Hi everyone,
I am running an event study model with firm level unbalanced panel data. The treatment happens at state level where some states are treated while other are not treated throughout the time period. There is also staggered policy implementation where states receive treatment at different times. I have two questions regarding the event study model that I am trying to run:
1. I am creating leads and lags variables using time till treat: lead1 takes a value 1 when it is one year after policy and 0 otherwise. This is also 0 when states are never treated. So this be the case or states that are never treated should have missing data?
2. Regarding the code for this:
Is this the correct way to do classic event study? I am planning to do the Sun and Abraham methods but for the baseline specification, should it be that or the one which gives missing values to never treated states? My code for that looks like this, where time_til_treat has missing values for never treated states.
Thank you.
I am running an event study model with firm level unbalanced panel data. The treatment happens at state level where some states are treated while other are not treated throughout the time period. There is also staggered policy implementation where states receive treatment at different times. I have two questions regarding the event study model that I am trying to run:
1. I am creating leads and lags variables using time till treat: lead1 takes a value 1 when it is one year after policy and 0 otherwise. This is also 0 when states are never treated. So this be the case or states that are never treated should have missing data?
2. Regarding the code for this:
Code:
areg y leads* lags* i.state i.year [aw= weight], absorb(firmid) cluster(state)
Code:
areg y ib18.time_til_treat i.state i.year [aw= weight], absorb(firmid) cluster(state)
Comment