Find Out How Active Your Fund Is

Investors paying for active management want to be sure they are getting what they pay for. Over the years, various measures have been created to gauge whether a fund is behaving like an index or charting a different course. The latest entry is a promising one that is based on portfolio holdings rather than performance.

Active share offers a couple of elegant solutions to problems with Modern Portfolio Theory statistics. Created by Yale's Martijn Cremers and Antti Petajisto, active share replaced a one-index-fits-all approach with one that measures funds against all indexes in a database to see which they are most like and uses data from the most similar benchmarks. (You may recall we had Petajisto present the paper at our 2007 Morningstar Investment Conference.)

It's also driven by portfolios rather than performance. While Modern Portfolio Theory stats line up a fund's performance with an index, active share compares portfolios to see how different the fund is based on actual holdings. It's easy to calculate. It's the inverse of a fund's holdings' overlap with its closest fitting index. On one end, a reading of 100 means that the fund has no overlap with the index. In fact, a fund could actually go above 100 if it has short names in the index and has no long positions in it. On the other, a reading of zero means that it is identical to an index.

It's a more precise reading and more up-to-date, as a fund could be positioned differently today than it was when many of the past stats were generated.

Our data suggest that active share is useful as an indicator of how closely returns will track that of the index and how likely it is to outperform. They confirm the work done by Petajisto and Cremers.

However, we found that the raw number (0 to 100) wasn't very useful for small- and mid-cap funds as the greater number of stocks in those universes leads to high active share figures for even funds that have broadly diversified portfolios. For example, the 23-stock large-cap  Ariel Focus (ARFFX) has an active share of 84 and the 907-stock Fidelity Low-Priced Stock (FLPSX) has an active share of 91. Yet Fidelity Low-Priced Stock has a higher R-squared indicating it is the one that's more like an index.

Rather than focus on the absolute number, we calculated a percentile rank, which proved to be more useful in understanding which funds were really index-huggers and which were charting a course far from their index. Fidelity Low-Priced Stock's active share percentile ranking in mid-blend is below average for its category, while Ariel Focus is in the top 10% of the large-value category on active share.

We set out to test active share on two fronts. One: Was it a useful gauge of whether a mutual fund closely tracks an index? Or, put another way, would it tell us how active a manager was? Two: Did it help to predict outperformance?

We've got some answers to those questions, but we plan to test more time periods in order to gather more evidence about predictive value.

How We Ran the Data
As I noted, active share in absolute terms tends to find much more active management in small caps than large, so we tested it as a relative measure within categories.

Cremers and Petajisto's study looked at performance from 1990 to 2003. We picked up from there by measuring the active share within each category in June 2004 and then examined their gross returns (expense ratios were backed out) over the ensuing five years. We used gross returns in case funds with higher active share differed in cost from those with lower active share.

The original active share paper grouped funds by active share on a monthly basis--something that makes for a rigorous academic test but isn't a realistic simulation of how investors behave. Investors want a fund they can hold on to for a while in order to reduce transaction costs and save the time and labor of constant switches.

Within each of the nine domestic-stock fund categories, we divided funds into quintiles based on their active share. We also calculated the percentage that ceased to exist over those five years in order to find out if our figures were being skewed by survivorship bias--in case higher active share funds were more likely or less likely to be liquidated or merged away than those with lower active share. Finally, we measured how close each fund's performance was to its closest fit index in order to test whether active share is useful for predicting whether a fund's returns will track the index.

Success on Two Fronts
Active share demonstrated its worth in both fields. First, it seemed to have some ability to predict returns. Rolling up the individual category data into an overall figure, we found that the least active quintile produced a 1.26% annualized loss, while the most active gained 0.03% annualized. In fact, in all nine of the categories studied, the highest active share quintile outperformed the lowest.

Over that period, mid-caps modestly outperformed small caps, with large caps lagging. Because we used active share within our fund categories, cap bias shouldn't have had much of an effect, but we'll need to test more periods to have a definitive answer.

Active share also worked well at predicting indexlike performance. We grouped returns into those within 100 basis points annualized of the index, those within 200 basis points, and those within 300 basis points.

Overall, the lowest active share quintile had 49% of results within 100 basis points, 77% within 200, and 91% with 300. By contrast, only 24% of the highest active share quintile produced returns within 100 basis points of their index, with 48% within 200 basis points, and 65% within 300 basis points. The pattern held up pretty well, except in mid-value, where there wasn't a pattern.

If funds with high active share were more likely to be killed off, then it might invalidate the exercise. If it were simply the case that such funds swung for the fences and were killed off if they failed, then the share differences in the above figures would be merely a product of survivorship bias. However, the number of funds killed off didn't vary much by active share. (Expense ratios and star ratings, though, are strong predictors of which funds will be killed off. It's one advantage they have over active share.)

What It Means
It makes sense that portfolios that are less like indexes lead to performance that is less like an index. So, it seems a safe bet that active share can help investors build better portfolios and have realistic expectations about performance.

The data on predictive power are encouraging, but we'll need to test more time periods to be certain. Even so, that makes active share a measure worth knowing, given how few measures are proved to have predictive power.

It's also worth remembering that these figures come pre-expense. That 100-basis-point pre-expense advantage we found would be wasted in a fund costing 100 basis points more than the cheapest funds in the category. And expense ratios are still the best proven predictor of future returns.

Who Is Most Active?
 Fairholme (FAIRX) is now probably the most popular and best-known focused fund, and active share confirms Bruce Berkowitz's nonconformity. This fund's asset growth might make a repeat very challenging, but Berkowitz certainly isn't about to let that turn this fund into a closet index fund. The fund has a 93% active share and ranks in the top 2% of large blend.

 Sequoia (SEQUX) gets a lot less publicity than Fairholme, but it is really the original focused fund.  Berkshire Hathaway (BRK.A),  Fastenal (FAST), and  Danaher (DHR) have been crucial to the fund's success against the S&P 500. Its 94% active share puts it right at Fairholme's level. With a more modest $3 billion asset base and a history of closing, this fund seems a good bet to remain unindexlike. With a great second generation of managers in place, it should continue to produce strong risk-adjusted performance.

 Primecap Odyssey Aggressive Growth's (POAGX) 91% active share and top 16% versus the Morningstar Mid-Growth results are less striking.With 80 stocks, this fund isn't as focused as Sequoia and Fairholme, but it does have a distinct character. Despite its strong performance, the fund still has just $600 million in assets.

 Janus Contrarian (JSVAX) is nominally a large-blend fund, but with 16% of assets in India and 22% more in other foreign markets, it's on the brink of the world-stock category. But even if it were moved to world stock, it would remain an iconoclastic fund. David Decker is a bold investor who aims to buy well-run companies with accelerating returns on invested capital. A miserable 2008 showed Decker doesn't walk on water, but his long-term record is excellent. The fund earns a 96% active share/top 1% measure versus Morningstar Mid-Value.

Who Is Least Active?
 Vanguard Explorer (VEXPX) has a 69% active share/bottom 2%, which ranks in the next-to-lowest percentile of small-growth funds. Given the fund's huge portfolio spread across seven subadvisors, this isn't a big surprise.

 Fidelity New Millennium's (FMILX) low active share of 68% (bottom 5% of mid-growth) marks a sharp departure from its past. Former manager Neal Miller ran it as a quirky aggressive fund, whereas John Roth has made it more of a middle-of-the-road growth fund. On the plus side, Roth has produced strong returns despite making rather small bets. This serves as a reminder that active share is certainly no guarantee of success or failure. It appears to point in the right direction, but it is not a measure of manager skill.

		FundInvestor Newsletter
	Morningstar FundInvestor is a 48-page newsletter dedicated to helping investors pick great mutual funds, build winning portfolios, and monitor their funds for greater		gains. This one-year subscription consists of 12 monthly issues, a Reader's Guide, and four free investing reports. Learn more.
	$129.00 for 12 Print Issues		$119.00 for 12 PDF Issues