Heat Strain at Work Reshapes the Social Cost of Carbon in Landmark New Analysis

One of the most consequential numbers in climate policy, the social cost of carbon dioxide, has just been substantially recalibrated. In a study published in Nature Climate Change, a team led by Frances C. Moore of the University of California, Davis, together with colleagues at Purdue University, Stanford University and the University of California, Davis, has for the first time folded the economy-wide costs of heat-driven labour productivity losses into a modern integrated assessment framework, while simultaneously revising the agricultural damage component downward using the latest evidence from the Intergovernmental Panel on Climate Change. The result is a social cost of carbon dioxide of US$179 per tonne for 2025, down slightly from US$204, but with a far more complete account of what warming actually does to human work and welfare.

The social cost of carbon, often abbreviated SC-CO2, attempts to answer a deceptively simple question: how much economic damage, in dollars, does one additional tonne of carbon dioxide emitted today inflict across the entire future? The concept traces back to Pigouvian welfare economics, and in recent years it has moved from academic obscurity to the centre of regulatory policy, informing everything from power plant standards to fuel economy rules. In 2022, a landmark analysis in Nature by Rennert and colleagues pushed central estimates sharply upward, and the United States Environmental Protection Agency subsequently adopted estimates incorporating recent scientific advances. Yet the damage functions underpinning these figures have remained incomplete, and one of the most glaring omissions has been the effect of heat on the human capacity to work.

The physiological mechanism is well understood. As wet bulb globe temperature rises, the human body must divert more blood flow to the skin for cooling, heart rate climbs, and workers instinctively take more breaks or reduce their working intensity to avoid dangerous heat strain. Occupational health standards, including the widely used wet bulb globe temperature index maintained by the International Organization for Standardization, codify exactly how much work time is lost at given heat levels. Decades of field studies, from Indian rice harvesters to West Bengal brick workers and Hong Kong construction crews, have documented these losses in practice, and economic research has confirmed measurable impacts on output in manufacturing and on cognitive performance as well. What has been missing is a rigorous translation of this physiological and empirical evidence into the global, sector-resolved economic accounting that the social cost of carbon requires.

The new study closes that gap with an unusually detailed modelling chain. Qinqin Kong and Matthew Huber produced bias-corrected projections of wet bulb globe temperature under the CMIP6 climate model ensemble, correcting known model biases and computing heat stress metrics explicitly rather than through crude approximations, which earlier work has shown can materially misestimate labour losses. These projections cover three levels of work intensity and both indoor and outdoor conditions. The team then applied two distinct labour response functions, one based on the ISO occupational standard and another drawn from a separate empirical framework, to convert heat exposure into losses of effective labour capacity by job type, economic sector and region.

Crucially, the researchers did not simply multiply lost labour hours by wages. Instead, they fed the labour productivity shocks into a general equilibrium model built on the Global Trade Analysis Project database, allowing prices, trade flows, sectoral reallocation and other economic adaptations to buffer or amplify the initial shock. This is a key distinction, because heat stress does not hit the world economy uniformly. It concentrates in already hot, labour-intensive economies, propagates through global supply chains as the prices of agricultural and manufactured goods shift, and triggers substitutions that general equilibrium modelling can capture but simpler accounting cannot. The resulting damages were then expressed as regional damage functions, relating warming to welfare losses as a percentage of initial income, and incorporated into the GIVE integrated assessment framework used in recent official estimates of the social cost of carbon.

The headline result for labour is striking: heat-related labour productivity damages amount to US$41 per tonne of carbon dioxide emitted in 2025, with a 90 percent confidence interval running from US$1 to US$108. Losses are heavily concentrated in South, East and Southeast Asia and in Africa, regions where outdoor and physically demanding work remains a large share of employment and where cooling infrastructure is least widespread. Under an illustrative warming level of 1.7 degrees Celsius, the maps of projected labour capacity loss reveal a world of profound inequality, with tropical and subtropical working populations bearing damages that temperate, wealthy economies largely escape. This geographic concentration matters not only for equity but also for policy design, since it identifies where adaptation investments such as shaded worksites, adjusted working hours, mechanisation and expanded access to cooling would deliver the greatest returns.

The second major contribution of the study is a downward revision of agricultural damages. Previous estimates, including the authors’ own earlier work, had translated the findings of crop-yield meta-analyses into damage functions that implied agricultural losses of US$95 per tonne of carbon dioxide. The Sixth Assessment Report of the Intergovernmental Panel on Climate Change, drawing on a much larger body of evidence including process-based crop models and studies accounting for adaptation, carbon dioxide fertilisation and changing growing regions, supports substantially smaller aggregate impacts. Incorporating that assessment reduces the agricultural damage component to US$29 per tonne. The revision is a reminder that damage estimates are only as good as the underlying impact literature, and that as climate impact science matures, policy-relevant numbers must be updated rather than fossilised.

Netted together, the two revisions lower the expected 2025 social cost of carbon dioxide from US$204 to US$179 per tonne, using a 2 percent near-term discount rate in 2020 dollars. But the authors emphasise that the more important change may be the treatment of uncertainty. By building labour damages from explicit physiological data, bias-corrected climate projections and structural economic modelling, and by grounding agricultural damages in an authoritative assessment synthesis, the study substantially narrows the confidence interval around the social cost of carbon. For regulators, who must defend these figures in courtrooms and rulemaking dockets, a central estimate backed by a transparent, reproducible evidence chain is arguably worth as much as the point value itself.

The findings land at a politically charged moment, as governments weigh how heavily carbon damages should weigh in cost-benefit analysis and as the scientific community continues to expand the catalogue of climate impacts, from mortality and morbidity to energy demand and coastal inundation. This study demonstrates both directions of that expansion: adding a previously missing damage category centred on the world’s most vulnerable workers, while trimming another that had likely been overstated. The complete methodological chain, from gridded heat stress datasets and damage module code to the revised integrated assessment calculations, has been made openly available, allowing other researchers to scrutinise and extend the work. As the evidence base grows, the social cost of carbon is becoming less of a contested abstraction and more of a measurable summary of what each tonne of carbon dioxide truly costs the human economy, and the newest answer is that it costs most dearly in the sweat of those who work under the sun.

Subject of Research: Estimating the social cost of carbon dioxide by incorporating heat-related labour productivity damages and updated agricultural damage functions

Article Title: New labour and agricultural damages improve climate cost estimates

Article References: Moore, F. C., Haqiqi, I., Kong, Q., Rennels, L., Baldos, U., Ganapathi, H., Huber, M., & Hertel, T. (2026). New labour and agricultural damages improve climate cost estimates. Nature Climate Change. https://doi.org/10.1038/s41558-026-02749-z

Image Credits: AI Generated

DOI: 10.1038/s41558-026-02749-z

Keywords: social cost of carbon, heat stress, labour productivity, climate change damages, agriculture, integrated assessment modelling, CMIP6, wet bulb globe temperature, general equilibrium, Nature Climate Change, climate economics, adaptation