TARKI European Social Report 2009 - Attitudes to Inequality

8/3/2019 TARKI European Social Report 2009 - Attitudes to Inequality

1/14

TRKI EUROPEAN SOCIAL REPORT2009

17ATTITUDES TO INEQUALITy

ATTITUDES TO INEQUALITy1.OrsOlyalelkes


2/14


18 ATTITUDES TO INEQUALITy

Introduction1.1.

Empirical studies highlight the fact that there is wide social support throughout thecountries of Europe for a reduction in inequalit. Using the European Social Surve(ESS) and the World Values Surve (WVS) (which cover 21 countries), we show thatthere is ver little relationship between support for reduction in inequalit and meas-ured income inequalit across countries, so we explore the theoretical and empiricalreasons for this.

Social consensus on inequalit reduction1.2.There is wide social consensus among people all over Europe that inequalit shouldbe reduced. An overall majorit of the population in the EU countries examined here(except Denmark) regards inequalit as too large and believes that governmentshould reduce differences in income levels. Support for greater redistribution isstrongest in the Eastern European countries, including Slovakia, Estonia, Slovenia, Po-land and Hungar (between 75% and 85%), France (79%) and the Mediterraneancountries of Greece, Ital, Spain and Portugal (8194%).

0

10

20

30

40

50

60

70

80

90

100

DK UK NL DE CZ LU SE BE IE AT Fl SK EE FR IT SI PL ES HU PT GR

%

ofpopulation

Agree strongly Agree

These high figures appear puzzling. Do the express disapproval of governmentactivities and social outcomes? If so, wh is this not converted into actual policies? Isit an institutional or a democratic failure? This would require exploration on the basisof political econom. Or alternativel, are there other, more important preferencesahead of these in peoples minds? Or are the perhaps inconsistent with other beliefs?These questions require a sociological exploration of attitudes, preferences and their

Figure 1.1: Attitudes toinequalit in 21 European Unioncountries: share of populationagreeing that governmentshould reduce differences inincome levels

Source: Own calculations basedon the European Social Survey(ESS), rounds 2 and 3.

Reference year: 2007, except forCzech Republic, Greece, Italy andLuxembourg, where it is 2005.


3/14



Figure 1.2: Attitudes andmeasured income inequalit(income quintile share ratio)

Source: Own calculations basedon the ESS, rounds 2 and 3 (2005,2007), and Eurostat online

statistics database (access date:15 June 2009).

Notes:

Attitudes: % of population who

agree strongly or agree withthe question whether govern-ment should reduce differences inincome levels.

Income quintile share ratio: Theratio of total income received bythe 20% of the population withthe highest income (top quintile)to that received by the 20% ofthe population with the lowestincome (lowest quintile). Income

must be understood as equival-ized disposable income.

Eurostat data (income quintileratio) refer to the same year asthe survey year for each country.

causes: how are these preferences determined? The most obvious question is: arepeople selfish? If the are, then their (relative) incomes (and income prospects) woulddetermine their preferences for further income redistribution b the state. In thischapter, we explore the patterns of preferences for redistribution, comparing alterna-tive surve questions and observing differences b social group.

Actual income inequalit and attitudes to1.3.inequalit

At the countr level, the relationship between perceived inequalit and actual incomeinequalit is ver weak. Some of the relativel equal countries have a strong prefer-ence for a reduction in inequalit (e.g. France, Slovenia, Hungar), while there ma berelativel little support for redistribution in countries with larger inequalit (the UnitedKingdom). Note that this relativel little support still encompasses the majorit ofthe respondents (57%).

0

10

20

30

40

50

60

70

80

90

100

DK UK NL DE CZ LU SE BE IE AT FI SK EE FR IT SI PL ES HU PT GR

%

ofpopulation

0

1

2

3

4

5

6

7

8

Incom

equintileshareratio

Attitudes Measured inequality (r ight axis)

On the one hand, it seems that Denmark has reached some sort of social opti-mum, where there is low income inequalit and people perceive inequalit to besmall. On the other hand, there seems to be large social support for inequalit reduc-tion in high-inequalit countries, including Poland and the Mediterranean countries ofItal, Spain, Portugal and Greece, in particular.


4/14



Figure 1.3: Attitudes andmeasured income inequalit

(Gini coefficient)

Source: Own calculations basedon the ESS, rounds 2 and 3 (2005,2007), and Eurostat online

statistics database (access date: 15June 2009).

Notes:

Attitudes: % of population whoagree strongly or agree with

the question whether govern-ment should reduce differences inincome levels.

Eurostat data (Gini coefficient)refer to the same year as the

survey year for each country. 0

10

20

30

40

50

60

70

80

90

100

DK UK NL DE C Z LU S E BE IE AT FI SK E E FR IT SI PL ES HU PT GR

%

ofpopulation

0

5

10

15

20

25

30

35

40

Ginicoefficient

Measured inequality (right axis)Attitudes

The weak relationship between attitudes to inequalit and measured inequalitare not attributable to the actual measure of inequalit used here: we find a similarlweak relationship when the Gini coefficient is used. The countr ranking b actualincome inequalit remains largel unaffected b the particular measure of inequalitused. Thus, at the countr level, actual income inequalities do not explain wh peoplethink that inequalit should be reduced.

The main explanations for this highlight the importance of comparison effects:comparison over time and comparison across people. It can be seen that peoplescurrent demand for redistribution and their satisfaction with their own financial situ-ation depend on their income mobilit: their past and expected future incomes (e.g.Kahneman and Tversk 1979; Kahneman, Knetsch and Thaler 1991; Clark, Frijters andShields 2008; Tth 2008). The same amount of mone ma seem adequate if it ishigher than past income, et inadequate if it is lower. When evaluating their currentsituation, people anticipate future income changes as well, and ma be more tolerantof inequalit if the expect their own incomes to be on the rise (Hirschman and Roth-schild 1973). This ma have been the case in Eastern Europe in particular, where ine-qualit could be seen as a sign of greater mobilit, especiall during the earl ears oftransition (Ravallion and Lokshin 2000; Senik 2004).

It ma perhaps sound trivial to sa that people do not evaluate their situation inisolation from others, et this idea has onl slowl gained ground in mainstream eco-nomics. Now it is widel accepted that economic decisions depend on the context on status concerns; but due to the complexit of these issues, the actual mecha-nisms have et to be full explored. Frank explains how people sort themselves intolocal hierarchies, where the ma prefer to be a big fish in a small pond rather thana small fish in a big pond (Frank 1985; 2008). He also explains how status seeking

prevails (and is inevitable) and wh collective restrictions are imposed on status seek-


5/14


6/14



Figure 1.4: Percentage of thepopulation supporting a

reduction in inequalit in twoalternative surves, with andwithout an explicit trade-off forinequalit

Source: Own calculations basedon the ESS, rounds 2 and 3 (2005,2007), and the WVS, wave 5(sampling date: 200509).

Notes:

ESS: % of population who agreestrongly or agree with thestatement The governmentshould reduce differences inincome levels.

WVS: % of population who gaveanswers with a value of 1 to 4.

1 = Incomes should be mademore equal.10 = Greater income differences

are needed as incentives ofindividual performance.

0

10

20

30

40

50

60

70

80

90

UK NL DE SE Fl EE FR IT SI PL HU

%

ofpopulation

ESS WVS

As Figure 1.4 shows, the percentage of people who think that incomes should bemore equal is much lower in the WVS, where people are offered a choice betweeninequalit reduction and inequalit increase (We need larger income differences asincentives for individual effort). The sample covers fewer countries, but in the overallmajorit of the 11 countries for which comparison is possible, social support for ine-qualit reduction crumbles to less than 50%. In some countries, including Sweden,Estonia, Ital and Poland, less than one person in three believes that inequalit shouldbe reduced. B contrast, support for this was over two-thirds in the ESS, where peo-ple simpl had to state their preference as to whether or not to reduce inequalit,

with no explicit trade-off available. The significant difference in these surve out-comes highlights the importance of the framing of these questions, and the need forcareful robustness checks before an far-reaching interpretation is formulated. A de-tailed stud on preferences concerning the provision of public services in Hungarhad a similar outcome. These preferences alter significantl once people are madeaware of opportunit costs, and of the real price of these services (Csontos, Kornaiand Tth 1998).

Note, however, that the sampling ear of the two surves is different in most

countries. While, for the majorit of countries, the ESS data come from 2007, the WVSsurve ear is tpicall 2005 or 2006. Could that possibl explain the apparent differ-ences in the level of support for inequalit reduction? As Figure 1.5 shows, this is notthe case: the degree of support for inequalit reduction does not change ver muchover time in most countries. More accuratel, the change is not statisticall significantin eight out of the 21 countries, and its extent remains below 10 percentage points(with the exception of the Czech Republic and Finland, where it is 1011 percentagepoints). The disparit between the observed values of the two surves the ESS and


7/14



Figure 1.5: Change inpreferences over time. Share ofpopulation supportinginequalit reduction in 21European Union countries,200307

Source: Own calculations based

on the ESS, rounds 1, 2 and 3(2003, 2005, 2007).

Notes:

Countries are ranked according tomean value of support forinequality reduction in the latest

survey year available.

Percentage of population whoagree strongly or agree withthe question whether govern-

ment should reduce differences inincome levels.

We tested the statisticalsignificance of changes over time(comparing the two end-points ofthe period) and found that it isnot statistically significant in eightcountries, including Belgium,Estonia, Hungary, Italy, Luxem-bourg, the Netherlands, Polandand Sweden.

the WVS is substantiall larger, which implies that the differences cannot possibl beattributed to the change in surve ears.

0

10

20

30

40

50

60

70

80

90

100

DK UK NL D E CZ LU SE BE IE AT Fl SL EE FR IT Sl PL ES HU PT GR

%

ofpop

ulation

2003 2005 2007

Preferences appear to remain generall stable within countries: the countr rank-ing does not change substantiall, and those countries with meagre support for ine-qualit reduction in 2003 remain unenthusiastic four ears later (e.g. Denmark, UK),and those with widespread support (Hungar, Portugal, Greece) maintain that sup-port.

Peoples concerns about inequalit are strongl determined b what the think ofthe origin of social inequalit. Inequalit ma be more accepted, or even considered astate of social equilibrium, if people tend to perceive their social sstem as being oneof equal opportunities and greater mobilit, where individuals tend to attain whatthe deserve. Such attitudinal differences prevail across the United States and West-ern Europe, and while inequalit is shown to reduce well-being in Europe, it is muchless so across the Atlantic (Alesina, Di Tella and MacCulloch 2004).


8/14



Figure 1.6: Average support forinequalit and self-reliance (%)

Source: Own calculations basedon the WVS, wave 5 (samplingdate: 200509).

Notes:

Countries are ranked according tothe mean value of support forinequality. Higher values mean

greater support.

Mean scores of support for

inequality/self-reliance on a scaleof 1 to 10.

Inequality: the % of those whogave answers between 1 to 4.

1 Incomes should be mademore equal.10 Greater income differencesare needed as incentives ofindividual performance.

Self-reliance: the % of those who

gave answers between 1 to 4.1 The government should takemore responsibility to ensure thateveryone is provided for.10 People should take moreresponsibility providing forthemselves.

0

1

2

3

4

5

6

7

8

DE HU RO SI FI FR CY UK ES NL BG IT SE PL

Meansvaluesofinequalitytolerance/self-relia

nce

Inequality Self-reliance

Preferences about inequalit and self-reliance are connected on an individual

level, although the relationship is rather weak in statistical terms (the correlation coef-ficient is 0.35, which signals a weak, but statisticall significant, relationship). This sug-gests that people who tend to believe that larger income differences are needed arealso more likel to support the view that people should take more responsibilit toprovide for themselves. This relationship becomes feeble at the countr level. Thereare a number of countries where support for individual self-reliance is quite strong,including Finland, France, the United Kingdom and Sweden. This group thus includescountries with rather different culture and welfare sstems. The greatest average sup-

port for government provision (vis--vis individual self-reliance) prevails in Bulgaria.There is great diversit across countries where there is strong social support for great-er inequalit, including Ital, Sweden and Poland. While there is strong support forself-reliance in Sweden, the Italians and Poles tend to favour more government re-sponsibilit.

These latter findings ma appear puzzling. Note that all these questions are re-lated to a given institutional setting. In a countr with widespread universal benefits(like Sweden), the statement that people should take more responsibilit to provide

for themselves has a completel different connotation than in another countrwhere means-tested support is prevalent (tpicall coupled with the problem of stig-ma attached to claiming) (e.g. Ital).

There is an extended literature on earning inequalit, in particular on what wagepeople perceive as fair in various occupations. Here, the data permit us to observeone particular aspect of this issue.


9/14



Figure 1.7: (In)tolerance ofearning inequalities. Percentage

of population agreeing thatearnings inequalities are not fairon the basis of performance

Source: Own calculations basedon the WVS, wave 5 (samplingdate: 200509).

Note: % of population answeringnot fair to the following surveyquestion: Imagine two secretar-

ies, of the same age, doingpractically the same job. One findsout that the other earnsconsiderably more than she does.The better paid secretary,however, is quicker, more efficientand more reliable at her job. In

your opinion, is it fair or not fairthat one secretary is paid morethan the other?

0

5

10

15

20

25

30

CY SI BG DE SE RO Fl HU PL IT ES

%

ofpopulation

A surprisingl large intolerance of earnings inequalit prevails. In Finland, Hungarand Poland over one person in five believes that it is unfair to reward someone who

works more efficientl with higher wages (see Figure 1.7). In Ital and Spain, this figureis as high as one person in four. These figures are likel to include those who do ac-cept earnings inequalit on the basis of age or occupation, but also those who mightnot accept inequalit at all.

Socio-economic differences in attitudes1.5.

In the following section, we explore the extent to which these attitudes var b socialgroup. A primar hpothesis is self-interest: we expect people to be selfish or to sup-port redistribution, so long as the are among the potential winners.

In the majorit of countries, income situation tends to be strongl related to atti-tude towards redistribution: people on low incomes are more likel to support redis-tribution than are those on high incomes. In Figure 1.8, we compare the situation ofpeople in the bottom fourth of the income distribution (bottom quartile) with thosein the top fourth (top quartile), defining income as equivalized household income. This

definition thus assumes that the personal consumption opportunities are most likelto be influenced b household income (adjusted for household size), rather than pure-l personal income. The difference between the poor and the affluent is particularlhigh in German, but also in the United Kingdom (23% points), Slovenia (23% points),the Czech Republic (21% points) and Slovakia (20% points). Man of these countriesare affected b a Communist legac, including German (due to the unification), orb large prevailing inequalit (the UK).


10/14



Figure 1.8: Support for thereduction of inequalit, b

income group

Source: Own calculations basedon the ESS, rounds 2 and 3 (2005,2007).

Note: Attitudes: % of populationwho agree strongly or agreewith the question whethergovernment should reducedifferences in income levels.

0

10

20

30

40

50

60

70

80

90

100

DK NL UK LU DE AT SE CZ Fl IE BE EE PL IT PT FR SK SI HU ES GR

%

ofpopulation

Bottom quartile Top quartile

Surprisingl, three countries appear to be outliers. In Austria, support among the

bottom quartile (75%) and the top quartile (70%) differs onl modestl, albeit in astatisticall significant wa. On the other hand, there is no significant difference be-tween the preferences of the poor and the affluent groups in Portugal and Greece.In Greece, this might be related to a data problem (a fairl strong pattern can be ob-served in another wave of the surve, contrar to the latest wave, used here), while inPortugal it appears to be due to a particular distribution of responses. When we fo-cus onl on those who strongl agree with the reduction of inequalit, we find astrong difference b income group: while 45% of the bottom quartile strongl

agrees with the statement, this applies to onl 35% of the top quintile.Overall, however, there seems to be a strong pattern of self-interest across coun-

tries: those on low incomes are much more likel to support state intervention toreduce inequalit than are the more affluent groups.

We find similar, although less prevalent, differences across groups b emplomentstatus: the unemploed tend to be more supportive of state redistribution than thosewith paid work (Figure 1.9). The difference across these two emploment groupstends to be smaller than among the different income groups. In a number of coun-

tries, the difference is not statisticall significant.


11/14



Figure 1.9: Support for areduction in inequalit, b

emploment status


Notes:

Attitudes: % of population whoagree strongly or agree withthe question whether govern-ment should reduce differences inincome levels.

Employed refers to thepopulation in paid employment.

0

10

20

30

40

50

60

70

80

90

100

DK NL U K LU DE AT SE CZ Fl IE BE EE PL IT PT FR SK Sl HU ES GR

%

ofpopulation

Employed Unemployed

On the one hand, emploment status ma reflect differences in income levels,with the unemploed tpicall having lower incomes (although the negative effect of joblessness ma be buffered b the incomes of other household members). Thesedifferences in income levels are also largel determined b the unemploment sup-port sstem of the given countr. On the other hand, emploment relations tend tobe essential social relations. As mentioned before, work colleagues pla a major roleas a reference group for earnings comparisons (Clark and Senik 2008). Thus, peopleconnected to the world of work are expected to have higher wage aspirations, or

perhaps greater frustration if the view their relative income as low (Clark and Oswald1996; Luttmer 2005). However, we cannot assess the reference group of the unem-ploed, which might also partl include former colleagues.

Finall, we explore whether gender plas a role in these differences, and whetherthere is an sstematic difference between the attitudes of men and women. We findthat, in most countries, women are more likel to support a reduction in inequalit.One ma argue that gender differences simpl reflect income differences, given thewide prevalence of a gender pa gap, with women earning less and spending more

time awa from the labour market (e.g. because of child care or other care). This,however, is not the case. According to our simple multivariate calculations, this dispar-it between men and women remains even once differences in income levels, maritalstatus and emploment status are accounted for. Similar gender differences havebeen found with respect to Hungar using a series of alternative specifications, in-cluding multivariate models (controlling for differences in income, age, education andother characteristics) (Tth 2008: 1074). This evidence ma prove useful in under-standing gender differences in pro-social behaviour. (In a series of studies, men were

shown to be helpful in different was to women: while men were more likel to help


12/14



Figure 1.10: Difference insupport for a reduction ininequalit, b gender


Note: Attitudes: % of populationwho agree strongly or agreewith the question whethergovernment should reducedifferences in income levels.

strangers and to intervene in risk situations, women were more likel to providelong-term help and care. See the review of Eagl and Crowle 1986.)

0

10

20

30

40

50

60

70

80

90

100

DK NL U K LU DE AT SE CZ Fl IE BE EE PL IT PT FR SK SI HU ES GR

%

ofpopulation

Women Men

Much of the more in-depth analsis is inhibited b data constraints. The survesanalsed here do not contain evidence on the objective or subjective income mobilitof individuals. Further analsis, however, could be based on the European Social Sur-ve with respect to the impact of reference groups on attitudes.

Conclusion1.6.Our results suggest that support for inequalit erodes when there is a price tagattached to it: when it is made explicit that greater equalit ma come at the cost ofless incentive for individual effort. Attitudes are also influenced b peoples views onsocial justice and on the role of individuals personal responsibilit: preferences aboutinequalit and self-reliance are connected (albeit weakl) at an individual level. Notsurprisingl, we find support for our hpothesis that people are driven b self-interest:

the tend to support redistribution if the are among the potential winners. Thesepotential winners include those on low incomes and those without a job. With re-spect to gender differences, we find that, in most countries, women are more likelto support a reduction in inequalit. According to our calculations, this cannot beexplained b income differences or differences in other socio-economic characteris-tics: the gender difference prevails even after these potential explanator factors areaccounted for.


13/14



References

Alesina, A., Di Tella, R. and MacCulloch, R. (2004) Inequalit and happiness: areEuropeans and Americans different? Journal of Public Economics88(910):200942.

Clark, A. and Senik, C. (2008) Who compares to whom? The anatomy of incomecomparisons in Europe. Paris School of Economics Working Paper No. 2008-65.

Clark, A. E. and Oswald, A. J. (1996) Satisfaction and comparison income. Journal ofPublic Economics61: 35981.

Clark, A. E., Frijters, P. and Shields M. A. (2008) Relative income, happiness and utilit:

an explanation for the Easterlin paradox and other puzzles. Journal of EconomicLiterature 46(1): 95144.

Csontos, L., Kornai, J. and Tth, I. G. (1998) Tax awareness and reform of the welfarestate: Hungarian surve results. Economics of Transition 6(2): 287312.

Eagl, A. H. and Crowle M. (1986) Gender and helping behavior: a meta-analtic viewof the social pschological literature. Psychological Bulletin 100: 283308.

Frank, R. H. (1985). Choosing the Right Pond: Human Behaviour and the Quest forStatus. Oxford Universit Press, Oxford.

Frank, R. H. (2008). Should public polic respond to positional externalities?Journal ofPublic Economics92(89): 177786.

Hirschman, A. O. and Rothschild, M. (1973) The changing tolerance for income inequalitin the course of economic development. Quarterly Journal of Economics87(4):54466.

Kahneman, D. and Tversk, A. (1979) Prospect theor: an analsis of decision underrisk. Econometrica 47(2): 26392.

Kahneman, D., Knetsch, J. L. and Thaler, R. H. (1991) Anomalies the endowment

effect, loss aversion, and status quo bias.Journal of Economic Perspectives5(1):193206.

Luttmer, E. F. P. (2005) Neighbors as negatives. Quarterly Journal of Economics120(3): 9631002.

Luttmer, E. F. P. and Singhal, M. (2008) Culture, Context and the Taste for Redistribution.NBER Working Papers 14268, National Bureau of Economic Research, Inc.

Ravallion, M. and Lokshin, M. (2000). Who wants to redistribute? The tunnel effect in1990s Russia.Journal of Public Economics76: 87104.

Senik, C. (2004). When information dominates comparison. A panel data analsisusing Russian subjective data.Journal of Public Economics88(910): 2099133.

Tth, I. G. (2008). The demand for redistribution: a test on Hungarian data. CzechSociological Review44(6): 106387.


14/14



Table A1: Logit regression, theprobabilit of supportinginequalit reduction


Notes:

The specification includes countrycontrols as well.

Reference categories: BottomIncome Quartile; Education:Primary; Female; Never Married;In Paid Work. Standard errors in

parentheses.

* significant at 5%;** significant at 1%.

Annex

Coefficient Standard errors

Second Income Quartile -0.146** (0.043)

Third Income Quartile -0.252** (0.043)

Highest Income Quartile -0.559** (0.043)

Education: Lower Secondar -0.091 (0.053)

Education: Upper Secondar -0.226** (0.052)

Education: Post-Secondar, Non-Tertiar -0.355** (0.066)

Education: Tertiar -0.634** (0.054)

Male -0.243** (0.028)

Married 0.075* (0.033)

Separated -0.072 (0.112)

Divorced 0.269** (0.055)

Widowed 0.231** (0.059)

Inactive 0.081** (0.031)Unemploed 0.324** (0.076)

Constant 1.222** (0.081)

Observations 29643

Log likelihood -15948.09

Documents

TARKI European Social Report 2009 - Attitudes to Inequality