Targeting performance of Bangladesh's social safety net against asset-based and multidimensional poverty - evidence from MICS7 Bangladesh 2025

Description

This dataset is the R replication package for the article "Targeting performance of Bangladesh's social safety net against asset-based and multidimensional poverty: eligibility-adjusted evidence from the 2025 Multiple Indicator Cluster Survey". The study compares receipt of Bangladesh's core social safety-net programmes with the MICS asset (wealth) index and a ten-indicator Alkire-Foster multidimensional poverty index (MPI) for 61,207 households, using design-based inference, headcount-matched comparisons, eligibility-adjusted programme analysis, rank-based metrics (AUC, concentration indices), counterfactual allocations, district-level decomposition and survey-weighted logistic models with average marginal effects. The package contains: (1) a master script (run_all.R) and numbered R scripts (code/) that read the official MICS7 SPSS files, construct all variables, and reproduce every table, figure and number in the manuscript and Supplementary Appendix; (2) the authors' aggregate result tables (CSV) and figures (PNG) in expected_output/, which the code compares against automatically; (3) a README mapping each manuscript and appendix table/figure to its output file. The MICS7 Bangladesh 2025 microdata are NOT included, because the MICS terms of use prohibit redistribution. They are available free of charge to registered users from UNICEF at https://mics.unicef.org/surveys (Bangladesh MICS 2025 SPSS Datasets: hh.sav, hl.sav, wm.sav, ch.sav). Only aggregate statistics are shared here.

Steps to reproduce

1. Download "Bangladesh MICS 2025 SPSS Datasets" from https://mics.unicef.org/surveys (free registration) and copy hh.sav, hl.sav, wm.sav and ch.sav into the data/ folder. 2. Install R (>= 4.1; results produced with R 4.3.3). Open Bangladesh_SSN_Targeting_Replication.Rproj in RStudio, or open R in the package folder. 3. Run source("run_all.R") (or "Rscript run_all.R" in a terminal). Missing CRAN packages (haven, survey, marginaleffects, pROC, dplyr, tidyr, purrr, tibble, readr, stringr, ggplot2, patchwork, scales) are installed automatically. 4. Tables are written to output/tables, figures to output/figures and logs to output/logs (about 15-30 minutes). The final step reports whether each table matches expected_output/. See README.md for the table/figure map and methodological settings.

Institutions

• Khulna University

  Khulna Division, Khulna

Categories

Licence