Thursday, December 30, 2010

What if we were wrong all along?

Funny what we can do with Google: type in just about anything and you'll be whisked away to one or more potential sites to provide you with information related to your inquiry. I entered the title for this post and was sent to a Yahoo site that addresses this topic from a creative writing / English lit perspective. This has little (make that no) relevance to my reason for using it.

Rather, I've begun to hint at this for some time, with my various rants and analyzes.From my questioning the almost universal love affair we have with time-weighting (at the expense of money-weighting), to my questioning of the acceptability of the aggregate method for composite returns, and most recently my challenge of asset-weighting of composites (versus equal-weighting) (the last two, of course, relate to GIPS(R) (Global Investment Performance Standards)), I've clearly been asking this question, though not necessarily with these words.

I don't set off with the intent of upsetting the apple cart, but rather innocently stumble upon these epiphanies. The word "epiphany" has a variety of meanings, with one being "a sudden, intuitive perception of or insight into the reality or essential meaning of something, usually initiated by some simple, homely, or commonplace occurrence or experience." This one fairly accurately describes how I've come upon the realizations that what we do, sometimes, appears wrong. Because I arguably have no skin in the game, that is, I never came up with any of these approaches, I am not required to say "heh, I was wrong." But, the reality is that at one time I supported all of these approaches but of late have come to question them. And so, yes, I have been wrong in supporting them.

This being my last blog post for 2010 I thought it fitting to close out the year by simply reminding you of my questioning. There are some in our industry who, for whatever reason, refuse to even consider a fresh look at what we have done for years and have accepted as perfectly acceptable ways of operating. I know there are some, too, who grow frustrated and impatient when some, such as I, raise these issues. But many others find these ideas refreshing. 

Just this past week I was asked to opine on a client's presentation where they, independent of my analysis, decided that the aggregate method was inappropriate for what they were doing: I was ecstatic to see this and so of course enthusiastically supported their work product. Others, too, have joined me in my crusade to see that money-weighting achieve its rightful place in our industry. 

I sent a letter to the GIPS Executive Committee a couple months ago, challenging the employment of the aggregate method. Although I understand they have discussed my paper I have yet to hear what their reaction is. And while I don't hold out much hope of seeing the method either removed from the standards as an acceptable approach (my preference) or at least carry a "warning label," anything is possible. None of the EC's members were involved in sanctioning this method, and so there should be no "pride of authorship" to influence them. 

I may send them another letter requesting a fresh look at the weighting approach for composite returns. Steve Campisi's support in response to my recent post enhanced my confidence that I was perhaps on to something. Here, too, the EC has no members who championed this approach when introduced by the FAF in the mid-80s and AIMR in the early '90s, so there should hopefully be complete objectivity and a willingness to step back and ask, "what if we were wrong all along?" That's my hope. 

1 comment:

  1. David

    You say that “monthly is sufficient for transaction-based attribution… By using the beginning sector weights, plus weighted flows (just as we do with Modified Dietz) we an accomplish our objective.”
    But, consider a fund that starts the month with $100 of cash and only one share of issue z in sector Sz priced at $100. On the first business day of the 20-day month the share price increases to $110 and the fund sells the share into (i.e. purchasing) cash. No more trades are done this month and cash receives no interest.

    In general, Modified Dietz weights and returns are
    Wj = (Voj + Wpj*Pj – Wsj*Sj)/Vo
    Rj = (Vfj – Voj + Sj - Pj)/ (Voj + Wpj*Pj – Wsj*Sj)

    Using Modified Dietz for the month we have
    For cash:
    Wc = (100 + 0.95*110)/200 = 102.25%
    Rc = (210 – 100 + 0 – 110)/(100 + 0.95*110) = 0

    For the issue z and, thus, for sector Sz:
    Wz = (100 – 0.95*110)/200 = -2.25%
    Rz = (0 – 100 + 110)/ (100 – 0.95*110) = -222%

    Rolling up to the fund level we find:
    Wc + Wz = 100% and
    R = (102.25%)*0 + (-2.25%)*(-222%) = 5%,
    Which is the correct fund level return: (210 – 200)/200 = 5%.

    Thus, in this long-only fund, the cash weight is greater than 100% and the issue and sector weight is less than zero. Also, this issue and its sector that did so well are assigned a return of less than negative 100%.
    Despite the fact that these weights and returns roll up to the correct fund-level values for the month, it is easy to see how such strange values for the weights and returns of the cash and the issue/sector can cause havoc in the subsequent attribution calculation that motivated this question. Also, matters get even worse if we tried to link sector Sz’s return of -222% for this month to its return for the next month where, say, we buy some other issue in the Sz sector and keep it till the end of the month where it achieves a positive return. The next month’s positive return makes the two-month period to date return even more negative.
    Quite reasonable examples like this lead me to think that it is not the case that the Modified Dietz approach to performance applied “monthly is sufficient for transaction-based attribution.”



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