically and the benefit length cut substantially (Knotz, Reference Knotz2014). Similarly, the Schröder government in Germany departed fundamentally from the continental trend by implementing the Hartz IV reforms, which converged towards a liberal route (Fleckenstein, Reference Fleckenstein2012). However, Italy, France, and Switzerland did not experience structural ALMP reforms, but increased demands only incrementally (Løedmel and Trickey, Reference Lødemel and Trickey2001; Knotz, Reference Knotz2014). Tellingly, the work-incentive intensity index proposed by Bonoli (Reference Bonoli2013: 34)Footnote 5 shows that Denmark, UK and Germany have the strictest rules followed by Switzerland, France and Italy.
Based on the institutional feedback logic, these reforms should influence the aggregate opinion, leading to increased support for demanding ALMPs in Denmark and Germany (Larsen, Reference Larsen2007). Because of these policy changes, I expect a distinction between high-aggregate support for demanding ALMPs in high-conditionality countries, i.e. the UK and, more recently, Denmark and Germany, and low-aggregate support for demanding ALMPs in low-conditionality countries, i.e. Switzerland, Italy and France (Hypothesis 3).
Data and operationalisation
The dataset
To operationalise attitudes towards demanding ALMPs in detail and with multiple indicators, a 20-minute online survey on public perceptions of unemployment policies was conducted in October 2010 in Denmark, France, Germany, Italy, the UK and SwitzerlandFootnote 6 . Approximately 1,500 valid responses were obtained in each country.
The response rates vary from approximately 4 per cent in the UK to 25 per cent in Switzerland (cf. Schemer and Wirth, Reference Schemer and Wirth2013, Table 1 in the online appendix). As discussed in Sax et al., (Reference Sax, Gilmartin and Bryant2003), low response rates are not a problem in and of themselves but they can become problematic if the sample is biased. This issue is widely discussed in the literature. The under-representation of particular populations can have different causes, for instance, satisficing answersFootnote 7 , different internet access opportunities, data security concerns and the reduced likelihood of respondents completing an online survey compared with them completing a postal survey after several reminders (Berrens et al., Reference Berrens, Bohara, Jenkins-Smith, Silva and Weimer2003: 3–4; Sax et al., Reference Sax, Gilmartin and Bryant2003). To avoid distortions in the analyses, I apply weights (cf. Schemer and Wirth, Reference Schemer and Wirth2013). The weighting variable considers age, gender and education and acts to correct for the under-representation of particular groups in specific countries. The weighting has been truncated at a maximum of eight. Truncation is suggested and applied in major electoral surveys, although the extreme value is subject to debate (De Bell and Krosnick, Reference De Bell and Krosnick2009; cf. Schemer and Wirth, Reference Schemer and Wirth2013).
To contextualise the validity of the present database, I compare the variable that is available both in the present database and in the ESS (Round 4 in 2008) and test whether the two surveys’ summary statistics are equivalent. Specifically, I use the ‘governments should reduce differences in income levels’ item, which is included in the ESS for Denmark, France, Germany, Switzerland and the UK (data for Italy is not available). When comparing the means, medians and standard deviations of this variable, I obtain slightly higher values for the new dataFootnote 8 . However, overall, the comparisons of the two databases suggest that the quality of the data at hand is adequate.
The dependent variables: attitudes towards demanding ALMPs
To operationalise attitudes towards demanding ALMPsFootnote 9 , I rely on an index that is constructed by running a factor analysis on four items of the questionnaire (Table 1). These questions were chosen because, as discussed above in the theory section, they measure precisely the three principal characteristics of demanding ALMPs.
TABLE 1. Operationalisation of demanding ALMPs
First, I operationalise increasing self-reliance, with the statement: ‘Unemployed people should accept more responsibility for themselves’. Second, sanctions are measured with ‘Tougher sanctions [should be taken] against people who refuse to accept reasonable job offers’. Last, individuals’ readiness to make concessions and to take suboptimal jobs is operationalised with the following items: ‘Unemployed people should be willing to accept privations (such as longer commutes or relocation)’ and ‘Unemployed people should make more of an effort to adapt to the needs of the labour market’. These items were all gauged on a scale from 1 to 5 (strongly disagree to strongly agree).
The factor analysis shows that the four items load on one factor and thus form a strong uniform scale with an eigenvalue of 1.77. In the following analyses, I use this continuous factor as the dependent variable.
The independent variables
I operationalise self-interest in terms of three sources of labour market risks. First, I use a dichotomous variable that captures whether the respondent was unemployed and available or actively looking for a job at the time of the survey. Second, a dummy variable captures whether the individual had been unemployed at least once in his or her life. Third, I measure indirect risks with a question regarding whether the respondent had a family member or friend who had been unemployed in the 12 months prior to the interview. Next, I measure political orientation on a 10-point scale (1=left and 11=right).
Attitudes towards ALMPs may be co-determined by numerous socio-structural variables: I control for part-time work (yes/no), gender (male/female), age (in years), income level (five categories), educational level (low/medium/high), retirement (yes/no) and inactivity, i.e. invalidity and sickness (yes/no). Moreover, I capture the respondents’ occupation according to the one-digit International Standard Classification of Occupations (ISCO-08)Footnote 10 , union membership (active/passive or non-member), and nationality (native/foreigner).
The individual-level models include all control variables and country fixed effectsFootnote 11 . As a robustness check, I estimate separate logit and OLS models for support for the individual policies (high/low support), and the results remain stableFootnote 12 .
Empirical results
Labour market risk and support for demanding ALMPs
In the first step, I test whether individuals who experience(d) objective or subjective labour market risks are supportive of demanding ALMPs.
The results show (Model 1, Table 2) that a current unemployment experience has the strongest negative effect on attitudes towards demanding ALMPs (H1a). This finding contradicts the assumption of the new social risk, social investment and some of the insider-outsider literature that vulnerable workers favour ALMPs. Rather, demands and conditionality appear to put off the beneficiaries of these measures.
TABLE 2. The effect of labour market risks on support for demanding ALMPs
°Retired were not asked about the probability of becoming unemployed.
Standard errors are in parentheses, * p < 0.05, ** p < 0.01, *** p < 0.001
Similarly, previous unemployment and indirect unemployment significantly decrease support for demanding ALMPs (H1b). The coefficients for the temporally distant or more indirect risks, as expected, are smaller. Furthermore, Model 2 shows that individuals perceiving a high degree of labour market risk have a conspicuously less favourable attitude towards demanding ALMPs than respondents who are sure to remain employed (H1c). Overall, the expectations from the first set of hypotheses are corroborated.
The models also reveal that some control variables function in line with the self-interest logic. Part-time work decreases support for demanding policies, perhaps because these individuals are more likely to be on fixed-term contracts or to lose their employment in the event of job cuts. Moreover, individuals with higher income levels support demanding policies more than individuals with the lowest income levels. Finally, union members, who might be more sensitised to unemployment issues, favour demanding ALMPs significantly less than ex-union members or non-members.
Political ideology and support for demanding ALMPs
Concerning political ideology, Model 1 (Table 3) shows that the more a respondent adheres to a right-wing political ideology, the more he or she supports demanding ALMPs.
TABLE 3. The effect of unemployment and ideology on support for demanding ALMP
°Retired were not asked about the probability of becoming unemployed.
Standard errors are in parentheses, * p < 0.05, ** p < 0.01, *** p < 0.001
Model 2 shows that the interaction effect of ideology and unemployment – which is the most direct measure of labour market risks – is highly significant. Figure 1 illustrates the relationship postulated in H2 and shows that unemployed individuals are always less supportive of demanding ALMPs compared with employed individuals (even though the effects for the extreme left are non-significant).
Figure 1. Predicted support for demanding ALMPs based on labour market risks and political orientation (all countries)
However, the difference between individuals on the political right is striking. We clearly observe that if a right-leaning individual is unemployed, he or she clearly favours demanding ALMPs less than an employed individual with the same ideological background. To render the magnitude of the effect more precisely, I estimated predictive values for different respondent profiles (see Table 1A in the Appendix). First, it is useful to mention that the dependent variable ranges from -2.5 to 1.3 and has a standard deviation of 0.85. Estimating the predicted level of support for demanding ALMPs for an unemployed person with an extreme left ideology, we obtain a value of -0.61. At the other extreme, we have an employed, right-wing supporter (0.53)Footnote 13 . The most interesting difference, however, is found between the unemployed right-wing supporter (-0.32) and the employed one (0.53), which amounts to one standard deviation of the dependent variableFootnote 14 .
If unemployment ‘offsets’ ideological preferences lastingly (Naumann et al., Reference Naumann, Buss and Bähr2015), the popularity of demanding ALMPs should suffer in the long run, particularly in those countries where youth and those in the more affluent strata (e.g. middle class) are affected by increasing labour market risk.
Country differences in aggregate support for demanding ALMPs
Last, in terms of aggregate support levels for demanding ALMPs, Figure 2 shows that, on average, respondents in Germany and the UK are more supportive and respondents in France and Denmark are less supportive of these ALMPs. Finally, Switzerland and Italy tend to show negative levels of support; however, the result is non-significant.
Figure 2. Predicted support for demanding ALMPs per country
In line with its ALMP model, respondents strongly favour demanding ALMPs in the UK. The same pattern applies to Germany, even though this country only recently departed from the continental route and moved towards the liberal route (Hartz IV reforms) (Fleckenstein, Reference Fleckenstein2012). German support for demanding ALMPs might have remained high (May and Schwanholz, Reference May and Schwanholz2013) because of its excellent labour market performance during the crisis. In fact, unemployment levels remained very low and, consequently, the public might have concluded that the new system contributed to positive economic developments (Eichhorst and Marx, Reference Eichhorst and Marx2011). In contrast, the Danes still seem to favour the original Nordic approach rather than the current conditionality- and demand-based system, which Knotz (Reference Knotz2014) shows is the strictest in Europe. As research on the negative media coverage of the activation programmes in Denmark shows (Kriesi et al., Reference Kriesi, Fossati, Bernhard, Hänggli, Bernhard, Kriesi, Fossati and Hänggli2017), it is possible that the public might oppose policies that are apparently unable to adequately address labour market challenges. Alternatively, the public could react to the misalignment of demanding ALMPs and the human capital framing of the original activation model. Further research is required to determine which of these explanations is more plausible and whether politicians will react to these aggregate preferences as expected by Brooks and Manza (Reference Brooks and Manza2006). Next, French respondents are very sceptical of demanding ALMPs, a position that might stem from France's occupational ALMP legacy and ideological tradition of prioritising self-determination and individual liberty over state intervention (Lødemel and Trickey, Reference Lødemel and Trickey2001). Finally, as expected, Switzerland is situated somewhere between the extremes. In sum, these results seem to lend support to the theoretical expectation that aggregate public opinion correlates with the institutional setting (H3).
Conclusion
Decades of research demonstrates that social policies are supported most strongly by their beneficiaries (e.g. Ferrera, Reference Ferrera1993; Hasenfeld and Rafferty, Reference Hasenfeld and Rafferty1989; Svallfors, Reference Svallfors1997). My study challenges this result, showing that this relationship does not apply to demanding ALMPs, which have successfully and lastingly diffused across OECD countries.
Indeed, the results indicate that unemployed individuals are less supportive of demanding ALMPs than employed individuals, even though these measures help them re-access the labour market. Additionally, I found that ideology moderates the effect of risk exposure. To be precise, employed right-leaning partisans have substantially more favourable attitudes towards demanding ALMPs than their unemployed counterparts.
At the aggregate level, support for demanding ALMPs correlates with ALMP models, as support is highest in countries with a history of liberal-based and lowest in countries with human capital-based ALMP legacies. Interestingly, even though Germany and Denmark departed from their original ALMP model, substantially increasing demands and conditionality, respondents support demanding ALMPs more strongly in Germany (cf. Knotz, Reference Knotz2014). Presumably, the excellent economic performance during the crisis conveyed to German respondents that the Hartz reforms worked well, while the poor development in Denmark possibly called into question these ALMP reforms.
My findings have several implications. First, clearly the diffusion of demanding ALMPs has not been demanded by their ‘beneficiaries’. Rather, even left-dominated governments (e.g. in Germany and Denmark) are willing to introduce demands, probably because they focus on insiders’ interests (Rueda, Reference Rueda2007; Tepe and Vanhuysse, Reference Tepe and Vanhuysse2013) and assume that the unemployed react with political apathy to their situation (Scholzman and Verba, Reference Scholzman and Verba1979). This resonates with the ‘blame avoidance strategy’, whereby governments seeking to retrench the welfare state will concentrate on those groups who are least likely to punish them electorally (Pierson, Reference Pierson1996).
However, since the middle class is increasingly affected by labour market risk (e.g. Häusermann et al., Reference Häusermann, Kurer and Schwander2014), in the future governments could begin to encounter opposition by a broad alliance of left partisans and vulnerable workers that is independent of their ideology or social origin. The middle class has always been pivotal in the welfare coalition and thus might mobilise against policies that increase demands on jobseekers, following Brooks and Manza (Reference Brooks and Manza2006), possibly halting demanding ALMPs on their road to success. Doubtless, more research is needed to clarify the long-term political implications of the changing labour market risk structure, especially considering the scarring effects affecting millions of young unemployed (Naumann et al., Reference Naumann, Buss and Bähr2015).
Second, support for particular policies tends to follow path-dependent patterns paralleling the institutional context, which here is the ideological orientation of the original ALMP models. Transferring the findings to other welfare domains, conditionality-based policies appear to be more easily defended in contexts where welfare state dependence is framed in terms of individual behavioural shortcomings. However, the results also suggest that this framing is not set in stone: successful political reforms and good economic conditions may convince a broad(er) audience of the usefulness of demanding ALMPs, even in contexts of spreading labour market risk.
Finally, since demands can be attached to any type of ALMP or social policy, we should analyse more closely the actual structure of ALMPs. The literature shows that different programmes mobilise different (coalitions of) supporters (Vlandas, Reference Vlandas2013; Bonoli, Reference Bonoli2013; Nelson, Reference Nelson2013). Disentangling the multi-dimensionality of ALMPs is a step in the right direction; however, up-to-date conditionality and demands have been neglected since data are often lacking. If increasing training expenditures at first sight benefit, and are supported by, parties catering to individuals suffering from labour market risk, this should only be the case if these schemes are non-demanding. Concisely, the design of a measure may change support patterns dramatically. Survey experiments could be the right instrument for studying the underlying mechanisms in more detail. Moreover, it would be interesting to approach this research question with longitudinal data to analyse the effect of economic developments on preferences, as well as to analyse the preferences of ALMP participants for both enabling and demanding measures to test whether actual participation changes support patterns compared to ‘mere’ risk exposition.
In the current era of economic instability and immigration, labour market vulnerability will remain high on European countries’ agendas. This is also likely to be true for welfare state reforms, particularly those pushing welfare support away from unconditional social rights. For future research, it will be important to consider the differences in the ‘terms and conditions’ of ALMPs, both to pinpoint the political determinants behind these reforms and to better understand their social consequences.
Acknowledgements
I would like to thank the editors and three anonymous reviewers for their comments. I am also grateful to Giuliano Bonoli, Lucio Baccaro, Fabienne Liechti, Anna Wilson, Philipp Trein, Marcello Natili, Daniel Auer and Delia Pisoni for comments on previous versions of this paper. This research was written with the support of the Swiss National Research Programme on the Challenges to Democracy in the 21st Century (NCCR Democracy 21).
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S0047279417000216
Appendix
TABLE A1. Predicted outcomes for selected profiles
