CSDiD Counterfactual?

[Mansour Mohseni]

Hi
I have a question regarding time-variant DiD models such as CSDiD. In a setting where treatment could happen at different times (staggered setting), we can estimate ATT I was wondering whether we can get a counterfactual estimation as well, for example, what if a treatment happened sooner? I appreciate your guidance.

[FernandoRios]

You need to modify your gvar for that case

[Mansour Mohseni]

You need to modify your gvar for that case

Dear Dr. Rios-Avila
Thank you so much for your time and response.

So you mean changing the time that subjects were treated artificially and reestimating the ATT? I deeply appreciate it if you please provide additional references or papers about this approach.

[FernandoRios]

It’s nothing novel
if a group was treated in 2000 assume it got treated in 1998
its a manual way of doing an anticipation test

[Mansour Mohseni]

It’s nothing novel
if a group was treated in 2000 assume it got treated in 1998
its a manual way of doing an anticipation test

Dear Dr. Rios-Avila

Can we interpret the results of such an approach as a what-if analysis? Like saying, what if the policymaker had intervened sooner?