libname in 'C:\Documents and Settings\Laura\My Documents\Cross-State Policy\CPS data\Data\Raw' ;
libname out 'C:\Documents and Settings\Laura\My Documents\Cross-State Policy\CPS data\TAXSIM' ;

options nofmterr;

%macro CPS1; * do the 1992-99 files first;
%do i=2 %to 9;
data temp9&i; set in.mar9&i;
* variables created below are those that need combining to be included as TAXSIM requires;
year=199&i;
**STATE recode;
if state=63 then STATER=1; if state=94 then STATER=2; if state=86 then STATER=3; 
if state=71 then STATER=4; if state=93 then STATER=5; if state=84 then STATER=6; 
if state=16 then STATER=7; if state=51 then STATER=8; if state=53 then STATER=9;
if state=59 then STATER=10; if state=58 then STATER=11; if state=95 then STATER=12;
if state=82 then STATER=13; if state=33 then STATER=14; if state=32 then STATER=15;
if state=42 then STATER=16; if state=47 then STATER=17; if state=61 then STATER=18; 
if state=72 then STATER=19; if state=11 then STATER=20; if state=52 then STATER=21;
if state=14 then STATER=22; if state=34 then STATER=23; if state=41 then STATER=24;
if state=64 then STATER=25; if state=43 then STATER=26; if state=81 then STATER=27;
if state=46 then STATER=28; if state=88 then STATER=29; if state=12 then STATER=30; 
if state=22 then STATER=31; if state=85 then STATER=32; if state=21 then STATER=33;
if state=56 then STATER=34; if state=44 then STATER=35; if state=31 then STATER=36;
if state=73 then STATER=37; if state=92 then STATER=38; if state=23 then STATER=39;
if state=15 then STATER=40; if state=57 then STATER=41; if state=45 then STATER=42; 
if state=62 then STATER=43; if state=74 then STATER=44; if state=87 then STATER=45;
if state=13 then STATER=46; if state=54 then STATER=47; if state=91 then STATER=48;
if state=55 then STATER=49; if state=35 then STATER=50; if state=83 then STATER=51;

* create TAXSIM status variable;
if FLSTAT ne 6; * drop those who are nonfilers (later give them zeroes for tax vars);
if FLSTAT=5 then STATUS=1; * single;
else if 1 le FLSTAT le 3 then STATUS=2; * joint;
else if FLSTAT=4 then STATUS=3; * head of household;

* create TAXSIM age gt 65 variable;
if FLSTAT=2 then AGE65=1; * close to #65 year olds because this is # filers of that age;
else if FLSTAT=3 then AGE65=2; 
else AGE65=0;

RENT=0; *because we have no rent indicator;
ITEMS=sum(FEDRET,FICA); *includes just payroll taxes because we have no other itemized expenses;
CCE=0; *because we have no child care expenses;
NDEP=_CHILD18; 
MORT=0;
SHORT=0;
LONG=0;

**** create head and spouse earnings variables and other taxable & nontaxable income variables;
** here is what we decided to do:
   heads get their wages
   spouses get their wages + the kids wages
   other family members who file singly get computed separately (as HEADWS var for TAXSIM)
   non family members who file get computed separately ;
** this means that the resulting TAXSIM output will get summed up by family
** any nonfilers get zero for all tax variables;

if FAMREL=1 then HEADWS=sum(_incwag,INCWG1,_incse,INCSE1); * HH heads;
else if FAMREL=4 then HEADWS=sum(_incwag,INCWG1,_incse,INCSE1); * other relatives (filing singly);
else if FAMREL=0 then HEADWS=sum(_incwag,INCWG1,_incse,INCSE1); * non relatives (single-person family);
else if FAMREL=2 then SPOUSWS=sum(_incwag,INCWG1,_incse,INCSE1);
else if FAMREL=3 then KIDERN=sum(_incwag,INCWG1,_incse,INCSE1);
OTHERT=sum(FMIRNT,FMIINT,FMIALM,FMIOTH,KIDERN);
OTHERNT=sum(FMIPA,FMICH2,FMICP2,FMISEC,FMIDIS,FMIVET,FMISRV);

proc sort data=temp9&i; by hhseq fampos; run;
data rollup9&i; set temp9&i; by hhseq fampos;
   retain HEAD SPOUSE fstatus;
   if first.fampos then do;
 	  HEAD=0;
      SPOUSE=0;
	  fstatus=status; * this keeps the first record filing status;
	  end;
	  HEAD=sum(HEAD,HEADWS); 
      SPOUSE=sum(SPOUSE,SPOUSWS,KIDERN); * this sums up their records to the family;
   if last.fampos then output;
run;

data all9&i; 
   merge rollup9&i(keep=hhseq fampos fstatus head spouse) temp9&i;  
   by hhseq fampos;
if 2 ne FAMREL ne 3; * drop kid and spouse records because they are accounted for in head record for TAXSIM;
if head lt 0 then head=0; * TAXSIM wont take any income <0;
if spouse lt 0 then spouse=0; * TAXSIM wont take any income <0;

proc sort data=all9&i; by hhseq fampos;
* the KEEP statement below lists the variables (and order) needed for TAXSIM;
data out.TAXSIM9&i(KEEP=RECNUM YEAR STATER fSTATUS _CHILD18 AGE65 HEAD SPOUSE FMIDV2 OTHERT 
                  FMIRET FMISS OTHERNT RENT PRPTAX ITEMS CCE FMIUC NDEP MORT SHORT LONG);
    retain RECNUM YEAR STATER fSTATUS _CHILD18 AGE65 HEAD SPOUSE FMIDV2 OTHERT 
           FMIRET FMISS OTHERNT RENT PRPTAX ITEMS CCE FMIUC NDEP MORT SHORT LONG; * reorder for TAXSIM;
set all9&i; by hhseq fampos; if stater ne 9; * drop DC; RUN;

proc means data=out.TAXSIM9&i; * see if it did what was expected;
   title 'Test means 199'&i ; run;
proc print data=out.TAXSIM9&i (obs=40) noobs; * see if it did what was expected;
   title 'Test print 199'&i ; run;

* output data in text format for submission to TAXSIM;
* these are the PROC EXPORT details from SAS, which, by default, output variable names in the first row,
  which we do not want for TAXSIM, so I am modifying the specific code (dont ask me what some of this means);
data _null_;
set  OUT.TAXSIM9&i                                 end=EFIEOD;
%let _EFIERR_ = 0; /* set the ERROR detection macro variable */
%let _EFIREC_ = 0;     /* clear export record count macro variable */
file "C:\Documents and Settings\Laura\My Documents\Cross-State Policy\CPS data\taxsim\Temp9&i..txt"
delimiter=' ' DSD DROPOVER lrecl=32767;
format RECNUM best12. ; * this best12 format was the default for SAS;
format YEAR best12. ;
format STATER best12. ;
format fSTATUS best12. ;
format _CHILD18 best12. ;
format AGE65 best12. ;
format HEAD best12. ;
format SPOUSE best12. ;
format FMIDV2 best12. ;
format OTHERT best12. ;
format FMIRET best12. ;
format FMISS best12. ;
format OTHERNT best12. ;
format RENT best12. ;
format PRPTAX best12. ;
format ITEMS best12. ;
format CCE best12. ;
format FMIUC best12. ;
format NDEP best12. ;
format MORT best12. ;
format SHORT best12. ;
format LONG best12. ;
do;
   EFIOUT + 1;
   put RECNUM @;
   put YEAR @;
   put STATER @;
   put fSTATUS @;
   put _CHILD18 @;
   put AGE65 @;
   put HEAD @;
   put SPOUSE @;
   put FMIDV2 @;
   put OTHERT @;
   put FMIRET @;
   put FMISS @;
   put OTHERNT @;
   put RENT @;
   put PRPTAX @;
   put ITEMS @;
   put CCE @;
   put FMIUC @;
   put NDEP @;
   put MORT @;
   put SHORT @;
   put LONG ;
end;
if _ERROR_ then call symputx('_EFIERR_',1);  /* set ERROR detection macro variable */
if EFIEOD then call symputx('_EFIREC_',EFIOUT);
run;
%end; %mend; %CPS1; run;

%macro CPS2; * now do the 2000-05 files ;
%do i=0 %to 5;
data temp0&i; set in.mar0&i;
* variables created below are those that need combining to be included as TAXSIM requires;
year=200&i;
**STATE recode;
if state=63 then STATER=1; if state=94 then STATER=2; if state=86 then STATER=3; 
if state=71 then STATER=4; if state=93 then STATER=5; if state=84 then STATER=6; 
if state=16 then STATER=7; if state=51 then STATER=8; if state=53 then STATER=9;
if state=59 then STATER=10; if state=58 then STATER=11; if state=95 then STATER=12;
if state=82 then STATER=13; if state=33 then STATER=14; if state=32 then STATER=15;
if state=42 then STATER=16; if state=47 then STATER=17; if state=61 then STATER=18; 
if state=72 then STATER=19; if state=11 then STATER=20; if state=52 then STATER=21;
if state=14 then STATER=22; if state=34 then STATER=23; if state=41 then STATER=24;
if state=64 then STATER=25; if state=43 then STATER=26; if state=81 then STATER=27;
if state=46 then STATER=28; if state=88 then STATER=29; if state=12 then STATER=30; 
if state=22 then STATER=31; if state=85 then STATER=32; if state=21 then STATER=33;
if state=56 then STATER=34; if state=44 then STATER=35; if state=31 then STATER=36;
if state=73 then STATER=37; if state=92 then STATER=38; if state=23 then STATER=39;
if state=15 then STATER=40; if state=57 then STATER=41; if state=45 then STATER=42; 
if state=62 then STATER=43; if state=74 then STATER=44; if state=87 then STATER=45;
if state=13 then STATER=46; if state=54 then STATER=47; if state=91 then STATER=48;
if state=55 then STATER=49; if state=35 then STATER=50; if state=83 then STATER=51;

* create TAXSIM status variable;
if FLSTAT ne 6; * drop those who are nonfilers (later give them zeroes for tax vars);
if FLSTAT=5 then STATUS=1; * single;
else if 1 le FLSTAT le 3 then STATUS=2; * joint;
else if FLSTAT=4 then STATUS=3; * head of household;

* create TAXSIM age gt 65 variable;
if FLSTAT=2 then AGE65=1; * close to #65 year olds because this is # filers of that age;
else if FLSTAT=3 then AGE65=2; 
else AGE65=0;

RENT=0; *because we have no rent indicator;
ITEMS=sum(FEDRET,FICA); *includes just payroll taxes because we have no other itemized expenses;
CCE=0; *because we have no child care expenses;
NDEP=_CHILD18; 
MORT=0;
SHORT=0;
LONG=0;

**** create head and spouse earnings variables and other taxable & nontaxable income variables;
** here is what we decided to do:
   heads get their wages
   spouses get their wages + the kids wages
   other family members who file singly get computed separately (as HEADWS var for TAXSIM)
   non family members who file get computed separately ;
** this means that the resulting TAXSIM output will get summed up by family
** any nonfilers get zero for all tax variables;

if FAMREL=1 then HEADWS=sum(_incwag,INCWG1,_incse,INCSE1); * HH heads;
else if FAMREL=4 then HEADWS=sum(_incwag,INCWG1,_incse,INCSE1); * other relatives (filing singly);
else if FAMREL=0 then HEADWS=sum(_incwag,INCWG1,_incse,INCSE1); * non relatives (single-person family);
else if FAMREL=2 then SPOUSWS=sum(_incwag,INCWG1,_incse,INCSE1);
else if FAMREL=3 then KIDERN=sum(_incwag,INCWG1,_incse,INCSE1);
OTHERT=sum(FMIRNT,FMIINT,FMIALM,FMIOTH,KIDERN);
OTHERNT=sum(FMIPA,FMICH2,FMICP2,FMISEC,FMIDIS,FMIVET,FMISRV);

proc sort data=temp0&i; by hhseq fampos; run;
data rollup0&i; set temp0&i; by hhseq fampos;
   retain HEAD SPOUSE fstatus;
   if first.fampos then do;
 	  HEAD=0;
      SPOUSE=0;
	  fstatus=status; * this keeps the first records filing status;
	  end;
	  HEAD=sum(HEAD,HEADWS); 
      SPOUSE=sum(SPOUSE,SPOUSWS,KIDERN); * this sums up their records to the family;
   if last.fampos then output;
run;

data all0&i; 
   merge rollup0&i(keep=hhseq fampos fstatus head spouse) temp0&i;  
   by hhseq fampos;
if 2 ne FAMREL ne 3; * drop kid and spouse records because they are accounted for in head record for TAXSIM;
if head lt 0 then head=0; * TAXSIM wont take any income <0;
if spouse lt 0 then spouse=0; * TAXSIM wont take any income <0;

proc sort data=all0&i; by hhseq fampos;
* the KEEP statement below lists the variables (and order) needed for TAXSIM;
data out.TAXSIM0&i(KEEP=RECNUM YEAR STATER fSTATUS _CHILD18 AGE65 HEAD SPOUSE FMIDV2 OTHERT 
                   FMIRET FMISS OTHERNT RENT PRPTAX ITEMS CCE FMIUC NDEP MORT SHORT LONG);
	retain RECNUM YEAR STATER fSTATUS _CHILD18 AGE65 HEAD SPOUSE FMIDV2 OTHERT 
                   FMIRET FMISS OTHERNT RENT PRPTAX ITEMS CCE FMIUC NDEP MORT SHORT LONG; * reorder for TAXSIM;
set all0&i; by hhseq fampos; if stater ne 9; * drop DC; RUN;

proc means data=out.TAXSIM0&i; * see if it did what was expected;
   title 'Test means 200'&i ; run;
proc print data=out.TAXSIM0&i (obs=40) noobs; * see if it did what was expected;
   title 'Test print 200'&i ; run;

data _null_;
set  OUT.TAXSIM0&i                                 end=EFIEOD;
%let _EFIERR_ = 0; /* set the ERROR detection macro variable */
%let _EFIREC_ = 0;     /* clear export record count macro variable */
file "C:\Documents and Settings\Laura\My Documents\Cross-State Policy\CPS data\taxsim\Temp0&i..txt"
delimiter=' ' DSD DROPOVER lrecl=32767;
format RECNUM best12. ; * this best12 format was the default for SAS;
format YEAR best12. ;
format STATER best12. ;
format fSTATUS best12. ;
format _CHILD18 best12. ;
format AGE65 best12. ;
format HEAD best12. ;
format SPOUSE best12. ;
format FMIDV2 best12. ;
format OTHERT best12. ;
format FMIRET best12. ;
format FMISS best12. ;
format OTHERNT best12. ;
format RENT best12. ;
format PRPTAX best12. ;
format ITEMS best12. ;
format CCE best12. ;
format FMIUC best12. ;
format NDEP best12. ;
format MORT best12. ;
format SHORT best12. ;
format LONG best12. ;
do;
   EFIOUT + 1;
   put RECNUM @;
   put YEAR @;
   put STATER @;
   put fSTATUS @;
   put _CHILD18 @;
   put AGE65 @;
   put HEAD @;
   put SPOUSE @;
   put FMIDV2 @;
   put OTHERT @;
   put FMIRET @;
   put FMISS @;
   put OTHERNT @;
   put RENT @;
   put PRPTAX @;
   put ITEMS @;
   put CCE @;
   put FMIUC @;
   put NDEP @;
   put MORT @;
   put SHORT @;
   put LONG ;
end;
if _ERROR_ then call symputx('_EFIERR_',1);  /* set ERROR detection macro variable */
if EFIEOD then call symputx('_EFIREC_',EFIOUT);
run;
%end; %mend; %CPS2; run;