Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: "Difference-in-differences" program needed


From   "Nils Braakmann" <nilsbraakmann@googlemail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: "Difference-in-differences" program needed
Date   Mon, 26 Nov 2007 08:41:07 +0100

Hi Pär,

(short answer) you can use either -regress- or -xtreg-, depending on
whether you have repeated cross-section or panel-data.

(longer answer) You want to estimate
E[Y|t=1, d=1] - E[Y|t=0,d=1] - (E[Y|t=1, d=0] - E[Y|t=0,d=0]),
where t=1 marks the post-treatment period, d=1 marks the treatment
group and Y is the outcome of interest.

This setup can be written in regression form as
(repeated cross section) y_i = alpha + theta*t_i + delta*d_i +
tau*(t_i*d_i) + epsilon_i or
(panel case) y_it = alpha_i + theta*t_it + delta*d_it +
tau*(t_it*d_it) + epsilon_it,
where tau is the effect you want to estimate (see e.g. Meyer (1995) or
Angrist, Krueger (1999), p. 1299, full references at end of text).
Note also that you can control for differences in observable variables
by adding "beta*X_i" (cross section) or "beta*X_it" (panel data) to
the above equations.

In Stata, this can be set up as
(repeated cross section) regress y time_dummy group_dummy
interaction_time_group controls or
(panel) xtreg y time_dummy group_dummy interaction_time_group controls, fe.

Hope this helps.
Best regards,
Nils

Literature:
Bruce D. (1995): Natural and Quasi-Experiments in Economics. In:
Journal of Business and Economic Statistics Jg. 13, Nr. 2, S. 151-161
Angrist, Joshua D. and Alan B Krueger (1999): Empirical Strategies in
Labor Economics. In: Ashenfelter, Orley and David Card (eds.): Hanbook
of Labor Economics, Volume 3a, Elsevier: Amsterdam et. al, p1277-1366.



On Nov 23, 2007 2:30 PM, Pär Sjölander <Par.Sjolander@ihh.hj.se> wrote:
> Hi!
> I would like to know if threre is a Stata program (e.g. that works like the psmatch2 stata module) where you can get the difference-in-differences (se e.g. Heckman).
>
> (Or is it necessary that I do the propensity score matching first, and then perform the diff-in-diffs? Or is there an existing full program where I can get the difference-in-differences automatically, just as I get the ATT in psmatch2.)
>
> Grateful for help!
> /Pelle
>
>
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index