Rev. Env. Econ. Policy, revised 2/11.
University of California at Berkeley, Berkeley, CA
Solomon M. Hsiang, Wolfram Schlenker, and
Adam H. Sobel
Columbia University, New York, NY.
This paper discusses data sets used in climate sciences and provides an introduction to climate models as well as the type of output they provide. We discuss the major dimensions across which climate models differ and how uncertainty related to model predictions differs from the traditional economic view of uncertainty. We identify five major issues related to the use of gridded historical weather products and climate model output: Spatially specific prediction error due to the coarse resolution of climate model output; spatially varying correlations between different weather indicators; differences in cross sectional versus time series variation, strong spatial correlation in weather variables and finally the endogeneity of weather coverage. We discuss the conceptual issues arising from these five issues for economic impact studies of climate change.