Disentangling the New Liberal Dilemma On the Relation between General Welfare Redistribution Preferences and Welfare Chauvinism

Transcription

1 Tim Reeskens Postdoctoral Research Fellow Wim van Oorschot Professor of Sociology Disentangling the New Liberal Dilemma On the Relation between General Welfare Redistribution Preferences and Welfare Chauvinism Abstract: In the present Age of Migration, public policy as well as social scientists envisage the New Liberal Dilemma (Newton, 2007), namely how to find popular support for welfare programs that were established in a less diverse era. In this paper, we would like to discover whether attitudes towards social rights for immigrants are rooted in general preferences towards welfare redistribution, and whether this association is moderated by the context individuals live in. Using the 2008 wave of the European Social Survey, we discover that individuals who prefer a redistribution based on merit are least chauvinist while those who find that the neediest are more entitled to benefits put the highest restrictions on welfare provisions for immigrants. Additionally, living in a diverse context strengthens this relation. In the face of the New Liberal Dilemma, one of the major ambitions for public policy is therefore to foster a sense of reciprocity without losing sight of general notions of altruism. Paper Prepared for Presentation at the General Conference of the European Consortium for Political Research Reykjavik, August 24-27, 2011

2 1. Introduction Politicians of European welfare states are continuously faced with what Kenneth Newton (2007) refers to as the New Liberal Dilemma : 1 in the present Age of Migration, reconciling the integration of immigrants with finding popular support for welfare programs that came in effect in times of cultural homogeneity is a difficult line to balance on. Especially in the present aftermath of the worldwide financial crisis, public opinion has further polarized on the issue of immigration with rising success for right-wing populist parties that fuelled debates on restricted welfare access to immigrants, not only in countries with established anti-immigration parties, like France, Belgium and Denmark, but also in established welfare states where populist parties recently gained widespread support, including the Netherlands, Sweden, and Finland. In the light of this New Liberal Dilemma, social scientists, too, have been puzzled with finding fertile soil in which the roots of popular support for immigrant social rights can flourish. Until now, research findings have been rather disappointing in this respect. Although immigrant flows to European countries seem to be immune to welfare state generosity (Hooghe et al, 2008), Boeri and colleagues (2002) have given evidence that immigrants rely more upon welfare provisions and are also perceived as more welfare state dependent. Despite the fact that citizens of foreign descent are more in need, they are nonetheless considered as far less deserving compared to native residents (van Oorschot, 2006; van Oorschot & Uunk, 2007). Furthermore, a recent interest in welfare chauvinism (Kitschelt, 1997) shows that the native population is not in favor of an unconditional immigrant access to social benefits and services (Mewes & Mau, 2013, Gorodzeisky & Semyonov, 2009). In this study, we would like to deepen the knowledge of welfare chauvinism by analyzing whether criteria people might apply before immigrants can be entitled to welfare provisions, like having citizenship or having worked and paid taxes, are rooted in general ideas about welfare state redistribution. In a long tradition, students of social solidarity (Deutsch, 1975; Miller, 1999; Arts & Gelissen, 2001; Aalberg, 2003) have identified three main principles of redistribution, namely equity (those who contribute most to the welfare state should be entitled to higher benefits), need (the neediest should be entitled to more welfare), and equality (everybody should be entitled to the same provisions). From a theoretical perspective, it is more than plausible to assume a direct link between redistribution preferences and criteria associated with immigrants social rights, as these criteria are themselves embedded in ideas about deservingness (van Oorschot, 2000). 2 1 The term liberal in this respect refers to the term widely used in normative theory, but is especially well addressed by the theory of liberal multiculturalism (Kymlicka, 2010), which defends the idea that in just societies, nation-states should reconcile minority group rights with liberal democratic values. 2 A different theoretical angle to this puzzle is conceiving welfare chauvinism as an example of the bi-dimensional values structure as proposed by Lipset (1959), i.e. an economic dimension (whether or not one favors egalitarianism) that cuts across a cultural dimension (the authoritarian perspective whether one preserves the welfare state to own people) (van der Waal et al, 2010). From an attitudinal perspective, the cultural libertarian-authoritarian dimension 1

3 Nevertheless, empirically, the link between general redistribution preferences and specific preferences towards immigrants access to social rights has not yet been identified. What is more is that especially the national context people live in might aid to comprehend this complex association between the general preference for redistribution principles and individual opinions about immigrants access to social welfare. In order to make heads and tails about the relation between general welfare redistribution preferences and specific stances towards social rights for immigrants in contemporary diverse welfare states, we will first of all review the literature and propose specific hypothesis that will guide the analysis. In the third section, then, the data and methodology to answer this research puzzle is proposed. In the fourth section, the results are presented and being discussed. Finally, we will conclude our paper with a thorough reflection on the implications of our finding for the future of social insurance schemes in the concluding fifth section. 2. Literature Review The relation between diversity and the welfare state is documented as rather tense (Alesina & Glaezer, 2004; James, 1993; Hero & Tolbert, 1996; Banting et al, 2006). Representing the state likewise the welfare state requires delineating who is in and out. Social theorists have frequently discussed the linkages between citizenship and welfare redistribution (Miller, 1999), as redistribution requires making sacrifices with anonymous others whom we do not know, will probably never meet, and whose ethnic descent, religion and way of life differs from our own (Kymlicka, 2001, p. 225). While the symbolic boundaries that delineated the anonymous others present within the national boundaries, often leading to social tensions about how welfare should be redistributed, once included social class and ideological discrepancies (Dalton, 2002) nowadays immigration with its encompassing distinction of the insider and the outsider dominates as a social cleavage that polarizes not only public opinion but also cuts across former social cleavages (Kriesi et al., 2006; Van der Brug & van Spanje, 2009). While the point at what European public opinion perceives immigrants as full citizens of the country fills many research agendas (Bail, 2008; Wright, 2011; Reeskens & Hooghe, 2010), a related approach regards the criteria citizens endorse towards newcomers before they can make appeal to welfare state provisions (van der Waal, 2010; Mau & Mewes, 2013). This idea of welfare chauvinism, i.e. the system of social protection for those who belong to the ethnically defined community and who have contributed to it (Kitschelt, 1997, p. 22), is rooted in the prejudice literature (Kitschelt, 1997; Gorodeizksy, 2009) whereas it penetrated the social policy literature has been frequently investigated (van der Waal et al, 2010; Mewes & Mau, 2013), while the economic equality-laissez faire dimension has hardly been given any attention. By bringing in general preferences about welfare redistribution, we hope to fill this gap in the literature as well. 2

4 only recently. Students of the welfare state have nevertheless already clarified that the general public regards immigrants as less entitled to social benefits and services (van Oorschot, 2006), although immigrants have cumulative risks that makes them more in need than native residents (Boeri et al., 2002). Being regarded as deserving depends upon several criteria, 3 and especially in the case of the social risk groups of immigrants, opposing criteria describe why public opinion is concerned about an unrestricted welfare access of immigrants. The need criterion posits that individual concern increases with others level of need. In this respect, there is a continuous stream of outcomes that shows that immigrants are among the most vulnerable socio-economic groups, with cumulative disadvantages in school and on the labor market (for an overview, see Heath et al., 2008). The internal causes of their deprived status are, according to Heath and colleagues (2008) shown to be largely attributed to socioeconomic background, but also to language proficiency, aspirations, and segregation. Additionally, as external cause why immigrants fare worse than natives, discrimination still seems to play an important role (Heath et al., 2008; Carlsson & Rooth, 2007). In sum, given their deprived socioeconomic status, immigrants as risk group are expected to be perceived as more deserving. Yet, there are several criteria in deservingness theory explaining why immigrants are nonetheless considered as less deserving. First of all, the criterion of control predicts that those who are able to exert control over their social risk are being regarded as less deserving (van Oorschot, 2000). Immigration, from this perspective, cannot be decoupled from the individual decision to move from country A to country B, often in response to labor market demands (Chiswick, 1999; Hooghe et al., 2008) but at present also in terms of family reunification, and humanitarian considerations (SOPEMI, 2010). In this respect, public opinion surveys show that tolerance is higher towards refugees, who fled their country to prevent prosecution, than towards general immigrants who have motives that might conflict with the economic interests of the native population (O Rourke & Sinnott, 2006). Additionally, the aforementioned internal causes for immigrants deprived socioeconomic status might be invoked as factors immigrants are able, to a certain extent, to control, i.e. follow language courses, enjoy training, and the like. In sum, there are several arguments that predict that immigrants are regarded as less deserving because they, crudely speaking, have a considerable control over the migration and integration process. Second, the deservingness criterion of identity explains that levels of deservingness depend upon the proximity with groups at risk. As inter-group research has clarified recently, hostile reactions towards immigration are mainly caused by the fear that national identity is under pressure (Sides & Citrin, 2007; Sniderman & Hagendoorn, 2007; Hainmueller & Hiscox, 2007). Furthermore, anti-immigration sentiments represent a hierarchy, with less hostile attitudes against those 3 An additional criterion that has been proposed is attitudes, i.e. people show higher levels of deservingness towards those groups who are grateful and compliance. As research into immigrant attitudes towards welfare state support is rather limited, we are not able to propose specific hypotheses in this respect. 3

5 economically, culturally and politically closer (Ford, 2011; Bogardus, 1925). In sum, immigrants are regarded as less deserving because they are in any respect more distant from the native population. Third, according to the reciprocity criterion, the level of deserving depends upon the extent to which you have earned your support (van Oorschot, 2000). Immigrants in this perspective lag behind, as they have an incomplete labor market trajectory in the national economy. While they may have contributed to the economy of their country of origin, their labor market trajectory in the country of destination is self-evidently incomplete. As a consequence, as they have not contributed substantially to the national wealth of the country of destination, native citizens might consider immigrants as less deserving. From a welfarist perspective, these deservingness principles applied to immigrants can implicitly be regarded as embedded in general ideas about how the welfare state should be organized. The first main principle of on the basis of which welfare can be redistributed is equity (Deutsch, 1975; Arts & Gelissen, 2001), according to which the right of high welfare provisions depends upon the individual contribution to the welfare state; working hard and paying taxes and related contributions will be rewarded equally in the case you are hit by a social risk. Consequently, equity can be related to the deservingness criteria of reciprocity, which, according to social scientists is in itself one of the driving forces in social solidarity (Konow, 2003): individuals give support and make contributions to the welfare state because the odds of encountering a social risks and relying on the welfare state are always far from zero. In a similar vain, it can be expected that those people who think that if you work hard you should get rewarded accordingly, will adhere to the color-blind opinion that if immigrants contribute to the national economy, they are entitled to social benefits and services. This leads thus to the following hypothesis. Hypothesis 1: A preference for the redistribution principle of equity is strongly associated with granting immigrants access to social rights after they have worked and paid taxes. A second organizational principle for the redistribution of welfare regards equality (Deutsch, 1975; Miller, 1999; Arts & Gelissen, 2001). According to this principle, everybody should be entitled to the same amount of benefits, apart from how much one has contributed or how needy one is. Nevertheless, the main issue put forward in this respect regards the description of everybody within the borders of any given society. According to social theorists, equality makes appeal to and is highly associated with citizenship-status, on the one hand because that when placed behind a veil of ignorance, an equal redistribution of resources should be preferred (Rawls, 1971), and alternatively that an equal redistribution prevents the creation of a secondary citizens class (Miller, 1999). Therefore, applied to deservingness criteria, the criterion of identity comes most closely to the redistribution principle of equality. As citizenship entails rights and duties, it aims at closing the distance between natives and immigrants, making that also immigrants are considered to be entitled to an equal share of the pool of welfare resources. 4

6 Linking the general principle of equality to the identity criterion, we can propose the following hypothesis. Hypothesis 2: A preference for the redistribution principle of equality is strongly associated with granting immigrants access to social rights after they have acquired citizenship. The last main principle of welfare redistribution regards need, according to which welfare mainly targets those that are worse off (Deutsch, 1975; Arts & Gelissen, 2001). Need is dominant in the means-tested liberal welfare state, insofar they target the deserving poor. The principle of need leads to two contradictory hypotheses. On the one hand, according to one strand in the literature, favoring a redistribution that supports the neediest should accordingly relate to the deservingness criterion of need from a perspective of enlightenment and altruism (d Anjou et al., 1995; Tyran & Sausgruber, 2006). This perspective allows thinking of immigrant social rights as being granted rather immediately, as immigrant groups are high on the hierarchy of groups with social needs. On the other hand is endorsing the need perspective also an expression of self-interest, as especially the underclass, which has a cumulative socioeconomic disadvantage, bears a higher risk to become welfare state dependent. Research has already pointed that feelings of chauvinist are most strongly endorsed by the lower social classes (Kitschelt, 1997; Gorodzeisky & Semyonov, 2009; Reeskens & van Oorschot, 2011). This perspective leads to the situation that an endorsement of the need principle goes together with high barriers for immigrants before they are entitled to welfare benefits, with the deservingness principle of control being mostly endorsed: immigrants cannot be considered as having access to welfare because they have taken the individual decision to move to another country. This elaboration leads to the following opposing hypotheses. Hypothesis 3a: A preference for the redistribution principle of equality is strongly associated with granting immigrants access to social rights as soon as they have entered the country. Hypothesis 3b: A preference for the redistribution principle of equality is strongly associated with never granting immigrants social rights under any condition whatsoever. In order to further qualify this complex relation between general preferences for welfare redistribution and specific preferences about social rights for immigrants, and precisely to grasp the complexity of the need -principle, i.e. whether the most solidarity stances do not travel across the own ethnic group, we can bring in evidence from realistic group conflict theory. This theory argues that intergroup hostility is the result of the salience of resource stress in the presence of a potentially competitive outgroup (Esses et al., 2001, p. 394). To be more concrete, this would mean that welfare chauvinism is higher in socioeconomic precarious contexts when confronted with a sizeable immigrant population. Although has not reached agreement on this association (Gorodzeisky & Semyonov, 2009; Mewes & Mau, 2013), a large immigrant group might and economic precariousness might moderate the relation between general preferences towards welfare redistribution and attitudes about social rights for immigrants. While the expectation is that the principle of equity should be rather immune in the light of diversity or economic stress 5

7 because equity underlines the importance of economic duties towards society, we expect that the relation between preferences of respectively equality and need, and welfare chauvinism operates through the national context. As aforementioned, as the equality mechanism reflects citizenship, the expectation is that in the face of diversity and socioeconomic deprivation, distinctions between the in- and out-group become more salient, leading to a negative association with a positive outlook on immigrant social rights. Of special interest is nevertheless the relation between need and welfare chauvinism in diverse societies. If the general preference for a redistribution based on need is an expression of enlightenment or general altruistic stances and therefore reflects the deservingness criterion of need, then the relation is expected not to be moderated by the national context. However, if need reflects self-interest and would correlate with an exclusionary stance, than the expectation is that in the light of a diverse context or in resource stress, individuals become more concerned about the in-group and are less likely to share their welfare with immigrants (Wright, 2011; Wright & Reeskens, 2010). Hypothesis 4: The relation between the preference for equality and welfare chauvinism is positively affected by diversity and resource stress. Hypothesis 5a: The relation between the preference for need and welfare chauvinism is not affected by diversity and resource stress. Hypothesis 5b: The relation between the preference for need and welfare chauvinism is positively affected by diversity and resource stress. 3. Data & Methods In order to respond the question how general attitudes about welfare redistribution influence preferences for welfare redistribution, we will analyze the 2008 wave of the European Social Survey (ESS). This comparative biennial survey project has been carried out in more than 25 countries, of which, due to data limitations, we include 23 in our analysis. The countries studied are: Belgium (BE), Bulgaria (BG), Cyprus (CY), Czech Republic (CZ), Denmark (DK), Finland (FI), France (FR), Germany (DE), Greece (GR), Hungary (HU), Ireland (IE), Latvia (LV), Netherlands (NL), Norway (NO), Poland (PL), Portugal (PT), Romania (RO), Slovakia (SK), Slovenia (SI), Spain (ES), Sweden (SE), Switzerland (CH), and United Kingdom (GB). After listwise deletion of respondents with missing information, we retain 39,413 respondents in 23 countries. Dependent variable The dependent variable that is analyzed in this paper concerns the extent to which individuals perceive immigrants as entitled to the same social rights and benefits as compared to the native population. In the ESS, this attitude of welfare chauvinism is measured by the question: When should immigrants obtain rights to social benefits/services? This question has been offered to respondents with five possible answer categories namely (1) Immediately on arrival, (2) After 6

8 living in [country] for a year, whether or not they have worked, (3) Only after they have worked and paid taxes for at least a year, (4) Once they have become a [country] citizen, and (5) They should never get the same rights. In order to increase the coherence between the response categories and the theoretical framework of deservingness, the first two response categories are combined, as they should reflect the enlightenment -perspective on need, the third category should responds to the reciprocity-dimension, while the fourth citizenship -item should reflect the notion of equality. Last but not least the fifth never get the same rights - response is expected to reflect the self-interest perspective of need. Independent variable In order to measure individual preferences towards welfare redistribution, i.e. whether one is more in favor of equity, equality or need, as independent variable we will analyze the survey question Some people say that higher earners should get more benefit when they are temporarily unemployed because they have paid more. Others say that lower earners should get more benefits because they are more in greater need. Which of the three statements on this card comes closest to your view? The answering categories were (1) higher earners should get more, (2) high and low earners should get the same amount, and (3) lower earners should get more. The first response category responds to the equity-principle, the second one to the equality-principle, and the third and last on to need. In our analysis, the equity higher earners should get more category will serve as reference. 4 At the national level, we have two main variables of interest, namely immigrant group size and unemployment rates. For immigrant group size, we use estimated 2008 figures which is obtained by a linear interpolation of the 2005 and 2010 share of immigrants data of the United Nations Population Division. Regarding socioeconomic resource stress, we use the 2008 rates of long-term unemployment, obtained from Eurostat. Control variables To assess the unique association between general welfare redistribution preferences and specific preferences towards immigrant social rights, we control for a number of variables that have previously shown to be related to welfare chauvinism, social solidarity, or attitudes towards 4 The European Social Surveys has offered its respondents also the question that relates to preferences regarding redistribution of old-age pensions. From a conceptual point of view, we prefer to report the results of the relation between unemployment benefits preferences and welfare chauvinism instead of similar analysis using the old-age pension question. First of all, conceptually, unemployment is an individual risk that individuals tend to assess as place behind a veil of ignorance, meaning that they tend to prefer redistribution on the basis of equality because unemployment is a risk that is less assessable than the risk of growing old, which is covered by old-age pensions. Second, concerning contemporary debates about the pressures of immigration on the welfare state, debates are often framed in terms of threats to the active population, i.e. that they take away (low-skilled) jobs but on the other hand also have too easy access to unemployment benefits and programs or alternatively social assistance. While these two reasons legitimate the choice of the unemployment benefits variable, the analysis has nevertheless also been done using the old-age pension variable. Although the results of the old-age pension variable, which are available upon request, are less significant, they are nevertheless in line with the results we present in this paper. 7

9 immigrants (Mewes & Mau, 2013; van der Waal et al. 2010; Gorodzeisky & Semyonov, 2009). The first variable is age, with the expectation that elderly place higher boundaries against immigrants. For gender, we expect that women are less chauvinist than men (reference). As is evident, we expect that people of foreign origin (having been born abroad or having at least one parent that is born abroad) are far in defense of immigrant social rights (reference), 5 just as the hypothesis is that people in metropolitan areas will also be less chauvinist compared to people in rural areas, as they are more in contact with immigrants and underclass residents. For socioeconomic status, we expect that a higher socioeconomic status, which is operationalized by various indicator variables, goes together with a more egalitarian outcome with regard to social rights for immigrants. Educational diploma is operationalized by lower (reference), lower secondary, higher secondary, and tertiary education. Work status is operationalized by having paid work (reference), being unemployed, being a student, being retired and having another status. We also look at whether one has been unemployed in the past or not (reference). One s current financial situation has been operationalized with one s subjective financial satisfaction in order to cope with the high nonresponse on the objective income question. Finally, we also look whether one is dependent on welfare or not (reference). As previous studies have also shown that welfare chauvinism is a reflection of cultural authoritarianism, we also control for it in the model. Last but not least, the expectation is that religiosity, operationalized by attending religious services, is associated with less chauvinist preferences. For more information about the control variables, check Appendix. Methodology The assumption that the welfare chauvinism is explained by simultaneously individual and country level covariates, and the interaction between both, requires the use of multilevel regression modeling (Hox, 2010; Gelman & Hill, 2006). This technique accounts for the clustered nature of the ESS data source individuals within countries and enables estimating nationallevel effects on individual outcomes. Since our dependent variable has the nominal measurement level, multilevel multinomial analysis is applied using the SAS Glimmix-procedure (Schabenberger, 2005). The reference category of the multinomial model will be the most chauvinist position that immigrants should never get the same rights. In the analysis section, we will proceed as follows. First of all, we will introduce the bivariate relation between general welfare redistribution preferences and specific attitudes about immigrant social rights in a first part. In a second step, we will add individual-level independent variables in a multilevel multinomial regression analysis in order to gauge the unique association between welfare redistribution preferences and welfare chauvinism. In a third and final part, we will 5 The question whether the inclusion of respondents of foreign origin into this paper is evidently legitimate. Prejudice research often excludes these respondents from the analysis as they are interested in the opinion of the native citizens. In this manuscript, however, we have nevertheless included them in the data and will, if necessary, document deviances of the respondents of foreign origin from the native sample. 8

10 analyze whether the relation between these general and specific immigration redistribution preferences are moderated by relevant national level covariates of diversity and economic deprivation. 4. Results 4.1. Bivariate Results Before going to more complex analyses, we turn to the distribution of our dependent variable. As Table 1 shows, there is a considerable variation in how people conceive of social security rights for newly arrived citizens, with a small faction that is chauvinist, saying that immigrants should never get the same rights as native citizens (7.47 percent). The second least preferred response is the opposition, namely a quasi-unconditional immigrant access to social rights (16.48 percent). People nevertheless are keen on imposing conditions to which immigrants should comply with before they can make access to welfare provisions. A large share of the European population (34.69 percent) considers immigrants entitled to social welfare after they have acquired citizenship. Nevertheless, the lion s share of the European public (41.36 percent) think that working and tax paying conditions should apply for having access to social rights. Thus, also for immigrants we find confirmation for the fact that reciprocity drives solidarity (Konow, 2003): although immigrants may not have a shared path with native citizens, making contributions to the economy is regarded as a prerequisite to enjoy full social rights. Of mere interest for our research question is however the bivariate distribution between how people in general think of welfare redistribution, namely in terms of equity, equality and need, and how it relates to placing high boundaries against immigrant social rights. According to Table 1, there is a statistically significant association (chisq = ; df = 8; p < 0.001; phi = 0.09) between general welfare redistribution principles and welfare chauvinism. Contrasting the general distribution of preferences for immigrant access to social rights, as discussed above, along the categories of general preferences for redistribution welfare redistribution principles, it seems that especially the respondents endorsing an equality position reflect the average distribution. People who think that higher earners should get more (equity) have a distribution of immigrant social rights that lean further away from welfare chauvinism and more towards tolerance, while people who think that lower earners should get more (need) seem to be more welfare chauvinist. 6 6 A cross-tabulation on the native-sample only did reveal that the distribution of the dependent variable leans more to chauvinism when excluding the immigrants; e.g percent chooses for the first (immediately or after one year) response, percent for the work-response, for the citizenship-category and 8.13 for the exclusionary standpoint. However, there don t seem to be large deviances between the categories of the independent variable: all cells seem to get affected by the trend more or less equally. 9

11 Table 1. Cross Tabulation between General Preferences of Welfare Redistribution and Specific Preferences for Redistribution towards Immigrants When should immigrants obtain rights to social benefits/services Higher or lower earners should get larger unemployment benefits Higher earners should get more - equity High and low same amount - equality Lower earners should get more - need Total Immediately or after one year After worked and paid taxes at least a year Once they have become a citizen They should never get the same rights Total N 13,401 21,571 4,441 39,413 Note: Entries represent column percentages of general preferences of welfare redistribution and specific preferences towards welfare chauvinism. In order to cope with the clustered nature of the data, the bivariate association has been tested in a multinomial multilevel model approach. In Table 2, we first of all see that the intercepts reflect the general distribution of Table 1: Europeans are in general of the opinion that reciprocity underlies immigrants access to social rights, followed by citizenship; in addition, a quasiunconditional access is preferred to a full exclusion of immigrants. Looking then at the parameters of the welfare redistribution preferences on preferences regarding immigrant social rights, the trend that has been displayed in Table 1 is confirmed: people endorsing the principle of need are, compared to people who prefer welfare redistribution on the basis of merit, less willing to grant immigrants rights when certain conditions apply. The largest discrepancy between the equity and need group is present on the preference of having worked and paid taxes (b = -0.75) over granting immigrants never social rights. When looking at whether the preference for equality over equity, it seems that equality leads to a lower willingness to grant immigrants social rights after they have worked and paid taxes for at least one year compared to those who favor redistribution on the basis of merit (b = -0.10). Thus, as expected, the equity-category rules out other categories when it comes to the condition of having worked and paid taxes. Nevertheless, in contrast to our expectations, there is no significant difference in the willingness to grant immigrant social rights after having acquired citizenship between those that endorse the equality principle and those who are in favor of equity. Especially the need -category stands out and reflect an outspoken chauvinist opinion. 10

12 Table 2. Bivariate Effect Parameters of Specific Preferences for Redistribution towards Immigrants Regressed on General Preferences of Welfare Redistribution When should immigrants obtain rights to social benefits/services? Rather immediate Worked/Taxes Citizenship Fixed Effects Param T-Value Param T-Value Param T-Value Intercept 1.03*** *** *** Redistribution (Ref: Equity) - Equality (same amount) - Need (lower earners) *** * -0.75*** *** Random Effects Param T-Value Param T-Value Param T-Value Covariance parameters 1.54*** *** ** 3.27 ICC 31.88% 20.15% 15.68% * p < 0.05; ** p < 0.01; *** p < Note: Entries represent the results of a multilevel multinomial regression model with they should never get the same rights as reference category. Thus, to summarize, first of all, accounting for country clustering reveals minor but nevertheless interesting discrepancies that were not present in the cross-tabulation. Individuals endorsing the equity and equality redistribution preference are rather similar in their attitudes, while those in favor of need are rather chauvinist. A second preliminary conclusion is that several hypotheses are refuted bivariately. For instance, while we expected that respondents who favor need would be more in favor of granting immigrants rights on arrival, the alternative hypothesis is accepted, namely that they are highly chauvinist. Additionally, while the expectation was that people endorsing equality would be more prone to give immigrants social rights after they have become a citizen, there is no observable difference in this compared to those in favor of equity. For equity our hypothesis is partially confirmed. On the one hand, we do see that people in favor of equity are more willing to grant immigrants social rights after they have worked and paid taxes. However, before we can draw strong conclusions on the basis of the bivariate relations, we first need to figure out whether these patterns are not spurious due to related respondent information Individual-Level Effects In Table 3, the results of a multinomial multilevel analysis are given, controlled for structural and ideational respondent information. Before discussing how the bivariate effects of preferences towards welfare redistribution shift when taking confounding respondent factors into account, it s first worth mentioning that all individual determinants are in a relation consistent with previous research (Mewes & Mau, 2013; van der Waal et al., 2010): preferences towards immigrants social rights are hardly related to age and gender, while people of foreign origin are, as is evident, highly more in favor of social rights for immigrants, just as people living in urban 11

13 areas are more open towards immigrant group rights. Controlling for these basic characteristics, socioeconomic status shows mixed patterns. Those with a higher diploma are also more open to grant social rights to immigrants, while regarding employment status only status differ from those that are employed, as they are more willing to grant immigrant access to social benefits and services. Also having been unemployed does not strongly differentiate opinions about welfare chauvinism. Also doing well financially leads to more inclusive views on immigrant social rights while being welfare dependent leads to, even though there are no differences in immediately granting social rights, less conditional stances when it comes to giving immigrants access to social security. Two ideational stances that in previous research have shown to be relevant for explaining attitudes towards redistribution show theoretically relevant patterns. On the one hand, those with a cultural conservative stance are also most chauvinist, on the other hand, frequent attendees of religious services are more in favor of immigrant rights. Coming then to the discussion of the effect parameters of welfare redistribution preferences on welfare chauvinism, we see that in general, the size and significance of the effects are reduced when controlling for related respondent information, making the effect of general welfare redistribution preferences on welfare chauvinism are partially spurious to other individual determinants. The pattern that appeared in previous bivariate analysis is nevertheless confirmed: people that have the general preference to redistribute their national wealth especially to the neediest are more chauvinist and consequently less likely to grant immigrants easily access to social benefits and services, which confirms Hypothesis 3b. For people preferring redistribution on the basis of equality, we do not find outspoken differences compared with those residents who prefer redistribution on the basis of equity. This also implies that respondents who endorse the equality principle are not more in favor of granting immigrants equal access to welfare after they have acquired citizenship, which means that Hypothesis 2 is refuted. Additionally, also Hypothesis 1, namely that equity is strongly associated with granting immigrants access to welfare when they have worked in the country, needs to be rejected, as there is no clear differentiation from the equality group. 12

14 Table 3. Multivariate Effect Parameters of Specific Preferences for Redistribution towards Immigrants Regressed on General Preferences of Welfare Redistribution When should immigrants obtain rights to social benefits/services? Rather immediate Worked/Taxes Citizenship Fixed Effects Param T-Value Param T-Value Param T-Value Intercept 2.97*** *** *** 7.43 Redistribution (Ref: Equity) - Equality (same amount) - Need (lower earners) *** *** *** Age Woman (Ref: Man) * Foreign origin (Ref: Native) 1.33*** *** *** 6.30 Level of urbanization 0.07*** ** Education (Ref: Lower) - Lower secondary - Higher secondary - Tertiary Work (Ref: Employed) - Unemployed - Student - Retired - Other 0.21* 0.38*** 0.99*** *** ** 0.49*** 0.92*** ** *** 0.52*** 0.92*** *** Ever unempl (Ref: No) Subjective income 0.19*** *** *** 6.51 Welf dependent (Ref: No) -0.17* *** *** Authoritarianism -0.88*** *** *** Religiosity 0.10*** *** *** 5.43 Random Effects Param T-Value Param T-Value Param T-Value Covariance parameters 1.24*** ** ** 3.23 ICC 27.37% 18.16% 12.96% * p < 0.05; ** p < 0.01; *** p < Note: Entries represent the results of a multilevel multinomial regression model with they should never get the same rights as reference category Cross-Level Interactions In a next step, we bring in the national context to determine how the theoretically informed national context information mediates the relation between general welfare attitudes and welfare chauvinism. In the first step (Model 1), the multivariate impact of the share of foreigners and total unemployment rates on welfare chauvinism is analyzed. The results confirm previous 13

15 studies using this chauvinism item, namely that the direct impact of the classic perpetrators of anti-immigration sentiments, namely immigrant group size and socioeconomic distress, is rather limited (Mewes & Mau, 2013). We can only observe that people prefer less that immigrants should be entitled to social rights after they have obtained citizenship over the reference that they should never have the same rights when they live in diverse societies or in societies with high levels of long-term unemployment. As especially the citizenship -category is affected by the share of immigrants and unemployment rates, one interpretation might indeed be that in those contexts people become more concerned about their national identity and the social boundaries of citizenship that describes them, an interpretation that is in line with recent evidence of symbolic threats as culprit for out-group hostility (Hainmuëller & Hiscox, 2007; Sides & Citrin, 2007). Table 4. Cross-Level Interaction of Immigrant Size and Unemployment on the Association between General Preferences of Welfare Redistribution and Specific Preferences for Redistribution towards Immigrants When should immigrants obtain rights to social benefits/services? Rather immediate Worked/Taxes Citizenship Fixed Effects Param T-Value Param T-Value Param T-Value Redistribution (Ref: Equity) - Equality (same amount) - Need (lower earners) *** *** *** Share of foreigners * LT unemployment rate *** Model 2 Param T-Value Param T-Value Param T-Value Redistribution (Ref: Equity) - Equality (same amount) - Need (lower earners) 0.39*** *** *** Share of foreigners LT unemployment rate *** Foreigners*Equality -0.04*** *** *** Foreigners*Need -0.05*** *** Model 3 Param T-Value Param T-Value Param T-Value Redistribution (Ref: Equity) - Equality (same amount) - Need (lower earners) ** *** *** Share of foreigners LT unemployment rate *** Unemployment*Equality Unemployment*Need * 0.03 * p < 0.05; ** p < 0.01; *** p < Note: Entries represent the results of three separate multilevel multinomial regression models, holding constant for all individual level controls of Table 3, with they should never get the same rights as reference category. 14

16 The mere interest of this paper is nonetheless whether the association between general preferences for welfare redistribution and welfare chauvinism is moderated by the national context of resource stress and the salience of an out-group. With regard to the moderating impact of the national context, we can be brief, as the analysis shows that total unemployment hardly affects the relation between welfare redistribution preferences and attitudes about social rights for immigrants. Concerning the moderating influence of ethnocultural diversity, we can see in Table 4 that the country s share of immigrants affects the relation between general welfare opinions and welfare chauvinism in a negative trend. Individuals endorsing the view that welfare should be redistributed in terms of equality or need over equity become less willing to redistribute their national wealth with immigrants in the face of increasing diversity. This trend seems to be consistent across all categories of views over welfare access to immigrants. Thus, when confronted with a salient out-group, people who in general tend to show higher levels of solidarity and stronger expressions of the economic equality nexus are also more prone to raise symbolic boundaries between insiders and outsiders, preserving welfare resources mainly to the native population, which confirms Hypothesis 4. Additionally, as we had two opposing hypothesis for how a diverse context influences the relation between general preferences for need and welfare chauvinism, we need to reject hypothesis 5a and accept hypothesis 5b: in the light of diversity, people who think that those in need should get higher benefits become more chauvinist in the light of a diverse context. 5. Conclusion In the face of the New Liberal Dilemma (Newton, 2007), in which governments are urged to maneuver between the challenge of managing unprecedented immigrant inflows and sustaining social horizontality, the question what defines individual preferences towards social rights for immigrants has become a hot topic among social scientists. In this manuscript, we have approached the conditions individuals might impose on immigrants before they can be entitled to welfare benefits from a deservingness perspective, and have associated these conditions with general economic left-right opinions about how welfare should be redistributed. As such, we have tried to give an answer to the question whether specific attitudes towards sharing national welfare with immigrants are rooted in general preferences towards welfare redistribution, and additionally whether this complex relation can be explained by the context, i.e. by bringing immigrant group size and socioeconomic deprivation in as explanation. The results of this study shed an additional light on the boundaries of social solidarity. First of all we find confirmation for the fact that reciprocity is the main motive for showing solidarity towards immigrants: if you pay your duties to society, ethnocultural origin is of no interest at all. Second, ceteris paribus, we can see that respondents who prefer equity and equality do not differ significantly in their opinions about welfare chauvinism, as they are in general quite outward 15

17 reaching towards immigrants. However, importantly is that those who are more in favor of a that gives the highest benefits to those in need are rather chauvinist and thus tend to be less willing to offer a slice of the welfare pie to immigrants. This finding disconfirms the expectation that the altruistic and enlightened idea of need travels towards everybody on the soil, but on the other hand confirmed the expectation that preferences for redistribution based on need are an expression of self-interest of the underclass, which conflicts highly with ethnocultural heterogeneity. Symbolic boundaries of us and them are more outspoken when a scarce pool of welfare resources is at stake among those who are the most vulnerable. Interestingly, the exclusionary tendency becomes even more outspoken, not in the light of socioeconomic stress, but in the face of a sizeable immigrant population. The negative association between being generous in redistribution the welfare to those that are at the bottom of society and sharing with immigrants operates most firmly in diverse societies. A first contribution of this paper to ongoing debates is that we shed an additional light to vivid discussions, mainly in social theory, about the tense relation between diversity and social solidarity. According to many, this relation should be tense, and also our results indicate that people who are more in favor of restoring socioeconomic imbalances in societies are least in favor of sharing welfare with immigrants, especially in diverse societies. The results of our studies clearly show that it reciprocity above altruism drives positive orientations towards immigrants social rights, especially faced with diversity. While the underlying causal mechanisms still need to be disentangled, it can be expected that respondents praising reciprocity might regard it as an obstacle for the productivity of society if immigrants are abstained of social welfare. Further research, predominantly of qualitative nature, should qualify this hypothesis. A second contribution of our paper regards important additions to present scholarly debates about the social consequences of increasing ethnocultural diversity. A number of recent studies have argued that theories of realistic group conflict have become obsolete in explaining outgroup hostility, as natives feel rather threatened for a loss of identity instead of that they are been target of competition on the labor market. This study has crossed socioeconomic resources with an identity perspective and gives additional weight to the symbolic threat perspective without loosing sight of the socioeconomic stress paradigm. It is precisely in the light of diversity that preferences for redistribution that target the underclass lead to stronger symbolic boundaries that define outsiders as less entitled to social welfare. In any case, as public policy makers envisage the new liberal dilemma with major concern, the policy implications of our work are nevertheless not straightforward. Faced with diversity, legitimacy towards welfare support towards immigrants is most prominent among respondents that express a general equity outlook on welfare redistribution. As equity in general reflects reciprocity you are entitled to what you have worked for throughout your life public policy should be concerned about how to foster generalized reciprocity among its citizens, as exactly 16