csdid with individuals moving

[Reese Alexander]

I'm running a staggered event study with zip code treatment of individuals. I observe individuals each year from 2007 to 2019. Over the course of the sample a portion (~38%) of my sample move at least once between zip codes. When I use csdid outcomevar, ivar(individualID) time(year) gvar(treatment_year) long2 what effects could this have assuming that individuals are changing their year of treatment via moving? I'm using a fairly large (100k individuals and 30 treated zipcodes across 7 years) sample for this so I assume asymptotic assumptions are satisfied.

[FernandoRios]

It will still use the first year of treatment as the relevant treatment. This is defined by gvar.
In other words, individuals Year of treatment should be constant, and equal to the first year they were treated

[Reese Alexander]

awesome thank you for the response!