Estimates of global economic damage from climate change assess the effect of annual temperature changes. However, the roles of precipitation, temperature variability and extreme events are not yet known. Here, by combining projections of climate models with empirical dose–response functions translating shifts in temperature means and variability, rainfall patterns and extreme precipitation into economic damage, we show that at +3 ^°C global average losses reach 10% of gross domestic product, with worst effects (up to 17%) in poorer, low-latitude countries. Relative to annual temperature damage, the additional impacts of projecting variability and extremes are smaller and dominated by interannual variability, especially at lower latitudes. However, accounting for variability and extremes when estimating the temperature dose–response function raises global economic losses by nearly two percentage points and exacerbates economic tail risks. These results call for region-specific risk assessments and the integration of other climate variables for a better understanding of climate change impacts.

对气候变化造成的全球经济损失的估计评估了年度气温变化的影响。然而，降水、温度变化和极端事件的作用尚不清楚。在这里，通过将气候模型的预测与经验剂量反应函数相结合，将温度平均值和变异性、降雨模式和极端降水的变化转化为经济损失，我们表明，在+3  ^° C时，全球平均损失达到国内生产总值的10%，在较贫穷的低纬度国家，影响最严重（高达 17%）。相对于年度温度损害，预测变率和极端事件的额外影响较小，并且主要由年际变率主导，特别是在低纬度地区。然而，在估计温度剂量反应函数时考虑变异性和极端情况会使全球经济损失增加近两个百分点，并加剧经济尾部风险。这些结果需要针对特定​​区域的风险评估并整合其他气候变量，以更好地了解气候变化的影响。