Replication files for “The Long-Run Effects of Colonial Institutions on Household Agricultural Assets: Evidence from Cameroon”
Description
This replication package accompanies the paper “The Long-Run Effects of Colonial Institutions on Household Agricultural Assets: Evidence from Cameroon.” It contains a harmonized replication dataset of 12,007 household observations from the 2011 and 2018 Cameroon Demographic and Health Surveys and the 2022 Cameroon Malaria Indicator Survey, together with a Stata master do-file and detailed replication instructions. The dataset includes household agricultural-asset outcomes, household characteristics, survey weights, geographic distance to the historical British–French colonial boundary, treatment-side indicators, measures of access to markets and administrative centres, and indicators of local conflict exposure. The master do-file reproduces the balance tests, geographic regression-discontinuity estimates, heterogeneity analyses, robustness checks, conflict specifications, mechanism analyses, tables, and figures reported in the paper. The analysis was conducted using Stata 18 and requires the rdrobust package, which is installed automatically by the master do-file if necessary. No additional data preparation is required to run the replication package. Users should consult the README file for information on the package contents, software requirements, file paths, and execution instructions.
Files
Steps to reproduce
The replication dataset was constructed from the 2011 and 2018 Cameroon Demographic and Health Surveys (DHS) and the 2022 Cameroon Malaria Indicator Survey (MIS). These surveys use nationally representative, stratified two-stage sampling designs. Household records from the three survey rounds were harmonized to construct comparable measures of agricultural-land ownership, land area, livestock ownership, livestock holdings, household composition, and survey weights. Georeferenced survey clusters were spatially matched to the historical British–French colonial boundary. For each cluster, the shortest distance to the boundary was calculated and assigned a positive value for clusters located on the former British side and a negative value for clusters located on the former French side. Additional geospatial variables measure distance to the nearest market and administrative centre. Conflict-exposure indicators were constructed by matching survey clusters to georeferenced conflict-event data, using information on proximity, fatalities, and event type. The resulting harmonized dataset contains 12,007 household observations from 2011, 2018, and 2022. It is provided as Data_replication.dta. All analyses were conducted in Stata 18 using the rdrobust package. To reproduce the results: 1-Download Data_replication.dta, Replication_Master_final.do, and the README file into the same folder. 2-Open Replication_Master_final.do in Stata 18 or later. 3-Modify the global DATAFILE path at the beginning of the do-file so that it points to the local location of Data_replication.dta. 4-Run the master do-file from beginning to end. The program checks whether the rdrobust package is installed and installs it automatically if necessary. It then loads the replication dataset and reproduces the balance tests, baseline geographic regression-discontinuity estimates, heterogeneity analyses, alternative RDD specifications, bandwidth-sensitivity tests, survey-round estimates, intensive-margin results, conflict robustness checks, mechanism analyses, tables, and figures reported in the paper. No additional data preparation is required to reproduce the reported results from the archived replication dataset.