In the first global, empirical study of mortality risk to capture both the costs and benefits of climate adaptation, we find the increased global mortality burden from climate change to be 3.7% of global GDP by the end of the century if past emissions trends continue. The analysis is conducted with the most comprehensive data set ever compiled on mortality, historical temperature, income, and climate simulations. For example, it is derived from data covering 399 million deaths all over the world over the last several decades—accounting for different incomes, climates, and varying levels of development — and modeling future population and income growth.
Utilizing a unique, big data approach to generate global information, we find that a metric ton of CO2 emitted today generates a median mortality risk of roughly $39 in a high emissions scenario. This value represents a “partial” social cost of carbon (SCC) that accounts for the value of lost life, as well as the net costs of adaptation to withstand temperature extremes, but excludes all other climate impacts. As a basis of comparison, a leading model that had not relied on similar data-driven approaches estimated the total mortality impacts of climate change at less than $1.50 per metric ton (Diaz 2014).
The analysis is conducted with the most comprehensive data set ever compiled on mortality, historical temperature, income, and climate simulations. For example, it is derived from data covering 399 million deaths all over the world over the last several decades—accounting for different incomes, climates, and varying levels of development — and modeling future population and income growth.
Our research demonstrates the importance of updating estimates of the SCC used in policymaking to incorporate empirically-derived damage estimates. For comparison, the Obama administration’s most recent estimate of the economy-wide cost of climate change was roughly $41 per ton. In 2017, the National Academies of Science recommended ways in which the U.S. government should update this estimate to incorporate the most recent scientific and economic research. The Impact Lab’s work shows that updating the mortality cost estimates alone would significantly alter any economy-wide number. Among our key findings:
- Even after accounting for adaptation, an additional 1.5 million people die per year from climate change by 2100 if past emissions trends continue. For comparison, road injuries killed roughly 1.4 million people worldwide in 2016, and diabetes, ranked as the seventh leading cause of death worldwide, killed 1.6 million people in 2016. These projections include net gains in many regions of the world where lives will be saved from fewer cold days.
- Extreme heat is measurably deadlier for the poorest third of the world, and the decline in cold-related deaths does not offset the harm caused by temperature rise. Higher incomes make societies more resilient to extreme heat, allowing people to make a range of protective investments, including in air conditioning and better building insulation. But for the most vulnerable developing countries, even optimistic economic growth projections do not provide complete protection. The findings show warming caused by an additional ton of CO2 harms 72 percent of the global population, while the rest benefit on net, primarily due to a decrease in cold days.
- Each of the study’s 24,378 regions exhibits a unique relationship between mortality and temperature. To capture these effects in a way that is relevant to policymakers, the value of lost life and the net costs of life-saving adaptations are expressed in death-equivalents. These calculations are done for all regions in the world. By 2100, for example, the researchers project that climate change will cause annual damages equivalent to approximately 3,700 deaths in Mogadishu, Somalia, but generate benefits equivalent to roughly 1,100 lives in Oslo, Norway.
- In dollar terms, this paper’s empirically-grounded estimates of mortality risks substantially exceed mortality risks in the models that underlie previous U.S. government estimates of the social cost of carbon. Only one of the three integrated assessment models supporting the figure — roughly $41 in per ton of CO2 emitted in 2015 — allows calculation of a partial cost assigned to mortality risk. The Climate Framework for Uncertainty, Negotiation and Distribution (FUND) values three comparable health impacts of climate change — diarrhea, vector-borne diseases, and cardiovascular/respiratory impacts – without data-driven evidence of how these outcomes are actually impacted by changing climates. The FUND estimate for these three impacts is less than $1.50 per ton, while the Climate Impact Lab values mortality risks at roughly $39, reflecting the importance of examining historical data from diverse and globally representative populations.
- The amount people spend to adapt accounts for roughly two-thirds of total damages, with the value of actual lives lost accounting for the remainder. Declining cold-related deaths will benefit some parts of the world, while the impacts of high temperatures will be lowest for some well-adapted regions. For example, Northern Europe, Singapore, the Andes, and Alaska, stand to gain from climate change. Several regions that are relatively wealthy and hot today, such as Australia, Saudi Arabia, and the southeastern United States, are already so heavily adapted to their hot climate that the findings show additional warming will lead to limited additional mortality or adaptation costs.
- Vulnerability to extreme temperatures depends on a location’s climate and its level of income, which must be tracked at a local scale. The costs of mortality are distributed unevenly around the world, and extreme heat is measurably deadlier for the poorest populations of the world. These findings were only possible due to the collection and analysis of high-resolution data covering nearly half of the global population, which also account for the positive impacts of reduced cold-related deaths as the world warms.
Our Reseach Summary explains more about the design and policy implications of this study.