Estimating the Consequences of Climate Change from Variation in Weather

I formally define the limits of what we can learn about the consequences of long-run climate change from short-run weather shocks. I show that conventional approaches to estimating climate impacts require assuming that payoffs are independent of capital and resource stocks. I derive a new indirect least squares estimator that bounds long-run climate impacts from short-run responses to weather under less restrictive assumptions. In an application to the U.S. economy, I project that climate change will reduce steady-state income per capita by at least 1.8% in the Midwest, by at least 1% in the Northeast, and by at least 0.23% in the West. Each lower bound implies damages beyond the 95% confidence intervals produced by the conventional approach.