Blog Archive

Friday, August 15, 2025

A Paper Published in Nature That Predicted Huge Drops in GDP Due to Climate Change Shown to Be Inaccurate When the Data Was Examined: The Paper was Widely Cited and Widely Adopted in Climate Risk Modeling

      An immensely influential paper published in Nature that predicted huge drops in GDP due to climate change was recently found to have a single data error that basically invalidates the paper’s conclusions. The error is not thought to be intentional, but an accidental oversight. The paper is now flagged with the following editorial note:

06 November 2024 Editor’s Note: Readers are alerted that the reliability of data and methodology presented in this manuscript is currently in question. Appropriate editorial action will be taken once this matter is resolved.”

     The paper claimed that the world was on track to lose 19% of GDP by 2050 due to impacts from climate change. It also claimed that by 2100, we would lose 62% of GDP under a high emissions scenario. Those predictions were triple the conclusions of other analyses and papers. Perhaps that itself should have been a red flag that the data should be examined more closely. One might ask why no other researchers had predicted anything close to this. Instead, the paper was the second most cited paper by the media in 2024, according to Carbon Brief. Shannon Osaka of the Washington Post notes that the paper’s analysis and dataset have been used for financial planning by the U.S. government, the World Bank, and other institutions.

     Apparently, the erroneous data for one country, Uzbekistan, tainted the entire analysis. On August 6, 2025, a commentary paper describing the data error was published in Nature with the following observations about the paper, known as KLW, referencing its authors:

“(1) data anomalies arising from one country in KLW’s underlying GDP dataset, Uzbekistan, substantially bias their predicted impacts of climate change, (2) KLW underestimate statistical uncertainty in their future projections of climate impacts, and (3) additional data-quality concerns in KLW’s subnational GDP data warrant further investigation. When Uzbekistan’s data are removed and statistical uncertainty is corrected to account for spatial correlations, KLW’s central estimate aligns closely with previous literature and their results are no longer statistically distinguishable from mitigation costs at any time this century.”

     Osaka notes:

With Uzbekistan removed from the dataset, the predictions dropped substantially — from 62 percent GDP loss in 2100 to 23 percent and from 19 percent by 2050 to 6 percent, said Solomon Hsiang, director of the global policy laboratory at Stanford University and one of the authors of the commentary.”

     This puts the conclusions in line with most other estimates. It was a case where a model was very strongly influenced by a single incorrect model assumption. Those who found the data error and wrote the commentary were surprised, noting that it was counterintuitive that a single error could influence a model’s conclusions to such a high degree.

     The paper’s original authors from the Potsdam Institute for Climate Impact Research, in Germany, however, think that correcting the Uzbekistan error only reduces their conclusion by about 2%, from a 19% GDP drop by 2050, to a 17% drop. They arrived at this conclusion by changing the setup of the model. They claim that the paper’s conclusions are still valid, according to a post on their web page. Others are quite skeptical. One of the commentary authors, Solomon Hsiang, noted:

Science doesn’t work by changing the setup of an experiment to get the answer you want,” Hsiang said. “This approach is antithetical to the scientific method.”

     Osaka writes that during peer review, according to a peer review document published by Nature, one anonymous reviewer wrote:

I find all of this well explained and fairly convincing, yet, purely subjectively, I have a hard time in believing the results, which seem unintuitively large given damages aren’t perfectly persistent.”

     After the paper was published, the Potsdam Institute for Climate Impact Research put out a press release claiming that there were $38 trillion in global climate impact damages each year. That number seems pretty ridiculous.

     According to climate impact scientist Roger Pielke Jr., some in the media relished the paper’s original conclusions:

Some went far beyond the paper’s claims. The Washington Post explained (incorrectly) that the paper showed why food prices were up last summer:

In March, a study from scientists at the European Central Bank and the Potsdam Institute for Climate Impact Research found that rising temperatures could add as much as 1.2 percentage points to annual global inflation by 2035. The effects are taking shape already: Drought in Europe is devastating olive harvests. Heavy rains and extreme heat in West Africa are causing cocoa plants to rot. Wildfires, floods and more frequent weather disasters are pushing insurance costs up, too.”

     He also notes that the paper found its way into some important policy groups, including the U.S. Congressional Budget Office, the OECD, the World Bank, and the UK Office for Budget Responsibility. In addition, the paper’s conclusion was accepted and adopted into the Network for Greening the Financial System (NGFS), a consortium of more than 100 central banks from around the world. It is a forum for the world’s central banks and regulators to work on a set of climate risk principles.

     The error prompted Greg Hopper of the Bank Policy Institute to write about the paper’s erroneous conclusions, calling it a “Flawed Analysis With Massive Economic Consequences” in the subtitle. Hopper wrote his opinion in December 2024, after Nature indicated that they were looking into the data, but before the source of the error was found. He emphasized the arbitrary nature of the modeling and the ability to derive very different conclusions from the same data:

The statistical procedure used to justify the new damage model is arbitrary and could easily have produced a damage function that would have predicted much smaller losses of global real income. Those much smaller loss projections should not be taken seriously either. In our assessment of the damage function, we found very limited statistical evidence for any causation between the climate variables and material economic damage and no statistical evidence for the purported influence of the temperature variables that drive almost all the economic damage.”

    As a result of the paper, the NGFS and affiliates, which may include the U.S. Federal Reserve, updated the ‘damage function’ of their climate scenarios to reflect the paper’s original conclusions. These should now be further updated, preferably without any influence from the paper’s questionable re-modeled conclusions. The data issue, which reveals a possible flaw in some models (where a single data point in a large data set can drastically alter the conclusion, albeit counterintuitive) and also shows the risks of climate risk modeling itself. Hopper describes the current state of academic damage function research as highly uncertain. Economic modeling is notoriously difficult and gets more difficult the further it goes into the future. Hopper’s article goes into great detail about damage function modeling and analysis. Below, Hopper explains how damage function models work.



Source: The NGFS’s New Climate Damage Function: A Flawed Analysis With Massive Economic Consequences. Greg Hopper. Bank Policy Institute. December 9, 2024. The NGFS’s New Climate Damage Function: A Flawed Analysis With Massive Economic Consequences - Bank Policy Institute



     Nordhaus noted that damage function modeling was challenging. Hopper points out one potential source of error in Nordhaus’s modeling that assumes global mean temperature equates to a similar change everywhere, when the reality is that temperatures change by much different magnitudes in different regions, such as the Arctic, where temperatures have risen much more than in other regions. This would theoretically skew the data.

     The graph below shows how much the conclusions changed when NGFS updated their damage function model to that of the single new paper, ignoring all the other papers. The change is pretty dramatic, and that in itself should have been a cause for concern or at least triggered a desire to look further into the data.


Source: The NGFS’s New Climate Damage Function: A Flawed Analysis With Massive Economic Consequences. Greg Hopper. Bank Policy Institute. December 9, 2024. The NGFS’s New Climate Damage Function: A Flawed Analysis With Massive Economic Consequences - Bank Policy Institute


     Hopper wrote another article, published just yesterday, after the source of the error was found, noting that the NGFS should have seen that something was amiss in their damage function model. He also critiques the new model put forth by the paper’s authors and discredits it. It is worth reading his conclusion. Below that is a graph showing what the conclusion would have been for the original model, had the erroneous Uzbekistan data been removed. This alone shows that moving to another model is really very likely an attempt to fit the data to pre-determined conclusions.

Conclusion

In November 2024, central banks and other regulators across the globe, under the auspices of the NGFS, put into their climate toolkit a climate damage function that could justify more stringent climate targets for banks, fines for banks that operate in Europe and potentially additional capital. About a month after the new damage function was released, a BPI study showed that the climate damage function had some serious statistical problems. It should not have been terribly surprising then to learn the climate damage function also had a serious data error that inflated its loss numbers by about a factor of three.

What is surprising is the reaction of the NGFS. The regulators in the NGFS could have, and should have, known about the data error before putting the new model into their climate scenario toolkit. They also could have investigated and acted when the academic paper the climate damage function was based on was called into question by the academic journal that published it. But they took no action and still take no action.

The revised climate damage model is even more flawed than the original, since the statistical problems remain and it now appears that the model update was cherry-picked to reach a pre-determined conclusion. The NGFS should immediately retract the damage model from its climate scenario toolkit. The NGFS must also review and correct its procedures for putting its climate models into production to prevent a model with a serious data error from again being adopted, especially when the error should have been caught. A general review and update of NGFS model control procedures is necessary to restore faith in their climate toolkit and data.

The more troubling questions remain unanswered: did any central bank or other regulator follow up with Nature or Kotz et al. when it was publicly revealed that there was a problem with the model they use to regulate banks? If they did not, why not? Are regulators putting their thumb on the scale, targeting a pre-determined conclusion rather than using the best scientific evidence?

 

    


Source: The Flawed NGFS Damage Function Is Even More Flawed Than We Thought. Greg Hopper. Bank Policy Institute. August 14, 2025. The Flawed NGFS Damage Function Is Even More Flawed Than We Thought - Bank Policy Institute


 

References:

 

This climate study made a big error. One piece of data was to blame. Shannon Osaka. Washington Post. August 6, 2025. This climate study made a big error. One piece of data was to blame.

Too Big to Fail: A major new scandal in climate science. Roger Pielke Jr. The Honest Broker. Substack. August 15, 2025. Too Big to Fail - by Roger Pielke Jr. - The Honest Broker

The economic commitment of climate change. Maximilian Kotz, Anders Levermann & Leonie Wenz. Nature volume 628, pages 551–557 (April 17, 2024). The economic commitment of climate change | Nature

Analysis: The climate papers most featured in the media in 2024. Carbon Brief. January 15, 2025. Analysis: The climate papers most featured in the media in 2024 - Carbon Brief

Data anomalies and the economic commitment of climate change. Tom Bearpark, Dylan Hogan & Solomon Hsiang. Nature volume 644, pages E7–E11 (August 6, 2025). Data anomalies and the economic commitment of climate change | Nature

Peer Review File. Manuscript Title: The economic commitment of climate change. 2024. TPR1

38 trillion dollars in damages each year: World economy already committed to income reduction of 19 % due to climate change. Potsdam Institute for Climate Impact Research. April 17, 2024. 38 trillion dollars in damages each year: World economy already committed to income reduction of 19 % due to climate change — Potsdam Institute for Climate Impact Research

Nature study on economic damages from climate change revised. Potsdam Institute for Climate Impact Research. August 6, 2025. Nature study on economic damages from climate change revised — Potsdam Institute for Climate Impact Research

The NGFS’s New Climate Damage Function: A Flawed Analysis With Massive Economic Consequences. Greg Hopper. Bank Policy Institute. December 9, 2024. The NGFS’s New Climate Damage Function: A Flawed Analysis With Massive Economic Consequences - Bank Policy Institute

The Flawed NGFS Damage Function Is Even More Flawed Than We Thought. Greg Hopper. Bank Policy Institute. August 14, 2025. The Flawed NGFS Damage Function Is Even More Flawed Than We Thought - Bank Policy Institute 

No comments:

Post a Comment

         In a guest post on Roger Pielke Jr.’s blog, The Honest Broker, Justin Ritchie, a professor at the University of British Columbia, a...