From OGJAL@coba.ttu.edu Sun Jun 16 09:04:35 1996
Received: from COBA2.TTU.EDU (coba2.ttu.edu [129.118.49.47]) by gsb-pound.Stanford.EDU (8.7.5/8.7.1) with SMTP id JAA14873 for <doncram@sphinx.stanford.edu>; Sun, 16 Jun 1996 09:04:34 -0700 (PDT)
Date: Sun, 16 Jun 1996 11:05:34 -0500 (CDT)
From: TERRY  JALBERT <OGJAL@coba.ttu.edu>
To: doncram@sphinx.stanford.edu
Message-Id: <960616110534.20801352@coba.ttu.edu>
Subject: SAS program for beta calculation
Status: RO

title 'this program calculates Scholes-Wms betas from CRSP information';
options ls=80 ps=45;

/*

 Terrance Jalbert's beta calculation program, in SAS, which was 
 mentioned on CRSP-L.  Terrance points
 out this program may be very specific to his situation.  It appears to
 me to draw on CRSP data installed by a FORTRAN program (as it
 addresses the CRSP file missing value codes -66, -77, etc.) rather
 than by SAS PROC DATASOURCE (which recodes missing values as .P, .T
 etc., see p. 19 of my GSB Technical
 Report 80.  It implements a beta calculation formula from the CRSP
 manual, a formula which Terrance reports was listed incorrectly in
 many copies of the CRSP stock file guide dated December 31, 1993, at least.
  Don Cram

> > With regard to calculating beta's.  I have done this before using sas.
> > The code is not difficult, I just used the formula in the CRSP mannual.
> > I do want to caution you on one thing though.... The formula for
> > calculating betas is wrong in some versions of the CRSP documentation.
> > So you want to make sure before you  do it that you check with CRSP
> > to get the correct formula.  I would be happy to supply you with the
> > sas code, however; the code that I have is very much specific to the
> > application that I was using it for, and thus I am not sure how
> > usefull that it would be for you.
> >
> > With regard to missing days ect... I believe that they handle this by
> > not calculating the beta's for anything that does not have at least
> > 50 or 100 days of information.  I do not recall how they handle a
> > specific day where data was missing.  On the Bid ask spread question.
> > If I remember correctly, they use the average of the bid ask spread.
> > The precise method by which they do it is outlined in the mannual.
> >
> > Terrance Jalbert
> > Texas Tech Univerisity.
*/


data one;
  infile 'DISK$5:[CRSP.INDEX94]dsp500';
input
   ldate vwretd vwretx ewretd ewretx totval totcnt usdval usdcnt spindx sprtrn;
   drop vwretx ewretd ewretx totval totcnt usdval usdcnt spindx sprtrn;
if vwretd = -99.0 then delete;
if Ldate < 810101 then delete;
VWR=(1+ VWRETD);
LLVWR=LOG(VWR);
MRET=LAG(LLVWR);
date=lag(ldate);
l1=lag1(LLVWR);
l2=lag2(LLVWR);
mRET3=(LLVWR + l1 + l2);	
PRODM=(MRET*MRET3);

drop l1 l2  VWR VWRETD VWRETX EWRETD EWRETX TOTVAL TOTCNT USDVAL
     USDCNT SPINDX SPRTRN LDATE LLVWR;


proc sort;
  by date;

title 'this program will read the daily return info from my file information';
options ls=80 ps=45;
data two;
 infile mercdr1;
input cusip $ date dr @@;
if dr = -66.0 then delete;
if dr = -77.0 then delete;
if dr = -88.0 then delete;
if dr = -99.0 then delete;
DR1=(1+DR);
RET=LOG(DR1);
DROP DR DR1;
proc sort; by date;


data threeb;
   merge ONE(in=inone) two(in=intwo);
   by date;
   if intwo ne 1 then delete;
   IF DATE < 820101 THEN D1 = 1;
   IF DATE > 811231 AND DATE < 830101 THEN D1=2;   
   IF DATE > 821231 AND DATE < 840101 THEN D1=3; 
   IF DATE > 831231 AND DATE < 850101 THEN D1=4;
   IF DATE > 841231 AND DATE < 860101 THEN D1=5;
   IF DATE > 851231 AND DATE < 870101 THEN D1=6;
   IF DATE > 861231 AND DATE < 880101 THEN D1=7;
   IF DATE > 871231 AND DATE < 890101 THEN D1=8;
   IF DATE > 881231 AND DATE < 900101 THEN D1=9;
   IF DATE > 891231 AND DATE < 910101 THEN D1=10;
   IF DATE > 901231 AND DATE < 920101 THEN D1=11;
   IF DATE > 911231 AND DATE < 930101 THEN D1=12;
   IF DATE > 921231 AND DATE < 940101 THEN D1=13;
   IF DATE > 931231 AND DATE < 950101 THEN D1=14;
   IF DATE > 941231 AND DATE < 960101 THEN D1=15;
PRODR=(RET*MRET3);

data three; set threeb;
PROC SORT; BY CUSIP D1; 


PROC UNIVARIATE DATA=THREE NOPRINT;
var MRET3;
by cusip D1;
output out=outMRET3 sum=sum1; 


PROC UNIVARIATE DATA=THREE NOPRINT;
VAR RET;
BY CUSIP D1;
OUTPUT OUT=OUTRET N=N1 SUM=SUM2 STD=STD;

PROC UNIVARIATE DATA=THREE NOPRINT;
VAR MRET;
BY CUSIP D1;
OUTPUT OUT=OUTMRET SUM=SUM3;

PROC UNIVARIATE DATA=THREE NOPRINT; 
VAR PRODM;
BY CUSIP D1;
OUTPUT OUT=OUTPRODM SUM=SUM4;

PROC UNIVARIATE DATA=THREE NOPRINT;
VAR PRODR;
BY CUSIP D1;
OUTPUT OUT=OUTPRODR SUM=SUM5;

DATA FOUR;
  MERGE OUTMRET OUTMRET3 OUTRET OUTPRODM OUTPRODR;
  BY CUSIP D1;
  IF N1 < 50 THEN DELETE;
 TOP=((SUM5)-((1/N1)*(SUM2*SUM1)));
 BOTTOM=((SUM4)-((1/N1)*(SUM3*SUM1)));
 BETA=TOP/BOTTOM;
 DROP SUM1 SUM2 SUM3 SUM4 SUM5 TOP BOTTOM;
PROC PRINT U;