The fair division of losses
Economic Theory
Final Report Abstract
In the project “The fair division of losses,” we conducted experiments investigating fairness behavior when subjects have to divide losses. To model fairness, we used the dictator game (DG) in every case, in several versions, and, in most cases, the ultimatum game (UG) as well. In the negative versions of the games, subjects divided a loss of €-10. All negative versions were compared to a standard version with a gain of €10. The experiments were carried out at the Magdeburg Experimental Laboratory of Economic Research and the Leipzig Experimental Laboratory for Social Science. In total, N=1,179 subjects participated in the experiments. Experimental Designs: Effect of loss treatment: In a first series (reference experiments) conducted in 2018, we ran baseline treatments of our experiments on losses; that is, we conducted experiments using a negative DG and a negative UG and compared the results to standard versions of the games over gains. In all experiments where both games were played, the order of DG and UG was randomized over sessions. The results of these baseline experiments were later used as control treatment for the other experimental settings and analyses. Normative expectations: In a second series of experiments, we investigated the role of norms in the behavior of the participants within the four distribution situations considered, namely DG over gains or losses and UG over gains or losses. These experiments were run in 2019. We used the same experimental setting as in the reference experiments, but changed the instructions by adding a norm focus. In one treatment, we reported the average amount given in the reference experiments to set a descriptive norm of what was usually done in the two games. In a second treatment, we focused participants on the game theoretical solution to the games. Since a fairness norm is an injunctive norm, there is always an incentive for the subjects to deviate from it and decide in a self-interested manner. By focusing on selfinterested behavior, we tested the robustness of the fairness norm. Increased anonymity: For social norms, such as fairness norms, social control is an important factor in compliance. Increased anonymity might lead to a decreased feeling of social control and thus to less fair decisions. In 2021, we replicated the reference experiments in an online setting. Conducting the experiments online might have increased anonymity, as the participants met neither the other participants nor the experimenter physically, and, because we used a webcam-off protocol, they had no visual contact throughout the experiment. This increase in anonymity might lead to an (additional) decrease in fairness behavior. Expectations of a third party: In 2021, we also ran a version of the DG with an additionally implemented punishment mechanism. The so-called third-party punishment DG consists of two stages: first, the standard DG over a loss or gain, and second, a stage in which a third party is shown the decision of the dictator and has the option to punish her. The punishment is costly, and we varied the strength of the punishment mechanism. Methodological question for experiments on losses: To induce losses, we used a prepaid mechanism: The money was paid at a first session two weeks prior to the experimental session. We used several questions to check the functionality of this prepaid mechanism. Main Findings. Across several experiments, we found an effect of the loss condition. Dictators, who divide a loss instead of a gain, become more self-interested. Hence, our data show a diminishing effect on fairness when losses have to be divided. However, we found only small effects for the norm-focusing treatments. Focusing on the average giving in the DG led to a smaller variance in the dictators’ offers. In the UG, however, the focus on the self-interested solution decreased fairness. Interestingly, both effects were found to be more evident in the loss condition than in the gains condition. We did not find a difference in behavior between laboratory and online settings, which means that the online experiments’ results were comparable to those obtained in the laboratory. The punishment mechanism introduced in the third-party punishment DG clearly showed an increasing effect on fairness, but only if the mechanism was strong enough. Another interesting finding is that the effect of giving less under the loss than under the gains condition vanished when the punishment mechanism was added to the DG. Regarding the methodological question, we found evidence that a meaningful proportion of participants in the loss treatment did indeed believe themselves to be in a loss frame when they played the negative games.
Publications
- (2017). Ultimatum bargaining over losses and gains– An experimental comparison. Social Science Research, 67, 49-58
Neumann, T., Schosser, S., & Vogt, B.
(See online at https://doi.org/10.1016/j.ssresearch.2017.08.009) - (2018). How to split gains and losses? Experimental evidence of dictator and ultimatum games. Games, 9(4), 78
Neumann, T., Kierspel, S., Windrich, I., Berger, R., & Vogt, B.
(See online at https://doi.org/10.3390/g9040078) - (2018). Non-acceptance of Losses—An Experimental Study on the Importance of the Sign of Final Outcomes in Ultimatum Bargaining. In Operations Research Proceedings 2017 (pp. 287-292). Springer, Cham
Neumann, T., Schosser, S., & Vogt, B.
(See online at https://doi.org/10.1007/978-3-319-89920-6_39) - (2020). Der Einfluss von Normen in Verlustexperimenten. Arbeitsberichte des Instituts für Soziologie, Nr. 78, Leipzig
Windrich, I., Berger, R., Kierspel, S.; Neumann, T., & Vogt, B.
- (2022). Experiments on norm focusing and losses in dictator games. Frontiers in Sociology, 7:930976
Windrich, I., Kierspel, S., Neumann, T., Berger, R., & Vogt, B.
(See online at https://doi.org/10.3389/fsoc.2022.930976)
