Over the past few decades, we have improved our understanding of the health impacts of climate change.1 Although many public health researchers have contributed to this knowledge, relatively few are aware of how their work may relate to the social cost of carbon. The social cost of carbon is a core economic concept in climate policy and one that can—and should—benefit directly from research produced by the public health community. The concept’s importance was recently highlighted by this past year’s Nobel Prize in Economic Sciences, which was awarded to William Nordhaus in part for his pioneering work developing models to estimate the social cost of carbon. Below we describe this concept, explain how it is calculated, and provide some brief guidance on how health research can improve its estimation.
THE SOCIAL COST OF CARBON
The social cost of carbon is the net economic damage to society that results from one additional ton of CO2 emissions. The concept is widely used in regulatory cost-benefit analysis to value the impacts of CO2 emissions, making it one of the most important metrics in climate economics. In the United States, e.g., the use of the social cost of carbon in cost-benefit analysis is mandated by law, and it has been applied to dozens of federal regulations ranging from vehicle emission standards to power generation, including the Clean Power Plan.2 Because the social cost of carbon represents the climate-related costs of releasing an additional ton of CO2 emissions, from an economic perspective it also reflects the optimal carbon tax level to apply in a carbon tax regime, should governments choose to adopt one.
Called “the most important number you’ve never heard of,”3 the social cost of carbon that is officially recognized by the United States government for regulatory analysis is calculated by equally weighting the outputs of three integrated assessment models (described below). Other governments (e.g., United Kingdom) have also used one or more of these models for the same purpose, or have based their own social cost of carbon metric on the US estimates (e.g., Canada, Mexico).
A distinguishing feature of these models is that their backbone is comprised of “damage functions”: sets of equations that compute the harm to society that results from an extra ton of CO2 emissions. These functions are based on empirical studies across multiple dimensions that estimate the potential impacts from climate change. These impacts are then monetized in an effort to place their relative harm to humans (economic and/or physical) onto a single scale.
Until now, most of the studies underlying damage functions have been undertaken by economists, including the studies quantifying health impacts; public health researchers have not been made active participants. In acknowledgment of this shortcoming (and others), a key recommendation of a recent National Academy of Sciences report was to improve damage functions by engaging all relevant experts and incorporating up-to-date disciplinary science.4 Therefore, conducting research applicable to damage functions is a unique opportunity for epidemiologists and other health professionals to directly inform climate policy.
To that end, in the remaining sections we summarize the damage functions of the three leading models used to calculate the social cost of carbon, and argue that health experts should be fundamental to the process of updating these functions. We also briefly describe specific study design features that can help make research results applicable to future damage functions.
DAMAGE FUNCTIONS IN LEADING SOCIAL COST OF CARBON MODELS
The three leading models used to estimate the social cost of carbon are the Dynamic Integrated Climate and Economy (DICE) model, the Policy Analysis of the Greenhouse Effect (PAGE) model, and the Climate Framework for Uncertainty, Negotiation, and Distribution (FUND) model. The former was developed in the United States and the latter two in Europe. These models each have an economic module and a climate module, which are linked (Figure). Changes in economic activity generate climate change by virtue of the emissions that are associated with that activity. At each time period, damages from climate change are monetized, aggregated, and then subtracted from the gross domestic product.
The three models differ in terms of which impacts are included in their damage functions and the degree of detail incorporated for each. Only FUND explicitly computes mortality and morbidity impacts, which are then transformed into economic damage. In contrast, the other two models (DICE and PAGE) do not directly estimate any specific type of impact but instead aggregate them in more generalized damage functions where a total amount of damage is computed, with human health implicitly assumed to comprise some percentage of that total. In this aggregated framework, the total damages are based on meta-analyses of individual studies which differ widely in terms of how health impacts are considered.5–8
Importantly, a number of climate-sensitive diseases that have become established in new locations—or are expected to in the future—are not currently featured (explicitly or implicitly) in existing damage functions. Examples include West Nile Virus, Lyme Disease, Zika Virus, and Chikungunya.9–13 Several other key health risks are also excluded (Table). However, even those health risks that are included are often based on incomplete or outdated estimates. For instance, one study used in DICE’s damage function relies on the 1996 edition of the Global Burden of Disease, whereas cardiorespiratory impacts in FUND are based on mortality rates in 1990.14,15 These shortcomings may partly explain why the health costs are sometimes small compared to those from many other sectors,8 which does not correspond to more recent work.16,17
An additional consideration is that some social cost of carbon models also estimate optimal decarbonization trajectories, derived by balancing mitigation costs against climate benefits. In the standard modeling framework, this tradeoff has ignored the potential near-term health impacts of mitigation—sometimes referred to as “co-benefits” or “co-harms.” Recent studies22–24 have demonstrated that, at least in the case of air pollution, these are likely to be net-positive overall despite opposing influences; the health co-benefits from improved air quality seem to outweigh the warming co-harms that may occur when reducing emissions of air pollutants such as SO2 and NOx that act to cool the atmosphere. The implication is that social cost of carbon models may be recommending artificially low levels of mitigation if they omit health co-benefits. However, this result again depends in part on the damage functions calculating climate impacts: if warming is much more costly than currently assumed, the loss of the aerosol cooling effect could counterbalance more of the health benefit, a trade-off often neglected in the literature,25 but one that recent studies have shown how to incorporate.22,24
IMPROVING FUTURE DAMAGE FUNCTIONS
The omission of numerous important health endpoints, together with the use of outdated data or methods to calibrate existing damage functions, underscores recent arguments that the human health impacts of climate change—and consequently the social cost of carbon—may be underestimated.17,26 As a result, there is a clear need for health professionals to contribute to this research agenda if these damage functions are to be comprehensively upgraded.
Fortunately, compared to 10 or 20 years ago when the studies underlying most of the current damage functions were conducted, our knowledge of climate–health relationships is substantially improved. This new knowledge is reflected in the publication of numerous recent studies projecting the health impacts of climate change, several of which report large potential disease burdens.16,17,27–32
How would a large correction to these damage functions affect the social cost of carbon? As a single example, the authors of one recent econometric study calculated a partial social cost of carbon—estimated from mortality impacts alone—and found that it may be comparable to the full metric employed under the Obama administration which included impacts across all dimensions, both health and otherwise.17 Other studies are needed to confirm, expand, or revise this type of result, and such research should be conducted in close collaboration with epidemiologists and other public health experts to ensure that the best possible science is applied.
New and improved damage functions that reflect our best scientific understanding require strong interdisciplinary collaboration as well as a clear procedure for noneconomists to contribute to model development. Prototypes of modeling platforms that streamline this collaboration are underway (e.g., see Resources for the Future’s Social Cost of Carbon initiative33 and the Mimi modular modeling framework34). In addition, there are a number of features particular to the damage functions that health experts should consider when designing research studies related to climate change impacts (also see Refs. 4 and 8). We highlight four:
- (1) Global and regional coverage: Social cost of carbon models are global models comprised of individual regions, each with its own damage function. As a result, the ideal study to improve their estimation of climate impacts—for example from heat waves or extreme weather—would also be global in scope and would use uniform methods to assess all regions differentially. Although epidemiologists may not be entirely comfortable making inferences for areas with data limitations, excluding these areas would require other (potentially nonspecialist) researchers to extrapolate the results from one region to another.
- (2) Scenarios and metrics: Climate damage functions are generally continuous with temperature, though some impacts may additionally result from changes in sea-level rise, CO2 concentrations, or ocean temperature. However, many health impact studies report results for only one or a few time-points or temperature/emission scenarios. These studies cannot be readily matched to social cost of carbon models. We recommend that researchers conduct their evaluations across as wide a range of scenarios and time horizons as possible. Additionally, because future paths are multi-dimensional (socioeconomic, climatic etc.),35 framing results solely as the number of additional deaths or life years lost from climate change can sometimes be difficult to contextualize; multiplicative effect estimates (e.g., incidence rates; fraction of attributable deaths) are preferable.
- (3) Vulnerable groups: Experts generally agree that the poor and disadvantaged will suffer the most from climate change.36,37 Therefore, it is important that human health research attempts to characterize climate risks in vulnerable populations. Groupings to consider include sex, age, urban/rural populations, income quantiles, and other populations of concern. Building such evidence will make it possible to take advantage of a powerful and unique feature of social cost of carbon models: their ability to analyze policy while explicitly accounting for societal preferences about fairness and equity.38,39 Specifically, the models include a social welfare function, an equation which translates economic measures (dollars) into estimates of wellbeing, thus giving researchers and decision makers the ability to compute the social cost of carbon with alternative choices about how much to prioritize specific populations. For example, previous work has shown that if climate change will in fact disproportionately harm the poor, then the social cost of carbon would be much higher—perhaps by an order of magnitude—than suggested by conventional estimates that ignore the distribution of burdens.39
- (4) Social change and adaptation: Over the coming decades, societies will experience rapid changes that may affect their vulnerability to climate impacts. Some types of change, such as demographic transitions, can affect climate vulnerability without adaptation, but the adoption of explicit adaptation measures is also likely. Many studies have already documented sizeable reductions in climate vulnerability over time, at least for certain health outcomes.40–42 Nevertheless, these features are still routinely neglected in estimates of future health burdens, which limits their use in social cost of carbon models.
The above criteria refer specifically to the quantification of health outcomes, but a change in health status across a population also produces a cascade of economic consequences, which is ultimately how climate damage functions value climate impacts. For instance, increased mortality or morbidity can affect labor supply and productivity, population growth, overall economic productivity and so on.6,16,40,43,44 A gold-standard study would include both the physical (e.g., deaths) and economic effects of climate-induced health burdens; the importance of these economic adjustments has been discussed elsewhere.4,8
AN EXAMPLE FROM THE LITERATURE
For further illustration, we highlight some of these issues by way of a recent study by Gasparrini et al (2017) that used a dataset of daily deaths from 451 locations in 23 countries to project future all-cause mortality from outdoor temperature. The authors first estimated location-specific exposure-response relationships, and then projected deaths under different climate change scenarios. Although this study uses state-of-the-art epidemiological methods, and is the largest of its kind, it does not meet all the criteria described above:
- (1) Despite the expansive dataset, the majority of the world’s countries and regions are excluded, including several populous regions that may be highly vulnerable to climate change (e.g., Africa, South Asia). Providing specific guidance on how to estimate temperature effects in data-sparse regions is beyond the scope of this article, but promising avenues include advanced spatial smoothing techniques and extrapolation based on correlated indicator variables, such as macroeconomic measures or population characteristics. We also note that Demographic and Health Surveys (DHS) are continually released and include ever more detail45; although not directly applicable to the study design of Gasparrini et al, DHS surveys allow for the estimation of temperature effects using other methods.20
- (2) The study did analyze multiple climate scenarios and time periods that would thus enable the estimation of burdens across a continuous range of temperatures.
- (3) The study does not consider subgroups and assumes no demographic changes. The latter has the advantage of isolating the impact of climate change but disregards the influence of population growth and composition. These effects could be at least partially incorporated by applying age-sex stratified exposure-response relationships—increasingly common in the literature46,47—to long-term mortality projections, which are available (by age and sex) from the United Nations as well as the Shared Socioeconomic Pathway project.48,49
- (4) The study assumes no adaptation. As with extrapolation to data-sparse regions, future adaptation effects could potentially be explored using projections of macroeconomic or population variables.
Much of this discussion centers on dealing with data limitations, whether in geographic regions, sub-populations or about the future. We are aware of several ongoing projects aiming to apply epidemiologically based approaches to these problems. Another method is to employ simple sensitivity tests that span the range of possibilities. This approach has been used with social cost of carbon models in the past, for example to bound the possible implications of alternative pathways of future economic development and to analyze within-country inequalities.39,50
The social cost of carbon is one of the most influential metrics in climate policy. It depends critically on the potential human health impacts of climate change, but our best estimates of these impacts are not currently incorporated into the leading social cost of carbon models. Likely reasons include a lack of communication between disciplines and a lack of true interdisciplinary teams, as well as the time gap between the production of evidence and its use in policy applications. Therefore, as the social cost of carbon modeling community prepares to move towards a new generation of models, it is imperative that we avoid previous pitfalls by ensuring that the public health community contribute its expertise, both through the inclusion of existing evidence and through the design of future studies.
ABOUT THE AUTHOR
About the authors can be found in https://links.lww.com/EDE/B551.
M.B.B. is supported by a Catalyst Award from the Gund Institute for Environment at the University of Vermont. A.G. is supported by grants from the Medical Research Council UK (ID: MR/M022625/1) and Natural Environment Research Council (ID: NE/R009384/1). F.D. acknowledges support by Yale NUS College (through grant number R-607-264-235-121). N.S., F.E., F.D., D.S., and M.B.B. thank Princeton’s Climate Futures Initiative for support.
1. Smith KR, Woodward A, Campbell-Lendrum D, et al. Field CB, Barros VR, Dokken DJ, et al. Human health: impacts, adaptation, and co-benefits. In: Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part A: Global and Sectoral Aspects. Contribution of Working Group II to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. 2014;Cambridge, NY: Cambridge University Press, 709–754.
2. Environmental Protection Agency. Regulatory Impact Analysis for the Clean Power Plan Final Rule. Research Triangle Park, 2015. Available at: https://www3.epa.gov/ttnecas1/docs/ria/utilities_ria_final-clean-power-plan-existing-units_2015-08.pdf
3. Wihbey J; Yale Climate Connections. Understanding the Social Cost of Carbon- and Connecting It to Our Lives. 2015. Available at: https://www.yaleclimateconnections.org/2015/02/understanding-the-social-cost-of-carbon-and-connecting-it-to-our-lives/
. Accessed 1 July 2019.
4. National Academies of Sciences. Valuing Climate Damages: Updating Estimation of the Social Cost of Carbon Dioxide. 2017.Washington, DC:The National Academies Press.
5. Warren R, Hope C, Mastrandrea M, Tol R, Adger N, Lorenzoni I. Spotlighting impacts functions in integrated assessment. Tyndall Centre on Climate Change Working Paper (91). 2006. Available at: https://ueaeprints.uea.ac.uk/20207/
6. Tol RS. The economic effects of climate change. J Econ Perspect. 2009;23:29–51.
7. Nordhaus WD. Revisiting the social cost of carbon. Proc Natl Acad Sci U S A. 2017;114:1518–1523.
8. Diaz D, Moore F. Quantifying the economic risks of climate change. Nat Clim Chang. 2017;7:774–782.
9. Paz S. Climate change impacts on West Nile virus transmission in a global context. Phil Trans R Soc B. 2015;370:20130561.
10. Brownstein JS, Holford TR, Fish D. Effect of climate change on lyme disease risk in North America. Ecohealth. 2005;2:38–46.
11. Paz S, Semenza JC. El Niño and climate change–contributing factors in the dispersal of Zika virus in the Americas? Lancet. 2016;387:745.
12. Ali S, Gugliemini O, Harber S, et al. Environmental and social change drive the explosive emergence of Zika virus in the Americas. PLoS Negl Trop Dis. 2017;11:e0005135.
13. Medlock JM, Leach SA. Effect of climate change on vector-borne disease risk in the UK. Lancet Infect Dis. 2015;15:721–730.
14. Nordhaus WD, Boyer J. Warming the World: Economic Models of Global Warming. 2000.Cambridge, England: MIT Press.
15. Anthoff D, Tol R. The Climate Framework for Uncertainty, Negotiation and Distribution (FUND), Technical Description, Version 3.9. Available at: http://www.fund-model.org/versions
. Accessed 29 November 2018.
16. Hsiang S, Kopp R, Jina A, et al. Estimating economic damage from climate change in the United States. Science. 2017;356:1362–1369.
17. Carleton T, Delgado M, Greenstone M, et al. Valuing the global mortality consequences of climate change accounting for adaptation costs and benefits. Working Paper 2018–51. 2018.Chicago, IL: University of Chicago.
18. Tol RS. On the optimal control of carbon dioxide emissions: an application of FUND. Environ Model Assess. 1997;2:151–163.
19. Hope C. The PAGE09 integrated assessment model: a technical description. Cambridge Judge Business School Working Paper. 2011;4(11)
20. Geruso M, Spears D. Heat, humidity and infant mortality in the developing world. NBER Working Paper Series. 2018.Cambridge, MA: National Beurea of Economic Research.
21. Garzon-Villalba XP, Mbah A, Wu Y, et al. Exertional heat illness and acute injury related to ambient wet bulb globe temperature. Am J Ind Med. 2016;59:1169–1176.
22. Ikefuji M, Magnus JR, Sakamoto H. The effect of health benefits on climate change mitigation policies. Clim Change. 2014;126:229–243.
23. Bollen J, van der Zwaan B, Brink C, et al. Local air pollution and global climate change: a combined cost–benefit analysis. Resour Energy Econ. 2009;31:161–181.
24. Scovronick N, Budolfson M, Dennig F, et al. The impact of human health co-benefits on evaluations of global climate policy. Nat Commun. 2019;10:2095.
25. Armstrong B, Hajat S, Kovats S, et al. Climate change: how can epidemiology best inform policy? Epidemiology. 2012;23:780–784.
26. Stern N. Economics: current climate models are grossly misleading. Nature. 2016;530:407–409.
27. World Health Organization. Quantitative risk assessment of the effects of climate change on selected causes of death, 2030s and 2050s World Health Organization, 2014. Available at: https://apps.who.int/iris/bitstream/handle/10665/134014/9789241507691_eng.pdf?sequence=1
28. Lloyd SJ, Kovats RS, Chalabi Z. Climate change, crop yields, and undernutrition: development of a model to quantify the impact of climate scenarios on child undernutrition. Environ Health Perspect. 2011;119:1817–1823.
29. Caminade C, Kovats S, Rocklov J, et al. Impact of climate change on global malaria distribution. Proc Natl Acad Sci U S A. 2014;111:3286–3291.
30. Guo Y, Gasparrini A, Li S, et al. Quantifying excess deaths related to heatwaves under climate change scenarios: a multicountry time series modelling study. PLoS Med. 2018;15:e1002629.
31. Martinez GS, Williams E, Yu SS. The economics of health damage and adaptation to climate change in Europe: a review of the conventional and grey literature. Climate. 2015;3:522–541.
32. Gasparrini A, Guo Y, Sera F, et al. Projections of temperature-related excess mortality under climate change scenarios. Lancet Planet Health. 2017;1:e360–e367.
33. Resources For The Future. Social Cost of Carbon Initiative. Available at: http://www.rff.org/research/collection/rffs-social-cost-carbon-initiative
Accessed 26 November 2018.
34. Moore FC, Rising J, Lollo N, et al. Mimi-PAGE, an open-source implementation of the PAGE09 integrated assessment model. Sci Data. 2018;5:180187.
35. O’Neill BC, Kriegler E, Ebi KL, et al. The roads ahead: narratives for shared socioeconomic pathways describing world futures in the 21st century. Global Environ Chang. 2017;42:169–180.
36. Hallegatte S, Bangalore M, Bonzanigo L, et al. Shock Waves: Managing the Impacts of Climate Change on Poverty. 2015.Washington, DC: The World Bank.
37. Olsson L, Opondo M, Tschakert P, et al. Field CB, Barros VR, Dokken DJ, et al. Livelihoods and poverty. In: Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part A: Global and Sectoral Aspects. Contribution of Working Group II to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. 2014;Cambridge, NY: Cambridge University Press, 793–832.
38. Adler M, Anthoff D, Bosetti V, et al. Priority for the worse-off and the social cost of carbon. Nat Clim Chang. 2017;7:443.
39. Dennig F, Budolfson MB, Fleurbaey M, et al. Inequality, climate impacts on the future poor, and carbon prices. Proc Natl Acad Sci U S A. 2015;112:15827–15832.
40. Carleton TA, Hsiang SM. Social and economic impacts of climate. Science. 2016;353:aad9837.
41. Vicedo-Cabrera AM, Sera F, Guo Y, et al. A multi-country analysis on potential adaptive mechanisms to cold and heat in a changing climate. Environ Int. 2018;111:239–246.
42. Deschênes O, Greenstone M. Climate change, mortality, and adaptation: evidence from annual fluctuations in weather in the US. Am Econ J Appl Econ. 2011;3:152–185.
43. Kjellstrom T, Kovats RS, Lloyd SJ, et al. The direct impact of climate change on regional labor productivity. Arch Environ Occup Health. 2009;64:217–227.
44. Bosello F, Roson R, Tol RS. Economy-wide estimates of the implications of climate change: human health. Ecol Econ. 2006;58:579–591.
45. AbouZahr C, de Savigny D, Mikkelsen L, et al. Civil registration and vital statistics: progress in the data revolution for counting and accountability. Lancet. 2015;386:1373–1385.
46. Achebak H, Devolder D, Ballester J. Heat-related mortality trends under recent climate warming in Spain: a 36-year observational study. PLoS Med. 2018;15:e1002617.
47. Scovronick N, Sera F, Acquaotta F, et al. The association between ambient temperature and mortality in South Africa: a time-series analysis. Environ Res. 2018;161:229–235.
48. United Nations Population Division. World Population Prospects, the 2017 Revision. Available at: http://esa.un.org/unpd/wpp/Download/Standard/Population/
. Accessed 11 December 2017.
49. Lutz W, Butz WP, Samir K. World Population and Human Capital in the Twenty-First Century. 2014.Oxford, United Kingdom: Oxford University Press.
50. Budolfson M, Dennig F, Fleurbaey M, et al. Optimal climate policy and the future of world economic development. World Bank Econ Rev. 2018;33:21–40.