Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Propensity Score Matching, fixing date variable when matching using psmatch2


From   Austin Nichols <austinnichols@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Propensity Score Matching, fixing date variable when matching using psmatch2
Date   Thu, 20 Oct 2011 10:47:27 -0400

Mario Reinhold <Mario.Reinhold@mailbox.tu-dresden.de>:
You want an exact match on year and a close match on ps, which has
been discussed on the list many times.  One way to do it:

g fake=year*10+ps
psmatch2 unmplyd, outcome(wage) pscore(fake) ate cal(0.0001)

I hope you are conditioning on multiple years of lagged employment and
wages (wage histories) when creating ps.
Also, the mean is only one moment of interest for reemployed persons;
see e.g. http://www.jstor.org/stable/1935837
and the difference in means may well vary across years and macro conditions.

On Thu, Oct 20, 2011 at 4:50 AM, Mario Reinhold
<Mario.Reinhold@mailbox.tu-dresden.de> wrote:
> For my diploma thesis I want to conduct Propensity Score matching. I try and
> extimate the wage reduction of persons that became unemployed and find a new
> job then after. In order to control for the macroeconomic circumstances in
> which a person finds a new job, I would like to 'fix' the year of the
> treated and untreated persons when matching, e.g. the age or the education
> does not have to be necessarily the same when matching, however the 'year'
> observed must be the same. One way would be to create subsamples, that is
> restricting the estimation to one certain year (or period) but that is not
> really what I want as I would have to conduct the same estimation 20 times
> and it would only tell me 20 times how the situation in year t=1 to t=20
> which is not sensible if I want to estimate a generalised effect over the 20
> years observed.
>
> The statistical model hence is:
> ATT= E[y_it|D_i=1, P(x)] - E[y_it|D=0, P(x)]
>
> with
> t... date
> i... individual
> y... outcome
> D... treatment
>
> instead of:
>
> ATT= E[y_i|D_i=1, P(x)] - E[y_i|D=0, P(x)]
>
> Is there a way to conduct this analysis using psmatch2?
> So far, my command is:
> psmatch2 unmplyd, outcome(wage) pscore(ps) ate cal(0.0001)
>
> where ps is estimated beforehands using probit.
>
> I am sorry but I could not find anything related here on statalist.
> I would be more than happy if I could receive some help.
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   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   |   Site index