Nicholas Stern argues that today's integrated assessment models for quantifying the economic and societal impacts of climate change are inadequate (Nature 530, 407–409; 2016). We disagree with his view on the superiority of more complex models such as DSGE (dynamic stochastic computable general equilibrium) models, which purport to account for a larger class of uncertain future events.

In our view, DSGE models have proved to be ineffective for policymaking, even in simple, short-term settings of pure economics, by failing to anticipate the onset of the recent recession (see P. Mirowski Never Let a Serious Crisis Go to Waste 275–286; Verso, 2013). Three decades of social-sciences research on science and politics make it clear that cost–benefit models cannot tame policy-relevant uncertainties or promote political agreement (see, for example, D. Collingridge and C. Reeve Science Speaks to Power 3–4, 59–60; Pinter, 1986).

Models that predict higher costs of climate change might make political intervention more palatable. But prescribing models that generate more precisely quantified estimates of a desired output is a political programme, not a scientific one. Responsible research requires responsible quantification and responsible acknowledgement of uncertainty.