I am doing a difference-in-differences estimation to see the effect of a parental leave reform on mothers' and fathers' wages. However, I am not sure that I specify the treatment and control groups as well as the time dummy correctly. What I want to specify in the end is
Code:
xtreg log(wage) i.treatment##i.post, robust
Code:
xtreg log(wage) i.treatment##i.post, *controls*, *fixed effects*, fe robust
The reform was implemented October 15th 1997. I have data running from 1994 to 2007. First, I created a time dummy equal to one if year is larger or equal to 1997:
Code:
gen post = (year>=1997)
I tried to respecify the treatment variable as
Code:
gen treatment = 1 if birthdate>tc(15oct1997) & birthdate<=(31dec1997) replace treatment = 0 if if birthdate>=tc(01au1997) & birthdate<=(15oct1997) replace treatment = 1 if birthdate>tc(15oct1996) & birthdate<=(31dec1996) replace treatment = 0 if if birthdate>=tc(01au1996) & birthdate<=(15oct1996)
Do you have any ideas on how I could specify the control and treatment groups correctly as I don't think it is correct right now?
Best,
Kamilla
0 Response to Specification of treatment and control group in diff-in-diff
Post a Comment