Schools of Fishermen: A Theory of Information Sharing in Spatial Search

Fishermen who compete for a resource in an open access setting generally do not share information amongst each other about where stocks are located. On the other hand, in fisheries where there are established property rights or fishing cooperatives, fishermen tend to share information with each other. Recent work in developing spatial models of fishermen behavior has generally ignored the effect that sharing information about where stocks are located can have. We develop a behavioral model of search amongst spatial resource harvesters that allows for varying degrees of information sharing. We demonstrate that informational cascades may lead to extremely inefficient spatial search by fishermen that do not share information and that this inefficiency may be persistent over time. We define a new parameter, information-dependent catchability, which captures the degree to which information sharing improves the efficiency of spatial search. We argue that institutions change the incentives to share information; information-dependent catchability will differ within a fishery depending on the specific management institutions adopted. This leads to theoretical predictions which depart from standard models but account for a wider range of field observations. In particular, we derive the conditions under which closing access to a fishery would have such a drastic impact on the incentives to share information (and hence, the efficiency of search) that total effort in the fishery will be unaffected by restricting access. We derive the equivalent conditions for steady state harvest and stock levels. Furthermore, we make an important distinction between Property Rights Rents and Information Sharing Rents (economic rents that can be attributed solely to changes in information-sharing). The shiroebi shrimp fishery in Shinminato, Japan provides an ideal natural experiment to test the impact of information sharing. Field observations from this fishery strongly accord with the predictions of the model presented in this paper.