DICESC: Optimal Policy in a Stochastic Control Framework

The effects of anthropogenic greenhouse gas emission to climate change are uncertain; how does this uncertainty affect the optimal carbon abatement policy? Tackling this classic problem is challenging; not only does it require integrated assessment analyses that consolidate an adequate representation of both climate and economy, but also a sensible understanding of hedging incentives and decision making under uncertainty. In the Integrated Assessment Model (IAM) community, this representation is severely restricted by computational burdens and model intractability (Webster, Santen, and Parpas, 2012). I provide a minimal framework that relies on the transparency of mathematical programming, to assess precautionary mitigation as a means to hedge against the risk and uncertainty of climate catastrophes. The scope of the paper encompasses the following: assessment of optimal abatement under parameter uncertainty and endogenous learning; decomposition of optimal abatement with respect to mitigatory incentives; assessment of policy sensitivity to assumed distributions of uncertain temperature thresholds. I introduce DICESC, a multistage stochastic control version of Nordhaus’ DICE model to demonstrate that policy optimization in which the hazard rates of climate catastrophes can be controlled significantly increases incentives for precautionary abatement. Despite the drastic level increase in precautionary abatement the proposed formulation achieves, I stress the importance of understanding the distributive properties of uncertainty for the robust assessment of optimal policy