search

Daniel-Pailanir / sdid

Synthetic Difference in Differences for Stata
GNU General Public License v3.0
72 stars 36 forks source link

Event study PLot #60

Closed anzonyquispe closed 8 months ago

anzonyquispe commented 12 months ago

Is it possible to construct event study plots for staggered treatment?

damiancclarke commented 12 months ago

Hello @anzonyquispe,

You would need to have some theory related to how you might group lags and leads to the policy if you wish to do event studies based on SDID (I presume you are referring to the event studies in the working paper). You could certainly do an event study for each adoption and present them separately, but would need to have some particular way to aggregate them (perhaps the methods discussed in Sun & Abraham, equation 28 could be an option). However, I unfortunately do not think there is theory yet that shows this is valid in the SDID setting. Best wishes, Damian