The pivotal mechanism revisited: some evidence on group manipulation
通过实验室实验研究关键机制是否容易被群体操纵，发现仅当允许组内沟通时才会出现合谋，而默契合谋几乎没有发生。
Abstract This paper studies the vulnerability of the pivotal mechanism with respect to manipulation by groups. In a lab experiment, groups decide on the implementation of various alternatives, some of which imply opposite interests for the two subgroups. We investigate the occurrence of tacit and explicit collusion by allowing for communication within subgroups in one treatment and prohibiting it in another. Even though all agents’ preferences are common knowledge and there exists a simple symmetric collusive strategy for one subgroup, we find little evidence for tacit collusion. Only when explicit communication is allowed, collusion is established. A behavioral model using quantal response equilibrium in which subjects have beliefs over the correlation of errors of same-type subjects helps explain the main features of our data.