Cooperation & Punishment

“Does the presence of punishment opportunities lead to an increase in cooperation level? – Under which circumstances does punishment help to establish cooperation?”

The basis of many economic theories is the concept of the homo oeconomicus, the rational thinking human who wants to maximize profits rather than cooperate for the sake of society. However, in reality individuals show their willingness to cooperate. The idea of society includes the necessity of cooperation. Therefore, getting people to contribute to the progress of society as a whole should be seen as a major goal.

In the past, cooperation was investigated thoroughly by many economists. Martin Nowak of Harvard University even considers it as one of the main forces behind evolution along with natural selection and mutation. The question of how cooperation can sustain against the self-interest that characterizes every life form on earth is still debatable. Darwin came up with the idea that kinship may help to explain cooperative behavior. Moreover, reciprocal altruism and strategy might be a possible motivation behind it. However, lab experiments have shown that individuals even cooperate with strangers even when there is little possibility of encountering the same cooperation partner again.

When it comes to cooperation, one favorable method is to make use of the standard public good game. It has been implemented in many modifications and applications. Basically, subjects have to decide how much to contribute from their initial endowment into a group account. Afterwards they get their payoff, which consists of the part from their endowment kept, plus the amount contributed by every subject multiplied with a factor. The highest overall payoff is reached when every participant contributes his or her whole fund. One can never know if the others will cooperate or defect. This is why subjects playing the public good game find themselves in a typical prisoner’s dilemma.

Experiments have shown that punishment opportunities seem to play an important role in establishing cooperation. When investigating the topic in greater detail, however, I find that this is not always the case. The question therefore is: Under which circumstances does punishment help to establish cooperation?

Fehr and Gächter tested a version of the public good game, which aimed to investigate the influence of punishment on cooperation. Their experiment consisted of 4 treatments: a stranger- and a partner-treatment with and without punishment opportunities. In the partner-treatment the same 4 subjects stayed in one group for each period, whereas in the stranger’s version new members were assigned randomly to the group in each period. The condition with punishment differed from the one without punishment in that the contribution stage was followed by a punishment stage in which subjects were informed about the other group members’ contributions and had the possibility to punish each other accordingly. For that purpose punishment points were assigned to the victim by the punisher. One of these points reduced the victim’s payoff by 10 % and was also costly for the punisher.

The authors predicted that individual contributions would be zero if subjects were rational and selfish. This even holds for the punishment condition due to the fact that rational individuals will not punish if it is costly for them as well. All contributors know that there is no threat of being punished and therefore no deterrent from free riding. In contrast to these theoretical assumptions the experiment showed that individuals are willing to punish uncooperative behavior. The question of whether cooperation can be achieved and held stable in the punishment treatment will be illuminated by looking at the experimental results in more detail:

  • Within the Stranger-treatment with punishment opportunities the average contribution level is much higher than in the stranger treatment without punishment. The same holds for the partner treatment.
  • Within the stranger treatment without punishment cooperative behavior converges to full free- riding.
  • Within the punishment condition of the Partner-treatment, subjects’ behavior converges towards full cooperation, whereas in the no-punishment condition full free-riding is the preferred action.
  • The greater the contribution amount deviates negatively from the average group contribution, the more severe is the punishment received.

With respect to the main question one can conclude that in the no-punishment condition subjects tend to free ride. However, in the face of punishment opportunities subjects are much more willing to cooperate, because they prefer not to be punished. They know that cooperative individuals won’t approve of non-cooperative behavior. In most of the cases punished subjects increased their contribution in the next period suggesting that “subjects seemed to have had a clear understanding of why they were punished and how they should respond to the punishment”. In this case it is obvious that punishment opportunities are suitable to help achieve and maintain cooperation. In terms of efficiency though, the payoff losses and gains have to be examined. In the beginning periods of the punishment treatment a payoff loss relative to the no-punishment condition can be observed. In the final periods there is a positive payoff gain. In the stranger treatment it is only positive in the last two periods. Therefore, if the experiment had lasted less than 8 periods, a payoff loss would have been observed. Achieving cooperation is a nice goal but is not sufficient when the costs exceed the benefits. Punishment is not only costly in the public good experiment but also in reality. When trying to achieve cooperation through punishment, the costs of doing so have to be taken into account.

With respect to the main question “Does the presence of punishment opportunities lead to an increase in cooperation level?” it can be discussed, that punishment methods can encourage cooperation if they are implemented in an appropriate manner.
However, lab experiments have a limited scope of validity and field experiments have to be taken into account to achieve more precise outcomes.


Fehr, Ernst; Gächter, Simon (2000): “Cooperation and Punishment in Public Goods Experiments”,The American Economic Review, Vol. 90, No. 4 , pp. 980-994.

Pennisi, Elizabeth (2009): “On the Origin of Cooperation”,, published by AAAS, Vol. 325, (2009), pp. 1196-1199.

2 thoughts on “Cooperation & Punishment

  1. First of all I wanted to add that your blog post is easy to read. I had a seminar on the topic before so it was quiet nice for me to follow your arguments. At least there came some points to my mind when reading your blog:
    As your blog deals with cooperation I think it should focus on the question of repeating games vs. one-shot games, because there are at least big differences between these games. However, there is no way to induce cooperation in a one-shot game because all players know that this is the last round and therefore the threatment of punishing others is not credible. Maybe this can be a point if you write your paper on the same topic.
    Furthermore the question came to my mind what you wanted to say by introducing punishment in an appropriate way. For me it is not clear what this means because punishment is costly for a minimum of two player, the punisher and the punished guy, so I did not get your argument right I think.
    Also comming to my mind was the question whether punishment is always the best method to induce cooperation. I think it would be less harmful to introduce cooperation by changing the incentives of the players e.g. with nudging.


    1. Thank you for your comment, Kevin!
      I can clearly see that you have already dealt with this topic in the past because your comment is very good and highlights some weak points/ further applications of this topic. I mainly focused in my essay on dealing with lab and field experiments and therefore the one shot vs repeated games situation also took place, so thank you for that. Even though you may not be able to apply cooperation in a one shot game you can see from studies, that people are willing to cooperate, even in a one shot situation. Therefore I also presented some situations in my paper that will cover this situation, where it might not be rational to cooperate. As we have already learned in our lecture there are different types of people (rational, quasi-rational and enculturated actor) that could explain such situations.
      Nudging is another interesting aspect that came to your mind and I also thought about implementing it to my essay. I did not focused too much on this point (due to lack of space) but in my opinion nudging can help to provide more cooperation solely by the point that people may have an incentive to contribute in several situations. Furthermore nudging has to happen in a rather cost free situation especially when it comes to developing countries in my opinion. People don´t have the mental capacity and the will power to evaluate such special and often unclear situations when for instance a state wants to nudge people to cooperate. There has to be some sort of cost-free incentive or even an intervention to promote cooperation where it can be beneficial for a group of people. If you are interested in reading my essay, just let me know and I will send it to you.
      Again, thank you for your criticism and thoughts about this interesting topic 🙂


Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s