This paper presents an experimental study of two different pollution compliance games: collective vis-à-vis random fining as a means to regulate non-point pollution. Result suggests the importance of considering subject pool differences in the evaluation of environmental policies by means of experiments, particularly if those policies involve certain forms of management decisions.
In this paper we present an experimental study of both mechanisms proposed by Xepapadeas (1991) for cases where only ambient pollution levels can be observed. The main issue is the analysis of the players’ behavior under collective vis-à-vis random fining, where the expected payoff in both mechanisms is identical. We designed two non-cooperative games that closely reflect the proposed mechanisms. One of the games reflects the non-budget-balancing mechanism, where firms face collective penalties if ambient pollution exceeds the desired target. The second game mimics the mechanism, where one firm is randomly chosen to bear the fine whenever measured ambient pollution is not optimal. Following Kritikos (1993) we did not incorporate budget balancing, i.e., we did not refund the fines to other players, mainly because we are interested in the response of agents to the scheme, and because risk neutrality cannot be excluded for small stakes. Using samples from both Costa Rican coffee mill managers and Costa Rican students, we find that the two games perform equivalently but, although they lead to efficient outcomes through Nash play in the majority of cases, the observed frequency of Nash play is lower than theoretically predicted. Moreover, we reject the hypothesis that managers and students behave equally. Off the equilibrium, managers tend to over-abate, whereas students tend to under-abate. This result suggests the importance of considering subject pool differences in the evaluation of environmental policies by means of experiments, particularly if those policies involve certain forms of management decisions.
Co-authors:
Till Requate, Albert Schram
