Abstract
The standard account of the climate–pollution relationship holds that the two share a root in fossil-fuel combustion, so that attacking one attacks the other. That is true about the source and it has done real good, since the immediate local health return is the strongest available argument for decarbonisation in countries indifferent to a distant temperature. This report is about the ways in which the inference fails when extended from the source to the policy. Three asymmetries separate the two problems: carbon dioxide is a stock with a residence time of centuries while fine particulate matter is a flow with a residence time of days; carbon's harm is global and non-excludable while particulate harm falls on the people who breathe the air; and, least often discussed, the two have opposite signs in the climate system, because sulphate aerosols from the same combustion scatter sunlight and brighten clouds.
The stock-and-flow asymmetry governs the penalty for delay, the reversibility of success and the metric each problem should be measured against. Delay on carbon is irreversible — the Global Carbon Budget 2025 puts the remaining 1.5°C budget at about 170 GtCO₂ against roughly 42 Gt emitted each year, some four years and exhausted before 2030 — while delay on particulates is physically recoverable, though not for the people who died in the interval. Conversely, an avoided tonne of carbon is banked permanently whereas clean air is never banked: a country that has cleaned its air can dirty it again in a single year, so air-quality gains require perpetual maintenance and climate gains do not.
The sign asymmetry is the report's centre. The IPCC's Sixth Assessment puts sulphate effective radiative forcing at about −0.9 W/m² over 1750–2019, with a range of −1.6 to −0.3, corresponding to roughly −0.5°C of offset warming: observed warming is around 1.4°C where without aerosols it would be near 2°C. Aerosol cooling peaked around 2000 and has declined since, so the unmasking is already under way. Two natural experiments make the mechanism concrete. The 2020 International Maritime Organization sulphur cap cut marine fuel sulphur from 3.5 to 0.5 per cent and shipping sulphur dioxide by roughly 80 per cent, producing a multi-model mean effective radiative forcing of 0.073 W/m² — comparable to three years of growth in carbon dioxide forcing, and appearing in no emissions inventory. China, meanwhile, cut particulate pollution by 41 per cent between 2013 and 2022 and gained about two years of life expectancy while its carbon emissions continued to rise, demonstrating that the two pollutants are separable in practice as well as in principle.
The practical consequence is a correction to the co-benefit literature and a re-ranking of interventions. Eliminating combustion cuts the carbon stock and the particulate harm together; filtering exhaust cuts the particulates and leaves the carbon entirely, removing a cooling agent and nothing else, which makes end-of-pipe control a near-term warming measure however valuable it is in health terms. The co-benefit therefore runs from climate action to health and not in reverse. The report's agenda follows: state carbon budgets conditional on an aerosol pathway, since a successful clean-air scenario implies a tighter budget rather than the same one; keep two targets and never infer progress on one from progress on the other; concentrate capital on interventions that remove combustion rather than clean it; and treat foreseeable aerosol reductions as scheduled events with radiative consequences. The report closes by rejecting explicitly the reading that aerosol cooling should be preserved — a mechanism whose operating cost is 8.1 million deaths a year, 709,000 of them children under five, is the harm itself and not an instrument, and because the aerosol is a flow while the carbon is a stock the trade fails on its own terms as well.
Executive Summary
Climate change and air pollution come out of the same chimney. They do not come out with the same sign, the same lifetime, or the same geography — and the differences decide what a combined strategy can and cannot do.
A stock and a flow
Carbon dioxide accumulates: what sets the temperature is cumulative emissions, so every tonne is permanently in the account and the remaining 1.5°C budget is about four years of current output. Fine particulates clear from the air in days. One problem punishes delay irreversibly; the other does not — except for the people who died while waiting.
Opposite signs
The same combustion emits a gas that warms and a particle that cools. Sulphate aerosols are currently masking roughly half a degree of warming: observed warming is about 1.4°C, and without aerosols it would be near 2°C. Cleaning the air removes the mask.
The co-benefit runs one way
Eliminating combustion cuts carbon and particulates together. Filtering exhaust cuts particulates and leaves the carbon entirely — removing the cooling and none of the warming. Health benefits flow from climate action; climate benefits do not flow from pollution control.
The standard account of the climate–pollution relationship is that the two share a root in fossil-fuel combustion, so attacking one attacks the other. An earlier report in this series made precisely that argument, and it is true about the source. This report is about the ways in which it is false about the policy.
Three asymmetries separate the two problems. Carbon dioxide is a stock with a residence time of centuries; fine particulate matter is a flow with a residence time of days. Carbon dioxide's harm is global; particulate harm falls on the people who breathe it. And — the asymmetry least often discussed — the two have opposite signs in the climate system, because sulphate aerosols from the same combustion reflect sunlight and cool the planet.
Two sentences need to sit at the front of this report so that nothing later can be misread. Aerosol cooling is a subsidy paid in lives. Eight million deaths a year, seven hundred thousand of them children under five, is not a climate policy; it is a catastrophe that happens to have a side effect on radiative forcing. Nothing here argues for slowing the clean-up.
What it argues is that the unmasking is a foreseeable, quantifiable consequence of doing the right thing, and that it should be budgeted rather than discovered. A carbon budget computed on the assumption of constant aerosol forcing understates the cut required. And an intervention ranking that treats all pollution control as climate-positive gets the sign wrong on an entire class of measures.
1. One Source, Two Signs
What comes out of a coal plant is not one pollutant with two names. It is at least two substances with different lifetimes, different geographies of harm, and opposite effects on the planet's energy balance.
Burning coal, oil or gas produces carbon dioxide, which is transparent to incoming sunlight and absorbs outgoing infrared radiation: it warms. The same combustion produces sulphur dioxide, which oxidises in the atmosphere into sulphate aerosol — fine particles that scatter incoming sunlight directly and, more importantly, seed brighter and longer-lived clouds. Sulphate aerosol cools. It also lodges in human lungs, which is why it appears in the mortality statistics as well as in the radiative-forcing budget.
Figure 1 — One source, two substances, two signs
Schematic. Sulphate is the dominant cooling aerosol and the one this report follows; black carbon, by contrast, warms, which is one reason aerosol accounting is harder than the diagram implies. Forcing directions are settled; magnitudes are not, and Section 3 states the range.
The shared-source framing is therefore correct and incomplete. It accurately identifies where both problems come from, and it licenses an inference that does not follow: that any measure which reduces one reduces the other. Three asymmetries break that inference, and the rest of this report takes them in turn.
2. A Stock and a Flow
The first asymmetry is in residence time, and it governs the penalty for delay, the reversibility of success, and the metric each problem should be measured against.
Carbon dioxide is a stock problem. What sets the eventual temperature is cumulative emissions rather than the rate in any given year, because the gas persists for centuries. That has an uncomfortable implication: a tonne emitted is permanently in the account, and the remaining room is finite and small.
Figure 2 — What is left of the stock
The Global Carbon Budget 2025 puts fossil CO₂ at a record 38.1 Gt, land-use change at a further 4.1 Gt, and atmospheric concentration at 425.7 ppm — 52 per cent above pre-industrial. The remaining budget for 1.5°C is assessed at about 170 GtCO₂, roughly four years at current rates and exhausted before 2030.
Fine particulate matter is a flow problem. Its atmospheric residence time is days: concentration tomorrow is very largely a function of emission tomorrow. Stop emitting and the air is clean within a week, and the health benefit begins accruing immediately and almost in full.
Three consequences follow, and each cuts differently.
- The penalty for delay. Delay on carbon is irreversible: the stock is permanent and the budget does not refill. Delay on particulates is, in the physical sense, fully recoverable — the air will be just as clean in 2040 if the clean-up happens then. But the harm in the interval is not recoverable, because the people who died of it are dead. Recoverability of the medium is not recoverability of the damage, and it is a serious error to read the flow property as permission to wait.
- The reversibility of success. A tonne of carbon avoided is banked forever; nothing can un-avoid it. Clean air is never banked. A country that has cleaned its air can dirty it again in a single year of coal expansion, which means air-quality gains require permanent maintenance while climate gains do not.
- The right metric. Carbon should be managed against a cumulative budget, because that is the physical quantity that matters. Particulates should be managed against an annual concentration, because that is the physical quantity that matters. Neither metric is informative about the other problem.
A third asymmetry travels with these and is treated at length elsewhere: carbon's harm is global and non-excludable, while particulate harm falls on the people who breathe the air. That difference is what makes air-quality policy unilaterally profitable and climate mitigation a collective-action problem, and it is examined in a companion report on the governance geometry of these goods. It is noted here because it compounds the stock-and-flow asymmetry rather than offsetting it.
3. The Cooling Subsidy
The second asymmetry is in sign, and it is the one that makes the two agendas interact adversarially in a single specific channel.
Sulphate aerosols scatter incoming sunlight and brighten clouds, and their net effect on the planet's energy balance is negative. The IPCC's Sixth Assessment puts the effective radiative forcing of sulphate over 1750–2019 at about −0.9 W/m², with a range of −1.6 to −0.3, which corresponds to roughly −0.5°C of offset warming globally, with a range of −0.1 to −0.9 (Carbon Brief's explainer on the AR6 assessment). Taking direct and cloud-mediated effects together, total aerosol cooling is presently estimated at around −0.55°C.
Figure 3 — The half-degree that is currently hidden
Observed warming today is around 1.4°C. In the absence of anthropogenic aerosol emissions, the best estimate of current warming would be roughly 0.5°C higher — putting the world near 2°C rather than 1.4°C. The masking estimate carries a wide uncertainty range, discussed below. Source: Carbon Brief, on IPCC AR6.
Two things follow from that figure, and they are of very different kinds.
The first is an accounting point. The warming already committed is larger than gross carbon accounting implies. Some part of the greenhouse effect humanity has built is currently being held off the temperature record by particles that will not persist. A carbon budget computed on the assumption that aerosol forcing stays where it is will understate the emissions reduction needed to hold any given temperature, because the baseline it is working from is artificially cool.
The second is a timing point. Aerosol cooling peaked around the year 2000 and has been declining since, as the world's largest emitters have cleaned up their air. The unmasking is not a hypothetical future event to be managed; it has been under way for two decades and will continue for as long as air-quality policy succeeds — which is to say, one hopes, indefinitely.
Where the uncertainty sits
The direction of aerosol forcing is not in scientific dispute. The magnitude is genuinely uncertain, and the range — roughly −1.6 to −0.3 W/m² for sulphate — is among the largest single uncertainties in estimates of climate sensitivity, because it interacts with clouds in ways that are hard to model. This matters for how the argument should be used. It is sound as a reason to treat committed warming as larger than it looks and to plan for unmasking. It is not sound as a basis for precise numerical claims about how much additional mitigation is required, and this report makes none.
4. The Natural Experiment
In 2020 the world ran, without intending to, something close to a controlled experiment on aerosol masking.
From 1 January 2020 the International Maritime Organization cut the permitted sulphur content of marine fuel from 3.5 per cent to 0.5 per cent, reducing sulphur dioxide emissions from international shipping by roughly 80 per cent. As climate experiments go this is unusually clean: a large, abrupt, geographically concentrated reduction in one aerosol species, over ocean surfaces where the cloud-brightening effect is strongest, with no accompanying change in carbon dioxide.
Figure 4 — The warming impulse from cleaning up shipping fuel
A multi-model assessment puts the effective radiative forcing from the 2020 sulphur cap at a mean of 0.073 W/m², with individual model means from 0.057 to 0.089 — about 0.1 W/m², comparable to the increase in CO₂ forcing between 2019 and 2022. Sources: Jordan and Henry (2024) in Atmospheric Chemistry and Physics; see also Yoshioka et al. (2024) and Carbon Brief's analysis.
The modelled reductions were largest over the busiest lanes — the North Atlantic, the Caribbean and the South China Sea — which is where the cloud brightening had been most intense and where its removal therefore shows up most strongly.
The right reading of this episode is precise. The sulphur cap was good regulation: sulphur oxides from shipping cause acid deposition and respiratory disease, and coastal and port populations bear the cost. It should have been done, and it should have been done earlier. It also delivered, as a by-product, a warming impulse of roughly the same size as three years of growth in carbon dioxide forcing — and it delivered it in a form that appeared in no country's emissions inventory and no mitigation plan.
That is the pattern this report is about. Not a trade-off to be resolved in favour of dirty air, but a consequence to be anticipated. IMO 2020 was a foreseeable unmasking event, and it was foreseen by the aerosol modelling community. It was not, in any meaningful sense, budgeted for.
5. Clean Air Without Decarbonisation
China's air-quality programme is the largest and fastest in history. It also demonstrates that the two agendas are separable in practice, not merely in principle.
Since the 2013 Air Pollution Action Plan, China has achieved the most rapid large-scale improvement in air quality on record. Particulate pollution fell by about 41 per cent nationally between 2013 and 2022, and Beijing province by 54.1 per cent in nine years.
Figure 5 — How fast clean air can arrive
On the University of Chicago's Air Quality Life Index, the average Chinese citizen can now expect to live about two years longer than at 2013 pollution levels. National population-weighted mean PM2.5 fell from 61.8 to 42.0 µg/m³ between 2013 and 2017 alone (Zhang et al., PNAS).
Two years of life expectancy for an entire population, delivered in under a decade, is among the largest public-health achievements of the century. It is also, for the purposes of this report, a demonstration.
Over broadly the same period, China's carbon dioxide emissions continued to rise. The particulate clean-up was achieved through end-of-pipe controls on power and industry, fuel-quality standards, dispersal of heavy industry, and coal-to-gas switching in northern cities — measures that attack the particles directly and leave most of the combustion, and therefore most of the carbon, in place.
So the shared-root inference fails empirically as well as theoretically. The world's largest emitter cut its particulate burden nearly in half while its carbon emissions grew, which is only possible because the two pollutants can be separated. It follows that the clean-up contributed to the global unmasking, and that the health dividend and the climate dividend were not, in this case, delivered together.
What this does and does not imply
It does not imply that China should have polluted for longer. Two years of life expectancy across 1.4 billion people is not a cost to be weighed against a fraction of a degree — and a planet kept cool by particulates is a planet kept cool by killing people, which is an indefensible instrument whatever its radiative properties. What it implies is that the clean-up should have been paired with carbon reduction rather than substituted for it, and that an accounting framework which credits particulate control as climate progress will systematically mis-rank interventions.
6. What Follows for Intervention Choice
Three classes of intervention, which differ not in degree but in which of the two problems they actually touch.
Table 1 — Interventions by what they remove
| Intervention | Cuts the carbon stock | Cuts particulate harm | Near-term net radiative effect |
|---|---|---|---|
| Eliminate combustion — renewables displacing coal, electrified transport and heat | Fully | Fully | Warming from lost aerosol, but the stock falls too |
| Efficiency — using less of the same fuel | Partly | Partly | Small, in both directions |
| Substitute the fuel — coal to gas | Partly | Largely | Ambiguous; methane leakage can reverse it |
| Filter the exhaust — scrubbers, particulate filters, fuel-sulphur limits | Not at all | Fully | Net warming: the cooling goes, the carbon stays |
| Disperse the source — relocating industry away from population | Not at all | Locally | Neutral to warming; harm is moved, not removed |
Ratings are analytical judgements on the direction and completeness of each effect, not measured magnitudes. The final column is the report's specific contribution: it is negative for exactly the measures most often counted as climate co-benefits.
Read the last two rows carefully, because they are the uncomfortable ones. End-of-pipe pollution control is, in near-term radiative terms, a warming measure. A scrubber removes the sulphate that was cooling the planet and removes none of the carbon dioxide that is heating it. So does a particulate filter, and so did the shipping sulphur cap.
This should not be read as an argument against scrubbers. It is an argument about where the co-benefit lies, and the direction matters enormously for how interventions are ranked:
- From climate action to health: the co-benefit is real. Eliminating combustion removes the carbon and the particles together. The health return arrives quickly and locally — which, as another report in this series argues, is the leg on which the whole bundle is most easily justified.
- From pollution control to climate: the co-benefit is negative. Measures that clean the exhaust while leaving the fuel remove a cooling agent and nothing else. They are excellent health policy and they make the near-term temperature problem slightly worse.
The practical ranking is therefore unambiguous even though the accounting is awkward. Prefer interventions that remove the combustion, because they are the only class that addresses both problems. Do the filtering too — 8.1 million deaths a year settles that question — but count it honestly, as health spending with a small adverse climate side effect, and pair it with mitigation rather than crediting it as mitigation.
7. Two Targets, One Plan
What changes if the two problems are treated as physically distinct rather than as one problem with two names.
Table 2 — What the two-signs view changes
| Common practice | What the physics implies |
|---|---|
| Treat air quality and climate as one agenda with shared metrics | Keep two targets — an annual concentration target and a cumulative carbon budget — and never infer progress on one from progress on the other. |
| Compute carbon budgets on today's aerosol forcing | State budgets conditional on an aerosol pathway. A successful clean-air scenario implies a tighter carbon budget, not the same one. |
| Count all pollution control as climate co-benefit | Count only combustion elimination that way. Filtering is health spending with an adverse near-term radiative effect. |
| Report air-quality success as climate progress | Report it as what it is — life expectancy gained — and separately disclose the unmasking it implies. |
7.1 Budget for the unmasking
- Publish carbon budgets conditional on a stated aerosol trajectory, so that the additional mitigation implied by successful air-quality policy is visible in advance rather than absorbed as a forecast error.
- Treat large, foreseeable aerosol reductions — a shipping fuel rule, a national coal clean-up — as scheduled events with radiative consequences, and assess them jointly at the point of adoption. IMO 2020 was predictable and was not planned for.
7.2 Keep the two targets separate
- Maintain distinct instruments and distinct reporting: concentration standards and health outcomes for air; cumulative budgets and emissions inventories for carbon. Success on either is not evidence about the other, and China's decade demonstrates how far the two can diverge.
- Prevent the metrics from substituting for one another in political accounting, where an air-quality improvement is the more visible and faster-arriving achievement and is therefore the more tempting one to report.
7.3 Rank by whether combustion is removed
- Give priority in capital allocation to interventions that eliminate combustion rather than clean it, on the grounds that they are the only class that reduces both the stock and the harm.
- Continue and accelerate end-of-pipe control on its own merits, budgeted as health protection, and offset its radiative effect explicitly rather than implicitly.
7.4 Do not treat the cooling as an asset
One reading of this report would be that aerosol cooling is worth preserving, or at least that clean-air policy should be slowed while mitigation catches up. That reading is wrong on two independent grounds, and it is worth closing it off explicitly.
The first is moral and decisive. The cooling is produced by particles that kill 8.1 million people a year, including 709,000 children under five. A cooling mechanism whose operating cost is measured in premature deaths is not a policy instrument; it is the harm itself, and proposing to sustain it deliberately is not a serious position.
The second is technical, and it holds even for anyone unmoved by the first. The aerosol is a flow and the carbon is a stock. Maintaining the mask requires emitting the particles continuously and forever, while the carbon dioxide underneath continues to accumulate. The strategy therefore commits to a permanent and rising death toll in exchange for a delay that grows shorter every year, and it ends with the mask removed anyway — at a higher stock, and therefore a higher temperature, than if the clean-up had happened first. The trade is not merely repugnant; it does not work on its own terms.
7.5 Conclusion
Climate change and air pollution are the two best-known consequences of burning fossil fuels, and the fact that they share a source has made it natural to treat them as one problem. That framing has done real good: it is the strongest available argument for decarbonisation in countries indifferent to a distant temperature, because the health return is immediate, local and very large.
But it is an argument about the source, and it has been quietly extended into an argument about the policy, where it does not hold. The two substances differ in residence time by five orders of magnitude, in the geography of their harm, and in the sign of their effect on the planet's energy balance. A strategy that does not distinguish them will mis-measure committed warming, mis-rank its own interventions, and be surprised by the warming that its air-quality successes reveal.
The correction is narrow and does not require anyone to accept less clean air. Keep two targets and two metrics. State carbon budgets conditional on the aerosol path, so that the cost of clean air appears on the ledger where it belongs rather than as a shock. And concentrate capital on the one class of intervention that removes the source rather than the symptom — because eliminating combustion is the only measure that cuts the stock and clears the air at the same time, and the only one for which the co-benefit story is true in both directions.
References
Every quantitative claim above is attributed inline. The principal sources are collected here.
- IPCC, via Carbon BriefExplainer: How human-caused aerosols are 'masking' global warming — sulphate effective radiative forcing of −0.9 W/m² (−1.6 to −0.3) over 1750–2019, about −0.5°C of offset warming, total aerosol cooling near −0.55°C, and the observation that aerosol cooling peaked around 2000.
- IPCCSixth Assessment Report, Working Group I: The Physical Science Basis — the underlying assessment of aerosol and greenhouse-gas radiative forcing.
- Atmospheric Chemistry and PhysicsMulti-model effective radiative forcing of the 2020 sulfur cap for shipping (2024) — multi-model mean ERF of 0.073 W/m², model means from 0.057 to 0.089.
- Atmospheric Chemistry and PhysicsWarming effects of reduced sulfur emissions from shipping (2024).
- Carbon BriefAnalysis: How low-sulphur shipping rules are affecting global warming.
- Global Carbon ProjectGlobal Carbon Budget 2025 — fossil CO₂ at a record 38.1 Gt, land-use change 4.1 Gt, 425.7 ppm, and a remaining 1.5°C budget of about 170 GtCO₂.
- Health Effects InstituteState of Global Air 2024 — 8.1 million deaths in 2021, the second-leading risk factor for death worldwide, including 709,000 children under five. Also at stateofglobalair.org.
- University of Chicago EPICChina's air quality policies have swiftly reduced pollution, improved life expectancy — a 41 per cent fall in particulate pollution from 2013 to 2022, 54.1 per cent in Beijing province, and about two years of life expectancy gained. See also the Air Quality Life Index summary.
- PNASDrivers of improved PM2.5 air quality in China from 2013 to 2017 — national population-weighted mean PM2.5 falling from 61.8 to 42.0 µg/m³.
- UNEPEmissions Gap Report 2025 — the distance between current policy and any Paris-consistent pathway.
- H HeuristicsEarlier reports in this series: the shared-root argument this report qualifies; the health leg as the fastest-paying case for the bundle; and the governance geometry of global and local goods.
Metadata
- Keywords
- air pollutionaerosol maskingsulphate aerosolradiative forcingclimate co-benefitscarbon budgetPM2.5IMO 2020shipping emissionsChina air qualityunmaskingstock and flow pollutantsend-of-pipe controlclimate policy
- JEL classification
- Q53, Q54, Q58, I18, Q48 — air pollution and environmental externalities; climate and natural disasters; environmental economics and government policy; government policy and health; energy policy
- Data and method
- This report synthesises the atmospheric-science literature on aerosol radiative forcing with institutional assessments of emissions and air quality, including the IPCC Sixth Assessment Working Group I as summarised by Carbon Brief; two 2024 multi-model studies in Atmospheric Chemistry and Physics on the radiative effect of the 2020 shipping sulphur cap; the Global Carbon Budget 2025; the Health Effects Institute's State of Global Air 2024; the University of Chicago Air Quality Life Index and Zhang et al. in PNAS on China's particulate decline; and UNEP's Emissions Gap Report 2025. Sources on aerosol forcing and the shipping sulphur cap were verified by search in September 2026. Every quantitative claim is attributed inline. Figure 1 is a conceptual schematic; it follows sulphate, the dominant cooling aerosol, and notes that black carbon warms, so aerosol accounting is more complex than the diagram implies. Table 1 records analytical judgements on the direction and completeness of each intervention's effect rather than measured magnitudes. The report states explicitly that while the direction of aerosol forcing is not in scientific dispute, its magnitude carries a wide uncertainty range which is among the largest in estimates of climate sensitivity, and it therefore makes no precise numerical claim about the additional mitigation implied. The report is analytical rather than predictive.
- Report
- H Heuristics Digital Report № 2026-08 · Published 12 September 2026
- Licence
- CC BY-NC-ND 4.0
- Cite as
- Hunter Hughes (2026). One Source, Two Signs: Addressing the global climate–pollution polycrisis when the two problems share a chimney but not a physics. H Heuristics Digital Report 2026-08. https://digitalreports.hheuristics.com/reports/climate-pollution-two-signs/