damiancclarke
commented
2 years ago At present, sdid is written to implement the estimation as described in the Arkhangelsky et al paper. As such, we only allow for the export of timing-specific graphs. Given the nature of the staggered rollout, having grouped time to treatment estimates is not necessarily trivial, as each time-specific adoption will have a different number of treatment lags and leads. So at present, we do not implement this. Perhaps if you are after something along these lines, you could look into papers such as "Synthetic Controls with Staggered Adoption" by Ben-Micheal et al: https://www.nber.org/system/files/working_papers/w28886/w28886.pdf
Best wishes, Damian
Hi. I am running sdid with staggered adoption of treatment. Is it possible to plot the estimate in one graph like the event study one? I mean the aggregate treatment effect for time -2, -1, 0, 1, 2,3,.... by the time to treatment?
The current one is very transparent, but it generates a lot of graphs if I have many different treatment timings.
thank you