Spotlighting the impacts functions in integrated assessments. Research Report Prepared for the Stern Review on the Economics of Climate Change.

This project reviews how the integrated assessment models DICE/RICE, MERGE, PAGE, and FUND simulate climate impacts under different socioeconomic scenarios, and compares this with the results of a parallel study, Understanding the Regional
Impacts of Climate Change. These models are commonly used in studies to determine the social cost of carbon, and in cost-benefit analyses to determine “optimal” climate change policy.

RICE and FUND simulate regionally specific impacts in a number of sectors (either in the aggregate or sector-specific), whilst PAGE, DICE and MERGE simulate aggregate global market and non-market damages, as well as damage due to rapid or
catastrophic climate change. Only FUND shows how damages expressed as %loss of GDP for a given temperature rise vary with socioeconomic scenario1. The versions of DICE/RICE, MERGE and FUND studied for this report are deterministic models. Only the PAGE model is probabilistic in nature2.

Non-market damages are normally estimated through a willingness-to-pay approach. All damages in the DICE/RICE, MERGE, PAGE and FUND models are output from the models in terms of % loss of global or regional GDP or $$. No physical metrics are used. Only FUND includes sector-specific equations based on simple physical processes, but it also does not produce outputs in physical metrics3. Adaptation is usually modelled implicitly through the calibration procedure, such that the models
reflect the assumptions of the underlying literature. Only PAGE can explicitly simulate adaptation.

The study found that all the models are based on literature from 2000 and earlier. Since this time, some predictions of climate impacts have become more pessimistic, for example predictions of the potential for rapid or catastrophic climate change. The implications of these predictions has, however, recently been studied using PAGE4.