Fairness of inequality and support for redistribution: directly comparing citizens and legislators

Abstract Economic inequality constitutes a defining challenge of our time and it remains puzzling why rising levels of inequality have not led to more redistribution. In this article a novel individual-level perspective is taken, with a focus on how much legislators and citizens agree on questions of redistribution and inequality, and what causes these mismatches. The study compares legislators’ views to a representative citizen sample in Switzerland. The results show considerable disagreement between the groups with legislators being more sceptical towards redistribution and seeing inequality as fairer outcome. The mismatch is only partially explained by legislators’ higher social status. Ideology plays a fundamental role as more polarisation according to ideological lines is found among elites and their attitudes are also more rooted in their ideology. In sum, the findings point to some underexplored angles of the puzzle of why not more redistribution has been observed and thus offer a valuable addition to the existing literature.

share the citizens' assessment of inequality being unfair.While there is some evidence at the macro level suggesting that the translation of preferences for redistribution to votes might be biased (Lascombes 2022;Lesschaeve 2017;Rosset and Kurella 2021), the literature on individual-level elite preferences on the topic is still scarce and does not allow us to draw direct inferences on differences with citizens.Are political elites thus not sharing their electorate's views on these important questions of public policy regarding redistribution?And if so, what individual-level factors help us explain these differences?These are the questions we embark on in this article.
We rely on a novel dataset that does not only provide firsthand information on how political elites see and judge economic inequality but includes the answers to the same questions in a large-scale citizens survey as well for Switzerland.This makes it possible to examine a potential mismatch in attitudes towards inequality and redistribution and the factors explaining those attitudes between political elites and citizens in order to draw direct inference.
A priori, we would expect high levels of opinion congruence for the issue of inequality and redistribution as shown for example by Brooks and Manza (2006) in their comparative study of social policy responsiveness.The literature agrees that overall, representation works and at the aggregate level, politicians seem to more or less do what citizens want them to do, at least on the left-right dimension (see e.g.Burstein 2003;Golder and Stramski 2010) although more detailed, issue-specific accounts show considerable variation in congruence (Lefkofridi et al. 2014;Schakel and Hakhverdian 2018;Stecker and Tausendpfund 2016;Thomassen 2012) and there is also a growing body of research suggesting that not all citizens are represented equally (Bartels 2008;Giger et al. 2012;Rosset and Stecker 2019;Sevenans Soontjes and Walgrave 2022).
However, the literature also points to a number of potential pitfalls for a strong individual-level correspondence in the attitudes of elites and citizens regarding this topic.Most important, it has been shown that to explain individual attitudes about inequality, the socio-economic status is highly relevant (Condon and Wichowsky 2020;Fatke 2018).Given the discrepancy in social backgrounds between political elites and citizens in Western democracies (e.g.Best 2007;Hemingway 2020;Rosset 2016), we could thus expect a gap in what citizens and elites think about inequality and how they see redistribution.Secondly, as highlighted by an extant literature, welfare and inequality issues are closely tight to the left-right ideology (Arts and Gelissen 2001;Bean and Papadakis 1998;Feldman and Zaller 1992;Jaeger 2006b).Together with some findings that political elites' beliefs and attitudes are more structured and more difficult to change (Uslaner 1999), we thus expect that ideology is a relevant factor here.Not only to explain what individuals think about inequality but in particular in the sense that attitudes of elites could be more polarised and elite's preferences more rooted in ideology.
In this manuscript, we pursue a double focus by examining preferences for redistribution but at the same time also evaluations of fairness of inequality.While there is an extant literature on demand for redistribution, fairness concerns have not received as much attention but are now seen as one key driving factor to explain why somebody is in favour or against high levels of redistribution (Cavaillé 2023;Lascombes 2022).
Analysing both concepts at the same time thus allows us on the one hand to capture the bigger picture while on the other hand advancing the literature in particular on the aspect of fairness of inequality and how this is understood by elites and citizens alike.With this, our study of the individual-mechanisms contributes to our understanding of the mechanisms underlying (in-)congruence found at the macro level.
Our key findings indicate that indeed political elites and citizens differ in how they evaluate inequality and how much they support redistribution.Only part of this can be explained by the higher social status of elites.We further present evidence that elites' attitudes are more polarised and more rooted in their ideology.These findings are another steppingstone to help us understand why we don't see more policies fostering redistribution enacted despite high levels of inequality.

Theoretical framework
In this study, we focus on how political elites see inequality and redistribution and how their attitudes map with citizens' preferences in this regard.We approach the topic from a comparative point of view and ask whether gaps in the positions of elites and citizens with regard to redistribution policies and the fairness of inequality exist, and how we can explain them.These are important questions in terms of how well representation works and are particularly relevant as they shed light on a long-standing debate in political economy, i.e. why we are not seeing more action against economic inequality despite the clear disadvantages such high levels of inequality have for societies (e.g.Neckerman and Torche 2007) and also against the prediction of the classical Meltzer-Richard model which postulates that higher inequality leads to more redistribution.
For these reasons we focus on preferences for redistribution but we also study how elites and citizens see inequality, i.e. how they judge its fairness.These fairness considerations are seen as an important driver of attitudes towards redistribution (Cavaillé 2023;Lascombes 2022) and our double focus thus allows us not only to broaden the scope of the analysis but to add a potential new point of view to the picture.
While we have vast evidence on citizens' support for redistribution (e.g.Finseraas 2009) and their fairness perceptions (e.g.Osberg and Smeeding 2006), much less is known about elites' view on inequality and redistribution.We thus first discuss general findings on what citizens think about inequality and redistribution before focussing on two factors we believe could explain why citizens and elites don't hold congruent opinions on inequality and redistribution: differences in socio-economic status and the (differential) influence of ideological beliefs.
A first general conclusion from the literature on citizens' attitudes towards redistribution is that citizens are highly supportive of the principle of redistribution: Large majorities agree that governments should reduce income differences between rich and poor in most Western countries and support for more specific welfare programs such as pension schemes are also generally very high (e.g.Jaeger 2006a;Finseraas 2009).It is also known that citizens are rather sceptical towards high levels of inequality; if asked directly about its fairness, many state that they consider economic inequality in their country as unfair (Giger and Lascombes 2019;Osberg and Smeeding 2006).We also observe variation in these attitudes and beliefs across the population with attitudes being stratified by income and social class groups (e.g.Andersen and Curtis 2015;Bullock 2017) as well as strongly linked to ideological preferences (Jost et al. 2009).
At the elite level, we have only scant knowledge of what political elite members think about inequality and redistribution (but see Lopez et al. 2022).For economic elites in the USA, the literature suggests that they tend to have 'merit-based' interpretations of inequality which makes them not only more acceptant of the phenomenon but also more conservative in general (e.g.Bramlett et al. 2011;Suhay et al. 2021).Based on a comparative sample of political elites in five European countries, Breunig and Loewen (2021) report that 'politicans' perceptions of inequality are not, it seems, well-calibrated to the facts of the ground' as they find no big differences in the evaluation of the trajectory of inequality (how income differences have changed over the past twenty years).Also, half of their sample thinks of economic inequality to be fair or unfair respectively.However, their design does not allow us to connect these findings to a citizens' perspective.
At first sight, existing evidence seems to suggest that ideological congruence between political decision-makers and citizens is quite high (e.g.Burstein 2003;Golder and Stramski 2010) and given the crucial status of redistribution in state policy-making this should apply to inequality and reducing income differences between rich and poor as well.Also, when looking at experiments directly comparing elites and voters on attitudes or traits, on average the differences are small as a recent meta-analysis by Kertzer (2022) suggests.But, if anywhere the domain of attitudes is where he finds the most pronounced differences.This is confirmed for example by evidence in decision-making criteria (Sheffer et al. 2018) or of perceptions of voter opinion by elites more largely (e.g.Sevenans and Soontjes 2022;Varone and Helfer 2022;Walgrave et al. 2022).In sum, there is evidence leading us to expect that political elites and citizens differ in their evaluations of inequality and their approach to redistribution.Next, we discuss how differences in the socio-economic status of the average citizen and legislator can be influential for their view on inequality and redistribution and also elaborate on the crucial role of ideology and its consequences on attitudes towards inequality.A last hypothesis deals with the sources of these attitudes and potential differences between elites and ordinary citizens which could help explain their differential views on the topic.
The social and economic status provides leverage on individual's self-interest (see e.g.Cusack et al. 2006;Iversen and Soskice 2001;Rehm 2011), but it is also related to personal experience of various kinds, such as poverty or affluence, and fair or unfair treatments on the job or elsewhere.We can thus expect it to be a powerful predictor of attitudes and beliefs towards inequality.Indeed, the literature provides clear evidence that the individual-level variation can be explained by the fact that the more and less affluent differ in their views on inequality.For example, it has been shown that high social classes are less critical of inequality and consider it as fairer (Giger and Lascombes 2019;Kluegel and Smith 1986;Svallfors 1993).
The widely acknowledged influence of the socio-economic status gives rise to the expectation that citizens and elites could differ in their views on the fairness of inequality and their support for redistribution.We know that in Western countries, politicians tend to have a more affluent background than ordinary citizens (Best 2007;Butler 2014;Carnes and Lupu 2015;Rosset 2016).While most classical political economy work examines objective social status, we focus here on the subjective socio-economic status.Subjective status is most crucial in individual's assessment of inequality and for the subsequent demand for redistribution.Indeed, a body of social-psychological literature testifies the importance of subjective status (e.g.Brown-Iannuzzi et al. 2015;Kraus 2015;Kraus et al. 2009) and has been taken up by recent work in political science as well (see e.g.Condon and Wichowsky 2020;Fatke 2018).Given that higher social class is associated with less critical and thus more acceptant views of high levels of inequality, political elites could be more acceptant of high inequality situations as well and also less likely to support redistribution.That is, the differences in socio-economic status between citizens and political elites could lead to a gap in their thinking about inequality and about redistributive policies and thus to a lack of congruence between representatives and represented.Such a pattern has been observed for the US where especially Democratic legislator of high status were more likely to be in favour of bills maintaining the high level of inequality compared to their less-affluent party colleagues (Kraus and Callaghan 2014).Thus, we expect that: H1: Differences in legislators' and citizens' support for redistribution and how unfair they consider inequalities can be explained by their socio-economic status.
In other words, once we control for their different status, we should not find any differences between representatives and the represented anymore.
Another important factor regarding the acceptance of inequality is the ideological space (Cochrane 2015; Jost et al. 2009).Conservative or right-wing ideology puts much emphasis on individual rights and thus inequality is seen as a necessary outcome when some individuals outperform others, while individuals with a leftist orientation put more emphasis on equality of opportunities and the individual protection from the market.Given these deeply rooted beliefs about the working of society, it has been widely shown that ideological beliefs structure attitudes towards inequality and redistribution (Arts and Gelissen 2001;Bean and Papadakis 1998;Breunig and Loewen 2021;Feldman and Zaller 1992;Haddon 2019;Jaeger 2006b).A recent study by Waldfogel et al. (2021) suggests that differences between egalitarians (i.e.leftists) and non-egalitarians extend to how much attention these groups pay to inequality and thus their ability to detect inequality in the first place.Breunig and Loewen (2021) show for a large set of legislators that while perceptions of inequality are slightly more homogenous among left than right-wing legislators, legislators from the left and right are very clearly distinct about their evaluations of inequality.Suhay et al. (2021) report a strong connection between explanations of inequality and subsequent attitudes among economic elites (the super rich) in the US.
While differences in attitudes towards inequality according to ideology are not indicative for a match or mismatch between political elites and citizens per se, they can become consequential under two circumstances: First, if evaluations of inequality and redistribution demand are more stratified according to ideology among one group than the other, leading to more polarised attitudes and thus a more wide-spread distribution which is not mirrored in the other group.Second, if ideology is a more important predictor for redistributive attitudes in one group than in the other, this has consequences more for the meaning these attitudes have for the groups, in particular how they see inequality.In fact, we have reasons to believe that exactly these two scenarios apply to our setting here.Uslaner (1999) shows that politicians' opinions on various topics are more fixed and less prone to change than among ordinary citizens.The higher stability can be explained first by their expertise and the higher salience of political attitudes in their daily live, i.e. psychological reasons, but also by external pressure, i.e. the publicity of their actions and attitudes, and their re-election calculus which makes elites opt for more stability in their attitudes.In a recent study, Varone and Helfer (2022) also show that politicians are more accurate at predicting their party electorate's opinion on issues their party owns.In sum, we formulate the following two hypotheses: H2: Differences in legislators' and citizens' support for redistribution and how unfair they consider inequalities can be explained by the larger polarization of attitudes among legislators.H3: Legislators' support for redistribution and how unfair they consider inequalities are more rooted in ideological beliefs than those of citizens.

Inequality in the Swiss context and our data
For our investigation, we draw on data from Switzerland.Switzerland is rather representative for many Western countries in terms of economic inequality and redistribution effort: the country is ranking 7th among a set of West European countries with a Gini coefficient of .296 in 2017 (Pontusson et al. 2020).Furthermore, economic inequality has remained very stable since the early 2000s with almost no change in the Gini coefficient visible.The welfare state is not very large in Switzerland and the country can considered part of the liberal welfare scheme if any (Armingeon et al. 2004).Switzerland has arguably some special political features such as extended direct democratic votes and a militia, rather professionalised parliament (Bailer and Bütikofer 2015;Linder and Mueller 2017).If anything, for the study of congruence and representation, these characteristics should make it harder to find differences between legislators and citizens as the two groups should be more in contact and in general closer to each other (Helfer, Wäspi and Varone 2021).The frequent direct democratic votes provide politicians free high-quality information on what the public wants and could thus lead to a closer congruence.
In order to investigate evaluations of inequalities and support for redistribution, we draw on answers from two separate surveys: a survey with active Swiss legislators and one with a representative sample of citizens conducted simultaneously.Both surveys use the exact same questions to gauge a respondent's opinion regarding the fairness of inequality, support for redistribution, their socio-economic status and other aspects.These data also allow us to pair legislators from a specific party with the electorate of that same party on their views on inequalities and redistribution.
For the citizen survey, a representative probability sample of 10,268 Swiss citizens obtained from the Federal Statistical Office and aged 18 and over from the German and French speaking regions were contacted by the University institute FORS with a letter containing a personal link to an online Qualtrics survey and a 10 CHF voucher.Of the 4,667 who participated in the survey, 1,036 chose to use the paper version offered with the second reminder.Data collection ran from May 28th to July 17th 2018.To make sure our estimations are valid for the Swiss population at large in our estimations focussing on the population at large, we created weights.Using post-stratification, we create a weight for the combined distribution of age, gender and education of the population.Then, we add a probability weight for partisanship based on information from the last general elections.More information on the sample and its representativity is available in Online Appendix A.
In the legislator survey, we focus on 236 Swiss elected members of the national parliament (both Chambers, excluding the six representatives from the region of Ticino) and one regional parliament with 100 legislators (Geneva).In Switzerland, policy areas such as social welfare fall under the jurisdiction of both the regional and national parliaments.In fact, there have been popular votes at the national level and in many regions on the redistributive issue we focus on, taxation (see below). 1  We control for the type of parliament in our regression models.To ensure legislators themselves and not their staff respond to our questions, project team members met with legislators who filled out the survey on tablet computers during those meetings.To schedule meetings, legislators were first contacted with a letter informing them about the study which was followed up by personal phone calls.The first contacts were made in August 2018, meetings were mostly held in September 2018 although data collection spread out through October of the same year for a few meetings to accommodate politicians' scheduling requests.With our personal approach, we obtained exceptionally high response rates of between 77% and 64%.These participants are largely representative for their parliaments with regards to age, gender and political experience.Because one version of the survey fielded in both populations did not ask respondents to indicate their support for redistribution, we make use of a smaller set of the participating MPs (N = 92, of which 38 from the regional parliament) and citizens (N = 1,492) in this article.

Dependent and independent variables
We use the exact same questions in both the citizen and politician surveys as only such a parallel design allows drawing valid comparisons between the populations.
Our two main dependent variables are support for redistributive policy and how fair income inequalities are perceived.To gauge support for redistributive policies, we focused on a well-known general policy statement about redistributive taxation.The statement read 'Taxes on high-income should be raised while taxes on low-income should be reduced' on a four-point scale from fully disagree to fully agree, with an additional 'Don't know/no opinion' option which we omit here.This question was part of a larger set of policy statements respondents were asked to evaluate covering additional issues such as national security or immigration.Hence, we do not expect that respondents were primed by this question or paid particular attention to it.In an unrelated section later in the survey, we gauged how unfair respondents considered income inequalities.Specifically, we asked: 'Would you consider the income differences in Switzerland as fair or unfair?' offering respondents an answering scale from 1 (unfair) to 5 (fair), plus a 'don't know' option which we leave aside here.This means we reverse-coded the answers from the survey to these latter questions so that higher values indicate respondents perceiving inequalities as less fair, in line with our measure of respondents' support for redistributive policies.
We include a number of independent variables in our analyses, both on the legislators' and citizens' side, to address our hypotheses.As main independent variables we include respondents' socio-economic status and their ideology.We focus here on subjective socio-economic status on a 10-point scale using the MacArthur scale (e.g.Adler et al. 2000;Kraus and Keltner 2013;Buchel, Luijkx and Achterberg, 2021) which we introduced as follows including a drawing of a ladder with 10 rugs: 'This ladder represents different positions of people in Swiss society.10 -at the top those who are doing the best with the best education, the most money and most prestigious jobs. 1 -At the bottom those who are doing the worst with the lowest education, little money and a not respected job or no job at all.Where would you place yourself on this ladder?'Much like in most other countries around the world, Swiss legislators have a higher socio-economic status than citizens (see also Rosset 2016).While the legislators score a high 7.45 (SD = 1.29) on our 10-point scale, our representative citizen sample averages substantially lower at just 6.34 (SD = 1.54).In Switzerland, politicians do not need to disclose their income thus making it difficult for us to compare subjective status with an objective measure in the citizen and politician sample in the same way.However, using the occupation politicians list on the official parliamentary website and other resources freely available online, we classify politicians into five social classes following Oesch and Rennwald (2010).For citizens, we use information from the survey where we asked their occupation, whether they were self-employed and the number of employees working for them.These explorations underline the relationship between respondents' placement on the subjective 10-point ladder of social status and their classification based on their occupation in both populations (see Online Appendix B for detailed results).
Identifying politicians' ideology is straight-forward through their party affiliation/choice.We distinguish between middle, right and left-wing.We classify parties with socialist roots and the Greens as left-wing (SP, Grüne, AL, EAG, Partei der Arbeit), the conservative liberal and populist right parties as right-wing (SVP, FDP, MCG, PNOS) and finally Christian (democratic) parties as middle (CVP, BDP, GLP, EVP, EDU, CSP-OW) -following established practice in Switzerland (see e.g.Bornschier 2015). 2  To identify ideological leaning in the citizen sample, we asked respondents to indicate the party they had voted for the in the last elections in 2015.For respondents who did not remember, we used the party they would vote for if elections were held today.Descriptives in Online Appendix A show that within both populations, respondents are spread very similarly across the three ideologies, albeit we see a slight overrepresentation of left-wing legislators in comparison to the citizen sample.
Finally, as control variables we use gender, as a measure of political sophistication a dummy for whether a person has University education and, in our estimations using the politicians sample, we control for the policy level by adding a dummy for regional-federal legislators.

Findings
We begin by mapping descriptively what political elites and citizens in our sample think about redistribution and the unfairness of inequality, see Figure 1.If we look at legislators' and citizens' level of support for redistribution, we find a striking difference.If we combine those who rather and fully agree with the redistributive policy, among legislators only 39% support such a policy while 73% of citizens are in favour.With regards to the fairness of inequalities, when combining those who think economic inequality to be unfair or rather unfair, we see that more that almost 54% of MPs share this view while the share is even higher with 65% of citizens.However, confidence intervals overlap and on average, this difference is only just significant (t = 2.13, p = .033).
We thus see a substantive mismatch between legislators' and citizens' support for redistribution and, albeit to a lesser degree, how unfair they consider inequalities.We thus now embark on explaining these differences with (a) differences in social-status (H1) and (b) differences in the prominence and polarisation of ideology (H2 and H3).
Our first hypothesis posits that support for redistributive policies and perceptions of the fairness of inequalities are both related to a person's own socio-economic status (H1).Considering the significantly higher socio-economic status of legislators, we expect that the mismatch between legislator and citizens' opinions regarding inequalities can be explained by their differing social status position in society.In more statistical terms: once we control for socio-economic status, we expect the difference between legislators and citizens to disappear (or at least be substantially smaller).
Results of our linear regression models based on the combined dataset are reported in Figure 2, the coefficients for the full models are reported in Online Appendix C and we have tested the robustness of our findings using an ordered logistic regression and bootstrapping (see Online Appendix D).The first 'null model' with a dummy for political elites simply replicates what has become visible already in Figure 1: there is a gap in elites' and citizens' attitudes towards redistribution and the fairness of inequality.Then, there are two important findings here: First, both models show a significant effect of the socio-economic status on support for redistribution (Figure 2a) and how unfair respondents evaluate inequality (Figure 2b).This is in line with previous literature on individuals and thus confirms this relationship also holds at the elite level.Second, the models show that the difference in legislators' and citizens' support for redistribution does not decrease or become insignificant once we control for socio-economic status, even though the adjusted R-square increases from 3.3% to 10.5% for support for redistribution as our dependent variable (panel a).The predictive power of the fairness perception model (panel b) also rises from less than 0.1% to 6.1% underlining the importance of socio-economic status as a predictor in the relationship.But in this case, after controlling for socio-economic status, the differences between elites and citizens are no longer significant.This shows that socio-economic status does explain differences we find between citizens' and politicians' views on the fairness of inequalities but does not play as big a role in explaining their support for redistribution.In sum, we find indicative support for our first hypothesis.
The control variables behave as expected: Especially the variable for gender is highly predictive of both our dependent variables.Indirectly, this could mean that the lower rate of women in politics could also be another mechanism to explain differences between elites and citizens regarding these topics.
While socio-economic status can explain some of the gap between elites and citizens as we have seen, open questions remain.Thus, we explore the role of ideology and how it can help us understand the difference between political elites and citizens.Based on literature suggesting that views on both inequalities and redistribution are strongly rooted in people's overall ideological beliefs, but even more so among elites than ordinary citizens, we expect more polarisation among elites than citizens (H2) and a stronger link between ideology and attitudes towards redistribution and fairness among political elites than citizens (H3).To address these hypotheses, let us first directly compare elites and citizens by ideology.Figure 3 shows the average support for redistribution and how unfair they consider inequalities for our three ideological groups and separated for political elites and citizens.
Our data presented in Figure 3(a) reveal a clear picture of a mismatch across the political spectrum.On the right, citizens show twice more support for redistribution than legislators. 3The difference persists for centre legislators and citizens, albeit less extreme, and on the left, confidence intervals of the mean support by citizens and legislators even overlap. 4When we look at how unfair citizens and legislators generally evaluate inequalities (panel b), we find an even more pronounced mismatch across the political spectrum.Left-wing voters consider the inequalities less unfair than their legislators, while for middle and right-wing parties, legislators think inequalities are significantly fairer than their voters.As with the support for redistributive policies, the difference is more pronounced on the right.Right-wing legislators think inequalities are almost 1.5 times fairer (32% more) than their voters.
Taken together, we find overall support for our H2 indicating larger polarising tendencies among elites than citizens, and the pattern seems to be more pronounced for fairness perceptions than support for redistribution.Legislators take more extreme positions, with legislators less in favour of change on the right of the ideological space and showing more support on the left of the ideological space than their voters.Probably most striking, when we look at the respective group's support for redistribution, the absolute mean difference between left-wing and right-wing voters on both questions is much smaller than that within the legislators with a difference of 2.32 between left-and right-wing legislators versus only .66 among voters for example.There is only one instance where this turns out not to be correct: For redistribution preferences of left-wing citizens and elites.Given the importance left-wing parties attach to the topic of redistribution this might not be very surprising.If this were the only explanation, we would expect a similar match for how fair these groups judge inequality.However, for the perception of the fairness of inequalities we find a divide between legislators and citizens across the political spectrum, and not just for those on the right.
In sum, while this set of analysis has already helped us understand what causes the mismatch in opinion between representatives and represented, the next and last empirical exercise addresses the question why elites hold more polarised views than citizens (H3).To address the question how entrenched views of redistribution and fairness perceptions are among our two groups of interest, we draw on regression models controlling for our main variables of socio-economic status and ideological groups.This allows us to compare the role the former play to explain redistribution demand and fairness concerns about inequality.Figure 4 shows the corresponding results.
Our results of the separate regression models for both groups 5 in Figure 4 confirm the large role played by ideological beliefs in structuring both preferences for redistribution as well as fairness evaluations for legislators.First, even when controlling for other factors we see that the ideological groups differ in their support for redistribution as well as their fairness perceptions, and this among legislators and citizens.Right-wing MPs have an estimated lower support for redistribution than centre ones, while left-wing ideology is associated with more support for both MPs and citizens.To look at the relative sizes of these effects, namely the proportion of the total variance in a model attributed to a specific variable, we use the partial eta2 which can be compared across different models.We find that ideology play a much more important role for legislators than citizens.Ideology only explains 9% of the variance in support for redistribution in the citizen model, while it explains 73% in the legislator model.With 5% and 11% respectively, the socio-economic status follows as the second most important variable.In sum, this indicates that ideology and support for redistribution are much more interlinked for political elites while more loosely connected for ordinary citizens.This is by the way also what a simple correlation analysis shows.
A similarly pronounced pattern emerges for how unfair the groups see inequalities.Here, socio-economic status does not have a significant effect for legislators, while for citizens higher socio-economic status is associated with perceiving inequalities as more fair.Again, the proportion of unexplained variance captured by ideology is much higher (50%) for legislators than in the model for citizens (8%).
Finally, another way of testing our third hypothesis is to look at how much of the difference in the attitudes between the two groups is attributable to average differences in the levels of the observed independent variables (e.g. less women in the legislator sample than in the general population) and how much is attributable to differential effects (βj) of these variables in the two groups.The Blinder-Oaxaca decomposition method provides exactly this information (Blinder, 1973;Jann 2008;Oaxaca, 1973), full tables in Online Appendix C. What becomes clear here again is that for legislators, the influence of ideology is much larger than for the general population as for both outcome variables the coefficients (i.e.βj) for ideology are significant suggesting that these variables have a differential effect depending on the group.In sum, both approaches to test our H3 point in the same direction: redistribution and inequality attitudes are more rooted in ideological beliefs for political elites than for citizens.This is in line with our hypothesis 3.
Taken together, our results underline not only the importance of ideology to explain redistribution preferences and fairness perceptions.The results are also informative for the initial puzzle of why citizens and political elites differ in their view on the important questions of public policy: They do not necessary disagree on the fundamentals but (a) politicians have more polarised views and (b) more coherent and structured belief systems according to ideology than citizens.These two tendencies have as consequence that we see overall a mismatch between citizens and political elites regarding redistribution and, albeit to a lesser degree, fairness evaluations.

Conclusions
We look at how politicians evaluate and judge inequality and redistribution in this article and could document that overall, Swiss political elites are less supportive of redistribution than the Swiss population.Moreover, politicians see economic inequality as a rather fair outcome, again disagreeing with the Swiss people to quite a bit.Importantly, differences in socio-economic status between politicians and citizens -which are present in Switzerland as we show -do not fully explain these divergent views, especially not regarding support for redistribution.A closer look at the role played by ideology tells us that attitudes of elites are more polarised across the left-right spectrum and that ideological beliefs more heavily linked to support for redistribution and fairness evaluations among elites than among citizens.In sum, this indicates that the belief system of politicians is likely firmer and especially more structured around ideological beliefs.
As such, ideology plays a larger role in determining attitudes towards redistribution and inequality for this group.This is good and bad news at the same time for scholars on (unequal) representation and congruence.First, as our results suggest that politicians are not completely off-track with what citizens want on inequality and redistribution, they simply hold more extreme opinions and also more ideologically rooted ones.This seems positive for culprits of representation since these more extreme opinions are unlikely to get enacted and as scholars of directional or compensatory models of voting state, to be extreme(r) can be a good thing for overall correspondence between policy-outcomes and preferences later on (e.g.Kedar 2005).
The incongruence on these important questions of public policy can on the other hand also be worrisome, especially because our analyses show that the gap is linked to stable, deeply-rooted aspects such as their socio-economic status within society and core ideological beliefs.These things are unlikely to change in the short run and thus these gaps in representation to remain in place.Further research should still explore under which conditions elites learn and eventually change their attitudes and their behaviour, focusing on redistributive issues.
We explored our research question with Swiss data.This poses inevitably the question about the generalisability of our results.However, if anything, Switzerland should be a rather hard case to find incongruence between citizens and political elites due to its institutional framework with extended direct democracy and high levels of federalism which should foster a close connection between representatives and represented.Moreover, our findings are substantiated by the replication of our findings for the highly politicised issue of immigration.The findings we report in Online Appendix D point to the uniqueness of the mechanism for redistribution issues.We see that citizens and legislators from the same political orientation are much more in tune for immigration issues than for questions of redistribution.
This being said, an obvious limitation of our article is its focus on opinion and their match or mismatch.While the parallel surveying of elites and citizens has allowed a direct comparison of their stances on the issues, we have only focused on what might be the starting point for policy representation and we also lack a more dynamic perspective on the topic where reactions on the sides of elites could be studied.Especially, we have not touched on actual policies in favour of redistribution, and in the translation of attitudes into policy might lay even more complexity.
Our findings point to some underexplored angles of the puzzle of why not more redistribution has been observed and thus offers a valuable addition to the existing literature.The findings presented in this article suggest that one of the problems of the Meltzer-Richard model lies at the representation stage where policy preferences of citizens are channelled into the system and taken up by representatives.While we could show that higher socio-economic status and deeply rooted ideological beliefs play a role in explaining this imperfect transformation, we are far from a definite answer to the puzzle.If anything, the questions raised in this manuscript as well as the empirical evidence presented should serve as food for thought to further explore what causes incongruencies between political elites and citizens, how to explain preference gaps regarding social policy and redistribution, but also beyond, and most importantly what the consequences of such disruptions are for the working of democracy and representation.

Notes
1.There were popular initiatives in the Swiss direct democratic system in recent years at the national level (November 30th 2014) and regional level, in the regional Geneva parliament (in 2014) and several others (SG in 2015, ZH in 2013, BL 2012, SO 2014).2. The ethical clearance we obtained for the project does not allow us to report results by party, i.e. models with a dummy variable for each party, instead of groupings.3. Right-wing citizens: M = 2.74, 95% CI [2.67, 2.80], right-wing legislators: M = 1.24, 95% CI [1.00, 1.49].4. Left-wing citizens: M = 3.34, 95% CI [3.26,3.41];left-wing legislators: M = 3.59, 95% CI [3.30,3.89]. 5. Full models in Online Appendix C, robustness checks with ordered logistic regression in Online Appendix D.

Figure 1 .
Figure 1.support for redistribution and perceptions of the unfairness of inequalities among legislators and citizens.note: weights applied in kernel density graphs.

Figure 2 .
Figure2.support for redistribution and unfairness of inequality evaluations among legislators and citizens controlling for their socio-economic status (ses).note.95%-confidence intervals reported, full models in online appendix c.

Figure 3 .
Figure 3. support for redistribution and unfairness of inequality evaluations of legislators and citizen by their ideological leaning.

Figure 4 .
Figure 4. comparing the role of ideology among citizens and legislators as predictors for their support for redistribution and unfairness of inequality evaluations.note.95%-confidence intervals reported, full models in online appendix c.