Economic consequences of the spatial and temporal variability of climate change

Damage functions in integrated assessment models (IAMs) map changes in climate to economic impacts and form the basis for most of estimates of the social cost of carbon. Implicit in these functions lies an unwarranted assumption that restricts the spatial variation (Svar) and temporal variability (Tvar) of changes in climate to be null. This could bias damage estimates and the climate policy advice from IAMs. While the effects of Tvar have been studied in the literature, those of Svar and their interactions with Tvar have not. Here we present estimates of the economic costs of climate change that account for both Tvar and Svar, as well as for the seasonality of damages across sectors. Contrary to the results of recent studies which show little effect that of Tvar on expected losses, we reveal that ignoring Svar produces large downward biases, as warming is highly heterogeneous over space. Using a conservative calibration for the damage function, we show that previous estimates are biased downwards by about 23-36%, which represents additional losses of about US$1,400-US$2,300 billion by 2050 and US$17-US$28 trillion by the end of the century, under a high emissions scenario. The present value of losses during the period 2020-2100 would be larger than reported in previous studies by $47-$66 trillion or about 1/2 to 3/4 of annual global GDP in 2020. Our results imply that using global mean temperature change in IAMs as a summary measure of warming is not adequate for estimating the costs of climate change. Instead, IAMs should include a more complete description of climate conditions.