Thursday, February 16, 2012

More on the aggregate method

Fear not!

I haven't given up on my arguments against the use of the aggregate method to derive the extremely important composite returns (which are the bedrock of the Standards).

I was conducting a GIPS(R) verification earlier this week, and stumbled upon the following on page 6 of the 2010 edition of the Global Investment Performance Standards:

"The composite return
is the asset-weighted average
of the performance
of all portfolios in the composite."
[emphasis added]

But, as I have pointed out repeatedly, this does not hold for the aggregate method, which calculates the return of the composite, which can yield a hugely different result. And so, IF this IS the definition, why allow the use of a formula that violates it? We have two measures which DO satisfy this definition, and should be the ONLY ones permitted.

Can I get an "amen" on this?

Now, in reality, I favor equal-weighting, but asset-weighting won't go away. But we can at least adhere to the intended definition and calculate it properly, can't we?

p.s., I learned this form of writing from reading NBA Hall of Famer Dennis Rodman's autobiography (yes, I read it).

p.p.s., Well, actually, Susan Weiner was my inspiration.

4 comments:

  1. Dave,

    Thank you! I am glad to be a small source of inspiration for your blogging.

    ReplyDelete
  2. Susan, actually you're more than a "small" source. Thanks for your great ideas!

    ReplyDelete
  3. Stephen Campisi, CFAFebruary 16, 2012 at 9:20 PM

    When, oh when will the performance industry (and the Rules Makers and Affirmative Bodies Overseeing All That Is Right and Good in Performance Calculation Methodologies) begin their methodology discussions by FIRST making sure that they understand the question they are answering? From my vantage point, we are making NO progress in this regard. Instead, we now have "camps" that are hunkering down into their "positions" on the "right" way or the "best" way or (heaven help us) the "most accurate" way to calculate something... without demonstrating any understanding of what that "something" IS.

    Let's start with this: what question are you answering when you calculate a size weighted composite return? I believe that would be "What is the return on the average client dollar invested?" And what question is answered by an equal weighted composite return? Most likely it is the question that clients are really concerned about: "What was the return the average client received?" And the dispersion around this equally weighted return provides useful information about whether a) the investment process is administered and applied consistently as well as b) whether all clients are receiving fair and equal treatment. (Of course, this may also provide some information on whether the composite is appropriately constructed in terms of the guidelines around account inclusion.) So, if GIPS is meant to help clients answer the questions most relevant to the process of selecting a manager based on performance history for a group of like-managed accounts (not dollars) then it is clear that the equal weighted methodology is the appropriate one.

    I believe that the so-called "controversy" around composite calculation is more a reflection of "confusion" around what question is being answered. Hence the need for "change" - in this case, a change in perspective about what you should understand before making rules about how other people do things. Let's get it straight as to what things we are trying to do. Otherwise, this is just another case of "ready, fire... aim!"

    (Who came up with those "3 C's of Performance" anyway? It's pretty catchy.)

    As to the "Aggregate Method" - well, I'm still waiting for someone to explain what compelling question this answers. So far, I've only heard unsubstantiated claims that it is more "accurate" - whatever that is supposed to mean. After all, it's not accurate to answer the wrong question, even when you do so with great "precision." I wouldn't use a micrometer to measure the circumference of my head within a tolerance of 1/1000 of an inch if I really needed to know my shoe size. Seems like the Aggregate Method is the performance industry's micrometer...

    ReplyDelete
  4. Steve, excellent points. I'm looking at this "one step at a time." First, to put an end to the aggregate method, which produces a useless number. Then to move the equal-weighted approach to the same level as asset-weighted: i.e., make it a requirement. Step by step.

    Thanks for your input!

    ReplyDelete

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