clear

*Create paths to folders 
global data "C:\Users\aapo.kivinen\Dropbox (Aalto)\PhD\teaching\Principles_of_Economic_Analysis\aapo\session two\data"
global output "C:\Users\aapo.kivinen\Dropbox (Aalto)\PhD\teaching\Principles_of_Economic_Analysis\aapo\session two\output"

*Import data set to Stata
import delimited "$data\vaestolaskenta.csv", clear

keep vuosi id ika isa_ika tuloluokka yo sp isa_amas

*Change variable names to english 
rename vuosi year 
rename ika age
rename tuloluokka income
rename isa_ika fathers_age
rename yo high_school
rename sp sex
rename isa_amas fathers_occupation

*Label sex variable 
label define sex 1 "male", add
label define sex 2 "female", add
label values sex sex

*Label education variable 
label define high_school 0 "No academic high school degree", add
label define high_school 1 "Has academic high school degree", add
label values high_school high_school

*Extend fathers occupation to all observations
egen aux = max(fathers_occupation), by(id)
replace fathers_occupation = aux 
drop aux 

label define fathers_occupation 1 "Employed", add
label define fathers_occupation 2 "Entrepreneur", add
label values fathers_occupation fathers_occupation

*Calculate fathers age when child is born
replace fathers_age = fathers_age-age
egen aux = max(fathers_age), by(id)
replace fathers_age = aux 
drop aux 

*Create age group variable 
global bin = 10
global binmax = 70

gen age_group = . 
replace age_group = ${bin}*ceil(age/${bin})
replace age_group = ${binmax}+1 if age> ${binmax}

label define age_group 10 "0-10", add
label define age_group 20 "10-20", add
label define age_group 30 "20-30", add
label define age_group 40 "30-40", add
label define age_group 50 "40-50", add
label define age_group 60 "50-60", add
label define age_group 70 "60-70", add
label values age_group age_group

save "$data\cleaned_data.dta", replace