MHE Data Archive

Data and programs from Mostly Harmless Econometrics

This page links some of the data sets and do files used to produce the estimates reported in my book with Steve Pischke, Mostly Harmless Econometrics.  The data sets are listed by original citation.  This list omits papers that I've contributed to.

Our thanks to the authors who prepared this material and have graciously agreed to have their data posted here.

Data sets and programs for papers that Angrist contributed to appear in the Angrist Data Archive.

This do file and these Stata data sets create the estimates report in Tables 3.3.2 and 3.3.3.  Data for this analysis are courtesy of Rajeev Dehejia.

Dehejia-Wahba.do

cps1re74.dta

cps3re74.dta

nswre74.dta

This link contains data and documentation for the full Lalonde sample:

http://www.nber.org/~rdehejia/nswdata.html

again, courtesy of Rajeev Dehejia.

The do file and data set linked below create the estimates in Table 8.2.1.  The data we used are from
http://www.heros-inc.org/data.htm
which has the public use version of the STAR data. The public use data doesn't have class identifiers, so these are imputed (based on school characteristics, experimental group, and teacher characteristics). Note that our analysis of these data uses the BRL and Moulton routines linked below.

krueger.do
webstar.dta

This excel spread sheet makes MHE Figure 5.2.3. In addition we've linked  the data and a STATA do file which produces the results for Table 3 in the original Pischke (2007) paper (these results are not in the book) and the numbers for Figure 5.2.3.

graderep.do
graderep.dta

The data used to make Figure 5.2.4 can be found in David Autor's data archive
along with data and programs to make the tables in his paper.

Documentation and data for this paper are linked below.  These data are the same as used in Chapter 2 of Card and Krueger's Myth and Measurement.  Data are courtesy of David Card.

njmin-readme.txt
njmin.zip

The data for Card (1992) are the same as those used in Chapter 4 of Card and Kruegers's Myth and Measurement.  The zip file containing data and programs for Chapter 4 of M&M comes courtesy of David Card.

chapt4.zip

This program runs the Monte Carlos discussed in Chapter 4 and creates Figures 4.6.1, 4.6.2, and 4.6.3 (bias of 2SLS):
mc_2sls.do

This program creates Table 8.1.1 (behavior of alternative covariance matrix estimators):
mc_robust.do

The zip file linked below includes Stata do files, help files, and other material to adjust standard errors for clustering using Biased-Reduced Linearization (BRL; Bell and McCaffrey, 2002) or a parametric Moulton (1986) correction factor.  Copy the do and help files into your stata working directory or use stata's net install command to access them remotely from a web page.
BRL_Moulton.zip

The brl.ado file has been written by Brigham Frandsen, who maintains an updated version fixing some bugs on his
web page.

Lee2008.zip

MHE p. 13, you know, the t-stats that go with the little unnumbered table at the top of the page

page13.do

nhis_13.dta