xThe AEA RCT Registry will be down for maintenance on Tuesday, April 20th, from 9pm EDT to 11pm EDT to perform necessary upgrades. The site will be in maintenance mode during that time. We apologize for any inconvenience this may cause.
Making and Breaking Coalitions: The Influence of Prosociality and Rationality
From a traditional rational choice perspective, coalitions are inherently unstable if the collective decision involves some kind of distributional conflict. While traditional explanations for the stability of coalitions refer to institutions, more recent theoretical developments argue that behavioral traits like actors' prosociality and rationality have important effects. The aim of this project is to provide a first empirical test of this theoretical claim. In laboratory experiments, we first gather information on subjects' prosociality and rationality. Then subjects are matched into three-person groups based on their revealed behavior in the previous part of the experiment and then play a coalition formation game in continuous time.
External Link(s)
Citation
Kittel, Bernhard, Jan Sauermann and Manuel Schwaninger. 2018. "Making and Breaking Coalitions: The Influence of Prosociality and Rationality." AEA RCT Registry. April 06. https://doi.org/10.1257/rct.2859-1.0.
We measure the time coalitions are maintained in a three-person bargaining experiment.
Secondary Outcomes (end points)
outcome distributions
Secondary Outcomes (explanation)
We measure which kind of coalitions are formed and how they distribute the payoff among the group members.
Experimental Design
We expect that more egalitarian (i.e. more prosocial) and more forward looking (i.e. more rational) subjects are more likely to form and maintain stable coalitions. To test these hypotheses, we study two treatments, in which we vary between two matching mechanisms. In the first part of the experiment, prior to the bargaining stage, all subjects play a dictator game and a beauty contest game. In the dictator game, we study groups consisting of three subjects and use the strategy method to elicit distributional preferences of all three subjects. The amount kept as a dictator serves as an estimate of individuals’ prosociality, while the number stated in the beauty contest approximates rationality. Subjects do not receive information about their payoff from the tasks, nor do they receive information about the action of other participants until the end of the experiment.
In the second part of the experiment we use a slightly modified version of the unstructured bargaining game with continuous payments developed by Tremewan and Vanberg (2016). Subjects are matched into three-person groups and negotiate over the distribution of payoffs. Negotiations last for at least 5 minutes, and groups bargain in continuous time. Hence, at any point in time subjects can propose and agree on new distributions. Payments flow in one-second intervals if at least two group members agree on a distribution of benefits. Without agreement of a majority of group members no payoffs are allocated. Maximally, the whole bargaining process lasts for 10 minutes while payments flow no more than 5 minutes. Hence, delay is costless if the time without majority agreement does not exceed 5 minutes.
The bargaining game has a two-dimensional graphical interface. Subjects can offer and agree on allocations by clicking on circles in a triangle, which represent a finite set of feasible outcomes. Each corner of the triangle represents a player and the closer a circle to the player, the higher the respective payoff. The design essentially excludes any proposer power, giving equal bargaining power to all three subjects. During the whole bargaining game, subjects are informed about their payoffs and the remaining time.
The two treatments differ with respect to the composition of groups. In Treatment 1, we match three subjects into groups based on their behavior in the dictator game in the first part of the experiment. We maximize between-group variance by matching the most egoistic subjects into groups and the most prosocial subjects into other groups. In Treatment 2, we follow the same procedure, by matching groups according to the stated number in the beauty contest game. If the prosociality and rationality of subjects affect the stability of coalitions, subjects who transferred higher amounts in the dictator game or stated lower numbers beauty contest should be more likely to form stable coalitions.
Experimental Design Details
Randomization Method
computer (ztree)
Randomization Unit
individual (experimental participants)
Was the treatment clustered?
Yes
Sample size: planned number of clusters
60 groups
Sample size: planned number of observations
180 participants
Sample size (or number of clusters) by treatment arms
30 groups in each treatment
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)