korenmiklos / expat-analysis

Analysis for Koren, Miklós and Álmos Telegdy, "Expatriate Managers and Firm Performance."

Requires confidential micro data to be placed in the input folder, like so:

input/balance-small/balance-sheet-1992-2016-small.dta
input/manager_position/pos5.csv
input/motherlode/branch.csv
input/motherlode/hq.csv
input/motherlode/manage.csv
input/motherlode/own.csv
input/motherlode/site.csv

Then run make install to install the beautiful 538 graph schemes. After that, make does all the data cleaning and runs all the analysis, like so:

STATA = stata -b do
output/estimate.log: temp/analysis_sample.dta estimate.do event_study.do regram.do
	$(STATA) estimate
temp/analysis_sample.dta: input/balance-small/balance-sheet-1992-2016-small.dta temp/firm_ceo_panel.dta variables.do
	$(STATA) variables
temp/firm_ceo_panel.dta: temp/manager_panel.dta firm_panel.do fill_in_ceo.do
	$(STATA) firm_panel
temp/manager_panel.dta: temp/managers.dta temp/positions.dta manager_panel.do 
	$(STATA) manager_panel
temp/managers.dta temp/positions.dta: input/manager_position/pos5.csv input/motherlode/manage.csv read_data.do
	$(STATA) read_data
install:
	stata -b ssc install g538schemes, replace all

Intermediate files after data cleaning, sample and variable definition etc are stored in the temp folder.

Regression output is save as "regression grammar," see regram.do (at some point I will document this as a package). Regressions are saved into tables, columns and rows within columns. Each cell has a point estimate, standard deviation, t-stat and p-value saved in YAML format. Just see output/regression/baseline_intensity/columns/exporter.yaml, for example. Having these numbers, you can create the table you want in the format you want.