Welfare Policy Preferences
Preferences regarding social policies have attracted much scholarly attention in the past decades (e.g., Andress & Heien,
2001; Häusermann & Walter,
2010; Kangas,
1997; Kulin & Svallfors,
2013; Rehm,
2009; Roosma et al.,
2013; Svallfors,
2003). This interest is related to the fact that these preferences are expected to influence individuals’ electoral choices and thus to be linked with public policy. As a result, citizens’ preferences are perceived as important for understanding cross-country differences or temporal evolution in the shape or size of welfare states in democracies (e.g., Svallfors,
1997,
2003).
Early work on the topic has considered individual
self-interest as a main driver for individuals’ preferences with regard to redistribution (Meltzer & Richard,
1981). Following that logic, individuals’ support for redistributive policies largely depends on whether they are net contributors or beneficiaries of the welfare state. However, while self-interest certainly influences individual attitudes, research has also established that current income can only explain a small part of the variation in preferences between individuals. Accordingly, recent studies have begun to incorporate other aspects less directly related to self-interest, such as individuals’ risk profiles (Kananen et al.,
2006; Rehm et al.,
2012; Rueda,
2005), their expectations about the future (Bénabou & Ok,
2001), their unemployment experiences (Naumann et al.,
2016), or “externalities of inequality” (Rueda & Stegmueller,
2016).
Likewise,
subjective experiences of poverty, including feelings of relative deprivation (e.g., Kreidl,
2000) and direct exposure to poverty in one’s surroundings (e.g., Hopkins,
2009), have been shown to affect people’s thinking about social inequality and their welfare attitudes. Going one step further, research has stressed the importance of normative orientations which are only dimly related to self-interest—including inequality aversion (e.g., Munro,
2017), personal and political values (e.g., Arikan & Bloom,
2015), ideology (e.g., Jaeger,
2008), or perceptions of the deservingness of the poor (e.g., van Oorschot,
2006).
A (Not So) New Construct: Explanations for Poverty
In comparison with the variables reviewed above, there has been relatively little interest in the effect of explanations for poverty on policy preferences. As we explain in more detail below, this neglect is somewhat surprising. Ever since it was first proposed by Feagin (
1972) in the early 1970s, the construct of “lay explanations for poverty” has been obviously related to matters of social inequalities and welfare policy. (The terms “explanations for poverty” and “poverty attributions” convey the same meaning and are used interchangeably in this article.) For one thing, unlike “expert” approaches to socioeconomic inequalities, “lay” poverty attributions are the explanations provided by ordinary people to account for the existence and persistence of poverty in contemporary societies. Although these attributions are certainly reflective of elite debates, media framing and policy changes (e.g., Bullock et al.,
2001; Iyengar,
1990; Kangas,
2003; Wacquant,
1999), they also constitute an independent source to understand the ebbs and flows of welfare policy support. Likewise, they may be useful to measure the convergence between attitudes of elites and mass public attitudes. Thus, for example, when some countries took austerity measures including shrinkage of social services in the wake of the 2008 crisis, this was at odds with a surge of the “social blame” attribution of poverty in the years 2009‒2014 (Marquis,
2020), which may help explain the social turmoil arising in this period.
At first sight, poverty can be seen to have a variety of sources. Accordingly, the literature on popular explanations for poverty has come up with various measurements and typologies. One strand of research, following in the footsteps of Feagin’s initial proposal, has elaborated a three-category typology of “individual,” “social,” and “fatalistic” explanations (e.g., Feather,
1974; Smith & Stone,
1989; see also Furnham,
2003; Hunt & Bullock,
2016). The empirical validity of the three-category typology has not remained unquestioned, though. For example, in his analysis of explanations for unemployment in Britain, Furnham (
1982) recognized the relevance of two types of fatalistic explanations: one stressing the incompetence of industrial management and one stressing “outside influence” (also including, interestingly, the item “just bad luck”). Van Oorschot and Halman (
2000) took notice of Furnham’s (and others’) suggestion and tried to provide a more systematic account of all possible explanations of poverty. They made clear that a fourth type should be added to Feagin’s three-type classification, namely a “social fate type.”
As shown in Table
1, the four categories of the typology correspond to the combination of (1) judgments regarding the location of the explanation for poverty at the individual level or at the social level, and (2) the perception that individuals/society are responsible for poverty (“blame”) or that poverty arises from circumstances and events beyond control of individuals or social institutions (“fate”). Table
1 summarizes the four-type typology, and the way each attribution type is usually operationalized in opinion surveys, including in the Eurobarometer which we use in our analysis. The standard text indicates the label given to each attribution type (see van Oorschot & Halman,
2000) and in italics the response to the following survey question:
Why in your opinion are there people who live in poverty? Here are four opinions: which is closest to yours?Table 1
Explanations for poverty according to the four-type typology
Blame (agency) | Individual blame Because of laziness and lack of willpower | Social blame Because there is a lot of injustice in our society |
Fate (non-agency) | Individual fate Because they have been unlucky | Social fate It is an inevitable part of the way the modern world is going |
Admittedly, the diffusion of the four-category typology in academic research was favored (or directly inspired) by its inclusion in international surveys readily available to scholars (Lepianka et al.,
2009: 427). Next to the Eurobarometer series which will be analyzed in this article, the four-type typology also features in the European Values Study (waves 2‒4), in the British Social Attitudes Survey (until 2010), and in the World Values Survey (waves 2‒5). Accordingly, the measurement of poverty attributions pursued in the current study has been validated by its use in many studies yielding similar findings (e.g., Kainu & Niemelä,
2014; Kallio & Niemelä,
2014; Lepianka et al.,
2010; Marquis,
2020; Niemelä,
2008).
Still, this measurement approach is essentially
data-driven, and it is important to remember that a significant part of research on poverty attributions has gone down different paths. For example, especially after the influential contribution by Kluegel and Smith (
1986), American research on poverty attributions has tended to favor a two-type measurement of poverty attribution, based on factor analyses of items suggesting the presence of individualistic and structuralist dimensions (e.g., Hunt,
2002,
2004; Merolla et al.,
2011). Although a third, “fatalistic” factor emerged in some of these studies, it was usually discarded because of the dearth or “unpopularity” of items tapping the fatalistic dimension (Hunt,
1996: 318; Hunt,
2016: 394; Hunt & Bullock,
2016) or because both fatalistic and structuralist items loaded on this additional factor (Bullock,
1999; Bullock et al.,
2003). Let us note that the two/three-type approach has also been applied in Europe, for example in Germany (Schneider & Castillo,
2015) or in the 11 European countries which participated in the International Social Justice Project (e.g., Kluegel et al.,
1995; Kreidl,
2000; Kluegel & Mason,
2004). As a matter of fact, however, most of the surveys using the four-type typology shown in Table
1 center on European countries, which may explain part of the difference between European and American research traditions and empirical results.
Explanations for Poverty and Welfare Policy Preferences
In fact, popular explanations of poverty have often been analyzed as dependent variables to be explained, more rarely as
independent variables to explain political attitudes and behaviors. More often than not, the links between poverty attributions, welfare state preferences and voting are taken for granted. Possibly one of the reasons for this lack of interest in the effect of poverty attributions on political preferences is the rather obvious link between the two. Harper (
1996) is more critical toward this neglect, pointing out “a startling lack of curiosity about what effects and functions these kinds of explanations [for poverty] might have. (…) In ignoring such difficulties, traditional attributional research on poverty explanations has been essentially conservative in its theory and methodology and has failed to deliver findings which might be of use in acting politically and socially against poverty” (Harper,
1996: 252). Another contentious point which might have restrained scholars from investigating the political consequences of popular explanations for poverty is the issue of
causality. In particular, there have been suggestions that poverty attributions are ex-post rationalizations of individuals’ ideological orientation or welfare preferences. For example, Paugam et al. (
2017) suggest that affluent people tend to
justify poverty on the basis of preexisting neoliberal and meritocratic ideological principles, while other scholars uncritically assume that causality runs from welfare attitudes to poverty attributions (e.g., Niemelä,
2008). In their seminal study, Kluegel and Smith (
1986: 267–270) have argued against that viewpoint, emphasizing that sources of poverty attributions lie, for the most part, outside the political realm (see also Iyengar,
1990; Gilens,
1999: 85–89). This argument is buttressed by studies showing that beliefs about poverty are acquired early in life, before political socialization per se occurs (e.g., Bullock,
2006; Chafel,
1997; Chafel & Neitzel,
2005; Leahy,
1990). This does not mean, of course, that poverty attributions are exclusively determined by childhood experiences; as a matter of fact, all available evidence shows that explanations of poverty can change over time according to macro-level and personal circumstances. But the point is that the existence of political preferences is neither a sufficient nor a necessary condition for popular explanations of poverty to develop at the individual level.
How do poverty attributions affect social policy preferences? The relationship may seem obvious but it has seldom been subjected to theoretical analysis. It thus seems necessary to reconcile research traditions which, “despite evident conceptual links,” have tended to develop “parallel agendas” (Schneider & Castillo,
2015: 264). To begin with, the legitimacy of welfare institutions and policies is deeply rooted in well-entrenched social norms such as equity, fairness, solidarity, distributive justice, and reciprocity (e.g., Aalberg,
2003; Deutsch,
1985; Kangas,
2003; Kluegel & Smith,
1986; Mau,
2004; Miller,
1992; Nozick,
1973; Rawls,
1971; Rothstein,
1998). Welfare policies tend to enjoy wide support as long as the actors involved in redistributive mechanisms (i.e., contributors and recipients of welfare policies, but also welfare institutions themselves) are perceived to conform to these social norms (Bowles & Gintis,
2000; Fong et al.,
2005). In contrast, when these norms are obviously violated, a breakdown of the pro-welfare consensus is likely to ensue. With respect to anti-poverty programs, “people are willing to help the poor, but they withdraw support when they perceive that the poor cheat or fail to cooperate by not trying hard enough to be self-sufficient and morally upstanding” (Fong et al.,
2005: 279). This withdrawal of support closely corresponds to the endorsement of the “individual blame” category in the poverty attribution typology presented above. When the poor are deemed responsible for their own fate, feelings of reciprocity and “moral obligations” to the poor seem to dissolve, thus undermining the legitimacy of welfare policies (Kangas,
2003).
In this article, we subscribe to the common argument that this blame mechanism extends to other sectors of state intervention. As Kluegel and Smith (
1986: 164) put it, “antiwelfare sentiment seems to be linked to a ‘victim-blaming’ view of the poor as lazy, lacking thrift and good morals, etc.: the items representing individual explanations for poverty.” As a matter of fact, several studies have empirically explored the link between poverty attributions and a wide array of welfare policy preferences (Alston & Dean,
1972; Bullock,
1999; Burgoyne et al.,
1999; Bradley & Cole,
2002; Bullock et al.,
2003; Bullock,
2004; Feagin,
1972; Fong,
2001; Hasenfeld & Rafferty,
1989; Habibov et al.,
2017; Iyengar,
1990; Kluegel & Smith,
1986; Mau,
2003; Stephenson,
2000; Tagler & Cozzarelli,
2013; Williamson,
1974; Zucker & Weiner,
1993). Virtually all of these studies have established significant relationships between poverty attributions and welfare preferences. Interestingly, poverty attributions were also found to affect the degree to which economic inequalities are perceived as just or unjust (Schneider & Castillo,
2015). In sum, poverty attributions seem to have a pervasive influence on how people conceive the legitimacy of the social stratification at large and, in turn, on their welfare attitudes.
For our present purposes, it is unfortunate that this strand of research has mostly focused on the distinction between individual attributions (above all “laziness”) and structural attributions. With few exceptions, it has failed to take into account the
agency dimension—are individuals or social institutions to blame for poverty, or is the problem beyond control of individuals and institutions? Thus, to take the perspective of Table
1, the “blame” and “fate” rows have been conflated within the “individual” and “structural” columns of the typology. Hence, we need to develop more definite expectations about the influence of the four attributional types. Following the general argument formulated above, people endorsing the “individual blame” and “social blame” categories should be the least and the most likely to support redistributive policies, respectively. The two fatalistic categories are expected to fall in between (for a similar analysis, see Halman & van Oorschot,
1999: 4–5; van Oorschot & Halman,
2000: 21–23; Da Costa & Dias,
2014: 1410). First, “individual fate” attributions (e.g., bad luck) should elicit willingness to help the poor (e.g., public relief services) and thus should foster some support for redistribution. However, since poverty is seen as stemming from fatality rather than from structural inequalities, there should be no real impetus for supporting “preventive” policies designed to fight the causes of poverty (unemployment, low education, insufficient pensions, etc.), which should be more popular among people endorsing a “social blame” attribution of poverty. Second, “social fate” attributions ascribe poverty to a normal state of affairs—poverty is determined by impersonal and uncontrollable social forces, so the “modern world” argument goes. According to the social fate attribution type, poverty is here to stay because it is a natural consequence of the capitalist system. However, social policies may be seen as a necessary tool to maintain the system in the long run. By dealing with social inequalities and by meeting demands for social protection, social policies can be seen to fulfill a social control function designed to keep disadvantaged groups quiescent and to prevent social unrest (Armour & Coughlin,
1985; Brisman,
2012; Kim,
2007; Piven & Cloward,
1971; Schneider & Ingraham,
1984; Soss et al.,
2011; but see Dodenhoff,
1998). Of course, we do not assume that all people attributing poverty to “social fate” are keen supporters of the capitalist system. In this regard, a detailed analysis of the four-category typology, where 16 specific causes of poverty are related to the four general attributions, indicates that the social fate type is the most heterogeneous and the most uneasy to interpret (Lepianka et al.,
2009). This “all-embracing character of the modern world category” (2009: 430) seems to result from the blending of constitutive elements of the “individual fate” and “social blame” categories; in contrast, the key element of the “individual blame” category (i.e., laziness) is rarely mentioned by those who choose the social fate type. Overall, then, people who attribute poverty to social fate are expected to display more support for welfare policies than people blaming poverty on the poor themselves, but less support than people endorsing a “social blame” explanation.
Covariates of Poverty Attributions
By covariates, we mean variables which are supposedly related to explanations of poverty, our main independent variables, and which might also be important for the formation of welfare attitudes, our dependent variables. Accordingly, the influence of poverty attributions on welfare preferences may be confounded with the influence of these covariates, and we will try to disentangle the effects of independent variables and covariates in a series of robustness checks (see Appendix). In the following discussion, we identify four covariates of poverty attributions: (1) deservingness judgments; (2) political and social trust; (3) political ideology; and (4) personal experience of poverty.
First, the perceived
deservingness of actual or potential welfare state beneficiaries has been identified as an important antecedent of welfare policy preferences (e.g., Delton et al.,
2018; Feather,
1994,
1999; Hansen,
2019; Koster,
2018; Larsen,
2006: chap. 4; Mau,
2003; Raven,
2012; Roosma et al.,
2016; Slothuus,
2007; van Oorschot,
2000,
2006,
2008; Van Oorshot & Meuleman,
2014; van Oorschot et al.,
2017). In a nutshell, empirical research shows that the support for various social policies is conditional on the degree to which different groups are considered “really worthy” of social protection. While certain groups like the elderly, disabled people, or children from needy families are widely recognized as legitimate beneficiaries of welfare assistance, other groups like unemployed people and immigrants typically enjoy much less support (Jensen & Petersen,
2017; Larsen,
2006; Petersen,
2012; Petersen et al.,
2010; Sirovátka et al.,
2002; van Oorschot,
2006). It can be argued that poverty attributions have a direct conceptual link with deservingness through the “individual blame” response option. The depiction of the poor as “lazy” or “lacking willpower” implies that they could change their situation and that, therefore, they are potentially undeserving or illegitimate recipients of welfare benefits. In contrast, if society is mainly responsible for poverty or if the poor owe their condition to “bad luck”, it does
not follow that welfare recipients are undeserving—even though they might be judged undeserving for other, independent, reasons. Accordingly, there is strong evidence that poverty attributions and deservingness judgments are empirically related, though the nature of the relationship is unclear (e.g., Aarøe & Petersen,
2014; Appelbaum,
2001; Gilens,
1999; Hansen,
2019; Jensen & Petersen,
2017; Petersen,
2012; Skitka & Tetlock,
1993; Sniderman et al.,
1991).
In sum, although deservingness judgments and poverty attributions seem to have similar consequences for welfare policy preferences, we argue that they are not one and the same thing. As a more general concept, poverty attributions enable us to make broader predictions regarding policy preferences. At the same time, unlike deservingness judgments, they do not allow to focus on specific disadvantaged groups. Thus, our approach will be to model the effects of both poverty attributions and deservingness judgments on welfare policy preferences, and to estimate the residual effect of poverty attributions controlling for stereotypical and affective reactions toward specific welfare recipient groups.
A second possible covariate of poverty attributions which may have a confounding effect on welfare policy preferences is
trust. As argued above, support for redistributive policies hinges on trust relationships between taxpayers and welfare recipients, but it may also depend on how much these two groups trust
welfare institutions themselves. On the one hand, at least some of the individuals who hold society responsible for poverty may not rely on society for solving it either, and hence they may not be particularly supportive of social policy. On the other hand, many individuals will support social policies as long as their participation to the financing of welfare programs (through taxes and social security contributions) is perceived as fair and efficient. This requires, among other things, that other taxpayers contribute equally (no tax evasion; see Cerqueti et al.,
2019; Scholz,
1998), that welfare recipients do not abuse the system (e.g., Habibov et al.,
2017; Kumlin et al.,
2017; Mau,
2003; Roosma et al.,
2016), that the welfare system does not encourage idleness and dependency, thus reducing poverty rather than perpetuating it (Schmidtz & Goodin,
1998; Mau,
2003: 123–126; van Oorschot et al.,
2012), or that government is perceived as impartial, uncorrupted and competent (Edlund,
2006; Rothstein et al.,
2012; Svallfors,
2013).
A handful of empirical studies (each focusing on a particular subset of the arguments presented above) has examined whether and how welfare policy preferences depend on trust attitudes. In this research, two variables stand out:
trust in government (e.g., Edlund,
1999,
2006; Hetherington & Husser,
2011; Kuziemko et al.,
2015; Svallfors,
1999,
2002; Yamamura,
2014) and
generalized social trust, i.e., the belief that “most people” (in contrast to “particular others” one identifies with) can be trusted (e.g., Algan et al.,
2016; Bergh & Bjørnskov,
2014; Nannestad,
2008; Scholz,
1998; Sturgis & Smith,
2010; Uslaner,
2000; Warren,
2017). Overall, these studies suggest that higher levels of government and social trust are beneficial for welfare state support, even though the patterns of findings are not entirely consistent across national and time contexts (Svallfors,
1999,
2002). More importantly, however, it is likely that the effects of trust on welfare policy preferences are not completely distinct from the effects of poverty attributions—for example, if “social blame” explanations are premised on beliefs about the government’s inefficiency or anti-welfare bias, or if “individual blame” explanations are based on beliefs that most other people are untrustworthy. Hence, to disentangle the effect of poverty attributions and trust variables, both types of variables should be used simultaneously to predict welfare policy preferences.
To better delineate the effects of poverty attributions, a third check consists in considering the role of
political ideology. This variable has been shown to affect
both welfare policy preferences (e.g., Arts & Gelissen,
2001; Gonthier,
2017; Jacoby,
1994; Jaeger,
2006,
2008; Naumann,
2014; Wilson & Breusch,
2003) and poverty attributions (e.g., Furnham,
1982; Hunt,
2004; Hunt & Bullock,
2016; Pandey et al.,
1982; Weiner et al.,
2011; Zucker & Weiner,
1993). As it turns out, then, political ideology is an exogeneous variable that influences both the independent (endogenous) variable (i.e., poverty attributions) and the dependent variable (i.e., welfare policy preferences) in similar ways—for example, left-wing orientations tend to foster “social blame” explanations of poverty
and pro-welfare stances, which are themselves related (see Sect.
"Explanations for Poverty and Welfare Policy Preferences"). Therefore, part of the influence of political ideology might be unduly ascribed to poverty attributions if ideology is left out of the predictive model of welfare policy preferences. Our strategy will be to include ideology in our predictive model and thus to provide a rather conservative test of the effect of poverty attributions.
Besides, political ideology may be helpful to control, at least in part, for the unobservable effect of general orientations toward
individualism and collectivism (see Triandis,
1995). Both at the individual level and at the cultural/country level, individualism and collectivism have been shown to influence the propensity of individuals to make internal or external attributions in general (e.g., Carpenter,
2000; Morris & Peng,
1994; Oyserman et al.,
2002; Triandis et al.,
1988), but also more specifically in relation to poverty attributions (e.g., Bray & Schommer-Adkins,
2016; but see Nasser & Abouchedid,
2006). Interestingly, research has demonstrated that individualism and collectivism are linked in various ways to left–right ideology. For example, studies have suggested that individualism/collectivism and left–right positions are correlated (e.g., Radkiewicz,
2017), causally related (e.g., Bréchon,
2021), “aligned” (i.e., have similar effects on social policy preferences; e.g., Yoon,
2015) or interact in predicting welfare attitudes (e.g., Toikko & Rantanen,
2020). Thus, left–right positions may serve as a proxy for orientations toward individualism and collectivism. For example, it should control for the fact that right-wing individuals have a more individualistic profile (Bréchon,
2021) and are more likely to make individual attributions of poverty.
Finally, a fourth covariate of poverty attributions is the
personal experience of poverty. To be sure, people from underprivileged backgrounds may differ from their more affluent counterparts in how they explain the causes of their own misfortune, but also in their views of what social policies should be implemented to alleviate their problems. However, the intuitive expectations that poor people attribute poverty to social injustice (or at least reject the individual blame explanation) and demand more social protection from the state are not always borne out by empirical research.
1 In part, this is because “feeling poor” is a matter of both objective circumstances (e.g., struggling to “make ends meet”, being a welfare recipient) and subjective evaluations (e.g., feeling comparatively disadvantaged against peers). On the objective side, “those who have economic problems are less inclined to support the individualistic explanation than those who have never experienced financial problems” (Kallio & Niemelä,
2014: 123; see also Bullock,
1999; Lepianka,
2007; Morçöl,
1997; Niemelä,
2008). However, “the evaluation of one’s circumstances has a more pronounced influence on poverty attributions than the objective circumstances” (Lepianka,
2007: 91; see also Lepianka et al.,
2010; Nilson,
1981). The enhanced influence of subjective assessments has been analyzed, in particular, in the “relative deprivation” literature. Relative deprivation is usually defined as comparative financial resources, either in relation to what one would need to “make ends meet” or in relation to what
other people have (Kreidl,
2000: 157; see also Halleröd,
2006; Pedersen,
2004; Walker and Smith,
2002). Deprivation does predict attitudes toward inequalities and poverty; more specifically, it tends to be positively correlated with structuralist attributions and negatively correlated with individualistic attributions (Kreidl,
2000). Finally, “social stigma appears to be an important element within the experience of poverty” (Hirschl et al.,
2011: 366; see also Garthwaite,
2016; Patrick,
2014; Walker,
2014). In turn, feelings of shame are sometimes related to poverty attributions. At least in contexts characterized by strong resentment against “scroungers” (e.g., Britain), needy people who are shameful of their condition tend to dissociate themselves from “the poor” and to put the blame on them (Shildrick,
2018; Shildrick & MacDonald,
2013). Against this background, we will test the effect of objective and subjective measures of poverty in one of our robustness checks (see Table A6).