Performance Perspectives Blog

But there’s no difference in the numbers!!!

by | Oct 22, 2010

In addition to doing GIPS(R) (Global Investment Performance Standards) verifications, we also conduct “non-GIPS” verifications, for firms or individuals who can’t comply with the Standards, but still want their numbers reviewed. And occasionally we discover return methods which seem a tad irregular. For example:

  • a calculation that treats all cash flows as if they occur on the first day of the month (even when they don’t)
  • a Modified Dietz implementation that only values the portfolio once a year (i.e., that weights the flows across the full year).

You’re probably not surprised to learn that we reject these methods, and insist that the client use a more standard method. They are understandably frustrated, however, when they find that the results aren’t very different from what they had previously calculated. This can be because their composite has a lot of accounts, because there aren’t that many flows that occur, or for some other reason. One client, before fully implementing the change, did a sample test and found the results to be relatively close to what they previously had. This, of course, makes us a bit uncomfortable: do we say “okay, you’ve proven that your method works, so don’t worry about it” or “sorry, your method is still faulty regardless of the comparability of the results, so you have to extend the exercise to the full set of portfolios”?

One could no doubt construct an argument for either approach. We have taken the more conservative one and require the client to fully implement a more appropriate method. To “sign off on” a verification report that used an irregular method would lend credence to the method and serve as an endorsement for it, which would totally conflict with the industry’s clear aim to provide the most accurate information possible. We don’t insist that clients who don’t comply with GIPS adhere to the GIPS rules, but we do expect them to use methods that have been deemed acceptable.