I attached my dataset for investigating purpose. A glance of the dataset is below:
Array
year1 is the real year for each observation, TYPE2 is specific code for each firm. treat1 equal to 1 if this firm belongs to control group, and 0 otherwise, post33 equals to 1 for the after event date (2019). post33treat1 is the interaction of treat1 and post33. FAT_w1 is the dependent variable.
Yesterday, when applying the code xtdidregress, I faced a problem that the min and max of the first treatment observed for each firms varies (due to missing variable)
Code:
. xtdidregress (FAT_w1) (post33treat1), group(TYPE2) time(year1) Number of groups and treatment time Time variable: year1 Control: post33treat1 = 0 Treatment: post33treat1 = 1 ----------------------------------- | Control Treatment -------------+--------------------- Group | TYPE2 | 274 264 -------------+--------------------- Time | Minimum | 2016 2019 Maximum | 2021 2021 -----------------------------------
I tried to conduct these codes to delete all observations of any TYPE2 that has the first post33treat1 =1 at year after 2019.
Code:
bysort TYPE2 post33treat1 (year): generate indicator=1 if _n==1 & post33treat1==1 egen treataf19 = total(indicator==1 & year1!=2019) , by(TYPE2) drop if treataf19
Code:
. xtdidregress (FAT_w1) (post33treat1), group(TYPE2) time(year1) Number of groups and treatment time Time variable: year1 Control: post33treat1 = 0 Treatment: post33treat1 = 1 ----------------------------------- | Control Treatment -------------+--------------------- Group | TYPE2 | 274 248 -------------+--------------------- Time | Minimum | 2016 2019 Maximum | 2021 2020 -----------------------------------
I also cross check whether there is any TYPE2 that has the first post33treat1 observation equalling to 1 associated with the year1=2020 but it returns 0. I am wondering what's wrong with my coding progress and how to fix it.
Code:
count if (indicator==1 & year1==2020) 0
Thanks in advance and best regards.
0 Response to Major concern about xtdidregress after checking
Post a Comment