Modified Dietz doesn't always do the job

We have a new client who wanted us to do a "non-GIPS(R) verification" of a composite. We do this type of work for clients who aren't able (or choose not) to comply with the Global Investment Performance Standards. We typically begin by understanding their rules since they're under no obligation to adhere to GIPS. We then validate that they have, in fact, abided by their rules. We obviously also test to see if their rules are appropriate and not misleading.

In this case we asked how they calculate returns and were told that they use Modified Dietz, which, of course, is a perfectly acceptable method to derive a time-weighted return (under most circumstances). In our client's case, because of their inability to value all portfolios monthly, they calculate the composite's return across a full year: that is, they value the composite (which is an aggregation of the member accounts) at the start and end of the year, and weight the flows across the year. Interesting.

The AIMR-PPS(R) actually allowed this approach for periods prior to January 1, 1993; GIPS has never addressed such a long period and only permits, at most, quarterly for periods prior to January 2001. And, we would argue that it's a far from appropriate approach given (a) the long period, (b) the market volatility, and (c) the presence of cash flows throughout the year. As my friend Carl Bacon likes to remind me, Modified Dietz is a money-weighted formula that achieves the status of being time-weighted by linking; in this case there would be no linking and the return is, essentially, a money-weighted one; hardly a viable way to measure a manager's performance.

We reached out to the client earlier today, and spent about a half hour explaining that while we could do this work, our report would have to include enough qualifying language to make the results less than ideal. They're now working on another approach.

Yes, Modified Dietz can work ... but not always!