Probabilistic stabilization targets

We study stabilization targets: common environmental policy recommendations that specify a maximum probability of an environmental variable exceeding a fixed target (e.g., limit climate change to at most 2°C above pre-industrial). Previous work generally considers stabilization targets under certainty equivalence. Using an integrated assessment model with uncertainty about the sensitivity of the temperature to greenhouse gas (GHG) concentrations (the climate sensitivity), learning, and random weather shocks, we calculate the optimal GHG emissions policy with and without stabilization targets. We characterize the range of feasible targets and show that the climate is difficult to control in the short run, although as learning resolves the planner eventually achieves the target with a sustained reduction in emissions over time. We find that uncertainty exacerbates the welfare cost of stabilization targets. First, the targets are inflexible and do not adjust to new information about the climate system. Second, the target forces the emissions policy to overreact to transient shocks. These effects are present only in a model with uncertainty. Introduction of a stabilization target into the baseline model with uncertainty results in a welfare loss of 4.7%, which is 66% higher than the cost of introducing the target in the certainty version of the model.