/* compseg.sas

A SAS program to read the Compustat Industry Segment file.  This data
file is not supported by the SAS PROC DATASOURCE command, at least not in SAS
versions 6.12 and below.  Haim Mozes at Fordham University kindly
shares this program, in response to a request on the CRSP-L.  This program
has been edited only by reformatting and the addition of an endsas statement
to finish the program.   Don Cram 6/4/97.

NOTE: This program as written may or may not do an adequate job of
replacing Compustat Universal missing value code of -.001 by SAS
missing value code.  Haim checked whether Assets and other variables
are less than zero rather than checking if the values = -.001
directly.  So I have commented out his missing value replacement code
and replaced it.  Don Cram 9/8/97.

From: Haim Mozes
Date: Wed, 4 Jun 1997
Subject: the SAS program to read the segment file

I am sending you the SAS program.  A few things you should know:
 1 - The program was writtten for a VAX system.
 2 - The program reads the file from disk. You have to dump it from
     the tape to the disk first.
 3 - I dropped some variables, because I didn't need them for my
     purposes. To change this, just alter the keep/drop statement.
*/

filename compseg 'dsk10:[gbapgms]new280sic.dat';    
libname segments 'dsk10:[gbapgms.util]'; 

data segments.segment ; 
    infile compseg lrecl=774;
    input dnum 1-4 cnum $ 5-10  cic 11-13 stk 14 smbl $ 15-22  file 23-24
        zlist 25-26 @55 coname $ 55-82  ssrce 83-84 fyr 85-86 year 87-88 
        fundf 89-94
        sucode 95 sid 96-97 segn 98-99 sname $ 100-127 pname1 $ 128-147 
        pname2 $ 148-167 pname3 $ 168-187 pname4 $ 188-207 cname1 $ 208-223  
        cname2 $ 224-239 
        cname3 $ 240-255 cname4 $ 256-271 cname5 $ 272-287 cname6 $ 288-303 
        ssic1 304-307 ssic2 308-311 psic1 312-315 psic2 316-319 
        psic3 320-323 psic4 324-327 @428  sales 10.3  opprofit 10.3  
        deprec 10.3 
        capex 10.3  identass 10.3  eqincome 10.3  
        eqinvest 10.3  nworkers 10.3  backlog 10.3  
        rDcust 10.3  rdcomp 10.3  dgovrev 10.3   
        fgovrev 10.3  csale1 10.3  csale2 10.3 csale3 10.3 
        csale4 10.3  psale1 10.3  psale2 10.3 
        psale3 10.3  psale4 10.3  
        @734 foot20 $  774  ;

/*
    if backlog not >  0 then backlog=.;
    if sales not > 0 then sales=.;
    if capex  not > 0 then capex=.;
    if rdcomp not > 0 then rdcomp=.;
    if identass not > 0 then identass =.;
    if nworkers not > 0 then nworkers =.;
                          
    if backlog = -.01 then backlog=.;
    if sales  =-.01  then sales=.;
    if capex  =-.01 then capex=.;
    if rdcomp =-.01  then rdcomp=.;
    if identass =-.01  then identass =.;
    if nworkers = -.01 then nworkers =.;
    if opprofit = -.01 then opprofit=.;
                          
    if backlog = -.001 then backlog=.;
    if sales  = -.001  then sales=.;
    if capex  = -.001 then capex=.;
    if rdcomp = -.001  then rdcomp=.;
    if identass = -.001  then identass =.;
    if nworkers = -.001 then nworkers =.;
    if opprofit = -.001 then opprofit=.;
*/                          
    roa=opprofit/identass;
    pm=opprofit/sales;   

 /* Replace Compustat Universal missing value code of -.001 by SAS missing 
 value code.
 Programmers note that for IBM360/370 format Compustat data, the missing 
 value code to be replaced is .0001.
 See Compustat Technical Guide and/or CRSP-L FAQ for explanation.
 */

        array numerics {*} dnum cic stk file
            zlist ssrce fyr year fundf
            sucode sid segn
            ssic1 ssic2 psic1-psic4
            sales opprofit deprec capex identass eqincome
            eqinvest  nworkers  backlog  
            rDcust  rdcomp  dgovrev   
            fgovrev  csale1-csale4 
            psale1-psale4;
        do i = 1 to dim(numerics);
            if numerics{i} = -.001 then numerics{i} = .;
            end;
        drop i;

        drop pname1 pname2 pname3 pname4 cname1 cname2 cname3 cname4 cname5
            cname6 foot20  ssic2 psic1 psic2 psic3 psic4 eqinvest 
            dgovrev fgovrev fundf sucode prevd  file zlist  ssrce eqincome
            rdcust rdcomp smbl;
                                       
endsas;