Nonlinear Effects of Weather on Crop Yields: Implications for Climate Change

We estimate the relationship between weather and corn, soybean, and cotton yields using a unique flne-scale data set of daily weather records spanning the entire United States from 1950-2004. Our dataset incorporates the distribution of temperatures between the minimum and maximum within each day and across all days in the growing season. We flnd a robust and signiflcant nonlinear relationship between temperature and yields, which increase linearly in temperature until about 29 ‐ C for corn and soybeans and 33 ‐ C for cotton. However, higher temperatures quickly become very harmful. This sharp nonlinearity implies that studies using average temperatures give a biased estimates of the average impact of temperatures. The estimated relationship between yield and temperature is stable over time and location. Although technological change has increased average yields, it has not increased relative heat tolerance. These flndings suggest there may be limited potential for technological adaptation of plants to changes in climate. The stability of the estimated relationship across regions and crops further suggests that the results may be transferable to other parts of the world. We use the estimated relationship to predict efiects of the latest warming scenarios on crop yields. The predicted impacts are large and signiflcant across speciflcations and predict reductions in yields of 72-80% for the three crops under the most rapid warming scenario in our preferred model speciflcation.