#delimit;
clear;
set mem 1000m;

use ebenstei_econ_berkeley_edu_001.dta;

***********;
* Renaming ;
***********;

rename ethncn race;
rename provcn province;
rename prefeccn prefecture;

gen cdied=chborn-chsurv;

*********************;
* Years of education ;
*********************;

gen yrsed=.;
replace yrsed=. if educchn==0 & year==1982;
replace yrsed=0 if educchn==1 & year==1982;
replace yrsed=3 if educchn==2 & year==1982;
replace yrsed=6 if educchn==3 & year==1982;
replace yrsed=9 if educchn==4 & year==1982;
replace yrsed=12 if educchn==5 & year==1982;
replace yrsed=16 if educchn==6 & year==1982;

**************;
* Birth month ;
**************;

gen month=0;

********************************;
* Relationship / Marital Status ;
********************************;

rename relate relold; gen relate=.;
replace relate=1 if relold==1000 & year==1982;
replace relate=2 if relold==2100 & year==1982;
replace relate=6 if relold==3000 & year==1982;
replace relate=8 if relold==4100 & year==1982;
replace relate=3 if relold==4200 & year==1982;
replace relate=5 if relold==4600 & year==1982;
replace relate=10 if relold==4910 & year==1982;
replace relate=12 if relold==5000 & year==1982;

rename marst marstold; gen marst=.;
replace marst=1 if marstold==100 & year==1982;
replace marst=2 if marstold==210 & year==1982;
replace marst=3 if marstold==330 & year==1982;
replace marst=4 if marstold==400 & year==1982;

********************************;
* Relationship / Marital Status ;
********************************;

* The census was carried out july 1st, 1982;

gen married=marst==2;
gen working=empstat==1000;
gen birthyear=1982-age;

**********;
* Weights ;
**********;

gen perwt=1000;

****************;

replace age=. if age>500;
drop if age==.;

tostring serial,generate(sernum);
capture drop serial;
egen serial=concat(sernum year), punct("-");

gen mypref=prefecture;
tostring mypref, replace;
rename prefecture oldprefecture;
rename mypref prefecture;

*keep year
/* Region */
province prefecture oldprefecture geotypcn
/* Individual and family identifiers */
serial pernum perwt
/* Family relationship variables */
momloc poploc relate relold
/* Core demographic variables */
age month sex married working marst marstold chborn chsurv cdied
/* Ethnicity */
race
/* Education and income */
lit edattain educchn empstat yrsed regstat occisco occ indgen ind;
compress;
sort serial pernum;

save ~/data/china/china1982/china1982, replace;
sort serial pernum;
keep in 1/10000;
save ~/data/china/china1982/china1982_small, replace;