Here's a statement of the obvious: The opinions expressed here are those of the participants, not those of the Mutual Fund Observer. We cannot vouch for the accuracy or appropriateness of any of it, though we do encourage civility and good humor.
I'm mulling over both Mr. R's piece and the underlying research. If I were prone to be skeptical, the skepticism would begin with the observation "researchers from a firm whose bloated funds have low active shares report that low active shares are good!" The fact that the Fidelity researchers seem to be resorting to the rhetorical tool of renaming an active share effect as "blowup potential" kinda suggests a "start with the marketing mandate and work backward from there."
If I were prone to be skeptical, the skepticism would begin with the observation that "an enthusiastic supporter of active shares is using circumstantial ad hominem to cast doubts on the criticism of active shares".
But I am reading both sides.
Finding it difficult to get past the definition..
What is the active share measure of a fund that is all in cash? Is it really the same as that of a fully invested index fund or am I missing something?
If that is really the case, I can't see how a measure that gives absurd results at boundary conditions can said to be sound enough to draw any inference from.
What is the active share measure of a fund that makes money by rotating from one overvalued stock in the index to another undervalued stock in the index in a regular fashion?
What is the active share measure of a fund whose manager regularly goes to significant cash in overvalued conditions?
Isn't the problem that Fidelity researchers seem to be pointing out that the inferences made from active shares are confusing dubious correlation for causality?
I haven't looked at this measure before but there does seem to be a number of conceptual problems in its formulation.
Perhaps, some of you have studied this more thoroughly. Would appreciate comments or clarifications on the above.
@davidrmoran, thanks for the suggestion. This might help get over the inertia for creating an account at M* or reviving the very old one I had. Rekenthaler might be objective about this, since it didn't come from M*. He certainly doesn't subject his own flawed investor return metric interpretations to the same level.
It's also exceedingly straightforward to alter the active share equation to include cash
Care to suggest a formulation?
Have been thinking about this and it should be simple if there is an agreement on what the concept is supposed to capture. It seems like there are two conflicting intuitions behind the concept and conflating them seems to be part of the problem.
If the intuition is to capture a meaaure of how much a fund differs from a corresponding benchmark in its composition (regardless of what it implies), an all cash portfolio should be 100 in the metric measurement.
If the intuition is to capture a measure of how "active" the fund is in trying to generate returns different from the benchmark, then an all cash fund should be 0 as calculated in the current equation.
Fidelity's objection seems to be a consequence of assuming that the active share metric tries to capture both of the above intuitions and hence the lion/lamb discrepancy.
Still trying to think through this. Just some initial thoughts.
I should read the underlying papers before giving you a response, but my initial mechanistic one using the definition in our active share section is to define cash as an asset that indexes have a weighted allocation of 0% to, while managers can hold it at allocations between 0% and 100%. You don't change the equation at all, you just make sure that the W(fund,i) are all measured relative to fund AUM and include cash as an asset.
As to the goal of active share measurement, my intuition would be to assume that differing from your corresponding benchmark is active choice made in order to generate current returns (for lack of better options) or future returns (as reserve in anticipation of future opportunities or against future withdrawals at inopportune times).
@jlev, I full agree with you that cash can be active and can have positive relative performance over losing stocks!
The problem is that you might then lose any correlation between performance and active share measure because funds with high cash that do so for managing volatility will have much less performance (all cash fund will do poorly over a long period) and most funds with cash positions fall in this category while the smaller number of funds that go to cash tactically may generate alpha.
This seems to be a fundamental problem with the metric. Being different from the index when cash is considered may not have much correlation with any alpha generated by a fund.
Comments
But I am reading it.
David
But I am reading both sides.
Finding it difficult to get past the definition..
What is the active share measure of a fund that is all in cash? Is it really the same as that of a fully invested index fund or am I missing something?
If that is really the case, I can't see how a measure that gives absurd results at boundary conditions can said to be sound enough to draw any inference from.
What is the active share measure of a fund that makes money by rotating from one overvalued stock in the index to another undervalued stock in the index in a regular fashion?
What is the active share measure of a fund whose manager regularly goes to significant cash in overvalued conditions?
Isn't the problem that Fidelity researchers seem to be pointing out that the inferences made from active shares are confusing dubious correlation for causality?
I haven't looked at this measure before but there does seem to be a number of conceptual problems in its formulation.
Perhaps, some of you have studied this more thoroughly. Would appreciate comments or clarifications on the above.
Have been thinking about this and it should be simple if there is an agreement on what the concept is supposed to capture. It seems like there are two conflicting intuitions behind the concept and conflating them seems to be part of the problem.
If the intuition is to capture a meaaure of how much a fund differs from a corresponding benchmark in its composition (regardless of what it implies), an all cash portfolio should be 100 in the metric measurement.
If the intuition is to capture a measure of how "active" the fund is in trying to generate returns different from the benchmark, then an all cash fund should be 0 as calculated in the current equation.
Fidelity's objection seems to be a consequence of assuming that the active share metric tries to capture both of the above intuitions and hence the lion/lamb discrepancy.
Still trying to think through this. Just some initial thoughts.
As to the goal of active share measurement, my intuition would be to assume that differing from your corresponding benchmark is active choice made in order to generate current returns (for lack of better options) or future returns (as reserve in anticipation of future opportunities or against future withdrawals at inopportune times).
"Doing nothing is also a choice."
The problem is that you might then lose any correlation between performance and active share measure because funds with high cash that do so for managing volatility will have much less performance (all cash fund will do poorly over a long period) and most funds with cash positions fall in this category while the smaller number of funds that go to cash tactically may generate alpha.
This seems to be a fundamental problem with the metric. Being different from the index when cash is considered may not have much correlation with any alpha generated by a fund.