Fat tails, exponents, extreme uncertainty: Simulating catastrophe in DICE

The problem of low-probability, catastrophic risk is increasingly central to discussion of climate science and policy. But the integrated assessment models (IAMs) of climate economics rarely incorporate this possibility. What modifications are needed to analyze catastrophic economic risks in an IAM? We explore this question using DICE, a well-known IAM. We examine the implications of a fat-tailed probability distribution for the climate sensitivity parameter, a focus of recent work by Martin Weitzman, and the shape of the damage function, one of the issues raised by the Stern Review. Forecasts of disastrous economic outcomes in DICE can result from the interaction of these two innovations, but not from either one alone.