Smart Beta 4: Factor Exposure's Curveball

By including factor tilts in smart beta’s definition, you get a mishmash of ETFs.

This blog is the fourth installment of a series transforming our ideas about smart beta. Part 3 showed that so-called alternative weighting serves many purposes, and fails an indicator of smart beta. Part 2 set the ground rules. Part 1 argued that defining smart beta in an ETF context is essentially impossible.

There is a strong movement to conflate smart beta with factor investing. For lots of folks, factor investing is the “smart” part of smart beta.

Circling back to our ground rules, sorting ETFs by factor exposure produces groupings that will cause rifts in the ETF community.

Factor investing targets specific drivers of returns. The most common factors are size, value, yield, momentum, volatility and quality. These factors show up in plain-vanilla indexes, of course, but can also be targeted and amplified, especially by index providers with huge databases and crafty analysts.

The past decade has brought us factor funds like the FlexShares Morningstar’s US Market Factor Tilt ETF (TILT | A-80) and the PowerShares S&P 500 Low Volatility Portfolio (SPLV | A-45). Older “smart” funds target factors too: value in PowerShares’ FTSE RAFI US 1000 (PRF | A-88), size in Guggenheim’s S&P 500 Equal Weight (RSP | A-75) and yield in iShares’ Select Dividend (DVY | A-67).

Designer funds promote the smart-beta label. However, branding all funds with factor exposure as smart beta will start arguments. In terms of our ground rules, sorting ETFs by factor exposure produces results that are not widely acceptable to the ETF community.

If factor exposure defines smart beta, then all funds with factor exposure must be smart beta. As you will soon see, most funds have some kind of factor exposure.

We might have to work a bit to define what “factor exposure” means, just as we had to transform “noncap-weighted” to “plain vanilla” back in part 2.

We could require a deliberate index-design process that targets factor exposure, with nonvanilla selection or weighting, too.

Even these nuanced definitions of factor exposure produce groups of funds that will not be widely acceptable within the ETF community.

Three groups of funds that have factor exposure but don’t seem like smart beta will prove that factor exposure doesn’t work as a smart-beta criterion.

1. Cap-weighted (and selected) size and sector funds often have small-cap or value exposure.
2. Cap-weighted (but not selected) value funds deliberately access the value premium.
3. Equal weighting adds size exposure wherever it is used, even in seemingly vanilla funds.

Size And Sector Funds

When the press calls me asking, “What’s the biggest smart-beta fund?” I can’t say, “IWM” (the iShares Russell 2000 ETF (IWM | A-84)) because reporters aren’t looking for a fuddy-duddy, plain-vanilla fund.

IWM clearly accesses the size factor.

Many plain-vanilla funds carry factor tilts. Sector funds are great examples.

The Utilities Select Sector SPDR Fund’s (XLU | A-94) weighted average market cap of $25 billion (as of April 1, 2014) is one-fifth that of its parent S&P 500 Index of $118 billion, and smaller than the iShares MSCI USA Size Factor (SIZE | B-79)’s $37 billion.

SSgA’s Energy Select Sector SPDR (XLE | A-96) is more of a value play than the world’s biggest value ETF. XLE’s P/E ratio was 16.7 as of April 1, 2014, versus 17.11 for the iShares Russell 1000 Value ETF (IWD | A-86).

Does anyone think IWM, XLU and XLE are smart-beta funds?

Until folks cast aside the “anything-but-cap-weighting” definition of smart beta, using factor exposure to define smart beta will create surprising, unacceptable groupings.

XLU’s and XLE’s size or value tilts are knock-on effects, reflective of their sectors. FirstTrust and PowerShares have deliberately crafted sector suites, with products like the First Trust Energy AlphaDex (FXN | B-56) that are designed to capture the value and growth factors, in which the tiered-weighting scheme introduces a distinct small-cap tilt. That doesn’t lessen XLU and XLE’s factor exposure.

Growth And Value Funds

There are highly engineered value and growth indexes that don’t seem like smart beta to many. The Russell 1000 Value and its counterparts from S&P, MSCI, CRSP and FTSE challenge the conflation of smart beta and factor exposure.

Because these indexes are cap-weighted (sometimes with complex buffering and treatment-of-borderline-security rules), they might not seem like smart beta. Look at a so-called smart value fund and a plain-vanilla one. Is one smarter than the other?

IWD and iShares’ MSCI USA Value Factor ETF (VLUE | B-86) each target the value effect, one via security selection, and the other via weighting. VLUE’s name has the magic word “factor,” and VLUE is factor-weighted, but VLUE and IWD are remarkably similar.

VLUE and IWD share a parent universe of large- and midcap U.S. stocks, represented in the table below by the iShares Russell 1000 (IWB | A-92) and the iShares MSCI USA (EUSA | B-96). VLUE includes all stocks in the MSCI USA Index, but weights them by factors like earnings and book value per share. IWD culls the value stocks from the Russell 1000, then cap-weights them.

Look carefully at the table below. Which better captures the value premium: VLUE or IWD?

IWD beats VLUE with the lower P/B—the classic value metric. IWD also has the higher dividend yield. VLUE sports the lower P/E ratio. VLUE’s correlation with IWD is 0.98, with a .99 beta. The two are largely indistinguishable, but if I had to pick a value fund, I’d go with IWD.

Security selection can produce powerful portfolio tilts, just as weighting can, because any security not in a portfolio has a weight of zero percent.

If VLUE is a factor fund, then IWD is a factor fund.

Will the ETF community accept cap-weighted growth and value funds like IWD as smart beta? Yes, I demolished alternative weighting as a smart-beta criterion in my most recent blog, but changing perceptions takes time. I’m not sure the ETF community is ready to put IWD in the same bucket as VLUE.

Even if the community were ready to include all the style funds as smart beta, I think there would still be pushback about the equal-weighted industry funds.

Equal-Weighted Funds

With no vestige of cap weighting, and with deliberate index design that promotes the small-cap effect, equal-weighted funds check every smart-beta preconception box.

Some equal-weighted funds don’t seem like smart beta.

Compare SPDR S&P Retail (XRT | A-45) with Guggenheim’s S&P 500 Equal Weight ETF (RSP | A-75). RSP tops nearly everyone’s smart-beta list. XRT, meanwhile, seems like a “plain Jane” retail fund even though XRT—like RSP—is equal-weighted.

Guggenheim and SSgA have vastly different reasons for equal weighting their funds, and it shows in the way they market their funds, and, in turn, in the way investors think about these funds.

Guggenheim’s claim that equal weighting improves risk efficiency helps differentiate RSP among the 66 U.S. large-cap funds. RSP’s asset base is about 1/20^th the size of SPY’s.

SSgA claims that XRT represents the U.S. retailing industry, probably because registered investment company compliance issues drive XRT’s equal weighting. Yet XRT tilts much smaller than the overall retail industry. XRT is about 15 percent of the weighted average market cap of U.S. retailers, according to Thomson Reuters. As long as SSgA says XRT represents U.S. retailers, hardly anyone will think XRT is a smart-beta fund.

XRT and RSP are equally equal-weighted. If RSP is smart beta, then XRT must also be smart beta. I want ringside tickets to the fight this will provoke.

Factor Exposure

Factor exposure is not a reliable smart-beta indicator. Many funds have some kind of factor exposure, whether by happenstance (XLU), by design (IWM and IWD) or as a defense against the taxman (XRT). Unless the ETF community is ready to accept these old-line funds in the smart-beta rolls, factor exposure won’t work as a smart-beta criterion—neither in its pure form, nor adapted to account for deliberateness, nor expanded to include alternative weighting.

Neither the anything-but-vanilla approach nor the factor exposure approach to defining smart beta works across the ETF universe. When you use these criteria to sort U.S. ETFs, you get all kinds of unexpected—and, for many, unwelcome—results.

I’m beginning to think the most reliable smart-beta flag lives in the marketing material, where issuers claim, subtly or boldly, their rules-based designer strategy will outperform a plain-vanilla index on a risk-adjusted basis.

In the fifth part of this series, I will look at the returns of some flagship smart beta funds, to test whether smart beta funds deliver risk-adjusted outperformance.

At the time this article was written, the author held long positions in IWM and IWD. Contact Elisabeth Kashner at [email protected].