priate the common-pool resource and thereby give rise to the “tragedy” predicted by Hardin (1968). Giving subjects the possibility to sanction each other, however, strongly improves the prospects for cooperative behavior. The reason is that many people sanction defectors. This is surprising because sanctioning is costly and therefore not consistent with the assumptions that provide the basis for Hardin’s pessimistic view, that is, that subjects are selfish and rational. A similar observation holds for the communication environment. Allowing for communication also increases cooperative behavior. The resulting efficiency improvement is again inconsistent with the behavioral assumptions underlying Hardin’s analysis because communication does not alter the material incentives.
Taken together, therefore, we have the following puzzle: In a sparse institutional environment, people tend to overharvest common-pool resources. In this sense the pessimistic predictions by Gordon (1954) and Hardin (1968), which are based on the assumptions of selfish preferences, are supported. At the same time, however, we find the efficiency-enhancing effect of informal sanctions and communication. This is in clear contradiction to the standard rational choice view, because why should a rational and selfish individual sacrifice money in order to sanction the behavior of another subject? And why should a money-maximizing subject reduce his or her appropriation level following some cheap talk? The question is more general: Why is the rational choice conception correct in one setting and wrong in another?
In this paper we suggest an integrated theoretical framework that is capable of explaining this puzzle. We argue that the reported regularities are compatible with a model of human behavior that extends the standard rational choice approach and incorporates preferences for reciprocity and equity. The basic behavioral principle that is formalized in our model is that a substantial fraction of the subjects act conditionally on what other subjects do. If others are nice or cooperative, they act cooperatively as well, but if others are hostile, they retaliate.2 Our model also accounts for the fact that there are selfish subjects who behave in the way predicted by standard rational choice theory. We formally show that the interaction of these two diverse motivations (reciprocity and selfishness) and the institutional setup is responsible for the observed experimental outcomes. In the absence of an institution that externally enforces efficient appropriation levels, the selfish players are pivotal for the aggregate outcome. However, if there is an institutional setup that enables people to impose informal sanctions or allows for communication, the reciprocal subjects discipline selfish players and thus shape the aggregate outcome. Moreover, our model shows that when the members of a group have a preference for reciprocity or equity, the common-pool resource problem is transformed into a coordination game with efficient and inefficient equilibria. If subjects are given the opportunity to communicate, they can, therefore, ensure that the equilibrium with the efficient appropriation level is reached.
In the presence of a preference for reciprocity and equity, communication is a coordination device that helps subjects to coordinate their behavior on the low—