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"

*To make nicer graphs 
*ssc install grstyle (Run this if you have not downloaded it yet)
set scheme s2color
grstyle init
grstyle set plain

*Import data set to Stata
use "$data\cleaned_data.dta", clear

*Plot the cumulative distribution function of fathers age 

ssc install distplot

distplot fathers_age if age_group == 10, xtitle(Fathers age)
graph export "$output\c_dist.pdf", replace  

*Plot the density function of fathers age 
twoway (kdensity fathers_age), xtitle(Fathers age)
graph export "$output\p_dist.pdf", replace  

xtile father_age_group=fathers_age, nq(10)
egen min = min(fathers_age), by(father_age_group)
egen max = max(fathers_age), by(father_age_group)

graph bar (mean) max, over(father_age_group) ytitle(Years) title(Fathers age by decile)  blabel(bar)
graph export "$output\percentile.pdf", replace  

drop if age<11

preserve 

collapse (mean) income, by(age_group sex high_school)
local sc_opt "c(l)"
local opt "xlab(20(10)70, valuelabel) xtitle("Age group") ytitle("Income decile") legend(lab(1 "Not Academic") lab(2 "Academic"))"
twoway 	(scatter income age_group if high_school==0 & sex==1, `sc_opt') ///
		(scatter income age_group if high_school==1 & sex==1, `sc_opt'), `opt' subtitle(A. Men) name(wages_men, replace)  
twoway 	(scatter income age_group if high_school==0 & sex==2, `sc_opt') ///
		(scatter income age_group if high_school==1 & sex==2, `sc_opt'), `opt' subtitle(B. Female) name(wages_women, replace)  
graph combine wages_men wages_women, ycommon name(combined1, replace)
graph export "$output\development.pdf", replace  

restore 

preserve 

* Growth in groups 

forvalues yy=20(10)70{
		
	egen mean_income_`yy' = mean(income) if age_group==`yy', by(high_school sex)
			
}
	
collapse (min) mean_income_*, by(high_school sex)
	
gen change = mean_income_50-mean_income_30
	
graph bar (mean) change, over(sex) over(high_school) blabel(bar, format(%9.2g)) ytitle(Change) title("Change in income decile from age group 20-30 to 40-50""by degree and sex", size(medium))
graph export "$output\change.pdf", replace  
	
restore