Social norms in repeated public good games

In this paper we explore the relationship between an individual’s preference for cooperation and the establishment of cooperative norms. Our aim is to provide an experimental test of the hypothesis (see among others, Fehr and Gachter 2002) according to which individuals are prepared to punish defectors in experimental social dilemma games, such as through sanctions against free riding behaviour, because they want to enforce a social norm and that this may lead to increased future levels of cooperation. We study the results of two experiments in which individuals decide both whether to participate in a common project and the institutional rules determining how the profits of the project were to be shared among each of the participants in the group. They could choose between 1) a regime where gains were shared equally, regardless of individuals’ contributions and without sanctions and rewards (System A); 2) a regime where individuals were paid according to their marginal contribution, (System B); or, 3) a regime in which gains were shared equally (as in System A), but individuals were allowed to punish (and/or reward) free riding (cooperative) behaviour as in Sefton, Shupp and Walker (2007). Our main research question is to analyse whether high contributors react to free riding behaviour by adopting punitive sanctions (System C). Our subjects were also required to fill a questionnaire composed of four sections, where their attitude towards cooperation and their opinions on civic values and free riding behaviour were thoroughly explored. We monitored the behaviour in the game and the institutional choices of potential free riders and cooperators – defined on the basis of their questionnaire responses. Our results partly contradict the initial hypothesis in as much as System A and B received the largest shares of votes in almost all rounds, and while System A was chosen by low contributors, high contributors mostly voted B. Thus, most individuals do not like sanctions, and their institutional preferences do not seem to be related to their willingness to cooperate. Inspection of individual data and the study of the questionnaire answers, however, reveal some interesting points. In fact, we can assert that System C was mostly chosen by very cooperative individuals in response to observed free riding behaviour. Furthermore, in all sessions, the share of high contributors selecting C increased towards the end of the game, and average contributions significantly increased when C was expected. Finally, analysing the different sections of questionnaires, the individuals who voted C show the highest propensity to condemn free riding behaviour on a moral basis as well as the highest ethical standards. Our conclusion is that, as far as the institutional choices are concerned, unobserved ethical and cultural individual preferences play an important role. There are individuals who see cooperation as the “right” thing to do, and therefore are prepared to implement institutional rules that may favour this collective outcome. The majority of high contributors in our experiments, however, did not share these same values and responded to free riding by adopting B.