/**** this file extract the hours worked and employment sample **/

cd H:\Giovanni\MyPapers\Peri_Ottaviano_aggregate\REDUX_2008\empirics\selection
capture log close
#delimit;
clear;

set memory 2g;
set matsize 400;
use "H:\ipums_data\state_data\ipums_60";

** sample selection of the supply construction;
* include self employed, people in school, but not non-civilian and convicts;
*** in the wage file we will also have more restrictive specifications;

* eliminate people in group quarters ;
drop if gq ==0 | gq==3 |gq==4 ; 

*select people in working age;
drop if age<18;

*working age is defined as above 16 years old (US dept of labor)
but for consistency with BGH we eliminate the 17 years old;

*select people who worked a positive amount of weeks;
drop if wkswork2==0; 
*we use only the weeks worked variable as it refers to last year;

* foreign-born/including US citizens born abroad;
generate byte foreign =(bpld>=15000);
replace foreign=0 if bpld==90011 | bpld==90021;

**male dummy;
gen byte male=(sex==1);

*define edu and exp codes;
generate byte edu=1*(educrec<=6) + 2*(educrec==7) + 3*(educrec==8) + 4*(educrec==9);

generate experience=age-17 if edu==1;
replace experience=age-19 if edu==2;
replace experience=age-21 if edu==3;
replace experience=age-23 if edu==4;
drop if experience<1 | experience>40;

generate exp=1*(experience>=1 & experience<=5) + 2*(experience>=6 & experience<=10) + 
3*(experience>=11 & experience<=15) + 4*(experience>=16 & experience<=20)
+5*(experience>=21 & experience<=25) + 6*(experience>=26 & experience<=30) 
+ 7*(experience>=31 & experience<=35) + 8*(experience>=36 & experience<=40);

** define hours and weeks worked ***;
generate hours=7.5*(hrswork2==1) + 22*(hrswork2==2) 
+ 32*(hrswork2==3) + 37*(hrswork2==4) + 40*(hrswork2==5) + 
44.5*(hrswork2==6) + 54*(hrswork2==7) + 70*(hrswork2==8) ;

generate weeks=6.5*(wkswork2==1) + 20*(wkswork2==2) + 33*(wkswork2==3) + 43.5*(wkswork2==4) + 
	48.5*(wkswork2==5) + 51*(wkswork2==6) ;

*supply in hours worked 4 groups;
gen howo=perwt*hours*weeks/2000;
gen howo_ma_us=howo*male*(1-foreign);
gen howo_ma_for=howo*male*(foreign);
gen howo_fe_us=howo*(1-male)*(1-foreign);
gen howo_fe_for=howo*(1-male)*(foreign);

*supply in bodycount 4 groups;
gen employment=perwt;
gen empl_ma_us=employment*male*(1-foreign);
gen empl_ma_for=employment*male*(foreign);
gen empl_fe_us=employment*(1-male)*(1-foreign);
gen empl_fe_for=employment*(1-male)*(foreign);

keep edu exp employment empl_ma_us empl_ma_for empl_fe_us empl_fe_for 
howo howo_ma_us howo_ma_for howo_fe_us howo_fe_for;

sort edu exp;
collapse (sum) employment empl_ma_us empl_ma_for empl_fe_us empl_fe_for 
howo howo_ma_us howo_ma_for howo_fe_us howo_fe_for, by(edu exp);
gen year=1960;

save supply_groups_60,replace;


**1970;

use "H:\ipums_data\state_data\ipums_70";


* eliminate people in group quarters ;
drop if gq ==0 | gq==3 |gq==4 ; 

*select people in working age;
drop if age<18;
** following BGH;

*** no need to select the older, as they are cut by experience limit;

*select people who worked a positive amount of weeks;
drop if wkswork2==0; 
*we use only the weeks worked variable as it refers to last year;


* foreign-born/including US citizens born abroad;
generate byte foreign=(citizen==2 | citizen==3);

**male dummy;
gen byte male=(sex==1);

*define edu and exp codes;
generate byte edu=1*(educrec<=6) + 2*(educrec==7) + 3*(educrec==8) + 4*(educrec==9);

generate experience=age-17 if edu==1;
replace experience=age-19 if edu==2;
replace experience=age-21 if edu==3;
replace experience=age-23 if edu==4;
drop if experience<1 | experience>40;

generate exp=1*(experience>=1 & experience<=5) + 2*(experience>=6 & experience<=10) + 
3*(experience>=11 & experience<=15) + 4*(experience>=16 & experience<=20)
+5*(experience>=21 & experience<=25) + 6*(experience>=26 & experience<=30) 
+ 7*(experience>=31 & experience<=35) + 8*(experience>=36 & experience<=40);

** define hours and weeks worked ***;
generate hours=7.5*(hrswork2==1) + 22*(hrswork2==2) 
+ 32*(hrswork2==3) + 37*(hrswork2==4) + 40*(hrswork2==5) + 
44.5*(hrswork2==6) + 54*(hrswork2==7) + 70*(hrswork2==8) ;

generate weeks=6.5*(wkswork2==1) + 20*(wkswork2==2) + 33*(wkswork2==3) + 43.5*(wkswork2==4) + 
	48.5*(wkswork2==5) + 51*(wkswork2==6) ;

*** supply in hours worked, 4 groups;
gen howo=perwt*hours*weeks/2000;
gen howo_ma_us=howo*male*(1-foreign);
gen howo_ma_for=howo*male*(foreign);
gen howo_fe_us=howo*(1-male)*(1-foreign);
gen howo_fe_for=howo*(1-male)*(foreign);

*** supply in bodycount, 4 groups;
gen employment=perwt;
gen empl_ma_us=employment*male*(1-foreign);
gen empl_ma_for=employment*male*(foreign);
gen empl_fe_us=employment*(1-male)*(1-foreign);
gen empl_fe_for=employment*(1-male)*(foreign);

keep edu exp employment empl_ma_us empl_ma_for empl_fe_us empl_fe_for
howo howo_ma_us howo_ma_for howo_fe_us howo_fe_for;

sort edu exp;
collapse (sum) employment empl_ma_us empl_ma_for empl_fe_us empl_fe_for
howo howo_ma_us howo_ma_for howo_fe_us howo_fe_for, by(edu exp);
gen year=1970;

save supply_groups_70,replace;


****1980;
use "H:\ipums_data\state_data\ipums_80";

* eliminate people in group quarters ;
drop if gq ==0 | gq==3 |gq==4 ; 

*select people in working age;
drop if age<18;
** following BGH;
**do not need to select the older, as they are cut by experience limit;

*select people who worked a positive amount of weeks;
drop if wkswork1==0; 
*we use only the weeks worked variable as it refers to last year;

* foreign-born/including US citizens born abroad;
generate byte foreign=(citizen==2 | citizen==3);

**male dummy;
gen byte male=(sex==1);

*define edu and exp codes;
generate byte edu=1*(educrec<=6) + 2*(educrec==7) + 3*(educrec==8) + 4*(educrec==9);

generate experience=age-17 if edu==1;
replace experience=age-19 if edu==2;
replace experience=age-21 if edu==3;
replace experience=age-23 if edu==4;
drop if experience<1 | experience>40;

generate exp=1*(experience>=1 & experience<=5) + 2*(experience>=6 & experience<=10) + 
3*(experience>=11 & experience<=15) + 4*(experience>=16 & experience<=20)
+5*(experience>=21 & experience<=25) + 6*(experience>=26 & experience<=30) 
+ 7*(experience>=31 & experience<=35) + 8*(experience>=36 & experience<=40);

** define hours and weeks worked ***;
generate hours=uhrswork;
generate weeks=wkswork1;

*** supply in hours worked, 4 groups;
gen howo=perwt*hours*weeks/2000;
gen howo_ma_us=howo*male*(1-foreign);
gen howo_ma_for=howo*male*(foreign);
gen howo_fe_us=howo*(1-male)*(1-foreign);
gen howo_fe_for=howo*(1-male)*(foreign);

*** supply in bodycount, 4 groups;
gen employment=perwt;
gen empl_ma_us=employment*male*(1-foreign);
gen empl_ma_for=employment*male*(foreign);
gen empl_fe_us=employment*(1-male)*(1-foreign);
gen empl_fe_for=employment*(1-male)*(foreign);

keep edu exp employment empl_ma_us empl_ma_for empl_fe_us empl_fe_for
howo howo_ma_us howo_ma_for howo_fe_us howo_fe_for;

sort edu exp;
collapse (sum) employment empl_ma_us empl_ma_for empl_fe_us empl_fe_for
howo howo_ma_us howo_ma_for howo_fe_us howo_fe_for, by(edu exp);
gen year=1980;

save supply_groups_80,replace;


*** 1990;
use "H:\ipums_data\state_data\ipums_90";

* eliminate people in group quarters ;
drop if gq ==0 | gq==3 |gq==4 ; 

*select people in working age;
drop if age<18;
** follow BGH;
*** do not need to select the older, as they are cut by experience limit;

*select people who worked a positive amount of weeks;
drop if wkswork1==0; 
*we use only the weeks worked variable as it refers to last year;

* foreign-born/including US citizens born abroad;
generate byte foreign=(citizen==2 | citizen==3);

**male dummy;
gen byte male=(sex==1);

*define edu and exp codes;
generate byte edu=1*(educrec<=6) + 2*(educrec==7) + 3*(educrec==8) + 4*(educrec==9);

generate experience=age-17 if edu==1;
replace experience=age-19 if edu==2;
replace experience=age-21 if edu==3;
replace experience=age-23 if edu==4;
drop if experience<1 | experience>40;

generate exp=1*(experience>=1 & experience<=5) + 2*(experience>=6 & experience<=10) + 
3*(experience>=11 & experience<=15) + 4*(experience>=16 & experience<=20)
+5*(experience>=21 & experience<=25) + 6*(experience>=26 & experience<=30) 
+ 7*(experience>=31 & experience<=35) + 8*(experience>=36 & experience<=40);

** define hours and weeks worked ***;
generate hours=uhrswork;
generate weeks=wkswork1;

*** supply in hours worked, 4 groups;
gen howo=perwt*hours*weeks/2000;
gen howo_ma_us=howo*male*(1-foreign);
gen howo_ma_for=howo*male*(foreign);
gen howo_fe_us=howo*(1-male)*(1-foreign);
gen howo_fe_for=howo*(1-male)*(foreign);

*** supply in bodycount, 4 groups;
gen employment=perwt;
gen empl_ma_us=employment*male*(1-foreign);
gen empl_ma_for=employment*male*(foreign);
gen empl_fe_us=employment*(1-male)*(1-foreign);
gen empl_fe_for=employment*(1-male)*(foreign);

keep edu exp employment empl_ma_us empl_ma_for empl_fe_us empl_fe_for
howo howo_ma_us howo_ma_for howo_fe_us howo_fe_for;

sort edu exp;
collapse (sum) employment empl_ma_us empl_ma_for empl_fe_us empl_fe_for
howo howo_ma_us howo_ma_for howo_fe_us howo_fe_for, by(edu exp);
gen year=1990;

save supply_groups_90,replace;

**** 2000;
use "H:\ipums_data\state_data\ipums_00";

* eliminate people in group quarters ;
drop if gq ==0 | gq==3 |gq==4 ; 

*select people in working age;
drop if age<18;
** following BGH;
*** do not need to select the older, as they are cut by experience limit;

*select people who worked a positive amount of weeks;
drop if wkswork1==0; 
*we use only the weeks worked variable as it refers to last year;

* foreign-born/including US citizens born abroad;
generate byte foreign=(citizen==2 | citizen==3);

**male dummy;
gen byte male=(sex==1);

*define edu and exp codes;
generate byte edu=1*(educrec<=6) + 2*(educrec==7) + 3*(educrec==8) + 4*(educrec==9);

generate experience=age-17 if edu==1;
replace experience=age-19 if edu==2;
replace experience=age-21 if edu==3;
replace experience=age-23 if edu==4;
drop if experience<1 | experience>40;

generate exp=1*(experience>=1 & experience<=5) + 2*(experience>=6 & experience<=10) + 
3*(experience>=11 & experience<=15) + 4*(experience>=16 & experience<=20)
+5*(experience>=21 & experience<=25) + 6*(experience>=26 & experience<=30) 
+ 7*(experience>=31 & experience<=35) + 8*(experience>=36 & experience<=40);

** define hours and weeks worked ***;
generate hours=uhrswork;
generate weeks=wkswork1;

*** supply in hours worked, 4 groups;
gen howo=perwt*hours*weeks/2000;
gen howo_ma_us=howo*male*(1-foreign);
gen howo_ma_for=howo*male*(foreign);
gen howo_fe_us=howo*(1-male)*(1-foreign);
gen howo_fe_for=howo*(1-male)*(foreign);

*** supply in bodycount, 4 groups;
gen employment=perwt;
gen empl_ma_us=employment*male*(1-foreign);
gen empl_ma_for=employment*male*(foreign);
gen empl_fe_us=employment*(1-male)*(1-foreign);
gen empl_fe_for=employment*(1-male)*(foreign);

keep edu exp employment empl_ma_us empl_ma_for empl_fe_us empl_fe_for
howo howo_ma_us howo_ma_for howo_fe_us howo_fe_for;

sort edu exp;
collapse (sum) employment empl_ma_us empl_ma_for empl_fe_us empl_fe_for
howo howo_ma_us howo_ma_for howo_fe_us howo_fe_for, by(edu exp);
gen year=2000;

save supply_groups_00,replace;


**** 2006;
use "H:\ipums_data\state_data\ipums_06";

* eliminate people in group quarters ;
drop if gq ==0 | gq==3 |gq==4 ; 

*select people in working age;
drop if age<18;
** following BGH;
*** do not need to select the older, as they are cut by experience limit;

*select people who worked a positive amount of weeks;
drop if wkswork1==0; 
*we use only the weeks worked variable as it refers to last year;

* foreign-born/including US citizens born abroad;
generate byte foreign=(citizen==2 | citizen==3);

**male dummy;
gen byte male=(sex==1);

*define edu and exp codes;
generate byte edu=1*(educrec<=6) + 2*(educrec==7) + 3*(educrec==8) + 4*(educrec==9);

generate experience=age-17 if edu==1;
replace experience=age-19 if edu==2;
replace experience=age-21 if edu==3;
replace experience=age-23 if edu==4;
drop if experience<1 | experience>40;

generate exp=1*(experience>=1 & experience<=5) + 2*(experience>=6 & experience<=10) + 
3*(experience>=11 & experience<=15) + 4*(experience>=16 & experience<=20)
+5*(experience>=21 & experience<=25) + 6*(experience>=26 & experience<=30) 
+ 7*(experience>=31 & experience<=35) + 8*(experience>=36 & experience<=40);

** define hours and weeks worked ***;
generate hours=uhrswork;
generate weeks=wkswork1;

*** supply in hours worked, 4 groups;
gen howo=perwt*hours*weeks/2000;
gen howo_ma_us=howo*male*(1-foreign);
gen howo_ma_for=howo*male*(foreign);
gen howo_fe_us=howo*(1-male)*(1-foreign);
gen howo_fe_for=howo*(1-male)*(foreign);

*** supply in bodycount, 4 groups;
gen employment=perwt;
gen empl_ma_us=employment*male*(1-foreign);
gen empl_ma_for=employment*male*(foreign);
gen empl_fe_us=employment*(1-male)*(1-foreign);
gen empl_fe_for=employment*(1-male)*(foreign);

keep edu exp employment empl_ma_us empl_ma_for empl_fe_us empl_fe_for
howo howo_ma_us howo_ma_for howo_fe_us howo_fe_for;

sort edu exp;
collapse (sum) employment empl_ma_us empl_ma_for empl_fe_us empl_fe_for
howo howo_ma_us howo_ma_for howo_fe_us howo_fe_for, by(edu exp);
gen year=2006;

save supply_groups_06,replace;


*POOL DATA ACROSS YEARS;
use supply_groups_60.dta, clear;
append using supply_groups_70.dta;
append using supply_groups_80.dta;
append using supply_groups_90.dta;
append using supply_groups_00.dta;
append using supply_groups_06.dta;

save supply_groups_60_06_specification_1, replace;
for any 60 70 80 90 00 06 : erase supply_groups_X.dta;