README FILE for code used in Shared Custody Increases After Divorce In USA


The source of the data used is from the Current Population Survey's Child Support Supplement (CPS-CSS) 1994-2014. 

There are three folders in the zip file. The folder questionnaire contains all the questionnaires. The folder data_compilation has the codes authored by Jean Roth at NBER which uses the raw data files to generate Stata datasets. The folder codes cleans and analyses the data.

DOWNLOADING THE DATA 
1. Go to https://www.nber.org/research/data/current-population-survey-cps-supplements-marchapril-match-child-support
2. Download the .zip files for years 1992-2014. These are the raw data files which will get downloaded in .dat format. 
3. Run the do files with names cpsaprxx.do (in the folder named data_compilation) where the xx represents the last two digits of the year. For example running cpsapr08.do will use the 2008 .dat raw data file to compile the data in Stata's .dta format and generate cpsapr08.dta 

RUNNING ANALYSES
The codes for cleaning and analyzing the compiled CPS-CSS data in Stata format used in the paper are in the three STATA do files submitted: 
1. finalcleaning.do
	This file cleans the compiled CPS-CSS data to generate the data everdivorcedcustodial.dta which is the main file on which the analyses are done. This file begins with a package check code, to check whether the packages used in this file are installed in the user’s STATA.
2. analysis.do
	This file takes the cleaned data generated by filecleaning.do and generates the tables and graphs in the paper. This includes both descriptives and regressions (main and robustness). 
3. runsforrobustness.do
	This file contains the regression and simulation codes for the robustness checks. It is called on within analysis.do to run the robustness checks. 

The codes have one global directory path for all data and output of results. Users should note to change the global directory path, before running the codes. Codes are in STATA version 16.1.