Improving statistical estimates of the economic cost of climate change

Climate change is affecting global productivity through many avenues that include changes in the frequency and severity of extreme events as well as the more gradual, longer-term changes in average values for temperature and rainfall. Morris et al. (2025) found in their literature review that there is an order of magnitude difference in damage estimates between papers based on structural models and papers based on statistical models. For example, with a 30C temperature increase, the structural models estimated the impact on GDP to be around a 2 % loss in global GDP while the statistical models estimated losses at around 20 %, with some losses reported above 55 % of global GDP. In this commentary, we provide an explanation for the gap and suggest that the large estimates can result from assumptions that incorrectly bias the damage estimates to give larger values than warranted.