generate treated_start date

[Yi Zi]

Hi,
I intend to do a time-varying diffindiff analysis, and I am a bit confused about how to generated a treated_start date before generating the period variable. The following is an example:

Id	Year	Billion	Event	Treated_i	Period_i	Post_extreme_it	Treated_start
1	2010	0	0	1	0	0	.
1	2011	0	0	1	0	0	.
1	2012	3	1	1	1	1	2012
1	2013	1	0	1	1	1	.
1	2015	1	0	1	1	1	.
1	2016	2	1	1	1	1	2016

so the treated_start for this treated group should be all 2012. I am wondering how to do this? I have already wrote other commands as follows:
gen event = 0

replace event = 1 if extreme_type = hurricane & billion > 1

gen treated_start=.





gen period =1 if year >= treated_start

gen treated = 1 if treated_start !=.

gen Post_extreme = treated*period

Thank you so much for your help.

[Fei Wang]

* Example generated by -dataex-. For more info, type help dataex
clear
input float(Id Year Event)
1 2010 0
1 2011 0
1 2012 1
1 2013 0
1 2015 0
1 2016 1
end

bys Id (Year): gen first_event = sum(Event)
replace first_event = . if !(Event==1 & first_event==1)
bys Id (Year): egen Treated_start = total(Year*first_event), missing
drop first_event

[Yi Zi]

Thanks a lot!