Tuesday, December 27, 2011

New Year's Resolutions

It's that time of year when folks begin to think about things they want to change come the new year.

It occurred to me that wisdom from one of the most renowned and powerful Jedi Masters, Yoda, might serve us well. After all, he was known for his legendary wisdom and mastery of the Force, not to mention his skills in lightsaber combat!

You must unlearn what you have learned

This is a tough one, I must confess. We tend to hold on to beliefs that we learned early on, without regarding the possibility that there could be any other way. And, I must confess that I have been guilty of this shortcoming too many times, myself.

I hardly expect that we will all agree on one set of beliefs, and there is always room for different approaches. But to at least be open to why one approach might be better than another would be wise. Here is an abbreviated list of areas where one might consider exploring the alternative(s):

Rates of return: be open to employing money-weighted returns. Why must time-weighting dominate?

Attribution: arithmetic versus geometric. Here is one area where I must try to see the benefit of geometric.

GIPS(R) (Global Investment Performance Standards): I invite others to see that the two broad approaches to derive composite returns (using beginning market values (with or without weighted flows) versus the aggregate method) yield two completely different results, meaning two different definitions for this value. In addition, for more PMPs (Performance Measurement Professionals) to see that equal-weighting is superior to asset-weighting.

Risk: be open to exploring alternative metrics, both ex post and ex ante.


  1. Stephen Campisi, CFADecember 27, 2011 at 5:33 PM

    I think this is one of your best and most influential postings, clearly applicable to all of us. While it's difficult to change, it's clear that none of us will do so unless we see the benefit of such change. In terms of investment performance, we should identify something new that a new approach will provide, or perhaps something that we can do better as a result.

    For those looking to see the benefit of geometric attribution, let me suggest that the ability to reconcile attribution of return with changes in levels of wealth (AKA "money") is a terrific benefit. Seems like something that should be of benefit to you, since you have advocated money weighted returns as something for others to embrace. A little "quid pro quo" no?

  2. Thanks, Steve for your comments. As for geometric, I will attempt to be more open to the benefits, but don't hold out too much hope for me!

  3. Some additions to your list of "things worth unlearning:"

    1. Understand the question you're trying to answer BEFORE pulling out the calculations. Turns out that the "best" calculation or performance analysis depends on the question you're trying to answer. One is not inherently "better" than the other, but one is clearly more "appropriate" than the other - meaning that it answers the question you're asking. The easiest and most relevant example of this is time weighted vs money weighted returns. The first answers questions about manager skill, while the second measures portfolio return. Two different questions with the most appropriate or "best" calculation for each.

    2. The Interaction Effect:
    After more than two decades, performance analysts are still desperately trying to explain its relevance to investment managers and clients. It's time to recognize that this is simply a very large residual that is really part of the selection impact upon the portfolio. It is not "bad" but it's not been used correctly. It's use is really to break out the pure analyst impact on the overall portfolio selection effect.

    3. The Brinson-Hood-Beebower model:
    This one is simply bad, which is why no one really uses it. So, why hold onto it? We can save face by quoting the authors' comment that it was never intended to be used at the sector level, but rather as a means of measuring total portfolio attribution effects. At the sector level, this confuses the allocation effect and simply degenerates into a subtraction of contributions to return (portfolio vs benchmark.) One needs only primer school algebra to prove this.

    4. Daily attribution:
    A silly waste of time, since portfolios (in real life) rarely change substantially from day to day.

    5. Issue level attribution:
    This has no real meaning. Investment managers and clients sometimes need issue level "contribution to return" and this is easy to calculate.

    I suspect that vendors will be the last to try moving past some of these errors in thinking that have come to be accepted as "high quality" in the performance world (as though more is always better.) However, this deserves serious though by us all. We could be using our resources more effectively to serve our clients.

  4. Steve, thanks for chiming in. And I will comment briefly. As to the interaction effect, I don't know of anyone "desperately trying to explain it," but perhaps there are some folks; to me it's simple: it's the interaction of two decisions, allocation and selection. To properly understand how each on its own behaves, we must observe them in their purest state, without any carryover from the other; this leaves us with the interaction effect. How's that? As to your #1, I agree; excellent point. As to BHB (#3), actually too many folks use it. I take some credit in bringing to light the differences between it and the BF model; recall that the late Damien Laker went to great lengths to say there were no differences; on this point he was mistaken. The only benefits to daily attribution, in my view, are (1) when one uses a holdings-based approach, to reduce the residual and (2) on those rare occasions when someone needs to discover something going on within the month. And finally, to your final (#5) point, agreed! I have been arguing against this for quite some time, and am pleased to have you as an ally!



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