Results are reported from an experiment examining how opportunities to anonymously reward and sanction other group members affect resource allocations in a voluntary contributions mechanism (VCM). Sessions begin with 10 rounds of a VCM game, where subjects allocate an endowment of tokens between a private account and a group account. Each of the following 10 rounds is characterized by a two-stage game. In the first stage of each round, group members participate in a VCM game. In the second stage of the round, group members are informed of the profile of group allocations of others and receive an additional endowment of tokens. In the REWARD (SANCTION) treatment subjects can allocate tokens to their private account or use them to increase (decrease) other group members earnings. In the COMBINED treatment they can be allocated to the private account or used to either increase or decrease other group members' earnings. In the BASELINE treatment second-stage tokens are automatically placed in the private account. In the COMBINED treatment, allocations to the group account are sustained well above those observed in the BASELINE. In the SANCTION treatment allocations to the group account are sustained above BASELINE, but below those observed in the COMBINED treatment. Group allocations in the REWARD treatment initially follow a pattern similar to those observed in the SANCTION treatment, but fall off abruptly in the last few rounds. Interestingly, in the COMBINED treatment, the frequency of rewards begins at a much higher level than the frequency of sanctions, but falls dramatically over the course of the session. Rewards represent a zero sum transfer from subjects giving rewards and those receiving awards. Sanctions represent a net loss, a cost to both the subject giving the sanction and the subject receiving the sanction. Initially, earnings in the COMBINED and REWARD treatment are above those in the BASELINE, while earnings in the SANCTION treatment are below BASELINE. Over the course of multiple decision rounds, however, earnings in the four treatments tend to converge. By the last round, the earnings difference between the four treatments is minimal.

Co-author Robert Shupp and James Walker