Tashkent, Uzbekistan (UzDaily.com) — The prestigious scientific journal Nature has acknowledged a serious error in a high-profile study on the economic impacts of climate change, which was largely caused by distorted statistical data from Uzbekistan.

The study, published in 2024, predicted a catastrophic reduction in global GDP: 19% by 2050 and 62% by 2100 under a high-emission scenario. These estimates were three times higher than previous forecasts and quickly entered global scientific and political discourse, with users including the World Bank, the US government, and other international organizations. According to CarbonBrief, the study was the second most cited in the media that year.

However, a new analysis published in Nature’s commentary section revealed that these dramatic projections resulted from major statistical distortions. Specifically, the inclusion of anomalous data from Uzbekistan significantly inflated the estimated economic losses.

After excluding Uzbekistan’s figures, the recalculations changed dramatically: the projected global GDP decline by 2100 dropped from 62% to 23%, and by 2050 from 19% to 6%. Solomon Hsiang, Director of the Global Policy Lab at Stanford University and co-author of the corrective analysis, explained that when countries were sequentially excluded from the database, only Uzbekistan had such a substantial impact on the final model. The original dataset suggested that Uzbekistan’s GDP in 2000 had plummeted by nearly 90%, while in 2010 it purportedly grew by over 90% in certain regions. Other sharp fluctuations were recorded that do not align with reality: according to official World Bank data, Uzbekistan’s GDP growth rates over the last 40 years ranged from -0.2% to +7.7%.

Hsiang noted that these extreme deviations severely distorted the model linking economic development to climate factors. He emphasized that even if the influence of a relatively small country on global estimates seems disproportionate, it underscores the critical need for meticulous verification of source data.

The original authors from the Potsdam Institute for Climate Impact Research acknowledged the error and stated they had recalculated their results using corrected data. Nevertheless, as the researchers claim, the key conclusions remain valid: GDP losses by 2050 were revised from 19% to 17%.

Co-author Leonie Wenz expressed gratitude to colleagues for identifying the issue and noted that error correction is an integral part of the scientific process. However, critics argue that maintaining the original conclusions required the authors to alter their methodology, which they believe calls the reliability of the findings into question. Hsiang, in turn, stressed that science should not become a process of fitting models to desired outcomes.

Nature’s editorial board announced the launch of an internal review of the case and declared its intention to take necessary measures. Editor Carl Zimelis reminded that the scientific process involves continuous scrutiny both before and after publication.

Hsiang commented that the incident serves as a compelling example of the effectiveness of the scientific method. He added that for those skeptical about science’s capacity for self-correction, such situations demonstrate its vitality and strength.