For one, the "integrated" damage model does not use any climate data other than temperature. Rainfall, for instance, is crucial for agriculture, but is ignored.

For one, the "integrated" damage model does not use any climate data other than temperature. Rainfall, for instance, is crucial for agriculture, but is ignored. Note the parameters of the model: https://web.archive.org/web/20141014141710/http://www.econ.yale.edu/~nordhaus/homepage/DICE_documentation_111706.pdf https://web.archive.org/web/20141014141710/http://www.econ.y... Secondly, the damage function is a quadratic polynomial (see the parameters a1 and a2 in the formula). But that is an unreasonable assumption, as heating can induce exponential or logistic effects (or worse). More explanations about this by prof. Keen starting 14:42 in the video: https://youtu.be/5LvyxH3O7kE?t=882 https://youtu.be/5LvyxH3O7kE?t=882 Take, for instance, a phase change. Water boils at 100°C extremely nonlinearly compared to temperature. In fact, in a stepwise manner.