Minimax-regret climate policy with deep uncertainty in climate modeling and intergenerational discounting

Integrated assessment models have become the primary tools for comparing climate policies aimed at reducing greenhouse gas emissions. Such policies have often been identified by considering a planner who seeks to make optimal trade-offs between the costs of carbon abatement and the economic damages from climate change. The planning problem has been formalized as one of optimal control, the objective being to minimize the total costs of abatement and damages over a time horizon. Studying climate policy as a control problem presumes that a planner knows enough to make optimization feasible, but in practice, physical and economic uncertainties abound. Manski, Sanstad, and DeCanio (2021) proposed and studied use of the minimax-regret (MMR) decision criterion to account for deep uncertainty in climate modeling. Here we study choice of climate policy that minimizes maximum regret with deep uncertainty regarding both the correct climate model and the appropriate time discount rate to use in intergenerational assessment of policy consequences. The analysis considers a range of discount rates to express both empirical and normative uncertainty about the appropriate magnitude of this parameter. The findings are novel and informative. The MMR analysis points to use of a relatively low discount rate of 0.02 for climate policy. The MMR decision rule keeps the maximum future temperature increase below 2 °C above the 1900–09 level for most of the parameter values used to weight costs and damages.