What if the same cohort is treated twice?

[azev77]

Suppose the same cohort receives two treatments on two different dates:
-example: the same state receives two minimum wage hikes
How can your estimator be applied in this case?

The only way I can imagine:
lets say unit i==1 is treated first at t==4 and again at t==8
In this case I'd create two artificial units, each treated once, & use Never Treated as controls:
i==1a only treated once at t==4
i==1b only treated once at t==8

There's gotta be a better way?!

[borusyak]

Hi,
I'm not sure viewing changes in minimum wages as a binary treatment is very intuitive. However, if you want to do it that way, you can specify the appropriate estimands via the wtr() options, e.g. separate compute the average effects after one treatment only vs. after being treated twice. I don't think creating artifical units is ever a good idea.

Best,
Kirill