Has active share already had its day?
Joseph Mezrich presents research suggesting the much-hyped ‘active share’ measure may have helped select good stock pickers only before 2004
Active share has become a popular measure for active fund management, especially when used together with tracking error. The conventional wisdom today seems to be that high active share helps active fund managers. However, high active share certainly was not the way to invest in 2014 – active stock-picking funds in the US had their worst year in decades, underperforming their benchmarks, on average, by over 200 basis points.
That result is no anomaly. Based on data going back to 2004, our conclusion is that active share has not been a reliable indicator of fund manager success, at least not over the past decade, and may be overrated as an investment tool.
A portfolio manager’s job requires making active investment selections against a benchmark. Tracking error has been the traditional measure indicating how aggressive these investment choices are. However, in the past few years, the portfolio’s active share has emerged as a different, potentially more relevant indicator of the manager’s deviation from the benchmark.
Introduced by Martijn Cremers and Antti Petajisto in 2009, active share measures the net deviation of the weights of stocks in the portfolio from their weights in the benchmark. One of the attractions of looking at active share together with tracking error is that it provides a good way to characterise investment approaches.
Stock pickers will tend, by definition, to have high active share. If they also have relatively low tracking error, they are diversified stock pickers. But if their tracking error is high, they have made concentrated stock picks. If managers have low active share, they will be making factor bets if they also have high tracking error. Factor bets in this context do not simply mean a tilt to growth, value or momentum, but also sector tilts that are not focused on specific stocks in the sector (as one might do with a sector exchange-traded fund, for example). If both active share and tracking error are low, then the investor is a closet indexer.
“The main conclusion we draw… is that there hasn’t been a consistent winning approach for active managers over the past decade”
Using this categorisation, Petajisto in 2013 found that low active share managers underperform their benchmarks after fees, but high active share managers can add value if they don’t have very high tracking error. In short, stock pickers can win if their portfolios are not too concentrated.
This finding has implications for the fundamental portfolio managers’ business since they are deemed more likely to succeed and be more attractive to investors, if their active share is high. In our research, we took a closer look at the utility of the active share measure based on the historical record of fund managers.
Our data came from the Center for Research in Security Prices (CRSP), which provides a survivorship bias-free US mutual fund database. Even though fund returns in this database go back to 1961, fund holdings that are necessary to calculate active share only go back to late 2003. The Petajisto (2013) results on active share are based on a longer history, derived from blending the CRSP data with Thomson Reuters’ longer history. Our analysis attempts to replicate his results, but does not find active share to be a reliable indicator of success. At least, our results show it hasn’t been useful since 2004.
We follow Petajisto’s (2013) methodology. The funds are sorted into quintiles, first by active share and then by tracking error. The managers in the highest quintile of active share are stock pickers, and those in the highest quintile of tracking error among the stock pickers are the concentrated stock pickers. The managers that are in the highest quintile in tracking error but below the highest quintile in active share are those making factor bets. Managers with the lowest active share, but below the highest quintile of tracking error were closet indexers. The average annual excess returns of these different types of active managers over the past decade are displayed in the figure, where we refer to the concentrated stock pickers as ‘concentrated’, and to the high-active-share stock pickers that are not concentrated simply as ‘stock pickers’.
As the figure shows, there was no consistent winner among the different approaches in the years between 2004 and 2014. The concentrated stock pickers (dark grey bars) were positive in one out of the 11 years. And in years when this approach did not win, the downside was usually the worst of the investment approaches. The main conclusion we draw from this is that there hasn’t been a consistent winning approach for active managers over the past decade.
This raises the question: why do we find no particular benefit associated with high active share, in contrast to Petajisto’s (2013) finding that stock pickers have, on average, delivered positive excess returns? He relied on a blend of CRSP and Thomson Reuters databases that let him go back to 1990. We were able to confirm that our results agreed with the post-2004 data provided in his publication (for 2008 and 2009). This leads us to suspect that our negative findings on the role played by active share are because active share may have been useful before 2004 but not since then.
Joseph Mezrich, is a managing director and head of equities quantitative strategies at Nomura Securities International