Are estimated costs of stringent mitigation biased?

We take issue with the claim by Tavoni and Tol (Clim Chang 100:769–778, 2010 ) that reviews of the macroeconomic costs of achieving the 2 °C climate target have been affected by selection bias and have underestimated the costs. Although many more cost estimates are available in the literature, they have restricted their survey to the data in the EMF22 study, with a limited set of model solutions for the 2 °C target. They have applied the methodology of observational meta-analysis inappropriately to policy meta-analysis, where the number of results is often very small and the basis for imputing a statistical distribution does not usually exist. They have mixed direct costs with net costs in terms of %GDP. Their method of “correcting” for missing data with (high) costs of stringent mitigation could equally be applied to correcting the data for omission of mitigation options such as biomass energy with carbon capture so reducing the cost estimates. And finally they implicitly assume that the same policy combinations and mitigation options are applied for all climate scenarios, when more stringent scenarios may require more stringent policies and options, such as regulation or BECCS. The conclusion from the literature is more appropriately that the costs are highly uncertain, that they can equally be positive or negative (gains) and that models which fail to solve for stringent mitigations are not fit for purpose. Copyright Springer Science+Business Media Dordrecht 2013