Regression after PSM and DD mathcing

Hi all,

I’ve been working with quasi-experimental methods, but have not been able to solve these naïve doubts.

I’m implementing PSM for my subset1 and DD matching for my subset2.

Let’s assume y is the outcome, t is the treatment and x1/x2 are the covariates. The regression command would be:

reg y x1 x2 t … for PSM
xtreg y year t t_year … for DD matching
(t_year=interaction variable of treatment and time)

The questions are:

How to “tell” the regression that observations in my data are matched? While some use the variable _weight (created by command pscore) as a frequency weight, other dont add anything to the above commands. Are there other options?

Thanks!
Wil