Hotelling's Rule in the limit: an agent-based exploration of the model space

Hotelling's Rule is the observation that the exploitation of a nonrenewable resource can only be economically e cient if the resource owner's marginal pro t increases at the prevailing discount rate. This has been a perennial topic in the literature of resource economics since the 1970s, with some authors extending the theory and others analyzing empirical data. This paper reports on the results from using agent-based modeling to assess the consequences of relaxing the optimality constraint to explore the ways in which the outcome space converges on Hotelling's Rule in the limit. The agent-based model (ABM) in this paper has one choice variable: increase, decrease, or maintain the current production level based on one rule: choose the change in production level that maximizes estimated discounted pro t. The results, based on a costless technology and a stylized demand function from Hotelling, indicate that total discounted pro t has low sensitivity to deviations from the optimum. In extending the basic Hotelling model to stylized production technologies with cost, the simple ABM falls short of the optimum by as much as ten percent, depending on the magnitude and whether the cost is xed, marginal, or based on the resource stock level. The optimization errors of the ABM are similar to the errors of a human production planner with incomplete information. The ABM also exhibits emergent collusion-like and Cournot-like behaviors when extended to a small oligopoly market.(JEL Q32)