Decision making under catastrophic risk and learning: the case of the possible collapse of the West Antarctic Ice Sheet

A collapse of the West-Antarctic Ice Sheet (WAIS) would cause a sea level rise of 5-6 m, perhaps even within 100 years, with catastrophic consequences. The probability of such a collapse is small but increasing with the rise of the atmospheric concentrations of greenhouse gas and the resulting climate change. This paper investigates how the potential collapse of the WAIS affects the optimal rate of greenhouse gas emission control. We design a decision and learning tree in which decision are made about emission reduction at regular intervals: the decision makers (who act as social planners) have to decide whether to implement the environmental or not (keeping then the flexibility to act later). By investing in the environmental policy, they determine optimally the date of the optimal emission reduction. At the same time, they receive new information on the probability of a WAIS collapse and the severity of its impacts. The probability of a WAIS collapse is endogenous and contingent on greenhouse gas concentrations. We solve this optimisation problem by backward induction. We find that a potential WAIS collapse substantially bring the date of the optimal emission reduction forward and increases its amount if the probability is high enough (a probability of 1% per year for the worst case), if the impacts are high enough (a worst case damage of 10% of GDP for a 3 degrees C warming) or if the decision maker is risk averse enough (for example a social damage due to pollution equal to 1% GDP for an atmospheric temperature of 3 degrees C). We also find that, as soon as a WAIS collapse is a foregone fact, emission reduction falls to free up resource to prepare for adapting to the inevitable. By contrast, adaptation (such as building dikes along the coast) postpones policy intervention because that strategy reduces the risk of catastrophic damages.