Abstract
Estimating the cost to society from a ton of carbon dioxide (CO₂) released into the atmosphere requires connecting a model of the climate system with a representation of the economic and social effects of changes in climate, as well as the valuation and aggregation of diverse, uncertain impacts across both time and space. The literature on this cost, termed the social cost of carbon (SCC), is large and growing. Prior work has largely focused on better constraining the values of parameters such as climate sensitivity, the discount rate, and the damage function. A growing literature has also examined the effect of varying more fundamental structural elements of the models supporting SCC calculations. These structural model choices—including the introduction of climate or economic tipping points, changing the structure of economic preferences, and accounting for the persistence of climate damages—have been analyzed in piecemeal, uncoordinated fashion, leaving their relative importance unclear. Here we perform a comprehensive synthesis of the evidence on the SCC, combining 1823 estimates of the SCC from 147 studies published between 2000 and 2020 with a survey of the authors of these studies. The distribution of 2020 SCC values reported in the literature, with a median of $39 per ton CO₂, a mean of $132 (after truncating the upper and lower 0.1%), and a 95th percentile of $519, is substantially right-skewed and shows evidence of heavy tails. Analysis of variance reveals important roles for structural elements in driving SCC estimates, particularly the inclusion of persistent damages via effects on economic growth, representation of the Earth system, and distributional weighting. However, our survey reveals that experts believe the literature is biased downwards due to an under-sampling of structural model variations, as well as biases in damage-function and discount-rate parameters. To address this imbalance, we train a random forest model on variation in the literature and use it to generate a synthetic SCC distribution that more closely matches expert assessments of appropriate model structure and discounting. This synthetic distribution has a median and mean of $185 and $284 per ton CO₂, respectively, for a 2020 pulse year (5%–95% range: $32–$874), higher than all official government estimates, including the most recent update from the U.S. Environmental Protection Agency.