What Do Economists Say Warming Will Cost?

Economists agree that each degree of warming costs more than the last, and disagree tenfold on how much. Richard Tol's early net benefit has vanished in his later surveys. At 3 °C, sector models put the loss near 2 % of world income, growth studies at 20 % or more. The social cost of carbon turns mainly on the discount rate.

By AI (Claude Opus 5.5, thinking level unavailable) Draft:  October 2026

"The initially positive impacts have vanished: The central estimate of the economic impact of global warming is always negative." — Richard Tol, the economist whose earlier surveys founded the claim that warming does net good, 20241

Summary

Economists agree on the direction and disagree on the size.  On every central estimate, each further degree of warming costs more than the one before, and by 3 °C every central estimate is a net loss.  How large that loss is depends on the method, by a factor of ten.  The early finding that warming had been a net economic gain has not survived its own author's later work.

In 2013 the science writer Matt Ridley wrote:  "Climate change has done more good than harm so far and is likely to continue doing so for most of this century. This is not some barmy, right-wing fantasy; it is the consensus of expert opinion."2  His source was the economist Richard Tol, whose model put the world's gain from the warming of the 20th century at "1.4% of GDP by 2000".3

  1. The "net benefit" curve has vanished. Tol's 2009 survey found initial benefits peaking at about 1.1 °C of further warming.4 Two of its estimates had lost their minus signs,5 and by 2018 his total turned negative just below 1.7 °C above pre-industrial.6 In 2024 he wrote: "The initially positive impacts have vanished".1
  2. Two families of estimates differ tenfold. Models that add up damages sector by sector put the loss at 3 °C near 2 % of world income.7 Studies of how output has moved with temperature put it at 20 % or more.89 The most prominent of these, a 2024 Nature paper claiming a 19 % loss already locked in, was retracted by its authors.10
  3. The warming so far has cost little, and its sign is uncertain. On William Nordhaus's damage curve, 1.3 °C costs about 0.4 % of world income.11 Tol's 2024 central estimate is below zero, inside an interval that includes zero.1
  4. The price of a tonne of CO2 turns on the discount rate. The same model on the same damages gives $19.70 a tonne at a 5 % discount rate and $128.50 at 2.5 %;7 a newer model gives $185 at 2 %.12 Choosing the rate is a moral judgment, not a measurement.
  5. The strongest economic case for acting is the tail. Martin Weitzman argues that the chance of an extreme outcome can dominate the expected loss, and Robert Pindyck that the case for strong action rests on that tail, which no one can measure.1314 Nordhaus answers that this says the risk is serious, not how much to spend on it.15

The earth has warmed about 1.3 °C since the late 1800s, as Is the Earth Warming? shows.  How fast more warming arrives depends on climate sensitivity, weighed in How Much Warming Does More CO2 Cause?, and on whether the models run hot, weighed in How Well Have Climate Predictions Done?.  Most estimates here are stated per degree of warming, so a lower sensitivity delays the date at which 2 or 3 °C arrives but does not change what 2 or 3 °C costs.

The measured effects that these estimates add up, from crops to sea level, are weighed one by one in Does Warming Do More Good or Harm?.  Deaths from heat and cold are weighed in Do Cold Deaths Outweigh Heat Deaths?.

Part 1: What the economic estimates say

Economists try to add every effect into one number:  the loss or gain in world income, as a percent of GDP, at each level of warming.  These estimates are the only place where crops, deaths, energy bills, floods and ecosystems are put on one scale.  They are also models, not measurements, and their authors say so.

The curve behind the "net benefit" claim, and what happened to it

The claim that warming has been good on balance comes from one economist, Richard Tol, and it has changed as he has updated it.

  1. 2009. Tol surveyed 14 published estimates and found that some "point to initial benefits of a modest increase in temperature, followed by losses as temperatures increase further." His fitted curve peaked "at about 1.1°C warming (with a standard deviation of 0.7°C)",4 measured from 2009, not from the 1800s. It crossed zero at about 2.2 °C above 2009, or roughly 3 °C above pre-industrial, which is the figure Ridley quoted.16
  2. 2014. Tol published a correction: "minus signs were dropped from the two impact estimates, one by Plambeck and Hope (1996) and one by Hope (2006)."5 Two of the positive estimates had been negative all along. William Nordhaus and Andrew Moffat then tried to replicate the corrected set and reported that "Of the 21 impact estimates in the 2014 study, 10 were incorrect or could not be replicated."17
  3. 2018. With 27 estimates, Tol found that "the total impacts turn negative just below 1.7°C warming above preindustrial levels", and that "the incremental impacts turn negative before that", around 1.1 °C.6 The incremental impact is what each extra tenth of a degree does. On this curve, by the time the world reached 1.3 °C, the total was still slightly positive but every further step was already a loss. He added that "the impact of climate change does not significantly deviate from zero until 3.5°C warming."6
  4. 2024. Tol's latest meta-analysis says: "The initially positive impacts have vanished: The central estimate of the economic impact of global warming is always negative. The effects of carbon dioxide fertilization and reduced winter cold on heating costs and human health are still positive, but newer studies reckon these are relatively smaller than earlier studies."1

So the economist on whom the "net benefit" claim rests no longer finds a net benefit in his central estimate.  He still finds the benefits real, he still finds the total small, and he still finds the uncertainty wide enough to include no harm at all.  Ridley wrote in 2013, before the correction.

Tol's 20th-century model, the source of Ridley's "1.4 per cent", is worth reading closely.  Its farming gain "is entirely due to carbon dioxide fertilization", while "The impact of climate change (temperature, precipitation, cloud cover, wind, etc) is actually negative".  Tol notes that this "is in contrast to the empirical literature" on crop yields.3  The benefit in that model is a CO2 benefit, not a warming benefit, and it depends on how large the fertilisation effect is assumed to be.  Does Warming Do More Good or Harm? weighs that effect against the field data.

How large are the losses at 2 °C and 3 °C?

The estimates fall into two families that differ by a factor of ten.

Study Method Estimate
Nordhaus 2017 (DICE model) Damage function fitted to surveyed studies 2.1 % of income lost at 3 °C, 8.5 % at 6 °C7
Nordhaus and Moffat 2017 Regression on 36 estimates, plus 25 % for omitted damages 2.04 % (± 2.21) at 3 °C17
Tol 2018 Average of 11 estimates 1.3 % at 2.5 °C; 3 of the 11 estimates positive6
Tol 2024 Average of 13 estimates 1.7 % at 2.5 °C1
Howard and Sterner 2017 Meta-analysis correcting for duplicates 7 to 8 % at 3 °C; 9 to 10 % with catastrophic risk18
Burke, Hsiang and Miguel 2015 Country incomes against temperature, 1960–2010 23 % of 2100 output under the highest emissions path8
Bilal and Känzig 2024 World output against global temperature swings "over 20%" per 1 °C, permanent9
Kotz, Levermann and Wenz 2024 Regional incomes, 1,600 regions 19 % by 2049; retracted 202519

The first family adds up damages sector by sector, or surveys studies that do.  It finds that 2 to 3 °C costs about 1 to 3 % of world income.  Even the pessimistic meta-analysis of Peter Howard and Thomas Sterner stays under 10 % at 3 °C.  The second family infers damage from how national or world output has moved with temperature in the past, and it finds losses ten times larger.

Which family is closer to the truth is the largest open question in the economics, and the record of the second family is uneven.  Burke's own paper gives its high-emissions loss a "probability 0.29" of being positive instead.8  The Kotz paper was withdrawn by its authors after Tom Bearpark, Dylan Hogan and Solomon Hsiang showed that "data anomalies arising from one country in KLW's underlying GDP dataset, Uzbekistan, substantially bias their predicted impacts".20  The authors' retraction note says the corrected work widened the uncertainty "from 11–29% to 6–31%", and that "these changes are too substantial for a correction".10  Tol's 2024 review finds that studies relating growth to the level of temperature "cannot agree on the sign of the impact".1  That a high-profile Nature result was retracted is evidence that the field corrects itself.  It is also a reason to hold the largest numbers loosely.

What the measured 1.3 °C has cost so far

The models can be read at today's warming.  On Nordhaus's quadratic damage function, 1.3 °C costs about 0.4 % of world income, rising to about 0.9 % at 2 °C and 2.1 % at 3 °C.11  On Tol's 2018 curve, 1.3 °C is still slightly above zero in total.  On his 2024 meta-analysis the central estimate is below zero, inside an interval that includes zero.1  Bilal and Känzig go furthest:  by their estimate "world GDP per capita would be more than 20% higher today had no warming occurred between 1960 and 2019".9  That cannot be tested directly, because there is no unwarmed world to compare with.

Whichever family is used, the honest summary is the same.  The economic effect of the warming so far is small compared with the growth of the world economy over the same years.  Its sign is not known with confidence.  What the estimates do agree on is the direction of travel:  every additional degree costs more than the one before.  The IPCC's own summary says aggregate damages "generally increase non-linearly with global warming levels", and that "the lack of comparability between methodologies, does not allow for identification of a robust range of estimates".21

Part 2: The price of a tonne of CO2, and the tail

Why the social cost of carbon swings so widely

The social cost of carbon is the estimated damage, in today's dollars, done by one more tonne of CO2 over all the years it stays in the air.  Most of that damage comes decades or centuries from now, so its value today depends almost entirely on the discount rate, the rate at which future costs are shrunk to present value.

Nordhaus's 2017 model gives $31 a tonne in his central case.7  The same model on the same damages gives $19.70 at a 5 % discount rate, $36.30 at 4 %, $79.10 at 3 %, and $128.50 at 2.5 %.7  Kevin Rennert and colleagues, with a newer damage model, give "$185 per tonne of CO2 ($44–$413 per tCO2: 5%–95% range, 2020 US dollars) at a near-term risk-free discount rate of 2%", against $80 at 3 %.12

The arithmetic shows why.  A dollar of damage 100 years from now is worth 13.8 cents today at a 2 % discount rate, 5.2 cents at 3 %, and 0.76 cents at 5 %.22  Choosing between those rates is not a measurement.  It is a judgment about how much people today should sacrifice for people not yet born, and whether future generations, who will probably be richer, should be treated as equals.  Does Warming Do More Good or Harm? returns to it.

Tail risk: the argument from what the models leave out

The strongest economic case for acting is not the central estimate.  It is the chance of something much worse.  Martin Weitzman of Harvard argued in 2009 that because the key physical numbers are themselves uncertain, the odds of extreme outcomes are higher than the usual bell curves admit:  "uncertain structural parameters induce a critical “tail fattening” of posterior-predictive distributions."13  His "dismal theorem" says the expected loss can be dominated by those tails.  Weitzman himself drew a modest conclusion:  the theorem "suggests as a policy response to climate change a relatively more cautious approach to GHG emissions, but how much more caution is warranted? I simply do not know".13

Robert Pindyck of MIT goes further against the models.  He calls their damage functions "completely made up, with no theoretical or empirical foundation", and adds that this "might not matter much if we are looking at temperature increases of 2 or 3◦ C, because there is a rough consensus (perhaps completely wrong) that damages will be small at those levels of warming."14  On his view, the case for strong action rests on the tail, and the tail is the part no one can measure.  Tol's 2024 review agrees that "The uncertainty about the impact is skewed towards negative surprises."1

The answer from the other side is that a tail argument proves too much.  Nordhaus replied that "the usual version of the Dismal Theorem actually contains no policy decisions":15  it says the risk is serious, not how much to spend on it.  Any policy also carries its own tail risks, such as energy shortages and slower growth for the poor.  The physical tails are real:  the IPCC cannot rule out sea-level rise "approaching 2 m by 2100" under very high emissions, at low confidence.23  How much to pay to avoid them is a judgment about risk, the subject of Part 4 of Does Warming Do More Good or Harm?.

Objections and responses

From the side that says warming helps

  1. "Richard Tol showed that warming has been good for the world." Tol's 2009 survey and 2011 model did find a net gain at low warming, and Ridley reported them accurately. Tol himself has since corrected the 2009 table, found in 2018 that the incremental effect turns negative near 1.1 °C, and concluded in 2024 that "The initially positive impacts have vanished".1 The benefit in his 2011 model came "entirely" from CO2 fertilisation, with the climate effect on farming negative.3 Citing the 2009 curve today means citing a result its author has withdrawn.
  2. "The damage models are made up." Robert Pindyck, who favours action, agrees.14 But the criticism cuts both ways. If the models cannot be trusted to show large damages, they cannot be trusted to show small ones, and the survey-based estimates that the benefit side cites come from the same family of models.

From the side that says warming harms

  1. "Kotz and colleagues showed a 19 % income loss is already locked in." That paper was retracted by its own authors.10
  2. "Growth-panel studies show losses of 20 % or more, so the survey estimates are far too low." They might be. But they also imply that the world would already be "more than 20%" richer without the warming since 1960,9 a claim that cannot be checked, and Tol finds that studies relating growth to temperature levels "cannot agree on the sign of the impact".1 Until the two families converge, the honest statement is that the cost at 3 °C lies somewhere between about 2 and 20 % of income, most likely toward the lower end on the evidence now in hand.
  3. "A near-zero discount rate is the only ethical choice." Some economists and ethicists hold this, and it multiplies the cost of carbon several times over.7 Others answer that it asks today's poor to sacrifice for richer descendants. That is a value judgment, discussed in Part 4 of Does Warming Do More Good or Harm?, and the data cannot settle it.

Issues and unanswered questions

  1. Which family of economic estimates is right? Sector-by-sector models put 3 °C at 1 to 3 % of world income; growth-panel studies put it near 20 % or more. Neither family has been tested out of sample over a long enough record to settle the question. The retraction of Kotz and colleagues shows the growth-panel approach is fragile; Pindyck's critique shows the sector models are partly assumption.
  2. Tol's published curves are given only as figures. His 2018 and 2024 papers print the turning points and averages but not the fitted values at 1, 2, 3 and 4 °C, so the numbers between those points in this article come from Nordhaus's formula, not Tol's.
  3. Not yet peer reviewed. Bilal and Känzig's estimate is a working paper.

Verdict

Economists no longer support the claim that the warming so far has been a net economic gain, and they cannot yet say how large the loss from more warming will be.

  1. The net benefit has gone. Tol's 2024 central estimate is negative at every level of warming, though at today's warming its interval still includes zero.1 He still counts CO2 fertilisation and milder winters as benefits, only smaller than earlier studies found.
  2. The cost so far is small and of uncertain sign. It is small compared with the growth of the world economy over the same years.
  3. Each further degree costs more than the last. At 3 °C the central estimates run from about 2 % of world income on the sector models to 20 % or more on the growth-panel studies. On the evidence now in hand the cost is most likely toward the lower end, but the higher family has not been ruled out.
  4. The social cost of carbon is set mostly by the discount rate, which is a judgment about how to weigh people not yet born, not a measurement.

Of the published total-impact estimates, Tol writes that researchers "agree on the order of magnitude":  "a century of climate change is not worse than losing a decade of economic growth."1  Whether that cost justifies a given policy is a question of discount rates, risk and justice to the poor, which Does Warming Do More Good or Harm? takes up and the data alone cannot settle.

Sources

  1. 1.

    Tol, Richard S. J.  “A meta-analysis of the total economic impact of climate change.”  Energy Policy 185:113922.  2024.  Tol's latest meta-analysis of published estimates of the total economic effect of warming, with a separate review of studies that relate economic growth to weather.  It overturns his own earlier finding of an initial benefit.

    1. 1a.

      Page 2:  “global warming of 2.5 ◦ C would make the average person feel as if she had lost 1.7% of her income. 1.7% is the average of the 13 dots at”. ↩

    2. 1b.

      Page 11:  “(1) The initially positive impacts have vanished: The central estimate of the economic impact of global warming is always negative. The effects of carbon dioxide fertilization and reduced winter cold on heating costs and human health are still positive, but newer studies reckon these are relatively smaller than earlier studies. (2) The confidence interval about the estimates has widened considerably, and now includes no impact for very considerable warming.” ↩ ↩ ↩ ↩ ↩ ↩ ↩

    3. 1c.

      Page 1:  “studies, which relate economic growth to temperature levels, cannot agree on the sign of the impact”. ↩ ↩

    4. 1d.

      Page 5:  “Researchers disagree on the sign of the net impact, but agree on the order of magnitude: The welfare loss (or gain) caused by climate change is equivalent to the welfare loss caused by an income drop of at most ten percent—a century of climate change is not worse than losing a decade of economic growth.” ↩

    5. 1e.

      Page 1:  “(4) The uncertainty about the impact is skewed towards negative surprises.” ↩

  2. 2.

    Ridley, Matt.  “Why climate change is good for the world.”  The Spectator.  Oct 19, 2013.  A magazine essay by a science writer and member of the House of Lords, the best-known statement of the claim that warming has done net good.  Its economic numbers come from Richard Tol's 2009 survey and Tol's 2011 model of the 20th and 21st centuries.

    1. 2a.

      Paragraph 1:  “Climate change has done more good than harm so far and is likely to continue doing so for most of this century. This is not some barmy, right-wing fantasy; it is the consensus of expert opinion.” ↩

    2. 2b.

      Paragraph 4:  “To be precise, Prof Tol calculated that climate change would be beneficial up to 2.2˚C of warming from 2009 (when he wrote his paper).” ↩

  3. 3.

    Tol, Richard S. J.  “The Economic Impact of Climate Change in the 20th and 21st Centuries.”  Copenhagen Consensus on Human Challenges, Assessment Paper.  Copenhagen Consensus Center.  2011.  A run of Tol's FUND model, country by country, backward over the observed 20th-century climate and forward over a 21st-century scenario, summing farming, energy, health, water, storms, sea level and ecosystems as a share of income.  Published in revised form in Climatic Change 117 (2013).  It is the source of the "net benefit until 2080" claim, and it is a model result, not a measurement.

    1. 3a.

      Page 7:  “After 1950, impacts became more positive, edging up to 1.4% of GDP by 2000. However, impacts roughly stabilize after that, reaching their maximum at 1.5% of GDP in 2025 and then precipitously fall to reach - 1.2% of GDP in 2100.”  The "-" before the line break in the original is a minus sign. ↩

    2. 3b.

      Page 10:  “Agriculture is the biggest positive impact in the 20th century, approaching 0.8% of GDP by 2000. This is entirely due to carbon dioxide fertilization, which makes crops grow faster and more water efficient. The impact of climate change (temperature, precipitation, cloud cover, wind, etc) is actually negative, reaching -0.3% in 2000.”  “This is in contrast to the empirical literature reviewed in the introduction”. ↩ ↩

  4. 4.

    Tol, Richard S. J.  “The Economic Effects of Climate Change.”  Journal of Economic Perspectives 23(2):29–51.  2009.  A survey of the 14 published estimates of the total welfare effect of global warming then available, with a curve fitted through them.  Its early-benefit hump was the basis of the claim that moderate warming does net good;  two of its signs were later corrected.

    1. 4a.

      Page 34:  “point to initial benefits of a modest increase in temperature, followed by losses as temperatures increase further.”  “The fitted line in Figure 1 suggests that the turning point in terms of economic benefits occurs at about 1.1°C warming (with a standard deviation of 0.7°C).” ↩ ↩

    2. 4b.

      Page 35:  “The central heavier line is the least squares fit to the 14 observations: D ⫽ 2.46(1.25)T ⫺ 1.11(0.48)T 2 , R2 ⫽ 0.51, where D denotes impact and T denotes temperature”.  “as a function of the increase in global mean temperature relative to today”.  The typesetting prints = and − as ⫽ and ⫺, and T² as "T 2". ↩

  5. 5.

    Tol, Richard S. J.  “Correction and Update:  The Economic Effects of Climate Change.”  Journal of Economic Perspectives 28(2):221–226.  2014.  Tol's correction of his 2009 survey, which restores dropped minus signs and adds seven estimates for 21 in all.

    1. 5a.

      Page 221:  “In Table 1 of that paper, titled "Estimates of the Welfare Impact of Climate Change," minus signs were dropped from the two impact estimates, one by Plambeck and Hope (1996) and one by Hope (2006).” ↩ ↩

  6. 6.

    Tol, Richard S. J.  “The Economic Impacts of Climate Change.”  Review of Environmental Economics and Policy 12(1):4–25.  2018.  Tol's updated survey of 27 estimates from 22 studies of the total economic effect of warming, with a fitted curve and an analysis of uncertainty and regional spread.

    1. 6a.

      Page 5:  “a global mean temperature increase of 2.5°C would make the average person feel as if she had lost 1.3 percent of her income (1.3 percent is the average of the 11 impact estimates for warming of 2.5°C).”  “researchers disagree on the sign of the net impact: 3 estimates are positive and 8 are”. ↩

    2. 6b.

      Page 7:  “As shown in figure 1, the total impacts turn negative just below 1.7°C warming above preindustrial levels. More importantly, the incremental impacts turn negative before that”. ↩ ↩

    3. 6c.

      Page 8:  “If we take the confidence interval at face value, the impact of climate change does not significantly deviate from zero until 3.5°C warming.”  The text layer of the PDF drops the degree signs, which print on the page. ↩

  7. 7.

    Nordhaus, William D.  “Revisiting the social cost of carbon.”  Proceedings of the National Academy of Sciences 114(7):1518–1523.  2017.  The 2016 revision of Nordhaus's DICE model, the most widely used model linking the economy, emissions, warming and damage.  It gives a social cost of carbon and shows how that cost changes with the discount rate and other assumptions.

    1. 7a.

      Page 1519:  “Eq. 3 assumes that damages can be reasonably well approximated by a quadratic function of temperature change.”  “Including all factors, the final estimate is that the damages are 2.1% of global income at a 3 °C warming, and 8.5% of income at a 6 °C warming.” ↩ ↩ ↩

    2. 7b.

      Page 1518:  “The study estimates that the SCC is $31 per ton of CO2 in 2010 US$ for the current period (2015).” ↩

    3. 7c.

      Page 1520, table 1:  “Alternative discount rates”.  “2.5% 128.5”.  “3% 79.1”.  “4% 36.3”.  “5% 19.7”.  Social cost of carbon for 2015, in 2010 international dollars per tonne of CO2, calculated along the current-policy path. ↩ ↩

    4. 7d.

      Page 1520, table 1:  “The Stern Review discounting”.  “Uncalibrated† 197.4”.  Social cost of carbon for 2015, in 2010 international dollars per tonne of CO2, against the baseline of 31.2. ↩

  8. 8.

    Burke, Marshall et al.  “Global non-linear effect of temperature on economic production.”  Nature 527:235–239.  2015.  A study of 166 countries' annual output against their annual temperature, 1960–2010, finding that productivity peaks near a country-year average of 13 °C and falls at higher temperatures, then projected to 2100.  The 13 °C is a national annual mean, not a global one.

    1. 8a.

      Section Results:  “climate change reduces projected global output by 23% in 2100 (best estimate, SSP5) relative to a world without climate change, although statistical uncertainty allows for positive impacts with probability 0.29”. ↩ ↩ ↩

  9. 9.

    Bilal, Adrien, and Diego R. Känzig.  “The Macroeconomic Impact of Climate Change:  Global vs. Local Temperature.”  NBER Working Paper 32450.  National Bureau of Economic Research.  2024.  A working paper that estimates how world output responds to year-to-year swings in global, rather than national, temperature, and builds a model of the cost of warming from it.  Its damages are far larger than those of the sector-by-sector models.

    1. 9a.

      Section Abstract:  “Exploiting natural global temperature variability, we find that 1°C warming reduces world GDP by over 20% in the long run.” ↩ ↩

    2. 9b.

      Page 6:  “According to our counterfactual, world GDP per capita would be more than 20% higher today had no warming occurred between 1960 and 2019.” ↩ ↩

  10. 10.

    Kotz, Maximilian et al.  “Retraction Note:  The economic commitment of climate change.”  Nature.  Dec 3, 2025.  The authors' notice withdrawing their 2024 paper, giving the reasons and the effect of the corrections.

    1. 10a.

      Section Retraction Note:  “post-publication, the results were found to be sensitive to the removal of one country, Uzbekistan, where inaccuracies were noted in the underlying economic data for the period 1995–1999.”  “These changes led to discrepancies in the estimates for climate damages by mid-century, with an increased uncertainty range (from 11–29% to 6–31%)”.  “The authors acknowledge that these changes are too substantial for a correction, leading to the retraction of the paper.” ↩ ↩ ↩

  11. 11.

    Nordhaus's damages are 2.1 % at 3 °C and 8.5 % at 6 °C on a quadratic curve 7, so D ≈ 0.236 × T² percent of income.  At 1.3 °C that gives 0.40 %;  at 2 °C, 0.94 %;  at 2.5 °C, 1.5 %;  at 3 °C, 2.1 %;  at 4 °C, 3.8 %.  The curve has no early benefit. ↩ ↩

  12. 12.

    Rennert, Kevin et al.  “Comprehensive evidence implies a higher social cost of CO2.”  Nature 610:687–692.  2022.  A social cost of carbon built from newer sector-by-sector damage studies, probabilistic emissions projections and a discount rate tied to market interest rates.  It underlies the U.S. EPA's 2023 estimate.

    1. 12a.

      Page 687:  “Our preferred mean SC-CO2 estimate is $185 per tonne of CO2 ($44–$413 per tCO2: 5%–95% range, 2020 US dollars) at a near-term risk-free discount rate of 2%”.  Section Results:  “3% ($80 per tCO2 mean, $12–$197 per tCO2: 5%–95% range)”. ↩ ↩

  13. 13.

    Weitzman, Martin L.  “On Modeling and Interpreting the Economics of Catastrophic Climate Change.”  Review of Economics and Statistics 91(1):1–19.  2009.  A theoretical paper by a Harvard economist arguing that uncertainty about the climate system's key parameters makes extreme outcomes much more likely than standard models assume, and that this can dominate cost-benefit analysis.  It gives no damage estimate at any particular warming.

    1. 13a.

      Page 1:  “Even when updated by Bayesian learning, uncertain structural parameters induce a critical “tail fattening” of posterior-predictive distributions.” ↩ ↩

    2. 13b.

      Page 13:  “Other things being equal, the dismal theorem suggests as a policy response to climate change a relatively more cautious approach to GHG emissions, but how much more caution is warranted? I simply do not know”. ↩

  14. 14.

    Pindyck, Robert S.  “Climate Change Policy:  What Do the Models Tell Us?”  NBER Working Paper 19244.  National Bureau of Economic Research.  2013.  Published in the Journal of Economic Literature 51(3):860–872 (2013);  page numbers here are those of the working paper.  A critique of integrated assessment models by an MIT economist who favours action on climate, arguing that their damage functions and discount rates are arbitrary and that the case for policy rests on catastrophic outcomes the models cannot capture.

    1. 14a.

      Page 13:  “The bottom line here is that the damage functions used in most IAMs are completely made up, with no theoretical or empirical foundation. That might not matter much if we are looking at temperature increases of 2 or 3◦ C, because there is a rough consensus (perhaps completely wrong) that damages will be small at those levels of warming.” ↩ ↩ ↩

  15. 15.

    Nordhaus, William D.  Economic Policy in the Face of Severe Tail Events.  2010.  Nordhaus's reply to Weitzman's dismal theorem, a draft of 17 May 2010, published in revised form as "The Economics of Tail Events with an Application to Climate Change", Review of Environmental Economics and Policy 5(2), 2011.  It examines what assumptions the theorem needs and whether they hold.

    1. 15a.

      Page 1:  “note that the usual version of the Dismal Theorem actually contains no policy decisions.” ↩ ↩

  16. 16.

    Tol's 2009 fit, D = 2.46T − 1.11T² with T measured from 2009 4, peaks at T = 2.46 ÷ 2.22 = 1.1 °C, where it gives a gain of about 1.4 % of income, and returns to zero at T = 2.46 ÷ 1.11 = 2.2 °C.  Adding the roughly 0.8 °C of warming before 2009 that Ridley assumes gives about 3 °C above pre-industrial 2. ↩

  17. 17.

    Nordhaus, William D., and Andrew Moffat.  “A Survey of Global Impacts of Climate Change:  Replication, Survey Methods, and a Statistical Analysis.”  NBER Working Paper 23646.  National Bureau of Economic Research.  2017.  A replication of Tol's damage surveys and a new systematic search that yields 36 usable estimates from 27 studies, with a regression of damage on warming.  Nordhaus, who built the DICE model, won the 2018 Nobel prize in economics for this line of work.

    1. 17a.

      Page 10:  “Of the 21 impact estimates in the 2014 study, 10 were incorrect or could not be replicated.” ↩

    2. 17b.

      Section Abstract:  “With this addition, the estimated impact is -2.04 (± 2.21) % of income at 3 °C warming and -8.06 (± 2.43) % of income at 6 °C warming.”  "This addition" is a 25 % judgmental allowance for damages the studies leave out.  The paper's summary gives −8.16 % for 6 °C. ↩

  18. 18.

    Howard, Peter H., and Thomas Sterner.  “Few and Not So Far Between:  A Meta-analysis of Climate Damage Estimates.”  Environmental and Resource Economics 68(1):197–225.  2017.  A meta-analysis of published global damage estimates that removes duplicated estimates and controls for method.  It finds damages several times higher than the DICE model and is the most pessimistic of the surveys of this kind.

    1. 18a.

      Section Conclusions:  “non-catastrophic damages are likely between 7 and 8% of GDP for a 3 ◦ C increase (from the pre-industrial period), compared to the 2.4% predicted by DICE-2013R (Nordhaus 2013), and are between 9 and 10% when factoring in catastrophic risks.” ↩

  19. 19.

    Kotz, Maximilian et al.  “The economic commitment of climate change.”  Nature 628:551–557.  2024.  A study of 40 years of income data from about 1,600 subnational regions that claimed a large, unavoidable income loss by mid-century.  Its authors retracted it in December 2025.

    1. 19a.

      Page 551:  “we find that the world economy is committed to an income reduction of 19% within the next 26 years independent of future emission choices”. ↩

  20. 20.

    Bearpark, Tom et al.  “Data anomalies and the economic commitment of climate change.”  Nature.  Aug 6, 2025.  A Matters Arising reply to Kotz and colleagues by economists from Solomon Hsiang's group, the same group that produced the 2015 growth-panel estimate, showing that one country's data drove the result.

    1. 20a.

      Section Abstract:  “data anomalies arising from one country in KLW's underlying GDP dataset, Uzbekistan, substantially bias their predicted impacts of climate change”. ↩

  21. 21.

    IPCC.  “Summary for Policymakers.”  Climate Change 2022:  Impacts, Adaptation and Vulnerability.  Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change, 3–33.  Cambridge University Press.  2022.  The government-approved summary of the IPCC's 2022 assessment of climate impacts.  It is the mainstream statement of the harm case, and it also records the benefits the assessment found.

    1. 21a.

      Section B.4.6, page 15:  “Projected estimates of global aggregate net economic damages generally increase non-linearly with global warming levels (high confidence).”  “The wide range of global estimates, and the lack of comparability between methodologies, does not allow for identification of a robust range of estimates (high confidence).” ↩

  22. 22.

    The present value of $1 paid in 100 years is 1 ÷ (1 + r)^100.  At r = 2 % that is $0.138;  at 3 %, $0.052;  at 5 %, $0.0076. ↩

  23. 23.

    IPCC.  “Summary for Policymakers.”  Climate Change 2021:  The Physical Science Basis.  Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change, 3–32.  Cambridge University Press.  2021.  The summary of the IPCC's 2021 assessment of the physical science, line by line approved by member governments.  It is the mainstream statement of what has been observed and what is projected.

    1. 23a.

      Section B.5.3, page 21:  “Relative to 1995–2014, the likely global mean sea level rise by 2100 is 0.28–0.55 m under the very low GHG emissions scenario (SSP1-1.9); 0.32–0.62 m under the low GHG emissions scenario (SSP1-2.6); 0.44–0.76 m under the intermediate GHG emissions scenario (SSP2-4.5); and 0.63–1.01 m under the very high GHG emissions scenario (SSP5-8.5)”.  “Global mean sea level rise above the likely range – approaching 2 m by 2100 and 5 m by 2150 under a very high GHG emissions scenario (SSP5-8.5) (low confidence) – cannot be ruled out due to deep uncertainty in ice-sheet processes.” ↩