Wednesday, May 11, 2011

Equal-weighting ... can it be done?

I had a call yesterday with a client who wanted to know if they can show equal-weighted composites. Recall that GIPS(R) (Global Investment Performance Standards) requires composites to be asset-weighted (something which I have grown to consider perhaps not to be appropriate, but that will be taken up at length in an upcoming article). And so, is it permitted to show an equal-weighted composite?

Yes! Provided (a) you have an asset-weighted version which is your actual presentation, (b) that you don't use the equal-weighted version as the presentation, but rather, (c) label it as "supplemental information."

Firms can use supplemental information to go well beyond the requirements of GIPS. The rules deal with the materials GIPS requires, but not items outside of GIPS, other than provisions as outlined in the Supplemental Information Guidance Statement, which you should familiarize yourself with.


  1. Andreas, EXCELLENT comments and insights. I have found cases of exactly what you say, as performance folks might tell their sales staff that they cannot do something because it's not in GIPS. GIPS was never intended to limit what firms can do; rather, it provides a foundational report which all compliant managers are to provide, which allows easy comparison by anyone seeking a manager to serve their needs. If a manager feels that equal weighted returns provide a better representation of their performance (perhaps because one huge account dwarfs all others), then to do so is encouraged.

    You have offered a great message for all involved with the standards. Thanks!

  2. Because of John Simpsons correction, my arguments don't fly in this particular example. But I think my point has some general validity.

    If equal-weighting is a recommendation, then the argument can be used by the manager's client: "Dear investment manager, equal-weighting is a recommendation, why don't you report it?!"
    Maybe this is why you were asked about permissibility by the investment manager: Insted of having to mess around with data and calculate equal-weighted averages, or maybe showing a result that looks much less attractive (for example, because a large account had virtually no management fees). In this case, the same issue arises: the GIPS are used AGAINST the client.

  3. Andreas, I think your comments still have value, as firms often restrict themselves to what is required, even when prospects ask for "non-GIPS compliant" items.

  4. Stephen Campisi, CFAMay 14, 2011 at 11:25 AM

    The real question of clients is this: "Will I get the return of the composite?" To answer this question, we may depend on the measure of dispersion in the composite. But this raises an important second question: "What is the right measure of this dispersion - an equally weighted or a size weighted measure?"

    Clearly, the equally weighted measure is the appropriate measure because it addresses the client's true concern, which is better stated as this: "What is the dispersion of return among clients?" rather than the less useful dispersion requirement which answers the wrong question: "What is the dispersion of the average dollar invested?"

    It's clear that a size-weighted return and a size weighted dispersion measure are biased measures. The danger is that they skew the results toward the largest accounts, which may be given special treatment by the manager. One way to test this is to compare the size weighted statistics against the equal weighted statistics. I suggest that any meaningful differences between size weighted and equal weighted statistics are indications that larger clients may be receiving special treatment.

    The confusion and controversy around questions of equal weighting vs size weighted are the result of performance analysts failing to understand the questions being asked by clients. The GIPS standards are supposed to serve clients, not investment managers. I suggest that size weighted statistics do not serve clients; they serve investment managers by overemphasizing the larger accounts. In doing so, they minimize the risks that smaller clients face, such as unequal treatment.

  5. Steve, I think that the framers of the AIMR-PPS were fully convinced that both asset-weighted composite returns, as well as asset-weighted dispersion, make perfect sense. But you and I agree that they do not. GIPS dropped the emphasis on asset-weighted standard deviation, though many in the industry haven't caught on; we advise all verification clients that it's an inappropriate measure, and many have switched. As for asset-weighted vs. equal-weighted returns, this is a bigger battle for us, but another one we will pursue.


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