Stata: kernel density with median line

* Kernel density plot

* realearnings base month is December 2016.
gen earnings202002 = realearnings/100*115.2
* now the base month is February 2020.
gen learnings202002 = asinh(earnings202002)
label variable learnings202002 "Monthly earnings (Feb 2020 rands)"

* use bracketweight for earnings analysis (see the PALMS User Guide)

capture kdens
if _rc == 199 {
ssc install moremata
ssc install kdens
}

summ earnings202002  [aw=bracketweight], detail
scalar MEDIAN = round(r(p50))
scalar lMEDIAN = log(MEDIAN)
scalar TOTALMEAN = round(r(mean))
scalar lTOTMEAN = log(TOTALMEAN)
kdens learnings202002 if earnings202002 > 40 & earnings202002 < 300000 [pw=bracketweight], ci name("total", replace) xline(`=scalar(lTOTMEAN)', lc(lime) lp(dash)) xline(`=scalar(lMEDIAN)', lc(blue)) note("Dashed light-green line shows mean monthly earnings (R`=scalar(TOTALMEAN)')." "Solid blue line shows median monthly earnings (R`=scalar(MEDIAN)')." "Source: QLFS 2017 Q4") xlabel(4.605 "100" 6.908 "1 000" 9.21 "10 000" 11.513 "100 000")  scale(1.1) title("Total earnings distribution in SA (2017 Q4)") // -kdens- uses the optimal bandwidth, and can generate the confidence intervals automatically
graph export "QLFS2017Q4 total earnings distribution.pdf", replace