Commentary: As factor-based investing grows, beware the combo platter

Factor investing is projected to grow to $3.4 trillion by 2022, up from $1.9 trillion as of Sept. 30, 2017, according to BlackRock. While enthusiasm for factor-based investing continues to grow, so too does a risk: piling on individual factor exposures to achieve an expensive form of beta.

Call it the combo-platter conundrum. The early days of factor-based investing were marked by traditional a la carte selection of individual factors such as size, value, momentum, quality and low volatility. Now it's increasingly common to see combination approaches that purport to provide exposure to several factors in a single product, often under the generic moniker "multifactor."

While we strongly support a multifactor approach, we have serious reservations about the ability of some so-called multifactor products to deliver on their promise of multifactor exposure. Some providers appear to have simply cobbled together existing single-factor products in various combinations to create new suites of multifactor strategies. We believe many of these hastily developed products suffer from the effects of dilution, or the unintentional offsetting of individual factor exposures. Dilution attenuates the total factor content of the multifactor strategy, often resulting in multifactor products with little or no net factor exposure. Investors essentially end up investing in an index, at a higher cost.

As an example, consider a simple multifactor portfolio of just two exposures: quality and value. To gain these exposures, suppose we invest half of our assets in a single-factor quality-only strategy and the other half in a value-only strategy. Note, however, that it is not unusual for value stocks, which for various reasons are trading at a lower price relative to fundamentals, to be of lower quality, while higher-quality stocks tend to be relatively more expensive. When these two groups of stocks are placed side by side in a multifactor construct, you can see how dilution ensues: high quality/expensive stocks offset low quality/cheap stocks. At the extreme, the stocks receiving an overweight in the value portfolio can be precisely those stocks that receive an underweight in the quality portfolio. If we don't explicitly control for these offsets by eliminating the low-quality orientation of the value product and the low-value orientation of the quality product, then it is clear how a simple quality-value multifactor strategy ends up with little or no aggregate exposure to quality or value.

The root cause of dilution is the failure to eliminate unintentional, disadvantageous exposures in the single-factor components of a multifactor portfolio. Correlations among factors are rarely zero and frequently strongly negative. That means corrective action needs to be taken to avoid dilution.

Perhaps ironically, the solution to the dilution problem has always been at hand. A 2016 paper by Steven Thorley, Roger Clarke, Harindra de Silva said that the fatal flaw with multifactor investing often lies predominately in the use of aggregated single-factor components which, owing to dilution, are not collectively efficient from a risk-reward standpoint. They find that a combination of four single-factor component portfolios — low beta, size, value and momentum — captures only half of the risk-adjusted return of a portfolio formed from individual stocks that are simultaneously low beta, small size, high value and high momentum. The latter is sometimes referred to as a "bottom-up" factor implementation and addresses the dilution problem directly, as stocks with disadvantageous exposures are not allowed in the portfolio or are given below benchmark weights. In this manner, more factor content is maintained as dilution is minimized. The resultant portfolio is also naturally less sensitive to changing correlations.

In a 2017 paper, Scientific Beta CEO Noel Amenc and others argue that the bottom-up approach promotes undue concentration as there are few stocks that are simultaneously low beta, small size, high value and high momentum. These authors suggest the resultant bottom-up portfolio would have very few names and thus potentially high levels of individual, sector and country risk. We feel this conclusion misses the point, as the benefit of a bottom-up approach is not necessarily the total elimination of dilution but rather the intelligent management and minimization of this malady. While no tactics will be perfect, the benefits of multifactor investing will be realized to a far greater extent.

To help evaluate efficiency and dilution, a conceptually simple metric can be used: the percentage of a factor-based strategy's total active risk that comes from the intended factors rather than extraneous noise. A more efficient strategy achieves more intended factor exposure per unit of unintended exposure, such as security, sector, region concentration or unintended factor tilts. We've shown that strategies that score high on the metric tend to have correspondingly higher risk-adjusted returns. This is not surprising and adds to the mounting evidence, including in a 2017 paper by Columbia University finance professor Kent Daniel showing that unintended exposures such as from sectors add to risk but not necessarily to return.

Dilution is an affliction that has caused many multifactor equity products to fail at their stated task. Instead of delivering robust multifactor exposures, these products have been shown to produce little, if any, net factor content. The problem of dilution stems from poor product design, usually in the form of an insistence on single-factor components that are naturally prone to dilution and sensitive to changing factor correlations. Although providers of multifactor products constructed with single factors have defended their approach, research supporting a stock-level or "bottom-up" implementation is gaining strength. Regardless of how one pursues factor-based investing programs, acknowledging, evaluating and controlling for efficiency and dilution are crucial to success, by ensuring that one's factor portfolio is indeed composed of factors.

In other words, not all combo platters are alike.

Michael Hunstad is head of quantitative strategies at Northern Trust Asset Management, Chicago. This content represents the views of the author. It was submitted and edited under Pensions & Investments guidelines, but is not a product of P&I's editorial team.