Reader Comments

Post a new comment on this article

The role of uncertainty

Posted by rrosen on 20 Feb 2017 at 21:35 GMT

Even though the authors acknowledge that the results for IAMs that they rely on for this analysis are highly uncertain, they do not really take their own warnings to this effect seriously when providing very detailed output to the reader. My guess is that if they had performed an uncertainty analysis for the scenarios that they report, the level of uncertainly would swamp the subtle signal they are trying to detect. Furthermore, they do not provide a convincing analysis of the IAMs relied on, especially of the FUND model, to show that the way in which the model responds to temperature change to calculate "impact" is credible.

No competing interests declared.

RE: The role of uncertainty

RichardSJTol replied to rrosen on 22 Feb 2017 at 08:19 GMT

Dr Rosen is right to point out the large uncertainties about our estimates of the impacts of climate change and variability. We think that sensitivity analysis is more insightful than uncertainty analysis and therefore omitted the latter from a paper that is already long and complicated. Besides, for most of the impact models used, the second-order stochastic dominance shown in the paper implies first-order stochastic dominance. FUND is the exception. An analysis of its uncertainties can be found in Anthoff and Tol (2013, Climatic Change).

As we argue in the paper, impact models cannot readily be validated, if at all, as impacts have not been measured in the past and will not in the future. As to the respective credibility of IAMs, FUND includes more of the established empirical regularities on impacts and vulnerability than do the other IAMs.

No competing interests declared.