Replication data and code for: "It will rain: The effect of information on flood preparedness in urban Mozambique"

Published: 12 July 2026| Version 1 | DOI: 10.17632/xdk6p6rf5d.1
Contributor:
Stefan Leeffers

Description

This package contains the replication data and code for Leeffers, Stefan (2026), "It will rain: The effect of information on flood preparedness in urban Mozambique," Journal of Development Economics 182, 103787, https://doi.org/10.1016/j.jdeveco.2026.103787. The package reproduces all tables in the published paper (Tables 1 to 4) and in the Online Appendix (Tables D1 to D10). It includes seven anonymized analysis datasets in Stata format (household baseline, endline, and scale-up surveys; community leader baseline and endline surveys; location and community characteristics) and the Stata do-files that generate every table from these datasets. Running 0_master.do executes the full analysis and writes one .tex file per table to the Output folder. A compiled Tables.pdf typesets all output with the table numbers and titles of the published paper for side-by-side verification against the published version. The README file documents the mapping from each published table to its do-file, output file, and dataset, along with variable descriptions and software requirements (Stata 16 or later; the user-written packages estout, icw_index, and ietoolkit are installed automatically). The study is a randomized controlled trial on risk information and flood preparedness in Quelimane, Mozambique, pre-registered at the AEA RCT Registry (AEARCTR-0008451).

Files

Steps to reproduce

Data collection. The data come from a randomized controlled trial covering 300 urban communities in Quelimane, Mozambique, pre-registered at the AEA RCT Registry (AEARCTR-0008451) and approved by the IRB of Nova School of Business and Economics (February 24, 2021, updated January 4, 2022). Household and community leader surveys were administered in person by trained enumerators before the 2021-2022 wet season (baseline, September to October 2021), during the video scale-up phase (October to November 2021), and after the wet season (endline, April to May 2022). Observational data on community cleanliness were collected during the wet season through repeated field team visits to streets, public spaces, and drainage canals, combining enumerator assessments with photo-based waste indicators derived from machine-learning classification of over 25,000 photographs. Community-level geographic controls (area, buildings, elevation, drainage canals, shared boundaries with treated communities) were constructed in QGIS. Data processing. The seven datasets in this package are de-identified extracts of the project's restricted-access files, trimmed to the variables required by the analysis. Their construction, including all derived variables and indices, is documented in Dofiles/_create_package_data.do, which is included for transparency but cannot be run without the original restricted data. Reproducing the results. Download all files, preserving the folder structure (Dofiles, Data, Output). Open 0_master.do, set the global "root" to the folder containing the package, and run it in Stata 16 or later. It installs the required user-written packages from SSC (estout, icw_index, ietoolkit), runs all analysis do-files, and writes one .tex file per table to the Output folder, reproducing Tables 1 to 4 of the published paper and Tables D1 to D10 of the Online Appendix. The full run takes a few minutes. To verify the output, read the .tex fragments directly or compile Tables.tex (pdflatex) to produce Tables.pdf, which typesets all tables with the numbers and titles of the published paper; a pre-compiled copy is included. The README documents the mapping from each published table to its do-file, output file, and dataset.

Institutions

Categories

Economics

Funders

Licence