This paper addresses the collective action problem of nonpoint-source pollution control in a small agricultural watershed. At issue is the stability of cooperative behavior among a group of farmers, who have voluntarily agreed to discontinue their use of the herbicide atrazine due to high concentrations of the herbicide in a local water supply. Continued cooperation among the group is threatened by the unexpected cancellation of cyanazine, an inexpensive and widely used alternative to atrazine. With cyanazine no longer available, the farmers will face a significant increase in weed control costs if they continue to use products that do not contain atrazine. Is cooperation among the farmers still possible despite the increased cost of cooperating? This research explores the economic and behavioral factors that influence the collective outcome of this social dilemma. The collective action is modeled as a recurrent coordination problem. The producers (farmers) are engaged in a repeated assurance game with imperfect public information, where producers' choices are driven by the desire to coordinate their actions with the others in the group. A producer's decision to cooperate or defect is based on a threshold approach; if the number of others believed to be cooperating exceeds the level of cooperation required to make cooperation beneficial, then the producer will choose to cooperate. Otherwise, the producer will defect. Since producers are unable to directly observe the choices of the others in the group, each producer must rely on a subjective assessment of the group's behavior based on the realization of the public outcome, the concentration of atrazine in the lake. Producers use a naive Bayesian learning process to update their beliefs about the joint actions of the group. The formal learning process is modeled using a sequential quasi-Bayesian procedure that is consistent with the fictitious play model of learning. The interaction between the producers and the impact of their collective behavior on the levels of atrazine in the lake is formulated as a computational multi-agent system (MAS). The MAS is an artificial representation of the collective action problem that integrates the economic, behavioral and environmental factors that influence the decision-making process of producers. The MAS is used to simulate the evolution of collective behavior among the group and to evaluate the effectiveness of selected incentive mechanisms in preventing the collapse of joint cooperation. The results suggest that without additional incentives, farmers are likely to abandon their voluntary agreement and resume their use of atrazine within the watershed. It is then demonstrated how a combination of policy instruments can be used to alter the underlying game configuration of the collective action problem, resulting in cooperative outcomes. An ambient-based penalty, when used in conjunction with a subsidy payment, is shown to produce divergent incentive structures that shift the classification of the collective action away from a coordination problem with two equilibria to a mixed configuration containing several different game structures and many possible equilibria. This result has important consequences in terms of the evolution of producer behavior and the set of possible collective outcomes. The analysis concludes with an example, which demonstrates that when a mixture of game structures characterizes the collective action, joint cooperation is not a prerequisite to the realization of socially desirable outcomes. By carefully selecting the combination of subsidy payment and ambient penalty, a policy maker can manipulate the underlying structure of the collective action, whereby producers with the smallest impact on water quality choose to defect while all others cooperate.

Issue Date:
Publication Type:
Conference Paper/ Presentation
Record Identifier:
PURL Identifier:
Total Pages:
Series Statement:
Selected Paper

 Record created 2017-04-01, last modified 2018-01-22

Download fulltext

Rate this document:

Rate this document:
(Not yet reviewed)