Hello all,
I have a panel data set (long format) which contains companies and corresponding variables over a period of 2000-2020. It is about the introduction of a law on country level and the impact of this law on CEO compensation. The adoption of the law takes place in different years in the countries, which makes the simple DiD (canonical) approach not work. Using the paper by Callaway and Sant'Anna (2021), I have tried to apply the DiD for multiple time periods. However, the implementation in Stata causes me problems, since I have only known the simple DiD approach so far:
gen DiD = treated*t
reg fte treated t DiD
where in my case the Treatment Group consists of the firms located in countries where the law was introduced in the time period. The Control Group consists of firms in countries without the law throughout the period. The dependent variable (CEO compensation = Compensation) indicates as a dummy variable whether there has been a change in the structure of compensation. Thus, it has the value 1 for the company and the year in which the adjustment has taken place or is maintained. In addition, there is the variable LawIndicator, which is 1 for the company and the year if the law was introduced or applies and 0 otherwise.
Accordingly, in canonical format (only 2 periods with pre and post treatment) the application would be:
gen DiD = TreatmentGroup*Year
reg Compensation TreatmentGroup Year DiD
Thus the question is whether the basic command in Stata can be adapted to the multiple time periods or whether one should work with xtset/xtreg. I appreciate any feedback! Thank you
I have a panel data set (long format) which contains companies and corresponding variables over a period of 2000-2020. It is about the introduction of a law on country level and the impact of this law on CEO compensation. The adoption of the law takes place in different years in the countries, which makes the simple DiD (canonical) approach not work. Using the paper by Callaway and Sant'Anna (2021), I have tried to apply the DiD for multiple time periods. However, the implementation in Stata causes me problems, since I have only known the simple DiD approach so far:
gen DiD = treated*t
reg fte treated t DiD
where in my case the Treatment Group consists of the firms located in countries where the law was introduced in the time period. The Control Group consists of firms in countries without the law throughout the period. The dependent variable (CEO compensation = Compensation) indicates as a dummy variable whether there has been a change in the structure of compensation. Thus, it has the value 1 for the company and the year in which the adjustment has taken place or is maintained. In addition, there is the variable LawIndicator, which is 1 for the company and the year if the law was introduced or applies and 0 otherwise.
Accordingly, in canonical format (only 2 periods with pre and post treatment) the application would be:
gen DiD = TreatmentGroup*Year
reg Compensation TreatmentGroup Year DiD
Thus the question is whether the basic command in Stata can be adapted to the multiple time periods or whether one should work with xtset/xtreg. I appreciate any feedback! Thank you
Comment