By: David Keith and Anthony Harding (Georgia Tech)
Climate change has risks—and those risks are only increasing. Many of these might be reduced by solar geoengineering. Solar geoengineering also has risks, and it requires rigorous, transparent research before it is deployed. One major question though is how does the reduction in risks from solar geoengineering compare to the additional risks its use entails? Or equivalently, how big are the benefits of solar geoengineering compared to its harms?
Our paper, Impact of solar geoengineering on temperature-attributable mortality, is a first effort to provide a quantitative risk-risk comparison for any solar geoengineering method. It answers the call of an emerging consensus that if we’re going to avoid the worst effects of climate change, we need to evaluate every potential solution—including solar geoengineering. Importantly, highly credible scientific organizations like the National Academies of Sciences, Engineering, and Medicine (NASEM), and a growing chorus of experts, are among those who recognize solar engineering deserves exploration.
In our new paper, we quantify and compare what we consider to be a few of the largest physical risks for sulfate aerosol injection—mortality risk from temperature, air pollution, and ozone loss.
Injecting sulfate aerosol into the stratosphere will cool the planet, reducing mortality from heat, one of the leading risks of climate change. Sulfate aerosol air pollution is a leading cause of environmental mortality worldwide, so it is one of the most obvious risks of sulfate aerosol geoengineering. Sulfate aerosols in the stratosphere can also damage the ozone layer, causing an increase in mortality from skin cancers.
Comparing these three risks, we find that the reduction in mortality from cooling—a benefit—is roughly ten times larger than the increase in mortality from air pollution and ozone loss—a harm. Like any statement about solar geoengineering this result depends on the scenario we evaluated along with a host of other assumptions.
Our view is that quantitative analysis of the expected benefits and costs of a possible policy action is crucial input to sensible debate about public policy. This perspective is particularly relevant to solar geoengineering given its uncertainties, risks, and distributional effects. Good benefit-cost analysis should consider structural uncertainties and consider the distribution of those impacts across different affected groups. Benefit-cost analysis should not (and does not) mechanically determine policy outcomes, but good policy analysis and debate should take benefit-cost analysis seriously. Note that our view about the importance of benefit-cost analysis is reflected by the fact that we have both taught this topic in public policy schools.
In the remainder of this essay, we offer some notes and then a set of answers to questions we imagine readers might ask.
Some notes:
- Our paper is a collaboration between us and Princeton collaborators Gabe Vecchi and Wenchang Yang.
- Our paper relies on a state-of-the-art method for estimating the impact of warming on local mortality led by my UChicago colleague Michael Greenstone. The estimate of added mortality due to the addition air pollution and ozone loss comes from Seb Eastham’s paper.
- Our paper links to two prior papers. Tony led a paper on the impact of solar geoengineering on income inequality that used related econometric methods. David was part of prior collaboration with Gabe Vecchi which produced an important estimate of solar geoengineering’s potential to reduce regional climate hazards.
- Many groups have called for risk-risk analysis of solar geoengineering including the National Academy, NASEM 2021, The Carnegie Climate Governance Initiative, C2G and the call-for-balance letter with Peter Singer, James Hansen, and Bjorn Stevens, as signatories, see paper.
- The air pollution mortality estimate by Eastham combines the direct impacts of injected aerosol that makes it to the surface with climate-mediate changes in the amount of surface ozone and particulate air pollution produced from given industrial emissions. Air pollution mortality due to particular matter increases as the climate cools, a larger increase in mortality than direct impact of the sulfate injected into the stratosphere (see Figure 2 of Eastham). But this effect depends on air pollution emissions which will likely be lower late in the century than is assumed in the Eastham paper. It may be better to compare our estimated change in temperature-attributable mortality to Eastham’s estimate of direct impacts of the descending injection mass—yielding a benefit-harm ratio of about 40:1.
- Our work examines only three risk pathways: temperature-attributable mortality, air pollution, and the impact of increased ultraviolet due to reduced ozone. It is just one step toward a broad quantitative risk–risk assessment of solar geoengineering. While not comprehensive, these are important risk pathways: temperature-attributable mortality may account for more than half of the monetized harms of climate change, and air pollution and ozone loss are among the most salient impacts of stratospheric sulfate geoengineering.
How will the positive and negative impacts of solar geoengineering be distributed geographically? Research consistently suggests that those who are expected to be most harmed by a warming world are those in poorer and hotter regions of the world. Broadly, we find the converse for solar geoengineering. Cooling by solar geoengineering reduces temperature-attributable mortality in hotter regions while it increases mortality in cooler regions (Figure 1 of paper). Global warming does the converse. This, combined with the fact that mortality impacts are greater when people are poorer, means that the benefits of solar geoengineering are concentrated in hotter and poorer regions.
Did we get our result by choosing an unrealistically positive scenario for deploying solar geoengineering? On the one hand, any statement about SRM is necessarily scenario dependent. Here’s how Parson and Keith put it:
SRM presents two fundamental policy-relevant scientific questions. How effectively could it reduce climate risks? And what additional harms or risks, of what severity, would it introduce? Answers to these questions about SRM’s effects rely partly on knowledge derived from scientific research, but they also depend on assumptions about how SRM is used, under what background conditions of greenhouse gas emissions and climate change. The three principal dimensions of choice in how SRM is used are how much global-average cooling or radiative forcing is pursued, how changes in radiative forcing are distributed around the world, and what SRM method is used.
If, for example, one deployed a massive amount of SRM in only one hemisphere the results would be bad.
On the other hand, our paper expresses many of its results as ratios either of risk-to-risk or of SRM’s effect to the effect if the same climate change were caused by removal of CO2, and these ratios are strongly dependent on the amount of SRM. We expect, for example, that the 13:1 risk-risk ratios would be very roughly the same if one was cooling the world only 0.1 C or as much as 2 C, and roughly independent of how quickly emissions were cut or carbon was removed. But our scenario does assume a roughly hemispherically balanced uniform SAI deployment.
How sure are you about your results? The abstract says there is only a 60% chance that benefits would outweigh the harms. The 60% figure is almost completely driven by the uncertainty in the effect of temperature-related mortality. Forget SRM and just think about CO2-driven climate change: though we don’t calculate it exactly, there is a large probability that the benefits of reduced deaths in cooler regions would be larger than the harms of increased deaths in hot regions—so our 60% figure really comes from the fact that the model we use doesn’t have high confidence that CO2-driven climate change is bad for average mortality. Things look very different if you look at a hot or cold region alone.
What next? Do you believe this type of climate intervention should be deployed? And who should have the authority to make this kind of decision? Our role as scientists is to expand the base of knowledge. Governments must decide if, when and how to put theory into practice. We hope the main impact of this paper is to spur our colleagues to provide more and better quantitative comparisons between risks and benefits.