Introduction

 Many papers study how self-interest and ideological opinions shape one's normative preferences and, in particular, preferences on redistribution (see Clark and d'Ambrosio, 2015, for a comprehensive survey). In a democratic framework, the aggregation of individual preferences generates, depending of the voting system, to collective social outcomes on redistribution (Meltzer and Richard, 1981; Piketty, 1995; Bénabou and Ok, 2001). However, since voting systems do not necessarily lead to a consensual collective rule, in reality, even within a democratic framework, people have to conform to rules that may conflict with their own views of a fair redistribution. More generally, outside the realm of redistributive issues, laws abound that deeply contradict many citizen's core beliefs.

In this paper, echoing the literature on the expressive vs. backlash effects of legal norms (Tyler, 2000; Cooter, 1998, 2000; Posner, 1998, 2000a, 200b; McAdams, 1997, 2000; Licht, 2008; Bénabou and Tirole, 2012; Carbonara et al., 2012), we seek to understand how people's normative preferences are affected by knowing about such laws. Do they internalize collective preferences, even if it is blatantly unfair from their point of view? Do they ignore them, or does knowing about them cause a radicalization of their beliefs against the collective rule?

This matter of whether political obligation exists or not (i.e. whether citizens should be morally bound to internalize and submit to legitimately produced social preferences) and whether civil disobedience is legitimate or not in the face of an illegitimate social organization have received widely diverse theoretical answers throughout History (see Dagger and Lefkowitz, 2014, and Brownlee, 2016 for a review). In Crito, Plato exemplifies Socrates' compliance with his own sentencing to death to advocate that private persons should abide by their city's legal judgements, on pain of the city ‘being turned down' (Plato, 360 B.C.E).

Absolutists, following Hobbes, claim that in order to escape from a natural state where mankind is trapped in a war of all against all, people must engage in a one-sided Social Contract with an absolute ruler, who will guarantee their security in exchange to complete submission. In this framework, individual social choice preferences should not be allowed, since any discussion of the ruler's legitimacy on the grounds of one's own conscience would undermine the social contract and provoke a return of the state of nature (Hobbes, 1651).

By contrast, liberals, following Locke, advocate that people engage freely in the Social Contract and that their compliance with the general will is conditional to institutional safeguards such as the separation of powers and the majority rule. More importantly, people have the right to legitimately rebel against treacherous rulers who, instead of ruling according to the collective, unbiased will, take sides or rule for their own benefit. In this framework, political obligation is partial and civil disobedience is a virtue: individual preferences on social choice should bow to legitimate but overrule illegitimate collective preferences (Locke, 1689).A distinct answer is proposed by Rousseau, for whom the Social Contract should result from a direct, un-mediated union of all citizens who alienate their individual wills to a collective Sovereign. Since it relies on direct democracy and equal participation of all citizens, the Sovereign's choices can only pursue the common good and must focus on general, collective issues. As a result, political obligation is absolute and individuals should internalize collective preferences: if one's beliefs about the common good differ from the Sovereign's, it is one's preferences that are necessarily mistaken and should be re-aligned on the collective preferences. Dissenters should be, if needed, ‘forced to be free' (Rousseau, 1762).

More recently, Rawls takes a more nuanced stance on political obligation with a limited scope of legitimate collective preferences. In Rawls' view, citizens have the political obligation to support and to comply with just institutions. Political obligation derives from a liberal principle of legitimacy, where political obligation is limited to a stable “overlapping political consensus” of the basic normative elements shared by free, equal and reasonable citizens who understand and respect the diversity of individual views on social choices (Rawls, 1993 [1971], p.137). In this framework, there is no general political obligation for obeying the law, except for, noblesse oblige, those who assume public office (Rawls, 1999, p.98). On the contrary, civil disobedience and conscientious refusal are an illegal but necessary stabilizing devices in a constitutional system: “By resisting injustice within the limits of fidelity to law, it serves to inhibit departures from justice and to correct them when they occur.” (Rawls, 1999, p.336).

The point of this paper is to provide some empirical insights on whether citizens internalize collective social preferences or are, rather, either indifferent or defiant in the face on a conspicuously unfair Social contract. Most of the literature on the expressive effects of legal norms focus on the behavioral effects of the internalization of the values conveyed in the norms (see, for example, ***). By contrast, here, the focus is normative, not behavioral: we are interested in expressive vs. backlash effects of normative collective preferences on individual normative preferences.

We chose a redistributive framing because of its relevance in terms of economic public polices and because such a framing is very likely to bring forth potential tensions between collective preferences, self-interest, and ideological opinions. To do so, we rely on an empirical social choice questionnaire where we present respondents with a blatantly unfair redistributive social preference (99% of a 10,000 euros inheritance for a millionaire vs. 1% for a minimum-wage earner) and check whether knowing about it affects the respondents' beliefs about fair redistribution.

Since we focus on people's responses to a collective rule, we pay special attention to their opinion on whether social justice should be embodied in fair rules (proceduralism) of should rather be measured in terms of fair outcomes (consequentialism). Indeed, converging evidence shows that people give intrinsic value to procedural fairness, sometimes against their own self-interest (see for example Anand, 2001; Frohlich et al., 2004; Wailoo and Anand, 2005; Dolan et al, 2007; Cappelen et al., 2007; Shayo and Harel, 2012; Mertins et al., 2013; Ku and Salmon, 2013). As a result, we include support of proceduralism among the more classical control ideological variables taken into account to explain law internalization.

Further, the key hypothesis of this paper is that knowing about collective rules may have different impacts on different-minded persons. To test this idea, our survey was presented to two groups of respondents who were selected because their potentially radically different view on proceduralism and law internalization: first-year students who chose to enroll in Law and in Economics the same year and at the same University. The point was to take advantage of the well-established fact that students who are self-selected in economics present singular normative frameworks compared with other students (see, among others, Maxwell and Ames, 1981; Carter and Irons, 1991; Franck et al., 1993; Yezer et al., 1996; Frey and Meier, 2003; Zsolnai, 2003; Faravelli, 2007; Hole, 2013).

Last, beside the classical regressions used in the literature, we use decision learning techniques in order to map the ‘normative frameworks' of the respondents. Decision trees and random forests (Breiman et al., 1984, Breiman, 2001) were introduced by Keely and Tan (2008) in the empirical social choice literature. Because they reveal the structural mapping of the respondents' normative framework, these data mining and machine learning techniques are useful to show how different combinations of self-interest and ideological factors lead the rallying of different-minded respondents to the same side of normative alternatives.

The paper is organized as follows. In Section 2 we present a synthesis of the main findings of the literature on (1) the determinants of redistributive preferences and (2) law internalization. In Section 3 we present the survey and discuss our empirical strategy. Section 4 presents and discusses the results on redistributive preferences and the internalization of collective preferences. Section 5 concludes the paper and calls for further work.

 A Questionnaire-Experimental Methodology

 Information on redistributive preferences is collected using a questionnaire experimental methodology commonly used in Empirical Social Choice. The questionnaire consists in a series of short and simple ‘vignettes' that depict social justice dilemmas. The point is, straightforwardly, to collect the respondents' “judgments of justice” (Yaari and Bar-Hillel, 1984) on abstract dilemmas, i.e. their beliefs on what ought to be done for a given situation to be just. By asking respondents their preferred solution to each dilemma, the objective is to elicit their normative framework, i.e. their support to alternative abstract social justice principles with as little cognitive effort as possible. This ‘vignette' methodology is relevant when, as it is the case here, 1) explicitly asking the respondents to choose between fully and precisely phrased normative concepts would be too time-consuming and cognitively tiresome for them and 2) the aim is to collect stated preferences and not to study behaviors.

In October, 2014, a total of 761 freshmen students of the University of Paris Nanterre took part in the survey. 255 of them were enrolled in Economics and 506 were enrolled in Law.

Participation was voluntary and there was no show-up fee paid. The anonymous paper surveys were distributed in controlled sessions during magisterial lectures. The total time for completing the questionnaire (instructions included) varied between 35 and 45 minutes, due to differences on class sizes. 432 students (56.8%) completed all the questions of the survey (155 in economics, i.e. 60.8% of Economics students and 277 in Law, i.e. 54.7% of Law students). The high level of attrition is mainly due to the fact that many students did not inform the socio-demographic part of the survey (especially the questions related to religious and political beliefs), despite the fact that answers were collected anonymously. The results presented in this paper are therefore based on the students that were comfortable enough with disclosing this information.

Table 1 provides information on the socio-demographic characteristics of the respondents, as well as a general overview of the variables used in the paper.

(1) The individual redistributive preferences of the respondent (variable redistribution) are collected asking about how a 10.000 euros inheritance should be shared between two heirs, respectively a millionaire and a minimum-wage earner. Since the aim of the article is not to address the responsibility issue, we explicitly state that both heirs are identical in all possible meaningful ways: age, personality, talent, effort, preferences and merit. The point is to know if respondents support the principle of a compensation of the less-advantaged heir (giving a greater share of the inheritance to the poor heir, redistribution = compensation) or believe in the general principle of an equal treatment of identical respondents (therefore giving an equal share of the inheritance to both heirs, redistribution = equal).

The respondents are then informed that the law (i.e., the collective, legitimate preferences) provides that the millionaire receive 99% of the inheritance. We then ask the respondents again about their opinion on how the inheritance should be shared between the heirs. The point is to see if respondents internalize collective preferences (variable Law internalization) by either reducing (Law internalization = compliant) or raising (Law internalization = defiant) the amount they previously chose to give to the poorer heir  , or if they are indifferent to collective preferences and do not modify their initial choice (Law internalization = indifferent). We present the respondents with an outrageously unfair collective preference that is sure to clash with both the compensation and the equal distribution principles. By doing so, the objective is to magnify how individuals react to collective preferences that conflict with their individual preferences on redistribution.

We also collect information on the main determinants of redistributive preferences discussed in the literature: self-interest, ideological beliefs, valorisation of procedural fairness, and some additional socio-demographic characteristics.

(2) Self-interest is measured in several ways. First, we focus on the declared current income of the respondent's household (variable income). The University Paris Nanterre's catchment area encompasses some of the wealthiest areas of Western Paris as well as classically under-privileged suburbs. We ask the respondents if they believe that their family is among the top 33% richest families in France (income = rich) or not (income = not rich). To measure self-interest with respect to the redistributive vignette described above, we focus on the ‘rich' end of the income spectrum in order to know which respondents would logically identify with the millionaire heir in the inheritance vignette.

The literature also suggests that belonging to a minority or a group that is discriminated against affects people's redistributive preferences. We therefore also collect information on the origin of the respondents (variable origin) in the form of the nationality of their parents, and differentiate the respondents with at least one parent with a foreign nationality (origin = foreign) and the respondents whose two parents are French (origin = French).

Beyond current self-interest, we also collect data on the respondents' optimism, risk aversion and declared effort in order to depict their expected trajectories. The variable optimism is built by asking the respondents about their expected income in 10 years' time. Optimistic respondents (optimism = high) believe that they will be among the top 33% income earners. The variable risk aversion is constructed by asking the respondents to imagine that they take part in a game where they are asked to choose between two boxes A and B. They know there are 100 euros in Box A, and that there is a 50 % chance that Box B is empty, and a 50% chance that it contains X euros. They are asked about the minimum amount X that would encourage them to choose Box B. Respondents whose amount is lower than the median amount (i.e., 200 euros) are tagged to have low risk-aversion (risk aversion = not high) while those who choose an amount equal or superior to 200 euros are tagged to be highly risk adverse (risk aversion = high). The variable declared effort is recorded by asking the respondents how many of their peers produce more effort than them in their studies. The variable effort is coded not high for the respondents who answer that about 50% or more of their peers are more effortful than them and high for the respondents who answer that they produce more effort than half of their peers.

(3) Ideological opinions encompass political and religious beliefs, support of equality vs. freedom and the position of the respondent on the responsibility issue.

To know about the political affiliation of the respondents (variable political), we ask if they are completely, rather or a bit left- or right-wing (political = left and political = right). To assess their support of equality vs. freedom (variable freequal), we ask respondents if they prefer (completely, rather or a bit) equality (freequal = equality) vs. freedom (freequal = freedom). The variable religious belief distinguishes between the respondents who are atheist or agnostic (belief = no religion) and the respondents who have a religion (belief = religion).

The position of the respondents in the responsibility issue is coded by the effort matters variable. To build this variable, we present the respondents with a vignette where the respondents are asked to share the joint production of two Gold Rush miners who display unequal productivity levels. In a first framing the respondents are not provided any information on the causes of the inequality in productivities while in a second framing they are informed that the low-productivity miner does not want to tire himself working on the mine. Respondents who decrease the share of the joint production given to the low-productivity worker in the ‘laziness' setting are coded as believing that effort matters and that laziness should be punished (effort matters = punish) while respondents who do not change their initial repartition are coded to be neutral on the responsibility issue (effort matters = neutral).

(4) Procedural fairness is assessed in two ways.

We focus on the support of proceduralism as a general justice principle (variable proceduralism). An immediate, simple way of knowing if proceduralism is supported notwithstanding potential consequentialist outcomes is to ask respondents to explicitly choose between a procedural country, “where the rules that define how society is organized are fair, but they do not necessarily lead to a fair distribution of wealth and income among the citizens”, and a consequentialist country, “where wealth and income are fairly distributed among the citizens, but the rules that define how society is organized are not necessarily fair”.

(5) Additional socio-demographic questions provide information on the gender the respondents primarily identify with (variable gender, coded male or female) as well as their University major (variable subject, coded law or economics). We also build a proxy for the respondents' altruistic tendencies (variable altruism) by presenting them with a non-experimental, simplified version of the ultimatum game, where they are asked to imagine that they are anonymously taking part in a game. In the game, they are given 10 dollars and are asked how much of this amount they are willing to give to another anonymous player. Respondents who give more than the median of the sample are classified as altruistic (altruism = high) while those who give the median amount or less are coded as not altruistic (altruism = low).

 Main results

In this article, we rely on original data to understand how the knowledge about collective redistributive preferences affects and structures individual preferences on redistribution. To do so, we focus on the classical determinants examined by the literature (self-interest, expected trajectories, and ideological opinions) as well as on proceduralism.

In line with the literature, we find that self-interest is a poor predictor of the internalization of individual redistributive preferences, by contrast with trajectories and ideological opinions political opinions. We also find that sensitivity to procedural fairness is a good predictor of both the respondents' redistributive preferences and their internalization of collective preferences. We also find that social inclusion may be an important factor of law internalization.

Interestingly, we find that the major difference between respondents in terms of redistributive preferences is not about the amount of redistribution that should be implemented, but if two persons who are essentially identical but economically unequal be treated differently in terms of economic redistribution. We also find that this preference governs the internalization of an (unfair) law that gives most of the inheritance to the richer heir.

These results pave the way to further research. First, the data used in this paper comes from a student survey. The study could be reiterated on a larger sample of the population, and on respondents of more diverse social and cultural backgrounds. Second, in this survey we tested the ‘resistance' of the respondents to an outrageously unfair rule. It may be interesting to test other, less extreme, collective rules, as well as, echoing Tyran and Feld (2006) to test the robustness of our findings across the degree of legitimacy of the law.

References

Alesina A., Giuliano, P. (2010). “Preferences for Redistribution”. In J. Benhabib, M. Jackson, A. Bisin (Eds.), Handbook of Social Economics, Vol. 1A. Amsterdam: North Holland.

Alesina A., La Ferrara E. (2005) Preferences for redistribution in the land of opportunities. Journal of Public Economics 89: 897–931.

Bénabou R., Tirole J. (2012) Laws and Norms. IZA Discussion Paper n°6290, 47 p.

Blaufus K., Hundsdoerfer J., Jacob M., Sünwoldt M. (2016) Does legality matter? The case of tax avoidance and evasion. Journal of Economic Behavior & Organization. 127 (C): 182-206

Carbonara E, Parisi F, von Wangenheim G. (2012) Unjust laws and illegal norms. International Review of Law and Economics, 32 (3): 285–299.

Clark A. E., D'Ambrosio C. (2015) Attitudes to Income Inequality: Experimental and Survey Evidence. In: A. B. Atkinson and F. Bourguignon (eds) Handbook of Income Distribution, Volume 2A, Chapter 13: 1147-1208.

Cooter R. (1998). Expressive law and economics. Journal of Legal Studies 27: 585–608.

Cooter R. (2000). Do good laws make good citizens? An economic analysis of internalized norms. Virginia Law Review 86: 1577–1580.

Costa-Font J., Cowell, F. (2015) Social identity and redistributive preferences: a survey. Journal of Economic Surveys 29 (2) 357-374.

Durante R., Putterman L., Weele van der J. (2014) Preferences for redistribution and perception of fairness: an experimental study. Journal of the European Economic Association 12 (4), 1059–1086.

Fong C. (2001) Social preferences, self-interest, and the demand for redistribution. Journal of Public Economics, 82: 225-246.

Keely L., Tan C. M. (2008) Understanding preferences for income redistribution. Journal of Public Economics 92: 944–961.

Kotsadam, A., Jakobsson, N., 2011. Do laws affect attitudes? An assessment of the Norwegian prostitution law using longitudinal data. International Reviewof Law and Economics 31 (2), 103–115.

Ku H., Salmon T. C. (2013) Procedural Fairness and the tolerance for income inequality. European Economic Review 64: 111-128.

Licht A. N., (2008) Social Norms and the Law: Why Peoples Obey the Law? Review of Law and Economics 4 (3): 715-750.

Piketty T. (1995) Social mobility and redistributive politics. Quarterly Journal of Economics 110 (3): 551–583.

Posner E. (2000) Law and Social Norms. Cambridge (MA), Harvard University Press.

Rodriguez-Lara I., Moreno-Garrido L. (2012) Self-interest and fairness: self-serving choices of justice principles. Experimental Economics 15: 158–175.

Schokkaert E, Devooght K. (2003) Responsibility-sensitive fair compensation in different cultures. Social Choice and Welfare 21: 207–242.

Skitka L. J. (2003). Of different minds: An accessible identity model of justice reasoning. Personality and Social Psychology Review, 7 : 286-297

Tyran J-R., Feld E. (2006) Achieving Compliance When Legal Sanctions are Non-Deterrent, Scandinavian Journal of Economics 108 (1): 135-156.

Tyran J-R., Sausgruber R. (2006) A little fairness may induce a lot of redistribution in democracy. European Economic Review 50 (2): 469–485.

Wailoo A., Anand P. (2005) The nature of procedural preferences for health-care rationing decisions. Social Science and Medicine 60: 223-236.

Yaari M. E., Bar-Hillel M. (1984) On dividing justly. Social Choice and Welfare 1: 1-24.