Friday, September 23, 2011

What does it mean?

Earlier this month I posted a question regarding what the definition of "composite return" is. Given the importance of this value within the GIPS(R) standards (Global Investment Performance Standards), there should be an answer.

I promised to expand on this subject further, and have done so in this month's Spaulding Group newsletter. To me this is a fundamental issue. Please let me know if you disagree.


  1. David, another thought provoking edition of the Perspectives, but I think that it would benefit from a more accurate reflection of history.

    You state that in the first edition of the AIMR-PPS, there is no definition of what a composite return represents and that only the BMV method is offered as a method of calculation. In my copy of the 1993 AIMR-PPS Book (page 27), there is a very clear description of what a composite return is intended to be:-

    "a single value that reflects the overall performance (the "central tendancy") of the set. The objective in reporting the returns of composites is to use a method for reporting the composite return that will give the same value achieved if the composite were treated as one master portfolio. That is, the value being calculated is the same value that would result if all of the assets and transactions of the individual portfolios/classes were combined and the return were computed using the procedures discussed earlier."

    On this basis, one clear vote for the aggregate method.

    The 1993 book then sets out four methods which might be used to obtain a composite return. These are the Equal Weighted method, the Asset-weighted return (BMV weights), the Asset-weighted and cash-flow weighted return method, and finally the aggregate method. The standard does require that one of the 3 weighted method returns must be presented, although equal weighted can be provided in addition.

    The 1997 book (page 48) is essentially unchanged.

    Although the first edition of GIPS simplified the language within the standards themselves, within the explanations included in the first edition of the GIPS Handbook much of the language and all of the intent is the same as that in the preceding AIMR-PPS books. The only significant change is that the equal weighted method is not longer described.

    Critically, the 2005 edition handbook still includesthe description that "the objective in calculating the composite's return is to use a method that will produce the same value as if the assets of all the individual portfolios in the composite are aggregated and a return is calculated for the one 'master portfolio'. The composite return is the asset-weighted average of the performance results of all the portfolios in the composite."

    Personally, I agree with your assertion that the weighted average of returns of all accounts in the composite for the entire period (monthly calculations), is a more meaningful measure of composite performance, but by your own standards it is not the one which best meets the existing clear definition of what a composite return is intended to present.

    There are many users of the standards who have not been around since the AIMR-PPS were first introduced, or even since that standard was replaced by GIPS, but it is a duty of those who have been to ensure that we represent the history of the standards accurately. Hopefully my comments will help to put the record straight.

  2. Brian, thanks for adding clarification. The "definition" that you found wasn't evident to me, as I explored, and is obviously lacking from the GIPS standards. I do recall discussions pre-1993, when IMCA and the ICAA voiced opposition to using asset-weighting: AIMR said at that time that they wanted a return that made the composite look like a portfolio. At that time I didn't give this matter much thought at all. As to it being the "central tendency" is perhaps open to question, regardless of the method employed, but that's not important or worth any further discussion.

    You are correct that if THIS is the definition, then aggregate is the way to accomplish it, no doubt, which of course would cause wonder as to why it wasn't (a) introduced in '93 (when the BMV method clearly doesn't accomplish this goal) and (b) why it's therefore not the only method permitted!

    I appreciate your support for my belief that the weighted (non-aggregate) methods are better. Whether or not the EC will even bother considering these points is unknown, and in my view, unlikely, which is also unfortunate.

    As to the "four" methods set out, while you are correct that they reference equal-weighting, this was the method that the ICAA and IMCA pushed for and as you state, it's not permitted as a compliant method. Since it's recommended within GIPS, it also isn't "supplemental." I intend to argue at a later point that in reality, equal-weighting SHOULD be the method, but first things first.


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