POPULAR PERCEPTIONS OF WELFARE STATE CONSEQUENCES. A multi-level, cross-national analysis of 25 European countries

Transcription

1 POPULAR PERCEPTIONS OF WELFARE STATE CONSEQUENCES A multi-level, cross-national analysis of European countries Wim van Oorschot Tilburg University, Tim Reeskens & Bart Meuleman University of Leuven Accepted for publication in Journal of European Social Policy

2 Page of Journal of European Social Policy Abstract: POPULAR PERCEPTIONS OF WELFARE STATE CONSEQUENCES A multi-level, cross-national analysis of European countries The societal effects of the welfare state are a perennial issue in the public debate. Critics accuse the welfare state of having unintended economic and moral consequences rather than producing its intended social goals. Popular perceptions of possible consequences of the welfare state are a crucial component of welfare state legitimacy, but have hardly received any scholarly attention. Using the 00 wave of the European Social Survey, we analyze how European citizens perceive the consequences of the welfare state, whether perceived positive consequences outweigh the negative ones, and to what extent consequence perceptions are determined by individual and country level factors. The conclusion is that the European public has a clearer eye for the positive social than for negative economic and moral consequences. Moreover, at the individual level these perceptions are mainly influenced by ideational factors, while they are affected by welfare state generosity at the country level. Interestingly, in more developed welfare states the public perceives the negative, as well as the positive consequences more strongly. Key words: welfare legitimacy, perceived welfare state consequences, European Social Survey, multi-level analysis, public opinion

3 Journal of European Social Policy Page of Introduction POPULAR PERCEPTIONS OF WELFARE STATE CONSEQUENCES A multi-level, cross-national analysis of European countries The societal effects of welfare benefits and services are a permanent issue of public debate (Gough, 00). Critics accuse the welfare state of doing more harm than good. Unintended economic side-effects would be that welfare provision increases labour costs, while income policies make labour markets too inflexible (Lindbeck, ). Neo-liberals emphasize unintended moral consequences by arguing that benefits sap the will to work and a sense of selfresponsibility (Murray, ). Also Christian-democrats and communitarians are worried that comprehensive social programs crowd out informal caring relations and social networks, fostering social isolation and self-centeredness (Etzioni, ). Remarkably, in public debate, the positive consequences of welfare provision, like that it prevents large scale poverty and social unrest, are often less heard of. Our interest is how European citizens view these issues that are persistently articulated in public debates. The legitimacy of European welfare states would be seriously affected if citizens would only have an eye for its negative unintended consequences without simultaneously having strong perceptions of the positive intended consequences. Especially in a situation, like the present aftermath of the 00 financial crisis, it is important to have an idea of how people balance the positive and negative consequences as they perceive them, and how this balance might be affected by welfare cutbacks. In the empirical literature on welfare state legitimacy, perceptions of welfare state consequences have hardly been an issue. Having knowledge of only four related studies, evidence is fragmentary. It has been shown that % of the Canadians (Gidengil et al., 00) and % of the British (Bryson, ) agree with the libertarian statement that the welfare state makes people less willing to look after themselves. Sihvo and Uusitalo (a) offer an explorative insight in the determinants of consequence perceptions where they show that older, richer, more rightist, and higher class Fins more strongly feel that social security and welfare services have made people passive and reduced their initiative. There is only one study, carried out in the Netherlands (van Oorschot, 00), that systematically questions individuals about a series of possible positive and negative welfare state consequences. This study found that perceptions of welfare state consequences crystallize in three dimensions, namely negative economic, negative moral and positive social consequences. About two-thirds of

4 Page of Journal of European Social Policy the Dutch outweigh the positive over negative consequences. This study also discovered that individuals consequences perceptions depend more on cultural beliefs, than on socio-economic position. However, the Dutch study mentions that its findings might be rather case specific. On the one hand because they regard findings from a relatively comprehensive and generous welfare state, which is known to have large and enduring overall popularity. It could be, for instance, that in less comprehensive welfare states, people are on average more sceptical about the positive social consequences and have a clearer eye for the negative ones. On the other hand, the Dutch data have been gathered in 00, a year in which the Dutch economy was healthy and unemployment was low. It could be, as the study suggests, that opinions are more sceptical in times of economic downturn. In this paper we take the field further by analysing cross-national data from the European Social Survey 00, which contains items measuring peoples perceptions of a number of possible welfare state consequences on the economy, individual morality and social life. These data allow us to analyse whether, and to what degree, the consequences perceptions of Europeans are indeed country specific. As the Dutch study suggests such opinions may be influenced not only by the personal, but also by national context characteristics. Social science research has indeed demonstrated that individual perceptions are influenced by the national context (Almond & Verba, ; Inglehart, ). Relevant context factors of the structural kind that will be considered relate to the comprehensiveness and generosity of welfare provisions, the present state of the economy, and the degree of unemployment, while cultural differences may relate to for instance the level of egalitarianism. By focusing explicitly on context effects our study responds to a gap being noted in the literature, which regards knowledge about the context specificity of welfare opinions (Sihvo & Uusitalo, b; Blomberg & Kroll, ). The questions we will address empirically are: () How do European people perceive the positive and negative consequences of the welfare state they live in? () What are the structural and cultural characteristics of individuals that determine their perceptions of consequences? () What, if any, are the national characteristics that influence people s consequences perceptions? Before giving an answer to these questions we will discuss our expectations about the individual and contextual influences on consequences perceptions, and we will explain our data and methodology.

5 Journal of European Social Policy Page of Individual and contextual influences on consequences perceptions Individual-level explanations We start from an assumption common in studies of welfare legitimacy that people s welfare perceptions are influenced by two types of individual factors, briefly referred to as interests and ideas (e.g. Kangas, ; Blekesaune & Quadagno, 00; Svallfors, 00). The interests-factor refers to structural characteristics of individuals indicating a degree of personal stake in welfare state provisions. Generally, it is assumed that people with a stronger interest in social protection have more positive attitudes towards the welfare state, its arrangements and outcomes (Kangas, ; Svallfors, 00). This perspective predicts that perceptions of negative welfare state consequences are less widespread among those who usually depend more on the welfare state, like women, older people, lower income groups, groups with more economic stress, people with lower educational level, employees in the public sector, the (previously) unemployed and welfare beneficiaries. Likewise, one could expect that these groups are more convinced of the positive social consequences. The ideas-factor refers to individuals welfare relevant values, attitudes and beliefs that complement interest variables. First and foremost, various welfare opinion studies have shown that people with a left-leaning political stance tend to be more positive on all aspects of the welfare state (Meier Jaeger, 00; Papadakis & Bean, ). In accordance with this, we expect that leftist people are less critical about possible negative consequences, while they perceive more positive social consequences (see also: Sihvo and Uusitalo, a; Van Oorschot 00). However, following Lipset s () study on class values, the left-right distinction has been unravelled into two mostly un-related ideological domains: one economic dimension covering favouring equality or inequality; and one cultural and moral dimension capturing preferences for individual freedom vs. authority (see Houtman, 00), where we assume that egalitarian people are less critical on economic consequences, and more positive on the social welfare consequences, while authoritarian people are assumed to be more critical about moral consequences. A second group of ideational factors consists of general opinions on the welfare state, its institutions and beneficiaries. Persons that prefer more state responsibility in welfare provisions ( range of government) and/or who prefer more spending of government in such fields ( degree of government, see Roller, ) are expected to endorse less critical views on possible moral and economic consequences of the welfare state, as well as stronger perceptions of positive social consequences. Likewise, we hypothesize that general trust in the country s political institutions spills over in positive evaluations of welfare state outcomes. Based on deservingness theory (van Oorschot 00), we hypothesize that those who endorse critical images about the moral character of benefit claimants will have a clearer eye for negative welfare state consequences. Additionally,

6 Page of Journal of European Social Policy persons with pessimist perceptions of their country s socio-economic situation and people who perceive a higher welfare burden (i.e. higher national rates of unemployment, of poverty and of disability) may be more concerned about negative economic consequences of their welfare state. A third group of ideational determinants regards religiously inspired stances. With a view on the Christian-democratic and communitarian critique, we expect religious people to perceive stronger negative moral consequences of welfare. Distinguishing further between denominations, one could expect different perceptions since it is argued that European welfare states have developed differently in Catholic and Protestant traditions (Manow, 00) and how anti-poverty policy programs are enacted (Kahl, 00). Therefore, scepticism about moral consequences of the welfare state is expected to be largest among Protestants because they would have more negative ideas about idleness and state intervention generally. Country-level explanations: institutions, structure and culture Our cross-national data allow going beyond individual level explanations by investigating whether perceptions of welfare consequences are affected by country characteristics. Many studies on social and political attitudes have clarified that they depend upon the social context (Almond & Verba, ; Inglehart, ). Since this is the first study to deal with perceptions of consequences of the welfare state, there is no available literature to fall back on. Nevertheless, following the arguments of Arts and Gelissen (00) regarding how preferences of welfare redistribution are influenced by context factors, similar causal reasoning can be used. The authors argue that the national context determines individual evaluations in two distinct ways. First of all, the nation-state and the debates that take place within it serve as a context in which individuals acquire a collective knowledge about the functions and functioning of their welfare state. Individuals are thus socialized in a specific national environment, and on the basis of the acquired knowledge they are able to make their inferences about the extent to which the welfare state produces intended positive or unintended negative consequences. Secondly, the national context also provides individuals with reference frames to which perceptions of consequences of the welfare state are expected to relate to. For instance, if people have the reference frame that the welfare state is performing well, e.g. by low poverty figures, then their perceptions of the welfare state are expected to be positive. The institutional characteristics of the welfare state are a first contextual factor that is considered to play a role in the socialization of citizens and to serve as reference frame In extended welfare states, the high share of the national income dedicated to welfare policies and the related high taxes and contributions are instances that can be expected to lead to stronger positive and negative perceptions. On the one hand, as countries allocate more of their wealth to welfare, individuals are on a perpetual basis made aware and may actually experience through their own use of benefits and services that combating poverty and inequality is more effective, which might

7 Journal of European Social Policy Page of increase the awareness of positive social consequences. On the other hand, simultaneously, in generous welfare states, individuals are also confronted with higher taxes as benefits tend to be higher and services more extended, providing in a basis for critical views on undesirable economic side-effects and the erosion of work ethic and personal responsibilities. Secondly, the expectation is that additional structural context factors serve as frames to which the consequences of the welfare state can be judge against. For various relationships, contradictory hypotheses can be formulated. In prosperous economies, concern for negative economic repercussions of the welfare state is expected to be less widespread because people might perceive a budgetary buffer against negative economic side-effects. On the other hand, it is possible that people in such countries are more fearful because they may experience a greater risk for losing their economic prosperity and employment. Also concerning poverty, there may be contradicting expectations. In countries with high poverty rates, people may feel that welfare spending stands in the way of necessary economic investment and development, and thus express more concern for the negative economic consequences of the welfare state. Or they may conversely see the welfare state as a most important and welcome solution against poverty and as a stimulator of economic consumption. As for the perceived social consequences, we assume that people living in countries that perform better on socioeconomic indicators, tend to be more aware of the positive social consequences welfare states can produce. Third, next to the institutional and structural context, also the national culture sets of values and moral beliefs that are, to a certain extent, shared in a country socializes residents. Additionally, shared values and believe also provide frames in ways described by the Thomas theorem: if men define situations as real, they are real in their consequences (Thomas & Thomas,, p., in Merton,, p. 0). Such dominant ideas, for instance about the perceived welfare burden, can serve as frames against which people can evaluate the positive and negative consequences of the welfare state. An additional theoretical argument for pursuing this cultural approach is provided by Huckfeldt (, p. ), who defined contextual effects as instances in which individual behaviour is affected by the presence of a societal property in a population regardless of whether the individual possesses the property in question. Our data allow us to explore the contextual effects of the aggregated cultural variables that we distinguished at the individual level. In more egalitarian countries, characterized with a prowelfare culture (in the sense of high support for a broad range and degree of the state taking up welfare responsibilities), there is the expectation that the positive social consequences of the welfare state will receive a stronger emphasis in the public debate. Therefore these positive intended consequences may be perceived to a larger degree by the general public, also by those citizens that not necessarily share these positions. Analogously, we expect individual perceptions of negative moral consequences to be more widespread in countries that are dominated by the neo-liberal or communitarian culture of strong authoritarian values, high religious involvement, more negative images of welfare recipients and high perceived welfare burden.

8 Page of Journal of European Social Policy Data and methods The data used to test these expectations are obtained from the 00 wave of the European Social Survey (ESS Round, 00). This comparative biennial survey project has been carried out in countries, of which we, for reasons of data availability, can include in our analysis, which adds up to, respondents. The countries studied are: Belgium (BE), Bulgaria (BG), Croatia (HR), Cyprus (CY), Czech Republic (CZ), Denmark (DK), Estonia (EE), Finland (FI), France (FR), Germany (DE), Greece (GR), Hungary (HU), Latvia (LV), Netherlands (NL), Norway (NO), Poland (PL), Portugal (PT), Romania (RO), Slovakia (SK), Slovenia (SI), Spain (ES), Sweden (SE), Switzerland (CH), Turkey (TR), and United Kingdom (UK). For each of these countries, a probability-based sample of at least,00 respondents aged years and older was interviewed. This selection of countries from all regions of Europe gives the most possible comprehensive picture of consequence perceptions among European people, and it additionally assures variation in the context variables. Dependent variables The dependent variables concern three strands of popular perceptions of consequences of the welfare state, namely economic, social and moral consequences. Respondents were confronted with eight Likert-items containing a variety of possible consequences. As the phrasing of the items makes clear (see Table ), the social dimension (D, D, D) reflects positive consequences of the welfare state while the economic (D, D) and moral (D, D, D) dimensions reflect negative consequences. Confirmatory factor analysis (CFA) on the pooled dataset showed that the eight items are sufficiently reliable and valid measurements of the three dimensions. Since it is known that certain respondents show the tendency of agreeing with all items using a balanced agree-disagree answer scale i.e. a form of measurement bias referred to as acquiescence bias (Billiet & McClendon, 000) an additional response style factor has been specified. Correcting for acquiescence bias is important in this study, since the group of respondents agreeing with all items produces artificial positive correlations between the positive social consequences on the one hand, and negative moral and economic consequences on the other. Additional multi-group CFA analyses evidenced that the measurements are sufficiently equivalent to allow cross-country comparisons (Steenkamp & Baumgartner, ). In the descriptive analyses, the sumscales are analyzed in order to respect the original response categories; in the multilevel analyses, the latent factor scores controlled for acquiescence bias are analyzed. TABLE ABOUT HERE TABLE ABOUT HERE

9 Journal of European Social Policy Page of Europeans in general are convinced of the positive social consequences of the welfare state: Between and % of the respondents (strongly) agree that the welfare state prevents poverty, creates equality and aids the combination between work and family life (see Table ). Perceptions of negative economic and moral consequences are less outspoken. Agreement with the items referring to negative consequences varies between and %. Thus, for Europeans the positive social aspects tend to outweigh negative economic and moral repercussions. This is a remarkable outcome, given that usually, the negative welfare state consequences dominate media portrayal and public debates. Table also shows a very strong positive correlation (0.) between the perceived negative economic and moral consequences, meaning that individuals who think that the welfare state is a burden for the economy and businesses also tend to think that it makes people lazy and less caring. As expected, we also see a negative association between positive social and negative economic and moral consequences. The fact that the correlation is weak, however, shows that having an eye for positive social consequences does not make one blind for possible negative side-effects. Independent individual-level variables The following structural covariates indicating possible interest in welfare provisions are included: age (in years), gender (male as reference), level of education (none or primary (reference); lower secondary; higher secondary; tertiary education). Employment status and the sector of employment are combined into a single variable with the following categories: private sector employee (reference); public sector employee; other employee (mainly self-employed); unemployed; student; retired; other. An additional dummy variable indicates whether one has previously been unemployed longer than three months. Financial situation is operationalized by the individual assessment whether one finds it difficult or comfortable to live on the present income. 0 A final structural covariate regards whether one s main source of household income is from social benefits. A second set of variables regards ideational dispositions. Several dispositions have been operationalized as scales consisting of several items. Endorsement of the principle of equality is measured by means three items gauging whether respondents consider income inequality as harmful for society. Authoritarianism is operationalized as a combination of preferences for authoritarian teaching styles and harsher sentences for those who break the law. Range of government indicates the preference for more state responsibility in various domains, such as health care, and ensuring a reasonable standard of living for the old and the unemployed. Political trust is a means scale capturing trust in parliament, politicians and political parties. The variable perceived welfare burden refers to respondents estimates of the omnipresence of various social problems, such as unemployment and poverty. Finally, negative attitudes towards

10 Page of Journal of European Social Policy claimants regards negative opinions about whether the unemployed are not eager to find a job, whether many who receive benefits are not entitled to, and whether employees often pretend they are sick to stay home. Exact question wordings and results of factor analysis results are given in Appendix Table. Other ideational variables are operationalized as single-item variables. Degree of government is measured by a single item capturing the preference for either tax decreases and more social spending, or tax increases and less generous social benefits. Political left-right stance has been categorized in four categories, namely left (reference), middle and right position, and a category for missing information to cope with the high item-nonresponse. Satisfaction with the national economy is ranges from 0 (extremely dissatisfied) to 0 (extremely satisfied). For religiosity, two indicators are used: religious service attendance (from : every day to : never) and religious denomination. The latter variable distinguishes between Roman Catholic (reference), Protestant, Eastern Orthodox, other denomination, and no denomination. Independent country-level variables The institutional indicator social expenditure per capita taken as well as the structural nationallevel variables are collected from Eurostat of the year of the ESS interviewing or closest (Eurostat, 00). The state of the national economy is represented by the GDP per capita (in PPS) as well as the real GDP per capita growth. Long-term unemployment rate is used as indicator of labour market malaise. For income inequality, the S0/S0-ratio representing the income of the highest income quintile over the income of the lowest income quintile is considered. Poverty is measured as the percentage of the population below 0 percent of the median equivalised income (after social benefits). As cultural context variables, relevant individual ideational variables in the ESS are aggregated to the national level, which provides us with information about the extent to which the population of a country () endorses the principle of equality, () is authoritarian, () has negative opinions about welfare state claimants, () perceives high levels of welfare burdens, () prefers a wide range of government intervention, () favours higher taxes to increase the welfare provisions (degree of government), and () attends religious services. The dominant religious denomination in the country has been taken from the Democracy Crossnational Data (Norris, 00), Methodology The assumption that individual welfare consequence perceptions are explained by individual and country level covariates requires the use of multilevel multiple regression analysis (Hox, 00). This technique accounts for the clustered nature of the ESS data source individuals within countries and enables estimating national-level effects on individual outcomes. To facilitate

11 Journal of European Social Policy Page 0 of comparisons between models, observations with missing information are listwise deleted from the multilevel analysis (they are included for the cross-national descriptive results), leaving us with,0 respondents sampled in countries. In the analysis section, we will proceed as follows. First, we will introduce the dependent variables in a descriptive section, presenting the univariate distributions across countries. Second, we will introduce individual-level determinants, distinguishing between structural and ideational covariates. Third, the impact of the country-level variables is estimated. In order to control for cross-country differences in sampling design, cases have been weighed by the ESS design weight.. Results.. Distribution of Perceived Welfare State Consequences across Countries Figure represents the distribution of the dependent variables across the ESS-countries. In of the countries, the average score on social consequences is above the scale midpoint (i.e. ), reflecting that perceptions of such positive consequences are relatively widespread. Perceptions of negative economic and moral consequences are less strong: averages above the midpoint are found in respectively and countries only. The positive assessment of the welfare state consequences is also reflected by the balance variable, which is the perceived social consequences score minus the average of negative moral and economic consequences scales. In all but three of the countries, perceptions of social consequences outweigh negative consequences. Large positive balance scores of. and up are found in all Nordic countries and the Netherlands, all characterized as extended welfare states, but surprisingly also in Cyprus, Greece and Estonia. The reason why these less developed welfare states show a relatively large positive balance is that their populations perceive substantially less moral consequences of welfare compared to more developed ones. Smaller positive balances (. or less) are found in the less developed welfare states of Central and Eastern Europe (Bulgaria, Czech Republic, Croatia, Poland, Romania, Slovenia). Also in France, where the three consequences scales score above the midpoint, the balance is positive but relatively small. In two countries, namely the UK and Slovakia, the negative consequences outweigh the positive ones slightly (with balance scores of -.0 and -.0 respectively). Finally, Hungary is a clear outlier, with its negative balance score of - scale points. This score is the result of relatively large majorities seeing negative economic and moral social consequences, combined with the EU-wide lowest perception of positive social consequences. The country rankings of perceived consequences give no clear reflection of welfare regimes (Esping-Andersen, 0) or families of nations (Castels, ). The only consistent pattern seems to be that the Nordic social-democratic welfare states tend to form a cluster with strong positive balance scores.

12 Page of Journal of European Social Policy FIGURE ABOUT HERE.. Individual-Level Determinants Since the most important contribution of our analysis relates to cross-national variation in consequences perceptions, the effects of individual-level determinants are discussed only briefly, as they largely confirm a previous the study on Dutch data (van Oorschot, 00). The effects of individual level factors were analyzed in two steps. In Model we modeled the structural characteristics of individuals and in Model we added the ideational variables (Table ). In our discussion we will concentrate on the Model findings, and refer to Model findings when necessary. Because the outlying position of Hungary could affect the results, this country is dropped in further analyses. Generally speaking, the structural individual characteristics do not have a strong (direct) influence on welfare opinions. This is surprising from an interest perspective. Nevertheless, several of the rather small effects are found to be statistically significant. Most of these effects are consistent with van Oorschot (00): Younger persons, the higher educated, employees in the public sector as well as persons receiving social benefits perceive less negative moral and economic welfare consequences. Males and those with a higher (subjective) income are more convinced of the positive social effects of the welfare state. Besides these similarities, a few contrasting findings are detected as well. Van Oorschot (00, p. ) reports that Dutch females are more critical about the economic repercussions of welfare, while no such gender-effect is detected in the pooled sample of Europeans here. Furthermore, we find that those employed in the other sector (mainly self-employed) perceive more negative moral and economic consequences than private sector employees. Van Oorschot (00), on the contrary, reports that the self-employed are not statistically different from private employees regarding economic and moral perceptions, but does find that this group has stronger perceptions of social welfare consequences. A comparison of the proportions of explained variance (R²) of Models and demonstrates that perceptions of welfare state consequences are much more strongly affected by peoples ideational stances than by their structural position in society, a finding similar to that in the Dutch study. Regarding the ideational variables, the similarities between our study and van Oorschot (00) are even more striking. Left-leaning respondents as well as individuals who prefer a larger degree and range of government intervention are more convinced of the negative moral and economic sideeffects of welfare. At the same time, these groups perceive less positive social consequences. Finally, no differences in welfare consequences are found between Catholics and Protestants. In our analysis, however, we do find that Eastern Orthodox believers (mostly Greek citizens) and those without religious denomination are less concerned about the impact of the welfare state on economy and morality. 0

13 Journal of European Social Policy Page of Compared to van Oorschot (00), our study includes a larger set of ideational variables. As expected, an egalitarian disposition (i.e. the economic aspect of left-right orientation) is found to weaken perceptions of moral and economic consequences, while authoritarianism (the cultural dimension of political orientation), perceived welfare burden and negative perceptions of benefit claimants have the exactly the opposite effect. Perceptions of positive social consequences are stimulated by endorsement of the principle of equality, high levels of political trust, satisfaction with the state of the economy, low levels of perceived welfare burden and positive attitudes toward benefit claimants. TABLE ABOUT HERE Summarizing, the overall picture of a person that sees positive social outcomes of the welfare state includes receiving benefits and being satisfied with ones income situation, having a leftish political stance, endorsing egalitarianism, trusting the political system, and supporting the welfare state more generally. More negative about the social consequences are those who perceive larger proportions of the population as being in a welfare dependent situation, and those who have more negative images of welfare claimants. The profiles of persons perceiving moral and economic consequences are very similar, which is not surprising giving the strong positive correlation (.) between these two dimensions. The general picture of a person who perceives the negative economic and moral effects of the welfare state more strongly includes having a rightist political stance, a non-egalitarian and authoritarian attitude, going to church more frequently, having a preference for small government, and a strong negative perception of the group of welfare claimants. Finally, it is remarkable that the same models offer a substantially better explanation for variation in perceived negative economic and moral consequences (R²s of % and %, respectively) than for the variation in perceived positive social consequences (R² = %). This difference reflects that perceptions of moral and economic consequences are strongly structured along ideological lines, while perceptions of positive social consequences are more evenly spread among the European populations... Country-Level Determinants Until now, we have paid no attention to the intraclass correlations. Nevertheless, an estimation of the empty model (without any control variables) has shown that about five, seven and percent of the variability in respectively perceived economic, moral, and social welfare state consequences can be attributed to the country level. When adding individual information, which corrects for compositional effects, these intra-class correlations drop to respectively four, six and

14 Page of Journal of European Social Policy six percent. This means that of the total variance in perceived negative economic consequences four percent can be attributed to country context, while this is six percent for the negative moral and positive social consequences. Thus, especially in the case of the perceived social consequences, compositional effects account for a large part of the differences between countries. Given that the multilevel models are estimated using a statistically speaking small number of countries, we have limited the number of country-level variables included simultaneously, using the following method. First, the country level effects are tested bivariately (i.e. under control of the individual level variables of Table, but without other country level variables included). In a second step we identified separately for each of the three sets country covariates (i.e. institutional, structural, and cultural) which variables have significant effects, under control of all country-level variables of the same set. In the final analytical step, the significant variables of each set are tested simultaneously (under control of the individual level variables), and models are trimmed to arrive at the most parsimonious explanation. Looking briefly at the main bivariate relations (see Appendix Table ) it shows that, as expected, perceptions of positive as well as negative consequences are more widespread in countries with higher social spending. Within the set of structural variables there are two general patterns: all three consequences are more strongly endorsed in countries that are wealthier in terms of GDP per capita. In addition, perceptions of the negative consequences are less strong in countries with a higher degree of income inequality. Regarding the cultural variables, a general interesting finding is that perceptions of negative economic and moral consequences are more strongly related to the hypothesized cultural variables compared to the perceived positive social consequences. Perceptions of negative consequences are on average less strong in countries with lower levels of egalitarianism, authoritarianism, welfare burden perception, religiosity, and in countries where there is less support for a broad range of welfare state responsibilities. Perceptions of positive social consequences are higher in countries where the general public refutes an authoritarian logic and where the population does not perceive a welfare burden, but on the contrary the opinion is widespread that taxes may be raised in order to ensure social welfare spending. However, as we explained earlier, since quite a few of the context variables are interrelated, model selection has been applied (Table ). Of the larger number of country variables that are bivariately related to perceived consequences, only a few turn out to have direct effects. To get an impression of the relative magnitude of the effects, the standardized parameters are also reported, as well as the explained variance at the country level. We can see that the retained models are able to explain about 0 percent of the variance between countries, with the best fit for the positive social consequences (. percent) and the worst for the negative economic consequences (. percent).

15 Journal of European Social Policy Page of Welfare state generosity positively affects individuals perceptions of negative economic and moral consequences, as well as their perceptions of positive social consequences. Apparently, in more generous welfare states with extended welfare policies but also elevated taxes and social contributions, there is a greater concern for undesirable economic side-effects of being a strain for the economy, as well as for the fact that work ethic and personal and social responsibilities are under pressure. At the same time, as generous welfare states are expected to be effective in combating poverty and inequality, social expenditure increases the general awareness of positive social welfare state consequences. The standardized effects show that the effect of welfare state generosity on the perceived social consequences is substantially larger than its effects on the negative economic and moral consequences. Thus, positive and negative perceptions are both stronger in countries with higher social expenditure, but positive perceptions relatively more than negative perceptions. TABLE ABOUT HERE We see that negative economic consequences are not only more strongly perceived in countries with higher social spending, but also in countries with higher long-term unemployment rates. In other words, in countries with a persistent labour market malaise, individuals more strongly perceive that the welfare state is harmful for the economy and local business. For the perceived negative moral consequences, only the aggregated preferred range of government exerts a direct effect, controlling for welfare state generosity. Thus, in a society where the general public is more in favour of more government intervention to ensure life support, individuals feel less that the welfare state makes people lazy and less caring for each other. Perceiving positive social consequences is related positively to the aggregated principle of equality, holding constant for social expenditure per capita. That is, in countries where the principle of equality is more popular, people are more inclined to see the welfare state as creating a more egalitarian society, keeping people out of poverty and enabling work and family life combination. These findings add to the substantial debate on the ecological levels in explaining social and political attitudes, as these effects emphasize the difference between the compositional and contextual influences of the culture on perceptions of welfare state consequences. For the negative effect of the preferred range of government on perceived negative moral consequences and the positive effect of the principle of equality on perceived social consequences, the story is additionally interesting since these variables show to have both at the individual and the country level a significant direct effect on the dependent variables. Thus, the effects do not only work for those individual who endorse these ideational stances, but also for those who refrain from these attitudes but are nevertheless surrounded by a culture that underscores these convictions.. Conclusion

16 Page of Journal of European Social Policy The current backlash of the 00 global financial crisis has faced European policy makers with both advantages and disadvantages of a generous social welfare system. While many European welfare states have prevented widespread poverty, this has come at the expense of increased public debts. Although in many, if not most, European countries, debates about the welfare state and the necessity of welfare reform are often negatively framed, suggesting that it hinders economic growth or makes people more dependent, lazy, and less caring, our cross-national analysis has shown that the European population on average outweighs the positive social consequences over the negative economic and moral ones. Only in three European countries Great Britain, Slovakia and especially Hungary public opinion perceives that the negative consequences prevail over the positive ones. Furthermore, we find a strong positive correlation between both negative perceptions, while perceptions of positive consequences stand alone. This means that negative and positive perceptions are not in a zero-sum relationship: people may, and many do, combine negative and positive perceptions at the same time. In our analysis of the determinants of perceptions of welfare state consequences we distinguished between individual and context level factors. The general picture of a person with strong perceptions of negative economic and moral consequences includes having a low educational level and no personal experience with being on welfare, having a rightist political stance, a nonegalitarian and authoritarian attitude, participating in religious services frequently, having a preference for small government, and a strong negative perception of the group of welfare claimants. This may be briefly summarized as the picture of a person with a conservative outlook on society. The determinants of positive perceptions are markedly different. The overall picture of a person that sees positive social outcomes of the welfare state includes receiving benefits and being satisfied with ones income situation, having a leftish political stance, being egalitarian, trusting government, and endorsing the welfare state more generally. However, more negative about the social consequences are those who perceive larger proportions of the population as being in a welfare dependent situation, and those who have more negative images of welfare claimants. People may see the first as a failure of welfare state performance, while the latter may tend to inhibit people to have an open eye for good performance. Additionally, contrary to expectation, consequence perceptions do not clearly reflect the interest that people may have in welfare arrangements, as users or as potential users. The structural variables, that may indicate such interests, only explain very little of the variation in the perceptions. We found that perceptions of welfare state consequences are more strongly influenced by people s ideas. Also, people generally seem to be more opinionated on the negative consequences of the welfare state, than on its positive consequences. This we deduce from the fact that the proportions of explained variance are substantially higher for the first, than for the latter.

17 Journal of European Social Policy Page of We need to be aware that the individual-level findings of our study are average effects based on a a large group of European countries. Nevertheless, as the Dutch study by van Oorschot (00) illustrates, country-specific deviations from the average effects reported here might exist. Future analysis on this topic might, in the first place, go into more detail whether the effects of individual level factors explaining perceptions of welfare state consequences are the same across all countries, and in the second place, in case these effects are different across countries, whether these differential effects can be explained by the welfare state people live in. However, even though the perceptions explicitly deal with respondents national welfare state, we found that the variation in consequence perceptions vary rather little across European countries. What is more is that a large part of the variation at the country level can be explained by composition effects. Nevertheless, in addition to composition effects, it showed that both the negative economic and moral, and positive social consequences are influenced positively by size of the welfare state that people are living in. In countries that have more elaborate social spending, the citizenry more strongly perceives that the welfare states creates a more egalitarian society, prevents poverty and facilitates reconciliation of work and family life. But at the same time it also becomes more sensitive to possible negative consequences However, the effect of welfare spending on perceptions of positive social consequences is larger than, and thus outweighs, the effects it has on the perceived negative consequences. As for policy relevance, our findings suggest that an increase in social spending would on balance favour an increase in positive perceptions of welfare state consequences over an increase in negative perceptions. On the other hand, spending retrenchment would increase the relative importance of the negative perceptions. Seen in this way, the conclusion can be that a higher spending welfare state promotes its social legitimacy by stimulating in people the idea that it is doing a good job, more than that it arouses their worries about its effect on the economy and morals. More generally, we feel that a major part of the social relevance of our study lies in the fact that it shows that citizens seem to have a more nuanced view on the consequences of the welfare state than policy makers, who have been pre-occupied with neo-liberal critiques on the welfare state in the past two decades and have given less weight to articulating the positive consequences. Admittedly, we do not have systematic and extensive empirical proof of this characterization of the public welfare debate in European countries; nevertheless, we are inclined to assume it is true since our findings regarding typical differences between the pair of negative consequences on the one hand and the positive consequences on the other would fit to it. First of all, compared to the positive perceptions, there is less country level variation in both negative perceptions, which may reflect that the neo-liberal paradigm has been dominating welfare reform debates in all European countries. Secondly, people are more strongly opinionated about the negative consequences, than they are about the positive consequences, which could be a result of higher issue salience of negative consequences in the public debate. Third, the perceptions of

18 Page of Journal of European Social Policy economic and moral perceptions are positively correlated, which can may be due to the fact that often in the critical debate about the welfare state the two criticisms are paired.

19 Journal of European Social Policy Page of References Almond, G. A., & Verba, S. (). The civic culture: Political attitudes and democracy in five nations. Princeton: Princeton University Press. Arts, W., & Gelissen, J. (00). Welfare states, solidarity and justice principles: Does the type really matter? Acta Sociologica, (), -. Atkinson, A. B. (). The welfare state and economic performance. London: London School of Economics. Barr, N. (). Economic theory and the welfare state: a survey and interpretation. Journal of Economic Literature, 0, - 0. Billiet, J. (000), & McClendon, M. J. (000). Modeling acquiescence in measurement models for two balanced sets of items. Structural Equation Modeling, (), 0-. Bird, E. (00). Does the welfare state induce risk-taking? Journal of Public Economics, 0(), -. Blekesaune, M., & Quadagno, J. (00). Public attitudes toward welfare state policies: a comparative analysis of nations. European Sociological Review, (), -. Blomberg, H., & Kroll, C. (). Do structural contexts matter?: Macro-sociological factors and popular attitudes towards public welfare services. Acta Sociologica, (), -. Bryson, C. (). Benefit claimants: villains or victims? In R. Jowell, J. Curtice, A. Park, L. Brook, K. Thomson & C. Bryson (Eds.), British Social Attitudes, the th report (pp. -). Aldershot: Ashgate. Castles, F. G. (). Families of nations: Patterns of public policy in Western democracies, Aldershot: Dartmouth Publishing. Esping-Andersen, G. (0). The three worlds of welfare capitalism. Princeton: Princeton University Press. Esser, I. (00). Why work? Comparative studies on welfare regimes and individuals' work orientation [Thesis]. Stockholm: Swedish Instititute for Social Research, Stockholm University. Etzioni, A. (). The spirit of community. London: Fontana Books. Eurostat (00). Eurostat Statistics Database [URL: page/portal/statistics/search_database]. Brussels: European Commission. Gidengil, E., Blais, A., Nadeau, R., & Nevitte, N. (00). Women to the left? Gender differences in political beliefs and policy preferences. In M. Tremblay (Ed.), Women and electoral politics in Canada (pp. 0-). New York: Oxford University Press.