1 Introduction

Since their expansion in the post-war period, European welfare states have received fairly stable support from the population: citizens across different countries support a strong role of government in providing benefits (Bean & Papadakis, 1998; Jakobsen, 2011; Roosma et al., 2016). The transformation ongoing since the 1990s towards a more activating type of social policy is most prominent in the field of unemployment. Activation policies, whose aim is to increase the employability of the jobless and favour their re-insertion in the paid labour market (Daguerre, 2007; Dingeldey, 2007; Seikel & Spannagel, 2018), have contributed to redefine the equilibrium between what citizens receive from the welfare state and what they are asked to do in return (Houtman, 1997). In particular, the demanding approach to activation—that was widely adopted in the European countries since the 1990s–has made the provision of welfare benefits increasingly conditional on certain strict requirements, exerting pressure on the beneficiaries to comply with these rules in order to keep their benefits (Dingeldey, 2007; Eichhorst et al., 2008; Knotz, 2018b). The introduction of these policies has raised questions not only about their efficiency in promoting employment, but also about their legitimacy among the population of European countries (Carriero & Filandri, 2018).

In this contribution, we aim to explore how public support for imposing work obligations on the unemployed is related to support for the welfare rights of this target group. Existing research has investigated this relationship from different perspectives. Some studies have focused on a linear, unidimensional relationship between support for social rights and for obligations of the welfare recipients (e.g., Laenen & Meuleman, 2019; Roosma & Jeene, 2017). These studies report that higher support for social rights of the unemployed tends to go together with less support for imposing obligations; in other words, that support for welfare generosity and support for welfare conditionality are “two sides of the same coin” (Laenen & Meuleman, 2019). Other studies move beyond this unidimensional conceptualization, and stress that citizens’ preferences on welfare rights and obligations are best characterised distinguishing two attitudinal dimensions, for example the right to social security versus the obligation to work (Houtman, 1997) or support for commodifying versus distributive reform (Achterberg et al., 2014). Going further along these lines, Jeene and van Oorschot (2015) cross two dimensions to construct a fourfold typology of preferred combinations for the Dutch population, and find that a relevant share of citizens combines high levels of support for unemployment benefits with a preference for welfare conditionality.

Following this last line of research, we investigate how citizens across Europe combine attitudes towards these two dimensions of the contemporary activating welfare states. We build a typology of individuals’ preferred balance between social rights and obligations by means of latent class analysis (LCA). This person-centred approach is aimed at distinguishing latent configurations of patterns of attitudes in the population studied (Meeusen et al., 2018; for application in the field of welfare attitudes, see Roosma et al., 2014). After discovering which attitudinal configurations are present in the data, we continue by exploring how these configurations are related to individuals’ socio-economic characteristics and ideological dispositions (Jaeger, 2006). The comparative design furthermore allows us to analyse whether the popularity of attitudinal configurations is linked to indicators of welfare generosity and conditionality at the country level. We thus aim to answer the following research questions: (1) How do Europeans combine support for social rights and work obligations of the unemployed? (2) What is the individual socio-economic and ideological profile of each of the configurations? (3) How are these configurations of preferred rights-obligations balance related to national levels of welfare generosity and welfare conditionality? For this purpose, we analyse data from the welfare attitudes module included in the European Social Survey 2016, round 8.

A main contribution of this approach is that this analysis links insights from the welfare attitudes literature, which is strongly focused on support for social rights, with emerging research on the legitimacy of benefit conditionality, a principle that has gained importance in the social and labour market policies of the European countries (Knotz, 2018a). Our analysis sheds light on how the public reacts to the introduction of these measures, and whether support for stricter obligations erodes the “too rosy picture of welfare state legitimacy” (Ervasti, 1998, p. 288), or can co-exist with support for social rights. Combining these two aspects of welfare legitimacy allows us to depict a more nuanced picture of citizens’ preferences regarding unemployment policies. In addition, contrary to previous studies on combinations of support for rights and obligations, we adopt a cross-national perspective. This allows us to investigate whether individual preferences are linked to country-level differences in institutional setting.

The remainder of the article proceeds as follows. First, we present the review of the main literature on the policy context, combinations of support for welfare rights and obligations, and their predictors at individual and country level. Second, we describe data and methods used. Third, we present the results of LCA and multilevel logistic regression models conducted on the European countries in the sample. The last section summarises the conclusions drawn from our findings, including a discussion on the implications of the study and avenues for future research.

2 Theoretical Background

2.1 Activating Welfare States: Social Rights Meet Work Obligations

Welfare states provide protection from market forces via social security systems and offer services in the areas of education, housing, health and poor relief (Esping-Andersen, 1990; Jakobsen, 2011). Redistribution of resources is the way through which the welfare state reaches the goal of social protection and of lessening unjust inequalities (Roosma et al., 2013). The economic crisis that started in the early 1970s, however, opened a phase of welfare retrenchment. As a response to mass unemployment, activating labour market policies were developed and implemented, with the aim of avoiding human capital deterioration and keeping the jobless busy (Bonoli, 2010, p. 446). From the early and mid-1990s, welfare policies have been reoriented towards activation of the welfare beneficiaries (Bonoli, 2013). Beneficiaries can still rely on the welfare state when they do not have a sufficient income, but they are asked to be actively engaged in the search of a new job, with the ultimate goal of re-integration in the paid labour market and in the society (Betzelt & Bothfeld, 2011; Eichhorst et al., 2008; Seikel & Spannagel, 2018). This goal might be pursued through different paths. On the one hand, enabling activation refers to policies aimed at promoting human capital development through incentives and services that are offered to the jobless. On the other hand, activation might take a more coercive or demanding nature, when it mainly imposes work and repressive sanctions in case the work obligations are unmet (Dingeldey, 2007; Eichhorst et al., 2008; Knotz, 2018b; Seikel & Spannagel, 2018).

Although the provision of welfare rights has always implied some form of conditionality –whether in the form of means-tested benefits, or contribution to the system to be eligible for benefits– the new reforms in the field of labour market policies have exacerbated benefit conditionality. In most of the European countries, stricter sanction rules for unemployment benefits have been introduced right after an increase in the budget deficit, and after a long-term increase in unemployment levels (Knotz, 2018b). Demanding activation policies are based on the idea that social rights need to be conditional upon the compliance with certain requirements, in a framework of rebalancing rights and responsibilities (Giddens, 1998; Houtman et al., 2008). Welfare beneficiaries are increasingly asked to meet obligations in order to keep receiving their benefits and to avoid incurring repressive sanctions (Eichhorst et al., 2008).

2.2 The Preferred Balance of Rights and Obligations Among the Europeans

The policy trends described above raise the question to what extent the shift in balance between provision of benefits and benefit conditions is seen as legitimate among the population at large. Despite the phase of welfare retrenchment that started after the economic crisis of the 1970s, citizens of advanced welfare states have remained highly supportive of a strong role of government in providing services and benefits for those in need (Bean & Papadakis, 1998; Jakobsen, 2011; Roosma et al., 2016). The active welfare state, with its attempt to reach a new equilibrium between rights to social security and correspondent duties (Groot & van der Veen, 2000), puts the concept of benefit conditionality to the fore in the societal debate. Recent studies have shown that support for benefit conditionality has increased in Europe in the last decade (for example, Carriero & Filandri, 2018). The question remains, however, how support for conditionality relates to support for social rights. Investigating the relation between preferences regarding rights and obligations offers opportunities to adjust the “too rosy picture” of welfare state support that plagues welfare attitudes research (Ervasti, 1998, p. 288). To date, this relation has been mainly investigated by assuming that the rights-obligations nexus can be described by a single linear pattern. This research has shown that there is a moderate negative relation between individuals’ support for welfare rights and for obligations of the unemployed: the higher one supports social rights, the weaker one prefers imposing obligations (Laenen & Meuleman, 2019; Roosma & Jeene, 2017).

Taking this linear approach might, however, veil more nuanced opinions on what benefit recipients should get and under which conditions. After all, people can combine preferences for social rights of the unemployed with support for work obligations in ways that go beyond the linear pattern. Two studies conducted in the Netherlands provide useful insights in this respect. First, a study by Houtman (1997) indicates that the majority of the Dutch population prefers to keep both the social security right for unemployed people and their obligation to work, whereas a rejection of both policy principles is highly unusual. Jeene and van Oorschot (2015) elaborate this idea further and construct a theoretical typology of the preferred balance of social rights and work obligations among the Dutch population. On the one hand, they detect two configurations that reflect the negative relationship between support for rights and for obligations. One attitudinal profile, labelled unconditional generosity, shows strong support for social rights and opposition against obligations. The work first profile combines low support for rights and high for obligations. However, two other profiles fall outside the logic of a negative linear pattern, combining high support for both rights and obligations (conditional generosity) or low support for both (laissez-faire). These last two combinations are highly relevant in societal debates on activation and reflect specific ideological profiles. First, the combination of strong support for rights as well as obligations resembles the logic of welfare populism, that is, a combination of support for redistribution and a strong criticism towards the welfare state (Abts et al., 2021; de Koster et al., 2013). A populist conception of redistribution, striving for equality, goes hand in hand with the idea that the welfare state is complex, non-transparent and it grants benefit to undeserving recipients (Derks, 2006). The imposition of strict obligations on the unemployed typical of the demanding activation is strongly rooted in the idea of punishing people that are dependent on the welfare state. Second, the laissez-faire option can be linked to a neoliberal view of the welfare state, in which the market has primacy over government redistribution and the governmental provision of welfare should be kept at the minimum (McCluskey, 2003; Spicker, 2013). People opting for this combination believe that the state should not intervene too much in the lives of the unemployed, either by providing generous benefits or by obliging them to do something in return for their benefit.

As a working hypothesis for our latent class analysis, we use this theoretical typology as a point of departure: we expect to observe two groups that follow the linear relation of support (high support for rights combined with low support for obligations; and low support for rights combined with high support for obligations), and we expect to find at least two more combinations that deviate from this linear relationship. Figure 1 illustrates these hypothesised support profiles.

Fig. 1
figure 1

Four combinations of preferred support for welfare rights and work obligations of the unemployed (adapted from Jeene & van Oorschot, 2015)

Full size image

2.3 Determinants of Preferred Balance

After identifying people’s combinations of preferences for rights and obligations, we investigate how these combinations can be contingent on individual characteristics as well as the institutional context at the national level.

2.3.1 Individual-level Predictors: Interests and Values

At the individual level, the preferred balance of rights and obligations could be affected by both socio-economic characteristics, as well as ideological beliefs (Hasenfeld & Rafferty, 1989; Jaeger, 2006). We start by listing expectations regarding the two sides (rights versus obligations) separately.

First, support for generous welfare arrangements is expected to be higher among welfare beneficiaries, and among those who feel at risk of unemployment, following self-interest explanations (Jaeger, 2006). Individuals demand more redistribution not only because they are in need, but also because of their higher labour market risks (Rehm, 2009). From an ideological perspective, stronger egalitarian and politically left-wing values are associated with stronger support for government intervention (Feldman & Zaller, 1992; Jaeger, 2006). Conversely, support for conditionality is expected to be higher among those with higher income, because of their stronger labour market position which makes them less subject to welfare dependency (Rossetti et al., 2021), and among those with stronger authoritarian and right-wing values (Achterberg et al., 2014; Fossati, 2018; Rossetti et al., 2021).

When looking at the two aspects together, we expect the preferred combinations of support for rights and obligations to be chosen by different profiles of individuals based on their socio-economic and ideological characteristics. First, we expect that the combination of strong support for rights and opposition against obligations (unconditional generosity) will be preferred in the first place by individuals who benefit from generous welfare rights, and see strict obligations as an obstacle to benefit access, namely those who are (currently or previously) unemployed and with stronger egalitarian values.

Second, we expect the profile combining strong support for welfare rights and for work obligations (conditional generosity) to be endorsed by individuals with strong authoritarian values, who give importance to preserving social order, obedience to authority, and compliance with the established rules (Feldman, 2003). For its resemblance with a welfare populist critique, which combines intolerance of non-conformism with economic egalitarianism, we expect this profile to be more prevalent among the lower socio-economic strata (Derks, 2006; Houtman et al., 2008; Van Hootegem et al., 2021). Third, we expect the combination of low support for both rights and obligations (laissez-faire) to be preferred by individuals opposing economic egalitarianism, and by those in a more secure position (high income and high education) as they are found to be less supportive of egalitarianism (Houtman et al., 2008). Finally, the combination of strong support for work obligations and weaker support for welfare rights (work first supporters) is expected to be more popular among right-wing individuals, and among higher income individuals.

2.3.2 Country-level Predictors: Institutional Context

The idea that the preferred balance of rights and obligations might differ across countries resonates with various theories stating that institutional contexts are systematically related to people’s attitudes towards the welfare state (e.g., Svallfors, 1997). Policy feedback theory (Pierson, 1993) postulates that there is a strong, positive link between public policies and citizens’ preferences. According to this argument, the policies that are implemented can be seen as a normative framework providing cues of what is ‘normal’ that shape public preferences. However, a differentiation should be made between established and relatively new policy domains (Raven et al., 2011). The increased welfare conditionality in social and labour market policy has occurred only in the last decades, thus conditions might not have been created yet for policy feedback mechanisms to develop. Another theoretical perspective that links policy design to attitudes is the thermostatic model. This model assumes that the public evaluates the current policies and adjusts its preferences in reaction to the current situation, acting as a thermostat for policymakers’ choices (Soroka & Wlezien, 2010). According to this model, public opinion does not automatically grow more supportive of the implemented policies. On the contrary, in settings where there is stronger policy effort in a particular direction, the public thermostat will probably become less supportive. To analyse how policy preferences relate to the institutional characteristics of the country, we focus on two highly relevant aspects of unemployment benefit systems, namely generosity and conditionality. Notwithstanding their relevance, these two aspects of ‘welfare stateness’ are notoriously difficult to measure (Kunißen, 2019). Welfare generosity is often operationalised by expenditure-based measures (most commonly social expenditure), which are far from being a ‘perfect indicator’ (Kunißen, 2019). Net replacement rates are suggested to be a better indicator, because individuals are more likely to have a basic knowledge of the generosity of unemployment benefits rather than a broader measure of generosity such as social expenditure (Kunißen, 2019). Welfare conditionality, on the other hand, comprises several aspects of strictness of unemployment benefits (Knotz, 2018a). A measure of overall conditionality of unemployment benefits has been developed by the OECD, which allows to assess the strictness of countries’ activation policies by taking into account different requirements for receiving unemployment benefits, such as how much freedom the unemployed person has to choose among available jobs, job-search-related efforts and sanctions in case of noncompliance (OECD, 2020).

It has to be noted that the theories on the link between institutional context and attitudes imply longitudinal processes that cannot be fully uncovered using cross-sectional data. Instead of attempting a stringent test of these theoretical arguments, we use them as background against which we interpret certain patterns. The two frameworks are useful in that respect, because they lead to opposing expectations. When a positive relationship between policy design and public preferences is found, this could be interpreted as a trace of policy feedback mechanisms. Thus, policy feedback mechanisms predict that in countries with more generous replacement rates, citizens are more supportive of generous benefits, while stricter conditionality can trigger support for these measures among the public (Fossati, 2018; Rossetti et al., 2022), thus leading more people to support conditional generosity or work first combinations. When the opposite pattern is found instead, this could be indicative of thermostatic reactions.

3 Data and Methods

3.1 Data

To investigate the preferred balance between social rights and obligations of the unemployed across the European population, we use data from the round 8 of European Social Survey (ESS Round 8 Data, 2016). Face-to-face interviews were conducted between August 2016 and December 2017 in each country on the basis of probability-based samples of adult population aged 15 or older. National response rates range from 30.6% (Germany) to 69.6% (Poland). The total sample for the LCA –which uses a Full Information Maximum Likelihood as a strategy to deal with item non-response (Enders & Bandalos, 2009)– includes 38,942 individuals in 21 countriesFootnote 1: Austria (AT), Belgium (BE), Czech Republic (CZ), Estonia (EE), Finland (FI), France (FR), Germany (DE), Hungary (HU), Iceland (IS), Ireland (IE), Italy (IT), Lithuania (LT), the Netherlands (NL), Norway (NO), Poland (PL), Portugal (PT), Slovenia (SI), Spain (ES), Sweden (SE), Switzerland (CH) and the United Kingdom (UK).

3.2 Variables

To measure support for welfare rights of the unemployed, we use the question on the range dimension of the welfare state (Roosma et al., 2014). Respondents were asked to indicate, on a scale from 0 (“not be governments’ responsibility at all”) to 10 (“entirely governments’ responsibility”), how much responsibility they think governments should have to ensure a reasonable standard of living for the unemployed. To measure support for work obligations of the unemployed, we use respondents’ answers to three items: “Imagine [someone] who is unemployed and looking for work. This person was previously working but lost their job and is now receiving unemployment benefit. What do you think should happen to this person’s unemployment benefit if this person (1) turns down a job because it pays a lot less than they earned previously? (2) Turns down a job because it needs a much lower level of education than the person has? (3) Refuses to regularly carry out unpaid work in the area where they live in return for unemployment benefit?”. Four possible answers were presented (here reversed): this person should (1) be able to keep all the unemployment benefit; (2) lose a small part; (3) lose about half; or (4) lose all the unemployment benefit. In the introductory part of the question, the condition ‘someone’ was randomly replaced by three other descriptions of the unemployed person: “someone in their 50s”, “someone aged 20–25”, “a single parent with a 3-year-old child”.

To investigate the effect of individual characteristics on the likelihood to choose one combination over the others, we include indicators for the socio-economic and ideological profile of individuals. We measure occupation by a variable combining information on the respondent’s main activity for the last seven days and the title of their main job for those in paid job (recoded into the EGP occupational class scheme; see Ganzeboom et al., 1992). Eight categories are obtained: service class, white-collar workers, blue-collar workers, self-employed, students, retired, unemployed and other. We calculate income (in quartiles) on the basis of the distribution of equivalised total net household income, using the OECD-modified scale (Hagenaars et al., 1994). Education is indicated by the highest level of education completed (lower secondary, upper secondary, tertiary). We use a dummy variable to measure previous experience of unemployment (“Have you ever been unemployed and seeking work for a period of more than three months?”).

Since the dataset does not include a measurement for authoritarian values, we use conformity-tradition values from the theory of basic human values (Schwartz, 1992), which are conceptually related to authoritarianism (Rohan & Zanna, 1996). These are measured as mean of the answers (ranging from “not like me at all” to “very much like me”) to the following items: (1) “He believes that people should do what they’re told. He thinks people should follow rules at all times, even when no-one is watching”; (2) “It is important to him always to behave properly. He wants to avoid doing anything people would say is wrong”; (3) “It is important to him to be humble and modest. He tries not to draw attention to himself”; (4) “Tradition is important to him. He tries to follow the customs handed down by his religion or his family”. Egalitarianism is included as mean of two items: (1) “For a society to be fair, differences in people’s standard of living should be small”; (2) “Large differences in people’s incomes are acceptable to properly reward differences in talents and efforts” (responses were recoded to indicate stronger egalitarian values with higher values). Political orientation is measured by four categories (left, centre, right, missing) based on the 11-point self-placement scale (0- left to 10- right). We control for age and gender.

At the country level, we include two macro-level indicators for the institutional context. To measure the generosity of unemployment benefit system we use the net replacement rate, as the proportion of income maintained after 60 months of unemployment (which accounts for the long-term unemployment). This indicator is retrieved from the OECD and calculated for a single person without children, with average earnings and excluding housing benefits and social assistance. As a measure of welfare conditionality, we use the OECD indicator overall score for strictness of activation requirements (OECD, 2023). Descriptive statistics of these indicators are included in Appendix Table 5.

3.3 Analytical Strategy

To build a typology of preferred balance between support for rights and obligations of the unemployed, we employ latent class analysis (LCA) (Vermunt, 2010). LCA makes use of the responses on multiple manifest indicators (that is, the survey items) to empirically construct a categorical latent variable classifying respondents into a typology. This is a person-centred approach, which allows to distinguish subgroups of individuals with particular configurations of attitudes, contrary to a variable-centred approach, which estimates a set of parameters averaged across all the respondents of the sample and thus precluding the occurrence of subpopulations with different attitudinal combinations (Meeusen et al., 2018; Morin et al., 2016). After building the LCA model for the set of selected items and identifying the number of latent groups that best represent attitudinal configurations, respondents are assigned to the latent groups on the basis of their response pattern. Subsequently, we perform a series of multilevel logistic regression models, using the membership of one of the latent classes –versus the other three latent classes– as dependent variable. In this way, we can account for the specific profile of individuals in one group. First, we test the effects of individual-level characteristics, followed by a one-by-one introduction of the country-level predictors in the model.

4 Results

4.1 Descriptive Findings

Comparing the means of our variables of interest across the 21 countries (see Appendix Table 3), we observe that the country averages of support for governments’ responsibility in providing for the standard of living of the unemployed are always above the midpoint (5). This is in line with previous findings that the European public has rather positive attitudes towards governments’ intervention to guarantee minimum standards of living for all citizens, including unemployed people (van Oorschot & Meuleman, 2012). On the contrary, support for work obligations of the unemployed seems to be less strong and uniform across countries. Recoding the answer categories to a scale 0–3, we can see that, in the majority of the countries, support does not exceed the midpoint (1.5). In most countries, the dominant opinion is against severe cuts in unemployment benefits when the beneficiaries do not comply with the obligations.

4.2 Latent Class Analysis: A Typology of Preferred Balance of Rights and Obligations

LCA is employed to uncover clusters of respondents preferring distinct combinations of attitudes towards social rights of the unemployed and their work obligations. To identify the most suitable solution among models with different numbers of classes, we rely on the comparison of model fit indices (Nylund et al., 2007). Given that the sample is quite large, looking at the lowest values of the Akaike information criterion (AIC), the Bayesian information criterion (BIC) and the sample size adjusted BIC (aBIC) would lead us to always increase the number of classes identified. A graphical representation of the reduction in these indices (Appendix Fig. 4) shows however that the four-class solution performs better than the three-class solution (looking at the reduction in AIC and BIC), while the five-class solution does not improve the model fit. For these reasons, we choose the four-class solution as final model (see detailed conditional probabilities in Appendix Table 4). Figure 2 plots how these classes relate to the support for welfare rights and work obligations (whereby the size of the bubbles represents the class size).

Fig. 2
figure 2

Four-class solution for the preferred balance of rights and obligations of the unemployed. The figure shows per latent class the average support for rights on the x-axis (from 0 to 10) and support for obligations on the y-axis (average over the three obligation-items, on a scale from 0 to 3). Class sizes are represented by the size of the bubbles. Design weights are applied in the calculation of the averages

Full size image

Three classes are largely situated along a linear continuum from strong support for social rights combined with opposition against obligations (right bottom corner) to weak support for rights combined with strong support for obligations (upper left). This linear continuum captures the negative relationship between rights and obligations that was uncovered in previous research (Laenen & Meuleman, 2019; Roosma & Jeene, 2017). At the extremes we find the unconditional generous (21.2%), which support high rights and no cuts on unemployment benefits, and the work-first supporters (21.3%), with the lowest support for government responsibility for the unemployed while being in favour of cutting about half of the benefits if the unemployed do not comply with the obligations. The largest class (36.3%) has an average support for welfare rights of the unemployed, and they want to cut unemployment benefits only to a small part if the unemployed do not fulfil their obligations; we label this group the mildly generous. One latent class, however, deviates from the linear pattern: the sanctioners (21.2% of the total sample) show an average support for welfare rights for the unemployed, but they endorse the harshest attitudes towards benefit cuts.

Figure 3 shows how these four classes are distributed across European countries. The mildly generous constitutes the largest group in most of the countries, with some exceptions. In Estonia, Hungary and Lithuania the largest group is the unconditional generous; in the Mediterranean countries (Spain, Italy, Portugal, Slovenia), plus Norway and Poland, the sanctioners is the largest latent class in the sample.

Fig. 3
figure 3

Distribution of the four classes across countries. Design weights are applied

Full size image

4.3 Explaining the Preferred Balance of Rights and Obligations: Individual and Contextual Predictors

We use multilevel logistic regressions to get insights into the profiles of individuals opting for one specific combination. Concretely, we estimate the effects of individual-level and country-level variables on the probability of belonging to one of the latent classes (versus belonging to one of the other three classes). Starting from the individual-level effects (Table 1), we observe significant differences regarding both structural characteristics and ideological values in the preferred balance of rights and obligations.

Table 1 Multilevel logistic models predicting unconditional generous, sanctioners, work first supporters and mildly generous (N = 37,737) – individual-level predictors
Full size table

Firstly, the characteristics of the unconditional generous class can be understood in terms of self-interest theory. Compared to white-collar workers, being currently unemployed increases the odds of opting for the unconditional generous combination (instead of choosing one of the other combinations) with factor 1.559. Similarly, past experiences of unemployment increase the odds of belonging to the unconditional generous class with factor 1.378. Furthermore, among the unconditional generous we find individuals with stronger egalitarian values (OR = 1.209), confirming previous findings regarding both support for government redistribution and for demanding activation policies for the unemployed (Jaeger, 2006; Laenen & Meuleman, 2019; Rossetti et al., 2021). Self-identifying with the political left increases the likelihood of preferring the unconditional generosity combination (OR = 1.490), while being right-oriented and scoring higher on the conformity/tradition values lower this likelihood (respectively, OR = 0.892 and 0.920).

Secondly, the socio-economic profile of the class of sanctioners fits partly the expectations. While we do not find a significant effect for blue-collar workers, we observe an overrepresentation of the lower educated among this attitude profile. Regarding ideological predictors, endorsing conformity/tradition values (OR = 1.108) and being right-wing oriented (OR = 1.205) increase the likelihood to belong to the sanctioning profile, while being left-wing oriented (OR = 0.878) and egalitarian (OR = 0.935) decreases the likelihood of opting for this combination.

Thirdly, the ideological profile of the work-first supporters is quite similar to that of the sanctioners. They are more likely characterised by stronger conformity values (although the difference is not so pronounced, as the odds ratio is very close to 1), however we do not observe a significant difference between right-wing oriented and individuals in the centre of the political scale. The egalitarian and the left-wing oriented are less likely to belong to this group. Regarding occupational status, being unemployed and having an experience of unemployment decrease the likelihood of opting for this combination.

Lastly, within the mildly generous class, right-wing people and those with stronger conservation values are underrepresented. However, it seems more difficult to grasp the socio-economic profile of those who opt for this combination. Individuals with previous experience of unemployment and those with up to upper secondary education are also less likely to be in this group.

Table 2 shows the results of the institutional predictors on the likelihood of being in one latent group (versus the other three), controlling for the individual-level predictors. Here the purpose is to analyse the effect of contextual characteristics on the likelihood of choosing one latent class over the others.

Table 2 Multilevel logistic models predicting unconditional generous, sanctioners, work first supporters and mildly generous (N = 37,737) – country-level predictors, controlling for individual-level predictors
Full size table

Looking at the generosity of unemployment benefits, we find mixed evidence for the existence of a link between institutional design and public preferences. In countries with a more generous unemployment system, the group of sanctioners tends to be smaller (OR = 0.681), suggesting that there is a positive feedback effect of the institutional design as people are more reluctant to oppose the current situation by adding strict sanctions to unemployment benefits. Generous unemployment benefits do not automatically lead to a greater willingness to sanction and punish the unemployed, as one might expect on the basis of the thermostatic model. At the same time, higher replacement rates increase the likelihood of opting for the mildly generous combination (OR = 1.240). This finding, however, is harder to interpret, given that this group gathers people with an average support for government’s responsibility, and has a less outspoken profile than the sanctioners group.

Welfare conditionality, on the contrary, does not seem to play a significant role in explaining country differences in choosing one preferred combination over the others. For none of the combinations, the effect of the level of strictness of activation requirements is significant. A stronger conditionality built into the policy design would thus neither trigger public support nor create a backlash. A possible explanation might be that these policies are still relatively new and thus not sufficiently institutionalised to shape public opinion (Raven et al., 2011).

5 Conclusions and Discussion

While welfare attitudes research has demonstrated that support for welfare redistribution and social security provisions is stable in Europe (Bean & Papadakis, 1998; Roosma et al., 2016), and more recent research has shown that conditional unemployment benefits are largely supported in Europe (Buss et al., 2017; Carriero & Filandri, 2018; Rossetti et al., 2022), still little is known about how Europeans combine support for welfare generosity and conditionality. Our contribution goes beyond the linearity assumed to link these two aspects of the welfare state (Laenen & Meuleman, 2019; Roosma & Jeene, 2017). Analysing ESS data of 2016, we find that there are four possible combinations of support for welfare rights and obligations for non-compliant unemployed. Three major conclusions can be drawn from this analysis.

First, the findings suggest that support for welfare rights and for work obligations represents to a certain extent the logic of “two sides of the same coin” (Laenen & Meuleman, 2019), but also that there is an exception: the combination we labelled ‘sanctioners’. The use of latent class analysis allows to find the preferred combinations of support among the population, combinations that would not reflect the reality if constructed by the researchers. While the latent groups found in the sample do not distance themselves in their average support for government intervention for the standard of living of the unemployed, they stand apart on their opinions on work obligations.

Second, these groups are characterised at the individual level by specific ideological and socio-economic profiles. While people with a current or previous experience of unemployment are more likely to combine strong welfare support with no conditions attached (‘unconditional generosity’), they reject those combinations with even minimum levels of sanctioning. Right-wing oriented people, and individuals with conformity values, are more likely to choose for a combination with higher support for strict sanctioning.

Third, the proportions of sanctioners–that is, the combination that deviates from the linear pattern––and mildly generous in a country are partly linked to the institutional design. Generous unemployment benefit systems are linked to less public support for harsh punishment measures for the unemployed. This evidences that generous unemployment policies do not necessarily create a backlash effect in public opinion. On the contrary, generous unemployment policies seem to soften the call for sanctioning policies. It is tempting to interpret this pattern as a residue of policy feedback mechanisms. However, caution is needed and we also need to consider that high replacement rates seem to increase the mildly generous group (i.e., combining average support for welfare rights with some support for sanctions). Despite being difficult to interpret, this result might suggest that the public is not completely against any form of punishment, on the condition that generous welfare benefits are maintained.

The significant relation found between the generosity of welfare system and people’s preferences for certain levels of rights and obligations can have important implications for policymaking and for electoral outcomes. Allocating more or less resources to specific benefits might have important consequences on public demands for more or less obligations. In the other direction, people’s preferred levels of generosity and conditionality might drive policymakers, as well as political parties, towards finding the appropriate balance between benefit generosity and conditionality that also satisfies voters’ preferences. In order to fully understand these relationships between public policy and public attitudes, longitudinal data are required, which would allow us to understand the effects of policies once they are institutionalised. However, the present study was not able to include a time dimension, due to data availability. More research in this direction is needed, to further understand the effects of preferred combinations on citizens’ electoral decisions. Additionally, future research is encouraged to explore whether people’s preferred choices for rights and obligations support are related to other macro-level factors that were not analysed here, such as income inequality.