Transaction-based attribution: daily or monthly? Does it really matter?

At this year's PMAR (Performance Measurement, Attribution & Risk) North America and PMAR Europe conferences, I provided an update on research I'm doing, contrasting the results by using monthly holdings-based attribution versus monthly transaction-based attribution. The differences can be quite huge, not only in the residual typically found with the holdings-based approach, but also with the misassignment of attribution effects. More details will follow.

I pointed out that I believe that there will be no difference if we move to daily processing for the transaction-based approach; I am now reconsidering this, and will opine further as I do some exploring. Today I'll briefly explain the basis for this "180."

For the monthly approach, I use Modified Dietz, which in this context produces a money-weighted return. There are several of us (including Steve Campisi, CFA and Stefan Illmer, PhD) who have argued for money-weighted attribution (because the manager controls the subportfolio cash flows (e.g., buys and sells) and is, by virtue of the securities selected, responsible for the remaining internal flows (e.g., corporate actions and income).

To move to a daily application would, I believe, move us to a time-weighted approach, given the daily revaluing of the portfolio and its constituents. And so, it should (a) produce a different set of results and (b) arguably the wrong set. Wrong, that is, in Steve, Stefan, and my opinions (as well as others who have "seen the light" as far as the use of money-weighting for attribution).

Please look forward to further discussion on this both here and The Spaulding Group's monthly newsletter.