clear all
cd data
use observation
sort observeid
save, replace
use occupation
rename occ_sobserveid observeid
sort observeid
merge m:1 observeid using observation, keepusing(obs_entry_date)
keep if _merge ==3
gen year = year( obs_entry_date/86400000)
drop if year < 2004
gen h = halfyear(obs_entry_date/86400000)
gen round = yh(year,h)
sort individid round
gen period = round[_n+1]-round
replace period = 109-round if individid!=individid[_n+1]
expand period
replace round = round-89
sort individid round
by individid round: gen p= _n-1
replace round = round + p
drop if round > 19
drop h period p _merge  obs_entry_date
tostring round, generate(rounds)
replace rounds = "0" + rounds if strlen(rounds) == 1
gen individkey = individid +  rounds
sort individkey
duplicates drop individkey, force
drop  occ_episodeid observeid occ_eobserveid
save rOccupation, replace