Heterogeneity in the Motivations for Punishment: An Experimental Study

Description

We explores the motivations behind costly punishment in social dilemmas, specifically focusing on reciprocity and inequality aversion. To distinguish these motivations, we introduce uncertainty into the payoff function of a public goods game, preventing participants from predicting others' contributions based on their payoffs. In this situation, participants should choose between others' contributions and others' payoffs as the criterion for their punishment. We also include a random income game to test the plausibility of our classification by removing intentionality from the contribution decisions. Our results show heterogeneity in the motivations to punish, leading to different types of punishers: the self-interested type, the reciprocal type, the inequality-averse type, and the ``other" type who is inconsistent. Furthermore, in the random income game, overall punishment tends to be lower, but only reciprocal punishment decreases with the existence of intention. Additionally, the reciprocal type strongly punishes free-riding behaviors but also imposes some punishment for payoff inequality. These findings reveal that inequality aversion is a critical motivation for punishment---that is, that a significant number of individuals rely on inequality aversion as their sole motivation for punishment while others incorporate it into their punishment based on reciprocity.

Steps to reproduce

Please read [Read_me.txt]. First, to estimate the type of each subject, open [Base_all_rounds.dta] and input codes in the [Read_me.txt]. And then, we determine the types of subjects through the following process: 1) We select coefficients with high significance levels with the lowest p value in both regression results. 2) When two or more coefficients have the same significance level of p = 0.0000 in both models, we compare their magnitudes and take highest coefficient. By using [Base(3,5,7)_RI.dta] and seconds codes in the [Read_me.txt], we can perform linear regressions of Table 3 in the paper.

Institutions

Categories

Funding

Licence