Comprehensive evidence implies a higher social cost of CO₂

The social cost of carbon dioxide (SC-CO₂) quantifies the societal damages from an additional tonne of CO₂ emissions. For over a decade, it's been a crucial metric in benefit-cost analyses for climate policies globally, influencing regulatory impact analyses and informing decisions on various regulations and policies, including energy efficiency standards, vehicle emissions, carbon capture tax credits, and corporate sustainability accounting. However, the 2017 US National Academies of Sciences, Engineering, and Medicine (NASEM) report highlighted that existing SC-CO₂ estimates lagged behind the latest research. The NASEM report recommended improvements to the scientific basis, transparency, and uncertainty characterization of SC-CO₂ estimates. This research directly addresses these recommendations, utilizing a new open-source model to incorporate updated scientific understanding across all components of SC-CO₂ estimation.

The existing literature on SC-CO₂ estimation highlights the use of Integrated Assessment Models (IAMs) that combine climate system and economic models. These models typically project population and GDP to estimate CO₂ emissions, which then drive climate models to project temperature changes and other impacts. These impacts are monetized and discounted to a present value. However, criticisms of previous IAMs include outdated climate models, damage functions not reflecting recent scientific advances, and inadequate characterization of uncertainty. Recent research has addressed some of these shortcomings with improved damage functions, climate models, and socioeconomic projections; however, a fully integrated and probabilistically consistent approach was lacking. This study aims to fill this gap, integrating recent improvements into a cohesive model.

This study uses the Greenhouse Gas Impact Value Estimator (GIVE), a newly developed open-source integrated assessment model built on the Mimi.jl platform. GIVE incorporates several improvements over previous IAMs: 1. **Probabilistic Socioeconomic Projections:** The study utilizes the Resources for the Future Socioeconomic Projections (RFF-SPs), which provide internally consistent probabilistic projections of population, per capita economic growth, and greenhouse gas emissions. These projections consider uncertainty across multiple centuries. 2. **Advanced Climate Modeling:** GIVE employs the FaIR v1.6.2 climate model and the BRICK sea-level model, which are more sophisticated than models used in previous SC-CO₂ estimations and more accurately represent the climate system dynamics. 3. **Updated Damage Functions:** GIVE incorporates updated damage functions from recent peer-reviewed literature reflecting impacts on agriculture, mortality, energy consumption, and sea-level rise. These functions are regionally disaggregated for 184 countries, enabling more accurate damage assessments and allowing for sensitivity analysis using global damage functions from meta-analyses. 4. **Stochastic Discounting:** GIVE employs an empirically calibrated stochastic discounting framework that is consistent with observed interest rates and economic growth, enabling a theoretically consistent valuation of risk associated with marginal CO₂ emissions. The model uses a Monte Carlo approach to sample the interconnected uncertainties in socioeconomic, climate, and damage function components, generating 10,000 SC-CO₂ values to fully represent uncertainty in the estimations. The model is also capable of estimating the social cost of other greenhouse gases.

The study's preferred mean SC-CO₂ estimate, using a near-term risk-free discount rate of 2%, is $185 per tonne of CO₂ ($44–$413: 5%–95% range) in 2020 US dollars. This is 3.6 times higher than the US government's current value of $51 per tonne of CO₂. Sensitivity analyses reveal that a lower discount rate and updated damage functions are the largest contributors to the increased estimate. The four climate damage sectors included (mortality, agriculture, sea-level rise, and energy) demonstrate varying contributions to the total SC-CO₂. Mortality from temperature changes is the largest driver, with agriculture exhibiting a large range due to the complex interplay of CO₂, temperature, and crop yields. Sea-level rise's contribution is relatively small, partly due to the inclusion of optimal regional adaptation responses. Energy costs show a modest contribution. The study also compares results to the widely used DICE model and assesses sensitivity to the choice of discount rate, showing that estimates increase with lower discount rates (e.g., $80 at 3%, $118 at 2.5%, $308 at 1.5%). Furthermore, replacing sectoral damage functions with aggregate global damage functions yields similar SC-CO₂ distributions. The results suggest the current SC-CO₂ estimations might be underestimated.

The significantly higher SC-CO₂ estimate presented in this study has important implications for climate policy. The increased value substantially raises the estimated benefits of greenhouse gas mitigation, bolstering the case for more stringent climate policies. The findings emphasize the importance of incorporating the latest scientific advances in socioeconomic modeling, climate projections, and damage assessments to accurately reflect the true costs of carbon emissions. The comprehensive probabilistic approach employed in this study offers a robust framework for evaluating the risks and uncertainties associated with climate change and provides a more informed basis for policy decisions. The open-source nature of the GIVE model allows for future refinement and expansion by adding further damage sectors such as biodiversity and conflict, leading to even more comprehensive estimates.

This study provides a substantially higher estimate of the social cost of carbon dioxide ($185/tCO₂), driven by updated scientific understanding and a rigorous probabilistic approach. This revised estimate significantly increases the perceived benefits of mitigation efforts, reinforcing the need for more robust climate policies. The open-source GIVE model provides a framework for continued improvement and extension of SC-CO₂ estimations, by incorporating additional damage sectors and further refining existing ones. Future research should focus on incorporating the remaining damage categories and further exploring the uncertainties associated with tipping points within the climate system.

While this study incorporates significant advancements, limitations remain. Other damage categories, including non-market damages beyond mortality (e.g., biodiversity loss, conflict), are not fully accounted for. Incorporating these could further increase SC-CO₂ estimates. The model's reliance on specific damage functions and climate models introduces inherent uncertainty. The characterization of coastal adaptation also holds limitations. Future research needs to address these limitations for more comprehensive SC-CO₂ estimations.