This repository contains all the program codes developed in the "Distributional Impact Analysis: Toolkit and Illustrations of Impacts Beyond the Average Treatment Effect", by Guadalupe Bedoya (World Bank), Luca Bittarello (Northwestern University), Jonathan Davis (University of Chicago) and Nikolas Mittag (CERGE-EI), available at the World Bank Policy Research Working Paper Series

To replicate the simulations in the toolkit and the analysis in Part 7, start on /estimation/master.do. This master file sets global paths, sets the current directory and loads auxiliary programs.

All simulated data are created by user-written commands in /programs/simulation_programs.do. The simulation results in the toolkit are run by /estimation/simulations/simulations.

Five files are available under /estimation/brazil/:

• brazil_setup.do sets up the data (the baseline dataset is available from AEJ-EP).
• brazil_2.do replicates Subsection 7.1.2 (treatment effects on summary statistics, weighted and unweighted).
• brazil_3.do replicates Subsection 7.1.3 (bounds on features of the distribution of treatment effects).
• brazil_4.do replicates Subsection 7.1.4 (conditional effects).
• brazil_5.do replicates Subsection 7.1.5 (power calculations). Adjust line 11 to replicate the different columns of Table 6.

These files involve bootstrapping, which may be lengthy. The folder /results/bootstrap/ contains our bootstrap samples, which you can use to replicate our tables without running the algorithms.

Three files are avaiilable under /estimation/gambia/:

• gambia_setup.do sets up the data (provided by M. Blimpo, D.K. Evans and N. Lahire (2015): WB Policy Research WP 7238).
• gambia_analysis.do replicates all of Section 7.2.
• gambia_simulations.do replicates the deconvolution estimates using simulated data designed to mimic our sample from The Gambia. (This file is called by /estimation/simulations/simulations.do.)

These programs are available under /programs/:

• dte.do computes changes the distribution of marginal outcomes and its features. dte.sthlp documents the syntax (from Stata, run help dte).
• bscr.do computes simultaneous confidence regions and critical values from a bootstrap sample of estimators according to Romano and Wolf (2010). bscr.sthlp documents the syntax (from Stata, run help bscr).

If you are familiar with GitHub, please feel free to fork the repository and submit a pull request with your suggested edits.

An easy but still very efficient way to provide any feedback on the code in this repository is to create an issue in GitHub. You can read issues submitted by other users or create a new issue in the top menu below worldbank/DIA-toolkit at https://github.com/worldbank/DIA-toolkit. While the word issue has a negative connotation outside GitHub, it can be used for any kind of feedback. If you have identified a bug or enhancement, please create a new issue to let us know. Your feedback is very important to us. Please read existing issues to check whether someone else has made the same suggestion or reported the same error before creating a new issue.

This software is released under the Creative Commons Attribution 4.0 International license. You can view a license summary here.

Suggested citation:

Bedoya, Guadalupe, Luca Bittarello, Jonathan Davis, and Nikolas Mittag. "Distributional Impact Analysis: Toolkit and Illustrations of Impacts Beyond the Average Treatment Effect." World Bank Policy Research Working Paper Series, No. WPS 8139. Washington, D.C. (July 2017)