Behind a thin veil of ignorance and beyond the original position: a social experiment for distributive policy preferences of young people in Greece. Nikos Koutsiaras* & Yannis Tsirbas** * National and Kapodistrian University of Athens ** National and Kapodistrian University of Athens ECPR-2017 General Conference Oslo 9/9/2017
Introduction Modern societies are characterized by economic inequalities. Reasons, inter alia: market failures historical and socio-political reasons the very definition and content given to the concepts of "inequality" or "justice" in every particular social context governmental policies In every society there are institutions which are responsible for redistributing wealth in order to mitigate the extent and the effects of economic inequalities (i.e. taxation). The degree to which such a redistribution will take place, as well as the ways in which it will take place consist a central question in political competition. 2
Introduction The relevant literature mostly deals with the supply of redistribution policies. At the same time, the relationship between economic inequalities and demand for redistribution policies is not well explored. Greece, being a country ridden by economic crisis and severe inequalities serves as a suitable case for investigating the formulation of the demand for redistribution policies. 3
Aim of the paper The aim of the paper is to investigate the formation of subjective preferences for redistribution policies, by applying an experimental research design to young Greek university students. 4
Studies on subjective preferences for redistribution Salience of redistributive policies in politics yet, political scientists and economists have focused mainly on the supply side, e.g. the effects of globalisation on government policies w.r.t. taxation, social transfers, the welfare state and redistribution (Gray, 1998; Pierson, 1996; Castles, 2001; Scharpf and Schmidt, 2000). However, during the last decade, theoretical and empirical research has also been developed in regard to preferences for redistribution (Benabu and Tirole, 2006; Alesina and Giuliano, 2011). Empirical research on subjective preferences has relied on survey data, thereby tracing individuals support for overall redistribution. In our work we focus on redistributive policies and the welfare state; preferences for redistribution vary with mechanisms, policies and institutions (Jaime- Castillo and Saez-Lizano, 2016). 5
Studies on subjective preferences for redistribution Two main theoretical approaches to explain preferences: a. the uitilitarian-rational explanation, associated with Meltzer-Richard, 1981; also, inter alia, Piketty s, 1995, rational learning theory (family income and own mobility prospects) and b. the ideological approach, entailing values and political orientations (Feldman and Zaller, 1992). A closer look at how redistributive preferences are formed and, especially, on a. the importance of the distance between perceptions and reality in regard to one s relative economic status and b. the effects of providing objective information (allowing for increased awareness of both the extent of inequality and one s relative position and employment/income prospects) may shed some light on the debate concerning the rational vs. ideological motivation of individual preferences; randomized survey experiments may be employed for that purpose (Kuzimieko et al., 2015, where the information elasticity of preferences is found to be generally small) 6
Literature review We employ the Rawlsian metaphor: People in the original position are unaware of their relative place in society or their participation in common resources and operate under a veil of ignorance (Rawls 1971). People, therefore, under the veil of ignorance, would agree to a social contract, under the terms of which inequalities would be dealt with so as to favor those least well off (ibid.). As people move away from the original position, their veil of ignorance becomes thinner and they (are supposed to) become more utilitarian: their preferences will match a) their actual economic situation and b) their relative economic position in society. Young people, including university students, have moved some way beyond the original position (they have not traveled the whole distance); thus, they have a relatively thin but not transparent veil of ignorance. 7
Literature review That distance between perceptions and reality may conveniently (and metaphorically) be thought of as a veil of ignorance; it may be operationalised and measured accordingly on the basis of experimental survey data. Utility maximizers would prefer policies which maximize their income and/or improve their relative position. Thus, enter Meltzer and Richard, 1981 : As inequalities widen demand for redistribution increases. Thus well-off individuals are generally not in favour of redistributive policies, whereas economically underprivileged individuals should support such policies. Hence the importance of actual personal economic situation and of perceived position in society. Yet ideology and worldviews about equality and fairness are found to have an impact on people s preferences. Also, participation and interest in politics are important. 8
Research question Does the provision of information (a) on economic inequalities and related policies in the country and (b) on young persons relative personal/family position, affect their redistribution preferences? In other words, is there a veil of ignorance and what difference does it make? 9
Research hypotheses H 0 : There is no relation between the provision of information (lifting the veil of ignorance) and redistribution preferences of young Greeks. H1: The lifting (or thinning) of their veil of ignorance changes young Greeks redistribution preferences towards the adoption of a more utilitarian/rational position. H2: The lifting (or thinning) of their veil of ignorance reinforces young Greeks redistribution predispositions. 10
Methodology Experimental design Greek students (undergraduate and post-graduate) from 3 Universities (National and Kapodistrian University of Athens, Agricultural University of Athens and Harokopeio University) Convenience sampling: N=533 students Online questionnaire, in-lab conduct, with supervision. Completion of data collection: May 2017. 11
Methodology Dependent variable: Preferences for redistributive policies composite variable consisting of 6 variables. Was the most important component of Principal Components Analysis (PCA) of a theory-informed battery of 13 variables, which returned 4 components. (Eigenvalue: 4.01, % of variance: 31.1%) Preferences for redistributive policies Strengthening of protection against dismissals Increase in the minimum wage Increase in employers contributions for social security Increase in public expenditure for the elderly (mainly pensions) Increase in high incomes tax rates Increase in public expenditure for unemployment benefits 12 Cronbach s Alpha=.756
Methodology Independent variables manipulation (treatments): Three random groups Group A (control): no treatment Group B (first treatment): Shown an 11-minute video, describing the economic situation in Greece, economic inequalities and the policies to tackle them, also in European comparative perspective. Group C (second treatment): In addition to the material of the second treatment, the members of the third group were shown a 14-minute video presenting data enabling them to figure out their own relative position and mobility prospects. 13
Methodology Control variables Gender, Age, Year of study Actual personal economic situation (deciles, according to the Hellenic Statistical Authority) L-R self-placement (0-10) Veil of ignorance (operationalized as the subtraction of real economic position (10 point scale) based on reported income in Euros from subjective economic position (10 point scale) Political participation/composite variable (have done/would do: petition, boycott, legal demonstration, unofficial demonstration, legal strike, illegal strike, public building occupation, destruction of property). Interest in politics Prior views on issues of fairness, equality and the welfare state (13 variables reduced to 4 components through Principal Components Analysis): Social Investment Welfare State Egalitarian welfare state Fairness, not equality Equality as fairness 14
The experimental design Total Sample (N=533) Demographic questions Questions on values political orientation and political participation (L-R self-placement, prior beliefs and attitudes on the issues of equality, fairness and the WS, interest in politics and participation) Random Assignment in Groups/Experimental treatments Group A (N=181) No Video. Group B (N=176) Short video about inequalities and economic situation in Greece. Group C (N=176) Long video with additional data enabling respondents to figure out their own relative position and mobility prospects. Measure the effect of the experimental treatment as differences between groups in the dependent variable Control for Demos and prior beliefs/attitudes to ensure Group similarity 15
Sample description (N=533) Gender Years of Age Level of study 16
Sample description/l-r self-placement (%) N=488 respondents who positioned themselves Mean: 4.41 17
The match of perceptions with reality: a rather thin veil of ignorance (%) Veil of Ignorance=Real economic position based on reported income in Euros (10 point scale) minus subjective economic position (10 point scale) Statistics Veil 2 Subposit-Econsit N Valid 493 Missing 40 Mean -3.03 Std. Deviation 2.668 Range 15 Minimum -9 Maximum 6 Either same perceived and real position or difference smaller than 1 Std. Deviation 18
DV: Preferences for redistributive policies (composite scale, 6 variables, 1-5, 1=totally agree) recoded (%) For Redistribution: Agree/rather agree: all 6 variables. Redistribution a la carte: Disagree/have no opinion: less than 3 variables (they agree more than they disagree). Opposition to Redistribution: Disagree/rather disagree /have no opinion: 3 variables and more (they disagree more than they agree). 19
Preferences for redistributive policies behind the (thin)veil of ignorance % within Veil of ignorance recoded Recoded DV1COMBO-Final * Veil of ignorance recoded Crosstabulation Recoded DV1COMBO- Final Rather thin veil Veil of ignorance recoded Lower perception than reality Higher perception than reality Total Support for redistribution 46,4% 60,3% 62,5% 53,2% Caution towards redistribution Opposition to redistribution 21,5% 15,9% 12,5% 18,7% 32,1% 23,7% 25,0% 28,1% Total 100,0% 100,0% 100,0% 100,0% Chi-Square Tests Value df Asymp. Sig. (2- sided) Pearson Chi-Square 10,305 a 4,036 Likelihood Ratio 10,370 4,035 Linear-by-Linear Association 7,792 1,005 N of Valid Cases 513 a. 2 cells (22,2%) have expected count less than 5. The minimum expected count is 2,99. 20
The (no) effect of lifting the veil of ignorance- ANCOVA Tests of Between-Subjects Effects Levene's Test of Equality of Error Variances a Dependent Variable: Preferences for redistributive policies F df1 df2 Sig. 1,352 2 307,260 Tests the null hypothesis that the error variance of the dependent variable is equal across groups. No statistically significant error differences among groups: OK to run ANCOVA No statistically significant differences in the DV between the three experiment groups: Dependent Variable: Preferences for redistributive policies Source Type III Sum of Squares df Mean Square F Sig. Partial Eta Squared Corrected Model 2581,183 a 14 184,370 19,715,000,483 Intercept 888,963 1 888,963 95,057,000,244 L-R self-placement 16,402 1 16,402 1,754,186,006 Political interest 59,249 1 59,249 6,335,012,021 Age 3,526 1 3,526,377,540,001 Veil of ignorance,571 1,571,061,805,000 Years of study,969 1,969,104,748,000 Political participation 21,949 1 21,949 2,347,127,008 Personal economic situation Social investment welfare state 6,201 1 6,201,663,416,002 323,218 1 323,218 34,562,000,105 Equality as fairness 281,912 1 281,912 30,145,000,093 Egalitarian Welfare State 945,489 1 945,489 101,101,000,255 Fairness, not equality 202,215 1 202,215 21,623,000,068 Gender 179,196 1 179,196 19,162,000,061 Video 7,645 2 3,823,409,665,003 Error 2758,804 295 9,352 Total 54606,000 310 Corrected Total 5339,987 309 a. R Squared =,483 (Adjusted R Squared =,459) The provision of objective information does not affect Preferences for redistributive policies. 21
The (no) effect of lifting the veil of ignorance Lifting the veil of ignorance has no effects on preferences. Prima facie evidence against rational motivations. Possible explanation: Prior beliefs, values and ideology are too strong to be changed by utilitarian info. 22
Discriminant Analysis Since lifting the veil of ignorance has no effect on policy preferences across the 3 experiment groups, we can treat the sample as a whole. Discriminant analysis can identify the best group of independent variables, which can discriminate between the three categories of the dependent variable (support for redistribution, redistribution a la carte, opposition to redistribution). It can identify the existence of more than one dimensions (functions), on which the IVs influence the DV. 23
Greek students: Discriminant analysis- Descriptives Redistribution Preferences Composite political Participation L-R Self placement Interest in politics Social Investment Welfare State Equality as fairness Egalitarian welfare state Fairness, not equality Veil of ignorance 3 categories Support for redistribution Mean Std. Deviation Valid N Mean Redistribution a la carte Std. Deviation Valid N Mean Opposition to redistribution Std. Deviation Valid N Mean Total Std. Deviation 5.24 2.12 245 4.06 1.73 81 4.00 2.10 124 4.69 2.14 450 3.77 2.07 245 4.91 1.62 81 5.34 1.72 124 4.41 2.03 450 1.84 0.85 245 2.15 0.76 81 1.78 0.88 124 1.88 0.85 450-0.21 0.74 245 0.09 1.02 81 0.25 1.20 124-0.03 0.96 450-0.19 0.82 245 0.06 0.99 81 0.30 1.23 124-0.01 1.00 450-0.28 0.88 245 0.14 0.92 81 0.50 1.06 124 0.01 0.99 450 0.23 1.03 245-0.06 0.91 81-0.40 0.88 124 0.00 1.01 450 0.58 0.56 245 0.43 0.55 81 0.45 0.55 124 0.52 0.56 450 Valid N 24
Greek students Discriminant analysis (N=450) Standardized Canonical Discriminant Function Coefficients Function 1 2 Composite political Participation -.114 -.194 L-R self-placement.173.341 Interest in politics -.230.801 Social Investment Welfare State.475 -.062 Equality as fairness.443 -.217 Egalitarian welfare state.640 -.105 Fairness, not equality -.431.386 Veil of ignorance 3 categories -.072 -.264 25
Greek students Discriminant analysis (N=450) Structure Matrix Function 1 2 L-R self-placement.565 *.357 Egalitarian Welfare State.549 *.033 Fairness, not equality -.425 *.081 Equality as fairness.336 * -.017 Social Investment Welfare State.330 *.143 Interest in politics # -.007.836 * Composite political participation -.426 -.540 * Veil of ignorance 3 categories -.172 -.310 * Pooled within-groups correlations between discriminating variables and standardized canonical discriminant functions Variables ordered by absolute size of correlation within function. # Interest in politics coding: 1-4, 1=Very interested. 26
Greek students Discriminant analysis (N=450) The ideological/political orientation dimension Function 1: Canonical correlation: 0.55 / 92.8% of variance Mean location of Respondents on the function (group centroids) Support A la carte Opposition The political activism and economic self-awareness dimension Function 2: Canonical correlation: 0.18 / 7.2% of variance Mean location of Respondents on the function (group centroids) Support Opposition A la carte 27
Greek Students Discriminant analysis (N=450) Original Crossvalidated b Count % Count % Classification Results a,c Redistribution preferences Predicted Group Membership Total Support for redistribution Redistributi on a la carte Opposition to redistribution Support for redistribution 217 4 24 245 Caution towards redistribution 49 6 26 81 Opposition to redistribution 51 6 67 124 Ungrouped cases 5 0 0 5 Support for redistribution 88.6 1.6 9.8 100.0 Caution towards redistribution 60.5 7.4 32.1 100.0 Opposition to redistribution 41.1 4.8 54.0 100.0 Ungrouped cases 100.0.0.0 100.0 Support for redistribution 214 6 25 245 Caution towards redistribution 51 3 27 81 Opposition to redistribution 53 6 65 124 Support for redistribution 87.3 2.4 10.2 100.0 Caution towards redistribution 63.0 3.7 33.3 100.0 Opposition to redistribution 42.7 4.8 52.4 100.0 a. 64.4% of original grouped cases correctly classified. b. Cross validation is done only for those cases in the analysis. In cross validation, each case is classified by the functions derived from all cases other than that case. c. 62,7% of cross-validated grouped cases correctly classified. 28 Largest prior probability: 54.4%
Conclusions 1. Provision of economic information and lifting of the veil of ignorance does not affect redistribution preferences. 2. The veil of ignorance itself is rather thin. More than 50% of the respondents have a fairly accurate perception of their relative economic position. 3. The majority of the respondents support redistribution policies. 4. Furthermore, preferences are not determined by economic motives at all, but only by 1. Ideological factors (beliefs in regard to fairness and the welfare state, L-R selfplacement) and 2. Civic engagement factors (political participation, interest in politics), which mainly affect the formation of firm positions (either for or against redistribution) as opposed to moderate ones. 3. The veil of ignorance itself: it seems to be related to the intensity of preferences for redistributive policies -but not to their content. 29
Conclusions 4. Possible explanations-ideas for further research: Youth a. "they are lured by egalitarian and socialist ideas, b. they participate more. Crisis: a. greater attention to personal and family economic conditions, so their perceptions are more accurate. b. the risk of personal economic failure (including business bankruptcy) is widespread and is getting closer and it even affects members of one s inner circle; even relatively well-off may opt for a higher degree of social insurance (including redistribution). c. distrust in government. Pragmatism: increased economic self-awareness, better knowledge of the workings of Greek social policy (and its legacy thus far) leads to more moderate positions. 30