timdeisemann/replication_files_seeds_of_misallocation

The data and programs replicate the analysis from "The seeds of misallocation: Fertilizer use and maize varietal misidentification in Ethiopia" by Nils Bohr, Tim Deisemann, Douglas Gollin, Frederic Kosmowski, Travis J. Lybbert (2024). Replication Package for: Nils Bohr, Tim Deisemann, Douglas Gollin, Frederic Kosmowski, Travis J. Lybbert, The seeds of misallocation: Fertilizer use and maize varietal misidentification in Ethiopia, Journal of Development Economics, 2024, 103349, ISSN 0304-3878, doi.org. (sciencedirect.com) By Tim Deisemann (EBRD, timdeisemann@gmail.com) and Nils Bohr (bohr.nils98@gmail.com). For any questions, please do reach out to us. This replication package produces all tables and figures shown in the paper and online appendix (except illustrative Figure A1). (1) Accessing the raw data We do include the raw data in the replication package. The ESS4 data is available through the World Bank Microdata Library at microdata.worldbank.org. The varietal data file was originally sourced from openicpsr.org. Following new information from breeders and scientists who reached out after the launch of the Ethiopian report, some correction were made to the file used for analysis, and available in the paper replication package ("Var_data_new.csv"). We expect to add this new version in the open repository after the finalisation of an upcoming report. (2) Executing the code for figures 1, A2, A3, tables 2, A1-A10 Execute "create_master_data.do" to prepare the merged main dataset used for the analysis, then execute the respective dofiles in order of figures and tables appearing in the paper to reproduce the reported results. Note that "create_fig1_data.do" produces "figure1_data.csv" which is then used to produce figure 1 (figure1.R). "Create_figA3_data.do" prepares the dataset used to produce Figure A3 including yields data. (3) Executing the code for the prediction/scaling exercise Machine learning algorithms are prone to yield varying results across different machine environments. The variations we observed did not …