Do smart beta exchange-traded funds (ETF), which systematically tilt holdings to capture one or more factor premiums (such as size, value, momentum, quality, beta and volatility), offer net value to investors? In the April 2015 initial draft of his paper entitled “How Smart are ‘Smart Beta’ ETFs? Analysis of Relative Performance and Factor Timing”, Denys Glushkov assesses whether smart beta funds beat their benchmarks and whether they effectively time targeted factor premiums. After categorizing smart beta ETFs into 15 portfolios based on factor themes and weighted by fund capitalizations (rebalanced monthly), he evaluates performance of these portfolios versus three types of benchmarks: (1) passive benchmarks chosen by the ETF providers, capitalization-weighted within 15 matching portfolios (rebalanced monthly to corresponding smart beta ETF weights); (2) risk-adjusted versions of these self-declared benchmarks; and, (3) tradable capitalization-weighted blends of market, value and size funds, matched to smart beta ETF portfolios based on inception dates (rebalanced annually to original weights). The blended benchmark addresses the concern that the returns of self-declared benchmarks are realistic/gross of reformation costs and tests whether smart beta funds add value to a simple and relatively passive capitalization-weighted portfolio with size and value tilts. Using monthly returns for 164 U.S. equity smart beta ETFs and 49 distinct benchmarks during 2003 through 2014, and detailed historical holdings of most of these funds, he finds that:
- Pitted against returns of self-declared benchmarks:
- 9 of 15 smart beta ETF portfolios outperform, by an annualized 1.3%.
- 6 of 15 underperform, by an annualized -2.3%.
- Only three (equal-weighted, risk-weighted and volatility) outperform during the final year of the sample period.
- Pitted against risk-adjusted returns of self-declared benchmarks:
- Based on Jensen’s alpha, only one smart beta ETF portfolio (value) significantly outperforms, by an annualized 2.1%.
- Based on Jensen’s alpha, the dividend-oriented portfolio, one of the most popular in terms of number of funds and assets under management, significantly underperforms by an annualized -3.9%.
- Fewer than a third of the smart beta ETF portfolios have annual Sharpe and Sortino ratios higher than those of self-declared benchmarks, with the volatility portfolio offering the largest boost.
- Across all smart beta ETF portfolios, annual Sharpe ratios of smart beta ETFs and their self-declared benchmarks are about equal.
- Pitted against the tradable blended three-factor benchmarks:
- None of smart beta ETF portfolios significantly outperform based on either raw returns or Jensen’s alpha.
- As for self-declared benchmarks, about a third of smart beta ETF portfolios have higher annual Sharpe ratios than the blended benchmarks, led by the buyback/shareholder yield portfolio.
- Across all smart beta ETF portfolios, annual Sharpe ratios of smart beta ETFs and their blended benchmarks are about equal.
- Long-term investors who can forecast up and down markets would do at least as well shifting between the tradable blended benchmark and the risk-free asset as they would trying to shift between aggressive and defensive types of smart beta ETFs.
- Many smart beta ETFs exhibit adverse unintended factor tilts which tend to offset the benefit of targeted factor tilts. For example, fundamentals-weighted ETFs tend to benefit from a value tilt but suffer from adverse momentum, quality and volatility tilts.
- Overall, smart beta ETFs do not exhibit successful timing of targeted factors.
In summary, evidence suggests that long-term investors can achieve performance at least as good as those of smart beta ETFs via a portfolio comprised of capitalization-weighted and annually rebalanced market, size and value funds.
Cautions regarding findings include:
- Portfolio returns are gross, not net (monthly/annual portfolio rebalancing is frictionless). Accounting for rebalancing costs would lower returns for all portfolios. The use of annual rather than monthly rebalancing mitigates this concern for the blended benchmark portfolios.
- The sample period is not long for analysis of fund performance, especially in terms of number of equity bull and bear markets and number of economic cycles.
