INCOME INEQUALITY AND PARTICIPATION: A COMPARISON OF 24 EUROPEAN COUNTRIES Bram Lancee and Herman van de Werfhorst GINI DISCUSSION PAPER 6 JANUARY 2011 GROWING INEQUALITIES’ IMPACTS Acknowledgement The authors would like to thank the participants of the workshop ‘Inequality measurement’ in Amsterdam on october 22 and 23 2010 for their valuable comments. January 2011 © Bram Lancee, Herman van de Werfhorst, Amsterdam General contact: [email protected] Bibliographic Information Lancee, B., Werfhorst, H.G. van de (2011). Income Inequality and Participation: A Comparison of 24 European Countries. Amsterdam, AIAS, GINI Discussion Paper 6. Information may be quoted provided the source is stated accurately and clearly. Reproduction for own/internal use is permitted. This paper can be downloaded from our website www.gini-research.org. Income Inequality and Participation A Comparison of 24 European Countries Bram Lancee University of Amsterdam Amsterdam Centre for Inequality Studies Herman van de Werfhorst University of Amsterdam Amsterdam Centre for Inequality Studies 26 January 2011 DP 6 Bram Lancee, Herman van de Werfhorst Page ! 4 Income Inequality and Participation: A Comparison of 24 European Countries Table of content ABSTRACT ......................................................................................................................................................................7 1. INTRODUCTION .........................................................................................................................................................9 2. FORMS OF PARTICIPATION ...........................................................................................................................................11 3. THE RELATION BETWEEN INCOME INEQUALITY AND PARTICIPATION ............................................................................................13 4. DATA AND MEASUREMENT ..........................................................................................................................................17 4.1.Inequality as the mean distance to the median income ........................................................................................17 4.2.Independent variables .............................................................................................................................................21 4.3.Dependent variables ...............................................................................................................................................21 4.4.Control variables ......................................................................................................................................................22 5. RESULTS ...............................................................................................................................................................25 6. CONCLUSION ..........................................................................................................................................................33 REFERENCES..................................................................................................................................................................35 GINI DISCUSSION PAPERS ................................................................................................................................................39 INFORMATION ON THE GINI PROJECT.....................................................................................................................................41 Aims ...........................................................................................................................................................................41 Inequality Impacts and Analysis ....................................................................................................................................41 Support and Activities ...................................................................................................................................................41 Page ! 5 Bram Lancee, Herman van de Werfhorst Page ! 6 Income Inequality and Participation: A Comparison of 24 European Countries Abstract Previous research suggests that where inequality is high, participation is low. Two arguments are generally put !"#$%#&'("')*+,%-.'(/-0'1.&-.23'4-#0(5'-.)67%,-(8'&)+#)00)0'+%#(-9-+%(-".':)9%70)'+)"+,)'/%;)'&-;)#2-.2'0(%(70)0'%.&' therefore fewer opportunities to share common goals. Second, people may participate more in social and civic life when they have more resources to do so. However, up till now, these explanations have been lumped together in empirical analyses. Using EUSILC data for 24 European countries, we analyse how inequality in different parts of the income distribution is related to civic, cultural and social participation. Results indicate that a substantial part of the impact of inequality manifests itself through resources at the individual and societal level. However, -.&)+).&).('"!'#)0"7#9)05'-('-0'0(-,,'(/)'9%0)'(/%('/-2/)#'-.)67%,-(8'<%2.-1)0'(/)'#),%(-".0/-+':)($)).'-.9"<)'%.&' participation. This is in line with a view that interindividual processes explain why inequality diminishes partici pation. Page ! 7 Bram Lancee, Herman van de Werfhorst Page ! 8 Income Inequality and Participation: A Comparison of 24 European Countries 1. Introduction Studies on the impact of inequality on a wide range of outcomes are increasingly of a multilevel nature. Unde sirable outcomes such as crime, bad health, low social trust, and low levels of civic participation are seen as being dependent on individual (or household) income position and on the income inequality in a society (Uslaner and Brown 2005; Kawachi et al. 1997; Daly, Wilson and Vasdev 2001; Huisman and Oldehinkel 2009; Wilkinson and Pickett 2009; Rothstein and Uslaner 2005). One domain of life where inequality effects are found concerns par ticipation in various social, political and cultural activities. Previous research suggests that in countries, states or neighbourhoods where inequality is high, trust and civic participation are low (Uslaner and Brown 2005; Alesina %.&'=%'4)##%#%'>???@'A,-;)#'BCCCDE' One of the explanations is that inequality depresses participation because certain conditions for social in teraction are not met: people have competing statuses and therefore fewer opportunities to share common goals (Wilkinson and Pickett 2009). Moreover, with higher levels of inequality, differences between people are larger. As a consequence, people may feel powerless and they will opt out of civic and social engagement (Uslaner and Brown 2005). However, no previous study allows for the possibility of income and inequality to interact. This is likely to be the case. If inequality matters for participation, it can be expected that one’s individual position in the income distribution has different implications for civic engagement under different regimes of inequality. That is, to be poor in an unequal society may be different from being poor in an equal society. There is, however, another argument why participation differs across countries that is not accounted for in previous research on the participatory consequences of inequality. Apart from the sociological consequences of diverging statuses, people may participate more in social and civic life when they have more resources to do so. Essential in the resource (or ‘neomaterial’, Lynch et al 2000) explanation of inequality effects is that equal socie ties not only have a more equal income distribution, but also provide all kinds of services more equally to members of society. F#);-"70'0(7&-)0'&"'."('0)+%#%()'(/)0)'%#27<).(0')<+-#-9%,,8E'G"'1.&'"7('$/)(/)#'-.)67%,-(8'%!!)9(0'"7(9"<)0' through sociological processes or through the (individual or collective) resources that are available to individuals, we need to take account of resources more fully and more explicitly than we have thus far seen in the literature. Most of the existing literature has examined a society’s level of inequality using a single measurement, most "!().'(/)'H-.-I9")!19-).(E'J79/'%'<)%07#)'-.!"#<0'70'".'(/)',);),'"!'-.)67%,-(8'-.'%'0"9-)(8'%9#"00'(/)'9"<+,)()' income distribution. However, the literature on economic inequality has demonstrated that trends in inequality, Page ! 9 Bram Lancee, Herman van de Werfhorst and variations between countries, are very different in different parts of the income distribution (OECD 2008; Salverda, Nolan and Smeeding 2009). The rise in income inequality in the past 25 years has mostly been observed at the topend of the distribution. With regard to belowmedian inequality, the trends are much less pronounced. Also country variations are more pronounced in the topend of the income distribution. The literature on the consequences of inequality has, however, not examined inequality in different parts of the distribution. If inequality increases, the density of the tails of the income distribution becomes larger. Such varying distributions could affect macrolevel outcomes like the average participation in society, even if the effect of household income is invariant between countries. In addition to varying distributions, household income could have a differential effect across different distributions. ' G/)' 9".(#-:7(-".' "!' (/-0' +%+)#' -0' /).9)' (/#))!",&E' 4-#0(5' $)' 0(7&8' (/)' -<+%9(' "!' &-!!)#).(' ,);),0' "!' -. equality at the bottom and at the top of the distribution on participation. Second, we study whether the impact of household income on participation changes with different levels of inequality. Such a design allows us to disen tangle crossnationally varying income distributions from varying effects of income under different conditions of inequality. Third, we control for the availability of resources by including, besides individual resources, welfare state expenditure and its interaction with individual income. By doing so, we can disentangle the sociological mechanism from the resource explanation. Our empirical analyses are based on crossnational data from 24 mem ber states of the European Union (EUSILC), and we apply a multilevel framework to account for the clustering of people within countries. Page ! 10 Income Inequality and Participation: A Comparison of 24 European Countries 2. Forms of participation We identify three forms of participation: social, civic and cultural. Rather than to contribute to the conceptuali sation and theoretical development of the concept of participation, our aim in this paper is to analyze the effect of income and income inequality on these three domains of participation. However, to make clear how we conceive (/)0)'9".9)+(05'-.'(/-0'0)9(-".'$)':#-)K8'&)09#-:)'(/)'(/#))'!"#<0E Social participation can be described as the informal bonds between people, or as the extent to which people interact with their friends and family members (Spencer and Pahl 2006; Van der Meer, Scheepers and Grotenhuis >??CDE'A!().5'(/-0'-0'%,0"'9,%00-1)&'%0'-.!"#<%,'0"9-%,'9%+-(%,':)9%70)'-('#)!)#0'("'(/)'#)0"7#9)0'(/%('9%.':)'%99)00)&' through these relations (Lin 2001; Pichler and Wallace 2007). The main distinction with the other forms of partici pation that we identify is that it is based on informal, as opposed to formal or institutionalized ties. In the section measurement, we operationalize social participation as the frequency of interaction with friends and family. Civic participation usually refers to involvement in and membership of organizations and activities that consti tute ‘civil society’. Since it involves ties that are embedded in an formally constituted organizations or activities, it is also labelled formal social capital (Pichler and Wallace 2007). Often, civic participation is referred to as involve ment in voluntary associations (Putnam 2000; Hooghe and Stolle 2003; Curtis, Baer and Grabb 2001; Ruiter and L)'H#%%!'>??M@'J9/"!)#'%.&'4"7#9%&)IH"7#-.9/%0'>??B@'N.&)#0).5'O7#(-0'%.&'H#%::'>??M@'P-,0".'>???@'Q70-9R5' Wilson and Bynum Jr 2000). It includes participation in organizations and/or activities of neighbourhood asso ciations, charitable organizations, political parties or groups, professional associations, environmental or peace groups and alike. The relevance of cultural participation can be traced back to the work on cultural capital of Bourdieu (1993). Ganzeboom (1989) describes cultural participation as a way of processing information. When people engage in cultural activities they process information. Depending on the complexity of the activity and the ability of the person, a certain amount of information is processed. In terms of measurement, it refers to engagement in cultural activities, like attending cinema, going to live performances, or visits to cultural sites (see also section ‘data and measurement’). Page ! 11 Bram Lancee, Herman van de Werfhorst Page ! 12 Income Inequality and Participation: A Comparison of 24 European Countries 3. The relation between income inequality and participation According to Uslaner and Brown (2005: 876), “[t]here are strong reasons to believe that high levels of in equality depress civic participation”. Although not separated in their empirical analyses, they provide two argu ments why this is the case: “The direct effect of inequality on participation arises when inequality of resources leads people in lower economic brackets to refrain from participating, either because they have fewer resources or because they believe that getting involved will be fruitless because the system is stacked against them.” This 0)9(-".'&-09700)0'(/"0)'($"'%#27<).(0E'4-#0(5'$)'&-09700'$/%('$)'$-,,',%:),'(/)'S0"9-","2-9%,T'%#27<).(@'0)9".&5' we discuss the ‘resource’ argument. Scholars who study the “contact hypothesis” (Allport 1979) contend that there are certain conditions that help social interaction to instigate attitudes like cooperativeness: equal status between citizens, opportunities for personal acquaintance, and opportunities for people to share common goals. The sociological argument stipulates that income inequality interferes with these conditions, resulting in less contact and participation in civic life. Wilkinson and Pickett (2009: 5) argue that “the scale of income differences has a powerful effect on how we relate to each other”. According to Wilkinson and Pickett, with higher inequality there are more differences in status between individuals which results in status gaps. These gaps trigger status competition and this detriments a range of desirable outcomes, including participation. However, not only status competition triggers low trust. J"9-)(-)0'$-(/'/-2/',);),0'"!'-.)67%,-(8'%#)'<"#)'0(#%(-1)&5'#)07,(-.2'-.',%#2)#'&-!!)#).9)0':)($)).'+)"+,)E'L-!!)# ences between people in the social structure cause feelings of threat, anxiety and stress. Inequality promotes an air of bitterness and resentment, which manifest itself in less societal interaction. According to Uslaner and Brown (2005; Rothstein and Uslaner 2005), people will perceive that their views are not represented in the political sys tem. As a consequence, people may feel powerless and they will opt out of civic engagement. Oxendine (2009: MD'0(%()0'(/%('UV-W.'%.'%(<"0+/)#)'"!')9"."<-9'0(#%(-19%(-".5'(/)'+""#'$-,,'!)),'&)2#%&)&5'$-,,':)').;-"70'%.&'$-,,' continually covet the riches they lack”. The central point in the sociological ‘interindividual’ argument is hence that inequality depresses participation because certain conditions for social interaction are not met: people have diverging statuses, which results in status competition and feelings of anxiety. As a result, people have fewer op portunities to share common goals and for personal acquaintance. Page ! 13 Bram Lancee, Herman van de Werfhorst However, besides the sociological consequences of income differences, resources theory argues that it is the availability of resources that determines participation. According to the resources argument (also known as the ‘neomaterial’ theory, Lynch et al. 2000), resources are needed to achieve desirable outcomes. In the words of =8.9/')('%,E'X>???3'B>?>D3'UY.&)#'%'.)"I<%()#-%,'-.()#+#)(%(-".5'(/)')!!)9('"!'-.9"<)'-.)67%,-(8'".'/)%,(/'#)K)9(0' a combination of negative exposures and lack of resources held by individuals, along with systematic underinvest ment across a wide range of human, physical, health, and social infrastructure”. In this view, the effect of inequality is not the sociological consequence of interpersonal processes, but an indicator of the (un)availability of resources. Essential in the resource explanation of inequality effects is that equal societies not only have a more equal -.9"<)' &-0(#-:7(-".5' :7(' %,0"' +#";-&)' %,,' R-.&0' "!' 0)#;-9)0' <"#)' )67%,,8' ("' <)<:)#0' "!' 0"9-)(8E' 4%9-,-(%()&' :8' higher tax revenues, governments in relatively equal societies make it easier for the poor to participate in various &"<%-.0'"!',-!)E'4"#')*%<+,)5'-.'<"#)'2).)#"70'$),!%#)'0(%()0'(/)#)'<%8':)'<"#)'"++"#(7.-(-)0'("'#)9)-;)'07:0-&8' for setting up associations and there may be subsidy on subscriptions, tickets or entrance fees. Moreover, welfare states provide an opportunity structure that facilitates the development of civic engagement. Evidence that in more extensive welfare states participation is higher is however mixed. There are indications that there is some variation in participation according to the degree of social expenditure or welfare state regimes (Kääriäinen and Lehtonen >??M@'Z%.'A"#09/"('%.&'N#(0'>??[@'F-9/,)#'%.&'P%,,%9)'>??\D5'%,(/"72/'"(/)#'0(7&-)0'&"'."('1.&'%.8')!!)9(5'"#'-.' (/)'"++"0-()'&-#)9(-".'XZ%.'A"#09/"('%.&'4-.0;)).'>??C@'Z%.'&)#'Q))#5'J9/))+)#0'%.&'H#"()./7-0'>??CDE Based on these two arguments, one would expect that higher inequality results in less participation. Uslaner and Brown (2005: 870), although only including countrylevel variables, state that: “We know that the poor participate less, but a large gap between the rich and the poor may be just as important a depressor of partici +%(-".'%0'-.9"<)',);),0E]'N,)0-.%'%.&'=%'4)##%#%'X>???D'%++,8'<7,(-,);),'%.%,80)0'%.&'%.%,8^)'(/)'#),%(-".':)($)).' -.)67%,-(8'-.'0(%()0'-.'(/)'YJ'%.&'!"#<0'"!'+%#(-9-+%(-".5'$/-,)'(%R-.2'-.("'%99"7.('-.&-;-&7%,'-.9"<)E'G/)8'1.&' (/%('<"#)'-.)67%,-(8'X<)%07#)&'$-(/'(/)'H-.-'9")!19-).(D'#)07,(0'-.',)00'+%#(-9-+%(-".E'N,(/"72/'(/)0)'1.&-.20'#)!)#' to the US, there is no reason to expect this to be different in a crossnational comparison. We hence formulate our 1#0('/8+"(/)0-0'%99"#&-.2,83 H1: Income inequality is negatively associated with participation, even when controlling for individual income. Page ! 14 Income Inequality and Participation: A Comparison of 24 European Countries None of the mentioned studies allows for the possibility of income and inequality to interact. This is likely to be the case. If inequality matters for participation, it can be expected that one’s individual position in the income distribution matters as well. According to Major (2004), one of the consequences of inequality is that it alters what people feel they deserve, or are entitled to receive from social relationships. Therefore, under high inequality, the disadvantaged believe that they deserve lesser outcomes and, more importantly, behave accordingly. That is, to be poor in an unequal country is different from being poor in an equal country. Pichler and Wallace (2009) analyze for European countries the relation between class, inequality and formal and informal social capital. Pichler and P%,,%9)'X>??C3'_>>D'%007<)'(/%('U`$/)#)'0(#%(-19%(-".'-0'0(#".2)0(5'(/)#)'$-,,':)',)00')*().0-;)'.)($"#R05':)9%70)' these are likely to be limited to particular classes. Otherwise, the distance between classes is larger in more unequal 9"7.(#-)0]E'G/)8'-.&))&'1.&'(/%('$-(/'/-2/)#',);),0'"!'-.)67%,-(8'&-!!)#).9)0'%9#"00'9,%00'-.'()#<0'"!'0"9-%,'9%+-(%,' are larger. A*).&-.)'X>??C3'\D'!"970)0'".'%!K7).('9-(-^).0'-.'(/)'YJE'J/)'%#27)03'U,-;-.2'%<".2'-.)67%,-(8'0/"7,&'%,0"' fuel feelings of hostility by dampening opportunities for meaningful crossclass interaction”. Oxendine indeed 1.&0'(/%('$)%,(/8'9-(-^).0'$/"',-;)'%<".2')9"."<-9')67%,-(8')*/-:-('/-2/)#',);),0'"!'9#"00I97((-.2'2#"7+'-.;",;) ment, more diverse personal friendships, and a stronger support for helping the poor. Conversely, Uslaner and Brown (2005: 876877) state: “Where inequality is high, those people with fewer resources may feel powerless.” Similarly, Pickett et al. (2005: 670) refer to ‘”the psychological stress of life near the bottom of a steeply hierarchi cal society”. As a result, we expect different levels of participation for somebody with a high income in a society with a high level of inequality, then for somebody with the same income in a more equal society. We argue that in coun tries where inequality is higher, participation is more likely to be determined by income. Conversely, in societies where income is distributed in a more equal manner, income is less important. This is formulated in hypothesis two. H2: In countries where income inequality is larger, the association between income and participation is stronger. However, since most studies have thus far used aggregate data (Uslaner and Brown 2005; Wilkinson 2005; Kawachi et al. 1997; Wilkinson and Pickett 2009), it has been impossible to disentangle the explanations referring ("'#)0"7#9)0'"#'("'0"9-","2-9%,'+#"9)00)0E'G"'1.&'"7('$/)(/)#'(/)'0"9-","2-9%,'!%9("#0'#),%()&'("'-.)67%,-(8'<%(()#' for participation, one needs to cancel out resource explanations relating to individual resources or collective re sources by means of welfare state provision. Hence, to fully capture the impact of resources, one needs to include Page ! 15 Bram Lancee, Herman van de Werfhorst 0(%()I,);),'$),!%#)'+#";-0-".'-.'%&&-(-".'("'/"70)/",&'-.9"<)E'47#(/)#<"#)5'0-.9)'-.)67%,-(8'-0'#),%()&'("'2#"00' domestic product (GDP), national income is important to control for. Given that high GDPcountries may have higher levels of participation due to the availability of resources (Lynch et al. 2000), we need to cancel out these effects to judge whether inequality is really associated with participation. To the extent that (individual and collective) resources explain inequality effects, there should be no inequal ity effects on participation once we hold constant for household income, education, national income and welfare state expenditure. Thus, the effect as hypothesized in hypothesis 1 and 2 should disappear. If, on the other hand, (/)#)'-.&))&'-0'%'0"9-","2-9%,'+#"9)00'-.K7).9-.2'+%#(-9-+%(-".'(/%('-0'#""()&'-.'-.9"<)'&-!!)#).9)0':)($)).'+)"+,)5' hypothesis 1 and 2 should hold, even when controlling for the availability of individual and collective resources, and their interaction effect. Hence, the sociological theory of inequality effects predicts that H3: The effect of inequality, and the interaction effect of income and inequality on participation hold even when controlling for the availability of resources on the macro and micro level. Previous studies on inequality point out that trends in and variations between countries in inequality are dif ferent from the bottom and the top of the income distribution (OECD 2008). To date, however, studies on the 9".0)67).9)0'"!'-.)67%,-(8'/%;)'."('-<+,)<).()&'(/-0'1.&-.2E'P)'%#27)'(/%('(/)0)'&-!!)#).9)0'<%(()#'%.&'(/%('(/)' relation between inequality and participation is most pronounced for inequality at the bottom half of the income distribution. The sociological effect of diverging statuses is likely to be largest with respect to differences at the bottom. That is, poverty is more visible than wealth and it triggers stronger reactions. Inequality below the median income therefore results in fewer opportunities to cooperate and creates more stress than inequality above the me dian. This is formulated in hypothesis four: H4: The relation between inequality and participation is more pronounced for inequality in the bottom half of the income distribution than in the top half. Page ! 16 Income Inequality and Participation: A Comparison of 24 European Countries 4. Data and measurement The empirical analyses are carried out with data from the 2006 wave of the European Union Study on Income and Living Conditions (EUSILC). The EUSILC is provided by Eurostat and contains the 27 member states of (/)'a7#"+)%.'Y.-".5'+,70'J$-(^)#,%.&5'b9),%.&5'c"#$%8'%.&'G7#R)8E'4"#'<"#)'-.!"#<%(-".'".'(/)'aYIJb=O'&%(%5' see Eurostat (2008). In 2006, a module was included on social networks and participation, which makes it possible to measure social participation. The sample consists of all people in the age of 2565 years who were included in the module ‘social participation’. Because not al social participation indicators were measured in all countries, the sample consists of about 137,000 individuals in 24 countries1. Since individuals are nested in countries and we want to explain crosscountry variation we estimate random intercept models2 (Snijders and Bosker 1999). Our aim is to empirically test the hypotheses derived from sociological and neomaterial explanations of in equality effects. Although our approach allows for a much more detailed investigation of such explanations than previous research, it must be born in mind that the crosssectional nature of the data does prohibit causal statements relating to inequality effects. 4.1. Inequality as the mean distance to the median income We devise a novel method to measure income inequality at the societal level (based on the work of Checchi, Visser and Van de Werfhorst 2010). Usually, inequalities in different parts of the distribution are measured using S+I#%(-"0T'"!'(/)'&-!!)#).9)'-.')%#.-.20'"!'2#"7+0'%('&-!!)#).('+)#9).(-,)0'"!'(/)'-.9"<)'&-0(#-:7(-".E'4"#')*%<+,)5' the ratio between the 90th percentile (P90) and the median (the 50th percentile, P50) informs us on inequality at the upper part of the distribution, and the ratio between P50 and P10 on inequality at the bottom. These pratios, /"$);)#'70)!7,5'/%;)'($"'&-0%&;%.(%2)0'!"#'"7#'+7#+"0)0E'4-#0(5'+I#%(-"0'0/%#)'$-(/'<"0('"(/)#'<)%07#)<).(0'"!' -.)67%,-(8'X-.9,7&-.2'(/)'H-.-I9")!19-).(D'(/%('-('-0'%'(#7)'<%9#"I,);),';%#-%:,)'(/%('-0'."('%'&-#)9('%22#)2%(-".'"!' individual positions. Such macrolevel indicators make us believe that inequality is ‘out there’ affecting citizens, but it disguises that macrolevel inequality is ultimately a construct of a combination of individuals or households. J)9".&5'0-.9)'(/)'0-^)0'X&).0-(-)0D'"!'(/)'(%-,0'%#)'1*)&5'+I#%(-"0'(),,'70',-((,)'%:"7(';%#8-.2'&-0(#-:7(-".0E'FB?'%,$%80' B' >' 4"#'0"9-%,'+%#(-9-+%(-".5'.Bd>['0-.9)'%,0"'&%(%'-0'%;%-,%:,)'!"#'c"#$%8E'P)'%,0"')0(-<%()&'<"&),0'"<-((-.2'c"#$%85':7('#)07,(0'&-&'."(' differ substantially. N0'%'1#0('0()+5'%.')<+(8'<"&),'-0')0(-<%()&'("'9/)9R'$/)(/)#'(/)#)'-0';%#-%(-".'%('(/)'9"7.(#8',);),E'G/-0'-0'-.&))&'(/)'9%0)3'(/)'-.(#%I9,%00' correlation (ICC) for social participation is .11, for cultural participation .08 and for civic participation.16. Second, to make sure that the variation is not due to individual characteristics only, we estimated a composition model including individual characteristics. The ICC’s %#)'EBB5'E?C'%.&'EBe'#)0+)9(-;),8@'-.&-9%(-.2'(/%('(/)#)'-0'07!19-).(';%#-%(-".'("':)')*+,%-.)&':8'9"7.(#8',);),'9/%#%9()#-0(-90E'G/-#&5'$)' ()0()&'$/)(/)#'(/)'0,"+)'"!'-.&-;-&7%,'-.9"<)';%#-)0'0-2.-19%.(,8'%9#"00'9"7.(#-)0E'G/-0'$%0'(/)'9%0)'!"#'%,,'(/#))'!"#<0'"!'+%#(-9-+%(-".' (p<.001). Page ! 17 Bram Lancee, Herman van de Werfhorst has 10 percent of the distribution to the lefthand side, and P90 always has 10 percent on the righthand side. This is unfortunate, because with larger levels of inequality (i.e. higher P90/P50 or P50/P10 ratios) we cannot directly see that the sizes of the tails increase in density. As we aim to disentangle varying distributions from varying effects, we need a measurement that captures the relationship between measured inequality and the density of the tails. The measurement of inequality below and above median incomes that we propose solves these two is 07)0E'4-#0('"!'%,,5'(/)'<)%07#)'-0'%'&-#)9('%22#)2%(-".'"!'/"70)/",&'-.9"<)'+"0-(-".05'<)%07#)&'%0'(/)'&-0(%.9)'("' the median as a proportion of the median (below and above median separately). Measured this way, inequality is directly a construct that results from a combination of individual households. Second, for this measure we can directly see that larger inequalities (i.e. larger average distances to the median income) are matched to larger densities at the tails of the income distribution. Our study validates this measurement by comparing it to the Gini 9")!19-).('%.&'("'+I#%(-"05':)!"#)'$)'#),%()'-('("'+%#(-9-+%(-".E' ' G/)'Q)%.'L-0(%.9)'("'(/)'Q)&-%.'b.9"<)'XQLQbD'-0'%'<%9#"'-.&-9%("#'(/%('#)K)9(0'(/)'<)%.'S&-0(%.9)T' of a household income relative to the median household income in a respective country. To be able to compare between countries, this number is standardized by the median household income. The MDMI for a given country is calculated as follows (separately above and below the median, subscripts a and b): $! H # medH '! MDMI a , b " 1 / n * &! )! %! medH (! ! (1) Wher)3'f'd')67-;%,-^)&'&-0+"0%:,)'/"70)/",&'-.9"<)'X!"#')%9/'-.&-;-&7%,5'%.&'9"##)9()&'!"#'/"70)/",&'.".I #)0+".0)D5'<)&f'd'(/)'<)&-%.'"!'%'9"7.(#8T0'f5'.'d'(/)'.7<:)#'"!'-.&-;-&7%,0'-.'(/)'9"7.(#8g0%<+,)E'b.'%&&-(-".' to separating measures of inequality below and above the median, we calculate an overall measure of inequality averaging the distance to the median (either below or above) for all individuals. Mathematically there is a strong 0-<-,%#-(8':)($)).'(/)'QLQb'%.&'(/)'H-.-'9")!19-).(5'%0'(/)'H-.-'9")!19-).(':%0-9%,,8'#)+#)0).(0'%'9"<+%#-0".' of everyone with everyone else. Given that we set off each household income to the median, this is very similar. f"$);)#5'7.,-R)'(/)'H-.-'9")!19-).(5'(/)'QLQb'9%.':)'9%,97,%()&'".'($"'0-&)0'"!'(/)'<)&-%.'0)+%#%(),8E In table 1, the MDMI is shown for the countries in the EUSILC 2006. In the left panel, the MDMI con sists of a simple country mean of the mean distance to the median income, ignoring whether the distance is above or below the median income. In the other two columns, the MDMI is computed separately for incomes above and below the (national) median income. The interpretation is that the value of .95 in table 1 for Portugal implies that the mean distance for people above the median income is almost one median income. Hence, people in Por tugal with an income above the (Portuguese) median income on average almost earn twice the median income. Page ! 18 Income Inequality and Participation: A Comparison of 24 European Countries The value of the MDMI below the median of .34 means that people that have an income below the median, earn on average .66 times the median income (since the distance is on average .34*medH under the median income). Comparing these numbers with the UK, one can see that whereas inequality at the bottom of the distribution is the same for both countries, inequality at the top is larger in Portugal than in the UK. Table 1. Mean distances to the median income Country MDMI MDMI above the Median MDMI below the Median Median H Austria Belgium Cyprus Czech Republic Germany Denmark Estonia Spain FinIand France Greece Hungary Ireland Iceland Italy Lithuania Luxembourg Latvia Netherlands Poland Portugal Sweden Slovenia Slovakia United Kingdom .37 .44 .48 .38 .44 .31 .53 .51 .42 .42 .58 .50 .60 .38 .49 .58 .48 .75 .35 .50 .66 .32 .35 .43 .54 .47 .55 .65 .51 .58 .38 .71 .65 .55 .55 .79 .67 .84 .51 .63 .78 .64 1.06 .46 .67 .95 .38 .43 .57 .72 .27 .31 .29 .22 .29 .24 .31 .36 .28 .29 .36 .31 .30 .25 .34 .33 .33 .36 .24 .35 .34 .27 .25 .26 .34 17864.34 16404.24 13374.45 4415.24 16905.33 24434.81 3200.69 10675.86 18689.25 16241.17 9111.16 3694.24 17283.75 28716.67 14560 2361.9 28634.67 2093.29 18773 3093.35 6824 18119.68 9322.71 3174.24 18584.02 Mean .47 .62 .30 13005.85 G"'9"<+%#)'(/)'QLQb'$-(/'(/)'H-.-I9")!19-).(5'4-27#)'B%'%.&'B:'+#)0).('+,"(0'"!'(/)'S0-<+,)T'QLQb':8' (/)'H-.-I9")!19-).('%.&':8'(/)'FC?gB?'#%(-"'X:"(/'&)#-;)&'!#"<'aYIJb=O'>??M3). The MDMI is almost identical ("'(/)'H-.-I9")!19-).('%.&'9,"0)'("'(/)'FC?gB?'#%(-"E'b.'4-27#)'>%'%.&'>:5'(/)'QLQb'%:";)'%.&':),"$'(/)'<)&-%.' %#)'+,"(()&'%2%-.0('(/)'FC?g[?'%.&'(/)'F[?gB?E'4-.%,,85'G%:,)'>'+#";-&)0'(/)'9"##),%(-".0':)($)).'(/)'QLQb'%.&' the regular measures of inequality. We conclude from these analyses that our measure strongly resembles existing _' =-R)5'(/)'QLQb5'(/)'H-.-'9")!19-).('-0':%0)&'".')67-;%,-^)&'&-0+"0%:,)'/"70)/",&'-.9"<)'X9"##)9()&'!"#'/"70)/",&'.".I#)0+".0)DE'G/)' Gini that we calculate and the Gini as provided by Eurostat have a correlation of .95. Page ! 19 Bram Lancee, Herman van de Werfhorst measures of income inequality. Given the advantages of examining inequality measured as distance below and above the median, we continue our analyses with these separated indicators. Figure 1 a and 1b. MDMI by GINI and the MDMI by the p9010 ratio Figure 2a and 2b. MDMI above and below the median income by the P50/10 and by the P90/50 ratio Table 2. Bivariate correlations between country level variables MDMI MDMI above MDMI below GINI P50/10 P90/50 MDMI MDMI above the median MDMI below the median GINI P5010 P9050 1 .996 .7932 .9646 .8272 .9324 1 .7488 .9526 .7869 .9365 1 .8835 .9683 .7793 1 .8777 .8968 1 .8357 1 Note: (n=25; for all p< .001) Page ! 20 Income Inequality and Participation: A Comparison of 24 European Countries 4.2. Independent variables As a measure of income inequality we include the MDMI calculated below and above the median income. Our measure of individual income is the equivalized disposable household income of an individual, standardized by the median income of the respective country4. By standardizing income, it is better comparable across countries. Since the EUSILC is a household sample, there are people in the sample that belong to the same household (mostly partners, plus some children that live at home). Such a sampling design violates the assumption of inde pendence of errors and can therefore bias the estimates. To account for this, we randomly drew one household member from each household and estimated the models with this reduced sample5. 4.3. Dependent variables Social participation. Respondents are asked about the frequency of contact and the frequency of getting to gether with relatives and friends (four items on a six point scale, ranging from ‘does not have’ friends/family to S&%-,8T5'O#".:%9/T0'%,+/%'dE\?DE'G/)'!#)67).98'"!'2)((-.2'("2)(/)#'#)!)#0'("'/"$'"!().'(/)'#)0+".&).('707%,,8'S2)(0' together’ with friends/relatives during a usual year. The frequency of contact refers to the frequency with which the respondent is usually in contact with friends/family, during a usual year, by telephone, letter, fax, email, sms. Only friends and family outside the household of the respondent are considered. O-;-9'+%#(-9-+%(-".E'4"#'(/)'<)%07#)<).('"!'9-;-9'+%#(-9-+%(-".5'1;)'-()<0'%#)'70)&'(/%('%0R'!"#'+)"+,)T0'+%# (-9-+%(-".'-.'9)#(%-.'%9(-;-(-)0'&7#-.2'(/)',%0('($),;)'<".(/0'X8)0g."DE'4-#0(5'%.'-()<'%:"7('+%#(-9-+%(-".'-.'%9(-;-(-)0' of neighbourhood associations, environmental organisations, civil right groups, peace groups and alike. Second, and item about participation in voluntary activities. It is asked if the respondent, during the last twelve months, participated in the unpaid work of charitable organisations, groups or clubs. It includes unpaid charitable work for churches, religious groups and humanitarian organisations (attending holy masses or similar religious acts is not included). Third, an item about participation in activities of recreational groups or organisations. Respondents are asked if, during the last twelve months, they participated in recreational/leisure activities arranged by a club, association or similar. It can be sport groups, hobby associations, or leisure clubs. The fourth item refers to partici pation in activities of political parties, political associations or trade unions (Participating in formal strikes/dem ".0(#%(-".0'-0'."('-.9,7&)&DE'G/)'1!(/'-()<'#)!)#0'("'+%#(-9-+%(-".'-.'%9(-;-(-)0'#),%()&'("'+#"!)00-".%,'%00"9-%(-".0' h' [' b.&-;-&7%,'-.9"<)'d' $!H # medH '!, where H and medH are the same as in equation 1. )! &! (! %! medH P)'%,0"')0(-<%()&'%,,'<"&),0'$-(/'(/)'!7,,'0%<+,)E'i)07,(0'%#)'0-<-,%#E'f"$);)#5'%'!)$'-.()#%9(-".0'%++)%#'0-2.-19%.(E'J-.9)'$)'&"'."(' know whether this result is due to the clustering or due to the larger sample size, we decided to take the more conservative estimate. ! Page ! 21 Bram Lancee, Herman van de Werfhorst Xi)9)-;-.2'(#%-.-.2'"#2%.-0)&':8'079/'%00"9-%(-".'-0')*9,7&)&DE'4"#'%,,'1;)'-()<05'%(().&-.2'<))(-.20'9"..)9()&' $-(/'(/)0)'%9(-;-(-)0'%#)'-.9,7&)&E'4",,"$-.2'i7-()#'%.&'L)'H#%%!'X>??M@':7('0))'%,0"'O7#(-05'H#%::'%.&'j%)#'BCC>D5' these items are combined in a dummy variable, assigning the value one to respondents who participated in the last twelve months in one or more of the activities mentioned, and zero to respondents who did not participate in any of the activities. Cultural participation. Three items are used that ask for people’s participation in cultural activities during the ,%0('($),;)'<".(/0'X.".)5'BI_'(-<)05'hIM'(-<)05'\IB>'(-<)05'<"#)'(/%.'B>DE'4-#0(5'(/)'.7<:)#'"!';-0-(0'("'97,(7#%,' sites (historical monuments, museums, art galleries or archaeological sites). Second, the number of times going to live performances (any live performance, whether it was performed by professionals or amateurs. Live perfor mances include plays, concerts, operas, ballet and dance performances. Visits to live sport events are not included). Third, the number of times going to the cinema. Since (unlike with civic participation) the answering categories of these items contain multiple and ordinal categories referring to attendance frequency, the three items are summed %.&'(%R).'("2)(/)#'%0'%'09%,)'XO#".:%9/T0'%,+/%'dEMCDE 4.4. Control variables We control for the following individual sociodemographics: gender, age (and squared), marital status, educa tional attainment (with a slightly collapsed version of the ISCED scale), and labour market status (either full time, parttime, unemployed, or not on the labour market). Since differences in participation can be expected between rural and urban domiciles (Ziersch et al. 2009), we include the degree of urbanization of the domicile of the re spondent (thinly, moderately or densely populated area). On the country level, we include GDP per capita in pur 9/%0-.2'+"$)#'0(%.&%#&0'X&)#-;)&'!#"<'a7#"0(%(DE'4"#'(/)'<)%07#)<).('"!'0"9-%,')*+).&-(7#)5'$)'-.9,7&)'(/)'("(%,' amount of social expenditure as a percentage of the GDP (averaged for each country over the years 20022005), provided by Eurostat. Table 3 presents the descriptive statistics. Page ! 22 Income Inequality and Participation: A Comparison of 24 European Countries Table 3. Descriptive statistics independent variables Mean Individual Income MDMI above the median MDMI below the median GDP per capita Social expenditure (as % GDP) Age Female Working fulltime Working part-time Unemployed Not on the labour market Lower secondary education and below Upper secondary education Post-secondary and tertiary education Never married Married Separated/Divorced Widowed Densely populated Moderately populated Thinly populated 0.23 0.63 0.31 103.04 23.7 45.58 Percentage 54 58 1 6 26 27 43 3 24 59 12 5 43 23 34 Range Min -21.57 0.38 0.22 52 12.53 25 Max 141.35 1.06 0.36 272 31.13 65 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 SD 1.07 0.13 0.04 38.68 5.18 11.28 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Source: EUSILC 2006, Eurostat Figures 3a-c. The MDMI by the country average of social, civic and cultural participation Page ! 23 Bram Lancee, Herman van de Werfhorst Page ! 24 Income Inequality and Participation: A Comparison of 24 European Countries 5. Results b.'4-27#)0'_%I95'(/)'(/#))'!"#<0'"!'+%#(-9-+%(-".'%#)'+,"(()&'%2%-.0('(/)'";)#%,,'QLQbE'=-R)'$)'$"7,&')*+)9(' based on previous research, there is a negative association: higher levels of inequality correspond with lower levels of participation. However, these are just simple bivariate plots, not taking into account the variation in individual income and other variables, and not taking into account the availability of resources on the country level. In tables 4 to 6, multilevel models are estimated for each of the three forms of participation. Models 1 and >'"!'(%:,)'_'0/"$'(/%('-.)67%,-(8'&")0'."('/%;)'%'0-2.-19%.(')!!)9('".'0"9-%,'+%#(-9-+%(-".'X9".(#",,)&'!"#'0"9-%,')* penditure and GDP). Income does have a strong effect. Importantly, models 3 and 4 show that the effect of income -0'<%2.-1)&'-.'<"#)'7.)67%,'0"9-)(-)0E'f).9)5'-.'<"#)'7.)67%,'0"9-)(-)05'(/)'&-!!)#).9)':)($)).',"$I-.9"<)'%.&' highincome households is larger than in more equal societies. Put differently, macrolevel inequality does have negative repercussions on social participation, but primarily by enlarging differences in participation between income groups. This is true both for macrolevel inequality above the median and below the median. Models 5 and 6 add the interaction term between income and social expenditure, thereby controlling for the resource explanations relating to egalitarian services provision in equal societies. The interaction term has a signif -9%.(')!!)9('".'0"9-%,'+%#(-9-+%(-".E'J"9-%,')*+).&-(7#)'&-<-.-0/)0'(/)'-.K7).9)'"!'-.9"<)'".'+%#(-9-+%(-".5'$/-9/' is in line with the neomaterialist theory of inequality effects. Compared to models 3 and 4, there is a substantial #)&79(-".'"!'(/)'-.()#%9(-".')!!)9(':)($)).'-.9"<)'%.&'-.)67%,-(85':7('-('#)<%-.0'0-2.-19%.(E'J"5');).'(/"72/'+%#(' of the inequality effect on social participation runs through the availability of resources, it is still the case that the income effect on social participation is larger in more unequal societies. This supports the view that resources only explain part of the impact of inequality on social participation, which leaves room for alternative, interindividual, sociological explanations why undesirable outcomes are more often encountered in more unequal societies. Page ! 25 m1 Individual income MDMI above median MDMI below median MDMI above median*income MDMI below median*income Social expenditure*income Social expenditure GDP per capita Individual level control variables Female Age Age squared Never married Married Separated Widowed Divorced Densely populated Moderately populated Thinly populated Not on the labour market Working fulltime Working part-time Unemployed Lower secondary education and below Upper secondary education Post-secondary and tertiary Constant Log-likelihood + 2 uij + 2eij ! ! Intra-class correlation m2 m3 m4 m5 m6 b se b se b se b se b se b se .017*** .016 (.001) (.155) .017*** (.001) -.031*** -.009 (.004) (.155) -.062*** (.008) .046*** .004 (.008) (.156) .014 (.009) -.156 (.532) -.218 (.533) -.202 (.536) .074*** (.006) .030*** (.007) -.002*** .007 .000 (.000) (.005) (.000) .187*** -.002*** .007 .000 (.025) (.000) (.004) (.000) .056*** -.026*** .000*** -.032*** ref. -.007 .015*** -.008** ref. .011*** .010*** ref. .048*** .041*** -.019*** ref. .050*** .058*** 1.825*** -34695.1 .011 .096 .100 (.002) (.001) (.000) (.002) .056*** -.026*** .000*** -.032*** ref. -.006 .015*** -.008** ref. .011*** .011*** ref. .048*** .041*** -.019*** ref. .050*** .057*** 1.898*** -34677.0 .011 .096 .100 (.002) (.001) (.000) (.002) .253*** (.025) .006 .000 (.005) (.000) .006 .000 (.004) (.000) .006 .000 (.005) (.000) .006 .000 (.004) (.000) .056*** -.026*** .000*** -.032*** ref. -.007 .013*** -.008** ref. .010*** .008*** ref. .049*** .041*** -.020*** ref. .050*** .058*** 1.839*** -34842.7 .010 .096 .099 (.002) (.001) (.000) (.002) .056*** -.026*** .000*** -.032*** ref. -.007 .014*** -.008** ref. .010*** .008*** ref. .049*** .041*** -.020*** ref. .050*** .058*** 1.910*** -34842.7 .010 .096 .098 (.002) (.001) (.000) (.002) .056*** -.026*** .000*** -.032*** ref. -.007 .014*** -.008** ref. .011*** .009*** ref. .049*** .041*** -.020*** ref. .049*** .057*** 1.851*** -34758.6 .010 .096 .099 (.002) (.001) (.000) (.002) .056*** -.026*** .000*** -.032*** ref. -.007 .014*** -.009** ref. .011*** .009*** ref. .048*** .041*** -.020*** ref. .049*** .057*** 1.927*** -34790.1 .010 .096 .099 (.002) (.001) (.000) (.002) (.006) (.004) (.003) (.002) (.002) (.002) (.003) (.004) (.002) (.002) (.181) (.006) (.004) (.003) (.002) (.002) (.002) (.003) (.004) (.002) (.002) (.204) Source: EUSILC 2006, Eurostat; * p<0.05, ** p<0.01, *** p<0.001 (twotailed tests). (.006) (.004) (.003) (.002) (.002) (.002) (.003) (.004) (.002) (.002) (.182) (.006) (.004) (.003) (.002) (.002) (.002) (.003) (.004) (.002) (.002) (.204) (.006) (.004) (.003) (.002) (.002) (.002) (.003) (.004) (.002) (.002) (.183) (.006) (.004) (.003) (.002) (.002) (.002) (.003) (.004) (.002) (.002) (.206) Bram Lancee, Herman van de Werfhorst Page ! 26 Table 4. Multilevel regression predicting social participation; n1=25, n2= 140,873 Table 5. Multilevel regression predicting cultural participation; n1=24, n2= 138,950 m1 + 2 uij + 2eij ! ! Intra-class correlation m3 m4 m5 m6 b se b se b se b se b se b se .352*** -.840 (.006) (.554) .352*** (.006) (.050) .018 (.037) (1.874) (.051) (.555) (.059) -3.715* .504*** -.819 .047 (1.876) (.025) (.554) -.019 -3.423 .340*** -.846 (1.874) (.159) (.045) -3.701* 1.196*** -.075 -.004*** .041* .007*** (.001) (.017) (.002) 1.139*** -.002* .047** .007*** (.161) (.001) (.015) (.002) .275*** -.099*** .001*** .415*** ref. .075 .033 .195*** ref. -.287*** -.523*** ref. .398*** .376*** -.162*** ref. .805*** 1.958*** 5.034*** -293789.1 .134 4.015 .032 (.012) (.004) (.000) (.014) .274*** -.099*** .001*** .414*** ref. .078* .034 .191*** ref. -.287*** -.522*** ref. .398*** .377*** -.159*** ref. .798*** 1.951*** 5.471*** -293765.2 .129 4.013 .031 (.012) (.004) (.000) (.014) .040* .007*** (.017) (.002) .046** .007*** (.015) (.002) .040* .007*** (.017) (.002) .046** .007*** (.015) (.002) .275*** -.099*** .001*** .415*** ref. .075 .031 .194*** ref. -.288*** -.526*** ref. .399*** .377*** -.163*** ref. .803*** 1.958*** 5.087*** -293795.9 .134 4.015 .032 (.012) (.004) (.000) (.014) .275*** -.099*** .001*** .415*** ref. .074 .031 .194*** ref. -.288*** -.526*** ref. .399*** .377*** -.163*** ref. .803*** 1.958*** 5.415*** -293795.5 .129 4.015 .031 (.012) (.004) (.000) (.014) .275*** -.099*** .001*** .415*** ref. .075 .031 .194*** ref. -.288*** -.526*** ref. .400*** .377*** -.163*** ref. .803*** 1.958*** 5.090*** -293795.8 .134 4.015 .032 (.012) (.004) (.000) (.014) .274*** -.099*** .001*** .414*** ref. .078* .033 .191*** ref. -.288*** -.523*** ref. .399*** .377*** -.159*** ref. .798*** 1.951*** 5.496*** -293767.2 .129 4.013 .031 (.012) (.004) (.000) (.014) (.039) (.026) (.019) (.014) (.014) (.015) (.021) (.025) (.015) (.016) (.655) (.039) (.026) (.019) (.014) (.014) (.015) (.021) (.025) (.015) (.016) (.726) Source: EUSILC 2006, Eurostat; * p<0.05, ** p<0.01, *** p<0.001 (twotailed tests). (.039) (.026) (.019) (.015) (.014) (.015) (.021) (.025) (.015) (.016) (.655) (.039) (.026) (.019) (.014) (.014) (.015) (.021) (.025) (.015) (.016) (.725) (.039) (.026) (.019) (.015) (.014) (.015) (.021) (.025) (.015) (.016) (.655) (.039) (.026) (.019) (.015) (.014) (.015) (.021) (.025) (.015) (.016) (.725) Page ! 27 Income Inequality and Participation: A Comparison of 24 European Countries Individual income MDMI above median MDMI below median MDMI above median*income MDMI below median*income Social expenditure*income Social expenditure GDP per capita Individual level control variables Female Age Age squared Never married Married Separated Widowed Divorced Densely populated Moderately populated Thinly populated Not on the labour market Working fulltime Working part-time Unemployed Lower secondary education and below Upper secondary education Post-secondary and tertiary Constant Log-likelihood m2 m1 m3 m4 m5 m6 b se b se b se b se b se b se Individual income MDMI above median MDMI below median MDMI above median*income MDMI below median*income Social expenditure*income Social expenditure GDP per capita Individual level control variables Female Age Age squared Never married Married Separated Widowed Divorced Densely populated Moderately populated Thinly populated Not on the labour market Working fulltime Working part-time Unemployed Lower secondary education and below Upper secondary education Post-secondary and tertiary Constant Log-likelihood .196*** -1.891* (.008) (.822) .196*** (.008) .205*** -1.886* (.036) (.822) -.111 (.064) .465*** -1.835* (.068) (.826) .017 (.079) -9.402*** (2.485) -9.664*** (2.477) -9.635*** (2.482) -.013 (.051) -.148* (.059) -.007*** .022 .009*** (.002) (.025) (.003) .859*** -.004** .030 .010*** (.207) (.001) (.020) (.002) -.218*** .011* -.000 -.057*** ref. -.033 .039 -.040 ref. .109*** .213*** ref. .267*** .264*** -.222*** ref. .525*** 1.096*** -2.191* -76624.3 (.013) (.005) (.000) (.017) -.218*** .011* -.000 -.058*** ref. -.032 .040 -.047* ref. .109*** .212*** ref. .271*** .267*** -.221*** ref. .521*** 1.093*** -.721 -76615.9 (.013) (.005) (.000) (.017) + 2 uij .295 .082 Intra-class correlation ! m2 .969*** (.201) .019 .009*** (.025) (.003) .028 .010*** (.020) (.002) .019 .009*** (.025) (.003) .028 .010*** (.020) (.002) -.218*** .010* .000 -.057*** ref. -.032 .039 -.042 ref. .108*** .210*** ref. .268*** .265*** -.222*** ref. .522*** 1.093*** -2.072* -76634.4 (.013) (.005) (.000) (.017) -.218*** .010* .000 -.057*** ref. -.032 .039 -.042 ref. .108*** .210*** ref. .268*** .265*** -.222*** ref. .522*** 1.093*** -.735 -76631.1 (.013) (.005) (.000) (.017) -.218*** .010* .000 -.057*** ref. -.032 .039 -.041 ref. .108*** .210*** ref. .268*** .265*** -.222*** ref. .522*** 1.093*** -2.075* -76634.3 (.013) (.005) (.000) (.017) -.218*** .010* -.000 -.058*** ref. -.032 .039 -.047* ref. .108*** .210*** ref. .271*** .267*** -.221*** ref. .519*** 1.091*** -.661 -76619.8 (.013) (.005) (.000) (.017) (.045) (.032) (.023) (.017) (.016) (.018) (.024) (.032) (.018) (.019) (.973) .226 .064 (.045) (.032) (.023) (.017) (.016) (.018) (.024) (.032) (.018) (.019) (.967) .295 .082 Source: EUSILC 2006, Eurostat; * p<0.05, ** p<0.01, *** p<0.001 (twotailed tests). (.045) (.032) (.023) (.017) (.016) (.018) (.024) (.032) (.018) (.019) (.974) .22444 .064 (.045) (.032) (.023) (.017) (.016) (.018) (.024) (.032) (.018) (.019) (.964) .298 .083 (.045) (.032) (.023) (.017) (.016) (.018) (.024) (.032) (.018) (.019) (.978) .225 .064 (.045) (.032) (.023) (.017) (.016) (.018) (.024) (.032) (.018) (.019) (.966) Bram Lancee, Herman van de Werfhorst Page ! 28 Table 6. Multilevel logit regression predicting civic participation; n1=24, n2= 136,780 Income Inequality and Participation: A Comparison of 24 European Countries Table 5 shows the results on cultural participation. Household income has a strong effect on cultural partici pation, alongside other individual variables (in particular education). Although the main effects of macrolevel -.)67%,-(8'%#)'."('0-2.-19%.(5'(/)'&-#)9(-".'"!'(/)')!!)9(0'+"-.('("',"$)#',);),0'"!'97,(7#%,'+%#(-9-+%(-".'-.'<"#)'7. equal societies. Moreover, when omitting social expenditure, both inequality above and below the median appear 0-2.-19%.('X."('0/"$.'/)#)D5'-.&-9%(-.2'(/%('(/)'#),%(-".':)($)).'-.)67%,-(8'%.&'97,(7#%,'+%#(-9-+%(-".'-0'<%-.,8' due to the distribution of resources. Model 3 shows that relation between income and cultural participation is not affected by income inequality at the topend of the distribution. Model 4, however, indicates that cultural participa (-".'-0',)00'0(#%(-1)&':8'/"70)/",&'-.9"<)'-.'0"9-)(-)0'$-(/'%'<"#)')67%,'&-0(#-:7(-".'"!':),"$I<)&-%.')%#.-.20E'b.' other words, the positive effect of more equality below the median on cultural participation (positive main effect) is particularly found for lowincome households (negative interaction effect). Models 5 and 6, which control for the differential effect of income depending on the level of social expenditure, show that both interaction terms :)($)).'-.)67%,-(8'%.&'-.9"<)'%#)'0-2.-19%.(E'Q"&),'M'9".1#<0'(/)'1.&-.20'"!'<"&),'h5'<)%.-.2'(/%('(/)'/-2/)#' levels of cultural participation in egalitarian income distributions particularly concerns lowincome households. Importantly, this effect apparently holds after controlling for varying income effects across countries with different levels of social expenditure, suggesting that sociological explanations for inequality effects may not be too far off the mark. Table 6 shows the same models for civic participation, a concept that has often been used in discussions on the ‘decline of social capital’ in modern western societies (e.g. Putnam 2000; Curtis et al. 1992; Ruiter and De Graaf 2006). Here we see most clearly the negative impact of macrolevel inequality on participation. In model 1 it is clear that, besides individual resources (income, education), more inequality at the top is associated with a lower likelihood to be active in a voluntary organization. Model 2 shows the same for belowmedian inequality; societies with a relatively small average distance to the median income (thus: low level of inequality) have higher participation rates than countries with larger average distances between household income and the country median income. Model 3 shows that the depressing effect of abovemedian inequality on participation is invariant across income groups. Or, in other words, the association between income and civic participation is not dependent on the level of abovemedian income inequality. Model 4 shows something different. The positive effect of belowmedian equality on civic participation is stronger for lowincome households than for highincome households. Models 5 and 6 control for the varying impact of income across countries with different levels of social ex penditure. In model 5 the negative interaction effect illustrates that, surprisingly, in more unequal societies, the effect of household income gets smaller. Model 6 shows that the interaction effect of model 4 remains of similar Page ! 29 Bram Lancee, Herman van de Werfhorst 0-^)5'%.&'0-2.-19%.(E'G/705'(%R-.2'%99"7.('"!'(/)'0)#;-9)0'+#";-0-".'-.')2%,-(%#-%.'$),!%#)'0(%()05'-('-0'0(-,,'(/)'9%0)' that more unequal societies have larger variations across income groups in civic participation. ' G/)'#)07,(0'"!'<"&),0'_'%.&'h'%#)'2#%+/-9%,,8'&-0+,%8)&'-.'127#)0'hIME'G/)'127#)0'0/"$'(/)'%00"9-%(-".'X%.&' corresponding standard errors) between income on participation for the observed range of the moderator variables X!",,"$-.2'j#%<:"#5'O,%#R'%.&'H",&)#'>??MDE'b.'127#)'h%5'-('0/"$0'(/%('(/)')!!)9('"!'-.9"<)'".'0"9-%,'+%#(-9-+%(-".' steeply increases with higher levels of inequality above the country median income, and that this effect is positively %.&'0-2.-19%.('!"#'%,,',);),0'"!'-.)67%,-(8E'4-27#)'h:'+#)0).(0'(/)'0%<)'!"#'-.)67%,-(8':),"$'(/)'<)&-%.E'P/)#)'(/)' &-0(%.9)'("'(/)'<)&-%.'-0'0<%,,)0(5'(/)')!!)9('"!'-.9"<)'-0'."('0-2.-19%.(,8'#),%()&'("'0"9-%,'+%#(-9-+%(-".E'P-(/'/-2/)#' ,);),0'"!'-.)67%,-(8':),"$'(/)'<)&-%.5'(/)')!!)9('"!'-.9"<)'-0');)#',%#2)#E'b.'127#)'[5'(/-0'-0'&".)'!"#'97,(7#%,'+%#(-9- +%(-".3'(/)'%00"9-%(-".':)($)).'-.9"<)'%.&'97,(7#%,'+%#(-9-+%(-".'-0'+"0-(-;)'%.&'0-2.-19%.(5'%.&'0(#".2)#'!"#'/-2/)#' ,);),0'"!'-.)67%,-(8':),"$'(/)'<)&-%.E'4-27#)'M'0/"$0'/"$'".)'0(%.&%#&'&);-%(-".'-.9#)%0)'-.'-.9"<)'%!!)9(0'(/)' probability to participate in one of the civic associations across the observed range of inequality below the median in 9"<)'X$-(/'%,,'9";%#-%()0'0)('("')-(/)#'(/)'<)%.'"#'<)&-%.DE'G/)')!!)9('"!'-.9"<)'".'9-;-9'+%#(-9-+%(-".'-0'0-2.-19%.(,8' positive and larger for higher levels of inequality, although the slope levels off for the highest levels of inequality. Page ! 30 Income Inequality and Participation: A Comparison of 24 European Countries Figure 4a and 4b. The effect of income on social participation for observed values of MDMI above and below the median income Figure 5. The effect of income on cultural participation for observed values of MDMI below the median income Figure 6. The effect of income on civic participation for observed values of MDMI below the median income Page ! 31 Bram Lancee, Herman van de Werfhorst Page ! 32 Income Inequality and Participation: A Comparison of 24 European Countries 6. Conclusion In this paper, we examined whether and why income inequality in a society is negatively related to various forms of participation. We were particularly interested in examining different theoretical explanations for a rela tionship between inequality and participation (or other undesirable outcomes, for that matter). Thus far, studies have mostly relied on aggregate data that showed that higher levels of inequality are related to lower levels of social trust, bad health, higher crime rates, and lower levels of participation. However, this relationship can be explained by two different theories, which are impossible to test with ag 2#)2%()'&%(%E'4-#0(5'-.9"<)'-.)67%,-(8'9%.'%!!)9('(/)0)'"7(9"<)0'(/#"72/'0"9-","2-9%,'+#"9)00)0'"!'0(%(70'&-!!)# ences and related levels of stress (Wilkinson 2005; Wilkinson & Pickett 2009). In more unequal societies, the poor abstain from participation, show lower levels of generalized trust, develop bad health, and are more often involved in criminal activities, because they are more distressed by their disadvantaged position than they would have been if society would have been more egalitarian. Second, inequality is related to bad outcomes such as low levels of participation, because of the resources that people have available to them (Lynch et al. 2000). Such resources come in different forms, both at the individual and collective level. At the individual level, income and education provide relevant resources for a number of outcomes, including participation (Gesthuizen, Van der Meer and Scheepers 2008; Pichler and Wallace 2009). In more unequal societies, the poor have fewer resources, so their participation is negatively affected in comparison to lower income groups in more equal societies. At the country level, resources include GDP and the extent to which the state provides welfare to its citizens (Van Oorschot and Arts 2005). Both these aspects are related to the level of inequality in a country. The ‘neomaterial’ theory of inequality effects argues that more equal societies provide many services to its citizens that eliminate obstacles to participation and good health (Lynch et al. 2000). None of the existing studies have seriously disentangled these different mechanisms. We have done so by looking at resources in great detail, by including household income, education, GDP per capita, and social expend iture as a percentage of GDP. To the extent that these microlevel and macrolevel resources explain why inequality is related to participation, there should be no differential effect of household income on participation across dif ferent levels of income inequality. However, our results showed that, after controlling extensively for resources by including interaction terms between household income and social expenditure, income effects on participation are <%2.-1)&'-.'<"#)'7.)67%,'0"9-)(-)0E'="$I-.9"<)'/"70)/",&'<)<:)#0'+%#(-9-+%()',)00'-.';",7.(%#8'"#2%.-^%(-".0' and social life than highincome household members, and this difference gets larger in more unequal societies. Page ! 33 Bram Lancee, Herman van de Werfhorst G/-0'1.&-.2'-0'."('(/)'#)07,('"!',"$)#'-.9"<)'2#"7+0':)-.2'<"#)'%:,)'("'+%#(-9-+%()'(/#"72/')*().0-;)'$),!%#)' provision, higher GDP, or a relatively higher individual income. Moreover, under conditions of high inequality, all forms of participation that we identify are more affected by individual income. It is therefore not the case that when inequality is higher, the poor shift to forms of participation that require fewer resources, like contacts with friends. To further examine how interindividual processes explain part of the negative effects of inequality, it would be desirable to include information on how exactly income inequality translates into status competition and higher ,);),0'"!'0(#)00E'4"#')*%<+,)5'-('$%0'."('+"00-:,)'("'-.9,7&)';%#-%:,)0'(/%('9%+(7#)'(/)')*().('("'$/-9/'-.)67%,-(8'-0' perceived as such (Osberg and Smeeding 2006). To conclude, although resources are important to explain participation, there seems to be support for the theory that resources are not fully able to explain inequality effects on participation. In addition to the material or neo material theory that explains inequality effects in terms of resources, we found support for the idea that inequality affects people of different income levels differently through nonmaterial, interindividual processes. Page ! 34 Income Inequality and Participation: A Comparison of 24 European Countries References N,)0-.%5'N,:)#("5'%.&'a,-%.%'=%'4)##%#%E'>???E'UF%#(-9-+%(-".'-.'/)()#"2).)"70'9"<<7.-(-)0E]'The Quarterly Jour nal of Economics 115:847858. Allport, Gordon W. 1979. The nature of prejudice. Cambridge, MA: Perseus Books. N.&)#0).5'i":)#(5'k%<)0'aE'O7#(-05'%.&'a&$%#&'H#%::E'>??ME'UG#).&0'-.'O-;-9'N00"9-%(-".'N9(-;-(8'-.'4"7#'L) mocracies: The Special Case of Women in the United States.” American Sociological Review 71:376400. Bourdieu, Pierre. 1993. The Field of Cultural Production, Essays on Art and Literature. Cambridge: Polity Press. Brambor, Thomas, Wiliam .R. Clark, and Matt Golder. 2006. “Understanding interaction models: improving em pirical analyses.” Political Analysis 14:6382. Checchi, Daniele, Jelle Visser, and Herman Van de Werfhorst. 2010. “Inequality and Union Membership: The b.K7).9)'"!'i),%(-;)'a%#.-.20'%.&'b.)67%,-(8'N((-(7&)0E]'British Journal of Industrial Relations 48:84108. Curtis, James E., Douglas .E. Baer, and Edward Grabb. 2001. “Nations of Joiners: Explaining Voluntary Associa tion Membership in Democratic Societies.” American Sociological Review 66:783805. O7#(-05' k%<)0' aE5' a&$%#&' H#%::5' %.&' L"72,%0' EaE' j%)#E' BCC>E' UZ",7.(%#8'N00"9-%(-".' Q)<:)#0/-+' -.' 4-!()).' Countries: A Comparative Analysis.” American Sociological Review 57:139152. Daly, Martin, Margot Wilson, and Shawn Vasdev. 2001. “Income inequality and homicide rates in Canada and the United States.” Canadian Journal of Criminology, april:219236. Eurostat. 2008. EUSILC database description Luxembourg: Eurostat. Ganzeboom, Harry. 1989. Cultuurdeelname in Nederland. Een empirisch theoretisch onderzoek naar determi nanten van deelname aan culturele activiteiten. Assen/ Maastricht: Van Gorcum. Gesthuizen, Maurice, Tom Van der Meer, and Peer Scheepers. 2008. “Education and Dimensions of Social Capital: Do Educational Effects Differ due to Educational Expansion and Social Security Expenditure?” European Sociological Review 24:617632. Hooghe, Marc, and Dietlind Stolle. 2003. Generating social capital: civil society and institutions in comparative perspective. New York: Palgrave. f7-0<%.5'Q5'%.&'NEkE'A,&)/-.R),E'>??CE'Ub.9"<)'-.)67%,-(85'0"9-%,'9%+-(%,'%.&'0),!I-.K-9()&'-.l7#8'%.&';-",).9)I related mortality.” Journal of Epidemiology and Community Health 6:3137. Kääriäinen, Juha, and Heikki Lehtonen. 2006. “The variety of social capital in welfare state regimes a compara tive study of 21 countries.” European Societies 8:2757. Kawachi, Ichiro, Bruce P. Kennedy, Kimberly Lochner, and Deborah ProthrowStith. 1997. “Social capital, in come inequality, and mortality.” American Journal of Public Health 87:14911498. Lin, Nan. 2001. Social capital: a theory of social structures and action. Cambridge: Cambridge University Press. Lynch, John W., George D. Smith, George A. Kaplan, and James S. House. 2000. “Income inequality and mor tality: importance to health of individual income, psychosocial environment, or material conditions.” BMJ 320:12001204. Q%l"#5'j#).&%E'>??hE'U4#"<'0"9-%,'-.)67%,-(8'("'+)#0".%,').(-(,)<).(3'G/)'#",)'"!'0"9-%,'9"<+%#-0".05',)2-(-<%98' appraisals, and group membership.” Pp. 293355 in Advances in Experimental Social Psychology, edited by Mark P. Zanna. New York: Academic Press. Page ! 35 Bram Lancee, Herman van de Werfhorst Musick, Marc A., John Wilson, and William B. Bynum Jr. 2000. “Race and formal volunteering: The differential effects of class and religion.” Social Forces 78:5391570. OECD. 2008. Growing unequal? Income distribution and poverty in OECD countries. Paris: OECD. Oliver, J. Eric. 1999. “The effects of metropolitan economic segregation on local civic participation.” American Journal of Political Science 43:186212. A0:)#25'=%#05'%.&'G-<"(/8'QE'J<))&-.2E'>??ME'UTT4%-#TT'b.)67%,-(8m'N((-(7&)0'("$%#&'F%8'L-!!)#).(-%,03'G/)'Y.-( ed States in Comparative Perspective.” American Sociological Review 71:450473. Oxendine, Alina R. 2009. “Inequality and Indifference: America’s Wealthy and CrossCutting Civic Engagement.” in Annual Meeting of the Midwest Political Science Association,. F-9/,)#5'4,"#-%.5'%.&'O,%-#)'P%,,%9)E'>??\E'UF%(()#.0'"!'!"<%,'%.&'-.!"#<%,'0"9-%,'9%+-(%,'-.'a7#"+)E]'European Sociological Review 23:423435. nE'>??CE'UJ"9-%,'9%+-(%,'%.&'0"9-%,'9,%00'-.'a7#"+)3'(/)'#",)'"!'0"9-%,'.)($"#R0'-.'0(#%(-19%(-".E]'European Socio logical Review 25:319332. Pickett, Kate E., Shona Kelly, Eric Brunner, Tim Lobstein, and Richard G. Wilkinson. 2005. “Wider income gaps, wider waistbands? An ecological study of obesity an income inequality.” Journal of Epidemiology and Com munity Health 59:670674. Putnam, Robert D. 2000. Bowling alone. The collapse and revival of American community. New York: Simon and Schuster. Rothstein, Bo, and Eric M. Uslaner. 2005. “All for All: Equality, Corruption, and Social Trust.” World Politics 58:4172. Ruiter, Stijn, and Nan Drik De Graaf. 2006. “National Context, Religiosity, and Volunteering: Results from 53 Countries.” American Sociological Review 71:191210. Salverda, Wiemer, Brian Nolan, and Timothy M. Smeeding. 2009. Oxford Handbook of Economic Inequality. Oxford: OUP. J9/"!)#5'a;%.5'%.&'Q%#-".'4"7#9%&)IH"7#-.9/%0E'>??BE'UG/)'J(#79(7#%,'O".()*(0'"!'O-;-9'a.2%2)<).(3'Z",7.(%#8' Association Membership in Comparative Perspective.” American Sociological Review 66:806828. Snijders, Tom A.B., and Roel Bosker. 1999. Multilevel analysis. An introduction to basic and advanced multilevel modeling. London: Sage Publications. Spencer, Liz, and Ray Pahl. 2006. Rethinking Friendship. Hidden Solidarities Today. Princeton: Princeton Uni versity Press. Uslaner, Eric M. , and Mitchell Brown. 2005. “Inequality, trust, and civic engagement.” American Politics Re search 33:868894. Van der Meer, Tom, Peer Scheepers, and Manfred te Grotenhuis. 2009. “States as molders of informal relations? A multilevel test on social participation in 20 Western countries.” European Societies 11:233255. Van Oorschot, Wim, and Wil Arts. 2005. “The social capital of European welfare states: the crowding out hypoth esis revisited.” Journal of European Social Policy 15:526. Z%.'A"#09/"(5'P-<5'%.&'a,,).'4-.0;)).E'>??CE'UG/)'$),!%#)'0(%()'%.&'0"9-%,'9%+-(%,'-.)67%,-(8E]'European Socie ties 11:189210. Wilkinson, Richard G. 2005. The Impact of Inequality: How to Make Sick Societies Healthier. New York: The New Press. Page ! 36 Income Inequality and Participation: A Comparison of 24 European Countries Wilkinson, Richard G., and Kate E. Pickett. 2009. The spirit level. Why more equal societies almost always do better. London: Allen Lane. Wilson, John. 2000. “Volunteering.” Annual Review of Sociology 26:215240. o-)#09/5'N..%'QE5'4#%.'j%7<5'bEH70(-'c27#%/'L%#<%$%.5'N..)'QE'p%;%.%2/5'%.&'i):)99%'kE'j).(,)8E'>??CE' “Social capital and health in rural and urban communities in South Australia.” Australian and New Zealand Journal of Public Health 33:716. Page ! 37 Bram Lancee, Herman van de Werfhorst Page ! 38 Income Inequality and Participation: A Comparison of 24 European Countries GINI Discussion Papers Recent publications of GINI. They can be downloaded from the website www.gini-research.org under the subject Papers. DP 6 Income Inequality and Participation: A Comparison of 24 European Countries + Appendix Bram Lancee and Herman van de Werfhorst January 2011 DP 5 Household Joblessness and Its Impact on Poverty and Deprivation in Europe Marloes de Graaf-Zijl January 2011 DP 4 Inequality Decompositions - A Reconciliation Frank A. Cowell and Carlo V. Fiorio December 2010 DP 3 A New Dataset of Educational Inequality Elena Meschi and Francesco Scervini December 2010 DP 2 Coverage and adequacy of Minimum Income schemes in the European Union Francesco Figari, Tina Haux, Manos Matsaganis and Holly Sutherland November 2010 DP 1 Distributional Consequences of Labor Demand Adjustments to a Downturn. A Model-based Approach with Application to Germany 2008-09 Olivier Bargain, Herwig Immervoll, Andreas Peichl and Sebastian Siegloch September 2010 Page ! 39 Bram Lancee, Herman van de Werfhorst Page ! 40 Income Inequality and Participation: A Comparison of 24 European Countries Information on the GINI project Aims The core objective of GINI is to deliver important new answers to questions of great interest to European societies: What are the social, cultural and political impacts that increasing inequalities in income, wealth and education may have? For the answers, GINI combines an interdisciplinary analysis that draws on economics, sociology, political science and health studies, with improved methodologies, uniform measurement, wide country coverage, a clear policy dimension and broad dissemination. Methodologically, GINI aims to: ! exploit differences between and within 29 countries in inequality levels and trends for understanding the impacts and teasing out implications for policy and institutions, ! elaborate on the effects of both individual distributional positions and aggregate inequalities, and ! allow for feedback from impacts to inequality in a two-way causality approach. The project operates in a framework of policyoriented debate and international comparisons across all EU countries (except Cyprus and Malta), the USA, Japan, Canada and Australia. Inequality Impacts and Analysis Social impacts of inequality include educational access and achievement, individual employment opportunities and labour market behaviour, household joblessness, living standards and deprivation, family and household formation/breakdown, housing and intergenerational social mobility, individual health and life expectancy, and social cohesion versus polarisation. Underlying long-term trends, the economic cycle and "#$%&'(($)"%*)+)&,+-%+).%$&/)/0,&%&(,1,1%2,--%3$%,)&/(4/(+"$.5%6/-,",&/7&'-"'(+-%,04+&"1%,)8$1",9+"$.%+($:%;/% increasing income/educational inequalities widen cultural and political ‘distances’, alienating people from 4/-,",&1<%9-/3+-,1+",/)%+).%='(/4$+)%,)"$9(+",/)>%;/%"#$?%+@@$&"%,).,8,.'+-1A%4+(",&,4+",/)%+).%9$)$(+-%1/&,+-% trust? Is acceptance of inequality and policies of redistribution affected by inequality itself ? What effects ./%4/-,",&+-%1?1"$01%B&/+-,",/)1C2,))$(7"+D$17+--E%#+8$>%F,)+--?<%,"%@/&'1$1%/)%&/1"1%+).%3$)$*%"1%/@ %4/-,&,$1% -,0,",)9%,)&/0$%,)$G'+-,"?%+).%,"1%$@*%&,$)&?%@/(%0,",9+",)9%/"#$(%,)$G'+-,",$1%B#$+-"#<%#/'1,)9<%$.'&+",/)% and opportunity), and addresses the question what contributions policy making itself may have made to the growth of inequalities. Support and Activities The project receives EU research support to the amount of Euro 2.7 million. The work will result in four 0+,)%($4/("1%+).%+%*)+-%($4/("<%1/0$%HI%.,1&'11,/)%4+4$(1%+).%JK%&/')"(?%($4/("15%L#$%1"+("%/@ %"#$%4(/M$&"% ,1%N%F$3('+(?%JINI%@/(%+%"#($$7?$+(%4$(,/.5%;$"+,-$.%,)@/(0+",/)%&+)%3$%@/').%/)%"#$%2$31,"$5 www.gini-research.org Page ! 41 Amsterdam Institute for Advanced labour Studies University of Amsterdam Plantage Muidergracht 12 1018 TV Amsterdam The Netherlands Tel +31 20 525 4199 Fax +31 20 525 4301 [email protected] www.gini-research.org Project funded under the SocioEconomic sciences and Humanities theme.
© Copyright 2026 Paperzz