Yves here. This study on the mortality costs of climate change may seem unduly simple in its construction but it still makes an important point. The lifespan cost of higher temperatures is greater than can be explained by the direct effects of heat, such as heatstroke, alone. More extreme weather patterns, increasingly out of line with historical planting and harvest cycles, endanger life, limb, and livelihoods.
By Adrien Bilal, Assistant Professor of Economics Stanford University, Diego Känzig, Faculty Research Fellow National Bureau Of Economic Research; Assistant Professor Northwestern University, and Krzysztof Lisiecki, Predoctoral Research Fellow, Stanford Institute for Economic Policy Research Stanford University. Originally published at VoxEU
Mortality is one of the most salient consequences of climate change. Existing estimates have largely focused on the direct effects of local heat exposure. This column argues that this perspective can miss an important part of the overall health burden. New evidence based on fluctuations in global temperature suggests that climate change also affects mortality indirectly through its effects on economic activity and living standards. Accounting for these broader channels substantially increases estimated mortality damages and highlights an important role for economic resilience in limiting the health consequences of climate change.
Western Europe has just experienced its warmest summer on record, while August 2026 was the warmest August ever recorded globally. These episodes have brought the health consequences of extreme heat to the fore. According to the World Health Organization (WHO), they have caused more than 200,000 deaths across Europe over the past four years (WHO Europe 2026). The mortality consequences of local heat exposure, the potential for adaptation, and the pressures on healthcare systems are by now well understood (Barreca et al. 2016, Carleton et al. 2022, 2025, Aguilar-Gomez et al. 2025).
Yet, local heat exposure may capture only part of the mortality consequences of climate change. Climate change also affects economic activity. If a warmer world lowers incomes, disrupts healthcare provision, or slows improvements in living standards, these economic effects can themselves translate into higher mortality. This observation raises the question: how much does global warming affect mortality once we account for the wider economic consequences of a changing global climate?
In new work, we address this question by estimating the effect of changes in global mean temperature on worldwide mortality (Bilal et al. 2026a). We find substantially larger mortality effects than implied by local heat exposure alone. Our evidence suggests that much of this difference can be traced back to the macroeconomic consequences of higher global temperatures.
From Local Heat to Global Temperature
Most empirical research on climate and mortality exploits variation in local weather. Researchers can, for example, compare mortality in years when France experiences unusually high temperatures with years in which it does not, while controlling for developments common to all countries.
Local heat exposure, however, captures only one dimension of a warming climate. Climate change is best understood as an increase in global mean temperature. A rise in global mean temperature is associated with broader changes in the climate system – not only hotter days on land, but also changes in ocean temperatures, precipitation and wind patterns, drought, and other weather extremes. Estimates based on local temperature variation may therefore miss some of the aggregate consequences of a changing global climate. In earlier work, we showed that this distinction matters for economic activity (Bilal and Känzig 2026). Here, we ask whether it also matters for mortality.
We assemble annual mortality data for more than 100 countries between 1960 and 2019, covering roughly 70% of the world population, and combine them with global temperature and economic data. A central challenge is that both temperature and mortality exhibit long-run trends. Simply comparing these trends could reveal a relationship between the two even if changes in mortality were driven by entirely different factors. To overcome this challenge, we exploit fluctuations in global temperature around its long-run trend, which arise in part from natural climate variability, such as El Niño, volcanic eruptions, and solar activity.
Figure 1 The effect of a global temperature shock on mortality
Notes: Impulse responses of global mean temperature and mortality rate (deaths per 1,000 individuals) to a 1°C global temperature shock. Shaded areas: 90% and 95% confidence bands.
Following a global temperature shock scaled to 1°C, mortality rises gradually and peaks after around 12 years at roughly one additional death per 1,000 people. Because the underlying temperature shock is persistent but eventually fades, we also translate these estimates into the effect of a permanent temperature increase. A sustained 1°C increase in global temperature raises annual mortality by approximately 2.5 deaths per 1,000 people in the long run.
This magnitude is considerably larger than what we obtain from local heat exposure. Using the same mortality data and a comparable empirical design, a permanent 1°C increase operating through extreme heat – holding income fixed – raises annual mortality by around 0.1–0.3 deaths per 1,000, a comparable amount to what existing research has established. This is far below the effect we estimate from global temperature. Moreover, reduced exposure to extreme cold partly offsets the mortality effects of additional heat, further widening the gap.
Why Does Global Temperature Affect Mortality So Much More?
One possibility why global temperature affects mortality so much more is that global temperature captures other extreme weather events missed by measures of local heat. We therefore examine droughts, extreme precipitation, and extreme wind alongside heat and cold. These events matter, but even taken together they account for only a fraction of the mortality response to global temperature.
The critical missing channel appears to operate through economic activity. There is a longstanding relationship between income and health: richer societies tend to have lower mortality, partly because higher income allows greater spending on nutrition, healthcare, sanitation, infrastructure, and medical innovation (Preston 1975, Cutler et al. 2006). At the same time, our earlier work estimates that higher global temperature causes large and persistent declines in world GDP (Bilal and Känzig 2026).
Figure 2 Direct exposure to extreme weather and the indirect income channel
>
Notes: Panel (a) shows how much of the mortality response can be explained by extreme weather, with the dashed line isolating direct effects, holding global income fixed. Panel (b) shows the mortality effect operating through global income, alone and combined with the direct effects of extreme weather.
Combining this temperature–GDP relationship with separate estimates of the effect of global GDP on mortality from a companion paper (Bilal et al. 2026b) allows us to quantify the income channel. The resulting indirect mortality response tracks the directly estimated effect of global temperature well. Quantitatively, declines in global GDP account for about two-thirds of the cumulative mortality response. Direct effects of heat and other extreme weather events account for another 23%.
We find further corroborating evidence for this mechanism. Cardiovascular disease and HIV/AIDS account for much of the mortality increase following global temperature shocks, while deaths explicitly attributed to heat and cold make up a relatively small share. Mortality effects are also larger in countries where public healthcare accounts for a smaller share of total healthcare expenditure. Together, these findings are consistent with economic disruptions and access to healthcare playing an important role in shaping mortality outcomes.
Mortality Changes the Welfare Cost of Climate Change
What do these estimates imply for the assessment of future climate damages? To answer this question, we embed our empirical estimates in a neoclassical growth model augmented with mortality risk and health spending. Households can adapt to higher mortality risk by spending more on health, and economic growth itself improves health through the link between aggregate income and mortality. We use the observed responses of output and mortality to temperature shocks to estimate the model’s climate-damage functions.
Consider a scenario in which global temperature rises by another 2°C between 2026 and 2100, reaching roughly 3°C above pre-industrial levels. Under our estimated damage functions based on global temperature, the model implies around six additional deaths per 1,000 people by the end of the century.
The mortality consequences are also quantitatively important for the social cost of carbon. Combining economic and mortality damages, our baseline estimate of the social cost of carbon is approximately $1,640 per ton of CO₂, measured in 2026 international dollars. Mortality accounts for around $400 per ton, or one-quarter of the total. Estimates based on local temperature and heat exposure instead imply a total social cost of carbon of roughly $250 per ton, with a smaller mortality component.
Two considerations are worth keeping in mind when translating these findings into long-run climate damages. First, our empirical estimates exploit moderate historical temperature fluctuations that are persistent but eventually fade. As with most estimates of climate impacts, applying these relationships to sustained warming of several degrees requires some degree of extrapolation. Reassuringly, however, we find little evidence of nonlinearities within the range of temperature shocks observed in the data.
Second, and to address this limitation, our model explicitly accounts for two important margins of adaptation: adjustments in private health spending and the health benefits of economic growth. Additional forms of technological, institutional, and behavioural adaptation may further shape the consequences of climate change in the future.
Figure 3 The social cost of carbon: Mortality and the economy
Notes: The bars show the social cost of carbon and its decomposition into consumption losses (output), indirect mortality losses through lower output, and direct mortality losses. The left bar uses damages estimated from global temperature, while the right bar uses local temperature and extreme heat.
Economic Adaptation Can Also Be Health Adaptation
The results also broaden the way we think about adaptation. Adaptation to the health consequences of climate change is often associated with air conditioning, heat warnings, or investments in hospital capacity. These are clearly important. But if a substantial part of climate-related mortality operates through economic activity, policies and investments that make economies more resilient to climate shocks may have health benefits as well.
This channel is quantitatively important in our model. Without background economic growth, the same warming scenario would result in around nine, rather than six, additional deaths per 1,000 people by 2100. By contrast, shutting down the ability of households to adjust private health spending has comparatively modest effects.
The broader conclusion is that the health consequences of climate change cannot necessarily be inferred from heat exposure alone. A warmer global climate changes local weather, but it also changes the global economy. If economic disruptions worsen health outcomes, the mortality effects of climate change extend well beyond the deaths that occur during heatwaves.
Understanding those indirect effects matters both for measuring the damages from climate change and for thinking about adaptation. Protecting people from a warmer climate may involve not only protecting them from heat, but also protecting the economic and health systems on which their lives depend.
See original post for references
