Investors love to chase performance. But as Morningstar and others have documented, a strong long-term record alone is not predictive of future performance. Many investors have discovered this the hard way, which has helped fuel the growing adoption of index strategies, including strategic-beta funds. These attempt to provide lower-cost, more transparent, and more-consistent exposure to securities with characteristics that have historically been associated with superior performance. But what if the apparent performance edge is not sustainable and investors are just relapsing into counterproductive performance-chasing? These are the provocative questions Rob Arnott and his Research Affiliates colleagues pose in a pair of recent articles, “How Can ‘Smart Beta’ Go Horribly Wrong?”1 and “To Win with ‘Smart Beta’ Ask if the Price is Right.”2
In these research pieces, Arnott and his co-authors argue that when a factor outperforms and investors get more excited about it, it tends to get more expensive. As a result, it is less likely to outperform going forward. The same process works in reverse as factors underperform. In other words, valuations matter, and it is necessary to account for the valuations of each factor to properly set expectations for its future returns. Arnott and his colleagues went one step further and made the bold assertion that much of the return premiums during the past 50 years to the low beta and gross profitability factors were attributable to rising valuations, which are unlikely to persist. (However, they still delivered better risk-adjusted performance, net of valuation changes.)
Using data from 1967 through March 2016, the Research Affiliates team found that as value stocks become cheaper relative to growth stocks, they were more likely to outperform during the next five years. It uncovered a similar inverse relationship between valuations and future performance for the small-cap, illiquidity, investment, and gross profitability factors. This inverse relationship for momentum was weaker in the long term than in the short term. This isn’t surprising because momentum is a high-turnover strategy and its future portfolio often bears little resemblance to its present one.
The group found no significant link between the valuation and future performance of the low-beta factor. They contend that this is because the low-beta stocks’ valuations have become stretched relative to history and that they have yet to revert to the mean, a possibility they cite as a key risk. The authors also caution that, as of March 2016, valuations for a few other factors looked rich relative to history. These included high dividend yield, gross profitability, investment, and momentum. Value is the only factor that looked particularly cheap.
While valuations matter (at least for low-turnover portfolios), investors probably shouldn’t use valuation spreads to time factor exposures, unless those spreads are extreme. There is a lot of noise in the relationship between factor valuations and future performance, and it can be skewed by extreme events, like the technology bubble. And as AQR’s Cliff Asness points out in his article, “The Siren Song of Factor Timing,”3 the success of timing factors with valuation signals is positively correlated with the value factor itself. (Arnott argues that this effect cannot fully explain the success of contrarian timing.) Asness suggests—and I would agree—that factor timers sacrifice diversification by concentrating their bets more heavily in the cheapest styles, which could increase the volatility of the portfolio.
Factor Index Analysis
In order to shed further light on the relationship between factor valuations and future performance, I ran a similar analysis to the one Research Affiliates conducted with several factor indexes that investors have access to via exchange-traded funds. I used the performance and valuation spreads between the indexes listed in Exhibit 1 for this analysis.
.png)
First, I looked at the ratio of the long (first) index’s valuation relative to the short (second) index and tracked the performance spread between the two indexes during the subsequent five years, similar to Research Affiliates. For each factor, I regressed the rolling five-year performance spreads against the valuation ratios at the start of each period. The results are presented in Exhibit 2. I ran this analysis once with price/book and again with price/earnings.
.png)
As expected and consistent with Research Affiliates’ findings, as small-cap and value stocks became cheaper relative to their large-cap and growth counterparts, they tended to do better during the next five years. But the strength of this relationship is only moderate. The r-squared values of these regressions indicate the percentage of the dispersion in subsequent five-year returns that the initial valuations explain. These values range from 0.33 for the price/book ratio in the case of small-value to 0.53 for price/earnings in the case of large-value. This indicates that the beginning-of-period valuations explain 33%–53% of the return dispersion during the following five years. That means there is a lot more going on that this simple model doesn’t capture. It is also important to note that the strength of this relationship was largely influenced by the tech bubble.
In part 2 of this article, we will continue to explore if timing factor strategies will work.
1 Arnott, R., Beck, N., Kalesnik, V., & West, J. 2016. “How Can ‘Smart Beta’ Go Horribly Wrong?” Research Affiliates. https://www.researchaffiliates.com/en_us/publications/articles/442_how_can_smart_beta_go_horribly_wrong.html
2 Arnott, R., Beck, N., & Kalesnik, V. 2016. “To Win with ‘Smart Beta’ Ask if the Price Is Right.” Research Affiliates. https://www.researchaffiliates.com/en_us/publications/articles/540_to_win_with_smart_beta_ask_if_the_price_is_right.html
3 Asness, C. 2016. “The Siren Song of Factor Timing aka ‘Smart Beta Timing’ aka ‘Style Timing’.” J. Portfolio Management. Special Issue 2016. https://www.aqr.com/library/journal-articles/the-siren-song-of-factor-timing
