A new analysis, published in a commentary to Nature, showed that such dramatic figures were the result of a statistical error. As it turned out, the data on Uzbekistan radically distorted the calculations, due to which the global forecasts were overestimated.
After excluding incorrect data on Uzbekistan, the forecasts changed significantly: the expected decline in global GDP by 2100 decreased from 62% to 23%, and by 2050 – from 19% to 6%. According to Solomon Hsiang, director of the Stanford Global Policy Lab and one of the authors of the new analysis, the error was discovered when different countries were removed from the database one by one. Removing Uzbekistan had the biggest impact on the final model.
In the original study, Uzbekistan’s GDP allegedly fell by almost 90% in 2000, and grew by more than 90% in some regions in 2010. At the same time, according to the World Bank, the country’s real GDP growth over the past 40 years has fluctuated between -0.2% and +7.7%. Hsiang noted that such extreme, fictitious jumps had a decisive impact on the calculations.
The authors of the original study from the Potsdam Institute for Climate Impact Research admitted their error and presented recalculated results. According to them, after adjusting the data, the projected loss by 2050 was 17%, which is close to the original estimates.
However, critics argue that the authors had to change their methodology to preserve the original results, which they say undermines the credibility of the study. Hsiang stressed that science should be based on accurate data, not on adjusting parameters to a desired result.
Nature’s editors have launched an internal review of the incident, with its editor Karl Tsimelis noting that science involves constant checking and self-correction. In summing up, Solomon Hsiang noted that this situation serves as a clear example of how the scientific approach allows us to find and admit mistakes. He stressed that if anyone doubts the effectiveness of the scientific method, then this case is proof of it.