use "C:\Users\Asus\Documents\AUCA\Teaching\2022-2023\Econometrics I\Seminars\Seminar 6\Data\hh1a.dta" 
save "C:\Users\Asus\Documents\AUCA\Teaching\2022-2023\Econometrics I\Seminars\Seminar 6\Data\New data\hh1a_new.dta"
keep hhid pid age h102 h104 h108
bysort hhid: egen hhsize=count(pid)
bysort hhid: egen males_over_15=count(pid) if age>15 & h102==1
bysort hhid ( males_over_15): replace males_over_15= males_over_15[1]
replace males_over_15=0 if males_over_15==.
bysort hhid: egen child_under_5=count(pid) if age<5
bysort hhid ( child_under_5 ): replace child_under_5 = child_under_5[1]
replace child_under_5 =0 if child_under_5 ==.
bysort hhid: gen female_hh_head=1 if h102==2 & h104==1
bysort hhid ( female_hh_head ): replace female_hh_head = female_hh_head[1]
replace female_hh_head =0 if female_hh_head ==.
bysort hhid: gen married_hh_head=1 if h108==1 & h104==1
bysort hhid ( married_hh_head ): replace married_hh_head = married_hh_head[1]
replace married_hh_head =0 if married_hh_head ==.
drop pid age h102 h104 h108
duplicates drop
duplicates report hhid
save "C:\Users\Asus\Documents\AUCA\Teaching\2022-2023\Econometrics I\Seminars\Seminar 6\Data\New data\hh1a_new.dta", replace
use "C:\Users\Asus\Documents\AUCA\Teaching\2022-2023\Econometrics I\Seminars\Seminar 6\Data\hh4b.dta"
save "C:\Users\Asus\Documents\AUCA\Teaching\2022-2023\Econometrics I\Seminars\Seminar 6\Data\New data\hh4b_new.dta"
gen health_expenses=h403 if n4b==5 & h404==1 & h405==1
replace health_expenses=h403/12 if n4b==5 & h404==1 & h405==2
replace health_expenses=h403*47.01 if n4b==5 & h404==2 & h405==1
replace health_expenses=h403*47.01/12 if n4b==5 & h404==2 & h405==2