«Urs Fischbacher Gerald Eisenkopf Franziska Föllmi-Heusi Learning and Peer Effects Unequal Opportunities and Distributive Justice Research Paper ...»
No. 57 september 2010
Learning and Peer Effects
Unequal Opportunities and
Research Paper Series
Thurgau Institute of Economics and Department of Economics
at the University of Konstanz
Unequal Opportunities and Distributive Justice
University of Konstanz &
Thurgau Institute of Economics
Post Box 131
Contact: email@example.com April 6, 2011
We provide experimental evidence on how unequal access to performance enhancing education affects demand for redistribution. People earn money in a real effort experiment and can then decide how to distribute it among themselves and another subjects. We compare situations in which randomly chosen people get access to performance enhancing education with situations in which either only luck or only performance determines outcome. We find that unequal opportunities evoke a preference for redistribution that is comparable to the situation when luck alone determines the allocation. However, people with unequal access to education are more likely to disagree about the appropriate distribution.
Keywords: Distribution, Inequality of opportunities, Negotiation, Education, Experiment JEL-Codes: D03, D31, I20 1 Introduction How do people redistribute if inequality is caused by unequal access to education? A huge literature shows that people are more willing to accept inequality in incomes if it results from hard work rather than from pure luck.1 Education is ambiguous in these dimensions. Random processes like high innate abilities or a favorable socio-economic environment enhance the chances to get education but the student herself still has to provide effort in order to acquire and improve her skill. Some students study hard but others relax. Furthermore, distorted beliefs confuse the assessment of distributional preferences. People may argue in favor of redistribution because they prefer equal incomes or because they incorrectly perceive the access to education as unfair. Alesina and Glaeser (2005, p. 5) argue that beliefs do not reflect the actual (in)equality of opportunities correctly. Instead, people base their beliefs on personal experiences rather than on econometric studies (Piketty, 1995) and they have a biased perception of these experiences (Benabou and Tirole, 2006). Therefore, our study uses an experiment to shed light on the question how people evaluate unequal access to education.
We investigate the demand for post-educational redistribution with a real-effort experiment. In our experiment, subjects are paired in groups of two. In a quiz task they create an output, which they contribute to a common pool. Then, they negotiate how to distribute their joint output. In all treatments, subjects get the opportunity to learn some of the questions of the quiz. Our focus is on the education treatments in which one of the two subjects in a group gets a better education because she can learn more relevant questions. As the number of correctly answered general knowledge questions determines a subject s contribution, knowledge was the relevant skill in this experiment. However, since one randomly chosen subject in each group in the education treatments received additional knowledge, it is obvious that luck was also relevant for contributions. The two education treatments differed with respect to the learning time. With short education subjects had to concentrate more in order to reap the benefits of the learning advantage. Long education allowed them to learn rather leisurely. We used two benchmark treatments in which we controlled the importance of skill and luck. As one benchmark we use a treatment, the skill treatment, in which a subject s contribution depends only on her ex-ante skills. A second benchmark is provided by the luck treatment, in which a lottery determines the contribution.
See for example the studies by Hoffman et al. (1994), Burrows and Loomes (1994), Ruffle (1998), Konow (2003), or Durante and Putterman (2009).
The control treatments in our study relate to two principles of distributive justice, the egalitarian one and the desert-based one. Strict egalitarianism advocates the allocation of equal material goods to all members of society (Lamont and Favor, (2007) in the online version of the Stanford Encyclopedia of Philosophy). According to the desert principle, people should be rewarded according to the value of their contribution to the social product.2 This means that this principle and similar meritocratic ideas have a concept of equality of opportunity rather than equality of outcome. If unequal opportunities exist then it is important to identify if and how far a person is accountable for an outcome. Roemer (1998) argues that accountability requires a comparison of people with the same exogenous characteristics.3 The share that a person gets should only increase in her relative contribution, i.e. the share of an educated person should be measured in relation to her educated peers and the share of an uneducated persons should be measured relative to other uneducated persons. Since education in our experiment is randomly assigned, educated people should, on average, receive the same share as uneducated people even if they contribute more than uneducated people.
As mentioned above, several studies have shown that people opt for more egalitarian distributions once luck rather than meritocratic criteria determine an outcome (see footnote 1).
Cappelen et al. (2007) used distribution decisions after an investment period with unequal rates of returns and investigated the importance of different fairness principles. They provide evidence for heterogeneity in the application of fairness principles among their subjects.
Konow (2000; 2003) provides detailed positive analyses on the accountability principle.
In our study, we implement unequal access to education, which creates an ambiguous situation with respect to these fairness principles. On the one hand, luck is relevant for the unequal access to education; on the other hand, performance alone determines the outcome after education has been received. Thus, with our study we can assess the importance of different fairness principle in this situation. We can investigate whether the fairness norms that are applied in situations with equal opportunities or purely randomly determined investments prevail in a situation of inequality of opportunities,4. Such a comparison reveals whether people make claims for more or less redistribution once they have correct information about the determinants of inequalities in opportunities. This comparison is It is important to distinguish between the desert principle and the provision of incentives. The latter implies a provision on the distribution of outcome before production has taken place while the former considers a distribution after production has taken place.
I say it is morally wrong to hold a person accountable for not doing something that it would have been unreasonable for a person in his circumstances to have done (p. 18).
Inequity averse people in the sense of Fehr and Schmidt (1999) do not accept inequality in outcomes even if the differences depend on choices only. On the other hand, libertarian thinkers such as Hayek (1960) are reluctant to accept redistribution even if luck has a strong impact on economic outcomes.
particularly important in the context of inequalities in the access to education. Educational choices depend on skills (or abilities) which are, at least to a certain degree, exogenous, unobservable and unevenly distributed productivity factors. Nevertheless, skill premiums are widely tolerated and meritocratic societies claim that the most able citizens do constitute their elite.
We expected redistribution in the situation of unequal opportunities to be in between the control treatments, in which luck or skill alone determines outcome. Interestingly though, our results reveal that subjects responses to unequal learning opportunities are similar to their responses when luck alone determines output. This means that if the access to education is saliently due to luck, people apply more egalitarian than desert based fairness principles. We also observe conflicting distribution norms between educated and uneducated participants.
The paper is structured as follows. The following section presents the experimental design.
Afterwards, we provide behavioral predictions. Section 4 presents the results of the experiment. Section 5 summarizes the paper and provides concluding comments.
2 Experimental design
We start with an overview of the experiment and, then explain all the steps in detail. The key feature of our design is the generation of unequal access to education. In our experiment, subjects have to solve multiple choice knowledge questions. As a preparation, they can learn some of them. We generate unequal access to education by differentiating how many of the questions are useful, i.e., how many of the questions are relevant in the real effort task. In order to manipulate the importance of the education for performance, we implemented two situations. In one situation, the long education treatment subjects could learn for 15 minutes, while in the short education treatment, subjects could learn for only 4 minutes. The latter condition creates higher variance within the educated group. Thus, high performance within the educated group is less associated with luck and could be considered as more deserved.
The production in the multiple choice question task determined the subject s contribution to a common pool. Subjects were informed about each other s contribution and could bargain how to share the common pool. The control treatments differ in how the contribution to the common pool is determined: In the skill treatment, the individual contribution of a subject to the joint output was determined by her skills (more specifically her general knowledge). In the luck treatment, luck determined the individual contribution. We will now present the three phases of the experiment in detail. The phases are, the learning phase, the production and contribution phase, and the negotiation phase.
Learning All subjects learned the correct answers for 60 knowledge questions. We used multiple choice versions of questions from the German standard version of the quiz game Trivial Pursuit which includes questions on geography, entertainment, history, arts and literature, science and technology as well as sports. The learning phase lasted for 15 minutes. In the short education treatment this time was reduced to 4 minutes.
In the learning phase, subjects could learn the correct answer to 60 questions. In this phase only the correct answer was displayed- The treatments differed in how many of the learned questions were relevant in the production phase. In the skill treatment, 5% of the questions from the learning period (i.e. 3 out of 60) reappeared in the production period. In both education treatments, one member in each group had learned 5% of the relevant questions while the other one had learned 95% (i.e. 57 out of 60 questions). In the luck treatment, each subject learned 50% of the relevant questions. In the skill and the luck treatment, the subjects were informed about the number of relevant questions at the beginning of the learning period. In the education treatments, the subjects were initially informed about the possible number of relevant questions. The actual assignment of the number of relevant questions and the information of the subjects occurred immediately after the learning period via the throw of a die.
Production and contribution to the common pool In the production phase lasted for 15 minutes in all treatments. Each subject had to answer 60 knowledge questions by choosing between 4 possible answers. Only one of the answers was correct. As Trivial Pursuit provides only the correct answers, the authors of this paper developed the alternatives on their own. The experiment included two payment components.