Thread by @ArielOrtizBobea

1/ My research article out today in @ScienceAdvances shows U.S. agriculture has become considerably more sensitive to climatic shocks since the 1980s in key core regions like the Midwest.
- Cornell Chronicle: bit.ly/2PyLwF2 (w/ video)
- Paper: bit.ly/2UGLpuP

2/ This is joint work with @Barefoot_Econ and Bob Chambers (Maryland) and was funded by @NSF and @USDA_NIFA. I also thank @AtkinsonCenter for their support.

3/ Most of the growth in agricultural production in the US has been driven by gains in productivity over the past 50 years. We wanted to know whether these profound transformations altered the sector's ability to cope with extreme climatic events.

4/ This is a critical *global* question, bc meeting the rising demand for food has to eventually come from higher ag productivity rather than through more inputs (e.g. more land/chemicals). So our study provides lessons beyond the U.S. Let me talk about the study!

5/ To capture the *whole* ag sector we relied on state-level measures of Total Factor Productivity (TFP) which we linked to detailed weather information. Previous work focuses on a limited set of commodities (e.g. corn) or partial measures of productivity (net revenue/acre).

6/ We found a clear link between summer temperatures >30C and lower productivity in the eastern portions of the country. In the West, things look flat, likely due to irrigation which we cannot account for in our model.

7/ Importantly, we found that the productivity-temperature relationship has changed over time! For instance, in the Midwest, a +2C warmer summer prior to the 1980s led to a 11.3% drop in TFP. Since then, the same conditions lead to a drop of 29.3%. Almost a 3-fold increase.

8/ We see some patterns in other regions but they are not as drastic or statistically significant as in the Midwest, where the signal is very clear.

9/ The big issue with this result is that damaging summers (>2C) happened about 6.4% of the time ~ once every 20 years. However, a uniform +1C warming would make them 4.5 more frequent, or once every 4-5 years. This is substantial.

10/ We wondered what was driving this result, so we looked at crop and livestock output. We found the same "signature" in crop output, but not in livestock production, so this seems linked to a rising sensitivity of crop production.

11/ An important observation is that the composition of the sector has evolved. Midwest is specializing more in non-irrigated field crops, whereas other regions focus more on livestock/hogs/poultry/dairy production (South and Northeast) or speciality crops (West).

12/ This specialization has contributed to the rising climatic sensitivity of the Midwest. In a nutshell, the Midwest has increasingly specialized in increasingly climate-sensitive production. And we can see it in our study.

13/ As economists we generally agree that specializing following regional comparative advantage is desirable. But are these trends efficiently accounting for the rising climatic risk? Are farm/energy policies (e.g. RFS) exacerbating these trends?

14/ It seems important we realize US agriculture is making some tradeoffs here, between higher productivity and higher sensitivity to climatic shocks, particularly in the Midwest.

Thanks for reading, and RT if you feel you learned something ...or just liked the pictures! :)

Coming soon: link to data and replication code

Many thanks to the amazing R-squared team at the @Cornell Institute for Social and Economic Research (CISER) for going through our code and making sure our recent @ScienceAdvances study is replicable on any platform! @CornellRsrch @williamblock. Link: doi.org/10.6077/f26v-x…

Quick note on #replication. When doing a bootstrap, make sure you set your seed for generating random numbers right before running your model! In this study we did not do that, so our replication does not *perfectly* match the published results to the smallest decimal point.