Highlights

  • Benjamin Graham's investment philosophy centered on buying stocks at no more than a modest multiple of conservatively estimated average annual earnings.
  • Decades later, Fama and French established price-to-book as a core value metric, and it was the only input in one of the most prominent value indexes of the '90s and early '00s.
  • Today's most popular off-the-shelf value indexes are somewhat more sophisticated, but they still employ simplistic inputs that stray from core Graham and Fama-French principles, making value especially promising territory for smart-beta ETFs.
  • Using insights from our proprietary Equity Trading Strategy model, we consider the variables that best reveal value. We then evaluate the existing lineup of smart-beta Value ETFs based on how well they provide exposure to those variables.

Feature

What makes a stock a value stock? How does an investor put value investing principles into practice? A half-century of discussion has failed to yield universally agreed-upon answers. Even Benjamin Graham's Intelligent Investor, billed on the front cover of its recent editions as "the classic text on value investing," and "the definitive book on value investing," never used the term.

Based on our reading of Graham, we submit that an essential element of value investing is the identification of stocks that are temporarily trading below their intrinsic value. The temporary drag may persist for a while - stock markets can remain oblivious to fundamentals for extended stretches - but it is ultimately expected to dissipate. From the long side, value investing is a play on negative overreaction or neglect, and it is not a pursuit for the timid. Value investors not only have to be contrarians, they have to be contrarians with elevated levels of conviction and staying power.

This Special Report summarizes the antecedents of value investing before examining how they've been woven into conventional value benchmarks. In our view, core value investing principles have been attenuated and distorted by an index construction process that most closely resembles a game of Whisper Down the Lane. The shortcomings of the established indexes present a golden opportunity for smart-beta ETFs, low-cost vehicles that track indexes custom-designed to provide exposure to a particular factor or factors. Using the insights of our proprietary Equity Trading Strategy ranking model, we examine the metrics that could be used to construct a better value index and assess the current crop of value factor ETFs to determine which provide the fullest exposure to the optimal value metrics.

The Father Of Value Investing

The temporary nature of undervaluation is a recurring theme in Graham's Intelligent Investor. The stock market's ever-present proclivity toward overreaction ensures a steady supply of value opportunities: "The market is always making mountains out of molehills and exaggerating ordinary vicissitudes into major setbacks."1 "[W]hen an individual company ... begins to lose ground in the economy, Wall Street is quick to assume that its future is entirely hopeless and it should be avoided at any price."2 "[T]he outstanding characteristic of the stock market is its tendency to react excessively to favorable and unfavorable influences."3

Graham viewed security analysis as the comparison of an issue's market price to its intrinsic value. He advised buying stocks only when they trade at a discount to intrinsic value, offering an investor a "margin of safety" that should guard against significant declines. His go-to measure for assessing intrinsic value was a sober, objective estimate of average future earnings, grossed-up by an appropriate multiple. A low price-to-average-earnings ratio was the linchpin of his margin-of-safety mantra.

The Chicago Boys

Chart 1
A Factor For All Seasons
Chart 1

Fullscreen        BAN

Decades after Graham's heyday, University of Chicago professors Eugene Fama and Kenneth French granted value investing the academy's imprimatur. Their landmark 1992 paper4 found that low price-to-book ("P/B") stocks consistently and convincingly outperformed high P/B stocks (Chart 1). The metric was duly enshrined in value indexes, albeit without regard for the qualification at the beginning of the paper: "We use all nonfinancial firms. ... We exclude financial firms because the high leverage that is normal for these firms probably does not have the same meaning as for nonfinancial firms, where high leverage more likely indicates distress."5

Whisper Down The Lane

Russell claims to have developed the first value and growth indexes in 1987, as academic research into the impact of fundamental factors on equity returns was reaching a fever pitch. Morningstar developed its now-ubiquitous Style Box in 1992, entrenching the concept of style investing, despite the lack of agreed-upon definitions of value and growth styles. The most prominent value index of the 1990s, the S&P/Barra Value Index, used price-to-book as its sole input, but value index metrics eventually coalesced around price-to-book, price-to-sales ("P/S") and price-to-earnings. Price-to-cash-flow and dividend yield are also familiar inputs and some indexes, like Morningstar's own, attempt to introduce a forward-looking element by using forward P/Es.

The final result bears no more than a passing resemblance to Graham's, and Fama and French's, work. Graham's perspective was squarely forward-looking; although he looked to the past for a sense of realistic earnings potential, he focused unwaveringly on future cash flows. Fama and French, meanwhile, explicitly excluded Financials from their sample. Their findings are most accurately expressed as, "low-P/B nonfinancial stocks outperform high P/B nonfinancial stocks," and yet value indexes would be unrecognizable without Financials (Chart 2).

The heavy Financials tilt has reduced the Value/Growth decision to a matter of sector preference, as Tech stocks' growth index share mirrors Financials' value index share (Chart 3). It also illustrates a problem: banks have taken up permanent residency in most value indexes. Graham sought to exploit the stock market's tendency to overreact, and indexes following his lead should therefore be characterized by regular churn, reflecting the waves with which various stocks go in and out of favor. Instead, for Financials in general, and banks in particular, conventional value indexes have come to resemble the Hotel California (Box 1).

Chart 2
Stagnant Methodology...
Chart 2

Fullscreen        BAN

Chart 3
...Has Reduced Style Indices To Sector Proxies
Chart 3

Fullscreen        BAN

BOX 1

The Bank Bias Problem

Chart 4
Low Margins = Low Multiples
Chart 4

Fullscreen        BAN

Banking is a low-margin commodity business, and banks borrow multiples of their equity base to earn an acceptable return. Outsize leverage renders banks structurally distinct from companies in other industries and frustrates cross-industry comparisons. BCA therefore excludes financial companies from its analyses of aggregate corporate debt, and Fama and French reasonably excluded them from their sample.

Banks earn low margins because their core function is to take deposits and make loans, and the width of lending spreads is limited.6 Since the early '80s, banks' average net lending profit margin has been less than 1.5% (Chart 4, top panel), based on their average 3.8% net interest margin (Chart 4, middle panel) and average 63% efficiency ratio (the share of each dollar of interest income that goes to pay non-interest expenses; Chart 4, bottom panel). Investors logically pay less for banks' comparatively low-return assets, and banks trade at persistently low multiples of book value. Similar analysis applies to other financial services firms, like insurers and specialty lenders, which deploy sizable balance sheets in the service of modest spreads.

The balkanized history of American banking regulation, which barred operation across state lines until the early '80s, also weighs on bank book multiples. Every bank with a regional or national footprint is the product of dozens, if not hundreds, of mergers, which have left considerable goodwill on the surviving entities' balance sheets. Investors haircut the value of goodwill "assets," further dragging down price-to-book multiples. Bank analysts typically look through banks' P/B multiples and focus instead on their tangible book multiples, acknowledging that sizable banking footprints were bought, not built.

An Ideal Value Index

We identify four broad shortcomings of off-the-shelf value indexes:

  • They exclusively use trailing multiples, a rear-view mirror metric.
  • They rely on simple price-to-book multiples, which flatter serial acquirers.
  • They rely entirely on reported earnings, which are an imperfect proxy for cash flow. A share of stock ultimately represents a claim on its issuer's future cash flows.
  • They make no attempt to place relative metrics into historical context. Without a mechanism to compare a particular segment's valuation relative to its history, structurally low-multiple stocks will be over-represented and structurally high-multiple stocks will be under-represented.

Our Equity Trading Strategy model goes a long way to rectifying these flaws in its value factor inputs. It includes forward P/E multiples alongside trailing multiples, it uses P/TBV instead of P/B, and it also incorporates price-to-operating-cash-flow ratios ("P/OCF"). It does not consider valuation relative to history, which might further limit the bank over-representation problem, but its top-rated value stocks have nonetheless far outpaced the Russell 3000 Value Index over time (Chart 5).

Chart 5
A Better Mousetrap
Chart 5

Fullscreen

An Empirical Review Of Value Factors

Our Equity Trading Strategy ("ETS") model's composite value inputs have outperformed the off-the-shelf value indexes by a comfortable margin, but that does not necessarily mean that each ETS component has added value. In this section, we review the empirical performance of each of ETS' advanced inputs relative to the basic value index inputs. We compare ETS' combined trailing and forward P/Es to the basic trailing P/E, we compare ETS' P/TBV to the basic P/B and we compare ETS' P/OCF to the basic trailing P/E.

Trailing And Forward P/E Versus Trailing P/E

An equally-weighted portfolio long the top 30% and short the bottom 30% of U.S. stocks ranked on a score incorporating both forward and trailing P/E would have outperformed a similar strategy investing in stocks ranked by trailing P/E only7 (Chart 6). Paying less for each dollar of earnings clearly matters, as each strategy piles up gaudy market-neutral returns over 18-plus years, but ETS' 7.8% CAGR is a meaningful improvement on the basic metric's 6.6% CAGR. The empirical result squares with theory: researchers have shown that share prices tend to respond with a delay to the information in analysts' earnings forecasts.8 The ETS model's express attempt to capture the resulting market anomaly by incorporating forward P/E in its value inputs has been rewarded over the last two decades.

Chart 6
Incorporating Forward P/E Improves Performance
Chart 6

Fullscreen

Price-to-Tangible-Book Versus Price-To-Book

An equally-weighted portfolio long the top 30% and short the bottom 30% of U.S. stocks ranked on the basis of P/TBV would have outperformed a similar strategy investing in stocks ranked by P/B9 (Chart 7). The tangible-book/book gap is smaller than the forward and trailing/trailing P/E gap, above, but the market-neutral returns (8.3% and 8% CAGR, respectively) are even greater. We conclude that price-to-book is a meaningful measure of value, and that using price-to-tangible-book marginally improves a value index's performance. The improvement is likely related to tangible book's slight correction for the advantage large acquirers gain from simple book, as mergers of equals generally fail to recoup the premia paid to consummate them.

Chart 7
Tangible Improvement
Chart 7

Fullscreen

Price-to-Operating-Cash-Flow Versus Trailing P/E

An equally-weighted portfolio long the top 30% and short the bottom 30% of U.S. stocks ranked on the basis of P/OCF would have outperformed a similar strategy investing in stocks ranked by trailing P/E10 (Chart 8). The 60 basis-point annual divergence between ETS' OCF metric and the conventional trailing earnings metric falls between the earnings and book-value divergences, but it was as wide as the forward/trailing vs. trailing earnings divergence through the P/OCF peak in July 2014. Earnings are a concept, and their calculation is subject to potentially significant management estimates. Operating cash flow, by contrast, is an unambiguous measure that may provide a better window into corporate performance, and the market-neutral strategy repeated throughout this section generated its highest return, 8.8%, when applied to P/OCF.11

Chart 8
Less Discretion Makes For A Better Metric
Chart 8

Fullscreen

Smart-Beta Fund Evaluation

The success of the ETS model's advanced metrics provides a sound empirical basis for using the composite value score to evaluate the constituent makeup of the 21 U.S. smart-beta value funds (Table 1). We also evaluate, on the same basis, the six Russell and S&P 500 Value Index trackers that form our off-the-shelf control group. Our control group ETFs consistently rank in the lowest deciles of our value ETF subset, confirming our intuition that value is a promising segment for smart-beta ETFs. Capsule reviews of some of the highest- and lowest-ranking ETFs follow. (For an explanation of the evaluation methodology, please refer to the Appendix.)

Table 1
Universe Of Value Smart-Beta Funds And Control Group Funds
Table 1

Fullscreen

Guggenheim S&P 500 Pure Value ETF (RPV)

RPV tracks the S&P 500 Pure Value Index, composed of the highest-ranked third of stocks in the S&P 500 Value Index. Filtering out the bottom two-thirds of the standard index limits membership to the cheapest stocks on a P/B, P/S and trailing P/E basis, and allows the Pure Value index to steer clear of the overlapping style index membership that plagues the standard index. RPV's ETS value scores have consistently ranked in the top two deciles, with high relative price-to-sales scores, and strong P/TBV and forward P/E showings. In line with its high ETS model scores, RPV has solidly outperformed IVE and IWD, its control group peers (Chart 9). With ample liquidity (average daily turnover exceeds $13 million) and a reasonable expense ratio (0.35% annually), we recommend RPV for investors seeking large-cap value exposure.

Chart 9
RPV Has Crushed Its Control Group ...
Chart 9

Fullscreen

Guggenheim S&P MidCap 400 Pure Value ETF (RFV)

Considerably lesser liquidity (average daily turnover below $1 million) and modest historical outperformance (Chart 10) make RFV, the mid-cap equivalent of RPV, a slightly more difficult call. We are sold on the less-is-more Pure Value methodology, though, and expect RFV will outperform IJJ by a wider margin over time. Large investors can overcome RFV's lesser activity by transacting directly with sponsors via creation and redemption units, and market makers' built-in arbitrage incentive12 should shield secondary-market investors. We recommend RFV as a source of U.S. mid-cap value exposure.

Chart 10
... And RFV May Yet Do So
Chart 10

Fullscreen

Powershares S&P 500 Value Portfolio (SPVU)

SPVU tracks the S&P 500 Enhanced Value Index, an even more selective and concentrated version of the S&P 500 Value Index than the Pure Value Index, including only the top 100 value stocks in the S&P 500 and weighting them more aggressively. The fund's limited history (it was launched in October 2015), modest AUM and sporadic trading activity keep us from ranking it ahead of RPV as a large-cap value option. Its head-of-the-class ETS model scores will keep it on our radar, however, and it could become a viable option if it can achieve critical mass.

Diamond Hill Valuation-Weighted 500 ETF (DHVW)

At the opposite end of the spectrum, DHVW consistently records bottom-decile ETS model scores. Its proprietary discounted cash flow index construction methodology may have some merit, but it will be hard-pressed to display it when selecting 500 of the 700 largest stocks. The less-is-more success of the concentrated Pure Value indexes suggests that investing in 71% of the stocks in a universe diffuses a model's skill.

Powershares Russell 2000 Pure Value ETF (PXSV)

The ETS model also consistently ranks PXSV and VLU among the least of the smart-beta value ETFs. We like a pure value approach, but applying it only to all stocks' price-to-book multiples, as the Russell indexes do, undermines PXSV's chances of uncovering true value opportunities.

SPDR S&P 1500 Value Tilt ETF (VLU)

Seeing the good in everyone is an estimable quality in a human being, but a dubious feature of a smart-beta ETF. With 1,359 holdings, VLU finds a place for 90% of the constituents in its benchmark index, too high of a proportion to support through-the-cycle outperformance.

Doug Peta, Vice President
Global ETF Strategy
dougp@bcaresearch.com

Jennifer Lacombe, Research Analyst
Global ETF Strategy
Jenniferl@bcaresearch.com

Philippe Morissette, Associate Vice President
Equity Trading Strategy
philippem@bcaresearch.com

  • 1 Graham, Benjamin, The Intelligent Investor, HarperCollins: New York, 2005, p. 97.
  • 2 Ibid, p. 15
  • 3 Ibid, p. 189
  • 4 Fama, Eugene F. and French, Kenneth R., "The Cross-Section of Expected Stock Market Returns," The Journal of Finance, Volume 47, Issue 2 (June 1992), pp. 427-465.
  • 5 Ibid, p. 429. The paper also examined the effect of company leverage on equity returns, but banks' capital structure also impacts their P/B multiples, as discussed below.
  • 6 Borrowers would borrow from non-bank lenders if bank lending rates were too high, and non-bank lenders and direct lending would draw deposits away from banks if bank deposit rates were too low.
  • 7 Stocks with market cap exceeding $50 million, ranked in reverse order of multiples.
  • 8 Elgers, Pieter T. and Lo, May H. and Pfeiffer, Ray J., Delayed Security Price Adjustments to Financial Analysts' Forecasts of Annual Earnings. The Accounting Review, forthcoming. Available at https://ssrn.com/abstract=272322
  • 9 Stocks with market cap exceeding $50 million, ranked in reverse order of multiples.
  • 10 Stocks with market cap exceeding $50 million, ranked in reverse order of multiples.
  • 11 From April 2005 to the OCF strategy's peak in July 2014, its CAGR was a whopping 10.1%.
  • 12 Please see Global ETF Strategy Special Report, "ETFs 101: How They Work, How They Trade, And Why It Matters," published October 26, 2016, available at etf.bcaresearch.com.

Appendix

Only ETFs that track indexes designed to provide specific exposure to a particular factor or factors in a way not offered by off-the-shelf indexes meet our definition of smart-beta ETFs. We exclude actively managed ETFs because they are not nearly as predictable as index-tracking funds, which rebalance and reconstitute their constituents at regularly defined intervals. Because of the paucity of non-U.S. smart-beta value ETFs (four, currently), we limit our analysis to funds investing in U.S-listed companies.

There are currently 21 U.S. smart-beta value ETFs that meet our criteria. We have evaluated them, along with six traditional value style ETFs that track all market-cap tranches of the two most widely used value index families (Table 1). We incorporate the six plain-vanilla style funds to serve as a control group and benchmark for our analysis of smart-beta strategies.

We have not filtered out any funds based on AUM or liquidity criteria. Our goal is to shed light on the index construction methodology and its ability to provide exposure to the factors we have found to have the potential to outperform over time. The smart-beta concept is still relatively new, and we have refrained from applying any filters lest they keep us from discovering potentially useful index-construction innovations.

We have calculated an ETS Value Score for each fund in our universe. The ETS Value Score is a composite of forward P/E, trailing P/E, P/OCF, P/S and P/TBV metrics, weighted by the individual metrics' proportional weights as of February 2017, which provides a reasonable approximation of their proportional weights over time. Using historical monthly ETF constituent weighting data from September 2011, we have compiled up to 65 monthly observations for every ETF in our smart-beta value universe.