Journal of Economic Behavior Organization Vol. 43 2000 75–89
Entitlements and fairness: an experimental study of distributive preferences
E. Elisabet Rutström
a
, Melonie B. Williams
b ,∗
a
Department of Economics, Darla Moore School of Business, The University of South Carolina, Columbia, SC, USA
b
US Environmental Protection Agency, Mail Code 2172, Ariel Rios Building, 1200 Pennsylvania Avenue NW, Washington, DC 20460, USA
Received 11 January 1999; received in revised form 14 February 2000; accepted 23 February 2000
Abstract
Under three different rules for allocation of initial income we elicit experimental subjects’ pref- erences for income redistribution using an incentive compatible elicitation mechanism. The three
income allocation rules are designed to capture preferences for distributive justice among sub- jects. The concern is motivated by claims in some of the experimental economics literature that
non-self-interested motives often underlie individual behavior. We cannot reject self-interest in fa- vor of any redistribution motives based on our observations. Almost all individuals chose the income
distribution which maximized their own income — high income individuals chose no redistribution and low income individuals chose perfect equality in income distribution. © 2000 Elsevier Science
B.V. All rights reserved.
JEL classification: D3; D63
Keywords: Equity; Distribution; Justice; Fairness; Experiments
1. Introduction
In this study we elicit preferences over income distribution in an incentive compatible way and test how such preferences relate to some simple notions of earnings-based justice. We
employ a controlled laboratory setting with no income uncertainty and no strategic consid- erations. Under such conditions, any revealed preferences that are non-payoff-maximizing
would be inconsistent with the traditional notion of self-interest, where the individual agent’s
∗
Corresponding author. Tel.: +1-202-260-7978; fax: +1-202-260-7875. E-mail address:
williams.melonieepa.gov M.B. Williams. 0167-268100 – see front matter © 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 7 - 2 6 8 1 0 0 0 0 1 0 9 - 8
76 E.E. Rutström, M.B. Williams J. of Economic Behavior Org. 43 2000 75–89
utility is independent of the utilities of other agents. Our results strongly favor self-interest and stand in stark contrast to previous experimental evidence.
Our interest in the question of individual preferences over income distribution arises from the discrepancy between theoretical predictions and previous experimental findings.
Mainstream political economy models predict payoff-maximizing behavior based on the assumption that agents are motivated by self-interest. Simply put, within these models
individuals’ decisions over economic policies do not reflect any preferences that involve a regard for others, whether based on justice or some other consideration. If this assumption
is correct, then any preferences over income distribution that are revealed among members of a group must be reflections of each member’s desire to maximize his own income and not
of any desire to satisfy others’ wants or needs or to respect others’ rights. Consistent with the assumption of self-interested economic decision-making, several theories of distributive
justice and social welfare further assume that decisions regarding what distribution is just can only be made by a representative individual in some “original position”, where the decision
maker by assumption would not be biased by his own self-interest.
1
Such decisions are sometimes referred to as being made behind a veil of ignorance. Individuals not positioned
behind such a veil of ignorance are therefore expected to always act in a manner that disregards justice.
Laboratory experiments have been employed to test whether economic agents are gen- erally motivated by self-interest. Based on observing subjects choosing to redistribute
income, many such studies have concluded that experimental subjects do not appear to be motivated solely by self-interest, and that this is true in many economic choice situ-
ations and not only behind a veil of ignorance.
2
Several experimenters now favor no- tions of fairness or altruism, in addition to monetary payoffs, as motivating individual
choice.
3
Revealed preferences over income distributions, however, appear to be very sensitive to variations in experimental design. Several experimental studies document results sug-
gesting that the origin of initial entitlements has an effect on the extent of apparently non-self-interested behavior;
4
when subjects are required to earn their initial entitlements, the frequency of apparently non-self-interested behavior is lower than when initial entitle-
ments are allocated according to pure chance. Hoffman and Spitzer 1985, p. 260 propose that experimental subjects behave as if they believe in an earnings-based notion of justice;
an unequal income distribution is unjust when initial entitlements are randomly assigned earned. Such results appear to be consistent with equity theory in social psychology, which
1
See, for example, Rawls’ 1971 definition of the “original position”, and Harsanyi’s 1955 distinction between “impersonal social considerations” and “subjective” preferences. Binmore 1998 employs the process of the
original position for the renegotiation of the social contract among players in the game of morals.
2
The idea that revealed preferences may reflect moral considerations is not new. Harsanyi offers a definition of subjective preferences that may incorporate moral considerations. Binmore 1994, 1998 offers an alternative
definitional classification of preferences into self-interest narrowly conceived and self-interest broadly conceived. He develops these ideas into a complete theory of revealed ethical preferences. See Andreoni and Miller 1998 for
a test confirming that distributive preferences obey the Generalized Axiom of Revealed Preferences, suggesting that there may be a trade-off between self-interest and justice motives.
3
See, for example, Güth et al. 1982, Rabin 1993, Fehr et al. 1993, and Eckel and Grossman 1995, 1996.
4
We review two such studies below: Hoffman and Spitzer 1985, and Burrows and Loomes 1994.
E.E. Rutström, M.B. Williams J. of Economic Behavior Org. 43 2000 75–89 77
is a theory of earnings-based justice.
5
According to equity theory, payments to subjects in experiments get allocated in a way that reflects the subjects’ “intrinsic inputs”. The
definitions of “intrinsic input” vary and may include the time spent, the amount of work completed, and also various personal characteristics such as intelligence and social status.
6
Some of these definitions capture the notion that the employment of inputs in the task car- ries an opportunity cost and therefore justifies a compensation. We refer to this aspect as
the resource cost of the input. Other definitions capture the notion that input use leads to expansions in output which rightfully should belong to the producer. In fact, some measures
of “intrinsic inputs” focus directly on the amount produced rather than on the inputs used. We refer to this aspect as the productivity of the input.
We propose a test, not only of the existence of distributive preferences, but also of whether revealed preferences over income distributions are robust with respect to variations in some
measure of the resource cost which we will call “effort” separately from variations in some measure of the productivity of the resources used. We therefore perform the following joint
hypothesis test: H
1
individuals have preferences over income distribution that differ from their own payoff maximization, and H
2
preferences over income distribution depend on how individuals perceive others’ “worthiness of compensation” as indicated by either the
amount of effort used or by the amount of production performed in the task. For example, worthiness based on productivity could very well depend on how much effort was needed.
In cases where productivity and effort cannot be separately observed, however, it is an open question whether it is the effort per se or the resulting production that determines worthiness.
That is, it is possible that even unproductive effort can be considered worthy. We therefore add the following sub-hypotheses to H
2
: H
2 ,
1
the effort used will warrant a compensation even when the effort is unproductive and H
2 ,
2
productivity, even with relatively little effort, will be considered worthy of compensation. Later, we will refer to the distributive
preferences expressed under the two sub-hypotheses as effort-based and productivity-based preferences. We test these hypotheses by employing a laboratory task with which we can
define and measure effort and productivity in a separable way. The task we employ is the Tower of Hanoi puzzle. Many other tasks that have been used in the literature, such as spell
correction tasks, do not allow the observer to distinguish between effort and productivity.
With the exception of dictator games and majority voting games, all experimental ob- servations of redistribution choices have been performed using institutions where strategic
considerations are important. We want to study individual preferences directly and there- fore our experimental design controls for strategic considerations. Given our interest in
eliciting individual preferences over income distribution, we also control for differences in decision-making power as exist in dictator games, which otherwise could provide a
motivation for redistribution. In our experiment both advantaged and disadvantaged agents are able to improve their own position at the expense of the other agents. Therefore, we
collect choice data not only from well endowed high income subjects, following previous dictator games, but also from poorly endowed low income subjects. Finally, we find it less
interesting to study situations with trivial monetary consequences where the opportunity
5
The origin of equity theory is found in Adams 1964 and Homans 1974. Much of equity theory has been developed in the context of industrial relations, see, e.g., Akerlof 1982 and Akerlof and Yellen 1990.
6
Hoffman and Spitzer 1985, p. 265.
78 E.E. Rutström, M.B. Williams J. of Economic Behavior Org. 43 2000 75–89
cost of revealing justice-based preferences is very low. It is more informative to see if such behavior survives at higher opportunity costs.
The basic experimental design involves two phases. In Phase I subjects are assigned a task which will determine their initial income entitlement in Phase II. They work at this task
individually. In Phase II subjects are brought together into groups of 12. Each subject is informed about the value of his initial income entitlement and how this has been determined.
He is then asked to choose the distribution rule that results in his preferred final distribution of income. Hence, subject decisions should not be influenced by risk attitudes as decisions
are made under complete information regarding income entitlements. Finally, to determine which distribution rule will determine the payoffs for the 12 subjects in the group, we
employ an incentive compatible mechanism called the Random Dictator rule under which everyone has the same chance of dictating the outcome and strategic considerations are
eliminated.
Our results are surprising in light of previous experimental evidence. We find that we cannot reject the hypothesis that self-interest is the sole motivation for subject behavior
in either of our earnings treatments. In our sample, 99 percent of subjects chose the dis- tribution rule which maximized their own final payoff. Because of the predominance of
apparently self-interested behavior, we implement an additional robustness test, comparing behavior under earned entitlements to behavior under a random entitlement mechanism. In
this session, we observe a slight increase in apparently non-self-interested behavior, albeit not in a way entirely consistent with earnings-based distributive preferences. The difference
is significant in the statistical sense, though the absolute magnitude is small. Despite this slight support for non-self-interested behavior, we conclude that the model of self-interest
in individual decision-making can explain our data very well.
The following section presents an overview of related literature. Section 3 details the experimental design, while the results are presented in Section 4. Conclusions and discussion
follow in Section 5.
2. Previous findings