SSRN Id3442539 PDF
SSRN Id3442539 PDF
SSRN Id3442539 PDF
Baruch Lev*
Philip Bardes Professor of Accounting and Finance, Stern School of Business, New York
University
Anup Srivastava
ABSTRACT
It is widely believed that the long-standing and highly popular value investing strategy—investing
in low-valued stocks and selling short high-valued equities—lost its edge in the past 10-12 years.
The reasons for this putative failure of value investing elude investors and academics, making it a
challenge to assess the likelihood of the return of value investing to its days of glory. Based on
extensive data analysis we show that value investing has generally been unprofitable for almost 30
years, barring a brief resurrection following the dotcom bust. We identify two major reasons for
the failure of value investing: (1) accounting deficiencies causing systematic misidentification of
value, and particularly of glamour (growth) stocks, and (2) fundamental economic developments
which slowed down significantly the reshuffling of value and glamour stocks (mean reversion)
which drove the erstwhile gains from the value strategy. We end up by identifying the type of
companies (stocks) that may still generate gains from value investing.
---------------------------
*Corresponding Author: E-mail blev@stern.nyu.edu, Leonard N. Stern School of Business, Kaufman Management
Center, 44 West Fourth Street, 10-92, New York, NY 10012. We thank Partha Mohanram, Dushyant Vyas, Yrjo
Koskinen, Alex David, Joshua Livnat, Yakov Amihud, and James Mackintosh for helpful comments. This research
benefitted from participants’ comments at seminars, conferences and webcasts organized by New York University,
Rutgers University, Bernstein Quant Conference, and Evercore ISI webcast.
(“value”) stocks and shorting highly-valued (“glamour”) equities, thereby capturing companies
fundamentals. The price reversals of these misvalued stocks drive the gains from value investing.
The value investment strategy yielded excess (abnormal) returns for decades, leading to the
proliferation of “value funds” and strategies offered to investors, yet around 2007, the strategy
appears to have lost its magic. A google search of the “death of value investing” and related
morbid terms yields hundreds of articles, including in Forbes, Barrons, The Wall Street Journal,
Seeking Alpha, Bloomberg, and Financial Times. The Economist (October 27, 2018, p. 71), under
the title “The agony of the value investor,” quoted a prominent value fund manager informing
investors: “Our results have been far worse than we could have imagined.” Another manager
wrote: “The market is telling us we are wrong, wrong, wrong about almost everything.” And yet,
all those self-flagellations by value managers and the value obituaries by media pundits don’t
• Why did value investing, which yielded steady excess returns for decades, abruptly lose its
• Is the recent funk of the value strategy temporary, or a long-term fall from grace? Is “value
We will answer both questions by showing, based on extensive data analysis, that flaws in the
accounting for intangibles, combined with fundamental economic shifts, have undermined the
value strategy.
The value strategy is often traced to Ben Graham1, and is based on the premise that among
low-valuation (out-of-favor) stocks are many undervalued ones, relative to fundamentals, due to
investors’ overreaction to bad news, such as an accounting scandal or an unexpected loss. Once
investors realize the undervaluation, the share prices of the undervalued stocks will rise and revert
stocks”―should yield above market returns. A similar argument can, of course, be made about
high-valuation stocks, which likely include overvalued equities, bound to fall in price over time.
excess returns. Thus, a combined strategy of “long on value and short on glamour”―the “value
Indeed, academic research consistently documented excess returns to the value strategy,
until the past 10-12 years. Lakonishok et al. (1994), for example, reported that investing every year
(on June 30) in the 30% of the stocks with the lowest market-to-book ratio―“value stocks”―and
selling short the 30% of the stocks with the highest market-to-book ratio―the “glamour stocks.”—
yielded over the period 1968–1989 a mean annual return of 6.3%, and a whopping 34.4% for a
rolling three-year holding period. A similar value strategy, also based on the market-to-book ratio,
is the HML (high minus low) methodology introduced by Fama and French (1992, 1993). It differs
slightly from Lakonishok et al. (1994) in that the long-short portfolios were determined after first,
dividing the sample firms into large and small by median market capitalization (thereby adjusting
for firm size), and then investing in the 30% lowest value stocks and shorting the 30% highest
valued stocks, of both large and small firms. The Fama-French HML strategy averaged almost 5%
1
See, Graham, Benjamin, 1949, The Intelligent Investor, Harper & Brothers, New York.
considerable attention, getting almost 33,000 academic citations by June, 2019, one of the highest
citation counts in the finance literature. The HML classification also became a standard control
factor (arguably, for risk) in studies examining abnormal stock returns. Based on the superior
While value investing is premised on identifying stocks whose prices understate their
intrinsic value and those who overstate it, there is no universally accepted way of implementing
the value strategy, leading to multiple value measures used in practice and academic research.2 For
example, in addition to the book-to-market ratio, Lakonishok et al. (1994) examined cash flow and
earnings to market value. Piotroski (2000) and Mohanram (2005) advance value strategies that add
“quality” parameters to the book-to-price (B/P) ratio. Obviously, we cannot examine all the value
indicators used by researchers and investors. To keep our analysis focused and readily
interpretable, we primarily identify value and glamour stocks by the market-to-book ratio, since it
is the most frequently used indicator by researchers (Lakonishok et al., 1994; Fama and French,
1992 & 1993; Kok et al., 2018; Ball et al. 2019.), and by investment institutions (Russell Value
Stocks, S&P, and MSCI Value indexes).3 For robustness check, we also examine another popular
2
MSCI comments: “Value investing is premised on identifying stocks whose prices seem to understate their intrinsic
value. While many institutional investors may agree with that premise, implementation of value-index strategies
differs widely.” See https://www.msci.com/documents/1296102/1339060/Factor+Factsheets+Value.pdf
3
In addition, S&P and MSCI use sales/price ratio and enterprise value to cash flow ratio (EV/CFO), respectively. See
https://us.spindices.com/documents/methodologies/methodology-sp-us-style.pdf and
http://www.ftse.com/products/downloads/Russell-US-indexes.pdf
methodology adjusts for firm size and has been subjected to most academic rigor, as is evident
It’s instructive to examine the 10-year growth of a dollar invested in the Fama-French value
strategy at the beginning of each decade, from the 1970s to the present. Note that this is a hedged
investment since we both buy and sell equal values of low market-to-book stocks and high market-
to-book stocks. Figure 1 presents each decade’s cumulative return from this annually-updated
long-short value strategy. Thus, for example, a dollar invested in the long-short portfolio on
January 1, 1970 yielded by the end of the decade $2.02 (102% return), and a dollar invested on
January 1, 1980 yielded by decade-end $1.75 (75% return). (The term investment is a misnomer
here because one dollar of stocks is bought while another dollar of stocks is sold short.) These
However, this seems to have been the “swan song” of value investing: From 1989 on, the
strategy faltered, mainly because of the tech bubble of the 1990s which elevated the valuations of
4
We follow Fama and French (2003) to classify firms into the value and glamour categories. Stock returns and
accounting data are obtained from CRSP and Compustat, respectively. All firms traded on NYSE, Amex, and
NASDAQ are initially included. Delisted stock returns are taken from CRSP. Book value for computing the book-to-
market ratio is taken from t−1 financials statement for June to December fiscal year end and t−2 for January to May
firms. Negative book value firms are excluded. The numerator, market value of equity, is obtained for the end of year
t−1. Portfolios are formed in June 30 of year t. All firms are divided into six equal groups [two groups by market value
and three groups by market-to-book (highest 30%, middle 40%, and lowest 30%)]. Portfolios with the highest (lowest)
30% market-to-book are called glamour (value). Monthly returns are for each portfolio from July 1 of year t to June
30 of year t+1, value weighted, based on market value as of June 30 of year t. HML returns are computed as one-half
the returns from going long on large and small value portfolios, and one-half the returns from shorting large and small
glamour portfolios. We use SAS code obtained from WRDS (Wharton Research Data Services) to identify stocks by
HML classification (https://wrds-www.wharton.upenn.edu/pages/support/applications/risk-factors-and-industry-
benchmarks/fama-french-factors/).
Fama-Frech HML strategy is originally based on the book-to-market ratio. We use in this paper the inverse
term market-to-book ratio due to its greater familiarity. Since our sample is restricted to positive book value firms, the
ranking of stocks is not affected using either metric.
losing proposition. A dollar invested in the strategy at the beginning of the 1990s would lose 10%
by the end of the decade. Moving forward in Figure 1, the first few years of the 2000s saw a brief
resurgence of the value strategy, driven primarily by the success of shorting glamour stocks, due
to the burst of the tech bubble: Prices of erstwhile glamour (growth) companies
plummeted―Nasdaq fell in 2000 by 55%―and 17% of the small glamour companies failed
altogether and were delisted. Shorting all those losers substantially boosted the long-short value
strategy, leading to good performance until 2006. The flight from the collapsing tech to the more
stable value stocks in those years further boosted the value strategy.
The good performance of the value strategy in the early 2000s apparently looms large in
the minds of relatively young investors who started investing in the 21st century, making it so hard
for them to fathom the “demise of value” since 2007. However, as Figure 1 makes clear, the value
strategy had already lost much of its potency in the late 1980s, and yielded negative returns in the
1990s, barring a brief resurgence in 2000-2006. This then leads to our rephrased question: What
caused value investing to lose its consistent edge by the late 1980s?
Figure 1: Growth in Value of One Dollar Invested in the Long-Short Value Strategy on
Jan 1 of the First Year of Each Decade, 1970-2018
the market-to-book ratio—is measured with considerable error. Many asset values on the balance
sheet are not updated to reflect their current values, and earnings—whose cumulative, retained
value-irrelevant (one-time) items. But an increasing source of book value mismeasurement is the
creating, intangibles, such as R&D, IT, brand development, and human resources. This expensing
of intangibles, which are obviously investments intended to generate future profits, leading to their
absence from book values. This omission of major investments from book values started to have
a major effect on financial data (book values, earnings) from the late 1980s, due to the growth of
was a major contributor to the failure of value investing which started in the late 1980s, recall
Figure 1.
In Ben Graham’s time and up to the late 1980s, corporate investments were primarily in
tangible (physical) assets (property, plant, equipment, structures, airplanes, etc.), which are
capitalized (considered assets) by accounting rules and, therefore, fully reflected (net of
depreciation) in companies’ book values (equity). This inclusion of most corporate investments in
book values was reflected, among other things, by the median market-to-book ratio of public
companies which hovered around 1.0 until the mid 1980s. Accordingly, market values, being lower
or higher than the book values, often reflected under- or overvaluation of stocks. From then on,
gross value added.5 Currently in the U.S., the intangible investment rate of the corporate sector is
roughly twice that of the tangible investment rate, and the gap keeps growing.6 In absolute terms,
Back to the value strategy, a firm investing heavily in R&D, IT, brands, or business
due to its understated denominator of the market-to-book ratio, whereas in reality its valuation
isn’t excessively high when book value is properly measured. Furthermore, two similar companies,
one generating its intangibles internally by developing patents, for example, whereas the other
acquiring patents from other firms, will have substantially different market-to-book ratios because
the former’s book value is significantly smaller than the latter’s. Same with the Price-Earnings
ratio. Reported earnings of companies with increasing investments in intangibles are understated,
5
Figure 2 is from Corrado and Hulten (2010), and kindly updated by the authors’ request.
6
Enache and Srivastava (2018) report a similar pattern from analyzing company-level data.
7
Intangible investments in other countries lag considerably behind the U.S. China comes second at about $700
billion annually, and Northern European countries at $100-200 billion each. Most other countries have negligible
investment in intangibles.
increasing size of corporate intangible investments, the misspecification of book values and
earnings is substantial and growing, and is represented, among other things, by a median market-
to-book ratio of close to 3.0 in 2018.8 Some observers may contend that value strategies relying
on cash flows overcome the above accounting deficiency. However, cash flows too are calculated
after the deduction of intangibles, and therefore, do not solve the accounting-deficiency discussed
above.
We run an experiment to show how the accounting mismeasurement of book value could
affect the returns from value strategy by recomputing companies’ book values after capitalizing
the expensed intangible investments. The essence of our capitalization procedure is as follows: For
every public company and year, we capitalize (consider as an asset) its annual R&D expense, and
amortize the cumulative R&D capital (the sum of the capitalized past annual R&D expenses),
according to industry-specific R&D amortization rates reported in Li and Hall (2018). The annual
amortization of the R&D capital thus replaces the expensing of current R&D outlays in firms’
income statements, and, importantly, the unamortized R&D capital—an asset—is added to book
value.
We also capitalize a part of SG&A (sales, general & administrative) expenses, following
Enache and Srivastava (2018), since many non-R&D intangible investments, such as in brands,
IT, business processes, and human resources are included in SG&A expenses in the income
statement. However, SG&A also includes regular expenses, such as sales commissions and
administrative salaries. Following Enache and Srivastava (2018), we separate expenses from
8
The wholesale expensing of intangibles also renders many firms ineligible for inclusion in our value strategy
because of their negative book values.
regression of SG&A expenses on current revenues (scaled by total assets) and two dummy
variables representing losing firms and sales-decreasing companies. The portion unexplained by
current revenues in this regression equation represents the average amount of intangibles included
in SG&A, for that firm and year. We match a firm to its industry progressively by four-digit, three-
digit, two-digit, or one-digit SIC code, depending on data availability, to estimate the regressions.9
We thus maintain a rolling stock of the capitalized intangibles in SG&A expenses, and amortize
Using this methodology, we adjusted the book values of all firms by adding to the reported
(GAAP) values both the unamortized R&D and SG&A capitals. We then recomputed companies’
market-to-book ratios, using the adjusted book values. We note that additional firms were included
in the feasible set of value and glamour stocks as their book values turn positive by the
causing substantial reassignments of firms to value and glamour portfolios. Finally, we recalculate
the returns to the adjusted value strategy. The impact of our book value intangibles adjustment is
quite dramatic. The recalculation of book values changed significantly the composition of the top
and bottom 30% of companies ranked by market-to-book. For example, in 2017, for the top 30%
(glamour) companies, 455 firms changed classification after capitalization (255 enter and 200
leave, of an initial total of 921), and 524 companies (271 enter and 253 leave, of an initial total of
929) changed classification in the bottom (value) 30% category. Thus, roughly 40-60% of value
and glamour stocks changed classification due to our intangibles book value adjustments in recent
years. These major changes had a significant effect on the returns from the value strategy.
9
The empirical procedure requires at least 15 firm-year observations by industry.
10
value strategy. The red (left) bars in each year are the returns from the original, unadjusted market-
to-book classification, whereas the blue (right) bars represent the returns from the adjusted
methodology. While our book value adjustment didn’t change much of the strategy’s performance
in 1970s, when intangible investments were low, it began to yield improved returns in the 1980s,
when intangibles gained prominence (see Figure 2). All in all, in 34 out of the 39 years examined,
1970-2018, the returns from the adjusted value strategy were higher than those of the conventional
strategy (based on GAAP-reported book values), and in most years the adjusted returns were
substantially higher.
Figure 3: Cumulative Returns to the Original and Adjusted Value Strategy in Each
Decade, 1970-2018
Thus, for example, in the 1980s, while a dollar invested in January 1, 1980 in the
conventional strategy grew to $1.75 by decade-end, a dollar invested by the adjusted strategy grew
to almost $2.86, a 68% difference! Strikingly, in the 1990s, when the conventional value
11
at decade-end. The differences in investment gains continued to be substantial in the early 2000s,
and even in the recent period, 2010-2018, when the conventional value strategy lost its edge and
yielded negative returns, the adjusted strategy still generated reasonably positive gains. Thus, and
this is reported here for the first time, the adjustment of book values for the glaring accounting
deficiencies of intangibles expensing could have quite a dramatic effect on the long-short value
The effect of our intangibles book-value adjustments are, as expected, more pronounced
for glamour than for value stocks, since most glamour (highly-valued) companies are intangibles-
intensive. And among glamour stocks, our adjustments had a larger effect on small than on large
companies, since small, high-growth glamour firms tend to invest heavily in intangibles, causing
their reported book values to be highly misspecified. Our adjustments thus improve significantly
adjustments were less consequential for value stocks, many of which don’t invest much in
intangibles. The additional gains from our book value adjustments in the long-only, pure value
strategy (no shorting) were modest in the 1990s, very large during 2000-2009, and essentially
vanished thereafter.10
So, while the conventional (GAAP) mismeasurement of book value explains a great deal
of the failure of value investing, it doesn’t completely resolve the issue as far as the past 10-12
years are concerned. Even with our book value adjustments, the long-short, and more so the long-
10
Since the long-only, pure value strategy is not a hedged investment, we consider the investment returns net of the
market return in each year.
12
To generalize our findings we replicated the tests conducted so far with the market-to-book
ratio using another popular valuation metric: the price-earnings (PE) ratio. Figure 4 portrays the
year-by-year gains from investing in the 30% lowest PE stocks (after adjusting for firm size) and
shorting the highest PE stocks at the beginning of each decade. The picture emerging from using
the PE metric resembles closely that of the market-to-book ratio in Figure 1, except that with the
PE, the 1990s were also profitable. Figure 4 also juxtaposes the PE annual returns over the MB
returns and shows a high correlation (89%) between the two, with a higher volatility to the PE
returns.
Figure 4: High Correlation between HML Returns of PE and MB value Strategies
We also adjusted the earnings of the price-earnings ratio for the intangibles’ capitalization:
adding back to earnings the annual R&D expense and the part of SG&A related to intangibles, and
subtracting from earnings the annual amortization of the R&D and SG&A capitals. As expected,
the effect of intangibles’ capitalization on earnings was substantially lower than that on book
13
Figure 5: Cumulative Returns to the Original and Adjusted PE-based Value Strategy in
Each Decade, 1970-2018
In any case, whether guided by the MB ratio or the PE ratio, the recent 10-12 years returns
from value investing were unusually low, leading us to continue the quest for the reasons for the
From a statistical point of view, the gains from value investing derive from the mean
reversion of the highest (glamour) and lowest (value) ranked stocks. When ranked by the market-
to-book ratio (or by alternative rankings, like price-to-earnings), some of the highly ranked stocks
will drop in value over time; being ranked at the top they can only remain at the top, or drop down
due to deteriorating operations (a sales decline), or from other revisions in investor valuations.
This mean reversion from the top generates the gains from shorting glamour stocks. Similarly,
14
operations or other factors and escape the lowly value class, thereby generating the gains from
The mean reversion of value and glamour stocks isn’t unique to capital markets. In fact,
mean reversion is ubiquitous to many phenomena in life, yet often misunderstood or overlooked.
Whenever subjects are ranked by a score, such as people ordered from highest to lowest levels of
blood cholesterol, or investment funds ranked from top to bottom on performance, some of the
observations at the top and bottom ranks will revert to the mean observation over time.11
The reason for mean reversion is that the specific ranking of a subject at a point in time,
say a football team, is due to systematic (fundamental) factors, such as players’ talent and size of
fan base, as well as to random (transitory) factors, like injuries. Overtime, the transitory factors
average out, and the subjects ranked at the top and bottom revert to the mean. Similarly, value and
glamour stocks are ranked at the bottom and top due to systematic factors, like a strong patent
portfolio or a weak product mix, as well as to random (transitory) factors, such as a contract
cancellation, or an SEC inquiry. The random (transitory) factors will average out over time, leading
to mean reversion of both value and glamour stocks. The extent and speed of mean reversion is
determined by the relative size of the random to fundamental factors—the larger the relative size
Lakonishok et al.’s (1994) explanation of the consistent gains from value investing as
earnings streak as a long-term trend—essentially argues that investors exaggerate the effect of the
11
On mean reversion in the investment context, see Mauboussin (2012).
12
Michael Jordan was ranked for many years at the top of professional basketball players: a record 50 appearances on
the cover of Sports Illustrated. This absence of mean reversion was due to the fact that the random element in Jordan’s
performance was negligible relative to his fundamental, exceptional talent.
15
one). The quick reversals of past sales and earnings trends documented by Lakonishok et al.
(1994), corroborated investors’ misinterpretation of temporary effects (past growth) for permanent
Accordingly, an understanding of the failure of value investing in the past 12 years should
start with exploring changes in the extent and speed of the mean reversion of value and glamour
stocks.13
We measure stocks’ mean reversion in three ways: rank correlation, length of stay in the
value or glamour categories, and large stock price upticks and downticks. Rank correlation
indicates the correlation of a stock’s market-to-book rank (relative to all stocks) at the end of a
given year, with its rank in the previous year. Fast mean reversion is reflected by low rank
correlation. Figure 6 portrays the grouped annual rank correlations of large value and glamour
stocks, during 1989-2018. It is evident that the rank correlations of both value and glamour stocks
increased quite substantially during 2007-2018, and are now the highest since the 1970s. For both
groups the rank correlations increased from the late 1980s, with glamour stocks jumping from 45-
47% to 60% in 2007-2018. This substantial increase in rank correlation reflects a significant
slowdown of the mean reversion of both value and glamour stocks during 2007-2018. Hence the
13
We focus in the subsequent analyses on the 50% largest value and glamour stocks, since our adjusted strategy
yielded above-market returns for the small value and glamour stocks. Since the returns from the value strategy
presented in this paper are value-weighted by capitalization, large stocks have a dominant effect on the returns of the
total sample (large and small stocks).
16
Our second mean reversion measure is the length of stay of a particular stock in the value
or glamour portfolios. The longer the stay, the lower the mean reversion. Figure 7 shows
substantial increases in the average length of stay: Value stocks increased from 2.5 years, on
average, during 1989-2006 to 3.3 years in 2007-2018 (a 32% increase), while for glamour stocks
the increase was from 3.5 to 4.5 years (a 28% increase). These increases of length of stay in
category corroborate the substantial slowdown of stocks’ mean reversion in recent years.
Our third measure of mean reversion is the one most directly related to the gains from value
investing: it reflects the frequency of large (10% or more) stock price upticks for value stocks and
17
reversion of value and glamour stocks, and they generate the gains from the long-short value
investing strategy. Figure 8 presents the periodic annual percentages of 10% or more upticks (for
value) and downticks (for glamour stocks). In both cases, the price-change frequency during 2007-
2018 was substantially lower than in previous periods: the 10% or more upticks for value stocks
decreased from 22% annually during 2000-2006 to just 10% in 2007-2018 (a 55% decrease), and
the downticks frequency for glamour stocks decreased from 18% annually in 2000-2006 to 10%
in 2007-2018. Note that these substantial decreases in upticks and downticks in the recent period
are not the result of a general decrease in stock volatility in recent years. For example, the
percentage downticks of 10% or more of value stocks, in fact, increased during 2007-2018 (not
Figure 8: Decrease in Price Downticks for Glamour Stocks and Upticks for Value Stocks,
1970-2018
Thus, all our measures indicate a substantial slowdown of the mean reversion of both value
and glamour stock in the past 12 years, accounting for much of the decline of the profitability of
value investing. But this, of course, is a statistical explanation, raising the question of the economic
18
will enable us to address the essential question whether the recent demise of value investing is a
It’s not a coincidence that the failure of value investing started in 2007—the first year of
the recent financial crisis. The 2007-2009 crisis and the subsequent deep recession had a
devastating and prolonged effect on the performance of several industries, particularly in the
financial sector, as well as firms relying on consumers’ demand, driving those industries and firms
to the ranks of value (low valuation) firms, and keeping them there. The two major adverse effects
of the financial crisis were the sudden contraction of bank lending and the sharp fall in consumer
demand. Hogan (2019, pp. 1-2) writes on the former: “One mystery of the slow recovery [post
recession] is why lending failed to respond to expansionary monetary policy. Bank lending
declined dramatically during the crisis, and despite the period of very low interest rates since,
lending has failed to recover…Instead, banks appear to have permanently decreased lending
relative to their other activities… Thus, bank loans have been persistently low since the financial
crisis despite high demand for loans during this period.” (Emphasis ours).
The prolonged decline of bank lending had a direct, adverse effect on the performance of
banks, most relying on lending as a major source of profits and growth, and an indirect effect on
low-valuation firms which rely on bank lending to finance investment in innovation and growth
(R&D, IT, acquisitions), being unable to issue stock due to their low valuation. Those “value
firms” were cutoff of much needed equity and bank financing sources to improve operations and
escape the low-valuation trap. Many of those firms were also hit by the second major effect of the
19
research (2018) concluded: “A substantial body of evidence now suggests that such [credit] factors
are important for the behavior of households, firms, and financial intermediaries… More
specifically, the empirical portion of this paper has shown that the financial panic of 2007-2008,
including the runs on wholesale funding and the retreat from securitized credit, was highly
disruptive to the real economy and was probably the main reason that the recession was so
unusually deep… In particular, it seems plausible that the weakening of household balance
sheets… was a significant headwind to recovery.” (pp. 58-59, emphasis ours). Similarly, Mian et
al. (2013) reported that the collapse in house prices during the Great Recession caused a sharp
drop in consumer demand by households. Relatedly, Authers and Leatherby (2019) showed that
household debt hasn’t recovered even after a full decade, leading to painful consequences, such as
the demand decrease and income-inequality increase; the only increase occurred in the student
debt.
The combination of the prolonged contraction of bank lending and the falling of consumer
demand, post-recession, drove industries and individual firms to the ranks of “value” (low
valuation) companies and largely kept them there for the past 10-12 years. It’s not surprising,
therefore, that during the past decade the five leading industries of value firms were: banking,
retail, insurance, wholesale, and utilities, accounting for roughly 50-60% of large value companies.
The causes of the significant slowdown in mean reversion during the 12-year period, 2007-
2018 and the consequent failure of value investing should now be clear. Value firms, since the
relative to previous periods (and relative to glamour companies), as is made clear by Figure 9,
portraying for both value and glamour firms their median return-on-equity (ROE), and the return
20
profitability trends of value and glamour firms couldn’t have been more different. Whereas,
glamour firms experienced in 2007-2018 their highest profitability since 1970, value firms
sustained in 2007-2018 their worst profitability. Other profitability indictors were consistent: for
example, the percentage of firms reporting annual losses was highest for value firms during 2007-
2018 (and lowest for glamour companies, not presented in graphs). Thus, by practically any
measure of operating performance the profitability of value firms deteriorated sharply since the
Figure 9: Increase (Decrease) in ROE and RNOA for Glamour (Value) Stocks, 1970-2018
This poor profitability precluded most value firms from improving their performance by
investing in innovation and growth (R&D, IT, brands, acquisitions). Such crucial investments
require massive funding (the average large glamour company invests now close to $1 billion a year
in R&D) which most value firms couldn’t afford. R&D investments, for example, require ample
21
by because of R&D’s uncertain outcomes, severe asymmetric information problems, and lack of
collateral value (Hall 2002). Figure 10 shows that the internal funds of value firms (earnings minus
dividends) were, on average, negative during 2007-2018. A historical low. Without internal funds,
and with very limited access to the stock and debt markets, most value firms were unable in recent
years to pick themselves up and rise in performance and value.14 Investment in this group of poor
The economic experience, post-crisis, was very different for glamour companies. The three
pharmaceuticals (including biotech), and electronics. The business models of firms in these
industries are largely based on scalable intangible assets (compare the revenue generation potential
of patents or software with that of a rental property or a retail store), strongly protected by patents
and brands. First-mover advantages, network externalities, and platforms and ecosystems built
around entrenched customer relationships further enhanced the performance of many glamour
14
During 2007-2018, less than 1% of value firms issued stock annually.
22
markets, due to their attractive business models, have no problems raising funds to maintain a high
Glamour companies are highly rewarded by investors, as made clear by Figure 11, showing
the dramatic increase of investors’ valuation of intangible assets, and the large valuation gap
between intangible and tangible valuations.15 Many of the large glamour companies also enjoy
strong entry barriers to their proprietary business models, keeping them at the top for long periods
of time, as evidenced by the increasing rank correlation (Figure 6) and their longer stay in the high
Figure 11: Investors’ Valuation of Tangible and Unreported Intangible Assets, 1970-2018
15
We estimated investors’ valuation of the intangibles which aren’t shown as assets on the balance sheet: internally-
generated R&D, IT, brands, business processes, etc., and term them unrecognized intangibles, as follows:
UnrecogIntan is the amount of the estimated unamortized stock of R&D and SG&A (see Section 4). We ran the
following regression by year to estimate investors’ valuation of property, plant, and equipment (PPE) and
UnrecogIntan:
Cashi,t AccountsReceivablei,t Inventoryi,t
SharePricei,t = β1,i + β2,i × + β2,i × + β3,i ×
CommonSharesi,t CommonSharesi,t CommonSharesi,t
PPEi,t RecogIntangi,t UnrecogIntani,t Liabilitiesi,t
+ β4,i × + β5,i × + β6,i × + β7,i ×
CommonSharesi,t CommonSharesi,t CommonSharesi,t CommonSharesi,t
+ β8,i ×NYSE_Indexi,t + εi,t
The coefficients β4 and β6 are investors’ estimated valuation of tangible and intangible assets, and shown in Figure
11.
23
as “creative destruction”—which drove the past gains from shorting these stocks, gave way to
increased stability of glamour companies. Just consider the leaders of the software, pharma,
telecom, electronics, Internet service providers, and media industries; these companies are at the
top of their industries for decades. Creative destruction and industry disruption seems to have little
effect on Microsoft, Pfizer, Apple, Amazon, and the like. Shorting these companies was obviously
a futile exercise. Thus, the increased stability at both the top and bottom of stock valuation ranks—
due to different economic and technological reasons—robbed value investing of its previous gains.
Importantly, the serious financial constraints on value firms, (bottom 30%), and to a
somewhat lesser extent on the middle 40% of firms, prevented these companies from spending
large amounts of money on R&D, brand development, and technology, required to unseat the top,
glamour firms from their coveted position. Thus, the considerable reshuffling of glamour
companies pre-financial crisis, driving the gains from shorting these firms, dwindled significantly
in recent years.
This then explains the failure of value investing since 2007. A striking bifurcation
developed between value and glamour firms: Most value companies now operate in regulated
industries (banks, insurance, and utilities), or in low valuation, tangible asset-rich sectors (retailers,
transportation). Escaping this low-valuation group requires massive investments in intangibles and
acquisitions, and often a radical restructuring of business models, which most value firms can’t
afford. Findings diamond in the rough of this group is increasingly challenging. Glamour firms, in
contrast, operate intangibles-based business models which enable longevity and high profitability.
24
For investors intent on finding diamonds in the rough of value, out-of-favor companies, it
would be instructive to identify the attributes of value companies which rose in valuation in recent
years and escaped the value trap. We address this question by distinguishing between firms in the
large “value” category (trapped), from 2008 to 2017, that remained in that category and those that
moved up to the medium (40%) and high (30%) market-to-book ratio categories, while retaining
large size. We performed this separation by a statistical Logit regression which focuses on a large
number of company and performance attributes, and highlights the attributes which significantly
(in a statistical sense) distinguished between the trapped (in “value”) and escaped companies.16
Estimates from this Logit regression are presented in Table 1. Variables with significance level
(right column) lower than 0.05 (5%) are generally regarded as statistically significant.
Table 1: Logit Regression to Distinguish Characteristics of Large Value Companies which
Remained in the Same Category From those that Rose to Higher Market-to-Book Ratio
Categories, 2008-2017
Variable Coefficient Standard error Significance
Intangibles to Assets 1.664 0.45 <.01
Capex (net of depreciation) to Assets 4.686 1.64 <.01
Percent Contribution to Sales from
Acquisitions 0.082 0.74 0.91
Last Three Year's Sales Growth 1.694 0.37 <.01
Industry Change -0.044 1.31 0.97
Debt to Assets 1.199 0.34 <.01
Free cash Flows to Assets 2.110 1.02 0.04
Share Repurchase to Assets 6.182 2.12 <.01
Dividend Payments to Assets 7.352 3.32 0.03
Cash to Assets -0.067 0.62 0.91
Age Since Listing -0.001 0.00 0.86
Return on Equity 0.004 0.02 0.83
Loss -0.485 0.14 <.01
Log of Assets -0.193 0.05 <.01
Year and Industry fixed effects Yes
Number of observations 2,338
Number of observations with value = 1 730
Likelihood Ratio 423 significant at <.0001
16
All continuous variables were winsorized at 1% and 99% by year to remove the effect of outliers.
25
• Intangible investment (R&D, IT, brands, etc.) relative to total assets. Escapees had a
substantially higher intangible investment rate than firms which stayed behind.
• Net capital investments (Capex) relative to total assets. Escapees also had a substantially
• Debt raised to total assets. Escapees raised substantially more debt than others, apparently
• Size (total assets). The negative coefficient of (log) assets indicates, surprisingly, that larger
• Recapitalization. Firms that could reduce their equity base by share repurchases had a higher
Not less interesting are some of the variables which failed to distinguish those that escaped
from companies which stayed behind. Corporate acquisitions, presumably made in order to escape
the low market-to-book class, appear largely ineffective, either due to overpayment for target,
and/or strategic misfit. An industry change also appears to be ineffective. Overall, internal
investments in traditional intangibles, like R&D, brands, information technology, and tangible
investments, as well as the lesser visible investments in organization capital, or management, were
the main drivers of growth in market value. Recent economic research clearly shows that
26
Unique business processes (recommendation algorithms, AI), employee incentive devices, and
managerial control and monitoring systems are major drivers of breaking out of the low-
capitalization crowd. A working business model (sales growth) was a key to success, and being
able to raise debt to finance investments was helpful. So, value firms aren’t entirely doomed to
stay in the trap; innovation and reinvestments are key to rising in value.
This is the question at the core of the value enigma: Will value investing soon return to its
days of glory? Every month, like September 2019, where value beats glamour, raises the hopes of
value investors. Will it last? Note that we move here from the solid grounds of factual analysis
which guided our discussion so far, to the shaky realm of speculation. But even here some facts
are useful. Some investors pin their hopes of a value rebound on believing that value stocks are
now much cheaper than in past periods, and therefore ripe for growth. Figure 12 (our final exhibit)
begs to differ: it displays the annual differences between the median market-to-book ratios of large
value and glamour stocks. At least by this yardstick, it doesn’t appear that value stocks are now
much cheaper than glamour (growth) stocks. In fact, the current differences between the medians
of the market-to-book ratios of value and glamour stocks in Figure 12 aren’t significantly larger
than those that prevailed in the late 1990s and early 2000s.
17
See J. Van Reenen, J. 2018. Increasing differences between firms: Market power and the macro-economy.
Available at
https://www.kansascityfed.org/~/media/files/publicat/sympos/2018/papersandhandouts/jh%20john%20van%20reene
n%20version%2020.pdf?la=en
27
What else could resurrect the value strategy? An unusual improvement in the performance
and valuation of financial institutions (banks and insurance companies) will do the trick, but how
likely is this to happen soon?18 The alternative is a significant improvement in the performance of
non-financial value firms such as retail, oil, wholesale, and utilities. To rise significantly in value,
trapped value companies require large investments, which, as we have shown above, most value
firms can’t afford. Finally, a collapse of glamour (high valuation) companies will resurrect the
long-short value investing, as it did in the early 2000s (see Figure 1). What’s the likelihood of this
to happen? Restrictive laws and regulations adversely affecting internet and pharmaceutical
many glamour companies. Overall, though, the above scenarios don’t seem to us highly likely in
18
Warren Buffett apparently believes in a resurgence of bank stocks. The Wall Street Journal (August 14, 2019)
reported that Berkshire Hathaway holds nearly $100 billion in financial-services stocks.
28
Authers, J., and L. Leatherly, 2019, The decade of deleveraging didn’t quite turn out that way,
Bloomberg, April 3.
Ball, R., J. Gerakos, J. T. Linnainmaa, and V. Nikolaev, 2019. Earnings, retained earnings, and
book-to-market in the cross section of expected returns. Journal of Financial Economics.
Bernanke, B., 2018, The real effects of disrupted credit, Brookings Papers on Economic Activity,
September, 1-60.
Corrado, C., and C. Hulten, 2010, How do you measure a “technological revolution”? American
Economic Review, 100; 99-104.
Enache, L., and A. Srivastava, 2018, Should intangible investments be reported separately or
comingled with operating expenses? Management Science, 64, 3446-3468.
Fama, E., and K. French, 1992, The cross-section of expected stock returns, The Journal of
Finance, 46, 427-466.
Fama, E., and K. French, 1993, Common risk factors in the returns on stocks and bonds, Journal
of Financial Economics 33(1): 3-56.
Hall, B. H. 2002, The Financing of Research and Development. Oxford Review of Economic
Policy 18(1): 35-51.
Li, W, and B. H. Hall. 2018. Depreciation of Business R&D Capital. NBER working paper No.
22473
Hogan, T., 2019, What caused the post-crisis decline in bank lending? Rice University, Baker
Institute for Public Policy, 1-7.
Kok, U.W, J. Ribando and R. Sloan. 2018. Facts about Formulaic Value Investing. Financial
Analysts Journal, 73(2): 81-99
Lakonishok, J., A. Schleifer, and R. Vishny, 1994, Contrarian investment, extrapolation, and risk,
Journal of Finance, 49, 1541-1578.
Mian, A, K. Rao, and A. Sufi, 2013, Household balance sheet, consumption, and the economic
slump, The Quarterly Journal of Economics, pp. 1687-1726.
Mauboussin, M., 2012, The Success Equation, Harvard Business Review Press, Boston, MA.
Mohanram, P., 2005, Separating winners from losers among low book-to-market stocks using
financial statement analysis. Review of Accounting Studies 10: 133–170
Piotroski, J., 2000, Value investing: The use of historical financial statement information to
separate winners from losers. Journal of Accounting Research 38: 1–41.
29