Thank you very much, I have this one working now ... View more

You refers to the first post? it would be this: Proc sql noprint;     Create table want as     Select date, portfolio, sum(value) as value,         sum(lag1(value)*return)/sum(lag1(value)) as return     from have group by portfolio, date; quit; return (t) = sum(value(t-1)*return(t))/sum(value(t-1)) so if I don't take the lag grouping those individual assets into portfolio works fine. But only for the return I have to take the lag ( previous month) value  ... View more

1. the weight is the value column 2. the R is the missing return, missing value. 3. No there are more different ids (about 15000 of them) 4. and the number of ids varies per portfolio. there are about 2500 different portfolios  ... View more

Thanks!! ... View more

The data look like this.      ID         date               value              return     portfolio I want to group them into the portfolio level so the monthly return of the portfolio would be the weighted return of the individual asset. The result table would be: portfolio     date               value                    return 200075     01/31/1997     138.449                    . 200075   02/28/1997    138.249     (0.149*0.005199 + 138.3*0.006061)/(138.3+0.149). ... ... View more

the left join - part did not work well. because the weight and return are all from previous day. The data I have are monthly data I should say previous obs above I think. The return is this month return and the weight is last month weight of different assets return (t) = sum(weight (t-1)*return(t))/sum(weight(t-1)) ... View more

yes, by lag I mean the previous date (t-1). ... View more

Thank you. ... View more

I only have sas 9.2 it did not work. ... View more