Optimal management of an ecosystem with an unknown threshold

We consider an ecosystem with two distinct equations of motion that are separated by a threshold value of the state variable. We find that increasing uncertainty (both uncertainty embedded in the natural system and uncertainty of the decisionmaker about the location of the threshold) can lead to nonmonotonic changes in precaution: a reduction in uncertainty can first increase and then decrease optimal precautionary activity. This nonmonotonicity can help to explain why regulators often give conflicting arguments about optimal abatement policies in the face of uncertainty. For example, some regulators argue for an immediate reduction in pollutant loading until uncertainty about the underlying process is reduced while others call for no costly reductions in pollutant loading until the same uncertainty is reduced. These statements can be consistent even if both sides agree on both economic objectives and the system dynamics, but have different priors on the uncertainty involved.Ecosystem management Threshold Stochastic dynamic programming