Hi there - trying to do an attribution analysis for my work where the portfolio holds a few asset classes/sectors that the benchmark does not. Can I still perform an accurate attribution in this case?
Intuitively, it doesn’t make sense to me especially regarding allocation effect because with allocation you are multiplying the difference in sector weights between the port and bench by either 1) return of the sector in the bench or 2) the excess return of the bench sector over the overall bench return. In either case, since your benchmark doesn’t hold that sector your allocation effect will always be either zero or negative…
How do jumping handle this? Does the port and bench HAVE to have the same number of sectors?
Even if you do not hold sectors/asset classes that the benchmark holds, you should be able to calculate all of: allocation, security selection, and interaction.
For instance, I don’t see the problem of having sectors where allocation effect is negative. You’re simply losing out on sectors where the benchmark was able to generate alpha in having positive weights in certain sector/asset class.
Sure.
For the sectors your portfolio has that the benchmark hasn’t, the benchmark weight is zero; assume that the benchmark sector return is 0%.
For the sectors the benchmark has that your portfolio hasn’t, the portfolio weight is zero; assume that the portfolio sector return is 0%.
Hi thanks for the reply. Like, yes mathematically this is how you’ll have to do it. But intuitively, it doesn’t make a lot of investment sense in my mind because if the portfolio sector you allocated to (which the benchmark does not have) did better than even the overall benchmark return, you’ll still get a zero or negative number for allocation effect since you’re multiplying the difference in sector weights by either the benchmark sector return or the benchmark sector excess return vs the overall benchmark (depending on if you use Brinson-hood-beebower or brinson-fachler method).
Hi thanks for the reply. I’m not saying a negative allocation effect is bad, because I understand that is useful information. But in this particular case a negative allocation effect doesn’t make a lot of intuitive investment sense in my mind because if the portfolio sector you allocated to (which the benchmark does not have) did better than even the overall benchmark return, you’ll still get a zero or negative number for allocation effect since you’re multiplying the difference in sector weights by either the benchmark sector return or the benchmark sector excess return vs the overall benchmark (depending on if you use Brinson-hood-beebower or brinson-fachler method).