Via Freedom to Coercion: The Emergence of Costly Punishment

forthcoming in Public Choice, 2021

Does the desirability of social institutions for public goods provision depend on the extent to which they include mechanisms for endogenous enforcement of cooperative behavior? We consider alternative institutions that vary the use of direct punishments to promote social cooperation. In one institution, subjects participate in a public goods experiment in which an initial stage of voluntary contribution is followed by a second stage of voluntary, costly sanctioning. Another institution consists of the voluntary contribution stage only, with no subsequent opportunity to sanction. In a third stage subjects vote for which institution they prefer for future interactions: do they prefer one that does allow sanctions or one that does not allow sanctions? Our results show that even though sanctions are frequently used when available, the clear majority of individuals vote for the institution that does not allow sanctions. Thus, a distinction is required between the principles that guide the choice of institutions and the principles that apply to actions guided by institutions. Our results indicate that it is the wealth generated by the institution that determines its desirability.

MAGKS Joint Discussion Paper Series in Economics, 2020

We use an experiment to test the hypothesis that groups consisting of like-minded cooperators are able to cooperate irrespective of punishment and therefore have a lower demand for a costly punishment institution than groups of like-minded free riders, who are unable to cooperate without punishment. We also predict that the difference in the demand for punishment is particularly large when members know about the composition of their group. The experimental results confirm these hypotheses. However, the information about the composition of the group turns out to be even more important than we expected. It helps cooperative groups to avoid wasting resources for an unneeded punishment institution. In uncooperative groups, it helps members to recognize the need for punishment early on and not to follow an uncooperative path that produces a persistently competitive attitude. These findings highlight the role of group composition and information for institution formation and that lessons learned by one group cannot be readily transferred to other groups.

Evidence from field and laboratory experiments indicates that a large fraction of the people behave like conditional cooperators in public good games. In this article, I investigate the implications of the existence of both conditional cooperators and egoists for optimal law enforcement strategies. When norms of cooperation exist between conditional cooperators, sanctions set by an authority can be lower than in a “Hobbesian" setting where everybody is egoistic. Moreover, if the authorities have private information about the fraction of egoists in society, low sanctions can be optimal because they signal that there are few defectors and thus “crowd in" trust and cooperation between agents. In social dilemmas where conditional cooperation is an important factor, as is the case in tax compliance, the model provides a rationale for the low observed sanctions in the real world and the mixed evidence on the effectiveness of deterrence.

Applied Mathematics and Computation, 2022

Punishing those who refuse to participate in common efforts is a known and intensively studied way to maintain cooperation among self-interested agents. But this act is costly, hence punishers who are generally also engaged in the original joint venture, become vulnerable, which jeopardizes the effectiveness of this incentive. As an alternative, we may hire special players, whose only duty is to watch the population and punish defectors. Such a policelike or mercenary punishment can be maintained by a tax-based fund. If this tax is negligible, a cyclic dominance may emerge among different strategies. When this tax is relevant then this solution disappears. In the latter case, the fine level becomes a significant factor that determines whether punisher players coexist with cooperators or alternatively with defectors. The maximal average outcome can be reached at an intermediate cost value of punishment. Our observations highlight that we should take special care when such kind of punishment and accompanying tax are introduced to reach a collective goal.

2007

Mounting evidence suggests that the outcomes of laboratory public goods games, and collective action in firms, communities, and polities, reflect the presence in most groups of individuals having differing preferences and beliefs. We designed a public goods experiment with targeted punishment opportunities to (a) confirm subject heterogeneity, (b) test the stability of subjects ’ types and (c) test the proposition that differences in group outcomes can be predicted with knowledge of the types of individuals who compose those groups. We demonstrate that differences in the inclination to cooperate have considerable persistence, that differences in levels of cooperation after many periods of repeated interaction can be significantly predicted by differences in inclination to cooperate which are manifested in the initial periods, and that significantly greater social efficiency can be achieved by grouping less cooperative subjects with those inclined to punish free riding while excludin...

Biology Letters, 2014

Punishment of defectors and cooperators is prevalent when their behaviour deviates from the social norm. Why atypical behaviour is more likely to be punished than typical behaviour remains unclear. One possible proximate explanation is that individuals simply dislike norm violators. However, an alternative possibility exists: individuals may be more likely to punish atypi- cal behaviour, because the cost of punishment generally increases with the number of individuals that are punished.We used a public goods game with third-party punishment to test whether punishment of defectors was reduced when defecting was typical, as predicted if punishment is responsive to norm violation. The cost of punishment was fixed, regardless of the number of players punished, meaning that it was not more costly to punish typical, rela- tive to atypical, behaviour. Under these conditions, atypical behaviourwas not punished more often than typical behaviour. In fact, most punishment was targeted at defectors, irrespective ofwhether defectingwas typical or atypical. We suggest that the reduced punishment of defectors when they are common might often be explained in terms of the costs to the punisher, rather than responses to norm violators.

Hardin’s (1968) “mutual coercion, mutually agreed upon” institutional solution is examined in view of Dawes’s (1980) reservation regarding the cost-effectiveness of a coercive mechanism. Specifically, we investigated the minimal effective coercion necessary to elicit cooperation and the degree of support for exploitative institutions, which use excessive coercion. Subjects played four versions of a modified repeated NPD game with “Central Authority”. In all modified games, players could choose between cooperation, defection, or joining the Authority. The Authority cooperates when cooperation is profitable; otherwise, it defects. Cost was deducted as a “fee” which the Authority charged its participants, and coercion was imposed as a progressive “fine” deducted from the payoffs of non-participants. With the progressive fine, the more support the Authority gets, the more coercive power it imposes on non-participants. Different levels of fees and fines result in different equilibria. The experimental conditions included several types of Authority games, in which the levels of fees and fines and the degree of inter-trial feedback were varied. Results showed that: (1) the overall cooperation under a Non-Coercive Authority condition (with fees=fines=0) did not exceed the cooperation level obtained in a standard PD game. We hypothesized that the efficient equilibrium for this condition, prescribing equal distribution of players between Authority support and defection, could not be maintained due to social comparisons, which enhanced the temptation to increase one’s payoff by switching from support of Authority to defection. This explanation was validated by results showing that support of Authority increased significantly when subjects were not given feedback about others’ payoffs. (2) For an Authority with intermediate levels of fees and fines, half of the groups converged to the cooperative equilibrium, at which all group members support the Authority. (3) In an “Exploitative Authority” condition (with high fees), only some groups reached the Pareto-deficient equilibrium, at which all group members support Authority, and most groups exited the equilibrium after short sequences of trials. (4) In a “Dominating Authority” condition, in which both fees and fines were progressive as a function of Authority support, all groups converged to the dominated equilibrium, in which all players supported the Authority. Taken together, the results suggest that coercion is necessary to enforce cooperation, but it may also serve institutions to exploit participants. Notwithstanding, non-coercive voluntary institutions could achieve intermediate levels of profitable cooperation, if “excessive” profits of defectors are not palpable.

Hardin’s (1968) “mutual coercion, mutually agreed upon” institutional solution is examined in view of Dawes’s (1980) reservation regarding the cost-effectiveness of a coercive mechanism. Specifically, we investigated the minimal effective coercion necessary to elicit cooperation and the degree of support for exploitative institutions, which use excessive coercion. Subjects played four versions of a modified repeated NPD game with “Central Authority”. In all modified games, players could choose between cooperation, defection, or joining the Authority. The Authority cooperates when cooperation is profitable; otherwise, it defects. Cost was deducted as a “fee” which the Authority charged its participants, and coercion was imposed as a progressive “fine” deducted from the payoffs of non-participants. With the progressive fine, the more support the Authority gets, the more coercive power it imposes on non-participants. Different levels of fees and fines result in different equilibria. The experimental conditions included several types of Authority games, in which the levels of fees and fines and the degree of inter-trial feedback were varied. Results showed that: (1) the overall cooperation under a Non-Coercive Authority condition (with fees=fines=0) did not exceed the cooperation level obtained in a standard PD game. We hypothesized that the efficient equilibrium for this condition, prescribing equal distribution of players between Authority support and defection, could not be maintained due to social comparisons, which enhanced the temptation to increase one’s payoff by switching from support of Authority to defection. This explanation was validated by results showing that support of Authority increased significantly when subjects were not given feedback about others’ payoffs. (2) For an Authority with intermediate levels of fees and fines, half of the groups converged to the cooperative equilibrium, at which all group members support the Authority. (3) In an “Exploitative Authority” condition (with high fees), only some groups reached the Pareto-deficient equilibrium, at which all group members support Authority, and most groups exited the equilibrium after short sequences of trials. (4) In a “Dominating Authority” condition, in which both fees and fines were progressive as a function of Authority support, all groups converged to the dominated equilibrium, in which all players supported the Authority. Taken together, the results suggest that coercion is necessary to enforce cooperation, but it may also serve institutions to exploit participants. Notwithstanding, non-coercive voluntary institutions could achieve intermediate levels of profitable cooperation, if “excessive” profits of defectors are not palpable.

Proceedings of The National Academy of Sciences, 2006

The evolution of cooperation within sizable groups of nonrelated humans offers many challenges for our understanding. Current research has highlighted two factors boosting cooperation in public goods interactions, namely, costly punishment of defectors and the option to abstain from the joint enterprise. A recent modeling approach has suggested that the autarkic option acts as a catalyzer for the ultimate fixation of altruistic punishment. We present an alternative, more microeconomically based model that yields a bistable outcome instead. Evolutionary dynamics can lead either to a Nash equilibrium of punishing and nonpunishing cooperators or to an oscillating state without punishers.

SSRN Electronic Journal, 2018

One of the important topics in public choice is how people's free-riding behavior could differ by group size in collective action dilemmas. This paper experimentally studies how the strength of third party punishment in a prisoner's dilemma could differ by the number of third parties in a group. Our data indicate that as the number of third party punishers increases in a group, the average punishment intensity per third party punisher decreases. However, the decrease rate is very mild and therefore the size of total punishment in a group substantially increases with an increase in group size. As a result, third party punishment becomes a sufficient deterrent against a player selecting defection in the prisoner's dilemma when the number of third party punishers is sufficiently large. Nevertheless, when there are too many third party punishers in a group, a defector's expected payoff is far lower than that of a cooperator due to strong aggregate punishment, while some cooperators are even hurt through punishment. Therefore, the group incurs a huge efficiency loss. Such over-punishment results from third party punishers' conditional punishment behaviors: their punishment intensity is positively correlated with their beliefs on the peers' punitive actions. Some possible ways to coordinate punishment among peers even when group size is very large, thus enabling the efficiency loss to be mitigated, are also discussed in the paper.