Event study for non-continuous time period

[Pier Han]

Dear Statalisters,

I am looking at the effect of a support program for provinces on provincial outcomes by using event study. I have a panel data over 4 time points, including 2012, 2014, 2016 and 2018. The event year ranges from 2013 to 2017. I have 2 questions.

1. Given my data structure, where the years are spaced by two-year intervals. I am not sure whether the way I create event_time is correct or not, especially for the cases where the event years are odd numbers, 2013, 2015, 2017.

* Example generated by -dataex-. For more info, type help dataex
clear
input str10 A str4 B str10 C
"event_year" "year" "event_time"
"2013"       "2012" "-1"        
"2013"       "2014" "0"         
"2013"       "2016" "1"         
"2013"       "2018" "2"         
"2014"       "2012" "-1"        
"2014"       "2014" "0"         
"2014"       "2016" "1"         
"2014"       "2018" "2"         
"2015"       "2012" "-3"        
"2015"       "2014" "-2"        
"2015"       "2016" "0"         
"2015"       "2018" "1"         
"2016"       "2012" "-2"        
"2016"       "2014" "-1"        
"2016"       "2016" "0"         
"2016"       "2018" "1"         
"2017"       "2012" "-3"        
"2017"       "2014" "-2"        
"2017"       "2016" "-1"        
"2017"       "2018" "0"         
end

event_year is the year at which the program started; year is the time of the sample

1. Shold I drop observations which started the program in 2017? Because for those observations, I only observe years before the event but I do not observe years after the event.
   
   Many thanks in advance.