What climate does not entail

“Are not poverty and need the greatest polluters?”

Two questions get glued together in climate talk. First: is anthropogenic warming real, and does it raise material risks? Second: does that fact require degrowth, permanent emergency, or the suspension of open, innovative order? The first is largely settled. The second is not entailed by the first — and treating it as if it were is how serious physics becomes bad systems design.

What follows takes the physics as given and asks what the evidence supports on capacity, carbon and growth, damages, and prices — including where serious estimates disagree.

What is settled

Anthropogenic climate change is real. Energy systems, land use, and industrial processes have altered the planetary carbon balance. The IPCC AR6 physical science summary is sufficient for that claim; this essay does not relitigate it.¹ Warming raises risks — heat, extremes, coastal pressure, agricultural stress in vulnerable places. Denying that is not seriousness.

Treating every risk as a warrant for command economies, permanent emergency, or deliberate impoverishment is not seriousness either. Risk is a design input for liberal institutions: prices, property, science, accountable government. The disagreement worth having is not “is the atmosphere a commons?” It is which institutions expand capability under that constraint without closing the open order that produces capability.

The case that has to be answered

The strongest version of the other side is not a cartoon of people who “hate growth.” It is this: the atmosphere is a shared budget; overshoot is cumulative; those least responsible for historical emissions often pay first; and “more of everything” can become a polite way of exporting harm. If you take distribution and carbon lock-in seriously, waiting for clean abundance can look like a luxury story told from high ground. You want hard limits, faster phase-outs, and a politics that does not treat the poor of the next generation as residual risk.

That side also has a number. The Stern Review (2006) put the welfare cost of business-as-usual climate change at the equivalent of losing at least 5% of global GDP each year, now and forever, rising toward 20% if a wider set of risks and impacts is counted — with the cost of strong action put near 1% of GDP per year.¹³ That is a different object from Tol’s comparative-static damage meta (Table C below): Stern aggregates risk over time with a very low pure-time discount rate and treats uncertainty as a reason for stronger early action. If those assumptions are right, my working prior is too calm. The steelman is not “feelings”; it is that figure under those assumptions.

That case still deserves a real answer, not a culture-war reverse. The question is whether the response that often follows — degrowth as moral project, permanent emergency as operating system — actually protects the people it claims. As system designs, extreme alarmism and degrowth thin the toolkit when adaptation, grids, denser housing, resilient agriculture, and cleaner baseload all require capital, knowledge, and institutional competence. Bjørn Lomborg has pressed this point for years: climate is serious, but policy that ignores opportunity cost and adaptation can leave people poorer without buying proportional safety.²

The pandemic was a natural experiment in less activity. Global fossil CO₂ fell by only about 5–6% in 2020 (roughly 5.4% in the Global Carbon Budget series) — a record annual drop, and still a small fraction of what deep Paris-compatible cuts would require on a sustained path — then rebounded by about 4.9% in 2021.¹⁴ The human bill was not small: World Bank estimates put on the order of tens of millions more people in extreme poverty in 2020 (updated tallies around ~90–100 million), with hunger and working poverty rising after decades of decline.¹⁴ Hitting Paris-scale emission cuts mainly by shrinking economic activity would be, in crude arithmetic, like replaying a pandemic-scale shock every year into 2030 without counting on a recovery rebound — and without the temporary character that made 2020 survivable as a one-off. That is not a climate strategy; it is a poverty strategy with a thin emissions coupon.

Gandhi’s Stockholm question lands here, not as folklore. Poverty and unmet need force the short-run choices that strip forests, burn dirty fuels, and postpone environmental priority; only past a threshold of prosperity do societies routinely invest in cleaner air, safer buildings, and long-horizon climate policy.¹⁶ Capacity compounds when knowledge compounds. Climate does not exempt us from that arithmetic. Terminal language crowds out high-leverage work: permitting reform, markets for clean power, industrial learning curves, carbon prices with teeth.

Mortality, wealth, and adaptive capacity

A useful prior on adaptation is not “nothing bad happens.” It is that on average and over time, open and richer societies get better at preventing death, recovering from shocks, and substituting better tools for worse ones — the pattern Steven Pinker maps across health, safety, and material welfare.³ Over the last century, global death rates from natural disasters have fallen by more than 90%, according to EM-DAT series summarized by Our World in Data — a rate decline even while population grew several-fold and reporting improved.⁴

That is not proof that climate risk is trivial. Insured losses can climb while deaths fall, because more wealth sits in harm’s way. It is proof that capacity matters: early warning, building codes, logistics, medicine, insurance, governance. Poor societies die more from the same storm class; richer ones absorb better. Johan Norberg’s progress work makes the operational claim: the floor of adaptive capacity rises with the tools and institutions of development, especially where people are still exiting energy poverty.⁵ Strategies that deliberately starve wealth and technology in the name of climate virtue trade present human safety for symbolic purity — unless they can show, with evidence, that the starved capacity buys more safety downstream than it costs now. That burden is rarely met.

A distributional caveat belongs here, not later. Aggregate charts can feel like an insult in a place with a ruined harvest. The correction is not to deny the global series. It is to refuse using the series as a shrug — and to insist that the remedy is more capability where people are exposed, not permanent impoverishment as climate virtue.

Inst 01 CAPACITY

OWID/EM-DAT: global death rates (per 100k) by decade — 1920s ≈ 26.5 → 2010s ≈ 0.6. Hover points.

Deaths per 100,000 / yr (decade mean) Table A points

Myth. If climate risk is real, disaster death rates must be rising. Mechanism. Global rates fell by more than 90% over the last century (OWID/EM-DAT; Table A: 1920s mean 26.5 → 2010s 0.6 per 100k) even as population and reporting grew — capacity, not milder hazards alone. Move. Raise adaptive capacity where people are still exposed. Hover points · full series on OWID.

Decoupling, carefully scoped

Liberalism is not a license to dump costs on the commons. It is a bet that when the right costs are priced and experiments are allowed, people invent toward lower harm. On carbon, that bet is no longer purely theoretical. Several rich economies have grown while cutting territorial CO₂. One clean window is the United Kingdom, 1990–2022. End-points from the OWID/GCP compilation:⁶

That is intensity and a large absolute territorial cut in one rich economy — and a smaller but still negative consumption-based cut. It is not a global absolution. Absolute global fossil CO₂ (fuels + cement, Global Carbon Budget scope) was about 36.8 GtCO₂ in 2023 and was projected near 37.4 GtCO₂ in 2024 — record territory, not a completed descent. Land-use change is additional (on the order of ~4 GtCO₂ in recent GCB years), so total anthropogenic CO₂ sits higher still.⁷ Intensity can fall while the global total stays high or rises, because other regions industrialize and energy demand grows. So “decoupling is possible” is not “the planet is done.” It is a narrower fact: the joint of prosperity and carbon per unit of output can be broken. Where it has been broken, the mechanism is technique and fuel substitution under pressure — including price pressure — not a demand that societies stop wanting better lives.

Vaclav Smil’s energy histories are the guardrail on that claim. National intensity can fall while global primary energy and materials use remain large: steel, cement, ammonia, and long-distance freight do not vanish by rhetoric, and transitions are capital-heavy and slow even when they are real.⁸ If the joint of prosperity and intensity can still open, the rational move is technologies and prices that make clean options the cheap default. That is conditional optimism in Matt Ridley’s sense: evidence of learning curves and substitution under open innovation, not faith that markets forgive every externality without institutions.⁹

Inst 02 DECOUPLE

UK 1990–2022 (Table B): GDP +74%, territorial CO₂ −48%, consumption CO₂ −25%. Accent = unhooked; ink = still coupled.

Unhooked path (UK-type) Still coupled Cleaner + richer direction

Myth. Prosperity and carbon are forever locked; any growth story is greenwash. Mechanism. UK 1990–2022 (Table B): real GDP +74%, territorial CO₂ −48%, consumption CO₂ −25% (OWID/GCP). Global fossil CO₂ still near record (~36.8–37.4 Gt, GCB) — local unhooking ≠ global descent. Move. Prefer techniques and prices that make clean options the cheap default. Click a point · scatter is teaching layout; Table B is the anchor case.

Prices, damages, and mind-change

Capacity without a price still dumps on the commons; a price without substitutes still lands first on people with the thinnest buffer. The instrument that matches both constraints is a predictable, broad carbon price under general rules — compressed information, not a revolving industrial plan that freezes technology menus or replaces discovery with administration. William Nordhaus’s climate-economics tradition is the clearest statement of that instrument: put a price on the externality so decentralized investment can respond, rather than pretending administrators can pick the entire energy menu in advance.¹⁰ Prefer the first; judge the second by evidence, not green branding. Ex-post reviews find that serious carbon prices tend to reduce emissions; Jessica Green’s 2021 review of ex-post studies is a useful entry point, without claiming magic or uniform effect sizes.¹¹

One jurisdiction makes the effect size concrete. British Columbia’s revenue-neutral carbon tax (phased in from 2008) is among the cleanest natural experiments: a broad price on fuels with proceeds returned through other tax cuts. Empirical and simulation reviews place the emissions reduction in the province at roughly 5–15% relative to a no-tax counterfactual in the early years of the policy, with little evidence of a large hit to overall economic activity in those studies.¹⁵ That is not “solve global climate with one province.” It is a measured price signal doing what Green’s survey says prices tend to do — cut emissions without requiring a permanent emergency or a collapse in GDP of the COVID type.

The bill is still real. A serious carbon price hits fuel, freight, and food-adjacent costs first where margins are thin. It can punish the urban poor before it rewires a grid. Revenue recycling and substitutes — transit, efficient housing, clean power — are not optional add-ons if the price is to be a bridge rather than a wall. British Columbia’s design stressed recycling for a reason. A climate politics that refuses that cost is incomplete liberalism, not “pro-market rigor.”

On economic damages, meta-analyses disagree — and the disagreement is the point. Two published objects side by side (both at about 3 °C, both as % of GDP, but not the same structural assumptions):¹²

Nordhaus’s integrated-assessment models sit in the same conversation from another angle: non-zero damages that rise with temperature, and an optimal path that still prices carbon rather than treating the median case as either trivial or unbounded catastrophe.¹⁰ I use Tol’s lower-to-mid level band (~1–2%) as a working prior for scale, not a comfort blanket: not zero, not civilizational wipeout in the median non-catastrophic cases. Howard–Sterner-type upper bands reweight urgency. Stern’s 5–20% welfare-equivalent BAU range reweights it harder still — if that risk aggregation is the right one.¹³

What would change my mind: robust evidence that non-catastrophic damages systematically sit well above the Tol-type ~1–2% band and that adaptation cannot buy them down at reasonable cost — or that absolute emissions pathways leave no room for the capacity story above — or that Stern-style welfare losses under careful discounting and risk aggregation dominate after honest comparison of objects. Meanwhile, if global product multiplies several times over the century under continued growth, the same percentage is a larger absolute number and a smaller share of a vastly larger capacity to pay for adaptation. That is the absorption point: the size of the system matters as much as the size of the shock. Count deaths, costs, substitutes, and trade-offs — including those that make a liberal prior uncomfortable. COVID showed that a global activity shock buys only a single-digit emissions cut at a vast human cost; it did not show that repeating the shock is a Paris pathway.

Inst 03 ABSORB

Table C: Tol ~1.5–2% level vs Howard–Sterner higher band at ~3 °C. Capacity multiplies; damage band held for scale check.

Tol-type ~1–2% level band System capacity growing

Myth. Median non-catastrophic damages are civilizational wipeout at a few degrees. Mechanism. Table C: Tol (2024) ~1.5–2% level near 3 °C vs Howard & Sterner (2017) preferred ~5–10% (wider with growth/catastrophe). Disagreement is the point. Capacity can multiply while a level band is held for scale. Move. Price the externality; compound capacity. Reweight if Howard–Sterner-type bands hold and adaptation cannot buy them down. Click to re-run · not a forecast · n. 12.

Implications

Prefer open systems that compound: abundant clean energy, industrial learning under competition, coastal and agricultural resilience, and prices that tell the truth about carbon without romanticizing poverty. Be suspicious of policies that require permanent fear to stay funded, and of solutions that require permanent administrators to stay legitimate. What scales is cost curves under real energy constraints, institutions that learn, and instruments that still work when the weather is not metaphorical.

None of this is triumphalism. Climate risk is real. Political risk is real. Distributional risk is real — including the risk that preferred tools land first on people with the least buffer. The sober move is the same as in any other complex domain: protect adaptive capacity, improve the information in prices and science, and build instruments that work under load.

1. Physical baseline: IPCC, Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (Cambridge University Press, 2021); Summary for Policymakers — ipcc.ch/report/ar6/wg1. This note takes anthropogenic warming and elevated risks as given; it does not re-argue attribution. ↩
2. Prioritization, opportunity cost, and adaptation under serious climate risk: Lomborg, B. (2020), False Alarm: How Climate Change Panic Costs Us Trillions, Hurts the Poor, and Fails to Fix the Planet (Basic Books); Lomborg, B. (2007), Cool It: The Skeptical Environmentalist’s Guide to Global Warming (Knopf). Agree or disagree with his policy ranking; the usable demand is to count costs, benefits, and alternatives rather than treat every risk as an unlimited warrant. ↩
3. Measured long-run gains in health, safety, and material living standards under open, science-using societies: Pinker, S. (2018), Enlightenment Now: The Case for Reason, Science, Humanism, and Progress (Viking), esp. chapters on life, health, and the environment. Not a climate damage model; supplies the prior that knowledge and capacity compound. ↩
4. Global death rates from natural disasters: Our World in Data — Natural disasters; grapher natural-disaster-death-rates (EM-DAT / CRED, UCLouvain). Table A: decade means of annual World rates (deaths per 100,000) computed from that series — 1920s 26.5, 1950s 2.1, 1970s 2.5, 1990s 0.8, 2010s 0.6, 2020s partial 0.7. OWID’s prose statement is a >90% century-scale rate decline; 1920s→2010s is on the order of 97%. Early decades are volatile (few very large events). Absolute deaths can also fall despite larger population; economic losses can still rise with exposure. Fig. 01 plots Table A points. ↩
5. Development, open institutions, and rising adaptive capacity: Norberg, J. (2016; expanded ed. 2020), Progress: Ten Reasons to Look Forward to the Future (Oneworld); related essays at johannorberg.net. Complements Pinker (n. 3) on the institutional side of capacity. ↩
6. UK 1990–2022 end-points (Table B): OWID CO₂ dataset owid/co2-data (owid-co2-data.csv), country = United Kingdom. 1990: gdp = 1.506×10¹²; co2 (territorial) = 601.9 Mt; consumption_co2 = 666.0 Mt. 2022: gdp = 2.615×10¹²; co2 = 311.1 Mt; consumption_co2 = 502.8 Mt. Changes: GDP +74%; territorial CO₂ −48%; consumption CO₂ −25%. GDP is OWID’s real international-$ series (linked to Maddison/World Bank pipelines in their docs); percent change is the decoupling metric used here. See also OWID — CO₂ and GDP. Territorial accounts can look cleaner than consumption-based footprints when carbon-intensive production is imported — Table B shows both. ↩
7. Global fossil CO₂ (fuels + cement; GCB treatment of cement carbonation as in source): Friedlingstein, P., et al. (2024), “Global Carbon Budget 2024,” Earth System Science Data (Global Carbon Project) — summary globalcarbonbudget.org; also OWID — CO₂ emissions. Approximate levels cited: ~36.8 GtCO₂ (2023); ~37.4 GtCO₂ projected (2024). Land-use change CO₂ is reported separately (~4 GtCO₂ in recent GCB years). Scope is fossil (+ cement), not all greenhouse gases. ↩
8. Energy systems, materials, and the pace of real transitions (power density, grids, steel, cement, fertilizers, freight): Smil, V. (2017), Energy and Civilization: A History (MIT Press); Smil, V. (2022), How the World Really Works: The Science Behind How We Got Here and Where We’re Going (Viking). Constraint set, not a brief for inaction: scale and inertia are part of the map. ↩
9. Innovation, exchange, and conditional optimism (problems solved by better tools under open orders, not by wishing for less ambition): Ridley, M. (2010), The Rational Optimist: How Prosperity Evolves (Harper); Ridley, M. (2020), How Innovation Works: And Why It Flourishes in Freedom (Harper). ↩
10. Carbon pricing and integrated assessment as mainstream climate-economics architecture: Nordhaus, W.D. (2013), The Climate Casino: Risk, Uncertainty, and Economics for a Warming World (Yale University Press); Nordhaus, W.D. (2019), “Climate Change: The Ultimate Challenge for Economics,” American Economic Review 109(6): 1991–2014 — AER (Nobel lecture summarizing the DICE tradition). Nordhaus is the pricing instrument and IAM framing, not a single GDP headline; read with Tol (n. 12) for damage meta-estimates and Green (n. 11) for ex-post price effects. ↩
11. Ex-post evidence that carbon prices reduce emissions (with heterogeneous effect sizes): Green, J.F. (2021), “Does carbon pricing reduce emissions? A review of ex-post analyses,” Environmental Research Letters 16(4) — IOP open access. Read with the British Columbia case study (n. 15). ↩
12. Damage meta-analyses disagree on magnitude (Table C). Tol: Tol, R.S.J. (2024), “A meta-analysis of the total economic impact of climate change,” Energy Policy 185, 113922 — ScienceDirect / open PDF via Tinbergen series. Comparative-static total-impact estimates; mean of estimates near 2.5 °C ≈ 1.7% of income (paper’s reading of the cloud); working prior here ≈ 1.5–2% of GDP level impact around 2.5–3 °C. Howard & Sterner: Howard, P.H., and Sterner, T. (2017), “Few and Not So Far Between: A Meta-analysis of Climate Damage Estimates,” Environmental and Resource Economics 68: 197–225 — preferred non-catastrophic specifications commonly summarized near ~5–10% of GDP at 3 °C; reported model span about 1.9–17.3%. Later update: Howard, P.H., et al. (2025), “Methodology Matters: A Careful Meta-Analysis of Climate Damages,” Environ. Resour. Econ. — non-catastrophic level 3.2% and with growth effects 9.2% at 3 °C (higher with catastrophe add-ons). Level ≠ growth-rate tails; objects differ. Working prior in this note remains Tol-type level; Howard–Sterner bands are one reweight trigger; Stern (n. 13) is another, under different discounting and risk aggregation. ↩
13. High-urgency welfare-equivalent damage range (steelman number): Stern, N. (2006), The Economics of Climate Change: The Stern Review (Cambridge University Press / HM Treasury). Summary of conclusions: without action, overall costs and risks “equivalent to losing at least 5% of global GDP each year, now and forever,” rising toward 20% or more with a wider range of risks and impacts; costs of strong action “around 1% of global GDP each year.” Archive summary: UK National Archives / HM Treasury. Object differs from Tol (n. 12): intertemporal welfare under low pure-time discounting and risk aggregation, not a single comparative-static level damage at 3 °C. Contested (see Nordhaus and others on discount rates); cited here as the best-known quantitative steelman of early, strong action, not as the working prior. ↩
14. COVID-19 as a natural experiment in less activity (not a climate policy). Emissions: Global Carbon Project / Friedlingstein et al., Global Carbon Budget 2021 cycle — fossil CO₂ down about 5.4% in 2020, rebound about 4.9% in 2021 (near pre-pandemic levels); rounded in the essay as “about 5–6%.” See GCP summary. Peak daily drops were larger (Le Quéré et al., Nature Climate Change, 2020); annual rates are the policy-relevant scale. Poverty: World Bank estimates of additional extreme poverty due to the pandemic cluster in the tens of millions to ~100 million for 2020 — e.g. updated estimate of about 97 million more people in extreme poverty in 2020 (World Bank Data Blog, June 2021); UN SDG reporting of a sharp 2019–2020 rise in the global extreme-poverty rate and setbacks on hunger/working poverty (UNSD SDG 1). The “Paris via less activity = pandemic every year to 2030” line is illustrative arithmetic about scale and human cost, not a formal integrated assessment: sustained multi-percent annual cuts from demand destruction alone would require repeated output shocks of COVID-like magnitude without assuming a full rebound. ↩
15. Carbon-price case study — British Columbia: Murray, B., and Rivers, N. (2015), “British Columbia’s revenue-neutral carbon tax: A review of the latest ‘grand experiment’ in environmental policy,” Energy Policy 86: 674–683 — ScienceDirect (working-paper version via Duke Nicholas Institute). Review of empirical and simulation evidence: emissions reduced by roughly 5–15% relative to counterfactual since implementation; negligible impact on overall economic activity in the studies they survey. Revenue neutrality (recycling via other tax cuts) is part of the design. Jurisdiction-scale, not global proof; effect sizes vary by model and period. ↩
16. Epigraph: Gandhi, I. (1972), address to the United Nations Conference on the Human Environment, Stockholm, June 1972. Exact wording: “Are not poverty and need the greatest polluters?” — often misquoted as “Poverty is the greatest polluter.” Context was development and environment, not a license for unlimited emissions: she linked environmental care to employment and basic needs. See e.g. Guardian retrospective; Down to Earth (2022) on the speech text. ↩