Example 11.9 CRSP Daily NYSE/AMEX Combined Stocks

This sample code reads all the data on a three-volume daily NYSE/AMEX combined character data set. Assume that the following filerefs are assigned to the calendar/indices file and security files that this database comprises:

The data set CALDATA is created by the following statements to contain the calendar/indices file:

   proc datasource filetype=crspdci infile=calfile out=caldata;
   run;

Here the FILETYPE=CRSPDCI indicates that you are reading a character format (indicated by a C in the 6th position) daily (indicated by a D in the 5th position) calendar/indices file (indicated by an I in the 7th position).

The annual data in security files can be obtained by the following statements:

   proc datasource filetype=crspdca
                   infile=( secfile1 secfile2 secfile3 )
                   out=annual;
   run;

Similarly, the data sets to contain the daily security data (the OUT= data set) and the event data (the OUTEVENT= data set) are obtained by the following statements:

   proc datasource filetype=crspdcs
                   infile=( calfile secfile1 secfile2 secfile3 )
                   out=periodic index outevent=events;
   run;

Note that the FILETYPE= has an S in the 7th position, since you are reading the security files. Also, the INFILE= option first expects the fileref of the calendar/indices file since the dating variable (CALDT) is contained in that file. Following the fileref of calendar/indices file, you give the list of security files in the order in which you want to read them. When data span more than one physical volume, the filerefs of the security files residing on each volume must be given following the fileref of the calendar/indices file. The DATASOURCE procedure reads each of these files in the order in which they are specified. Therefore, you can request that all three volumes be mounted to the same drive, if you choose to do so.

This sample code illustrates the following points:

• The INDEX option in the second PROC DATASOURCE run creates an index file for the OUT=PERIODIC data set. This index file provides random access to the OUT= data set and may increase the efficiency of the subsequent PROC and DATA steps that use BY and WHERE statements. The index variables are CUSIP, CRSP permanent number (PERMNO), NASDAQ company number (COMPNO), NASDAQ issue number (ISSUNO), header exchange code (HEXCD), and header SIC code (HSICCD). Each one of these variables forms a different key which is a single index. If you want to form keys from a combination of variables (composite indexes) or use some other variables as indexes, you should use the INDEX= data set option for the OUT= data set.

• The OUTEVENT=EVENTS data set is sparse. In fact, for each EVENT type, a unique set of event variables are defined. For example, for EVENT=’SHARES’, only the variables SHROUT and SHRFLG are defined, and they have missing values for all other EVENT types. Pictorially, this structure is similar to the data set shown in Figure 11.4. Because of this sparse representation, you should create the OUTEVENT= data set only when you need a subset of securities and events.

By default, the OUT= data set contains only the periodic data. However, you may also want to include the event-oriented data in the OUT= data set. This is accomplished by listing the event variables together with periodic variables in a KEEP statement. For example, if you want to extract the historical CUSIP (NCUSIP), number of shares outstanding (SHROUT), and dividend cash amount (DIVAMT) together with all the periodic series, use the following statements.

   proc datasource filetype=crspdcs
                   infile=( calfile secfile1 secfile2 secfile3 )
                   out=both outevent=events;
      where cusip='09523220';
      keep  bidlo askhi prc vol ret sxret bxret ncusip shrout divamt;
   run;

The KEEP statement has no effect on the event variables output to the OUTEVENT= data set. If you want to extract only a subset of event variables, you need to use the KEEPEVENT statement. For example, the following sample code outputs only NCUSIP and SHROUT to the OUTEVENT= data set for CUSIP=’09523220’:

   proc datasource filetype=crspdxc
                   infile=( calfile secfile)
                   outevent=subevts;
      where cusip='09523220';
      keepevent  ncusip shrout;
   run;

Output 11.9.1, Output 11.9.2, Output 11.9.3, and Output 11.9.4 show how to read the CRSP Daily NYSE/AMEX Combined ASCII Character Files.

   filename dxci "dxccal95.dat" RECFM=F LRECL=130;
   filename dxc "dxcsub95.dat" RECFM=F LRECL=400;
   
   /*--- create output data sets from character format DX files ---*/
   /*- create securities output data sets using DATASOURCE  -------*/
   /*- statements                                                 -*/
   proc datasource filetype=crspdcs  ascii
                   infile=( dxci dxc )
                   interval=day
                   outcont=dxccont
                   outkey=dxckey
                   outall=dxcall
                   out=dxc
                   outevent=dxcevent
                   outselect=off;
      range from '15aug95'd to '28aug95'd ;
      where cusip in ('12709510','35614220');
   run;
   
   title3 'DX Security File Outputs';
   title4 'OUTKEY= Data Set';
   proc print data=dxckey;
   run;
   
   title4 'OUTCONT= Data Set';
   proc print data=dxccont;
   run;
   
   title4 "Listing of OUT= Data Set for cusip in ('12709510','35614220')";
   proc print data=dxc;
   run;
   
   title4 "Listing of OUTEVENT= Data Set for cusip in ('12709510','35614220')";
   proc print data=dxcevent;
   run;

Note in Output 11.9.4 that there were no events in range for cusip 35614220. See Chapter 33, The SASECRSP Interface Engine, for more on CRSPAccess Data access.