Wooldridge staggered TWFE with multi-valued discrete treatment (jwdid)

[Tim Kooijmans]

Hi all,

I am wondering if there are insights into whether the TWFE results of Jeff Wooldridge (2021) extend to a setting with a multi-valued discrete treatment variable.

If I understand correctly, the Wooldridge method captures variation in treatment intensity due to units being treated at different points in time, by identifying treatment cohorts (units treated at the same time). So intuitively it seems that in a setting where treatment intensity is also determined by units (i) receiving either full (f) or half (h) treatment [at different points in time (t)], we may be able to add a series of additional (half-treated) cohorts in a specification something like this:

Or could this be not so simple?

I would be interested in comparing Average Treatment Effects (ATTs) of units that got full treatment, versus units that got half treatment. In Stata, following the “estat simple” method of FernandoRios jwdid command, I could estimate the ATTs for all observations with the full treatment, and separately the ATTs of all the observations with half treatment?

If anyone has ideas about this, I would be very keen to hear.

Thank you,
Tim

[FernandoRios]

Yes you can do it.
A) get latest jwdid from ssc (check is version 2)
use your intensity as another control.
b) use option estat simple, over(intensity)
c) or estat event , ores(intensity==1/2)
See the references in helpfile for new features