log using annual05, text replace
set memory 1000m
set more 1
*primary family and child variables added 2000-04-20 by Jean Roth, jroth@nber.org
*Y2K compliance by Jean Roth, jroth@nber.org, 2001-01-15
*Imputed highest grade completed, ihigrdc, added by Jean Roth, jroth@nber.org, 2003-01-20
*Stata 8 compliance by Jean Roth, jroth@nber.org, 2003-02-28
*matching variables added by Jean Roth, jroth@nber.org, 2003-02-12
* run with nohup stata -b do annual2002 & to run in batch mode
do rw
program define aef2
local dct_name "aef`3'.dbd"
local raw_name "../raw/morg`2'"
display "dct_name `dct_name'" 
display "raw_name `raw_name'" 
*quietly infile using "`dct_name'" if age>15&age!=. , using("`raw_name'") clear 
quietly infile using "`dct_name'" if age>15&age!=. in 1/100 , using("`raw_name'") clear
*quietly infile using "`dct_name'" in 1/100 , using("`raw_name'") clear
*quietly infile using aef if age>15&age!=.
display "infile successful"
#delimit cr
generate int year = `1'
*summarize
tab intmonth if `1'==2005
* Geography
display "Geography"
if  `1' == 2005 {
   replace serial = serial04 
   replace hhnum = hhnum04 
   display "dropping cmsacode msafips smsa93"
   drop cmsacode msafips smsa93
   replace smsa04 = . if smsa04<2 & intmonth >= 5
}
display "smsasize = . if smsasize <=0"
capture replace smsasize = . if smsasize <=0
display "replace smsastat = . if smsastat==3 | smsastat<0"
replace smsastat = . if smsastat==3 | smsastat<0
display "replace smsasize=. if smsasize<=0"
**should line below be commented out???
capture replace smsasize=. if smsasize<=0  
display "drop smsasize"
capture drop smsasize
* Demography
display "Demography"
if `1' >  1991 { 
    replace grade92 = . if grade92 < 0 
}
if `1' >= 1998 { 
    replace ged      = . if ged < 0
    replace gedhigr  = . if gedhigr < 0
    replace yrcoll   = . if yrcoll < 0
    replace grprof   = . if grprof < 0
    replace gr6cor   = . if gr6cor < 0
    replace ms123    = . if ms123 < 0        
    generate double ihigrdc = .  
    ** Jaeger
    label var ihigrdc "Imputed highest grade completed"
    replace ihigrdc = 0   if grade92==31  
    replace ihigrdc = 2.5 if grade92==32   
    replace ihigrdc = 5.5 if grade92==33  
    replace ihigrdc = 7.5 if grade92==34 
    replace ihigrdc = 9   if grade92==35  
    replace ihigrdc =10   if grade92==36  
    replace ihigrdc =11   if grade92==37  
    replace ihigrdc =12   if grade92==38  
    replace ihigrdc = 0   if grade92==39 & ged==2 & gedhigr==1  
    replace ihigrdc = 2.5 if grade92==39 & ged==2 & gedhigr==2  
    replace ihigrdc = 5.5 if grade92==39 & ged==2 & gedhigr==3  
    replace ihigrdc = 7.5 if grade92==39 & ged==2 & gedhigr==4  
    replace ihigrdc = 9   if grade92==39 & ged==2 & gedhigr==5 
    replace ihigrdc = 10  if grade92==39 & ged==2 & gedhigr==6  
    replace ihigrdc = 11  if grade92==39 & ged==2 & gedhigr==7  
    replace ihigrdc = 12  if grade92==39 & ged==2 & gedhigr==8  
    replace ihigrdc = 12  if ged==1
    replace ihigrdc = 12  if grade92>=40 & grade92<=42 & yrcoll==1
    replace ihigrdc = 13  if grade92>=40 & grade92<=42 & yrcoll==2
    replace ihigrdc = 14  if grade92>=40 & grade92<=42 & yrcoll==3
    replace ihigrdc = 15  if grade92>=40 & grade92<=42 & yrcoll==4
    replace ihigrdc = 16  if grade92>=40 & grade92<=42 & yrcoll==5
    replace ihigrdc = 16  if grade92==43 & grprof==2
    replace ihigrdc = 17  if grade92==43 & gr6cor==2
    replace ihigrdc = 18  if grade92==43 & gr6cor==1
    replace ihigrdc = 17  if grade92==44 & ms123==1
    replace ihigrdc = 18  if grade92==44 & ms123>=2 & ms123<.
    replace ihigrdc = 18  if grade92==45 | grade92==46
}
if `1' >= 1994 { 
    replace penatvty = . if penatvty < 0
    replace pemntvty = . if pemntvty < 0
    replace pefntvty = . if pefntvty < 0
    replace prcitshp = . if prcitshp < 0
    replace prcitflg = . if prcitflg < 0 
}
if `1' > 1989  { 
    replace ethnic = . if ethnic < 0 
    replace ethnic = . if ethnic >= 10
}
if `1' == 2005 {
    display "replacing veteran and dropping vet1-vet4 if intmonth <=7"
    replace veteran  = . if veteran  < 0 & intmonth <= 7 
    replace vet1=. if intmonth <= 7 
    replace vet2=. if intmonth <= 7 
    replace vet3=. if intmonth <= 7 
    replace vet4=. if intmonth <= 7 
}
if `1' == 2005 {
    display "setting veteran to missing and vet1-vet4 to missing"
    replace veteran = . if intmonth >= 8
    replace vet1 = . if vet1 < 0 & intmonth >= 8
    replace vet2 = . if vet2 < 0 & intmonth >= 8
    replace vet3 = . if vet3 < 0 & intmonth >= 8
    replace vet4 = . if vet4 < 0 & intmonth >= 8
}

* Employment
display "Employment"
#delimit ;
;
display "docc00" ;
if `1' >= 2000  { 
    display "generate int docc00=occ00;";
    generate int docc00=occ00;
    display "recode docc00";
    recode docc00  (10/430=1) (500/950=2) (1000/1240=3) (1300/1560=4) (1600/1960=5) 
    (2000/2060=6) (2100/2150=7) (2200/2550=8) (2600/2960=9) (3000/3540=10) (3600/3650=11) 
    (3700/3950=12) (4000/4160=13) (4200/4250=14) (4300/4650=15) (4700/4960=16) 
    (5000/5930=17) (6000/6130=18) (6200/6940=19) (7000/7620=20) (7700/8960=21) 
    (9000/9750=22) (9840=23);
} ;

display "dind";
if `1' >= 2000 {
    display "replacing dind for `1' >= 2000 " ;
    gen dind02 = ind02;
    recode
    dind02 (170/180 290=1) (190/280=2) (370/490=3) (770=4) (2470/2590=5)
    (2670/2990=6) (3070/3290=7) (3360/3390=8) (3470 3490=9) (3570/3690=10)
    (3770/3870=11) (3890=12) (3960/3990=13) (1070/1290=14) (1370 1390=15)
    (1470/1790=16) (1870/1990=17) (2070 2090=18) (2170/2290=19) (2370/2390=20)
    (4070/4590=21) (4670/5790=22) (6070/6390=23) (570/690=24) (6470/6490=25) 
    (6570 6590=26) (6670=27) (6675=28) (6680 6690=29) (6692 6695=30)
    (6770 6780=31) (6870/6970=32) (6990=33) (7070=34) (7080/7190=35)
    (7270/7490=36) (7570=37) (7580/7780=38) (7790=39) (7860/7890=40)
    (8190=41) (7970/8180 8270 8290=42) (8370/8470=43) (8560/8590=44) (8660 8670=45)
    (8680 8690=46) (8770/8890=47) (8970/9090=48) (9160/9190=49) (9290=50)
    (9370/9590=51) (9890=52) ;
} ;
#delimit cr
**Wages (If earnings are 0, then wage rate is missing, not zero).
display "Wages"
replace paidhr   = . if paidhr  <=0
replace paidhre  = . if paidhre <=0
replace earnhr   = . if earnhr  <=0
replace earnhre  = . if earnhre <=0
replace earnwke  = . if earnwke < 0
if `1' >= 1989 {
   replace eligible = 2 if eligible <= 0 
}
if `1' >= 1994 {
   replace I25a     = 0 if I25a <= 3
   replace I25b     = 0 if I25b <= 3
   *replace I25a = . if I25a < 0
   *replace I25b = . if I25b < 0
   replace earnhre  = . if earnhre ==1
   replace uhourse  = . if lfsr94 >2 | lfsr94 <1
   replace lfsr94    = . if lfsr94 < 0 
}
if `1' >= 1996 { 
   replace I25c = . if I25c < 0
   replace I25d = . if I25d < 0  
}
** Hours, Unions
display "Hours, Unions"
if `1' >= 1994 {
   replace reason94 = . if reason94< 0
   replace absent94 = . if absent94< 0
   replace studftpt = . if studftpt< 0
   replace schenr = . if schenr < 0
   replace schlvl = . if schlvl < 0
   replace ftpt94   = . if ftpt94  < 0
   replace hourslw  = . if hourslw < 0
   replace hourslw  = 99 if hourslw >99 & hourslw != .
   replace uhourse  = . if uhourse < 0
   replace why3594  = . if why3594 < 0
   replace unionmme = . if unionmme< 0
   replace unioncov = . if unioncov< 0 
}

** Save and exit
compress
saveold /home/data/morg/annual/morg`2',replace
summarize
clear
*! /usr/bin/rm -f aef.dct /tmp/aef.raw 
*! /bin/rm -f aef.dct /tmp/aef.raw
! /usr/bin/chmod a+r /home/data/morg/annual/morg`2'.dta 
! /bin/chmod a+r /home/data/morg/annual/morg`2'.dta
! /usr/bin/chmod g+w /home/data/morg/annual/morg`2'.dta 
! /bin/chmod g+w /home/data/morg/annual/morg`2'.dta
! /usr/bin/chgrp web /home/data/morg/annual/morg`2'.dta 
! /bin/chgrp web /home/data/morg/annual/morg`2'.dta
end

 *Each block of commands does one year of the data. The data must be
 *decompressed, and possibly have dashes converted to blanks.
 *Then a dictionary for the particular year is prepared by
 *concatenating a one line header (with the file name) to a generic
 *dictionary body that covers several years of data that used the
 *same format.
 *Lastly, the data is read, modified, summarized and saved.

aef2 2005 05 03