1. Introduction
Over the last decades, intense debates on welfare reforms emerged that fundamentally reinterpret the social contract between citizens and the state. Much-debated proposals such as the means-testing of benefits, the activation of benefit recipients and the implementation of basic income schemes aim to recalibrate the burdens and benefits of social welfare but appeal to very different logics of (re)distribution (Borosch et al., Reference Borosch, Kuhlmann, Blum, Schubert, de Villota and Kuhlmann2016; Gugushvili and van Oorschot, Reference Gugushvili and van Oorschot2020; Häusermann, Reference Häusermann, Bonoli and Natali2012; Van Parijs and Vanderborght, Reference Van Parijs and Vanderborght2017). Means-tested benefits primarily rely on the distributive justice principle of need by only granting benefits to those with insufficient financial resources (Marx et al., Reference Marx, Salanauskaite and Verbist2016; van Oorschot, Reference van Oorschot, Townsend and Gordon2002). Demanding activation policies attempt to increase labour market mobilization by making access to unemployment benefits increasingly restrictive and conditional (Bonoli, Reference Bonoli2010; Eichhorst et al., Reference Eichhorst, Kaufmann, Konle-Seidle, Reinhard, Eichhorst, Kaufmann and Konle-Seidle2008; van Berkel and Borghi, Reference van Berkel and Borghi2008) and are rooted in the principle of equity. By granting an unconditional income to all citizens without means test or work requirement, basic income schemes subscribe to the principle of equality (Van Parijs and Vanderborght, Reference Van Parijs and Vanderborght2017).
Such logics of distributive justice not only structure the blueprints of these policy reforms (Clasen and van Oorschot, Reference Clasen and van Oorschot2002; Ennser-Jedenastik, Reference Ennser-Jedenastik2018; Ervik et al., Reference Ervik, Kildal, Nilssen, Ervik, Kildal and Nilssen2015; Gugushvili and van Oorschot, Reference Gugushvili and van Oorschot2020; Reeskens and van Oorschot, Reference Reeskens and van Oorschot2013; Van Parijs and Vanderborght, Reference Van Parijs and Vanderborght2017), but also serve as normative reference points for the public at large. According to the moral economy and policy feedback theories, mass attitudes towards welfare do not merely reflect material interests and cost-benefit calculations. Instead, public opinions are grafted on the moral principles and social norms that are embodied by welfare institutions and policies (Mau, Reference Mau2004; Sachweh, Reference Sachweh2012; Taylor-Gooby et al., Reference Taylor-Gooby, Hvinden, Mau, Leruth, Schoyen and Gyory2018). This argumentation postulates that public support for a particular welfare policy depends substantially on the distributive justice principles that are ingrained within their ideal-typical design.
In the case of traditional policy domains, such as pension or unemployment systems, the relevance of justice preferences for public support has been evidenced empirically (Andreß and Heien, Reference Andreß and Heien2001; Jaime-Castillo, Reference Jaime-Castillo2013; Lewin-Epstein et al., Reference Lewin-Epstein, Kaplan and Levanon2003). It remains unclear, however, to what extent attitudes towards newer welfare reforms are rooted in social justice beliefs as well. After all, recent or planned reforms might not yet be sufficiently institutionalized for citizens to be socialized within their underlying conceptions of fairness (Raven et al., Reference Raven, Achterberg, van der Veen and Yerkes2011). Instead of encompassing policies that are newly invented, ‘recent’ or ‘contemporary’ reforms refer in this article to policy proposals that are being introduced and discussed on a much wider scale in more recent years, with a weaker institutionalization across European societies than traditional compensation policies. This lack of institutionalization is certainly the case for basic income proposals, but also means-tested benefits and activation policies are still being experimented with in many different forms and with a variety of policy instruments (Eichhorst et al., Reference Eichhorst, Kaufmann, Konle-Seidle, Reinhard, Eichhorst, Kaufmann and Konle-Seidle2008; Marx et al., Reference Marx, Salanauskaite and Verbist2016; Sainsbury and Morissens, Reference Sainsbury and Morissens2002). Furthermore, while the architecture of new welfare policies is clearly built on normative justice ideals, this is less the case for their actual implementation. In practice, welfare policies often come with unintended side-effects that complicate the realization of their central redistributive goals. Selective policies – such as means-tested benefits, for instance – open debates about deservingness and stigmatization of benefit claimants, which might steer attention away from its underlying objective (Larsen, Reference Larsen2008). Heated public discussions on welfare reforms often refer to a host of alternative solidarity principles and citizens seem to apply a variety of heuristics in formulating their opinions (Arni et al., Reference Arni, Lalive and Van Ours2013; Perkiö et al., Reference Perkiö, Rincon, van Draanen and Torry2019; Rossetti et al., Reference Rossetti, Abts, Meuleman and Swyngedouw2020a; van Oorschot, Reference van Oorschot, Townsend and Gordon2002; Wiggan, Reference Wiggan2012; Zimmermann et al., Reference Zimmermann, Boljka, Rakar and Hrast2020).
This empirical study analyses to what extent support for welfare reforms - that are heavily contested and weakly institutionalized - is rooted in social justice beliefs. Concretely, we investigate to what extent preferences for the principles of equality, equity and need explain support for means-tested benefits, demanding activation policies and basic income schemes. Shared conceptions of justice constitute the pre-eminent moral foundation of welfare attitudes and exploring this relationship hence enables to answer whether attitudes towards these policies have a strong or weak normative foundation (Mau, Reference Mau2004; Sachweh, Reference Sachweh2012; Taylor-Gooby et al., Reference Taylor-Gooby, Hvinden, Mau, Leruth, Schoyen and Gyory2018). This study contributes to the literature for three reasons. First, we evaluate the legitimacy and political feasibility of three prominent contemporary reforms that are founded on very different principles of welfare distribution. Second, investigating attitudes towards these three reforms simultaneously provides novel insight into the diffuse or specific drivers of support. Finally, we contribute to the moral economy and policy feedback literature by testing whether these logics also apply to rather weakly institutionalized yet highly debated welfare policies. To realize this empirically, we analyse data from the CRONOS panel (linked to the European Social Survey) by means of structural equation modelling.
2. Theoretical framework
2.1. Contemporary policy reforms: means-tested benefits, demanding activation policies and basic income schemes
We distinguish three particularly politicized policy trends that each recalibrate the traditional social contract substantially. First, means-tested benefits (that is, taking the financial resources of benefit claimants into account to grant welfare support) are being introduced for new target groups and on a wider scale (Borosch et al., Reference Borosch, Kuhlmann, Blum, Schubert, de Villota and Kuhlmann2016; Gugushvili and van Oorschot, Reference Gugushvili and van Oorschot2020; van Oorschot, Reference van Oorschot, Townsend and Gordon2002). Means-tested benefits are implemented in many different forms, but the most important distinction is between benefits based on resource tests that restrict access to those in poverty and based on affluence tests that exclude the well-off (Sainsbury and Morissens, Reference Sainsbury and Morissens2002). We focus particularly on the former, as discussions on the deservingness of the poor have generally been more salient and politicized than debates about the wealthy (Skilling and McLay, Reference Skilling and McLay2015). These types of means-tested benefits target low-income households and aim to offer poverty relief by implementing vertical redistribution (Marx et al., Reference Marx, Salanauskaite and Verbist2016). Although means-tested benefits are generally selective in nature and associated with a liberal welfare model, they could be combined with universal entitlements to realize effective redistribution (Gugushvili and van Oorschot, Reference Gugushvili and van Oorschot2020; Marx et al., Reference Marx, Salanauskaite and Verbist2016). Yet, the practice of means-tested benefits is often criticized for demotivating the poor, obstructing social participation and creating a non take-up of social rights (van Oorschot, Reference van Oorschot, Townsend and Gordon2002).
Activation policies, as a second reform, set out to decrease welfare dependency by stimulating economic self-reliance and labour market participation (Borosch et al., Reference Borosch, Kuhlmann, Blum, Schubert, de Villota and Kuhlmann2016; Häusermann, Reference Häusermann2006, Reference Häusermann, Bonoli and Natali2012; van Berkel and Borghi, Reference van Berkel and Borghi2008). Usually two forms of activation policies are distinguished: enabling policies that focus on investments in human capital; and demanding activation policies that apply benefit cuts and sanctions to push people into employment (Bonoli, Reference Bonoli2010; Eichhorst et al., Reference Eichhorst, Kaufmann, Konle-Seidle, Reinhard, Eichhorst, Kaufmann and Konle-Seidle2008). We focus on demanding activation, since especially this form is the subject of intense political debates. Demanding activation is an umbrella term, as it includes a multitude of different measures, such as restrictions on the level and duration of benefits, sanctions for non-compliance and mandatory participation programs (Eichhorst et al., Reference Eichhorst, Kaufmann, Konle-Seidle, Reinhard, Eichhorst, Kaufmann and Konle-Seidle2008). Demanding activation policies contain elements that appeal to a social insurance logic as well as to neoliberal discourses that individualise the causes and solutions for unemployment (Wiggan, Reference Wiggan2012). While demanding activation policies do encourage people to leave unemployment, critics argue that they simultaneously reduce job quality and employment stability of those mobilized into paid work (Arni et al., Reference Arni, Lalive and Van Ours2013).
Third, basic income schemes are gaining momentum among a wide audience of policy makers, politicians and academics. In its ideal-typical form a basic income refers to a periodic cash payment provided to all citizens on an individual basis without means-test or work requirements (Van Parijs and Vanderborght, Reference Van Parijs and Vanderborght2017). Although basic income schemes have to date not been fully implemented, there has been a plethora of real-life experiments and a strong societal debate on the feasibility and desirability of its introduction (Widerquist, Reference Widerquist2018). The debates on basic income schemes stretch beyond the universal and egalitarian objectives of this scheme and include questions of how they affect people’s work ethic and to what extent they are effectively able to reduce poverty (Perkiö et al., Reference Perkiö, Rincon, van Draanen and Torry2019; Zimmermann et al., Reference Zimmermann, Boljka, Rakar and Hrast2020). There are many varieties of basic income proposals that deviate from the ideal-typical proposal (De Wispelaere and Stirton, Reference De Wispelaere and Stirton2004). A participation income, which is conditional on participation in socially appreciated activities and a targeted basic income, which excludes people with an income above a certain threshold, are just two examples of variations that diverge from the ideal-type.
2.2. The normative foundations of contemporary welfare reforms: the role of distributive justice
The design of the three above-mentioned reforms is based on three distinct principles of distributive justice that guides the allocation of goods and services in the welfare state – namely, equality, equity and need (Deutsch, Reference Deutsch1975). While equality grants resources to all citizens without additional requirements, equity makes distribution dependent on contributions to the common good. On a societal level, the need principle entails a selective and liberal type of distribution that only allocates resources to deserving groups in need (Kittel, Reference Kittel, Traub and Kittel2020). Since the principles of equality, equity and need provide the normative foundation of the three welfare reforms (Clasen and van Oorschot, Reference Clasen and van Oorschot2002; Ennser-Jedenastik, Reference Ennser-Jedenastik2018; Ervik et al., Reference Ervik, Kildal, Nilssen, Ervik, Kildal and Nilssen2015; Gugushvili and van Oorschot, Reference Gugushvili and van Oorschot2020; Reeskens and van Oorschot, Reference Reeskens and van Oorschot2013; Van Parijs and Vanderborght, Reference Van Parijs and Vanderborght2017), the welfare attitude literature assumes that citizens’ policy preferences tend to align with the dominant moral principles embedded in institutional arrangements (Mau, Reference Mau2004; Sachweh, Reference Sachweh2012; Taylor-Gooby et al., Reference Taylor-Gooby, Hvinden, Mau, Leruth, Schoyen and Gyory2018).
The moral economy and policy feedback literatures argue that institutions socialize citizens within normative frameworks and shared conceptions of social justice, which serve as a reference point in the formation of individual preferences (Mau, Reference Mau2004; Sachweh, Reference Sachweh2012; Taylor-Gooby et al., Reference Taylor-Gooby, Hvinden, Mau, Leruth, Schoyen and Gyory2018). If welfare policies function as norm-shaping institutions, individual attitudes are structured by existing welfare arrangements and tend to align with the (distributive) logics inherent to policy designs. In the case of traditional welfare arrangements, this claim has been corroborated repeatedly (Andreß and Heien, Reference Andreß and Heien2001; Jaime-Castillo, Reference Jaime-Castillo2013; Lewin-Epstein et al., Reference Lewin-Epstein, Kaplan and Levanon2003). However, it remains unclear whether this logic also applies to more recent, less institutionalized welfare policies. Indeed, Raven and colleagues (Reference Raven, Achterberg, van der Veen and Yerkes2011) demonstrate that while policy feedback effects do occur for well-established welfare arrangements, relatively recent welfare proposals have not yet been sufficiently institutionalized to impact citizen’s opinions.
In this article, we revisit the study of normative foundations of welfare reforms by investigating to what extent citizens’ attitudes towards means-tested benefits, demanding activation policies and basic income schemes are shaped by norms of (re)distributive justice. These three contemporary welfare reforms are each anchored in a distinct principle of social justice. On the one hand, one could hypothesize that the main underlying justice principle constitutes a solid normative foundation, in the sense that support for a policy reform is driven by the preference for the justice principle to which the reform refers. On the other hand, arguments exist to expect a weaker normative foundation. That the concrete operationalization and implementation of the reforms is often quite diverse (De Wispelaere and Stirton, Reference De Wispelaere and Stirton2004; Eichhorst et al., Reference Eichhorst, Kaufmann, Konle-Seidle, Reinhard, Eichhorst, Kaufmann and Konle-Seidle2008; Sainsbury and Morissens, Reference Sainsbury and Morissens2002) might impede the socialization within a single redistributive logic and provoke interpretations from a variety of perspectives. Furthermore, the normative anchoring of policy reforms among the general public could be obstructed by the unintended outcomes arising from their implementation and alternative solidarity principles surfacing in public discourses. In other words, weak institutionalization and political contestation of policy reforms could hinder the crystallization of public opinion along the principles that form the bedrock of their moral foundation. Below, we apply this argumentation to the cases of means-tested benefits, activation policies and basic income schemes (see Table 1 for a summary of our expectations).
Expected relationships between distributive justice preferences and support for contemporary welfare reforms
++ = strong positive relationship; + = positive relationship; -- = strong negative relationship; - = negative relationship.
First, the reasoning behind introducing means-tested benefits is strongly based on the need principle, since these policies make the distribution of resources dependent on income or wealth (Ennser-Jedenastik, Reference Ennser-Jedenastik2018; Reeskens and van Oorschot, Reference Reeskens and van Oorschot2013). Means-tests are often portrayed as the most suitable instrument to capture ‘true need’ and the most efficient strategy to target those with insufficient resources (Clasen and van Oorschot, p. 91). Indeed, proponents argue that ideal-typically means-tested benefits are meant to allocate a larger share of redistributive budgets to the poor and those in need (Nelson, Reference Nelson2004). However, weak institutionalization, variations in concrete implementation and political contestation might blur the exclusive link between preferences for the need principle and support for means-tested benefits. These types of selective policies, for instance, open deservingness discussions by explicitly distinguishing groups who are entitled to and excluded from benefits (Clasen and van Oorschot, Reference Clasen and van Oorschot2002; Larsen, Reference Larsen2008). In this sense, the debate around means-tested benefits could shift much more to criteria such as identity, attitude and control. In addition, critics of means-testing argue that it creates poverty traps, leads to larger non take-up of benefits and stigmatizes claimants, which are all counterproductive to helping those in need (van Oorschot, Reference van Oorschot, Townsend and Gordon2002). Advocates argue that this reform tries to reduce inequalities by implementing vertical distributive mechanisms. This might lead citizens who support egalitarian distributions to favour means-tested welfare as well (Gugushvili and van Oorschot, Reference Gugushvili and van Oorschot2020). Preferences for the principle of equity are not likely to be linked to support for means-tested benefits, as this reform does not consider the previous contribution record of citizens, but only their current needs.
Second, demanding activation policies rely most clearly on the principle of equity, as both value labour market participation and contributions to the common good (Clasen and van Oorschot, Reference Clasen and van Oorschot2002; Ervik et al., Reference Ervik, Kildal, Nilssen, Ervik, Kildal and Nilssen2015; Reeskens and van Oorschot, Reference Reeskens and van Oorschot2013). Demanding activation is rooted in the idea of ‘justice as reciprocity’, which is grounded on a conditional contract between citizens and state (Ervik et al., Reference Ervik, Kildal, Nilssen, Ervik, Kildal and Nilssen2015). However, political proponents of demanding activation refer also to the principle of need to justify it. Need-based distribution is selective in nature and encourages self-reliance, which connects closely to the political aims of activation policies (Clasen and van Oorschot, Reference Clasen and van Oorschot2002; Dwyer, Reference Dwyer2000; Fossati, Reference Fossati2018). The principle of equality fits less closely for activation policies. Aspiring to make the allocation of resources dependent on strict behavioural requirements clashes with an egalitarian conception of universal rights (Watts and Fitzpatrick, Reference Watts and Fitzpatrick2018).
Last, basic income schemes relate most clearly to the principle of equality, as this universal and unconditional benefit is granted to everyone without requirements (Van Parijs and Vanderborght, Reference Van Parijs and Vanderborght2017). Given that basic income is a radical manifestation of universalism (Birnbaum, Reference Birnbaum2012; Clasen and van Oorschot, Reference Clasen and van Oorschot2002), egalitarian-minded individuals should be more likely to favour a basic income. Even though need generally equates with more selective welfare state distribution, a broader conception of need as a solidaristic responsibility to care for others is regularly called upon to defend a basic income (Rossetti et al., Reference Rossetti, Roosma, Laenen and Abts2020b). The basic income scheme is considerably more at odds with equity-based distribution as it violates logics of contribution and contradicts the idea of a work society that upholds the moral duty to work as a fundamental part of the social contract (Rossetti et al., Reference Rossetti, Roosma, Laenen and Abts2020b). Although some types of basic income proposals do appeal to logics of conditionality and selectivism (De Wispelaere and Stirton, Reference De Wispelaere and Stirton2004), its ideal-typical form is argued to undermine the central elements of equity.
2.3. Traditional explanatory frameworks: self-interest and ideologies
Although we focus on the normative foundations of support for the three contemporary welfare reforms, we also control for the role of alternative frameworks explaining welfare attitudes. First, self-interest theory argues that welfare policies and redistribution are supported by (potential) beneficiaries of social benefits and services (Jaeger, Reference Jaeger2006; Roosma et al., Reference Roosma, van Oorschot and Gelissen2014). Following this logic, vulnerable social groups could express more support for means-tested benefits that are solely targeted at those in need as well for basic income schemes that provide an unconditional income for all citizens, while disfavouring demanding activation that constrains the level and duration of benefits (Gugushvili and van Oorschot, Reference Gugushvili and van Oorschot2020; Rossetti et al., Reference Rossetti, Abts, Meuleman and Swyngedouw2020a; Vlandas, Reference Vlandas2020). Yet, the relationship between the social structural indicators and preferences for policy reform measures is not always clear-cut. Lower social strata could, for instance, oppose means-tested benefits that facilitate further welfare backlashes and basic income proposals that are likely to replace existing unemployment benefit schemes, while accepting activation policies that overcome social stigma (Fossati, Reference Fossati2018; Kangas, Reference Kangas1995; Vlandas, Reference Vlandas2020).
Second, according to the ideology approach, welfare attitudes are driven by coherent systems of cultural and political motivations (Staerklé et al., Reference Staerklé, Likki, Scheidegger and Svallfors2012). This logic assumes that policy preferences do not necessarily reflect material interests, but are embedded in a broader set of political norms (Jaeger, Reference Jaeger2006). In empirical research, political left-right placement is often used as a broad ideological indicator. This perspective expects left-wing individuals to be more in favour of basic income schemes, while right-leaning individuals should be prone to support activation polices and means-tested benefits, as they rely on conditional and selective solidarity notions (Clasen and van Oorschot, Reference Clasen and van Oorschot2002; van Oorschot, Reference van Oorschot2006). However, alternative patterns could possibly emerge as well. A basic income is heavily contested among certain segments of the political left and these schemes equally appeal to right-wing ideals, such as efficiency and individual freedom (Chrisp and Martinelli, Reference Chrisp, Martinelli and Torry2019; Schwander and Vlandas, Reference Schwander and Vlandas2020). Activation policies that promote social investment and re-integration programs might similarly be supported by leftist segments and means-tested benefits can also appeal to progressive ideas by trying to alleviate poverty (Gugushvili and van Oorschot, Reference Gugushvili and van Oorschot2020; Häusermann, Reference Häusermann, Bonoli and Natali2012).
3. Data and method
3.1. Data
This article uses data from the Cross-National Online Survey (CRONOS) panel, which is an online probability panel constructed as a follow-up of the European Social Survey (ESS) round eight (2016/2017). ESS consists of probability-based samples of the resident population of 15 years and older that are interviewed by means of face-to-face surveys. After completing the interview of the ESS, respondents from Estonia (EE), United Kingdom (UK) and Slovenia (SI) that are 18 or older were invited to participate in six online surveys spread out over twelve months. In this paper, data from the third wave of CRONOS are used, which are linked to the data from the main questionnaire of the ESS round eight. Of the 5285 respondents (EE = 2019; UK = 1959; SI = 1307) that were interviewed in the face-to-face stage of ESS, 2437 respondents (EE = 806; UK = 926; SI = 705) participated in CRONOS. As the inquiry of this article is not comparative, our analyses are conducted on the pooled dataset (yet taking country fixed effects into account). Design weights were applied in the analyses to control for unequal probabilities of selection in the sampling design.
3.2. Indicators
3.2.1. Dependent variables
This study includes three dependent variables, which were all included in the face-to-face ESS survey (full sample). First, support for means-tested benefits is operationalized by the following single item: “Would you be against or in favour of the government providing social benefits and services only for people with the lowest incomes, while people with middle and higher incomes are responsible for themselves?” (four-point scale ranging from “Strongly against” to “Strongly in favour”).
Second, support for demanding activation policies is measured by three items, which ask what should happen to the unemployment benefit of a benefit claimant who turns down a job because it pays a lot less than they earned previously, who turns down a job because it needs a much lower level of education or who refuses to regularly carry out unpaid community service. The four answer categories range from “This person should lose all their unemployment benefit” to “This person should be able to keep all their unemployment benefit” and are reversed so that higher scores refer to more support for activation measures. These questions were part of a survey experiment, in which respondents were randomly assigned to four conditions wherein the characteristics of the benefit claimant varied (“Someone”, “Someone in their 50s”, “Someone aged 20-25” and “A single parent with a 3-year old child”). Since the main analytical interest is not in differences across these categories, assignment to these experimental conditions is included as a control variable in the structural model. Measurement invariance is tested for this latent concept across the three countries (see Table A1 in Appendix). As the metric invariant model shows good fit and strong factor loadings in each of these countries (see Table A2 in Appendix), factor scores for this model are saved and included in the final regression model.
Last, support for basic income schemes is measured by asking respondents to what extent they support implementing a basic income that has the following characteristics: a monthly income granted by the government, which replaces many other social benefits, guarantees a minimum standard of living, gives everyone the same amount regardless of whether or not they are working, lets people keep the money they earn from other sources and is paid by taxes (four-point scale, ranging from “Strongly against” to “Strongly in favour”).
Figure 1 provides descriptive statistics for support for each of the reforms. As the question formats and answer categories vary considerably across the three dependent variables, the descriptives should not be compared directly across reforms. On average each of these welfare reforms receives a moderately high degree of public support. For means-tested benefits, 46 percent is (strongly) in favour of granting only benefits and services to those with the lowest incomes. A small majority (53 percent) (strongly) favours introducing the basic income scheme. A reduction of benefits as a sanctioning strategy is chosen by 71 percent of respondents (lose small part, half or all). An inspection of the correlations indicates that attitudes towards these welfare reforms are hardly related. There are very weak, yet significant, positive correlations between support for means-tested benefits and support for both activation policies (r = 0.05) and basic income schemes (r = 0.04), whereas the association between attitudes towards activation and a basic income is insignificant and negative (r = −0.03).
Percentages per answer category for attitudes towards means-tested benefits, activation policies and basic income schemes on the pooled dataset Footnote 1 .
3.2.2. Independent variables
Preferences for the distributive justice principles of equality, equity and need are each measured by a single item on a five-point agree-disagree answer scales and are only included in the CRONOS sample, stemming from the BSJO scale (Hülle et al., Reference Hülle, Liebig and May2017). Although the CRONOS sample includes two questions per distributive justice principle, assessments of measurement quality indicate that these items cannot be integrated into single and unambiguous latent concepts (Van Hootegem et al., Reference Van Hootegem, Meuleman and Abts2021). As an alternative, single items are chosen that provide the clearest indicator of each distributive principle. Using single items has the disadvantage that random measurement error is not taken into account, and that effect parameters might be biased downward. However, a clearly formulated single item is to be preferred over combining multiple items that do not measure the same concept into a single score. For equality, respondents are asked whether they believe that “For a society to be fair, differences in people’s standard of living should be small”. For equity, the item states that “A society is fair when hard-working people earn more than others” and for need it is formulated as “A society is fair when people who look after their children or their relatives in need of care receive special support and financial benefits”. Although the need item is measured in a way that diverts from the liberal interpretation of need that is represented in means-tested benefit (for instance, Kittel, Reference Kittel, Traub and Kittel2020) it still captures people’s support for granting resources and ensuring the well-being of those with higher needs. Scores are reversed so that higher values point to more support for each of these principles.
In addition, the social structure and left-right position are used as explanatory variables (all stemming from the ESS main survey). Education is categorized into three groups: no to lower secondary education, higher secondary education (reference category) and tertiary education. On the basis of the Erikson-Goldthorpe-Portocarero scheme (Ganzeboom and Treiman, Reference Ganzeboom and Treiman1996), occupation is divided into six categories: the service class, white collar workers, blue collar workers (reference category), the self-employed, the unemployed and the retired and other non-actives. To limit the number of missing values, income is operationalized as subjective rather than objective income. We differentiate between the following three income categories: living comfortable on present income (reference category), coping on present income and finding it (very) difficult on present income. Political ideology is measured as left-right placement, which is operationalized by a single item on an eleven-point scale (0 = left; 10 = right). Gender (female = reference category), country (UK; Slovenia; Estonia = reference category) and age are included as control variables. Descriptive statistics for all the study variables per country are displayed in Table A3 in Appendix.
3.3. Statistical modelling
Because of the considerable drop-out between the face-to-face survey and CRONOS, information for the distributive justice items is missing for 53.9% of the respondents. Instead of conducting complete case analysis (that assumes that the missingness is completely at random), we use multiple imputation (MI) techniques. Multiple imputation encompasses replacing the missing values by multiple draws from a distribution conditional on the known information, thereby creating multiple datasets. Even in the presence of a high share of missing values for the justice preferences, correctly conducting MI is still superior to complete case analysis, which has more stringent missing data assumptions, can result in biased estimates and reduces power substantially (Azur et al., Reference Azur, Stuart, Frangakis and Leaf2011; Graham, Reference Graham2009) Footnote 2 . After the imputation stage, structural equation modelling (SEM) was applied across all the imputed datasets to obtain averaged estimates of regression coefficients. We specifically apply SEM, to estimate regression coefficients on support for all three welfare reforms simultaneously. Model fit is nearly perfect (df = 6; RMSEA = 0.000; CFI = 1.000; TLI = 1.089; SRMR = 0.002).
4. Results
4.1. Structural equation model
Table 2 displays the results from the structural equation models that predicts support for means-tested benefits, activation policies and basic income schemes. Given that the three dependent variables are measured quite differently, we do not attempt direct numerical comparison. Yet, to make the broad predictive patterns more comparable across the three dependent variables, the regression coefficients are based on standardization of the dependent variable and the metric independent variables. The dummy variables are not standardized, so that these parameters refer to the number of standard deviations a group differs from the reference category.
Standardized regression coefficients for a structural equation model averaged over the imputed datasets (N = 4392)
* p≤0.05; ** p≤0.01; *** p≤0.001;
a For support for activation policies, the experimental conditions of the survey question were included as covariates. However, as support for means-tested benefits and basic income schemes were not regressed on these conditions and they do not constitute the primary research interest of this article, they have been omitted from the table.
Table 2 shows how the distributive justice preferences relate to support for the three welfare reforms, which enables to analyse their normative foundations. Contrary to our expectations, preferences for equality, equity and need do not have a significant impact on public support for means-tested benefits. Although the principle of need clearly structures the ideal-typical design of this reform (Clasen and van Oorschot, Reference Clasen and van Oorschot2002; Ennser-Jedenastik, Reference Ennser-Jedenastik2018; Gugushvili and van Oorschot, Reference Gugushvili and van Oorschot2020), citizens do not appear to base their opinions on its underlying redistributive logic. This might be in part related to the lack of full institutionalization of this reform, which obstructs a strong socialization within the principle structuring its blueprint (Raven et al., Reference Raven, Achterberg, van der Veen and Yerkes2011). In addition, the ineffective realization of the outcomes set out by the need principle, in terms of guaranteeing better well-being for those in need (van Oorschot, Reference van Oorschot, Townsend and Gordon2002), might lead people away from this normative idea when formulating their opinions. Instead, the discussions around means-tested benefits could occur much more around the deservingness of benefit recipients and centre around a stigmatizing picture of those in need of assistance (cf. Larsen, Reference Larsen2008). Despite the finding of previous research that means-tested benefits are also defended from an egalitarian point of view (Gugushvili and van Oorschot, Reference Gugushvili and van Oorschot2020), a preference for the principle of equality does not stimulate support for this welfare reform. Preferences for equity do not explain why people accept or reject means-tested benefits either. Clearly, attitudes for means-tested benefits are not structured by the normative principles.
As expected, demanding activation policies appeal to a logic of reciprocity that is built into the design of this policy (Ervik et al., Reference Ervik, Kildal, Nilssen, Ervik, Kildal and Nilssen2015). The regression coefficient for equity is significant but weak in strength (b = 0.10), which indicates that the main redistributive logic on which demanding activation is founded only limitedly forms its attitudes. A preference for the principle of equality, in contrast, significantly lowers support for work obligations for the unemployed (b = −0.05). Although this coefficient is also relatively small, the relationship is as anticipated and conforms to the observation that support for work obligations and for broad government redistribution are generally negatively associated (Laenen and Meuleman, Reference Laenen and Meuleman2018). While a preference for need-based distribution has a significant impact, the coefficient is, contrary to what was expected, negative (b = −0.07). Despite the strong neoliberal elements in the demanding activation discourse (Wiggan, Reference Wiggan2012) and the differentiation between deserving and undeserving recipients that characterizes both need-based distribution and demanding activation (Clasen and van Oorschot, Reference Clasen and van Oorschot2002; Dwyer, Reference Dwyer2000; Fossati, Reference Fossati2018), opinions on this welfare reform are negatively and only weakly related to the need principle. The combination of being implemented in many different forms that often deviate from the core distributive principle underlying its design, and of being surrounded by varying political interpretations (Eichhorst et al., Reference Eichhorst, Kaufmann, Konle-Seidle, Reinhard, Eichhorst, Kaufmann and Konle-Seidle2008; Wiggan, Reference Wiggan2012), makes support for demanding activation only weakly related to the normative principles of social justice.
Consistent with our expectation, support for a basic income scheme is especially stimulated by a preference for the equality principle (b = 0.09). Yet, once more, the size of the effect is rather weak, which indicates that the equality-based foundation of the ideal-typical basic income does not consolidate into a strong normative basis for its public support (Van Parijs and Vanderborght, Reference Van Parijs and Vanderborght2017). The relationship with equity is, as anticipated, negative (b = −0.07), which might stem from the criticism on the unconditional nature of basic income schemes that does not consider previous contribution records. Although there are types of basic incomes that do rely more strongly on conditionality and heuristics of reciprocity also permeate discussions on this proposal (De Wispelaere and Stirton, Reference De Wispelaere and Stirton2004; Perkiö et al., Reference Perkiö, Rincon, van Draanen and Torry2019; Zimmermann et al., Reference Zimmermann, Boljka, Rakar and Hrast2020), support for an ideal-typical basic income is negatively, yet weakly, associated with a preference for equity-based distribution. Last, the regression coefficient of the need principle is small and positive (b = 0.08), indicating that basic income schemes not only connect to egalitarian values, but are also argued by its proponents to function as effective means to help those in need (Birnbaum, Reference Birnbaum2012). All relationships are relatively weak, which is not surprising for a reform type that has not been institutionalized at all and where citizens are exposed to a lot of varying interpretations of its design.
Next, we shift attention to the effects of the social structural variables. For means-tested benefits, older respondents and those having difficulties on their present income express more support, while respondents with tertiary education and white-collar workers are significantly less in favour of this reform than those with higher secondary education and blue-collar workers, respectively. These findings are in line with the self-interest thesis, as vulnerable groups in more precarious positions, such as the elderly and low-income individuals, generally have a stronger interest in targeted welfare (Gugushvili and van Oorschot, Reference Gugushvili and van Oorschot2020). In addition, while older respondents express more support for demanding activation policies, those in the highest educational category, individuals who have difficulties on their present income and both the service class and the unemployed are significantly less in favour of benefit sanctions and work obligations. Although the lower support for demanding activating measures of low-income groups and the unemployed conform to their self-interest (Fossati, Reference Fossati2018), the higher popularity among older respondents and those who have not completed tertiary education contradicts this thesis. However, the larger support among those who have not completed tertiary education might be related to their stronger authoritarian dispositions and support for a work ethic (Rossetti et al., Reference Rossetti, Abts, Meuleman and Swyngedouw2020a). For basic income schemes, only age and income have a significant impact: younger respondents and those having difficulties on their present income express more support for unconditional and universal social protection. The relationship with income is in line with self-interest theory, but the higher support of younger respondents is more ambivalent. The country parameters indicate that respondents from the United Kingdom are more accepting of means-tested benefits and basic income schemes and Slovenian individuals are more supportive of basic income schemes and less in favour of means-tested benefits than Estonian respondents.
Political ideology has a significant, yet moderate, impact on support for each of the policy reforms. Right-wing individuals express more support for means-tested benefits and demanding activation policies, while left-wing individuals are more in favour of a basic income. These results are in line with the predominant expectations, as means-tested benefits and activation policies rely on more conditional and selective notions of solidarity that especially appeal to the political right and basic income schemes traditionally find most support among the left (Chrisp and Martinelli, Reference Chrisp, Martinelli and Torry2019; Clasen and van Oorschot, Reference Clasen and van Oorschot2002; Fossati, Reference Fossati2018; Rossetti et al., Reference Rossetti, Abts, Meuleman and Swyngedouw2020a; van Oorschot, Reference van Oorschot2006; Vlandas, Reference Vlandas2020).
5. Conclusion
Means-tested benefits, demanding activation policies and basic income schemes are increasingly debated and put forward as viable welfare reforms. The ideal-typical design of each of these reforms relies on a different redistributive logic and hence on one of the distributive justice principles of equality, equity and need. According to the moral economy and policy feedback literatures, citizens are socialized within these normative principles and use them as motivational reference points in formulating their opinions. However, as these proposals are not yet fully institutionalized and their normative roots are often contested in their implementations as well as discourses, citizens might not be as strongly socialized within these logics of distributive justice and might interpret these reforms differently. This article examined the normative foundations of contemporary welfare reforms by considering the impact of preferences for equality, equity and need on support for means-tested benefits, demanding activation policies and basic income schemes.
Results indicated that for means-tested benefits none of the distributive justice principles exert a significant influence. In spite of the reliance of the policy design of means-tested benefits on the need-principle (Clasen and van Oorschot, Reference Clasen and van Oorschot2002; Ennser-Jedenastik, Reference Ennser-Jedenastik2018; Gugushvili and van Oorschot, Reference Gugushvili and van Oorschot2020), its public opinion is not grounded on this justice ideal. This could in part be related by the predominance of an alternative discourse that stereotypes the poor, discourages the take-up of benefits and encourages self-reliance, instead of truly offering relief for those most in need. Support for demanding activation policies was slightly higher among those preferring the principle of equity to organise distributions. Yet the weak relationship indicates that activation policies do not necessarily strongly build upon the underlying justice conception that values contributions to the common good and a quid-pro-quo welfare model (Ervik et al., Reference Ervik, Kildal, Nilssen, Ervik, Kildal and Nilssen2015). Although we also anticipated the need principle to stimulate support for work obligations and benefit sanctions, preferences for need and equality lowered support for activation policies. For basic income schemes, all relationships were as expected, as equality and need significantly strengthened support for basic income, while a preference for equity weakened its popularity. Yet as for the other welfare reforms, the relationships were not very strong, which indicates that public support for a basic income only limitedly builds on the normative principles inherent to its institutional blueprint. In addition to the justice preferences, self-interest and ideology had a significant, yet moderate, impact on attitudes towards each of these policy reforms.
All in all, public opinion on means-tested benefits, demanding activation policies and basic income schemes is not strongly crystallized according to social structural, ideological or distributive dividing lines. Indeed, debates on these new types of policy reforms do not only seem to partly transcend traditional class coalitions and partisan alliances (Häusermann, Reference Häusermann2006), but also to not be fully based on the classical organizing principles of the welfare state (Clasen and van Oorschot, Reference Clasen and van Oorschot2002). Each of the reform types are not fully institutionalized yet, are debated from a whole array of perspectives and are interpreted with reference to many different heuristics (Arni et al., Reference Arni, Lalive and Van Ours2013; Perkiö et al., Reference Perkiö, Rincon, van Draanen and Torry2019; Rossetti et al., Reference Rossetti, Abts, Meuleman and Swyngedouw2020a; van Oorschot, Reference van Oorschot, Townsend and Gordon2002; Wiggan, Reference Wiggan2012; Zimmermann et al., Reference Zimmermann, Boljka, Rakar and Hrast2020), which might explain why these attitudes are not strongly consolidated according to the normative foundations of their ideal-typical design.
This article nevertheless has a few limitations. To begin with, the measurements used for both the welfare reforms and the justice principles were not necessarily ideal. Not all attitudes towards welfare reforms were measured in the same way, which complicates comparability, and the single justice items do not capture the multifaceted nature of the distributive ideas. Second, a pooled analysis across the countries was conducted to get a first indication of the extent to which justice norms are ingrained in support for the reforms. Yet, as different types of distributive reasoning can be predominant across various welfare regimes (for instance, Taylor-Gooby et al., Reference Taylor-Gooby, Hvinden, Mau, Leruth, Schoyen and Gyory2018), it would be interesting to inspect these relations across different social security systems. Last, as the moral economy and policy feedback arguments seem to only limitedly offer an explanation for why these reforms are supported, future research would benefit from inspecting alternative mechanisms driving these public preferences.
Acknowledgments
This work was supported by the National Science Foundation – FWO-Vlaanderen [Grant Number: 11H2619N].
Declaration of interest
Competing interests
The authors declare none.
Appendix
Measurement invariance for the latent concept of support for demanding activation policies
Question wording and standardized factor loadings of support for demanding activation policies for the metric invariance model
Descriptive statistics for all study variables per country
Open access
