In a recent study, Wolfram Schlenker and I set out to develop a better statistical model linking weather and U.S. crop yields for corn, soybeans, and cotton—the largest three crops in the U.S. in production value. Corn and soybeans are of particular interest because they are really important for global food prices and the U.S. contributes about 40% of the world’s production of these crops, and a much larger share of world exports for these crops. The goal was to find the causal links between observed climate and yields so that we might predict how yields will change as the climate changes.
The novelty of our work is that it carefully accounts for variation in temperatures over time and space. This contrasts with earlier studies that compare yields to average weather outcomes, like average temperature. The problem with averages is that they dilute nonlinearities—effects of the extremes—that are clearly important for crop growth and yield.
Our major new finding is that extreme heat is critical to yield outcomes. A key measure of extreme heat is given by how much temperatures exceed about 29°C (84°F) during the growing season. The threshold varies somewhat by crop—29°C (84°F) for corn, 30°C (86°F) for soybeans, and 32°C (90°F) for cotton. Below the threshold, warmer temperatures are more beneficial for yields; damages stemming from temperatures much above the threshold can be staggeringly large.
Another important finding is that the non-linear relationship between weather and yields across time in a fixed location is nearly an identical match to the non-linear relationship found when comparing weather distributions and yields across locations with different climates. This indicates farmers in southern areas have been unable to adapt to their warmer climates.
When we use the estimated relationship to predict yield outcomes under projected climate change scenarios, we find that between 2070 and 2099 nationwide average yields on corn, soybeans, and cotton are projected to fall 30–46% under the slowest Hadley III warming scenario and 63–82% under the fastest Hadley III warming scenario. Predicted declines are substantial even in the more immediate future (2020–2049).
What are the implications of these dramatic findings? Well, it’s hard to tell. The results should provide good motivation for seed companies and plant scientists to develop more heat tolerant plants. Some think CO2 fertilization will offset a large portion of losses from climate change. But it also seems clear that farmers will want to change the kinds of crops they grow and possibly their planting dates. It’s also possible gains in cooler parts of the world will offset losses in the U.S. The extent to which new areas become arable will determine how crop prices will change. To me, the inescapable conclusion is that the magnitude of climate changes currently anticipated will cause the face of agriculture worldwide to change dramatically. And to some extent this is true regardless of whether or not we markedly reduce CO2 emissions in the near future.
Wolfram Schlenker and Michael Roberts’ study entitled “Nonlinear temperature effects indicate severe damages to U.S. crop yields under climate change” is published in the Proceedings of the National Academy of Sciences (PNAS). View abstract. View his blog post on the study.
Michael J. Roberts
Department of Agricultural & Resource Economics
North Carolina State University
Publisher of Greed, Green and Grains