My favorite source for word meanings, Dictionary.com, provides the following for "insight":
- an instance of apprehending the true nature of a thing, especially through intuitive understanding
- penetrating mental vision or discernment; faculty of seeing into inner character or underlying truth.
To me, performance attribution is a huge source for insights, as it allows for the reader (to paraphrase the above definition) to apprehend the true nature of the source(s) of the return or excess return. By employing a flexible attribution model, one can twist and turn their portfolio about, seeing more than would otherwise be visible (for example, Ron Surz's "attribution with style" is just one way to look at the data from a different perspective (see below for this word) to gain additional insights into what's going on). We can penetrate the data and discern what is occurring, seeing into the inner character of the portfolio and the truth within.
In conducting performance attribution, we should strive to gain as many insights as possible, so that we are able to fully comprehend what has occurred.
The website dictionary.com offers a variety of meanings for this word, and the following are best for our purposes:
- the state of one's ideas, the facts known to one, etc., in having a meaningful interrelationship.
- a way of regarding situations, facts, etc, and judging their relative importance.
- the proper or accurate point of view or the ability to see it; objectivity.
A GIPS(R) verification client of ours asked me to review supplemental information that they propose including with their composite presentation. What they've done is broken their portfolio up into various sectors. These are not "carve-outs," because they don't contain cash, and if they did, it wouldn't have resulted from the cash being managed separately. It is quite common to see managers show the returns of asset classes (e.g., equities, bonds, cash), but for GIPS purposes this would be considered "supplemental." In our client's case, they break the data up into several categories, not necessarily mutually exclusive (see below).
I am gaining my own insights into this information, and realizing it caused me to consider performance attribution a bit more. I don't recall if it's ever been stated before (perhaps because it seemed obvious), but when we conduct attribution, the portfolio's components (e.g., sectors) which we analyze must be mutually exclusive; i.e., we couldn't have the same security sitting in two sectors simultaneously.
I liken what our client does to a form of attribution, since it provides the reader some insights into where the returns are coming from; however, as already stated, it wouldn't qualify as traditional attribution, given that it fails the "mutual exclusivity" requirement.
A question also arises regarding what is being presented; what's the perspective? If, for example, we're looking at U.S. large cap stock performance, is the manager saying "this is how the large cap sector of the portfolio performed" or "if I were to manage a large cap strategy, this is how I would have done"? These are vastly different, are they not?
Consider just a few of the differences:
- Cash: with the former, the exclusion of cash is appropriate, since we are only concerned with showing how the large cap stocks within the portfolio performed; with the latter, we'd expect to see cash present, since there would bound to be cash in a large cap portfolio, the relative composition of which might vary from time to time.
- Gaps: I have written in the past that I can see one justifying crossing gaps in performance, if you want to show how the avoidance of a sector (e.g., banking) might have benefited the overall performance, vis-à-vis the benchmark. And so, for the former I'd permit gap-crossing. However, for the latter I wouldn't, since the likelihood of someone managing a portfolio solely devoted to banking or any strategy going completely empty, would be extremely unlikely (I'm open to how you think this might occur); perhaps this would take place if the client decided to temporarily shift all their funds away from that strategy, but the resulting gap wouldn't be crossed.
- Makeup: for the former, we wouldn't necessarily have any expectation for a "minimum number of securities." However, for the latter, since it is equivalent to a portfolio being managed to a strategy, we would expect to see several securities present. While one or two might be okay for the former, they most likely wouldn't for the latter.
- Return methods: in the former case we would push for money-weighting, since the manager is controlling the cash flows; in the latter, an argument could be made for time-weighting, for if this were a separate portfolio, even though the cash flows might be under the control of the manager, we typically see time-weighting employed here.
Insights and perspectives: two words that really do have a lot to do with performance measurement. I may have more to say on this in our September newsletter, as I think it's an interesting topic; hope you do, too!