Fairness and Excuse Seeking-Behavior

2023-04-21

2023-04-28

I collect the following primary outcomes:
- Distribution to himself and to partner
- Beliefs about the probability of getting high and low piece rate for self and partner.
- Implemented inequality

Below we outline how we will use our primary outcomes and the key hypothesis.

Hypothesis 1: under uncertainty (partial information treatment) participants will distribute more to themselves in comparison with full information.

Hypothesis 2: participants will report a higher probability that their partner got the high piece rate in comparison to the Bayesian belief, and a lower probability that they got the high piece rate.

Hypothesis 3: Inequality implemented will be higher under partial information when the decision maker has the higher earnings of the pair, and inequality will be lower under partial information when the decision maker is the low earning of the pair.

Distribution to himself and to partner will be measured both in absolute and relative terms.

Implemented inequality will be measured as the relative difference in earnings after redistribution for each decision.

Participants perform a task and the money they earn depends on effort and luck. Then they are paired and distribute the sum of earnings between themselves. Two treatments, one with full information, and another one where participants do not know the role luck and effort had in determining their earnings.

The experiment will be conducted at the Pittsburgh Experimental Economic Laboratory. Subjects will participate in one session of an hour that consists of two parts.

In the first part of the experiment, participants perform the counting zero task of Abeler et al. (2011), where they see a 15x10 table filled with 0 or 1, and they have to count how many 0 are; after they submit an answer, a new table appears. Participants will work on this task for 25 minutes. Participants are paid according to a piece rate, where there are two possible piece rates with 50% each. In the high piece rate, participants get 50 cents for each table counted correctly, and in the low piece rate, they get 50 cents for 3 correctly counted tables. At the moment of performing the task, participants do not know which piece rate they have. While they are performing the task, they do not know if their answer is correct or incorrect.

Before starting the task, participants are told the payment earn in this part is provisional and the payment from the experiment will depend on the decisions made by them or others from the second part. They answer comprehension questions to make sure they know this.

In the second part, the participants are paired and have to decide how to distribute the sum of his and his partner's earnings between each other. Both participants of the pair will be making this decision, and one of them will be implemented. There are two treatments, full information, and partial information.

For the full information treatment, when making their decisions, participants will know the piece rate, the number of correctly counted tables, and the earnings of him and his partner.

In the partial information treatment, participants will not know the piece rate each of them got and neither the number of correctly counted tables. They will only know the earnings of each of them. In this treatment, before making the distribution decision, participants will be asked their beliefs about the piece rate he got and the piece rate his partner got. The beliefs are incentivized using the binarized scoring rule and following the recommendations of Danz et al. (2022). Participants will be provided with a histogram distribution of corrected counted tables from the first 3 sessions of the full information treatment, run on April 21, 2023. The elicit beliefs will be compared with the Bayesian belief generated by this distribution.

For both treatments, participants will be making distributive decisions (and belief elicitations, in the partial information treatment) for 11 different scenarios. 10 of these scenarios are hypothetical and one is the real one that corresponds to the information of the partner. The participants do not know which scenario is the real one, in the survey at the end of the experiment, they are asked which scenario they thought is the real one. The 10 hypothetical situations are the same for every participant in any treatment, where 5 of the hypothetical situations correspond to partners that got the high piece rate, and the other 5 to the low piece rate. The effort level was chosen such that most of the participants had to make decisions for earnings and effort levels above and below their own, and at the same time they have to be attainable effort levels. To determine the effort for the hypothetical situations, the data from Zimmermann (2020) and Abeler et al. (2011) was used.

Finally, the experiment finishes with a demographic survey.

The first 3 sessions of the full information treatment were conducted on April 21. The partial information treatment and one more session for the full information treatment will be conducted from April 27 to April 31.

References:
Abeler, Johannes, Armin Falk, Lorenz Goette, and David Huffman. 2011. "Reference Points and Effort Provision." American Economic Review, 101 (2): 470-92.
Danz, David, Lise Vesterlund, and Alistair J. Wilson. 2022. "Belief Elicitation and Behavioral Incentive Compatibility." American Economic Review, 112 (9): 2851-83.
Zimmermann, Florian. 2020. "The Dynamics of Motivated Beliefs." American Economic Review, 110 (2): 337-61.

Randomization of piece rate is done via the computer, in oTree.
Randomization between treatments will be based on which sessions participants sign up for. Where the first 3 sessions (already run) were of the full information treatment.

The randomization of the piece rate is done by individual.
The randomization of treatment is done at session level.

No

No Clusters

I expect to have 200 participants.

I plan to have 80 participants in the full information treatment and 120 participants in the partial information treatment.

The effect of the treatment will depend on the effort performed by participants, the distortion of beliefs, and the fairness views of participants (the decisions of libertarian and egalitarian should not be affected by the treatment). I have decided to have 200 participants, 80 in the full information treatment and 120 in the partial information treatment. By running 1000 simulations with this sample size, in the partial information treatment, the p-value of participants distributing more to themselves is significant at the 0.05 level. To estimate these simulations, the performance of the real effort task is simulated from a normal distribution with mean 22.5 and standard deviation 8.5. This is based on data from Zimmermann (2020) and Aberler et al. (2011). The piece rate is chosen with 50% chance. Following Cappelen et al. (2022) approximately 15% are libertarian, 5% egalitarian, and 80% meritocrat. And I assume the beliefs of meritocrats about their partner’s piece rate will be on average 40% higher than the Bayesian prediction with standard deviation 0.1 (following the results of Di Tella et al. (2015)). References: Abeler, Johannes, Armin Falk, Lorenz Goette, and David Huffman. 2011. "Reference Points and Effort Provision." American Economic Review, 101 (2): 470-92. Cappelen, Alexander W., Thomas de Hann, and Bertil Tungodden. 2022. "Fairness and limited information: Are people Bayesian meritocrats?" SSRN Electronic Journal. Di Tella, Rafael, Ricardo Perez-Truglia, Andres Babino, and Mariano Sigman. 2015. "Conveniently Upset: Avoiding Altruism by Distorting Beliefs about Others' Altruism." American Economic Review, 105 (11): 3416-42. Zimmermann, Florian. 2020. "The Dynamics of Motivated Beliefs." American Economic Review, 110 (2): 337-61.