This study (
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2597122) documents that the active share measure, which has attracted so much attention in the past few years, has no performance predictive power once the analysis is done carefully. This is a real problem given that many financial advisors and consultants use this measure to pick funds for their clients. In addition, some funds seem to heavily advertise having high active shares, potentially misleading investors.
Comments
"Active Share correlates with benchmark returns, but does not predict actual fund returns; within individual benchmarks, Active Share is as likely to correlate positively with performance as it is to correlate negatively. We conclude that neither theory nor data justify the expectation that Active Share might help investors improve their returns."
That makes sense to me. Active Share is just a tool that can help guide the initial phase of the selection process. High Active Share numbers can guide an investor to a group of funds. These funds have characteristics that differentiate them positively or negatively from a benchmark. It is up to the individual investor to do additional research to determine which if any of the funds deserve to be selected for investment.
That's my simple minded take on it....
Does that mean they're wrong?
Nope. It means that they have a vested interest in the results and, consequently, their conclusions need to be viewed as one voice in a debate between partisans rather than as a dispassionate judgment.
Other voices in the debate include Callan & Associates, which looks at product-pairs (two funds run in the same style by the same management team) and concludes that the more-focused of the funds tends to have significantly higher A/S, alpha and beta. They conclude that A/S is a useful predictive construct, but only one of many you need to consider.
A more-partisan response comes directly from Cremers and Petajisto who allege that AQR, like the others, misrepresents their benchmarking strategy. "They were able to replicate the main results in our paper, showing strong evidence that active share has been strongly predictive of mutual fund performance once you adjust for the benchmark performance.”
My own take, as you know, is that active share makes logical sense: if your portfolio closely replicates the portfolio in your benchmark index, your portfolio has an extremely low prospect of producing results that differ from your benchmark's. That is, if my actively managed fund is the S&P499-Plus-One, it's likely to track closely the S&P500 index - and to lose out based on the drag of fees and indirect costs. That said, merely being different clearly does not automatically translate to being better. That leads to our focus on a bunch of "soft" factors like whether the portfolio strategy makes sense, the manager can articulate his approach to risk-management, the advisor communicates clearly, the insiders are deeply and directly invested in the strategy and the fund (or strategy) has negotiated rocky markets in the past. None of that is perfect, but collectively it seems to point in a generally good direction.
For what that's worth,
David
However, I echo David's sentiment that we need to have a way to differentiate funds by how active they are. But my concern is that given the hype associated with the active share measure, some fund families are using their active share aggressively to advertise their funds, which could mislead investors. I think we also need a way to filter out "fake" active funds. Perhaps this could be done by looking at the past performance record of funds' large active bets, i.e., positions where they deviated a lot from the benchmark.
The AQR paper linked above includes a chart (Exhibit 1, page 6) that displays the fund data used by Petajisto. It displays the average active share percentage for the funds included in each of the benchmarks. It also looks more broadly at how average active share varies between the Large Cap, All Cap, Mid Cap, and Small Cap fund categories. It shows the average active share at about 75% for the funds included in the Large Cap benchmarks, 80% for the All Cap benchmarks, 88% for the Mid Cap benchmarks, and 92% for the Small Cap benchmarks. This supports the conclusion that sorting according to active share simultaneously sorts according to market cap.
I leave it to other to determine whether the data supports the proposition that high active share increases the odds that a fund will have superior fund performance. My point is that the data does appear to indicate it is most useful to compare active shares numbers only between funds that share the same benchmark...or at least that share the same broad categories. Both Exhibit 1 of AQR's paper and the response Cremers and Petajisto made to that paper in the link @David_Snowball provided above appear to support this conslusion.