a, b, Country share of global GDP at baseline (a) and by the end of the century (b) under SSP1, assuming no climate change. c, Distribution of global GDP by temperature, under baseline (black) and the end of the century SSP1 without climate change (red dashed); absent climate change, a substantial portion of global GDP is projected to be produced in countries with hotter average temperatures. d, Climate-model-predicted average global warming under RCP2.6 by the end of the century (x axis) versus the correlation between country-level baseline average temperature and country-level predicted warming in each model. In models that warm less at the global scale, countries that are currently warm tend to exhibit relatively larger warming, while in models that warm more at the global scale, countries that are currently cool tend to exhibit relatively larger warming. Future impacts on global GDP are a sum of country-specific impacts, which are a function of where each country is on the temperature response function (Fig. 1a) and the projected amount of future warming in that country; a given percentage impact in a country with a large GDP has a larger effect on global GDP than the same percentage impact in a country with small GDP. For high-temperature-optimum response functions (for example, Fig. 1g), impacts can be negative at +1 °C but positive at +2 °C because (i) absent climate change, a much larger proportion of total global GDP is projected by SSP1 to be produced in countries that are currently warmer than the optimum, and (ii) climate models with lower overall global warming projections under RCP2.6 tend to have higher relative warming in countries that are currently warm. This generates negative impacts at about 1 °C, where impacts are dominated by negative effects in warm countries (largely in the developing world), but positive impacts at about 2 °C, where high-latitude countries instead warm disproportionately and experience benefits that outweigh the damages in tropical countries.