I'm doing a DiD for a pension reform law that went into effect in 2016.
I am using micro data from national labor force surveys between 2013-2019, monthly data with PIDs to track individuals
So i want to know if treatment affects hours worked, but I also want to test sub groups for heterogenous effects on the hours worked, namely by age, sex, income, education

I have something like the following, after declaring my panel set:
Code:
xtset treated time, fe
Where:
Code:
treated = individual affected by reform
time = date after reform
did = treated*time
Code:
age= 1 young, 2 = prime, 3 = old
firm=1  is 499-999 employees, 2= 1000+
sex= 1 male, 0 female
inc=7 $10,000-$15,000 annual, 8= $15,000+
edu =1 highschool grad, 2=college grad, 3=university, etc
outcome variables
Code:
ln_hrs = log(weekly hrs).

Is this a reasonable way to proceed? What assumptions should I be testing next, parallel trends assumption against the two groups?
Also, I wanted to evaluate my control groups. Its my understanding the treatment and control should be as similar to each other as possible.
But I can't think of a way to separate control from sample without making a new data set

Thanks for any input