Content uploaded by Laurence B. Siegel
Author content
All content in this area was uploaded by Laurence B. Siegel on Nov 04, 2017
Content may be subject to copyright.
Draft
WINTER 2000 THE JOURNAL OF INVESTING 1
Value investing, loosely defined as
the practice of buying stocks that
appear underpriced and holding
them until the market realizes their
true worth, has been a poor performer lately.
From December 31, 1997 (a recent value high
point), to February 29, 2000, the Russell 1000
Value index, which is representative of the
value style, underperformed the Russell 1000
Growth index by 39.9%.1
While value and growth have always had
wide swings relative to one another, this is one
of the widest recorded, and one of the speed-
iest. “Deep” value stocks, which represent
the statistically most attractive company valu-
ations, have had even worse performance.
Given this miserable track record, many
investors are asking 1) whether “value” stocks
still represent good value for one’s investment
dollar; 2) if so, under what conditions they can
be expected to generate good performance;
and 3) to what extent “classic” methods of
value investing need to be adapted to the cur-
rent, puzzling market environment.
One current argument in favor of value is
that growth stock indexes currently sell at much
higher multiples (for example, price/earnings
ratios) than value stock indexes; if these multi-
ples revert toward a more normal relationship,
it is argued, market leadership will shift in favor
of value. We summarize this analysis and add to
it by demonstrating that comparative earnings
growth rates, as well as multiples, help to fore-
cast the value-growth cycle.
The current case for value investing can-
not, however, be established by relying on
mean reversion—or value would have started
outperforming growth long ago. We must
demonstrate that selected value companies
represent good value in the sense of having a
fundamental worth substantially higher than
the current market price. We perform this
analysis on individual stocks rather than at the
style index level. This approach enables the
sensibility of value investing at this time in his-
tory to come alive for investors contemplating
the use of an active value manager with a lim-
ited number of positions.
We seek to demonstrate that:
• Discounted cash flow (DCF) is still a
legitimate framework, and arguably
the best one, for deciding what a
company and its stock are worth.
• The DCF assumptions required for
today’s top-performing stocks to be
good values are heroic, while
• By the same criterion, selected
“value” stocks really are a good value.
SOME OBSERVATIONS ON
CURRENT MARKET CONDITIONS
Is Value Junk?
The disfavor into which value stocks
have fallen reflects a change in perception.
We can remember, not so long ago in the
The Future of Value Investing
LAURENCE B. SIEGEL AND JOHN G. ALEXANDER
LAURENCE B.
SIEGEL is director of
investment policy
research in the
investment division
of The Ford Founda-
tion in New York.
JOHN G.
ALEXANDER is an
equity portfolio man-
ager with INVESCO
Capital Management
in Atlanta.
Draft
1980s, when value stocks were widely regarded as good
companies that were unloved by the market and that
were merely waiting for a takeover or other catalyst that
would cause the market to realize their true, much higher
value. Growth stocks in this period were still tainted by
the experience of the “Nifty-Fifty” in the 1970s—once-
bitten, twice-shy, investors were leery of buying into
growth stories that could quickly dematerialize due to
economic downturn, foreign competition, or inflation.
Today, the perception is the opposite. Many value
issues are considered to be “junk stocks,” representing sun-
set industries, low margins, bad balance sheets, and poor
management. Growth, at this time, is associated not only
with rapidly growing earnings but also with low risk, high
earnings “quality” or acyclicality, progressive manage-
ment, and everything else good and beautiful.
This set of perceptions creates some opportunity for
value investors, since it has knocked down the prices of
good companies along with the bad. There may really be
some junk in the value sector; many deep-value managers
have been torpedoed by earnings disappointments and
other bad news, driving their returns well below that of
the benchmark. In this market, naive value screening
doesn’t seem to work, and portfolio choices must be sub-
jected to careful fundamental analysis.
Funds Flow-Related Pressure
on Value Stocks
The current extreme downtrend in value stocks has
a purely technical component. Funds flows into the stock
market as a whole have slowed from
the 1997-1998 pace, so that the rise
in technology and other growth
stocks cannot be powered by new
money alone. Investors have to sell
something to get the money to make
purchases.
Since many investors sell their
poorest performers, they have tended
to dump value stocks (especially
“deep” value stocks) and mutual
funds holding these stocks, acceler-
ating these stocks’ decline and send-
ing them into something like a free
fall. This trend has been exacerbated
by the tax-motivated desire to real-
ize losses to offset big gains in growth
issues.
Whenever a technical factor is
powerful enough to cause a substantial price move in a cat-
egory of securities, it creates an opportunity for bettors
against the trend—as long as they have the patience and
the capital to withstand further procyclical movement.
Value Managers versus
Value Benchmarks
Another reason that value managers have been
underperforming their own style benchmarks is that some
of the benchmarks are extraordinarily broad, including
something like three-quarters of the stocks in the market.
(As of April 10, 2000, for example, the S&P value index
comprised 388 stocks and the S&P growth index only
112, because the indexes are constructed to have the
same total market capitalization, and growth stocks have
much larger capitalizations on average.)
A “value” benchmark with such extreme breadth
must necessarily include many core, non-value issues
with some growth characteristics. These core stocks have
been performing well, causing “deep value” managers
who stick to their style to appear to underperform. (The
core issues also tend to have larger market capitalizations
because of their better performance, so they dominate the
benchmarks.) Few managers, or their clients, have the
stomach to consistently underperform their benchmarks
in a down market for value, by staying true to the style,
in the hope of outperforming when the cycle turns.
It is an open question whether a value manager
should stay faithful to the tenets of the style or broaden
2THE FUTURE OF VALUE INVESTING WINTER 2000
EXHIBIT 1
Ratio of Value Index to Growth Index—December 1975–February 2000
0
0.5
1
1.5
2
2.5
Dec-75
Dec-76
Dec-77
Dec-78
Dec-79
Dec-80
Dec-81
Dec-82
Dec-83
Dec-84
Dec-85
Dec-86
Dec-87
Dec-88
Dec-89
Dec-90
Dec-91
Dec-92
Dec-93
Dec-94
Dec-95
Dec-96
Dec-97
Dec-98
Dec-99
Ratio (12/31/1974 = 1.00)
Value index: S&P 500 Value, 1975-1978; Russell 1000 Value, 1979-2000
Growth index: S&P 500 Growth, 1975-1978; Russell 1000 Growth, 1979-2000
Draft
focus to track the benchmark better, but clients should be
aware of the situation so they can decide which type of
manager they want to hire.
RESTATING THE TRADITIONAL
CASE FOR VALUE INVESTING
The traditional case for value investing rests on two
points: 1) value has beaten growth in the very long run,
and 2) value stocks are timely now that many growth
stocks are overpriced. We revisit these basic ideas, and
introduce a new study showing that earnings growth
rates as well as valuation indicators can be used to predict
the relative returns of value and growth stocks.
The Long-Run Premium of
Value Over Growth
The traditional evidence cited for the value effect is
the outperformance of value indexes since the starting
point of widely available style index data in 1975, shown
in Exhibit 1. This argument has become steadily less
compelling as the growth run of the last few years has
caused this cumulative outperformance to shrink just
about to the vanishing point.
Fama and French, have backdated their work on the
three-factor model (beta, size, and book-to-price), con-
structing style indexes back to 1927, and find that value
very decisively outperformed
growth when measured over
that long period, which includes
numerous episodes of inflation
and deflation, boom and bust,
and technological change
and stability.
2
Exhibit 2 shows that a
dollar invested in the Fama-
French value index on June
30, 1927, grew to $5,881 by
the end of October 1999,
while the growth index grew
to only $1,345 over the same
period—a compound annual
return premium of 2.3% for
value over growth.
Timing would still have
been important—value per-
formed miserably in the Great
Depression and has disap-
pointed again recently, so that
all of the net gain takes place between 1949 and 1980—
but, this study bolsters the contention, now widely
doubted by investors, that value provides superior (or at
least competitive) returns in the very long run.3
Exhibit 3 shows the ratio of the two Fama-French
indexes, making it easy to see the long waves of perfor-
mance as we did for a more recent time period in Exhibit
1. Note that the current growth run, while dramatic, is
hardly unprecedented; growth also beat value by a huge
margin in the 1927-1939 period, and by smaller but still
significant margins at other times prior to the conventional
1975 start date of value/growth studies.
Fama and French’s finding is particularly notable
because it constitutes an almost independent experiment
confirming that a value strategy worked in periods, such
as the 1950s and 1960s, in which it had not been previ-
ously tested. This is very encouraging for today’s value
investors.
While studies of value that depend on more recent
data (1975 to the present) emphasize the need for declin-
ing interest rates or economic recovery for value to per-
form well, Fama and French show that value stocks added
much of their historically superior performance during the
best years of the postwar expansion. Much of that period
was characterized by rapid economic growth, the emer-
gence of new industries and decline of old ones, low
unemployment, and low inflation. Some of these condi-
WINTER 2000 THE JOURNAL OF INVESTING 3
EXHIBIT 2
Fama-French Growth and Value Indexes 1927–1999
0.1
1
10
100
1000
10000
Jun-27
Jun-31
Jun-35
Jun-39
Jun-43
Jun-47
Jun-51
Jun-55
Jun-59
Jun-63
Jun-67
Jun-71
Jun-75
Jun-79
Jun-83
Jun-87
Jun-91
Jun-95
Jun-99
Growth of $1.00 Invested 6/30/1927
$5,881
$1,345
Draft
tions are echoed in today’s economy, although monetary
policy was easier over much of that period than it is
today.
Risk of Value versus
Growth Stocks
Value stocks (with lower earnings growth rates,
weaker balance sheets, and less liquidity) would appear on
their face to be riskier than growth stocks, but this does
not show up in the recent data (1975 and forward). From
January 1975 through September 1999 the Russell 1000
Value index has an annualized standard deviation of 14.1%
compared to 16.9% for the Russell 1000 Growth index.
Yet according to the Fama-French data over 1927-
1999, value was much riskier than growth; the value
index had a standard deviation of 25.9% compared to only
18.9% for the growth index. All this extra risk took place
in the late 1920s and, more pointedly, in the Great Depres-
sion of the 1930s; over this time span, the value index fell
by 90.2% from high to low, while the growth index fell
by only 81.5%.
The long-term data, then, suggest that the higher
return on the value index is the delivery of a risk premium.
Investors who look only at the data starting in 1975
would be hard pressed to find the risk, but it is obvious
in the longer-term data. The risk is that in a major depres-
sion, the lower quality of value companies (reflected in
their higher leverage and other characteristics) makes
their stock prices much more
vulnerable.
Mean Reversion and
the Timeliness of
Value Investing
If the growth-value cycle
is mean-reverting, one can
make useful forecasts whether
or not one expects a long-run
premium for value in the
future. Mean reversion of the
growth-value cycle can be
measured in a number of ways,
but is probably best under-
stood in terms of the relative
price/earnings, or P/E, ratios,
of value and growth indexes.
The bars in Exhibit 4
show the P/E range of the middle three-fourths of the
stocks in the BARRA high-capitalization universe sorted
by P/E. That is, one-eighth of the stocks had a P/E
higher than that represented by the top of the bar, and
one-eighth had a P/E lower than that represented by the
bottom of the bar.
For example, the exhibit shows that, as of Decem-
ber 1979, three-quarters of the stocks in the BARRA
high-cap universe had P/E ratios between 5.3 and 11.7,
with the remainder of the stocks outside that range. The
ratio of the two P/E ratios, 11.7/5.3, or 2.2, is shown
underneath the bar for December 1979. This “ratio of
ratios” is a measure, over time, of the valuation disparity
between growth and value styles.4
Remarkably, although the general level of P/E ratios
rose over 1979-1997 as the bull market unfolded, the
ratio-of-ratios was stable and powerfully mean-reverting.
5
That is, rises in the ratio-of-ratios tended to be followed
by declines, and vice versa, so that fluctuations tend to can-
cel over time and cause the ratio-of-ratios to vary around
its long-term mean of about 2.5. As recently as Decem-
ber 1997, the ratio-of-ratios was at an 18-year low of 2.2.
The explosion in the ratio-of-ratios after December
1997, to an unprecedented 11.3 as of March 2000, sug-
gests that the P/E premium of growth over value is too
high and ought to decline. We would caution that the
same prediction could have been made in March 1999
when the ratio-of-ratios reached a then-unprecedented
4.2, and it would have been early, or wrong.
4THE FUTURE OF VALUE INVESTING WINTER 2000
EXHIBIT 3
Ratio of Fama-French Value-to-Growth Index—6/30/1927 = 1.00
0.1
1
10
Jun-27
Jun-30
Jun-33
Jun-36
Jun-39
Jun-42
Jun-45
Jun-48
Jun-51
Jun-54
Jun-57
Jun-60
Jun-63
Jun-66
Jun-69
Jun-72
Jun-75
Jun-78
Jun-81
Jun-84
Jun-87
Jun-90
Jun-93
Jun-96
Jun-99
Draft
A Two-Factor Style Timing Model. In an enhance-
ment to our simple P/E analysis, Asness et al. [2000]
study two factors: 1) the ratio of the P/E of growth stocks
to the P/E of value stocks, or “value spread,” and 2) the
difference in expected earnings growth rates of these
indexes, or “growth spread.”
Exhibit 5 shows that the value and growth spreads
have historically tended to move together, because growth
companies are worth more than other companies when
their earnings growth rates, again relative to other com-
panies, are especially fast. Starting about January 1998,
however, the value and growth spreads began to move in
opposite directions, with the result that the value spread
was at an all-time high in April 2000, while the growth
spread was near its all-time low. That is, investors in growth
stocks today are being asked to pay record valuations for
earnings growth that is actually below average when com-
pared to the earnings growth rates of value stocks.
Using the behavior of the growth and value spreads,
Asness et al. constructed a model of the one-year expected
relative return of value versus growth. The authors note that
the greater the value spread and the smaller the growth
spread, the higher the forecast return for value versus
growth going forward. They observe that:
At the time of this writing (end of October 1999),
the model forecasts a 52% return spread (3.6 stan-
dard deviations above the historical average)
between value and growth over the coming year.
Clearly this is near historic highs for value versus
growth [2000, p. 58].
That forecast is very nearly as great a one-year return
differential between value and growth as has been expe-
rienced over the entire length of the Fama-French study.
6
Fundamental Reasons for Mean Reversion. Statisti-
cal evidence of the relative mispricing of growth versus
value is, however, not enough for one to invest with
confidence. One must identify fundamental logical reasons
why mean reversion will occur. One of the most persua-
sive reasons is described by Grantham and Gray:7
All advantages erode with time. While the U.S. has
been able to attract overseas capital due to higher
WINTER 2000 THE JOURNAL OF INVESTING 5
EXHIBIT 4
Spread Between High and Low P/E Quartile Midpoints*
10
20
30
40
50
60
70
80
90
100
Dec-79
Dec-80
Dec-81
Dec-82
Dec-83
Dec-84
Dec-85
Dec-86
Dec-87
Dec-88
Dec-89
Dec-90
Dec-91
Dec-92
Dec-93
Dec-94
Dec-95
Dec-96
Dec-97
Dec-98
Mar-99
Jun-99
Sep-99
Dec-99
Mar-00
Price/earnings ratio
* "Quartile midpoint" = median P/E of
stocks in the top (bottom) quartile of
Barra high-capitalization universe. That
is, at each time, three-quarters of stocks
in universe had P/Es in the range defined
by the bar.
Annual
data
Quarterly
data
High/low
PE
2.2 2.8 2.5 3.1 2. 5 2.2 2.4 2.4 2.4 2.3 2.5 2.6 2.6 2.3 2.5 2.4 2.6 2.3 2.2 3.5 4.2 4. 2 5. 4 8.0 11.3
Source: Martingale Asset Management.
Draft
profitability and better economic management,
these abnormally high investments will in turn
bid exceptional returns down. That’s how the
process is meant to work.
In a world driven by a plentiful supply of brains
and ambition, and most of it dedicated to copying
good ideas and avoiding bad ones, reversion to the
mean is universal. It seems like a bad bet that after
centuries of data and experience this would change.
But occasionally things do take their time getting
back to average [1999, pp. 50-51].
Thus, the high margins being earned in the largest
and best-run U.S. companies, and the concomitant high
multiples for these companies, look permanent (and may
indeed “take their time getting back to average”) but are
temporary. The margins are a tempting, although not easy,
target for competition by foreign companies, smaller com-
panies, and entrepreneurs. This competition will tend to
drive down the margins, earnings growth rates, and rela-
tive stock prices of today’s market leaders (which, one
should remember, are different names from the market
leaders of only a decade ago) and help the returns of value-
oriented companies.
8
THE FUNDAMENTAL CASE FOR VALUE
The Logic of Discounted Cash Flow Analysis
The traditional (statistical) case for value investing is a
valuable backdrop for a discussion of that investment style, but
we find fundamental analysis more compelling. We use a dis-
counted cash flow (DCF) model to demonstrate that, under
quite reasonable and arguably conservative assumptions, a
number of prominent value names are underpriced in an abso-
lute sense, even in today’s expensive market. That is, these
stocks are worth more than the market price and represent
good “value” in the plain-English sense of the word.
An article written for investment professionals should
not have to explain why using a DCF model is valid—it
should go without saying. With many of today’s market
leaders priced with no known relation to their DCF
value, and with many sponsors and portfolio managers
consequently being richly rewarded for disregarding DCF,
however, a few remarks are in order.
The only thing investors really want is cash, since
consumer goods, debts, and taxes must be paid for in cash.
Investment in financial assets such as stocks, bonds, and real
estate is just a stratagem to increase the amount of cash that
will eventually be available to the investor. The DCF
approach to asset valuation originated as a means of fig-
6THE FUTURE OF VALUE INVESTING WINTER 2000
EXHIBIT 5
Value and Growth Spreads—January 1982-April 2000
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Jan-82
Jan-83
Jan-84
Jan-85
Jan-86
Jan-87
Jan-88
Jan-89
Jan-90
Jan-91
Jan-92
Jan-93
Jan-94
Jan-95
Jan-96
Jan-97
Jan-98
Jan-99
Jan-00
Value spread
(Ratio of P/Es of growth to value stocks)
0%
2%
4%
6%
8%
10%
12%
14%
16%
Growth spread
(Difference in expected earnings growth rates
between growth and value stocks)
Growth spread
Value spread
Source: ACR Capital Management.
Draft
uring out the present value of the actual cash—dividends,
in the case of a stock—that is expected to be generated
by an investment. One then compares this present value
to the market price, and if value is appreciably greater than
price, the investment is a “buy.”
As investors realized that dividends are taxed twice,
and that companies are sometimes better than investors at
deploying spare cash, dividends became less highly prized,
and the dividend discount version of the DCF model
became less relevant.8Under current conditions it tends
to yield low estimates of a stock’s value.
As analysts searched for an alternative valuation
approach that focuses on cash, many settled on free cash
flow—the cash that a company could afford to pay out as
dividends, after providing for capital expenditures both to
maintain existing assets and to create new assets or busi-
ness lines for future growth.
Not all analysts, of course, rely on DCF models as
their primary tool. Comparison of market value to a mea-
sure of cost (say, book value or replacement cost) is also
widely used. Such a method, however, presumes that a
DCF analysis is being conducted by other people, and seeks
to piggyback on it. This is because the corporate assets rep-
resenting the book value or replacement cost measure are
themselves worth only the discounted present value of the
cash they are expected to generate.
Methodology
The model used to identify strong value “buys”—
as well as “sells” chosen from among companies prized for
their growth rates—is the free cash flow to the firm
(FCFF) version of the DCF. We first discuss the inputs to
the model, and then we present the results.
Definition of Free Cash Flow to the Firm. An exten-
sion of the free cash flow concept introduced earlier, free
cash flow to the firm (FCFFis the amount of cash a com-
pany can afford to pay out to all claimants (stock plus
debtholders) while meeting capital expenditure
requirements. We define it for our purpose as:
FCFF= EBIT(1 – Tax Rate) + Depreciation
and Amortization – Capital Expen-
ditures - ∆Working Capital
where EBIT = earnings before interest and taxes.
Note that FCF
F
is calculated before the deduc-
tion of interest expense, and must therefore be dis-
counted at the company’s weighted-average cost of
capital (WACC) to arrive at a valuation for the whole
enterprise (stock plus debt). The market value of the debt
is then subtracted to arrive at the estimate of stock value.
Growth Rates. We begin our analysis by estimating
a current or baseline FCFF, which we take to be the
average of 1) the most recent 12 months’ FCFF, and 2) the
trailing 12 months’ FCFFas of one quarter earlier. We then
grow this amount according to a four-stage model.
Growth rates are based on IBES long-term earnings
growth rate estimates (three- to five-year), and “decay”
toward the economywide growth rate (or average for all
companies) over time. Details of the decay function are
shown in Exhibit 6.9
Note that we assume all companies have the same
growth rate in the twenty-first year and beyond—by
then, many growth stocks will have become value stocks,
and value stocks growing more slowly than the average
company will have accelerated to the average.10
Discount Rates. The forecast FCFFis then dis-
counted to arrive at a present value. We use the capital
asset pricing model (beta model) for the equity part of the
company’s capital structure, and a market cost of debt for
the debt part.11 Our CAPM assumes a 6% riskless rate and
a 5% equity risk premium. We then weight the equity and
debt costs by their market value weights in the company’s
total capital structure to arrive at the weighted-average cost
of capital (WACC). This company-specific WACC is the
discount rate for the first 20 years’ FCFF.
Beyond 20 years (that is, for the perpetual-growth
portion of the model), we assume that all companies have
the same discount rate (which amounts to assuming they
all have the same beta and capital structure). This discount
rate is assumed to be 11%, again consisting of a 6% risk-
less rate and a 5% equity risk premium.
The 5% equity risk premium is the required, not the
expected, return on the stock market index in excess of
the riskless rate. (Required returns equal expected returns
WINTER 2000 THE JOURNAL OF INVESTING 7
EXHIBIT 6
Weighted-Average Calculation of
Company Growth Rates for
Free Cash Flow Discount Model
Forecast time period
Weight of company-specific
IBES estimate of 3- to
5-year growth rate
Weight of economywide
growth rate (assumed to be
5% nominal)
Years 1-3 90% 10%
Years 4-10 50% 50%
Years 11-20 20% 80%
Years 21+ 0% 100%
Draft
only if the market is fairly priced.) Since stocks are expen-
sive, and the equity risk premium currently offered by
broad market indexes is currently less than 5%, our model
tends to regard stocks as overpriced, and produces a lot
more “sells” than “buys.” Under such conditions, the
undervaluation of the stocks we regard as strong “buys”
is even more compelling than it would be if we had
rigged the model to make the overall market fairly valued.
Results of the Study
Exhibit 7 shows DCF estimates of the fundamental
value of five selected value stocks and five selected growth
stocks, calculated according to this model. The funda-
mental value (in the column labeled “fair value of equity
value per share”) is compared to the
market price, and various model
inputs are also shown for each stock.
The value stocks in the exhibit
are selected for their favorable pricing
and other characteristics and should
not be considered representative of
the whole value category, but they
give an idea of how value investing
works—each stock is selling for a
fraction of its fundamental or fair
value, calculated using the conserva-
tive assumptions we have just out-
lined. These fundamental values may
be treated as price targets, and if real-
ized the investor would probably sell
the stock. (On average over the five
stocks, the price targets are just under
one and one-half times the current
market price.) If the earnings esti-
mates used in our model prove to be
realistic, each of these stocks repre-
sents excellent value.
The growth stocks in the
exhibit are selected on the basis of
unfavorable pricing and other char-
acteristics, and, as with value stocks,
these names should not be considered
representative of the whole style.12
They are just examples. The extent of
overpricing in these popular names is
remarkable, however. On average the
stocks are selling for more than twice
their fundamental value.
It is possible to back into the
growth rates that would be required for these growth
stocks to be worth their current market price. Exhibit 8
displays the results of this analysis, using the same “decay”
structure for growth rates as in Exhibit 7. That is, we
assume that all companies’ growth rates would fade toward
the economywide average of 5%, and we vary the start-
ing point, with intermediate growth rates also varying
according to Exhibit 6.
Note that the required growth rates are astonishingly
high for some companies, in the extreme case resulting in
an 86-fold explosion in free cash flow over the next two
decades (Qualcomm). While growth at rates such as those
shown in Exhibit 8 is not literally impossible—it has
occurred in a few companies historically—betting on a
8THE FUTURE OF VALUE INVESTING WINTER 2000
EXHIBIT 7A
Free Cash Flow Analysis of Selected Value and Growth Companies
Weighted
Average of Enterprise
Fair Value
Current Leverage Years Years Years Years Cost of (stock + debt)
Company FCF Beta in % 1-3 4-10 11-20 21+ Capital ($ billions)
Crown Cork & Seal 302$
F
ab
66 9.1 7.3 5.9 5.0 11 6,613$
Dillards 301$ 1.0
1.0
55 9.5 7.5 6.0 5.0 11 6,798$
Mylan 103$ 0.8 0 20.8 13.8 8.5 5.0 10 5,116$
Sherwin-Williams 429$ 0.8 34 11.3 8.5 6.4 5.0 9 5,200$
Whirlpool 271$ 1.0 49 9.5 7.5 6.0 5.0 10 6,469$
Ciena 124$ 1.5 0 27.5 17.5 10.0 5.0 14 7,855$
Intuit 178$ 1.3 2 18.5 12.5 8.0 5.0 13 6,502$
KLA-Tencor 69$ 1.6 5 23.0 15.0 9.0 5.0 14 3,237$
Medtronic 668$ 1.1 0 16.7 11.5 7.6 5.0 11 24,005$
Qualcomm
*Economywide WACC = 11.
293$ 1.4 21 36.5 22.5 12.0 5.0 12 36,674$
Growth Rates (%)
*
EXHIBIT 7B
Free Cash Flow Analysis of Selected Value and Growth Companies
Fair Enterprise 3/20/00 P/E Based
Value as Fair Value
of Equity
Fair Value
of Equity
Price Price Actual on Fair
Multiple of per as % of Current Value
Company Current FCFF($ billions) per Share Share Fair Value P/E of Stock
Crown Cork & Seal 21.9 3,215$ 26.38$ 16.31$ 62% 6.6 10.7
Dillards 22.6 3,244$ 30.92$ 16.13$ 52% 6.2 11.8
Mylan 49.6 5,116$ 39.60$ 28.25$ 71% 16.5 23.1
Sherwin-Williams 12.1 4,576$ 27.50$ 21.75$ 79% 9.7 12.3
Whirlpool 23.9 5,697$ 76.54$ 57.06$ 75% 8.1 10.8
Ciena 63.2 7,855$ 56.84$ 142.63$ 251% 234.9 93.6
Intuit 36.5 6,464$ 34.12$ 54.50$ 160% 90.1 56.4
KLA-Tencor 46.9 3,237$ 35.99$ 80.00$ 222% 51.0 19.8
Medtronic 35.9 23,987$ 20.47$ 52.69$ 257% 55.4 24.9
Qualcomm 125.0 36,013$ 55.72$ 136.25$ 245% 119.8 49.0
a
b
$ millions, stated as annual rate. FCF = free cash flow to the firm (stock plus debt).
Market value of debt/market value of debt plus equity.
F
Draft
whole portfolio of stocks priced for such growth seems
to us the height of folly.
The results in Exhibits 7 and 8 are powerful because
we have used the same model for every company (growth
or value). We have made no company-specific judgments,
and have used only publicly available data.
While the estimates of fundamental value are sub-
ject to wide confidence bounds (due to the inherent dif-
ficulty of making long-term forecasts), especially for
growth stocks with low current cash flows, the examples
illustrate a remarkably wide gulf between the valuations
of growth and value securities.
STYLE TIMING: WHEN CAN VALUE
BE EXPECTED TO OUTPERFORM?
While the traditional case for value investing notes
that value has historically been superior on average over
time and suggests that it might be in the future, no one seri-
ously suggests it is superior all the time. We have demon-
strated that value investing is currently timely in a statistical
sense, and fundamental analysis confirms this observation.
But to have any confidence in value investing, con-
sidering the battering it has just been through, one should
also seek to figure out under what economic circumstances
value performs well, so that one can adjust one’s alloca-
tions to value versus growth managers.
Classic value investing as taught by Benjamin Graham
[1973], among others, was concerned with finding great
companies at low prices. The kind of company identified as
a value stock by Graham was “large, prominent, and con-
servatively financed,” having, among other criteria, debt
less than 110% of current net assets; a very long history of
paying dividends (say, 20 or
more years); positive earn-
ings for 10 years running
(the company’s earnings,
thus, could be cyclical but
not negative); and current
assets at least twice current
liabilities. Finally, a com-
pany that passed these
screens had to have a low
P/E ratio, which presum-
ably was the result of “over-
shooting” by investors who
were pessimistic about a
company for the wrong
reasons.
Over much of investment history, such companies
could be found, and Warren Buffett was among the
investors who made great fortunes following this disci-
pline.Today, however, with broad market indexes selling
near all-time record high valuations, most high-quality
companies are priced as growth stocks. Going forward,
“deep” value stocks—and many other stocks in value
indexes—are likely to have some strong negatives, includ-
ing one or more of the characteristics following:
• High financial leverage.
• Cyclicality of earnings.
• High operating leverage.
• Inefficient management.
• Sunset industries.
Now, what economic environments would tend to
help stocks with these characteristics?
High Financial Leverage
Declining interest rates tend to help highly leveraged
stocks more than low-leverage stocks because the more
highly indebted companies can refinance their substantial
debt burdens at lower rates. Growth companies typically
do not have these burdens and do not get this direct ben-
efit of lower interest rates.
Some analysts have argued the contrary: that growth
stocks should benefit more from declining interest rates
because growth stocks have a longer duration.13 Yet in
most declining interest rate environments, except for the
brief and bizarre 1998 episode when Treasury yields
plummeted in a “flight to quality,” the leverage effect has
WINTER 2000 THE JOURNAL OF INVESTING 9
EXHIBIT 8
Analysis of Growth Rates Required to Justify Current Market Price of
Selected Growth Stocks
Year 20
FCF
F
as
Multiple
Years Years Years Years Current of Current
1-3 4-10 11-20 21+ FCF
F
*Year 4 Year 10 Year 20 FCF
F
Ciena 41.4 25.2 13.1 5.0 124$ 440.7$ 1,700.9$ 5,824.2$ 46.84
Intuit 25.6 16.4 9.6 5.0 178$ 410.7$ 1,023.0$ 2,551.7$ 14.32
KLA-Tencor
35.1 21.7 11.7 5.0 69$ 206.9$ 672.5$ 2,030.0$ 29.42
Medtronic 31.1 19.5 10.8 5.0 668$ 1,799.3$ 5,242.7$ 14,626.1$ 21.90
Qualcomm 50.1 30.1 15.0 5.0 293$ 1,290.8$ 6,250.6$ 25,349.6$ 86.43
*$ millions, stated as annual rate. FCF
F
= free cash flow to firm (stock plus debt).
Re
q
uired Growth Rates
(
%
)
Re
q
uired FCF
F
Draft
trumped the duration effect and more-leveraged compa-
nies have outperformed.
Cyclicality of Earnings
Earnings cyclicality is helped by recovery from an
economic recession. Corporate earnings are a leveraged
bet on GDP growth, and during recessions they fall by a
high multiple of the GDP decline. In recoveries, earnings
soar. This effect is much more pronounced for value than
for growth stocks (in part because of the financial and
operating leverage referred to above, but also because of
cyclical changes in demand for the company’s products).
This is one reason many value stocks are also “cyclicals.”
Recovery-related value rallies occurred in 1975-
1976, 1983-1984, after the crash of 1987, and in 1992-
1993. Later in the economic expansion, growth stocks
outperform. Note that recovery from an industry-specific
recession can be just as helpful to the value stocks in that
industry as recovery from a general recession.
High Operating Leverage
High leverage causes a company to be helped by
increased demand for its products and services, making the
company cyclical (as we just observed).
14
If fixed costs are
insensitive to inflation, however, high operating leverage may
cause a company to be helped by increased pricing power.
For most value companies, the inflation benefit
from operating leverage is minor, but in a few businesses
it is dominant. It costs roughly the same, for example, to
grow an acre of wheat or mine an ounce of gold whether
the price received is high or low. For such companies,
which are typically value companies, inflation in the
commodity they sell is a tremendous boost to earnings and
the stock price.
Inefficient Management
Inefficient companies will become takeover candi-
dates. The new owners often force out the inefficient
management and cut costs, to the delight of investors who
then push the stock price up. Merely being a promising
takeover candidate, however, does not cause the takeover
to happen. It helps if the stock market is richly valued and
capital costs are low, so that the acquirer can borrow or
offer a strong “currency,” in the form of its stock, to the
acquiree. Alternatively, cash-rich companies can make
takeover bids.
Because these are bull market and low-inflation con-
ditions, this effect works somewhat counter to the first three
that we discussed. The takeover activity of the 1980s, with
a strong bull market and rapidly declining inflation rates,
helped realize the “value” in many value stocks and is the
best example of this effect. Given today’s high stock prices
and cheap capital, one might be cautiously optimistic
about a takeover-related value rally looking forward.
Sunset Industries
Sunset industries also favor takeovers, but with
breakup rather than managerial efficiency in the mind of
the acquirer. A company that is in a sunset industry may
have declining earnings or even persistent losses, but there
are usually some assets that can be profitably sold by the
acquirer. A breakup play often improves the stock price for
severely beaten-down companies that have the most value.
The conditions that promote takeovers of inefficiently
managed companies are also favorable to takeovers of com-
panies in sunset industries—of which there are many today.
Other Economic Variables
Some of the stocks prominent in today’s value indexes
do not closely fit our profile. Coca-Cola, Compaq, and var-
ious pharmaceutical and financial stocks come to mind. We
do not have a theory as to why these growing, financially
healthy companies should be battered by the equity mar-
kets the way they have been. Perhaps they are simply vic-
tims of the overshooting we referred to earlier. We note,
however, there is something new and different in every
cycle, and it is possible to learn from historical patterns
without expecting them to be repeated exactly.
RETOOLING VALUE MANAGEMENT
FOR TODAY’S MARKETS
It is possible that the tide will simply turn in favor
of value as it is traditionally understood (as it may have
already). The classic approach to value investing—buying
stocks that are low-priced compared to earnings, cash
flows, book value, or dividends—may need to be retooled
a bit all the same.
Where’s the Catalyst?
Undervalued stocks do not become repriced to fair
value by magic. There typically has to be a “catalyst” on the
10 THE FUTURE OF VALUE INVESTING WINTER 2000
Draft
corporate as opposed to the investor side. (The investor-side
effect, that the stock rises because value-oriented investors buy
it, is the statistical arbitrage that we earlier said should not be
relied on.) Examples of corporate-side catalysts include:
• New management or a revamped corporate
strategy.
• Takeover or the threat of one.
• Changes in consumer demand.
• Changes in technology.
• Changes in the legal or regulatory environment
(say, a government program or new trade
policy).
Fundamental, company-focused research is the only
way to determine the presence or absence of such cata-
lysts in a value “story.” The value process must include
such research rather than relying on statistical indicators
of underpricing.
Industry Adjustment
Several researchers have noted that the value effect
is more powerful, and that it works over a greater per-
centage of historical time, if one compares underpriced
to overpriced companies in the same industries.15 One
possible reason is that value metrics, such as earnings and
book value, are not closely comparable across industries
because of differences in earnings growth rates and in the
way depreciation, amortization, and other accounting
variables are treated.
A portfolio that takes advantage of this insight could
conceivably have market weights in the various industrial
sectors, holding the companies within each sector that rep-
resent the best value. In practice, most value managers will
continue to overemphasize certain industries associated
with “value,” but not to the extent that would occur if one
were to compare the valuations of all companies without
regard to what industry they are in.
Sunset Companies and Industries
We have mentioned that some (perhaps many) stocks
appear to be, but are not, good values because the com-
panies are likely to experience earnings declines, to lose
money, or to go out of business. Such companies are
often in industries called sunset industries, in which there
is declining demand for the product (buggy whips are the
iconic example), or where foreign companies are becom-
ing the low-cost providers (as in the steel industry), or
where other some other unfavorable change in funda-
mentals is taking place. When looked at using only sta-
tistical tools, these companies appear cheap.
Fundamental analysis (including economic and com-
petitive market analysis) is again the only discipline that
will identify sunset companies and industries and enable
the value investor to distinguish pearls from junk in a
turbulent era when many industries may be experiencing
their sunset years.
Value managers may also learn a lot by studying
growth companies. The “creative destruction” (to use the
economist Joseph Schumpeter’s phrase) caused by today’s
rapid technological change affects every company, not just
the Internet and telecommunications firms selling the
technology. The value manager with a strong under-
standing of how creative destruction works, and how
changes in technology affect the companies he or she is
following, is likely to be better at separating true bargain
companies from those merely appearing to be bargains.
Funds Flows and Momentum
Because large moves in groups of stocks can occur
for purely technical reasons, momentum and funds flow
analysis should be incorporated into the value process—
just as these tools are used by growth investors.
As we noted earlier, one technical effect observed
in the first quarter of 2000 was the steepening of the rate
of decline in some widely held deep-value stocks because
of value fund redemptions unrelated to any news or
change in fundamentals. The value fund redemptions
are, in turn, due to poor performance of these funds in
immediately prior periods. Other technical effects come
from style index reconstitutions.
A value manager who is aware of these kinds of
effects may gain some performance advantages.
Defining Value as a
Nexus of Characteristics
Index constructors need rules or screens for choos-
ing value stocks that are mechanical, easily communicated
in advance, and efficient to implement. An index so con-
structed does not necessarily form the best basis for an
active portfolio.
To develop additional insight for beating, not match-
ing, indexes constructed according to simple rules, active
value managers may want to:
WINTER 2000 THE JOURNAL OF INVESTING 11
Draft
• Use more than one accounting ratio to screen for
value.
• Use different standards for different industries.
• Incorporate measures of company performance
(such as the FCFFvariable) that are more relevant
to valuation than the metrics used to construct
the indexes, and that are less “political” than
earnings in the sense that companies can use
them to influence the stock price.
CONCLUSION
Buying securities that are cheap relative to some
concept of fundamental value—value investing in the
broad sense—must work, in some form, perhaps with
the retooling we have described or perhaps with some
other adaptation, over some time horizon shorter than
infinity. If it did not work, both financial markets and the
real economy would seize up as capital allocated to over-
priced securities became progressively less productive.
In the long run—which, we admit, feels like an end-
less wait this time—the prices of financial assets must
necessarily converge on their fundamental values. Some
progress in this reconciliation is being made as of this writ-
ing, in April 2000, with a falling Nasdaq and a rising Dow
Jones Industrial Average.
Some things have changed. The network comput-
ing and telecommunications revolution of the last few
years will produce a reshaping of the real economy that
no investor can ignore. Most people probably underesti-
mate, not overestimate, the alterations in daily life and in
the structure of the economy that will eventually be
wrought by the Internet and other cheap, fast methods of
information transfer.
What has not changed is the idea that the worth of
an investment is the present value of the future benefits
it can produce—loosely speaking, the DCF model. This
is a law of economics and of human behavior and is not
dependent on circumstances.
The great challenge now facing investors is to fig-
ure out to what extent the changes observed and forecast
for the real economy are already reflected in market prices
—and, more broadly, to distinguish evolving economic
realities from unchanging “laws of nature” in a world that
provides no clear guideposts to tell them apart.
Looking ahead, value is a good bet to beat growth
over the intermediate term—a few years, perhaps—if only
because the starting point is so atypical in terms of the rel-
ative pricing of growth and value. (While growth portfo-
lios will continue to have almost all the “home runs,” they
will also include many stocks for which earnings disap-
pointments or other setbacks result in sharply negative
returns.) Over the longest time horizons (decades), how-
ever, our best guess is that the pendulum of performance
will continue to swing between the major investment
styles, and we recommend that investors hold portfolios
that are balanced with respect to these approaches.
ENDNOTES
The authors thank Clifford Asness, Kenneth French,
Roger Ibbotson, William Jacques, and Arnold Wood for their
contributions of data and exhibits. Conversations with Linda
Strumpf, Halliday Clark, Eric Doppstadt, and Clinton Steven-
son were enlightening. Mike Harhai, Emma Hastings, Nick
Mencher, and Jim Skesavage provided valuable guidance, as did
Nina Lesavoy. Jason Zweig and Clinton Stevenson provided
detailed editorial comments. Some of this material appeared in
an Invesco Research Summary, first quarter 2000.
1That is, the ratio of the two cumulative indexes (Rus-
sell 1000 Value divided by Russell 1000 Growth) fell by 39.9%.
In other words, a dollar invested in the growth index on
12/31/1997 grew to $1.85 by 2/29/2000, while a dollar
invested in the value index grew to only $1.11 over the same
period. $1.11 is 39.9% less than $1.85.
2Each year, all large-cap NYSE stocks (that is, stocks
having capitalizations greater than the median NYSE capital-
ization) were sorted by book/price ratio to construct the style
indexes, with value defined as the top 30% by book/price
ratio, and growth defined as the bottom 30%. The middle
40%, or core, are not shown. Returns are capitalization-
weighted. The Fama-French data are from Stocks, Bonds,
Bills, and Inflation 2000 Yearbook, updating the research of
Roger G. Ibbotson and Rex A. Sinquefield. See website
http:/web.mit.edu/kfrench/www/data_library.html.
3Lakonishok, Shleifer, and Vishny [1994] provide some
behavioral explanations as to why value ought to outperform
growth on average over the long run. In particular, they say that
investors tend to extrapolate past earnings growth rates too far
into the future, overreact to good and bad news, and make other
“cognitive errors” causing them to overestimate the value of
growing, well-run companies and to underestimate the value
of other companies. If these conditions hold, value stocks
should outperform on average over time.
4Exhibit 4 was provided by William Jacques of Mar-
tingale Asset Management in a private communication. An ear-
lier version of this graph appears in Jacques [1993, pp. 33-36].
The sorting of stocks into fractiles is performed by count (not
capitalization), where P/E is defined as the price divided by the
IBES estimate of the next 12 months’ reported earnings. Stocks
are actually sorted on E/P (the reciprocal of P/E) so that com-
12 THE FUTURE OF VALUE INVESTING WINTER 2000
Draft
panies with negative or zero earnings would be included in the
ranking. For March 2000, Martingale constructed the high-cap
universe using the method that BARRA employed for the pre-
ceding dates.
5A more complete description of mean reversion, and
of statistical tests for it, is in Poterba and Summers [1988].
6The return differential forecast by Asness et al. cannot be
compared precisely with those in the Fama-French study because
the definitions of growth and value differ.
7This article is highly recommended to the value skeptic.
8There is also strong evidence of mean reversion for
earnings growth rates and for ROEs of companies that are well
above or below average in these respects.
8Dividends are taxed once at the corporate level
(because dividends are not deductible from income subject to
corporate income tax), and once at the individual level (because
dividends are included in income subject to personal income
tax).
9For the first stage (years 1-3), we give only a 90%
(rather than 100%) weight to the IBES three- to five-year esti-
mate because these estimates are usually too optimistic; we also
terminate the first stage after three, not five, years.
10There is some event risk in this model, since slow-
growing value stocks occasionally become loss-making com-
panies that go out of business.
11We use the before- rather than after-tax market cost
of debt to avoid a “zero divide” problem with highly leveraged
companies. The problem occurs when high tax rates and lever-
age ratios combine to produce a WACC that is near, or even
below, the perpetual growth rate of the company.
12Only growth stocks with positive current FCFFwere
analyzed. Many growth stocks have negative current FCFF,
but our method does not produce a useful result for these
companies.
13Duration is the present value-weighted average time
to receipt of cash flows from an investment. With lower div-
idends, cash flows, and earnings in the near future and higher
flows in the far future, the duration of a growth stock is long;
with higher dividends and so forth, a value stock has a shorter
duration.
14Operating leverage equals fixed costs divided by the
sum of fixed and variable costs.
15See, for example, Asness, Porter, and Stevens [1999].
REFERENCES
Asness, Clifford S., R. Burt Porter, and Ross Stevens. “Pre-
dicting Stock Returns Using Industry-Relative Firm Charac-
teristics.” AQR Capital working paper, New York, 1999.
Asness, Clifford S., Jacques A. Friedman, Robert J. Krail, and
John M. Liew. “Style Timing: Value versus Growth.” The
Journal of Portfolio Management, Spring 2000, pp. 50-60.
Graham, Benjamin. The Intelligent Investor 4th revised edition.
New York: Harper & Row, 1973.
G
rantham, Jeremy, and Jack Gray. “New Paradigm or
Mean Reversion?” Investment Policy, vol.
2, no. 1 (Septem-
ber-October 1999), pp. 45-52. http://www.investment-
policy.com
Jacques, William E. “Low P/E Investing: Why It Works and
How to Capture the Returns.” The CAPM Controversy: Policy
and Strategy Implications for Investment Management. Charlottesville,
VA: Association for Investment Management and Research,
October 15, 1993.
Lakonishok, Josef, Andrei Shleifer, and Robert Vishny.
“Contrarian Investment, Extrapolation, and Risk.” Journal of
Finance, 49 (1994), pp. 1541-1578.
Poterba, James M., and Lawrence H. Summers. “Mean Rever-
sion in Stock Prices.” Journal of Financial Economics, October
1988, pp. 27-59.
Stocks, Bonds, Bills, and Inflation 2000 Yearbook. Chicago:
Ibbotson Associates, 2000, pp. 147-172.
WINTER 2000 THE JOURNAL OF INVESTING 13
