Fig. 3: Relation between the cumulative GDP losses until 2100 (in 2005 $US) in the absence of climate policy and the economically optimal warming until the end of the century, given uncertainty in the estimates of the historical impact and uncertainty in the climate sensitivity value.

Scattered points give the uncertainty ensemble in the historical relation between temperature increase and economic growth for three different climate sensitivities; red points for 4 °C climate sensitivity, black points for the original climate sensitivity calibration in the DICE-2013R model, and blue points for 2 °C climate sensitivity. Each point depicts the DICE-2013 model output for a damage function calibrated according to one of the 1000 bootstraps of the historical regression. Curves in the main plot represent the best fit for the relation between cumulative damage costs and optimal warming. The histograms below and on the left give the frequency of the model results as well as the medians and likely ranges for each of the three climate sensitivities. The likely rage of optimal end-of-century warming is approximately located between 2.3 °C and 3.4 °C with a median of 2.5 °C for the climate sensitivity of 4 °C, between 1.8 °C and 3 °C with a median of 2.1 °C for a climate sensitivity of 2.9 °C and between 1.3 °C and 2.5 °C with a median of 1.7 °C for a climate sensitivity of 2 °C. The results of the original DICE versions are located outside the likely ranges as shown by the black brackets.