Incomplete information, learning, and natural resource management

The problem of resource extraction developed in Levhari and Mirman (1980) is reconsidered under a situation of incomplete information. Specifically, players do not have information about other players' benefit functions. It is assumed that each player relies on simple, non probabilistic beliefs about the other players' behaviour. Basically, players assume that a variation of their own consumption has a first order linear effect on the consumption of others. We define a simple learning procedure where players' beliefs are updated through observations of resource levels over time. Convergence, viability, and local stability of the procedure are proved. Comparisons are made with the full information benchmark case provided by Levhari and Mirman. For a large set of situations, the steady state of the resource lies between the non-cooperative and cooperative solutions in the benchmark case.