I was on a conference call recently during which the presenter was creating a screen of investment managers as the first part of a selection process.  He said, “And let’s use a maximum tracking error of . . .”  Wait, I thought, let’s stop right there.  Maximum?

In choosing investment managers, many seem to hold dissonant beliefs:  Yes, we want active management, but we don’t want the managers to be too active, lest they stumble.  (Unless, of course, they fit in our “alternatives” bucket.)

If you spend any time in places where financial advisors or plan sponsors gather (online, at a conference, wherever), you know that at a moment’s notice the discussion can devolve into a theological battle about the relative virtues of active versus passive management.  We are going to explore that territory now, so pick sides if you must, although this discussion is about understanding the degree of “activeness” of managers and looking for useful ways to apply what we find.

To do so, we will examine the concept of “active share,” which comes from the work of K. J. Martijn Cremers and Antii Petajisto, as published in The Review of Financial Studies.The Review of Financial Studies | This is the paper as published.  Earlier versions of the working paper are available at SSRN. The goal of the authors was to move from the industry practice of primarily depending on tracking error to determine how active a manager is, their hypothesis being that tracking error is good at determining the extent of a manager’s bets on systematic risk factors and not very good at measuring its bets on individual positions versus a benchmark.  Active share focuses on the latter.

How different is a manager from the benchmark?  Is the manager really a “closet indexer,” playing it close to the vest while earning higher fees than a truly passive manager?  Or does he stray further afield in search of returns and in what way?

The authors make a persuasive case for including active share as an analytical tool to explore those questions.  More importantly for those seeking grist for the active management debate or for those just looking to find the best managers, the main conclusion of the research is revealing:  The most active managers (as measured by active share) deliver the best performance.

In one sense, that comes as no surprise.  The rise of hedge funds is in large part due to their willingness to be different, to hold positions regardless of their representation in an index, and to hold them in great concentration (in both senses of the word).  And, a less scientific view of mutual funds indicates that good performance over time usually comes from a manager who is willing to stand apart from what everyone else (read “the market”) is doing.  However, as typified by the anecdote at the top, the gears of manager selection (for and by individual and institutional investors) often grind in ways that lead to supposedly safer choices.

The slicing and dicing of data in the article includes other information of interest, including the impact of size on the amount of differentiation in a typical fund’s portfolio (not good news for big funds) and historical information that shows a decline in active share among supposedly non-passive managers over time.

As you might expect if you have read any of my other postings on academic research,the research puzzle | Here are some postings in that regard. I have a few quibbles with the article.  The most notable was the choice of benchmarks for the calculation of active share.  The authors considered a wide variety of benchmarks to find the one for each mutual fund that had the “greatest amount of overlap.”  So, to measure how different the funds were from their benchmark, an index to which they are most similar is used?  That seems a bit illogical.  It would be preferable to look at active share versus the stated prospectus benchmark; I believe you should try at all times to measure managers versus what they say they are trying to do.

Nevertheless, the work by Cremers and Petajisto is an important foundation for analysis.  I have been surprised that the concept of active share has not gotten more play in the years since it first appeared in a working paper by the authors.  There have been a few articles about it and there are some initial signs that the concept is gaining a minor foothold in the business, but it deserves greater attention and more study (by academics and investment professionals) to see if the concept truly has legs.  Those reviews should include longer time periods, management structures other than mutual funds, and asset classes other than equities.

To date, the evidence supports the view that portfolio differentiation as measured by active share is a key metric to assess manager skill.  Interestingly, the article also argues that tracking error is not very good at doing so, although given a choice I’d opt to have more of it rather than less of it.  If I hire active managers, I want them to be active, across whatever dimension they can use to their advantage, and I want measurement tools like active share that help me judge them outside of the conventions of the day.