Thursday, March 4, 2010

Attribution and GIPS

I just got off the phone with a verification client who wanted to discuss attribution and the requirements of the Global Investment Performance Standards (GIPS(R)). Many firms want to include attribution in their marketing materials, and this is great! It provides the recipient with an idea as to the source(s) of the firm's excess return and (hopefully) validates their claims.

There are no rules when it comes to GIPS and attribution, other than that attribution should be labeled as "supplemental information." You have two choices:
  1. show attribution of the full composite, where you treat the composite as a single portfolio
  2. show attribution for a "representative portfolio" within the composite.
Two problems with representative portfolios:
  1. Are they truly representative?
  2. What happens when they leave? You'll have to find another portfolio to use and link the results.
It's really up to you which way you go.  Some composite systems allow you to aggregate the portfolios so that your composite looks like a single portfolio while others don't provide this capability.

We were asked to provide some ideas on disclosure language:
  • Supplemental Information: the attribution results are of a representative portfolio which is a member of the composite. We believe these effects are representative of the composite, in general, as well as the other accounts within it.   
  • Supplemental Information: these results are of a representative portfolio which is in this composite. As with all representative portfolios there's a risk that the one selected presents the results the manager wishes to display. All portfolios within this composite are managed in a very similar manner, and therefore we believe these results truly represent the composite and all the accounts within it.
  • Supplemental Information: these results are of a representative portfolio which is in this composite. As with all representative portfolios there's a risk that the one selected presents the results the manager wishes to display. To avoid this we chose the portfolio with the longest history. All portfolios within this composite are managed in a very similar manner, and therefore we believe these results truly represent the composite and all the accounts within it
  • Supplemental Information: these results are of a representative portfolio which is in this composite. As with all representative portfolios there's a risk that the one selected presents the results the manager wishes to display. To avoid this we selected the portfolio at random. All portfolios within this composite are managed in a very similar manner, and therefore we believe these results truly represent the composite and all the accounts within it
I think any of these would be fine; full, complete, and accurate disclosure is always best. You don't want to mislead, so ensure that your language is accurate.

2 comments:

  1. Stephen Campisi, Client Portfolio ManagerMarch 6, 2010 at 10:13 PM

    I believe the guidance supplied here is quite reasonable and should be adopted. This would be a huge step forward in meeting the true intent of the GIPS standards: fair representation and full disclosure. Just as it is necessary to preset both the return of a composite along with information about the risk taken to earn that return, it is also necessary to understand both the excess return of the composite along with the source of that excess return through an attribution analysis.

    As an example, I recall hiring a global bond manager whose excess return was quite high. This raised questions about the investment process. It was only when an attribution of excess return was provided that the questions about the level of excess return were answered. And, it was a somewhat general answer that was the most helpful. The manager indicated that about 1/2 of the alpha was the result of active currency management. Given the volatility of currency, this was a reasonable statement. The other 1/2 of the alpha was the result of duration bets and sector bets. Not surprisingly, selection was not a big driver of their excess return.

    Without the attribution, the composite results would have been unpersuasive. With the attribution, we had confidence in the manager's track record. Five years later, we are happy with the manager, and the portfolio's results are in line with the expectations from their GIPS-compliant presentation. Should attribution be provided with the guidance provided in this blog posting? I definitely think so.

    ReplyDelete
  2. Attribution should be, one would hope, a standard component of any manager's marketing materials, just as we expect managers to comply with GIPS today. With GIPS 2010 we will have a requirement for risk; perhaps in the future we'll have a requirement for attribution, too. Or, at least a recommendation that it be provided.

    ReplyDelete

Note: Only a member of this blog may post a comment.