On 29th July 2021, Nature published a paper titled "The Mortality Cost of Carbon" and many publications have repeated a rather sensational part of the abstract of the paper (emphasis added):

Our central estimate 2020 MCC [mortality cost of carbon] also implies that reducing (adding) 4,434 metric tons of carbon dioxide in 2020 saves one life (causes one excess death) in expectation globally between 2020 and 2100. In all, 4,434 metric tons is equivalent to the lifetime emissions of 3.5 average Americans, 146.2 Nigerians, and 12.8 average world people.

This has been repeated by The Independent, Yahoo and many others. Here's the headline from the Guardian, 29th July 2021:

Three Americans create enough carbon emissions to kill one person, study finds

I am skeptical as there is no explanation of how the calculations of emissions and the calculations of estimated excess deaths can possibly be matched together and conclusions be drawn in such a simplistic way.

COVID-19 figures have demonstrated that only a few countries have the administrative infrastructure to reliably calculate the correct statistic for excess deaths and these very countries are, inherently, not the ones with the highest populations.

But the main feature of this published statement is that no explanation is offered as to the decease of the supposed individual:

  • Were they never born, because emissions caused a terminated pregnancy?
  • Were they never born because their mother never reached reproductive age, due to emissions?
  • Did they reach middle age?
  • Old age?
  • Was their life merely cut short by a few months?

And what of the "three and a half" Americans? Did they perhaps have shortened lives due to emissions? Did they lose a few years? Or many years?

To publish such a statistic from the mere matching of other (only vaguely related) statistics is highly debatable and leaves me very skeptical indeed.


@IMSoP comments and quotes the following definition from the paper:

Excess deaths are deaths attributable to climate change that occur prematurely relative to a counterfactual scenario in which the marginal emission did not occur.

I appreciate the definition and am grateful for it being quoted. But still it leaves questions as to how the statistics are calibrated to one another and as to what exactly 'premature' means.

I have worked night shifts for twenty-five years and I know that I shall live a shorter life because of it. But it is a risk I accept. I may live a few years less.

The 'death' caused by 'three and a half' Americans needs to be quantified as to when, exactly, this death will occur, (just how premature is 'premature') I would strongly suggest.

  • 1
    Did the paper explain how it calculated the carbon emissions of an "average American"? Simply dividing the country's carbon emissions by its population would paint a misleading picture IMO, as much of that pollution is caused by corporate activity rather than individuals.
    – F1Krazy
    Jul 31, 2021 at 12:03
  • 4
    For this to be true you'd have to expect multiple-billion climate-related deaths in the next 80 years (since Americans are not the only people producing carbon) . The only way this is possible is if you assign about 25% of all deaths to climate change, which requires a very generous definition of climate-related deaths.
    – antlersoft
    Jul 31, 2021 at 14:21
  • 2
    Excess mortality is a recognised technical concept, there's no need to bring in wild interpretations like "people who were never born". The paper spells it out in the introduction: "Excess deaths are deaths attributable to climate change that occur prematurely relative to a counterfactual scenario in which the marginal emission did not occur." There's also not a "supposed individual" involved; this is all about statistics and probabilities, not "the Smith family killed Joe Blogga"
    – IMSoP
    Aug 1, 2021 at 9:20
  • @IMSoP Your comment much appreciated and an edit made appropriately. Thank you.
    – Nigel J
    Aug 1, 2021 at 9:38
  • @Schmuddi Yes, agreed, in retrospect. Edited.
    – Nigel J
    Aug 3, 2021 at 19:09

1 Answer 1


As so often, the claim has not been translated well into headlines, and makes a compelling soundbite without a very deep meaning.

The actual central claim is this:

adding 4,434 metric tons of carbon dioxide in 2020 causes one excess death in expectation globally between 2020 and 2100

This is compared against various things, among them the lifetime emissions of 3.5 Americans. An accompanying graph makes this comparison clearer, labelling it "Average citizen's lifetime emissions if all added in 2020".

So the claim is not that every year 10 people die for every 35 Americans; it is that if 35 Americans blinked out of existence, the reduction in carbon dioxide would save 10 people from climate-related deaths over the course of the next century.

Indeed, even that is probably overstating it, because the model used includes feedback factors, so a tonne of carbon dioxide emitted in 2020 is counted differently from one emitted in 2099 in terms of its impact in 2100. So the 10 deaths would be spread further into the future, as some would be caused by emissions later than 2020.

  • I think you have hit the nail on the head. If none better arrives, I shall accept this as the answer. For now, up-voted +1.
    – Nigel J
    Aug 1, 2021 at 9:39
  • 3
    I've somewhat talked myself into skepticism here 😆 The choice of "lifetime per citizen" as the comparison seems to have no justification other than dramatic effect, and somewhat undermines the rest of the paper, which talks about measuring the effect of policy decisions.
    – IMSoP
    Aug 1, 2021 at 10:09
  • I’m confused about what’s going on here. The original question is 90% asking about the actually modeling performed in the study. But this answer doesn’t talk about any of this, and instead just repeats (in different words) the exact claim in the first quote of the question.
    – FifthArrow
    Aug 2, 2021 at 2:02
  • 1
    @KobeGote I interpreted the question as about how the two statistics can be "matched together", rather than the individual statistics themselves. It is a slightly vague question, though.
    – IMSoP
    Aug 2, 2021 at 7:52
  • 1
    @fredsbend The source is the same paper linked to in the question (as distinct from the news summaries also linked); I'm not sure what other sources would be relevant. As I said in a previous comment, I agree the question is somewhat open to interpretation; apparently the person who asked it thought this answered what they wanted to know, but I wouldn't be in any way offended if someone posted an answer from a different angle. For that matter, I wouldn't object to the question itself being closed on that basis.
    – IMSoP
    Aug 4, 2021 at 16:03

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .