Unlike the Kyoto Protocol, which set legally binding emission reduction targets (as well as penalties for non-compliance) only for industrialized countries, the Paris Agreement requires all countries – rich, poor, developed and developing – to take their share and reduce their greenhouse gas emissions. To this end, the Paris Agreement provides for greater flexibility: commitments that countries should make are not included, countries can voluntarily set their emissions targets and countries will not be penalized if they do not meet their proposed targets. But what the Paris agreement requires is to monitor, report and reassess, over time, the objectives of individual and collective countries, in order to bring the world closer to the broader objectives of the agreement. And the agreement stipulates that countries must announce their next round of targets every five years, contrary to the Kyoto Protocol, which was aimed at this target but which contained no specific requirements to achieve this goal. Below, we submit our results to numerous robustness tests. First, we add to the climate change sensitivity analysis in the main text by taking into account a total probability density function for ECS values. Second, we look at the effects of uncertainty in BHM`s estimates. In this regard, we take into account other estimates of 1 and 2, on the one hand, and different model specifications, on the other. This analysis is followed by a comparison with the DJO estimates. Third, we study the influence of uncertainty on the socio-economic future by recalibrate the DICE model based on a selected group of SSPs.
As a by-product of this calibration, we obtain mitigation and cost functions that emulate the costs of a detailed process model and thus represent a further evolution of the DICE model. The derivation of these functions allows us to test the sensitivity of our results to these alternative emission reduction costs. We complete this section by providing more information on the robustness test for the preferred parameters shown in the main text. The resulting distribution of economically optimal temperatures in 2100 inherits the properties of the distribution of the probabilities of the ECS. As the main text shows, higher ECS values imply a higher temperature target because of the limited room for manoeuvre to achieve lower temperatures with climate policy. In addition, the more detailed sensitivity analysis confirms that the most likely temperature targets are 2oC. Nevertheless, there is a certain probability, although very low, that the economically optimal temperature target may be significantly higher, perhaps up to 4oC.