Genetically Engineered Crop Adoption Support Yields and Cultivation under Climate Change

The code files in this object with extension ".m" are written in MATLAB R2024b, the ones with entension ".r(.R)" are written in R version 4.5.1, and the ones with extension ".do" are written in StataSE 17.0. 

The data for land use model estimated in the main text is "land_model_data.dta". Using this data and run "land_model_main_without_int.do" will generate result provided in Table 3 in the Extended Data for Main model results. Using this data and run "land_model_main_with_int.do" will generate result provided in Table 4 in the Extended Data for Main model results. Using this data and run "centroid_prediction.do" will generate centroid prediction results used in Figure 4 of the main text. Using this data and run "land_use_prediction.do" will generate land use prediction data used in Figure 5 of the main text, and the predicted land use difference between with GE and without GE in 2020 is stored in "2020diff_land_pre.csv". Using this data and run "Centroid_bootstrap.do" will generate the bootstrap results for centroid latitude and longitude prediction differences displayed in Table 5 in the Extended Data. All p-value of wald z-tests can be found in the regresion result tables in Stata after the run.

Before running the yield model codes in Stata, "reg2hdfespatial.ado" needs to be put in the ado folder of Stata. The data for yield model estimated in the main text is "yield_model_data.dta". Using this data and run "yield_model_stage3" will generate result provided in Table 1 in the Extended Data for yield model results. Using this data and run "yield_model_stage2" will generate result provided in Table 2 in the Extended Data for yield risk model results. Using this data and run "corn_yield_prediction.do", "soybean_yield_prediction.do", "cotton_yield_prediction.do" separately will generate yield prediction data and 95%CI used in Figure 3 of the main text. Using this data and run "corn_yield_risk_prediction.do", "soybean_yield_risk_prediction.do", "cotton_yield_risk_prediction.do" separately will generate variance of log yield prediction data and 95%CI used in Figure 3 of the main text. All p-value of wald t-tests can be found in the regresion result tables in Stata after the run.

"centroid_calc.do" generates historical land use centroid for corn, soybean and cotton used in Figure 2 of the main text. The calculated centroid is embedded in "Corn_real_centroid.R", "Soybean_real_centroid.R" and "Cotton_real_centroid.R" to produce Panel A, B and C of Figure 2 in the main text.

"Historical_land_use_change.do" generate historical land use percentage change data for corn, soybean and cotton used in Figure 1 of the main text. The data is stored in "data7820.csv". Using this data and run "Land_use_percentage_change.r" to produce Panel A, B and C of Figure 1 in the main text.

"Corn_predicted_centroids.r", "Soybean_predicted_centroids.r" and "Cotton_predicted_centroids.r" have predicted corn, soybean and cotton embedded in them respectively. Run them separately to get Panel A, B, and C of Figure 4 in the main text.

Run "Land_use_diff_predicted.r" using data stored in "2020diff_land_pre.csv" to get Panels in Figure 5 and Panels in Extended Data Figure 6.

Any data and code used in the robustness tests can be posted upon request.