Lars Peter Hansen, University of Chicago
Climate science documents uncertainty induced by different emission scenarios, alternative models, and ambiguous physical interactions. Moreover, for some purposes, it constructs tractable approximations to initially complex models. To engage in credible policy analysis requires that we acknowledge and confront the limits to our understanding of dynamic mechanisms by which human inputs impact the climate. Our research stresses limits to our understanding, with particular focus on the channel from emissions to atmospheric concentration and the channel from concentration to temperature. Even in the most recently developed Atmospheric Ocean General Circulation Models, there is wide variation across models in key policy parameters produced by such models for the same amount of anthropogenic forcing. This variation persists when considering impacts for a wide range of time scales, including shorter time scales that are more typical in economic investigations. We find it productive to pose hypothetical social planning problems and embrace recent advances in decision theory designed to confront uncertainty. The decision problems allow us to assess more formally the consequences of uncertainty. Especially for problems of this nature, we find it important to think of uncertainty as broadly conceived to include risk within a given model, ambiguity across models and potential model misspecification. We draw on asset pricing insights to understand better the consequences of climate impacts that depend on the horizon to the adverse outcomes might be realized. We show the potential importance of both state dependence and horizon dependence in marginal valuations of uncertainty. Because our analysis acknowledges the limits to our knowledge, it implies specific forms of caution but not inaction.