The GIPS(R) standards require compliant firms to include a measure of dispersion for each year in which they have six or more accounts present for the full 12 months. But, which return should they use, gross-of-fee or net-of-fee? What about for the 36-month, ex post, annualized standard deviation? Gross or net-of-fee?
The ideal return to use is …
Okay, so you won’t have to wait long for the answer:
Perhaps we should begin with some background.
WHY do we report dispersion? [I’ll give you a few seconds to reflect on this]
To provide the prospect with a sense of how consistent the manager was across the various accounts within the composite. That is, is the strategy implemented in a fairly consistent manner, or might there be wide differences from one account to another?
Back in 1997, the AIMR Performance Presentation Standards (AIMR-PPS) introduced its second version, and this is where dispersion first became a requirement. Prior to, a firm need only report returns. BUT, seeing a return and having some confidence that, had you been a client you would have gotten it, can be quite different.
And so, by mandating the reporting of dispersion, we can see whether the manager is quite consistent (with a lower standard deviation or a rather low range) or “all over the place” (with a wide range or large standard deviation).
So, what’s my point?
With dispersion we’re interested in gaining some insight into the degree of consistency of the strategy’s implementation. This can best be done with gross-of-fee returns.
And why do I say that?
Well, net-of-fee are gross-of-fees that have been netted down by the fee. And, for many managers the fee schedule can be quite wide, meaning that the resulting net-of-fee returns can, by virtue of fee differences, cause wider dispersion than is correct. What if all accounts were managed exactly the same, so that their gross-of-fee returns were identical. However, if the accounts pay various fee levels, their net-of-fee returns would be quite different. Therefore, seeing a rather large dispersion wouldn’t convey anything about the actual portfolio management.
This thought occurred to me recently while conducting a verification for a client. I was trying to see what series of returns they were using for their standard deviation. It became pretty clear it was gross-of-fee. And, by calculating standard deviation for both sets of returns, I could see some wide differences. For example, for one composite the standard deviation was 0.36 for gross-of-fees, but when done against net-of-fees it was 1.06. QUITE a difference, yes?
What value is it for the prospect to see the net-of-fee dispersion? Essentially none.
What about the 36-month, ex post, annualized standard deviation: which return to use?
I would again recommend gross-of-fee, but for a slightly different reason.
If a client is paying a fee, chances are that fee percentage doesn’t change very much. And so, the collection of account returns wouldn’t change much as a result of fees, except for the following: when they’re actually paid!
Most managers assess fees quarterly. And so, each quarter accounts are hit with a fee, resulting in a drop in performance relative to gross-of-fee. For the first two months, the gross- and net-of-fee returns are identical. But, when the fee hits, there’s a drop, which introduces volatility, which says nothing at all about the market.
And so, if you’re using net-of-fee returns in your 36-month standard deviation, you’re introducing a source of volatility that can mislead the prospect and hurt the manager. And so, ideally use gross-of-fee returns here, too!
Does this make sense? Please chime in!