The Trouble With Active Share
It's a descriptor, not a predictor.
The following first appeared in the June issue of Morningstar FundInvestor.
Last month, I wrote about the predictive power of fund expense ratios. Using the success ratio (the percentage of funds that survived and outperformed), I showed that your chances of a successful outcome grow dramatically as you move to cheaper funds. Cheap funds are more than thrice as likely to survive and outperform as high-cost funds.
But what about active share? It’s a highly touted measure that determines how different a fund’s portfolio is from its benchmark. It is the inverse of overlap, so a high active share means a fund is very different and a low one indicates a fund is similar to its benchmark. For example, a fund with an active share rating of 100 relative to the S&P 500 would have no holdings in common with the index, while a fund with an active share rating of 0 would have identical holdings (such as an S&P 500 fund). Early advocates said this measure was a good predictor of future performance.
More recently, some have said that it’s OK to have high expenses if you have high active share. My response: Balderdash!
I put active share through the same tests as expense ratios. As with fees, I broke U.S. equity funds into quintiles based on their active share relative to their peer group as of Jan. 1, 2011. Then, I looked to see how they did over the subsequent five years ended December 2015.
What Did I Find? The quintile with the highest active share among U.S. equity funds had a meager 29% success ratio, followed by 27% for the second quintile, 32% for the middle quintile, 40% for the fourth quintile, and 43% for the least-active quintile. Returns of the surviving funds likewise showed that active share hurt performance. The most active quintile had returns of 9.3% compared with 9.7% for the middle quintile and 10.6% for the least active quintile. This is a little worse than previous times I ran the test. In those tests, active share made closer to a neutral impact on results than the negative impact we see here.
For the sake of comparison, the cheapest quintile of U.S. equity funds had a success ratio of 64% versus 16% for the priciest quintile. The high active-share quintile’s 29% success ratio was comparable to the 28% success ratio registered by the fourth (or second-priciest) quintile of fees.
Active share’s ability to predict investor returns and risk-adjusted returns as measured by Morningstar Ratings was also weak. The highest active-share quintile had a 24% success ratio for investor returns compared with 28% for the least active share. Using five-year star ratings, the highest active-share quintile led to an average star rating of 2.6 while the least active had 3.2. It’s not surprising that risk-adjusting returns through the lens of the star rating would not help active share’s case, as more-focused funds are more volatile in general.
So, what’s wrong with active share? I see a few problems. First, it’s often mistaken for a measure of skill, but that’s not what it is. I could take a portfolio and add some short positions or out-of-benchmark long positions to boost active share, but would that necessarily improve returns? Only if I had skillfully chosen those shorts and nonbenchmark long positions. Second, it just doesn’t add up. You could carve the S&P 500 into 50 separate portfolios with high active share but you would not improve returns by 1 basis point.
I do find active share useful in a similar way that R-squared and tracking error tell you whether a fund is behaving like an index. Only in this case it measures holdings overlap with an index rather than correlation of returns. Our analysts monitor changes to a fund’s active share for signs that a manager is changing his strategy. For example, if a fund gets a lot of assets in a short period of time, active share may go down, and that tells me asset bloat is having an impact on a fund’s strategy. So, it’s a fine descriptor, just not a predictor.