Smart Beta 3: Sinking Noncap Weighting

If you thought smart beta means not cap-weighted, think again.

If you thought smart beta means not cap-weighted, think again.

This blog is the third installment of a series transforming our ideas about smart beta. Part I set the stage, arguing that defining smart beta in an ETF context is essentially impossible. Part 2 defined the ground rules for the analysis.

So you think smart beta is anything that’s not cap-weighted? You’re hardly alone. “Anything but cap-weighted” is the simplest definition of smart beta. But it falls apart if you rigorously use it to sort ETFs.

As a preview, consider the SPDR Dow Jones Industrial Average Trust (DIA | A-71). This is a price-weighted fund. It’s not cap-weighted, but I can’t imagine anyone would think DIA is a smart-beta fund.

And so it goes, as you’ll see as I cart out a number of examples in this blog, the third in a series that lays bare the impossibility of defining smart beta. Part I explored the problem broadly, while Part 2 introduced the tools for testing smart-beta definitions.

In any case, many funds, such as the PowerShares FTSE RAFI US 1000 Portfolio (PRF | A-88), the Guggenheim S&P 500 Equal Weight ETF (RSP | A-75) and the iShares Select Dividend ETF (DVY | A-68), stake their smartness on their fundamental-, equal- and dividend-weighting schemes, respectively.

True, none of these funds, which seem to be legitimate smart-beta strategies, is cap-weighted. It’s so tempting to think that alternative weighting defines smart beta. And so wrong.

ETF.com tags funds by their weighting schemes in our ETF Classification System, which we call “ECS.” (Expand the “More Filters” section for full effect.) Notice two critical fields: “Selection” and “Weighting.”

Selection explains the process by which indexers choose which securities to include. Weighting shows how the indexer allocates dollars to each chosen security. These two fields describe the ways an index veers away from broad representation of its universe.

What Does ‘Cap-Weighted’ Mean?

It’s not clear what the phrase noncap-weighted means in a smart-beta context. Try it out in ECS—again, our ETF Classification System, by sorting all U.S.-listed ETFs according to their weighting scheme. You’ll find that some cap-weighted funds seem pretty smart, while plenty of noncap-weighted funds look remarkably simple.

As of April 9, 2014, a total of 912 of the 1,573 U.S.-listed ETFs, or 58 percent of them, are cap-weighted.

I mentioned the cap-weighted Vanguard Dividend Appreciation ETF (VIG | A-67) in the introduction to this series, but I didn’t show how, by selection alone, VIG builds a portfolio that is very different from the U.S. total market.

VIG restricts its portfolio to U.S. companies that have 10 or more years of annual dividend increases. VIG’s 145 constituent portfolio is smaller-cap than the U.S. total market, with lower price-earnings multiples (P/E ratios) and higher yields.

Indeed, many folks think VIG is a smart-beta fund, despite its cap-weighting.

I suspect that, when we talk about cap-weighting, we’re not talking about funds like VIG. To get to what we want, we need to get a little smarter about how we sort funds. I suggest we look for what we at ETF.com call “plain vanilla” funds.

Plain Vanilla

Plain vanilla means that an index fully represents the long side of the opportunity set of the market it targets. Critically, plain-vanilla indexes are market-cap-oriented in two ways: selection; and weighting. We do make allowances for S&P’s committee-based selection process and for adaptations needed in the futures markets, where the net exposure is zero.

Filtering ECS—our ETF Classification System—for plain-vanilla funds gives more sensible results than just filtering by market cap. If you try it (instructions in the endnotes), you’ll find the plain-vanilla funds really do seem to be broad-based representations of an overall market.

As of April 9, 2014, 810 U.S. ETFs are plain vanilla—that’s 51 percent.

It’s the other group, the 49 percent, that will make you squirm. That 49 percent is like the parking lot crowd outside a Grateful Dead show: You can find just about anything there—the smart, the stupid, the jugglers, the clowns and the wastrels.

More formally, alternative weighting fails as a smart-beta criterion because when it’s applied to all funds according to their structure it produces groups that aren’t widely acceptable within the ETF community.

Anyone who knows the history of indexing should not be surprised by this outcome.

Stock market indexes date back to the 19^th century. Today’s indexes serve myriad purposes, from offering broad access to a market, to maximizing tradability, to capturing narrow themes. Often, indexers use specialized selection and weighting processes to create investable products.

The seven test cases below show the folly of relying on selection and weighting schemes to define smart beta. I‘ll use these as examples to show you, one by one, how indexers ingeniously use selection and weighting to create investable products in challenging markets.

Their solutions are usually smart, but the indexes they build are nothing like the fundamental, dividend or low-volatility schemes you call to mind when you think of smart beta.

Funds That Text The Noncap-Weighting Aspect Of A Smart-Beta Definition

Every one of these funds’ indexes solves a problem: from hedging duration exposure, to maximizing tradability in an illiquid market, to providing compliant exposure to narrow industries or creating thematic exposure.

All of them are adaptive, though some have not aged well. They are all edge cases, testing the usefulness of a plain-vanilla smart-beta definition.

Hedged Exposure In The 51 Percent

There are two problems with looking for smart beta by flavor (vanilla versus everything else). There are some pretty exotic funds in the vanilla group, while some plain-Jane ETFs make it to the not-vanilla bunch.

Let’s start with the easier of the two problems, the one type of by-the-book plain-vanilla fund that seems smart.

The ProShares’ High Yield - Interest Rate Hedged ETF (HYHG | C) is cap-weighted and selected, but has an overlay that grossly alters its pattern of returns. HYHG shorts U.S. Treasury futures to manage interest-rate risk.

So, HYHG’s returns will be different from the overall high-yield market any time interest rates move. Hedging exposures is a clever, complex strategy, but HYHG’s cap weighting and selections leaves it stuck in the vanilla bucket.

Really, though, this isn’t the end of the world. I’m sure that if the hedged funds were the only problem with using alternative weighting as a smart-beta definition, we would deal with it as a minor exception.

But the mishmash of strategies in the alternatively weighted group—the 49 percent—is the bigger challenge. Let’s walk through the remaining test cases and I’ll show you why anything-but-cap-weighting fails as a smart-beta definition.

Flotsam In The 49 Percent

There’s a group of legacy funds in the 49 percent. These funds solve the problems of yesteryear, but live on in the ETF landscape. Our test cases Nos. 2 through 4 show all too well how some not-exactly-smart funds live alongside the hot new factor funds in the alternatively weighted bucket.

Idiosyncratic funds like the Nasdaq 100 ETF—the PowerShares QQQ (QQQ | A-52)—and the SPDR Dow Jones Industrial Average Trust (DIA | A-71) are not plain vanilla, so they land in the 49 percent.

The $46 billion QQQ restricts its selection universe to the Nasdaq. QQQ has its oddities—it excludes financials, limits the weight of top holdings, rebalances unpredictably and sandwiches an equally weighted tier between two cap-weighted ones.

In other words, QQQ tries to be smart, but not in the way you might think.

Nasdaq, advancing its status as an exchange, chose the index’s caps and the redistribution scheme to adapt to high concentrations in its listings like Microsoft or Apple. Although modern indexing’s breadth makes QQQ look ridiculous, the problem of concentration is still real. Bottom line: QQQ ain’t smart beta.

Also in the 49 percent is DIA, which tracks the Dow Jones industrial average, the granddaddy of all indexes. No question, measuring the stock market was a huge innovation in the 19^th century, but the Dow hasn’t aged well.

The Dow is selected by a committee and holds one share (split-adjusted) of each of its 30 constituents. Today it’s clear that the Dow is about as dumb an index as could be. Nobody wants to put DIA on a smart-beta list.

The 49 percent has other legacy funds too, some from much more recent vintages. Launched in 2002, when access to emerging markets was difficult and the iShares MSCI Emerging Markets ETF (EEM | B-100) was highly optimized, the BLDRS Emerging Markets 50 ADR (ADRE | B-35) large-cap and ADR-only construction created a tradable and hedge-able basket.

ADRE clocks in at 93 percent large-caps, and overweights Brazil, telecoms and energy. ADRE is not vanilla, but it’s not smart beta, either.

Avoiding Taxes Puts Equal Weighting In The 49 Percent

Our fifth test case, the SPDR S&P Bank ETF (KBE | A-54) tracks an equal-weighted index, as do all of the SPDR S&P U.S. industry funds. Equal weighting sure seems smart in RSP, but I would guess that the ETF community would be reluctant to give S&P’s industry suite a smart-beta label. Let’s see why.

Internal Revenue Service rules for registered investment companies require diversified funds to hold no more than 25 percent in any single security and no more than 50 percent cumulatively in positions above 5 percent. In narrow industries like banking, this matters.

As of April 1, 2014, the top five U.S. banks comprised 71 percent of the industry, according to Thomson Reuters. No. 1, Wells Fargo, clocks in at 19.4 percent, while No. 5, U.S. Bancorp, takes up 6.4 percent of the weight. No diversified fund registered under the Investment Company Act of 1940 could track a vanilla banking index.

MSCI solves this problem by capping the weight of the biggest positions; others create tiers. S&P’s equal weighting is somewhat of a drastic solution, but it’s better than double taxation. Bottom line? KBE is equal weighted, but it’s probably not smart beta.

Oddballs And Hippies In The 49 Percent

Our final two test cases show how indexers use selection and weighting to craft another world, where everyone can act like an accredited investor and where good deeds are rewarded and wrongdoers punished.

To capture specialized exposures such as private equity, social media, initial public offerings or up-and-coming Nashville, indexers turn to nonmarket-cap-related rule sets to isolate an economic theme.

The PowerShares Global Listed Private Equity Portfolio (PSP | F-58) faces a hard problem: Delivering access to private equity returns. Private equity by definition is not publically listed, but some general partners like business development companies and MLPs are exchange-traded. The Red Rocks Global Listed Private Equity Index tracks these publically listed firms.

The index attempts to mimic what Red Rocks describes as a typical institutional private equity portfolio by differentiating between early, mid- and late-stage firms. Their weighting looks through to the listed firms’ portfolios, targeting 65 percent late-stage, 25 percent mid-stage, and 10 percent early stage exposure. The weight of each tier is fixed; and within the tiers holdings are cap-weighted.

This is an inventive solution, perhaps even an advance for democracy. But, I suspect it’s not the sort of indexing revolution that smart beta envisions.

The iShares MSCI KLD 400 Social ETF (DSI | B-90) is a cap-weighted fund that deliberately selects companies that safeguard the environment, communities, human rights and other sensitive areas. MSCI’s socially aware methodology helps some investors feel like they can align their investments with their beliefs. DSI’s portfolio isn’t vanilla, but it’s not really smart beta either.

Leave Your Cap On

The simplicity of noncap-weighted is its downfall as a smart-beta definition. Indexers have been skillful, adaptive and inventive in creating investable products that access quirky parts of the market, using a rich set of selection and weighting schemes.

These adaptations are smart, and surprisingly common in the ETF landscape. Not every alternatively weighted fund fits our presumptions of what smart beta is.

As I mentioned when I introduced the smart-beta definition ground rules, the groupings that result from definitions of smart beta have to make sense to the people who will use it. Alternatively weighted funds fail this test.

Factor exposure does no better, and for the same reasons. So, my wrecking ball will slam into factor exposure in my next blog, and I hope you join in the fun.

Endnotes: using ETF.com’s Finder to replicate my results

Pure cap weighting:

Weight = market cap or value

Weight = single asset, niche excludes optimized

Plain vanilla: all of the above, plus:

Active per SEC = no (applies in all cases)

Case 1) Selection = market cap or value, Weight = market cap or value

Case 2) Weight = market cap or value, underlying index = S&P, Selection = proprietary (this step includes the S&P US 1500 and Global 1200 series, which are subject to profitability screens and committee selection but are nonetheless quite representative of the opportunity set).

Case 3) Weight = single asset, niche excludes optimized

Case 4) Selection = futures liquidity, Weight = production (this selects for the GSCI, which is the most market-caplike commodities index).

Case 5) Weight = market cap or value, Selection = AMT Free, credit downgrade, credit rating, developed market currencies, financials, industrials, maturity, revenue backed, time since listing, US$ denominated, utilities. (Sometimes fixed-income funds are so specific that they need four ECS fields to describe their market segment.)

At the time this article was written, the author held no positions in the securities mentioned. Contact Elisabeth Kashner, CFA, at [email protected].