set memory 500m


use "/home/eva/CPS_QR/CPS6203_raw.dta", clear




/* rename variables to more friendly names */


for @ in any region state popstat marstat year race: rename _@ @
rename _esr emp_stat

rename _wkslyr weeksly

rename _clslyr classly

rename occlyr occly




/* drop observations for children and for Armed Forces members, i.e. restrict sample to civilian adults */


keep if popstat==1

drop popstat




/* for years with two samples, keep only one of them */


drop if year=="1988b"

drop if year=="2001"

replace year="2001" if year=="2001s"


destring year, replace


compress




/* load the values of the personal consumption expenditure deflator for each year */


gen pce=20.016 if year==1962

replace pce=20.250 if year==1963

replace pce=20.484 if year==1964

replace pce=20.773 if year==1965

replace pce=21.079 if year==1966

replace pce=21.628 if year==1967

replace pce=22.168 if year==1968

replace pce=23.033 if year==1969

replace pce=24.077 if year==1970

replace pce=25.212 if year==1971

replace pce=26.292 if year==1972

replace pce=27.211 if year==1973

replace pce=28.687 if year==1974

replace pce=31.641 if year==1975
replace pce=34.225 if year==1976

replace pce=36.089 if year==1977

replace pce=38.475 if year==1978

replace pce=41.221 if year==1979

replace pce=44.868 if year==1980

replace pce=49.712 if year==1981

replace pce=54.097 if year==1982

replace pce=57.158 if year==1983

replace pce=59.598 if year==1984

replace pce=61.795 if year==1985

replace pce=63.920 if year==1986

replace pce=65.478 if year==1987

replace pce=67.972 if year==1988

replace pce=70.628 if year==1989

replace pce=73.708 if year==1990

replace pce=77.102 if year==1991

replace pce=80.056 if year==1992

replace pce=82.496 if year==1993

replace pce=84.468 if year==1994

replace pce=86.170 if year==1995

replace pce=88.151 if year==1996

replace pce=90.041 if year==1997

replace pce=91.788 if year==1998

replace pce=92.770 if year==1999

replace pce=94.300 if year==2000

replace pce=96.695 if year==2001

replace pce=98.649 if year==2002

replace pce=100.000 if year==2003



/* infer region variable from state variable for years with missing region data */
/* turn missing values into 0 for the region variable */


replace region=1 if state>=1 & state<=5 & ((year>=1968 & year<=1975) | year==1963)

replace region=2 if state>=6 & state<=10 & ((year>=1968 & year<=1975) | year==1963)

replace region=3 if state>=11 & state<=18 & ((year>=1968 & year<=1975) | year==1963)

replace region=4 if state>=19 & state<=21 & ((year>=1968 & year<=1975) | year==1963)

replace region=0 if region==.
drop state




/* infer hispanic variable from spanish ethnicity variable 1971 onwards */

/* hispanic is set to missing for years prior to 1971 */


gen byte hispanic=0 if year>=1971

replace hispanic=1 if _spneth<=7 & year>=1971

drop _spneth




/* replace missing occupation and weight observations with 0 */

/* replace missing and not-in-smaple class of worker observations with 0 */


replace occly=0 if occly==.

replace wgt=0 if wgt==.

replace classly=0 if classly==. | classly==6 | classly==9




/* harmonize the definition of employment status by changing all unemployed to 3 */

/* infer employment status for 1994 from other monthly labor force status indicator */


replace emp_stat=3 if emp_stat==4

replace emp_stat=1 if mlr==1 & year==1994

replace emp_stat=2 if mlr==2 & year==1994

replace emp_stat=3 if (mlr==3 | mlr==4) & year==1994

replace emp_stat=7 if mlr>=5 & mlr<=7 & year==1994

drop mlr




/* generate education group indicator 1992 onwards from original CPS variable that indicates highest grade/degree attained */

/* prior to 1991 use the recoded variable created by Unicon that indicates the highest grade completed */

/* to insure the least break at the time of the data collection switch, we categorize workers with 17 years of education as college grads */



gen byte educ=1 if grdatn<=38 & year>=1992

replace educ=2 if grdatn==39 & year>=1992

replace educ=3 if grdatn>=40 & grdatn<=42 & year>=1992

replace educ=4 if grdatn==43 & year>=1992

replace educ=5 if grdatn>43 & grdatn~=. & year>=1992


replace grdcom=1 if grdcom==6

replace grdcom=2 if grdcom==7

replace _educ=_educ+0.5 if grdcom==2 & year<=1991

replace educ=1 if _educ<=11.5 & year<=1991

replace educ=2 if _educ==12 & year<=1991

replace educ=3 if _educ>=12.5 & _educ<=15.5 & year<=1991

replace educ=4 if _educ>=16 & _educ<=17.5 & year<=1991

replace educ=5 if _educ>=18 & _educ~=. & year<=1991

drop grdatn _educ grdcom




/* the variable wage_sample is 1 for observations that are in the sample from which we calculate mean wages, etc */

/* the variable emp_sample is 1 for observations that are in the sample from which we calculate employment statistics */

/* all variables are initially in both samples, then we make sample restriction choices later in the file */


gen byte wage_sample=1

gen byte emp_sample=1

gen byte week_sample=1

gen byte occ_sample=1




/* impose age restriction on wage and employment sample, note that since employment refers to last week, age is this year's age */


replace wage_sample=0 if age<23 | age>66

replace week_sample=0 if age<23 | age>66

replace emp_sample=0 if age<22 | age>65

replace occ_sample=0 if age<23 | age>66



/* the employment sample requires that employment status is observed */


replace emp_sample=0 if emp_stat==0




/* determine which observations represent full-time/full-year observations, only these should remain in the wage sample */

/* when changing the definition of full year replace 40 in the lines below by 48 */


replace weeksly=0 if weeksly==.

replace ftpt=0 if ftpt==. | ftpt==5 | ftpt==9
gen byte ftfy=0

replace ftfy=1 if ftpt==2 & weeksly>=40 & year==1962 

replace ftfy=1 if ftpt==1 & weeksly>=40 & year>=1963 & year<=1967

replace ftfy=1 if (ftpt==1 | ftpt==3) & weeksly>=40 & year>=1968

replace wage_sample=0 if ftfy==0
replace occ_sample=0 if ftfy==0




/*drop observations where the worker is working without pay */

/* or is self-employed */


replace wage_sample=0 if classly>=3

replace occ_sample=0 if classly>=3




/* harmonize wage observations so that they are always 0 when observation not in universe */


replace incwag=0 if incwag==99999 & year>=1968 & year<=1975

replace incwag=0 if incwag==. & year>=1962 & year<=1967




/* generate a variable indicating whether the observation was topcoded or not */


gen byte topcode=0

replace topcode=1 if incer1>200000 & year==2003

replace topcode=1 if incer1>150000 & year>=1996 & year<=2002

replace topcode=1 if incer1>=99999 & year>=1989 & year<=1995

replace topcode=1 if incwag>=99999 & year>=1985 & year<=1988

replace topcode=1 if incwag>=75000 & year>=1982 & year<=1984

replace topcode=1 if incwag>=50000 & year>=1968 & year<=1981

replace topcode=1 if incwag>=99900 & year>=1962 & year<=1967




/* replace topcoded wage observations by 1.75 the top-code value */


replace incwag=incwag*1.75 if topcode==1 & year<=1995




/* generate real annualized weekly wages */


rename incwag wage_total
gen wage=wage_total/weeksly*52 if weeksly>0 
 
/* annualize weekly wages to 52 weeks */

gen real_wage=wage/pce*100




/* do not include in the wage sample if annual real wage is less than a sixth of the real minimum wage ($1/hour), times 35 hours/week times 40 weeks */


replace wage_sample=0 if real_wage<35*40




/* do not include in the weeks sample if observation is not in the universe */


replace week_sample=0 if match~=0 & year==1962

replace week_sample=0 if match~=1 & year>=1963 & year<=1967




/* generate occupation dummies for phh and farm workers */


rename occly occ

gen byte ph=0

replace ph=1 if occ==20 & year==1962

replace ph=1 if occ==21 & year>=1963 & year<=1967

replace ph=1 if occ>=801 & occ<=804 & year>=1968 & year<=1970

replace ph=1 if occ>=980 & occ<=986 & year>=1971 & year<=1982

replace ph=1 if occ>=403 & occ<=407 & year>=1983 & year<=2002




gen byte farmhand=0

replace farmhand=1 if occ==23 & year==1962

replace farmhand=1 if occ==24 & year>=1963 & year<=1967

replace farmhand=1 if occ>=901 & occ<=903 & year>=1968 & year<=1970

replace farmhand=1 if occ>=821 & occ<=823 & year>=1971 & year<=1982

replace farmhand=1 if occ==846 & year>=1971 & year<=1982

replace farmhand=1 if occ>=477 & occ<=479 & year>=1983 & year<=2002




gen byte farmer=0

replace farmer=1 if occ==6 & year==1962

replace farmer=1 if occ==7 & year>=1963 & year<=1967

replace farmer=1 if occ>=200 & occ<=222 & year>=1968 & year<=1970

replace farmer=1 if occ>=801 & occ<=806 & year>=1971 & year<=1982

replace farmer=1 if occ>=473 & occ<=476 & year>=1983 & year<=2002




/* generate occupation dummies for all workers */



gen byte occat=.



/* 1983-2002 */


/* Managers & professionals */

replace occat=1 if occ>=1 & occ<=94 & year>=1983 & year<=2002

replace occat=1 if occ>=113 & occ<=154 & year>=1983 & year<=2002

replace occat=1 if occ>=166 & occ<=173 & year>=1983 & year<=2002

replace occat=1 if occ>=178 & occ<=202 & year>=1983 & year<=2002

replace occat=1 if occ>=229 & occ<=229 & year>=1983 & year<=2002

replace occat=1 if occ>=243 & occ<=243 & year>=1983 & year<=2002



/* White collar */

replace occat=2 if occ>=95 & occ<=112 & year>=1983 & year<=2002

replace occat=2 if occ>=155 & occ<=165 & year>=1983 & year<=2002

replace occat=2 if occ>=174 & occ<=177 & year>=1983 & year<=2002

replace occat=2 if occ>=203 & occ<=228 & year>=1983 & year<=2002

replace occat=2 if occ>=230 & occ<=242 & year>=1983 & year<=2002

replace occat=2 if occ>=244 & occ<=344 & year>=1983 & year<=2002

replace occat=2 if occ>=359 & occ<=402 & year>=1983 & year<=2002



/* Blue collar */

replace occat=3 if occ>=345 & occ<=358 & year>=1983 & year<=2002

replace occat=3 if occ>=403 & occ<=904 & year>=1983 & year<=2002




/* 1971-1982 */


/* Managers & professionals */

replace occat=1 if occ>=1 & occ<=21 & year>=1971 & year<=1982

replace occat=1 if occ>=23 & occ<=31 & year>=1971 & year<=1982

replace occat=1 if occ>=34 & occ<=63 & year>=1971 & year<=1982

replace occat=1 if occ>=65 & occ<=73 & year>=1971 & year<=1982

replace occat=1 if occ>=91 & occ<=99 & year>=1971 & year<=1982

replace occat=1 if occ>=102 & occ<=140 & year>=1971 & year<=1982

replace occat=1 if occ>=161 & occ<=161 & year>=1971 & year<=1982

replace occat=1 if occ>=170 & occ<=170 & year>=1971 & year<=1982

replace occat=1 if occ>=175 & occ<=195 & year>=1971 & year<=1982

replace occat=1 if occ>=201 & occ<=219 & year>=1971 & year<=1982

replace occat=1 if occ>=222 & occ<=225 & year>=1971 & year<=1982

replace occat=1 if occ>=227 & occ<=259 & year>=1971 & year<=1982

/* White collar */

replace occat=2 if occ>=22 & occ<=22 & year>=1971 & year<=1982

replace occat=2 if occ>=32 & occ<=33 & year>=1971 & year<=1982

replace occat=2 if occ>=64 & occ<=64 & year>=1971 & year<=1982

replace occat=2 if occ>=74 & occ<=90 & year>=1971 & year<=1982

replace occat=2 if occ>=100 & occ<=101 & year>=1971 & year<=1982

replace occat=2 if occ>=141 & occ<=160 & year>=1971 & year<=1982

replace occat=2 if occ>=162 & occ<=169 & year>=1971 & year<=1982

replace occat=2 if occ>=171 & occ<=174 & year>=1971 & year<=1982

replace occat=2 if occ>=196 & occ<=200 & year>=1971 & year<=1982

replace occat=2 if occ>=220 & occ<=220 & year>=1971 & year<=1982

replace occat=2 if occ>=260 & occ<=400 & year>=1971 & year<=1982




/* Blue collar */

replace occat=3 if occ>=221 & occ<=221 & year>=1971 & year<=1982

replace occat=3 if occ>=226 & occ<=226 & year>=1971 & year<=1982

replace occat=3 if occ>=401 & occ<=986 & year>=1971 & year<=1982




save "/home/eva/CPS_QR/Final/CPS6203.dta", replace