Note that Figure 5-2 shows the symmetric case, where the other players’ appropriation decisions are equal.


Because there is a whole range of equilibria, the question of equilibrium selection arises. This issue is discussed in the later section on coordination.


We restrict attention to the cases wherein equilibrium appropriation is less than in SNE.


We concentrate on the behavior of subjects in the final periods to exclude the possible confound of repeated games effects and to make behvior comparable to our one-shot predictions.


A sanctioning institution was first studied by Yamagishi (1986).


Moir (1999) studies the impact of monitoring additional to sanctioning. He points out that pure monitoring does not help to overcome excess appropriation. Institutions with a high level of monitoring but a low level of sanctioning may even lead to more apprpriation than institutions without any monitoring. A different design is suggsted by Casari and Plott (1999) in which monitoring and punishment are compacted in a single decision. In their treatment, efficiency is also higher compared to a baseline treatment without monitoring/punishment.


Notice that according to Proposition 4, the reason for c to increase in k is not because the costs of punishment can be shared between more punishers.


Other models predict a very different pattern of punishment. In Bolton and Ockenfels (2000), for example, punishment is not addressed individually, but directed toward a group average. This could imply, for example, that those who deviate are not punished whereas those who cooperate are punished. This is at odds with the experimental findings. Moreover, it does not take account of the ptential of reciprocal or equity preferences to establish and enforce social norms. For a detailed discussion of this point, see Falk et al. (2000b).


See also Ostrom and Walker (1991).


On communication, also see the paper by Kopelman et al. (this volume:Chapter 4).


See, however, the literature on equilbrium selection in, for example, Harsanyi and Selten (1988).


On coordination game experiments, see Ochs (1995).


For an overview on public goods experiments, see Ledyard (1995).


The results of the meta-study refer to public goods games where the group size is smaller than 10. There is also an experiment where the group size is substantially higher (40 and 100). In this experiment, contributions in the final period(s) are higher compared to small groups (Isaac et al., 1994).


See also Gintis (2000), Sethi and Somananthan (2000), and Huck and Oechssler (1999). See also de Waal (1996), who shows that conditional behavior is observed among chimpanzees. Their food-sharing behavior exhibits some reciprocal pattern: A chimpanzee is ceteris paribus more willing to share food with another champanzee if the latter has shared with the former in the past.


Notice that the latter model is built on the assumption that there exist selfish and reciprocal (tit-for-tat) types.


On the complementary relationship between reputation and reciprocity, also see the paper by Ostrom (1998).


Adams, J.S. 1963 Toward an understanding of inequity. Journal of Abnormal and Social Psychology 62:422-436.

Agell, J., and P. Lundborg 1995 Theories of pay and unemployment: Survey evidence from Swedish manufacturing firms. Scandinavian Journal Of Economics 97:295-307.

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement