طرد شدگی و ارائه یک کالای عمومی: شواهد تجربی

We analyze the effects of ostracism on cooperation in a linear public good experiment with fixed partner design. Our results show that introducing ostracism increases contribution levels significantly except in first and last periods. Despite reductions in group size due to ostracism, the net effect on earnings is positive. This effect is in contrast to most alternative mechanisms aimed at increasing cooperation rates studied in the literature on public good experiments.

The practice of excluding members of communities, groups and teams to enforce norm conformity and cooperation has been evidenced in almost all civilizations and known cultures (Gruter and Masters, 1986 and Williams, 2001). Ostracism,1 the act of excluding, has been widely documented and analyzed in many disciplines. Indeed, regulating and sanctioning behavior by ostracizing non-contributing members can be found in animals (Goodall, 1986 and Lancaster, 1986), and humans (Wiessner, 2005, Kurzban and Leary, 2001, Williams et al., 2000, Boehm, 1999 and Mahdi, 1986) alike. Ostracism has, however, not only been observed in animal groups, primitive cultures and ancient civilizations, but has also been observed in modern societies – for instance, with respect to strike-breakers or whistle-blowers, as well as to low performers in team production settings (see Durkheim, 1933, Gordon, 1975, Francis, 1985 and Williams, 2001). There exist many economic contexts, such as team production or other social dilemma problems, where it is feasible to ostracize individuals for the lifetime of the project when monitoring and punishment are decentralized. Moreover, Gaspart and Seki (2003) provide an empirical example of a local fishery demonstrating elements of ostracism. In general, ostracism is the consequence of breaking an established rule or a social norm. In the former case, ostracism is normally enforced exogenously, while in the latter case ostracism is imposed endogenously according to self-established criteria. The objective of this paper is to analyze the effects of ostracism using a controlled laboratory experiment. Our experimental design models situations, such as team work involving peers, when the latter have some influence upon team membership, a constellation that arises in particular in project teams within public administrative bodies, firms or in small-scale entrepreneurial activities, but also in local fisheries. More specifically, we investigate the effects of ostracism on contribution levels to a public good as well as on the overall welfare by applying a multi-period public good game with a partner matching design. Our experimental design provides the members of six-person groups with information on the contribution levels of the other five and with the possibility of casting a vote in order to exclude one member. A member is ostracized from the group if she obtains at least 50% of the votes. Excluded members no longer share any of the benefits of the public good and are also excluded from voting. In other words, ostracized members are fully excluded from all future group activities.2 Although even a zero-contributing member does not directly reduce the other group members’ payoffs, and in that respect is harmless to the group, we find that participants frequently ostracize the lowest contributor, thereby reducing the potential group productivity. A large fraction of voting activity and actual ostracism takes place during the first periods of the experiment. Remarkably, most of the remaining members usually converge towards full or almost full contribution, a situation that is sustained in most periods, although cooperation breaks down towards the end of the experiment. Many different forms of punishment exist, and previous experiments have demonstrated how sanctions among peers work quite effectively in fostering cooperation in groups. Exclusion is a powerful instrument when other, direct forms of punishment (from pay cut to imprisonment), usually enforced by formally established institutions, are not available. Especially in contexts where group interaction acquires a social meaning, exclusion or the threat thereof can constitute a powerful deterrent against behavior that does not correspond to the social norm. The possibility of ostracism can take different forms: people can be excluded from ongoing or future projects, from the team or from other relationships that form a natural complement to work (such as lunch or social events). An example, often cited in the economics literature on punishment, is Francis (1985), where ostracism for “scabs” breaking strikes is discussed.3 Again, the existence of social relationships outside work reinforce the motivational effectiveness of exclusion mechanisms, and this is possibly one of the reasons for fostering social activities outside the workplace such as excursions, parties or sports tournaments. In the following section we discuss the related literature. In Section 3 we present the experimental design and procedures and in Section 4 we derive predictions, first based on standard game-theoretic assumptions and second considering “social preferences”. In Section 5 the results and in Section 6 the concluding remarks are presented.

In standard public good experiments, the average contribution level typically declines over time and converges to around 10–15% of the endowment in the last period. The experimental literature shows various instances in which sanctions positively affect contributions. Cooperation increases especially when sanctions are endogenous, i.e. their existence and/or implementation depends on decisions by group members. In particular, this is also the case when peer sanctioning consists of the possibility of exclusion. Our paper contributes to the literature of endogenous sanctions by allowing group members to exclude others based on strict majority voting. The effect of ostracism as a sanctioning device is that the group size is reduced, undermining the potential efficiency of the group in all future periods. This is in contrast to various other forms of punishments tested in the literature which entail a direct monetary effect in that period, without affecting potential maximum overall welfare from the public good in future periods. We find that group members do tend to vote to ostracize other members and, unsurprisingly, these votes go mainly to low contributors. In fact, the ostracized members were always the lowest contributors. Despite the potential negative impact on maximum overall welfare, such behavior is effective in fostering cooperation among the remaining group members. As a result, the overall contribution level is substantially and significantly higher in the public good game with ostracism among those not ostracized as opposed to the standard public good game. Unsurprisingly this cooperation-enhancing effect vanishes as the game approaches the end, when the adverse consequences of being ostracized become less significant. This suggests that mechanisms based on ostracism are effective in the presence of sufficiently large – albeit not “infinite” – time horizons. Our experimental results corroborate the findings from experiments on sanctioning and exclusion as well as those from field studies on the effectiveness of ostracism as a regulatory instrument to enforce norm conformity and cooperation in groups. There is scope for extending this research further in the direction of exploring how ostracism can foster cooperation, both with reference to the length of time a participant is ostracized and with respect to the information feed-back of previous contribution levels. Information on contributions and votes cast could be given throughout the whole session. This would allow reputation to play a more prominent role and would possibly also allow more gradual “warnings” instead of direct exclusion.