Evaluating the Key Drivers of the US Government's Social Cost of Carbon: A Model Diagnostic and Inter-Comparison Study of Climate Impacts in DICE, FUND, and PAGE

Abstract

The social cost of carbon (SCC) is a monetary estimate of the climate change damages to society from an additional emission of carbon dioxide (CO2). US agencies are now required to apply the SCC to assess the potential benefits of CO2 reductions in federal regulations, including rules and proposals affecting appliances, transportation, industry, and power generation. This paper presents the first in-depth model diagnostic and inter-comparison examination of the three integrated assessment models used to estimate the SCC – DICE, FUND, and PAGE – to reveal how they uniquely determine damages from climate change. Specifically, we reviewed the source code, published documentation, and underlying literature of each of the three models, and then performed controlled experiments to diagnose the contribution of particular sectors, regions, and other model assumptions (parametric and structural) to the resulting cost. We find that DICE and PAGE project substantially higher climate damages, and therefore higher SCC values, than FUND, which includes the potential for net benefits in the near-term. Despite the fact that FUND is highly-disaggregated sectorally and regionally, 95% of its SCC can be explained by a few damage function parameters related to cooling, agriculture, avoided heating, and water resources, particularly those for China. The DICE SCC can only be decomposed into sea level rise damages and an aggregation of all other damages, with the latter category being the dominant SCC driver. The PAGE SCC is mostly driven by non-economic damages, with all costs distributed globally though greatest in the US and least in the former Soviet Union. In all three models, impacts from sea level rise contribute less than one-tenth of the SCC. This study's diagnostic analysis improves public understanding of the SCC, informs future SCC estimation, and sets research priorities for climate impacts modeling.