How smart has your beta been lately?

Despite overwhelming evidence that active managers underperform their benchmarks, Yale professors Martijn Cremers and Antti Petajisto offer evidence that not only can active management outperform (we knew that), but that predicting successful active manager performance in advance is also possible.

We examine these claims pursuant to smart/alternative beta ETFs in Canada, and study recent down-cycle performance of these strategies.

Readers are cautioned to see beyond the performance tables. When examining a single-market period that does not include a full market cycle, it’s dangerous to draw any serious conclusions. But spot-checking how smart-beta strategies have fared in a down market is instructive.

A word about smart beta

Smart-beta strategies set rules for security selection or weights, often replicating active-management approaches that worked in the past. These methodologies represent static views of the world, hoping past outperforming characteristics will persist over the holding period. However, changes in economic or market conditions can lead to significant variance in returns, find Jacobs and Levy in their Journal of Portfolio Management article, “Smart Beta versus Smart Alpha” (2014).

Rules transparency is an advantage of smart beta, but low cost compared with active managers employing similar strategies is the real benefit. All smart-beta approaches represent active management because they must rebalance to their original strategies. Don’t mistake static for passive—these are all active strategies.

Active share of alternative beta strategies

In “How Active is Your Fund Manager? A New Measure That Predicts Performance” (The Review of Financial Studies, 2009), authors Cremers and Petajisto (C&P) popularized the idea that active share (AS)—the extent to which a portfolio differs from its benchmark—is a good determinant of future outperformance. Covering 2,650 funds from 1980 to 2003, they found the funds with an AS of more than 80% beat their benchmarks by 1.49% to 1.59%, after fees. A cap-weighted passive fund that exactly replicates its benchmark would have zero active share, while a fund whose holdings are completely different than its benchmark would have an AS of 100%. The paper has emboldened active managers and contributed to the interest in smart-beta ETFs.

We calculated active share as half the absolute difference in an ETF’s holdings versus its benchmark as at March 11, 2016. For simplicity, we used the TSX Composite as the reference benchmark for all ETFs, except Horizons S&P/TSX 60 Equal Weight Index, for which the S&P/ TSX 60 was used (disclosure: the author sub-advises for Horizons). Table 1, shows the performance from April 15, 2015 to January 20, 2016, representing the most recent peak to most recent trough.

Observations

Smart-beta portfolios are concentrated in factors they attempt to capture, so they may be less diversified than their capitalization-weighted cousins when comparing holdings and weighting differences. The possible exceptions are the two RAFI Fundamental Indices—CRQ and PXC—that would be closest to C&P’s definition of closet indexers (AS below 20%).

The low-volatility phenomenon (see AER January 2016 and July 2012) is considered an anomaly, because low-risk stocks are observed outperforming high-risk stocks, contrary to the capital asset pricing model’s hypothesis.

If expected risk equals expected return, this strategy should not work. But it worked during the recent market downturn from April 15, 2015 to January 20, 2016, with all ETFs that used the strategy losing less money than other strategies and indices. There was a slight tendency for low-volatility ETFs with higher active share—ZLB and TLV’s AS were 70.6% and 70.0%, respectively—to outperform. But earlier caveats hold. ETFs with the highest AS, and particularly those with more than C&P’s 80% threshold (FXM, PZC, WXM), should have done better than they did. All value-based ETFs underperformed modestly except XCV, with an AS of 46.9%; that AS would not have predicted outperformance. Growth, multi-factor, and momentum strategies all performed in line with the market.

Summary

In our limited study, getting the strategy correct seems more important than chasing the highest AS score. Subsequent research by Frazzini, Friedman and Pomorski (“Deactivating Active Share,” Financial Analysts’ Journal, 2016) has shown that benchmarks are more important than AS in determining performance. When adjusted for benchmarks, even when using C&P data, the advantage of high over low AS is eliminated. Meanwhile, in a paper commissioned by Vanguard, Schlanger, Philips and LaBarge (“The search for outperformance: Evaluating ‘active share’,” 2012) showed that high levels of AS did not predict outperformance for periods subsequent to C&P’s paper (post-2009).

Not surprisingly, it’s not enough to be different from the benchmark. Better-performing stocks must be overweight, and/or underperformers underweight, to achieve success. Further, the high dispersion of returns means pursuing high AS could easily lead to under- as well as over-performance. AS is useful to show how different a strategy is from its benchmark. Active, smart-beta ETFs demonstrate they are far from closet indexers. Picking the winners in advance, however, remains as elusive as before. Table 1: Active share of smart- and alternative-beta ETFs and recent market performance

Mark Yamada

Investments

Mark Yamada is president of PÜR Investing Inc., a software development firm specializing in risk management and defined contribution pension strategies.