Future climate change impacts on U.S. agricultural yields, production, and market

Abstract

This study provides estimates of climate change impacts on U.S. agricultural yields and the agricultural economy through the end of the 21st century, utilizing multiple climate scenarios. Results from a process-based crop model project future increases in wheat, grassland, and soybean yield due to climate change and atmospheric CO₂ change; corn and sorghum show more muted responses. Results using yields from econometric models show less positive results. Both the econometric and process-based models tend to show more positive yields by the end of the century than several other similar studies. Using the process-based model to provide future yield estimates to an integrated agricultural sector model, the welfare gain is roughly $16B/year (2019 USD) for domestic producers and $6.2B/year for international trade, but domestic consumers lose $10.6B/year, resulting in a total welfare gain of $11.7B/year. When yield projections for major crops are drawn instead from econometric models, total welfare losses of more than $28B/year arise. Simulations using the process-based model as input to the agricultural sector model show large future production increases for soybean, wheat, and sorghum and large price reductions for corn and wheat. The most important factors are those about economic growth, flooding, international trade, and the type of yield model used. Somewhat less, but not insignificant factors include adaptation, livestock productivity, and damages from surface ozone, waterlogging, and pests and diseases.

Keywords: Agricultural sector; And livestock; Climate change; Crops; Economics; Modeling.

Fig. 1.

U.S. agricultural impacts modeling approach.

Fig. 2.

U.S. crop production response from LPJmL showing the percent change between 1980–2005 and 2075–2100. RCP4.5 is shown in green and RCP8.5 in orange. The darker hues show the combined effects of climatic factors (precipitation and temperature) together with CO₂ fertilization. The lighter hues show the isolated effect of CO₂ fertilization. The boxplots show the distribution of effects across 6 CMIP5-LOCA climate models. The horizontal bars within the boxes indicate the median, the ends of the whiskers extend to 1.5 times the interquartile range.

Fig. 3.

U.S. crop production responses from the LPJmL and empirical crop models driven by downscaled CMIP5 climate simulations in comparison to the GGCMI-CMIP6 ensemble, showing the percent change from 1980–2005 to 2075–2100. All climate model inputs for these simulations were from the RCP8.5 scenario. For each crop, the symbols on the left side show the overall production change, whereas the symbols on the right show the CO₂ effect alone. The orange box plots show the distribution of results from the GGCMI ensemble, including 12 crop models driven by 5 CMIP6 climate models (Jägermeyr et al., 2021 for details). The horizontal bars within the boxes indicate the median across all climate-crop model combinations, the boxes show the interquartile range, and the whiskers extend to 1.5 times the interquartile range. The open red circles show the LPJmL response driven by the CMIP6 models used in the GGCMI ensemble. The solid red circles show the LPJmL response driven by CMIP5-LOCA climate model inputs. The triangles show similar results but driven only by the CO₂ effect. The closed grey circles show the econometric model response driven by LOCA climate model inputs.

Fig. 4.

Time series through 2100 of LPJmL production responses to LOCA-downscaled outputs from the 6 CMIP5 models used in this study for RCP4.5 and RCP8.5. Results are shown as a 30-yr moving mean (bold line), annual data as the mean across GCMs (thin line), and the range of individual GCM realizations (lightly shaded areas).

Fig. 5.

Projected end-of-century yield changes from LPJmL driven by CMIP5-LOCA RCP4.5 and RCP8.5 and different CO₂ concentrations (relative change between 1980–2005 and 2075–2100). ‘2015 CO₂’ refers to simulations with CO₂ concentration held constant at the 2015 level and ‘transient CO₂’ to simulations with transient CO₂ as used in CMIP6. Results are masked for current cropland extent (Portman et al., 2010).

Fig. 6.

CO₂ effect on nutritional content as shown by the C:N ratio under default (i.e., transient) divided by the C:N ratio under constant 2015 CO₂. Higher C:N ratios can generally be interpreted as lower nutritional quality. Results are from LPJmL driven by LOCA-downscaled outputs from the 6 CMIP5 models used in this study for RCP4.5 and RCP8.5.

Fig. 7.

Average annual ensemble welfare changes at different arrival degrees under the 2020 base economy (results in billions of 2019 USD).

Fig. 8.

Shifts in ensemble average weighted centroid of crop production by selected crops and arrival degree. Centroids of each crop are indicated by continuum from a light to a darker color as the arrival degrees increase.

Fig. 9.

Changes in economic welfare associated with each of the sensitivity analyses described in this section.