What climate costs
How economists put a figure on damage, why their results diverge so widely, and what the social cost of carbon really covers.
Updated on 15 November 2026 · Policy and governance, Natural hazards, Health and populations
What does a tonne of CO₂ cost? The question looks technical; it is the most value-laden in the whole climate file. Published answers span more than an order of magnitude, and the spread does not come from calculation errors.
Three costs not to be conflated
Public debate routinely mixes three distinct quantities.
The cost of damages is what warming destroys: lost harvests, damaged buildings, unworkable days, shortened lives. It is suffered.
The cost of mitigation is what cutting emissions costs: investment, retraining, assets abandoned before amortisation. It is incurred deliberately.
The social cost of carbon is an estimate of the first, brought down to the tonne: the discounted value of all future damage caused by a tonne emitted today. It is an object of calculation, not an observation, and it is where the disagreements concentrate.
The discount rate, or the weight of the future
Discounting brings a future damage back to a present value. A rate of 5 % reduces a damage occurring a century hence to under a hundredth of its value; a rate of 1.5 % leaves it at nearly a quarter.
Two justifications are opposed. The descriptive approach takes rates observed in markets, on the grounds that they reflect real trade-offs. The prescriptive approach objects that a market rate expresses living agents' preference for their own consumption, and that there is no ethical reason to apply the same discount to generations with no say.
No statistic settles this. It is wholly normative, and an estimate that does not publish its rate cannot be interpreted.
Health damages, long under-counted
Classical damage functions aggregated losses in points of GDP, which captures mortality poorly. More recent work isolates that component and converts it into deaths attributable to a given quantity of emissions.
Bressler, 2021The exercise remains sensitive to the value placed on a statistical life, another normative parameter, but it shifts the result: the health component turns out to be of the same order as the entire economic damage previously counted.
On the side of damages already incurred, attribution now allows the share of an extreme event's costs imputable to warming to be quantified.
Newman et al., 2023Risk as finance sees it
The financial sector uses a different split, into two families. Physical risk is that of damages suffered: destroyed assets, interrupted supply chains. Transition risk is what a rapid decarbonisation would create for the assets that depend on fossil carbon — reserves rendered unexploitable, industries whose model rests on carbon being free.
The two move in opposite directions along a trajectory: vigorous action lowers the first and raises the second. No trajectory cancels both, which is why a banking supervisor cannot work from a single scenario.
Crona et al., 2021What policy evaluation shows
Beyond prospective estimates, a retrospective literature examines what schemes already in force have produced. The synthesis of ex post evaluations of carbon pricing finds real but generally modest reductions, strongly dependent on the level of the price and on the breadth of exemptions.
Döbbeling-Hildebrandt et al., 2024This is a useful result because it bears on observed facts rather than projections. It does not say the instrument fails; it says its effect is proportionate to its ambition, which is unremarkable and yet rarely absorbed into debates about its principle.
Kikstra et al., 2022Reading a cost figure
Three questions are enough to place an estimate. What discount rate? Which damage components are included, and which are absent for want of being quantifiable — ecosystems, migration, conflict? And does the estimate aggregate losses weighted by income, in which case a death in a poor country mechanically counts for less than a death in a rich one?
An estimate that answers these three questions is usable even if one disputes its choices. One that does not is not.
Sources
Studies cited in this brief
The mortality cost of carbon
Open accessR. Daniel Bressler · Nature Communications · 2021
- Health and populations
- Policy and governance
Rebecca Newman et al. · Nature Communications · 2023
- Natural hazards
- Policy and governance
The Anthropocene reality of financial risk
Open accessBeatrice Crona et al. · One Earth · 2021
- Policy and governance
- Natural hazards
Jarmo Kikstra et al. · Geoscientific model development · 2022
- Policy and governance
- Modelling and projections
Easwaran J. Narassimhan et al. · Climate Policy · 2018
- Policy and governance
Systematic review and meta-analysis of ex-post evaluations on the effectiveness of carbon pricing
Open accessNiklas Döbbeling-Hildebrandt et al. · Nature Communications · 2024
- Policy and governance