Global Economic Exposure to Future Temperature Changes

In global-scale analyses of future climate change, "global average temperature change" is a commonly used summary statistic. Unfortunately, this statistic may not be useful for many types of economic analyses because it is an average over the planet's entire surface and is therefore dominated by changes over oceans and other uninhabited regions. Here, we attempt to summarize projected temperature changes in a manner that is more useful for economic analyses: we construct the distributions of future temperature exposure for a randomly selected person, a random hectare of cropland, and a random dollar of value-added. Our results streamline global cost analyses, enabling future studies to estimate global losses by combining their locally derived loss-functions with our estimates of global exposure. We demonstrate this application by estimating that low and middle income populations may suffer income losses of 9% annually due only to the effects of thermal stress on workers, a mechanism previously omitted from global cost estimates. In ancillary findings, we also document that (1) when exposure distributions are substituted for global average temperature change in standard models of economic costs, projected annual losses increase by trillions of dollars; (2) low and middle income populations will be twice as exposed to harmful temperatures as high income populations, based only on their locations; and (3) it is unlikely the direct effects of warming can have a positive net impact on the global economy.