In the United States, the social cost of carbon (SCC) is one of the foremost tools for calibrating the socially optimal approach for climate change policy. The SCC is estimated using climate-economic models with implicit temperature-damage relationships. Given the vast uncertainty surrounding climate impacts, meta-analyses of global climate damage estimates are a key tool for determining the relationship between temperature and climate damages, so as to communicate the current state of knowledge to model developers. Using a larger dataset than previously assembled in the literature, this paper highlights several methodological improvements that address bias present in previous meta-analyses of the temperature-damage relationship. Specifically, due to limited data availability, previous meta-analyses of global climate damages potentially suffered from multiple sources of bias: duplication bias, measurement error, omitted variable bias, and publication bias. By expanding our dataset (to include additional published and grey literature estimates), including methodological variables, and correcting the specification of temperature (to account for different reference periods), we are able to address and test for these biases. Estimating the relationship between temperature and climate damages using weighted least squares with cluster robust standard errors at the model level, we find strong evidence of duplicate bias. Using these results as an input in the DICE model – to update the DICE damage function – we determine that duplication and omitted variable bias significantly impact the damage-temperature relationship in past meta-analyses and the resulting SCC estimates. Focusing exclusively on non-catastrophic climate impacts, we find that the temperature-damage relationship estimated in Nordhaus and Sztorc (2013) is biased downwards by approximately 179% to 264%, depending on how climate change’s impacts on productivity are treated. This implies a downward bias in DICE’s SCC estimate by 203% to 314%, depending on the treatment of productivity. If we also consider catastrophic impacts, the potential bias in the SCC increases to 344% to 469%.


Downloads Statistics

Download Full History