/********************************************************************************/
/**** This program reads in the sas dataset and contains data steps to create     ****/
/**** the 13 experimental poverty measures for 2001.
/**** This will enable analysts to replicate the experimental poverty        ****/
/**** measures or appropriately recombine thresholds and income measures to  ****/
/**** examine new ones. Also included in this file are variable and value    ****/
/**** label statements.                                                      ****/
/**** For details on construction of various income and threshold measures,  ****/
/**** refer to the P60-205 report on Experimental Poverty Measures           ****/
/**** and its Appendix C.   
/**** THIS PROGRAM WAS REVISED IN DECEMBER 2008 BY TRUDI RENWICK IN ORDER TO 
CREATE A LIST FILE THAT WOULD MATCH THE LISTS FILE CURRENTLY ON THE WEB SITE. 
POVERTY DEFINITIONS POOR5,POOR8,POOR3,POOR9, POOR4, POOR7 WERE RELABELED TO 
POOR3, POOR4, POOR5, POOR7, POOOR8, POOR9 TO MATCH THE PUBLISHED OUTPUT. THE 
VARIABLE HINSTAT WAS ADDED TO THE PUBLIC USE DATA FILE TO ENABLE CREATION OF 
THE CROSSTABS BY HINSTAT;
/********************************************************************************/


libname public 'H:\NAS Measures\2001PublicUseFile';                     

proc format;

value agefmt 
	0-17='Under 18 years' 
    18-64='18 to 64 years'
	65-high='65 and over'     
                            ;

value racefmt 
	1='White'
    2='Black'
    3='AIAN'
    4='API'  
              ;

value hofmt 
	1-7='Hispanic'
    other='not Hispanic'   ;

value kindfmt 
	1='in MC family' 
    2='in male hh fam' 
    3='in female hh fam'   ;

value regfmt 
	1='Northeast' 
    2='Midwest' 
    3='South'
    4='West'  ;

value typefmt
	1='Primary family'
    2='Nonfamily householder'
    3='Related subfamily'
	4='Unrelated subfamily'
	5='Secondary individual'
              ;

value msfmt
  1=  'married sp present'
  3=  'married sp absent'
  4=  'widowed'
  5=  'divorced/separate '
  7=  'never married'  ;

value age11fmt
  1=  'lt 3'
  2=  '3 to 5'
  3=  '6 to 11'
  4=  '12 to 17'
  5=  '18 to 21'
  6=  '22 to 44'
  7=  '45 to 54'
  8=  '55 to 59'
  9=  '60 to 64'
  10=  '65 to 74'
  11=  '75+'      ;

  value sexfmt
  1='Male'
  2='Female';

  value racerfmt
  1='White non-Hispanic'
  2='White Hispanic'
  3='Black non-Hispanic'
  4='Black Hispanic'
  5='AIAN non-Hispanic'
  6='AIAN Hispanic'
  7='API non-Hispanic'
  8='API Hispanic'
  9='Hispanic';

 value fpersonr
  1='1 person'
  2='2 persons'
  3='3 persons'
  4='4 persons'
  5='5 persons'
  6='6 persons'
  7-high='7 or more persons ' ;

value adwk
  0='no workers in family'
  1-high='workers in family';

 value educfmt
  0=  'Not in universe (under 25)'
  1=  'No high school'
  2=  'high school dip'
  3=  'some college'
  4=  'college degree'  ;

 value prcitshp
  1=  'native'
  2=  'naturalized citizen'
  3=  'not a citizen' ;

value disfmt
 1='disabled'
 2='not disabled'
 3='niu (under 16 or over 64 years of age)' ;

value sevfmt
 .='niu (under 16 or over 64 years of age)'
 0='not severely disabled'
 1='severely disabled';

value health
1='excellent'
2='very good'
3='good'
4='fair'
5='poor'
6='niu' ;

value fheadfmt
0='not family head'
1='family head';

run;

data two; set public.povpurev01;

    array poor{13} poor1-poor13;
    
curdate=177.1;
nl1=17036*177.1/166.6 *    scaleb * geo2;         /* cpi99, geo */          
nl2=17036*curdate/166.6 *    scaleb;                /* cpi99, NO GEO */         
nl3=18709 *    scaleb * geo2;         /* ce,geo */             
nl8=18709 * scaleb;                   /* ce, no geo */         

/***compute foop thresholds  using cpi spreadsheet figures *****/
/** use same percentage as ce threshold                    ****/
/**  p1 based on 1999 threshold                           ****/
nl4=(mooplin1 + fcsulin1) * geo2;             /*  cpi fcsum geo  */
nl6=(mooplin1 + fcsulin1);             /*  cpi fcsum nga*/

/***compute foop thresholds  using ce spreadsheet figures *****/

nl5=(mooplin2 + fcsulin2) * geo2;             /*  ce fcsum geo  */
nl7=(mooplin2 + fcsulin2) ;             /*  ce fcsum nga  */

/*********************    income definitions  ******************/

   def001= ftotval;

   def002 =  def001 -ffica + (fcapgain-fcaploss) - ffedtax-fsttax+feitc   /*def013*/
           + f_mv_fs + (input(fhoussub, 3.) * 12)
           + fmvfsl + fmvrsl +fengval;

   def003 = def001 -ffica + (fcapgain-fcaploss) - ffedtax-fsttax+feitc   /*no houssub in*/
           + f_mv_fs
           + fmvfsl + fmvrsl +fengval;

   def004 = def003 - combexpc + fmrcap - newmoop;    /*exp1 betson  */

   def005 = def003 - combexpc + fmrcap;              /*exp3  moopitt  */

   def006 = def003 - combexpc + fmrcap - newmoop + mooplin1;      /*exp4 cmb cpi nl*/

   def007 = def003 - combexpc + fmrcapce - newmoop;    /*exp1 betson ce */

   def008 = def003 - combexpc + fmrcapce;              /*exp3  moopitt ce */

   def009 = def003 - combexpc + fmrcapce - newmoop + mooplin2;      /*exp4 cmb ce nl*/

/**** compute poverty status  ***********/

do i=1 to 13;
  poor(i)=0;
end;

if ftotval lt fpovcut then poor1=1; /*Official poverty*/

if def004 lt nl1 then poor2=1;    /* msi - cpi   */
if def004 lt nl2 then poor3=1;     /* msi - cpi-nga  */
if def007 lt nl3 then poor4=1;    /* msi - ce   */
if def005 lt nl4 then poor5=1;    /* mit - cpi  */
if def005 lt nl6 then poor6=1;    /* mit - cpi-nga   */
if def008 lt nl5 then poor7=1;    /* mit - ce  */
if def006 lt nl4 then poor8=1;    /* cmb - cpi*/
if def006 lt nl6 then poor9=1;   /* cmb - cpi-nga  */
if def009 lt nl5 then poor10=1;   /* cmb-ce */
if def007 lt nl8 then poor11=1;    /*msi-ce-nga */
if def008 lt nl7 then poor12=1;   /* mit-ce-nga */
if def009 lt nl7 then poor13=1;   /* cmb-ce-nga*/

label
  poor1='official poverty '
poor2= 'msi-cpi'    
poor3= 'msi-cpi-nga'
poor4= 'msi'
poor5= 'mit-cpi'
poor6= 'mit-cpi-nga'   
poor7= 'mit'
poor8= 'cmb-cpi'
poor9= 'cmb-cpi-nga'
poor10= 'cmb ' 
poor11= 'msi-ce-nga'
poor12= 'mit-ce-nga'
poor13= 'cmb-ce-nga'
;
/** subgroups computed here   */

agecat=put(a_age, agefmt.) ;

race=put(a_race, racefmt.);

horg=put(a_reorgn, hofmt.);

famkind=put(fkind, kindfmt.);

famtype=put(ftype, typefmt.); 

age11=put(age, age11fmt.);

mms=put(ms, msfmt.);

sex=put(a_sex, sexfmt.);

racer=put(rac_hisp, racerfmt.);

fpersonsr=put(fpersr, fpersonr.);

adwk=put(adwkfmt, adwk.);

reg=put(region, regfmt.);

health=put(hea, health.);

sevdisab=put(severe, sevfmt.);

prcitshp=put(citshp, prcitshp.);

disab = put (dis, disfmt.);

educat=put (educ, educfmt.);

fhead=put (famhead, fheadfmt.);

label 

  curdate = 'Current CPI-U.  Obtained from the Bureau of Labor Statistics website'
  nl1='threshold used to estimate poor2 status'
  nl2='threshold used to estimate poor3 status'
  nl3='threshold used to estimate poor4 status'
  nl4='threshold used to estimate poor5 and poor8 status'
  nl5='threshold used to estimate poor7 and poor10 status'
  nl6='threshold used to estimate poor6 and poor9 status'
  nl7='threshold used to estimate poor12 and poor13 status'
  nl8='threshold used to estimate poor11 status'
  age11='11 category age recode'
  agecat='3 category age recode'
  race='3 category race recode'
  hisp='Hispanic indicator'
  adwk='presence of working adult in family'
  horg='Hispanic origin'
  def001 = 'income definition 1'
  def002 = 'income definition 2'
  def003 = 'income definition 3'
  def004 = 'income definition 4'
  def005 = 'income definition 5'
  def006 = 'income definition 6'
  def007 = 'income definition 7'
  def008 = 'income definition 8'
  def009 = 'income definition 9'
  reg='Region'
  sex = 'Sex - Male/Female'
  mms='marital status recode'
  famkind = 'kind of family'
  famtype = 'type of family'
  health='health status'
  racer='race/hispanic origin recode'
  fhead = 'Family head of household'
  educ = 'educational attainment'
  f_mv_fs='family food stamps income'
  f_mv_sl='school lunch income'
  fengval='family energy assistance subsidy amt'
  feitc='family EITC amt'
  ffica='family social security taxes amt'
  ffedtax='family federal taxes amt'
  fsttax='family state taxesamt'
  fcapgain='family capital gains amt'
  fcaploss='family capital losses amt'
  geo2='geographic adjustment to threshold'
    nl1='threshold used to estimate poor2 status'
  nl2='threshold used to estimate poor5 status'
  nl3='threshold used to estimate poor8 status'
  nl4='threshold used to estimate poor3 and poor4 status'
  nl5='threshold used to estimate poor9 and poor10 status'
  nl6='threshold used to estimate poor6 and poor7 status'
  nl7='threshold used to estimate poor12 and poor13 status'
  nl8='threshold used to estimate poor11 status'
    scaleb='three parameter scale'
  fpovcut='official pov threshold for family'
  ftotval='total family income'
  agecat='3 category age recode'
  educ='educational attainment'
  race='3 category race recode'
  hisp='Hispanic indicator'
  fpersons='# persons in family'
  adwkfmt='presence of working adult in family'
  dis='disability indicator'
  severe='severe disability indicator'
  hea='health indicator'
  fkind='kind of family'
  ms='marital status indicator'
  famhead='family head indicator'
  ftype='family type'
  marsupwt='March supplement final weight'
  horg='Hispanic origin'
  combexpc = 'Combined work expenses'
  newmoop='medical out-of-pocket expenditures'
  fmrcap='Capped housing subsidy based FMR'
  fmrcapce='Capped housing subsidy based on CE thresholds'
  mooplin1='MOOP part of CPI FSCUM threshold'
  mooplin2='MOOP part of CE FSCUM threshold'
  fcsulin1 = 'FCSU (Food Clothing Shelter and Utilities) part of CPI FCSUM threshold'
  fcsulin2 = 'FCSU part of CE FCSU threshold'
  def001 = 'Income definition 1'
  def002 = 'Income definition 2'
  def003 = 'Income definition 3'
  def004 = 'Income definition 4'
  def005 = 'Income definition 5'
  def006 = 'Income definition 6'
  def007 = 'Income definition 7'
  def008 = 'Income definition 8'
  def009 = 'Income definition 9'
  fmvfsl = 'free school lunch'
  fmvrsl = 'regular priced school lunch'
  rac_hisp='Race and Ethnicity recode'
  region='Region'
  citshp = 'Citizenship Recode'
  famkind = 'kind of family'
  famtype = 'type of family'
  age = '11 category age recode'
  fpers = 'Number of persons in family recode 1-10+'
  fpersr = 'Number of persons in family recode 1-7+';
run;


proc freq;
     tables poor1-poor13 
	 poor1-poor13*agecat
	 poor1-poor13*racer
	 poor1-poor13*hinstat 
	 poor1-poor13*race
	 poor1-poor13*hisp
	 poor1-poor13*adwkfmt
	 famtype*poor1-poor13*famkind
	 poor1-poor13*reg
	  ;
     weight marsupwt;
run;

proc contents data = two;
run;