With Smart Beta, Classification Matters—Here’s How We Tackle It

Over the last 10 years, smart beta ETFs have been among the most popular strategies to launch as investors have become more familiar with the idea of factor investing. Or rather, investors have become more familiar with the notion of harnessing well-documented premia (such as value and momentum) in search of returns in excess of broad-market beta. During this time, smart beta assets have ballooned by 2,371%, growing from $14 billion to $357 billion.¹ Moreover, the asset mix has moved away from traditional dividend strategies toward more factor-based approaches, with the share of smart beta ETF assets held by dividend funds falling from 59% to 44%.²

This shift in buying behavior indicates investors are increasingly viewing their portfolios through a factor lens. They are choosing among smart beta funds that provide specific exposure to a range of factors, including minimum volatility and size, or opting for multi-factor funds that offer exposure to multiple factors. Lately, fund launches have become more sophisticated, with many multi-factor blends and quantitatively-driven (e.g., optimized) products coming to market. The chart below illustrates these trends.

As smart beta ETFs garner more attention and become more sophisticated, more in-depth due diligence is required. In our view, this begins with defining the selection universe, an opaque topic in the field of smart beta investing. Here we explore the current challenges of smart beta classification and illustrate how our team approaches the subject.

Before we tackle classifications, it’s helpful to level-set by stating a few central beliefs about smart beta strategies. It is commonly held that:

• There is a core set of factors which are the most persistent in driving equity returns: value, size, volatility, quality, and momentum.
• These factors are well documented in academic literature and have persisted through time because they capture systematic risk, behavioral biases, and market structure issues.
•  While factors outperform over time, individual factor performance can vary by economic environment.
• Due to the cyclical nature of individual factor performance, a diversified mix of factors provides the best opportunity to outperform a cap-weighted benchmark over time.
• Dividend-based strategies should be recognized as smart beta as they are used to target specific portfolio outcomes.

The smart beta classification conundrum

Most smart beta classifications adopt either a selective “I’ll know it when I see it,” or a throw-it-all-in-there “kitchen sink” mentality. As a result, unlike traditional size and style strategies, there is no clear and concise definition for smart beta. The lack of consensus around what constitutes smart beta results in strategies with seemingly similar names—but in reality, very different construction methodologies.

For example, let’s consider “value” funds as identified by FactSet. At year-end 2018, there were 29 value funds focused on large- or broad-market US equities. The return dispersion between those 29 strategies in 2018 was 12.5%! The best-returning fund was down 4.05% while the worst declined 16.54%. From a risk perspective, the volatility of these strategies ranged from 10% to more than 15%. The chart below illustrates this risk and return dispersion among strategies sharing the same “value” moniker. The dispersion looks like someone just broke the rack in a game of pool, not a breakdown of similar strategies.

Furthermore, the average number of holdings in the funds was 302, ranging from as few as 21 holdings to as many as 1,322.³ The vastness of this range can catch investors off guard unless they carefully conduct in-depth due diligence, much like they would when choosing an actively managed fund. A checklist—such as the one we recently created—can help investors look beyond the label and ensure they’re asking the right questions. All due diligence, however, starts with identifying the applicable universe, and therefore starts with classifications. The old adage of “know what you own” definitely applies here.

Step-by-step: How we approach smart beta classification

Creating classification schemas is a difficult task. There is no right or wrong answer, and therein lies the main problem with smart beta investing. If major data providers—and for that matter, fund providers—cannot agree on what constitutes a smart beta strategy, how will investors be able to fully implement it as part of their investment process? Some include a fund tracking the price-weighted Dow Jones Industrial Average as smart beta, which feels a bit off. It’s unlikely Charles Dow was the innovator of smart beta strategies back in 1884 when he first put together a table of 11 stocks in the Customer's Afternoon Letter, a daily two-page financial news bulletin which was the precursor to The Wall Street Journal.

The chart below depicts the central issue of seemingly random walk down smart beta street. It illustrates the total number of smart beta ETFs and their assets as identified by three major data providers and according to the SPDR Americas Research definition. Given that each data provider has a different definition, we see different figures for assets and number of funds. Traditional style exposures are included by data providers, which we exclude. But that is not the only wrinkle we apply.

On the SPDR Americas Research team, we take five steps to identify and classify smart beta strategies:

• Step 1: As a starting place, take everything Morningstar deems to be “Strategic Beta.”
• Step 2: Determine which ETFs are sector-based strategies.
  - Rationale: We want to measure smart beta adoption in broad-based strategies, rather than sector funds that perhaps use equal weightings in order to more efficiently capture a narrow industry. We acknowledge that some sector funds may employ value or momentum screens to outperform traditional sector beta strategies. As a result, we have created a different category for non-equal weighted smart beta sector strategies.
• Step 3: Remove any Morningstar-identified smart beta strategies that are market-cap weighted.
  - Rationale: In our view, the purest form of smart beta should be alternatively weighted based on a metric associated with the factor targeted (e.g., the price-to-book metric for value strategies). Equal weight is ok, as it is balanced and mitigates stock-specific risk, allowing for the grouping of stocks to drive returns (e.g., 50 cheapest securities based on price-to-book).
• Step 4: Add back in market-cap weighted dividend strategies.
  - Rationale: We understand smart beta strategies can target outcomes—more specifically, income generation—therefore, we add back in market cap weighted dividend strategies. Classifications aren’t clean, and our approach may not be perfect, but we are attempting to increase precision while maintaining a systematic process. This step reinforces why smart beta dividend strategies should be separately identified.
• Step 5: Tally up what is left, then manually classify by strategy type to ascertain the targeted factor exposure(s).
  - Rationale: By performing this laborious step, we are able to supply analytics like those displayed in the first chart and decompose flow trends, as illustrated in our quarterly Smart Beta Dashboard.

Our aim: Provide clarity in a nebulous field

The last chart below is a recreation of the earlier chart, but this time with the sub-asset classes (sector, dividend) broken out. By creating different classification tiers for smart beta and being transparent about our process, we aim to provide clarity in a nebulous field that features an abundance of different strategy types—all carrying the same name. As more smart beta strategies come to market, our classification schema and other resources can be valuable tools in the due diligence process, helping investors go from identification to implementation.

For our latest smart beta insights, you can follow SPDR® Blog or visit the smart beta section  of our website.