Abstract Active Share is a new concept that has the potential to be gigantic for the support of active management over passive investing. Let me rephrase that. Active Share is a new concept that has the potential to be gigantic for the support of active management over passive investing. When combining Tracking Error with Active Share, recent empirical research seems to be able to better capture manager skill.

Body  The support for active management in the academic literature has been rather meager, to be very kind. Countless research papers have shown that active managers do not outperform. Passive investing is supported with arguments of market efficiency and the lack of performance persistence (or predictability). Even managers who do show out-performance will be met with, what I call, the “if-merely-random-we-should-expect-some-to-outperform” argument (it’s a technical term).(*)

That is, until recently. Cremers and Petajisto (2009) and Petajisto (2010) have questioned the way that active management is measured. A new metric called “Active Share”, when combined with Tracking Error, seems to be better able to capture manager skill.

A manager can essentially add value in two ways: through stock selection or factor timing (sector/industry selection). The traditional method used to compare a fund to a benchmark, tracking error, is only able to pick up the latter. I discuss tracking error in our training sessions. Tracking error, though, is pretty holdings agnostic. Active share, on the other hand, looks at the holdings themselves. For long-only funds, the number ranges between 0% and 100% and represents the percentage of the holdings that is different from the index.

Paraphrasing the author’s example, let us consider an index that encompasses 50 sectors, each containing 20 stocks. If a fund were to invest in all 1000 stocks, it would have an active share of 0%, as the fund matches the holdings of the index perfectly. It would also have a very low tracking error. If, on the other hand, the fund were to invest in 1 stock in each sector, the manager would grab systematic exposure from each sector and be pretty diversified across sectors. The manager’s tracking error is likely to still be rather low; however, the manager would have a pretty high active share of 95% (since only 5% of the manager’s holdings are identical to that of the index).

Every month, the authors compute the tracking error and active share of every fund to the best fitting benchmark, and then break up each into quintiles. They then break up the space into four different kinds of managers.

As far as performance is concerned, one of the four groups has clearly, and statistically significantly, outperformed the others, as well as their benchmarks after fees. Below are the reported annualized alphas(**) for each group, as well as the overall group. The blue numbers are gross of fees, the red numbers are net of fees.

A few points are worth noting.

  • Overall funds lose .41% after fees. This supports prior literature that active management under-performs overall.
  • Diversified Stock Pickers” rock, as the only group that has outperformed.
  • Active share is not a statistic that is easily obtainable. It is based on the tracking errors and active shares of all mutual funds.
  • High active share, by itself, is not sufficient. It seems to need to be combined with low tracking error.

(*) The argument simply says that if you have a field with a lot of managers, we should expect quite a few to do well, even if beating an index is a coin flip. Say you have a 1024 managers, each with a 50/50 chance out outperforming every year. After 8 years of history, with no other factors but luck:
– 4 will have outperformed ever single year
– 32 will have beaten the index 7 out of the 8 years
– 112 will have beaten the index 6 out of 8 years

(**) I provide the standard alphas to make the exposition more practical, not the four-factor alphas which are now more commonplace in the academic literature.


Category: CAIA, CFA, CIMA

About the Author


Leave a Reply

Your email address will not be published. Required fields are marked *

March 2012
M T W T F S S
« Feb   Apr »
 1234
567891011
12131415161718
19202122232425
262728293031  

Categories




Hit Counter provided by Curio cabinets