Replication Materials for “Who Bears the Hidden Time Cost? Children’s Water and Firewood Collection, Socioeconomic Inequality, and Rainfall Variability in Bangladesh”
Description
This repository contains replication materials for the manuscript “Who Bears the Hidden Time Cost? Children’s Water and Firewood Collection, Socioeconomic Inequality, and Rainfall Variability in Bangladesh.” The study examines children’s participation in water collection and firewood collection, the weekly time burden associated with these unpaid household activities, socioeconomic inequalities in that burden, and its relationship with short-run rainfall variability in Bangladesh. The empirical analysis combines the Bangladesh Multiple Indicator Cluster Surveys (MICS) 2019 and 2025 with district-level precipitation information derived from the Climate Hazards Center InfraRed Precipitation with Station data (CHIRPS). The analysis uses the MICS selected-child age 5–17 modules together with relevant household characteristics. Child-level outcomes include participation in water collection and firewood collection, weekly hours spent on each activity, their combined weekly time burden, and indicators of relatively high time burdens. Socioeconomic gradients are examined using household wealth information when available. Rainfall exposure is linked to the timing and district of each child interview. With daily CHIRPS data, the preferred exposure measure is rainfall accumulated over the seven days ending on the day before the interview, standardized relative to district-by-calendar-month rainfall climatology. The workflow also constructs lower- and upper-tail rainfall indicators and alternative 14-day and 30-day exposure measures for robustness analysis. The statistical workflow includes descriptive analyses, wealth-gradient comparisons, multivariable regression models, heterogeneity analyses, mechanism-oriented specifications, and rainfall-exposure robustness checks. The analysis is observational; estimated relationships between rainfall variability and children’s collection activities or time burdens should therefore be interpreted as conditional associations rather than automatically as causal effects. The public replication archive contains Stata 14-compatible harmonization and analysis code, input-variable schemas, source-provenance documentation, aggregate-output procedures, privacy-preserving diagnostics, reproducibility documentation, and a file-integrity manifest. Respondent-level Bangladesh MICS 2019 and 2025 microdata and locally generated child-level analytical datasets are not redistributed. Full execution requires authorized MICS inputs and a district-level CHIRPS rainfall file matching the documented schema. The package does not reconstruct the upstream geospatial extraction from native CHIRPS imagery to district-level rainfall data.
Files
Steps to reproduce
Download and extract the replication ZIP while preserving the folder structure and read README.txt before running the analysis. The workflow requires Stata 14.0 or later. Bangladesh MICS 2019 and 2025 respondent-level files are not redistributed. Researchers wishing to reproduce the complete analysis must obtain authorized selected-child age 5–17 and household files through the official MICS data-access process. For each survey round, place the selected-child file (fs.sav, fs.dta, or fs.csv) and household file (hh.sav, hh.dta, or hh.csv) in the corresponding 00_raw_data/MICS2019/ and 00_raw_data/MICS2025/ folders. A district-level CHIRPS rainfall file must also be placed in 00_raw_data/climate/. The preferred input is chirps_bgd_district_daily.csv; the supported monthly fallback is chirps_v3_district_month_2015_2025.csv. Accepted variables and file structures are documented in 05_documentation/RESTRICTED_INPUT_SCHEMA.csv and 05_documentation/CLIMATE_INPUT_SCHEMA.csv. Open Stata, set the current working directory to the extracted package root, and run do "00_RUN_ALL.do". The workflow harmonizes the two MICS rounds, constructs child water- and firewood-collection time measures, merges relevant household characteristics, prepares rainfall exposures, links rainfall to interview timing, and generates descriptive, socioeconomic-gradient, regression, heterogeneity, mechanism and robustness outputs. Locally generated respondent-level analytical files are written to 02_prepared_data/ and are not part of the public deposit. Aggregate tables, figures and diagnostics are generated under 03_outputs/. Users should review the merge diagnostics before interpreting the statistical outputs.
Institutions
- Jagannath UniversityDhaka Division, Dhaka