Hey, I think I might know what the problem is then. The shown here was only a simplification of the actual data. This is how my data looks: data have; input pcusip date permno ret cap anncmt code shrout prc; datalines; 1 19990202 a 0.01 400 19990209 3 100 4 1 19990202 b 0.04 100 19990209 3 100 1 1 19990203 a 0.01 400 19990209 3 100 4 1 19990203 b 0.04 100 19990209 3 100 1 2 20000101 c 0.01 800 20000310 3 400 2 2 20000102 c 0.02 800 20000310 3 400 2 3 19300812 d 0.03 750 19300810 3 750 1 3 19300813 d 0.03 750 19300810 3 750 1 ; My Code: proc sql; /*Capital weight the returns of same CUSIP with multiple permnos on same date and combine the permnos to one*/ create table want as select pcusip, date, code, anncmt, sum(shrout) as shrout, sum(prc) as prc, min(permno) as permno, /*for the two cases in this sample the smaller permno is the one that existed first*/ sum(ret*cap)/ sum(cap) as ret, sum(cap) as cap from have group by pcusip, date; select * from want; quit; From your comment I guess the problem are the variables anncmt & code right? Since they are not part of any calculations but also are not in the group by statement. Is there any way around this problem? Thank you for your help once again. BR nici
... View more
