Tobias Carlisle: “Deep Value Investing” | Talks at Google

MALE SPEAKER: Welcome, everyone. We have a very special
guest here today with us, Toby Carlisle. Welcome to our Talk at
Google, at Author series. He’s going to be talking about
his new book, “Deep Value: Why Activist Investors
and Other Contrarians Battle for the Control
of Losing Corporations”. Toby has a unique perspective
on value investing, much like Google’s. His approach to
value investing is data driven, just like Google’s
approach to engineering. In his book he uses statistical
analysis and in-depth research to give evidence for a simple,
yet counter-intuitive idea. Losing stocks with
failing businesses and uncertain
futures can sometimes offer unusually attractive
investment potential. His work in this area is
very much like Ben Graham’s classic work in his 1934
“Security Analysis” book. The funny thing about classics
is everyone’s heard about them, very few people have
actually read the classics. Toby is one of those rare,
few individuals– investors, I would say– who has
been living and breathing Ben Graham’s deep value
investing philosophy. So please help me
welcome Toby Carlisle. [APPLAUSE] TOBIAS CARLISLE: Thank you. Hi folks, thank you very much
for having me here today. I’m very much looking forward
to speaking to Google. Just a little bit about me,
my name’s Tobias Carlisle, I’m the managing director of
Eyquem Investment Management, and I manage the Fund and some
separately managed accounts. My most recent book
is “Deep Value”, and before that I published
“Quantitative Value”, co-authored with a gentleman
who did his qualitative research through Booth. I blog at Greenbackd
and I’m just launching a new site called
the Acquirer’s Multiple that is yet to go live. The talk today
proceeds in four parts. The first part we’re
going to discuss is the philosophy of deep value. What is is, its genesis. Then we’re going to look at
some of the behavioral reasons that stocks become undervalued
in Contrarians at the Gate. Third, we’re going
to examine some ways that we can avoid making
those behavioral errors that create undervalued stocks. And finally, we’re going to
examine the simple metric for generating outstanding
returns in the market. In 1927, a 33-year-old
Benjamin Graham started teaching a
night class at Columbia that he called
Security Analysis. He offered practical
counsel for the students who took the class, but the
thrust of what he taught was that a security
had an intrinsic value, and it was distinct from
its market price, which you could observe on the
stock market in any given day. And he argued that when there
was a sufficient mis-pricing, either the price
was at a discount to the intrinsic value,
or, in some instances where the price was at a
premium to the intrinsic value, you could trade long or
short and rely on the fact that there would be
some mean reversion to bring the price back
to intrinsic value. In 1929, the stock market
had its worst crash, it continues to be the
worst crash to date. From its peak it to its
trough it fell almost 90%. In 1932, fully three years
after the bottom of that crash, the stock market still
hadn’t recovered. Graham authored a series of
articles for Forbes Magazine where he pointed out some
research that he had undertaken that said that of the 600
issues on the stock market at that time, 200 of them traded
for less than their liquidation value. So they traded for less
than what the shareholders in those companies
could get out of them if they wanted to wind them up. And then a sizable portion
of those traded at a discount to their net cash
backing, which is after paying out all of the
liabilities of the company, there would still be cash
lift in a bank account and you could buy the
entire company for less than that amount. And he said– he likened it to
corporate gold dollars being traded for $0.50 or less
with strings attached. That sort of enduring iconic
image continues to be the way that we think about
value investing, that it is an attempt
to buy a $0.50 dollar. Some years later,
John Burr Williams added some additional
color to the way that we go about
valuing a company, and that is to look at all of
the cash inflows and outflows, discount them back
to the present day, and that gives you
an intrinsic value. So as you hunt in the market
for these particular stocks, you can find those
at the greatest discount to intrinsic value
will tend to underperform and those at the greatest
premium to intrinsic value will tend to– sorry,
will tend to underperform. So this chart shows, if we rank
every stock in the universe– and this is a global universe,
so 22,000 positions– if we give them all a ranking, price
to cash flow, price to book, price to earnings, we
take the average of those and then we rank them
into five groups. The glamour quintile, which
is the most expensive, tends to underperform
on average, and the value quintile,
which is the cheapest, tends to outperform. So when we’re seeking to
invest in these companies, one thing that we can
observe is that price tends to fluctuate more
than intrinsic value, which only moves much more slowly. And so when you find these
occasional moments when there’s a large discount
or a large premium, you can invest to
capture that discount. The question that
I’m often asked is what is it that causes
the discounted price to return to its
intrinsic value? And it’s a question
that Benjamin Graham got when he was appearing before
a Senate committee in 1955, and the chairman said,
how do you close that gap? Is it by advertising,
or what happens? And Graham gave one
of his great answers where he said it’s
a mystery, and it’s as much of a mystery to him
as it is to everybody else. But we understand it
to mean mean reversion, and so what this
talk is about is what are the actual
mechanics of mean reversion. What are the things that
make mean reversion occur. This is actually Graham’s list. He talks about, in
one, that’s just general improvement
in the industry. And this is a very
common occurrence that the industry
is in a trough, the business is
in a trough, it’s got a period of bad luck,
perhaps some bad management, and either the
incumbent managers turn the company around
or external managers are brought in. Less commonly, a
sale or a merger, or some sort of liquidation. Some time later, behavioral
researchers DeBondt and Thaler examined securities on
the basis of a variety of different metrics, but
they’re looking, particularly in this case, for undervaluation
and overvaluation measured by price to book value. So they found that by
selecting the decile, which is 1/10 of stocks that
were most undervalued, and the decile of stocks
that were most overvalued, and then examining the
trend in earnings per share, you find this unusual phenomenon
where the earnings for the most undervalued stocks have
been falling for the years previous to the
portfolio’s being formed, and for the most
overvalued stocks, they’ve been increasing. And that’s the reason
that one is overvalued and the other is undervalued. But then this very
unusual thing happens once the portfolios are formed. The overvalued
stocks still continue to grow but they
slow a great deal, and the earnings per share
of the undervalued stocks begin to improve at this stage. And four years
hence, on average, the undervalued stocks which
have been selected simply based on their undervaluation
have actually earned more and are earning more than
the overvalued stocks which had the highest rate of growth. It’s a phenomenon that we
see over and over again in the research. So this is another
interesting study subsequent to the
Debondt and Thaler research by Lakonishok,
Shleifer, and Vishny. They said, well, if
it is in fact the case that there is this main
reversion in the fundamentals of the business, if
there’s mean reversion in the earnings, if the
undervalued stocks do start creating earnings growth and
the overvalued stocks’ earnings slow, then what if we went
and selected portfolios specifically on– we
could find an undervalued portfolio with very slow growth
and an overvalued portfolio with very high growth, and an
undervalued portfolio with very high growth, which would
be the better portfolio? So they sorted the universe
into three portfolios on the basis of valuation,
from most overvalued to the median average, and
to the most undervalued. And then they divided
each of those portfolios into the highest rate of
growth, the mid rate of growth, and the slowest rate of growth. So there were nine
portfolios formed. I’ve selected three of them
here to show you the phenomenon. So the glamour portfolio has
the highest rate of growth, and you can see an earnings,
cash flow, book value, operating earnings, all of
those are growing higher than the other two
value portfolios. But you can also
see that you have to pay a higher multiple
to acquire these stocks, so at 19.6 times 10.8 times cash
flow, they’re expensive stocks. On the other end
of the scale, these are two of the three
value portfolios. The high growth value
portfolio is still growing at quite a
high rate of growth, but it’s available for
a much cheaper multiple, and the contrarian
value portfolio, which is the very
low growth portfolio, is growing at a low rate and
available on a low multiple. So as you would expect,
the value portfolios outperformed the
glamour portfolios. But this is the really
interesting thing. The low growth
undervalued portfolio outperforms the high growth
undervalued portfolio, and this is something
really unusual, and I think it’s something
that is a little bit counter-intuitive
because we sort of like to think that high
growth undervalued stocks are bargains, gems in the rough. And it turns out
that mean reversion acts on those portfolios as well
to pull the high growth back and to improve the low growth. So this is just that previous
slide with the glamour portfolio removed, just so
we can see specifically. The thing that I want to
draw your attention to here is that the contrarian
value portfolio, in many of these categories,
is actually a little bit more expensive than the
high growth portfolio. The reason possibly is
that earnings are little bit more anemic, cash flow
is a little bit more anemic. So you can see on
the book value basis, the high growth value portfolio
is slightly more expensive. But it’s very definitely growing
much faster than the contrarian value portfolio. So the real driver
of returns here was the undervaluation and
not the rate of growth. And then there was some
mean reversion as well which pushed down the
high rate of growth and made the high
growth value less attractive over a
period of time compared to the contrarian value. This was an earlier examination
conducted by Michelle Clayman. There was a book written
by Tom Peters that came out in the early ’80s called
“In Search of Excellence”, and Peter said that he’d
identified the characteristics that created
excellent companies. He looked at a variety of these
quantitative financial type elements. The book has been described
as the book of the century, possibly by Tom Peters himself. I’ve never read it, but
I’ve read the research, and the research is compelling. These are his criteria
for an excellent company, so they have very high rates of
asset growth, very high rates of equity growth,
they’re expensive, and they have excellent returns
on capital, assets, and equity. So Clayman said, well, let’s
create the opposite portfolio. Let’s call them
unexcellent companies, and we’re going to find
those with the worst asset growth, equity growth, low
valuation, anemic returns on capital, equity,
and assets, and then we’re going to track
their performance over a period of time. The thing that I’ll
draw your attention to is just the valuation alone. You can see here,
this is some research that appears in the book. It was conducted by Barry
Bannister at Stifel Financial. The unexcellent portfolios
have vastly outpaced the excellent portfolios. For an early period
of time, and it’s difficult to see on this
chart, the excellent portfolios did actually
outperform the market. But not since you get a better
return investing in the S&P 500 than you do investing in these
very high quality, high growth companies, these
excellent companies. And you get a much
better return investing in the ugliest of the ugly. The question that
naturally people ask when they see
that data is well, is there some
advantage to investing in the excellent
portfolios because they offer some sort of
risk protection? Is it a downside
protection phenomenon, is it that when the world
goes into a recession, you want to be in
these things that generate higher rates of growth? And the answer is no. Bannister looked at periods
where the global growth was below average
and above average. The shaded area is where
global economic growth is below average,
and the lighter areas are where global and economic
growth exceeds the average. And you can see quite
clearly, there’s no real method to
when unexcellent outperforms or
excellent outperforms. It does outperform occasionally,
but it doesn’t necessarily need a bad state of
the world to do so. I always like to use this motif
to describe the nature of what occurs in these
undervalued companies. Most companies are going
to go through this cycle. There are a very small
number of companies that can avoid this cycle, but
a very small number of them. This is from a woodcut
from Albrecht Durer in 1494 from a book called
“The Ship of Fools”, which is about some fools on
a ship to a fool’s paradise. And they go through, they
have a number of follies. One of them is hubris about
where they are and disregarding the role of luck in their
own lives, good and bad. So it’s the wheel of fortune,
that’s Fortuna’s hand puppeting the wheel in the top
left hand corner. The ass at the top
is regno, “I reign”. And you can see that he’s
at the very pinnacle, and he’s about to descend
on the other side, but instead he reaches
out for the sun. And this is what, in the
academic financial literature, they describe as
naive extrapolation. You imagine that
your current state will continue on
when, in actual fact, mean reversion is a much
more likely outcome. So the next,
regnavi, “I reigned”. He’s slowly turning
back into a man as he descends down
the other side. And this is, perhaps,
where deep value investors want to stake their claim. Sum sine regno, “I am
without a kingdom”, he’s not even visible
on the wood cut. And then back up the other
side, regnabo, “I will reign”, maybe that’s another
good time to invest, too. If what we’re looking for is
fundamental weakness, declining earnings, and
undervaluation, the thing that makes them difficult
to buy for other investors makes it equally
difficult for us to buy. So how can we avoid making
the behavioral errors that other investors are making? How can we not be the naive
extrapolation investor, and become a contrarian
value investor that captures that upside performance
from buying companies with ugly fundamentals? There’s been a great
deal of research into this area of the role of
statistical prediction models and experts. It’s a counter-intuitive area
of study for this reason. So a professor went and
studied 1,000 admissions to hospital of
depression or psychosis, and apparently, on
initial presentation, they can appear very similar. Then there’s a subsequent
diagnosis several weeks or months later where they’re
able to determine properly whether the patient
who presented was in fact psychotic
or depressed. So he created, by
examining the admission records and the
subsequent diagnoses, he created a simple
questionnaire that could be administered
by psychologists when these people presented
to the hospital and they could make
a determination. Without the benefit
of the model, the worst psychologists got it
right about 55% of the time, on average about 62%, and
the very best psychologists got it right 67% of the time. The simple model in back test
got it right 70% of the time. The professor distributed
the questionnaire to a variety of different
psychologists in the hospital. There were two
basically inexperienced who were students,
and then there were clinical psychologists
who were actually treating patients,
and they found that without the
benefit of the model, the inexperienced
students got it right about 59% of the time,
the experienced clinical psychologists got it right
about 64% of the time. The inexperienced
psychologist, with the benefit of the model but the ability to
override the model’s output got it right about 2/3 of the
time, and the experience psychologists, with the
benefit of the model, got it right 75% of the time. But this is the really
incredible thing. The simple model by itself
outperformed all of them. The reason is that
we make errors when we attempt to apply a model
and we don’t follow the model’s output, and that leads
us to underperform what the simple
model does by itself. The reason that people– when
people observe this research, they often say, if
you have a case that’s so different from
the base, right, from the statistical research
that you’ve conducted, doesn’t it make sense then to
be able to override the model? And the example
they always give, you have some sort of algorithm
that predicts whether John will go to the theater on
Friday, and now you know that he has a
broken leg, shouldn’t you be able to factor
that into your model to determine whether he will,
in fact, go to the theater this Friday. And the answer is no. And the reason is that we
tend to find more broken legs than there really are. And particularly in this
type of deep value investing, all of the stocks have what
appear to be broken legs. You might be familiar
with “The Little Book That Beats the Market”, it’s
a book by Joel Greenblatt. He took Warren
Buffett’s exhortation to buy wonderful
companies at fair prices, and he translated that
into a quantitative model that we’ll examine in
detail in a moment. But for the moment,
you just need to know that it meant
high quality, however that’s defined, and good value. He had an experiment
in his own firm where he handed out lists of the
stocks selected by the formula, and he allowed people to, in
their own separately managed accounts, to either choose
the stocks that they wanted, or to have him apply the
formula automatically. He found that the automatically
applied formula did, in fact, outperform the market
over two years, and by quite a
substantial margin. He also found that when
people were allowed to manage the
portfolio themselves, they tended to
underperform, and the reason is that they cherry-picked
out the very best stocks, and they were the stocks
that were the ugliest. You might say well,
they were not experts, they were people who were
relying on his expertise. But then Greenblatt says, we
attempted to do the same thing, and he found he had
the same outcome. He underperformed his own model. So that brings us to the golden
rule of statistical prediction rules, and that is that simple
models outperform experts. That simple models continue
to outperform experts even when they have the
benefit of the model. When we’re thinking
about designing a quantitative system,
there are several things we need to be aware of. One of them is that the
rules must be simple and they must be concrete. They must be simple so that
they’re able to be followed, and concrete so that they’re
able to be understood. This is a picture of a pipe,
this is not, in fact, a pipe. This is a picture of
a picture of a pipe. And when we’re valuing
companies, in many ways, this is sort of
what we’re doing. We’re using some proxy. And there are a variety of
different models or proxies that we can select. We can choose the
liquidation value, we can look at the franchise
value, the growth value, earnings power value,
the acquirers multiple. We can use any number of
these simple multiples, but we’re not really getting
the truth of the company. And so it’s important
that we recognize, first, that there are
limitations to the model. Recognizing that the model
is somewhat imperfect, we can recognize
then that we do have some tool for making a decision. So the 80/20 rule does apply to
information in investing, too. You get 80% of the way
there with a simple model, and the urge of
most investors is to continue to find that
final bit of information that will remove the
uncertainty that they have. But often by the
time that occurs, the uncertainty has been
removed for everybody else as well, and
so those low prices that you were
attracted to are gone. The simplest rule, I think,
and one of the most effective, is the net current
asset value rule. Graham wrote about this in 1934. He described it as a rough
measure of liquidating value. So you can see, you look
simply at the balance sheet, you ignore the earnings,
the cash flow statement– it didn’t exist at the time
that Graham put this together, but you can ignore it for the
purpose of this analysis– and then you treat only the
current assets as having value. So cash is worth cash,
receivables, there’s some discount applied
to them, inventories would vary depending on whether
it’s food that can spoil, high fashion that will be less
valuable in 12 months time, or something that’s going to
continue to be valuable well into the future, and then you
can look at the fixed assets and determine whether
they have a value. The research shows
quite comprehensively that net current asset
value massively outperforms. And it’s an extraordinary thing
that is such a simple analysis. So the market in these
instances is always a small capitalization,
micro-capitalization equivalent to what we’re looking
at for the net current asset value stocks. You can see in the US, that’s
a study from 1970 to 1983, the net current
asset value portfolio did 29% versus 11.5%
for the market. This is a study that
I conducted in the US, ’83 to 2010, just continuing
on the second study. So it was a second 27
years and a similar sort of outperformance,
maybe more so. Which is surprising,
because you would think that with all of
the access to information, a lot of these positions would
have been arbitraged away, but they haven’t been. They continue to exist in the
UK and they exist globally as well. But there are some
unusual things that you find when you
conduct this sort of research. The first one is that the
individual net current asset value stock is more
likely to go to 0 than the rest of the market,
and the rate is about 6% for a net current
asset value stock versus 2.5% for the average
stock in the market. But as a portfolio,
there are fewer down years than the market. And this is perhaps the most
controversial finding, the most counterintuitive, and that’s
that loss-making net-nets actually outperform net-nets
that are profitable. Within the profitable net-nets,
those that don’t pay a dividend outperform the ones
that do pay a dividend. So your instinct might
be to find a net-net that pays a dividend
with positive earnings, and that would lead you
to underperform what you can do just
by net-nets alone. Many of you will know that
Warren Buffett started out as a student of Benjamin
Graham’s, then he worked with him in his
Graham-Newman Corporation, and then he went out and he
set up his own partnership. In the early days
of his partnership, he was very much a
Graham-type investor, looking at liquidation value. One of the early positions
that he put into the fund was Sanborn Map. You had a portfolio
of securities and it was trading at $0.65 on
the dollar of that portfolio of securities, with
no value given at all to the business, which was
making about $100,000 a year. He got on the board,
he got control, he liquidated the security
portfolio, paid it out in a tax-effective manner, and
so he got a 50% return just on the portfolio, and then
the business remained. And the business
continues to exist to this day as a geographic
information system business in the US. He had this evolution after
meeting Charlie Munger, and in 1963, he found American
Express mired in a scandal. American Express is
a financial company, and it had provided
some warehouse receipts to Anthony “Tino”
De Angeles, who was a commodities
broker and trader. He had figured out that he could
bring soybean oil to the port, show the inspectors that his
tanks contained soybean oil, and then through
pipes and valves he could fill up with sea water
what had been soybean oil. Then they would check
a new tank and it would contain the
old soybean oil. At one stage, he controlled
almost 10 times more soybean oil than there was in existence. And that was the secret
to his low, low prices. Eventually he went bankrupt. American Express had
said that the warehouses did in fact contain
the salad oil. His broker went bankrupt
when the people who had lent against those
warehouse receipts came looking for
some deep pockets, they found American Express. And American Express
owed something in the order of
$175 million, which was 10 times its average annual
earnings of the last few years. So there was a real risk
that American Express would go bankrupt at that stage. Buffet put 40% of his portfolio
into the stock, it recovered. He bought $13 million
worth of that stock. That position today,
if he had held it– which he didn’t all the way
through because he’s changed investment entities–
that position today is worth something in
the order of $14 billion, which is an enormous return. He learned from
that that it wasn’t necessary for these things
to have a liquidating value. He could, in fact, buy them
on the basis of a franchise. He made sure that
people were still continuing to use
American Express cards, and so it taught
him that there was a different method of investing. In 1972, he found See’s Candies. It was earning something
like $2 million on $8 million of invested capital. They paid $27 million
dollars for it. He assessed the value in
the order of $45 million, so he got a fairly
substantial discount. And he said at the time when
someone asked him about it, are you still a
Grahamite-type investor. He said I’m 85% Graham, 15%
Phil Fisher, who recommended the scuttlebutt
method of investing, which is you go and
find as much information as you can about the
quality of the business and its ability to grow. So See’s Candies, between 1972
to 2011, returned $1.35 billion to Berkshire Hathaway, which
they’ve continued to invest. And it’s required only
something like $70 million reinvested in the business
to generate those earnings. So that’s what’s
known as a franchise. The lesson that he took from
investing in See’s Candies was that you’re much better
off with these businesses that are able to grow over
a long period of time. He said, though the cigar butt
might have a single puff left in it, and that puff is
pure profit, after you’ve smoke that puff,
there’s nothing left. So he stopped being
a cigar butt investor and became an investor
in wonderful companies at fair prices. Looks at Buffett’s
investment methodology and says that’s
been very lucrative for a very long period of time. Assuming we don’t have
Buffet’s great mind, are we just able to create
a quantitative version of the methodology that Buffet
describes in his letters, and he wrote a book
about that process. So he decided that good
quality, as defined by Buffett, means a high return
on invested capital. So invested capital
is the money that you need invested in
the business to run the business, the assets of
the business that are actually used to produce income. The higher the return
on invested capital, the better the business, the
faster it’s able to grow. And for valuation, he
uses an endings yield, earnings before
interest and taxes, because it’s agnostic
to the capital structure where interest payments on debt
affect the tax that you pay. If you back out
interest and tax then you get this idea of
the operating earnings that are coming
into the business. We tested this
qualitative value, and we found that the
magic formula does, in fact, beat the market, a
comparable market of stocks. And quite comprehensively, it’s
beaten it by 3.5% each year from 1974 to 2011. What is really shocking is
that the earnings yield alone, what I describe as the
acquirer’s multiple, beat the magic formula itself,
and the quality measure actually underperformed
the market. The quality measure actually
led the magic formula to underperform the
earnings yield alone. And it’s not just
a performance, it’s not just a raw return story. It’s, in fact, a risk-adjusted
return story as well. The earnings yield alone
generates a better Sharpe ratio and a better Sortino
ratio, which is basically the amount of growth relative
to the amount of variability in the returns. So you get better
returns and you get better risk adjusted returns
just using the earnings yield. And the reason is that
there is this mean reversion in return on invested capital. Michael Mauboussin
has tested this. He’s examined a number of
the thousand largest listed stocks in the US, ranked
them in order of return on invested capital, and then
divided them into five groups. So the very highest,
and then he’s ranked them into very
highest and the very lowest. And then he’s examined those
same companies 10 years hence. And what he finds is that the
highest return on invested capital mean reverts
towards the mean return, and the lowest return
on invested capital also slightly improves. So I think if the highest
return on invested capital as being regno, at the
very peak of the wheel, and sum sine regno at the
very bottom of the wheel. What is the acquirer’s multiple. Well, it’s the
enterprise multiple. The reason it’s called
the acquirer’s multiple is it’s the metric used
by leveraged buyout firms, private equity firms,
activists, to look through into the hidden
value of the business and, in terms of
the mechanics of it, it’s market
capitalization plus debt. Because the business
has to fund the debt, you’re able to use the
cash to pay off the debt. You’re also liable for
the preferred stock, you’re also liable for
any minority interests, and underfunded pensions, off
balance sheet liabilities. So it’s the real cost
to buy the business. And then you get access
to EBITDA or EBIT, which is the cash
flow coming in. It turns out that
it doesn’t really make much difference
which you choose. But when we tested them using
data from 1964 to 2011– so we tested a variety of
different possible metrics. Earnings yield, which is the
inverse of the price earnings metric, just to
highlight the two. The acquirer’s
multiple using EBIT, the acquirer’s
multiple using EBITDA. Free cash flow on enterprise
value, gross profits yield, so that’s just
revenues minus cost of goods sold to give you
the third line on the income statement, and book
to market, which is the inverse of price
to books so that they’re arranged in the same way. We found that the acquirer’s
multiple outperformed. And again, not just
on a raw return basis. It outperforms on a
risk-adjusted return basis, too, and comprehensively. The four things I want
to take away from today are that the very deepest
undervalued stocks outperform the highest quality
and the highest growth, even in the
undervalued portfolio. So undervalued portfolios
divided into high growth and low growth, the low
growth undervalued portfolios will outperform. So it’s better to be
assuming that there’s going to be some mean reversion,
positive and negative, rather than to naively
extrapolate out the growth in earnings. Simple models,
applying these ideas will always outperform
expert discretion. So at the beginning
of your process, you decide what is important
in the assessment of value, and then you rigorously apply
that without fear of failure. And the acquirer’s multiple
is the best multiple. If you’re looking for a very
simple application, a very simple rule, this
is a very good one. This is a plug
for the book, so I go into these studies in a great
deal more depth in the book. Each one of them– there are
several different versions of this study, so you can see as
we go through the entire book. The site Acquirer’s
Multiple has a place to capture your email
and your details if you’d like to learn more
when it’s up and going. It will basically
provide a free screen of acquirer’s
multiple companies, and some commentary and some
research as it goes along. So if you have any questions,
I’d be happy to hear them. AUDIENCE: This question is
about the mean reversion. Amongst the universe
of companies, do you think there are
some particular companies or industries that are more
flexible towards mean reversion and others that are
not, or did you think it’s the entire universe that
this mean reverts over there. TOBIAS CARLISLE:
There are certainly some companies do demonstrate
persistence in their ability to maintain a high return
on invested capital. And the question is that if you
look at a large enough universe of stocks, would you
just expect that there would be no persistence at
all, would you find some just by random chance? So it’s not entirely
clear whether there’s a reason for their persistence,
or whether it’s just the luck of looking
at a 10 year period. [? Maberson ?] has looked
at that specifically, and he found that
pharmaceuticals and biotechnology
and another group did demonstrate
some persistence, so they kept the high
returns on invested capital. But he wasn’t able to
determine the factors that– so prospectively, if you look
at a data set without knowing the outcome, can you determine
which ones are going to persist and which ones are
going to mean revert, and he hasn’t been
able to do that yet. AUDIENCE: You mentioned
the EBIT divided by EV is actually better than
the magical formula. Is Joel Greenblatt
aware of that? I cannot imagine that he
doesn’t know that, right? It doesn’t make a
lot of sense to me. TOBIAS CARLISLE: He’s almost
certainly aware of that. AUDIENCE: But you
think he knows that, but he still put
[? RYC ?] into the book. TOBIAS CARLISLE: Well, the
magic formula does outperform. The magic formula
does beat the market. AUDIENCE: Yeah, but
I’m sure that he also has done the study on the
separate metrics, right? Do you think he
hasn’t done the– TOBIAS CARLISLE: Well,
I’m sure that he has. AUDIENCE: But, OK, OK. I guess– TOBIAS CARLISLE: So
why do it that way? Well, you know there are– AUDIENCE: You understand
my question, yeah. TOBIAS CARLISLE: I do. There are elements
to the business. It’s a marketing business as
much as it’s a return business. AUDIENCE: But I
have the impression that Joel Greenblatt
doesn’t really– I mean, he has enough
money and he doesn’t really care about making more money. I mean, I could be wrong. TOBIAS CARLISLE: In different
states of the world, the magic formula will
outperform the earnings yield. So when the markets are
racing up, in a bull market, it’s about even. But there are some
periods of time– the late 1990s
for example, which was an unusual period in the
markets– but the magic formula did outperform the
pure earnings yield. But the earnings yield has
outperformed the magic formula over the last 15 years. So it may be a job
security type Idea that if you apply the
magic formula rather than the earnings
yield, you’ll have fewer years of underperformance. When I test that,
that’s not what I find. Basically, the earnings yield
outperforms the magic formula pretty consistently,
and pretty consistently over five and 10
year rolling periods. I think it’s the better
metric for the reason that I’ve outlined here, that
return on invested capital, mean reversion, is a
very real phenomenon. And it’s an easily
explained phenomenon. When you find companies
that are very profitable, it invites competitors to
go into those industries. In industries that low return,
people leave the industry. That happens all the time. It’s happening right now
in the oil and gas industry with low rates of return,
because the oil price is so low, they’ll
stop drilling holes. So that has a flow on effect. AUDIENCE: Joel Greenblatt
talks about the fact that a lot of these economical
companies are too cheap and don’t have enough liquidity
for the fund managers to own, and also that a lot
of the fund managers just want saleable product. That’s the popular
names, and so on. So could you comment
on the liquidity issue and moving in and out of
some of these smaller issues? TOBIAS CARLISLE: The
book, “The Little Book That Beats the
Market”, shows a return for the magic formula
that’s much higher than you can achieve if you adjust
for liquidity and for size, but you can still
get outperformance applying it in a very large
capitalization universe. And market
capitalization weighted, which means you’ve sized the
positions in the portfolio relative to the market
capitalization of the company that you’re buying rather
than equally weighting them, so that means that
you’re putting more money into the bigger companies. That is testing for
precisely that problem. So the results that
I was showing in here were actually a market
capitalization weighted. So you get slightly
better results, again, if you equal weight,
for the simple reason that you’re putting more
money into smaller positions. So market
capitalization weighting sort of adjusts for that. The problem with something like
the net current asset value is that it’s really not
an investable strategy for anybody other
than an individual. They’re just not
around often enough, and they’re not big enough or
liquid enough to invest in. If you have a million dollars
of investable capital, it’s probably too small for you. But the magic formula and
the acquirer’s multiple scale beautifully. You can take the acquirer’s
multiple in an S&P 500 universe, select
the 5% or the 10%, so the 25 or 50 stocks in that
universe, roll it once a year, and you’ll find
pretty consistently that you get quite a lot of
outperformance doing that. As a broad problem,
liquidity and size are issues that reduce returns,
but these metrics still do work in large universes. You get the incredible
outperformance if you’re able to invest
in the little stuff, which small managers and
individuals can. Does that answer the question? AUDIENCE: Yeah, that’s
a very good answer. AUDIENCE: So I think your
last statements have answered a lot of my questions
I was going to ask. But for the magic
formula backtest, I would like to
know more details. So one is, you said the
rebalance is once per year. TOBIAS CARLISLE: Yes. AUDIENCE: OK. So how does the backtest deal
with the survival buyers? TOBIAS CARLISLE:
That’s a good question. So the database that
we use, Compustat, keeps companies that
have failed in the data. So if it failed in 1975, the
backtester would have bought it and could have
bought it in 1974, because the data
remains in the database. So we do a number of things. In quantitative value,
we outline in some detail the process we go
through to backtest. Basically we do a
number of things. We lag the data, so one
of the problems you have is that you have this
look-ahead bias, which is the possibility for
trading on information that you don’t already have. And in addition to that, you
have this January effect, which is quite pronounced. So there’s some tax loss
selling at the end of the year, and then if you invest
assuming that you can buy your entire position on
the very first day of January, you capture quite a
bump in performance. So what we did is we rebalance
the portfolios on June 30, and we used data from
the preceding December. So we use the K data and
we rebalanced in June. So we’re avoiding
that January effect, and we’re avoiding
the look-ahead bias, and then we use a
very good database. AUDIENCE: So the basket is
about 30 stocks, in the basket? TOBIAS CARLISLE:
It varies depending on the size of your universe. The universe, and depending
on the size of how much you’re investing and how much time
you want to spend doing it, the fewest stocks that you
could possibly do it with might be 20. And the most stocks, you
really don’t get much benefit beyond about 30. So somewhere
between 20 and 30 is the appropriate size for
portfolio, equally weighted. That also means that you can
buy them on a quarterly basis. So you might buy–
if you’re buying 20, you buy five on a
quarterly basis, and then you rebalance those
five 12 months and one day to capture the tax effect. Does that make sense? AUDIENCE: OK. Yeah. Thank you. AUDIENCE: So if more people
use the same strategy, wouldn’t your strategy just
be equal to the market? TOBIAS CARLISLE: Possibly. But Joel Greenblatt
wrote that book in 2006. The magic formula doubled the
return of the stock market last year. So the magic formula
has continued to work. And the reason that
it continues to work is that there aren’t that many
value investors out there. It’s a new strategy, and
within that new strategy, the deep value investment
is a new strategy. Most guys who are
value investors would try to emulate
Buffett, they’re franchise-type investors. There are very few guys who
are really deep value guys. And there’s also this
problem that you’ll find. If you look at the data–
I’ll give these positions away for free on, and you can go ahead there
and you’ll be able to see, they’re quite
frightening positions. You’ll be buying very cheap
iron ore mining companies, you’ll be buying very cheap
oil and gas companies. A few years ago, you’d have been
buying for profit education. They’re frightening to buy. And that’s the real thing
that drives returns. That’s why I went through that
part of the presentation that says you have to
follow the model. That’s the most important
thing, without fear of failure, you can’t cherry pick. Because what you do
when you cherry pick is you do what everybody
else does, and you avoid the things that generate
the really big outperformance. AUDIENCE: OK. I’ve got a followup question. Have you tried the same
strategy in different markets? TOBIAS CARLISLE: Yes. Yes, so the magic formula works. And I discuss this “Deep
Value”, but the magic formula works in Japan, it
works in the UK, and it works in
Europe ex the UK. In each of those instances,
the acquirer’s multiple alone outperforms. The only place where
that hasn’t happened for the period of the
data that we examined, which was about 14
years, because there just isn’t that much
international data. But Japan, the
magic formula seem to outperform the acquirer’s
multiple alone in that one single region over that 14
year period that we examined. AUDIENCE: Hey. So I was wondering, what
does your company do then? Do you just follow
the model as well? TOBIAS CARLISLE:
So basically, we have some additional
things that we do. Quantitative value goes
through, quite comprehensively, a very big model
that you could use. And we sort of discuss. You can do things like
avoiding companies at high risk of
financial distress, avoiding fraud, avoiding
earnings manipulators, looking for financial
strength, looking for quality of earnings. So that’s making sure that the
cash flow into the business matches the accounting earnings. All of those things add a little
bit of additional performance at the edges. But the big muscle movement,
the big driver of performance is the acquirer’s multiple. So yes, that’s what I do. That’s what I apply. AUDIENCE: One question. Why won’t Buffett move
to liquidator to operate. Is it because he
had too much money? And another question
related to the first book of Joel Greenblatt. He’s talking about
the spin off and also the special certification. I think his point is that just
buying those would be good, but if you can pick your spot,
yeah, that would be better. I think his
performance in test was like 50%, that’s the
highest I’ve ever known. Then his way is like this is
the area you want to pick from, and then you want
to pick the best. That is kind of not dissimilar
from what you are saying here. You’re saying things like
don’t involve any human being. Just buy everything, right. Yeah. Just want to know your– TOBIAS CARLISLE: His first
book, “You Can Be a Stock Market Genius”, which is a terrible
name for a really good book, the process that he
describes in there is a reasonably complex
investment strategy. It’s a strategy that really only
a human being could implement at the moment,
because it requires reading unusual filings and
finding spin-offs or companies about to pay out a big dividend,
special situation investing. I don’t think he
could have invested as much money in that strategy
as he can in the magic formula, and I think that that’s
a more difficult strategy to implement. You can certainly get
expertise in an area and understand better
than another area. Spin-offs, special situations
are really a very broad basket of potential things
that you can do, and you can become
good in an industry or good at a particular
thing, and that might lead to
better performance. I sort of think that the
broader your strategies, and the broader your
potential universe of stocks, the better performance
you’re going to have. I think to maintain that
very high rate of return, he was paying out lot of
the capital that he was– so he’d make the profit and pay
it out, and then do it again the next year. Whereas the magic formula’s
a compounding strategy. You roll over whatever
you get and you reinvest that compounded
amount, and it’s a strategy that really shows
how good it is over a longer period of time, because
you get to that point where you’re investing
larger sums of money at a higher rate than a special
situation one, which requires that you keep on paying
it out and you’re sort of limited a little bit
in where you can apply. AUDIENCE: [INAUDIBLE]? TOBIAS CARLISLE:
Why did he change? AUDIENCE: Yeah, yeah, yeah. Why’d he change? TOBIAS CARLISLE:
Possibly the challenge. Maybe he had too
much money to invest. AUDIENCE: So one of
the biggest criticisms of any model created
looking at backward data is you are
fitting data, right. And so the strategy that is
outlined feels a bit like, well, let’s ignore quality and
go for the liquidation value, the enterprise value,
buying price of the company. And that could be kind
of a valid criticism. What would be your
answer to that? TOBIAS CARLISLE:
Well, we’ve looked at it in different markets, too. So I certainly didn’t go into
it– it’s not a great marketing strategy to tell somebody
that what we’re going to do is buy the lowest quality
stocks that we can find, or not really care
about quality. It’s a much better marketing
strategy to go and say, what we do is we look for
really high quality stocks that are undervalued, and
that’s what we buy. Because that sounds like a
really safe strategy to me. If you said I’m going
to ignore quality, people think you’re crazy. So I’ve written an
entire book explaining why I ignore quality. We didn’t set out to find that. It was just
something– it’s sort of unavoidable in
the literature. You find this every
single time you read. I give a sample of the
studies in the book, there are a lot more
studies in the book. There’s another thing
that looks at admiration. So what are your positive
or negative feelings about a company, and they
rank all of the companies on the basis of how
liked or hated they are, then they examine their
performance the following year. The most hated companies
outperform the most admired companies. Morningstar gives
rankings to companies, this is an A company, A plus,
this one’s a B, this one’s a D, this one’s an E. The
Es comprehensively outperform the As. Every single time
you look at research, and it’s so counter
to everything you see everywhere
else, where everybody’s saying find the really
high quality ones. So I didn’t go looking for it. But when you look at this
sort of strategy outside of this market, you find
almost the same thing. The one exception
to that is Japan. For whatever reason,
the magic formula over the period of the
data that we examined outperformed the
earnings yield alone. Japan’s an outlier in a number
of different strategies. Momentum hasn’t worked
in Japan, whereas that’s worked in lots of other markets. Even though Japan has sort
of been gently falling– this is just an interesting
side comment– even though it’s been gently falling since
1990, value investing strategies in Japan
have worked really well, just buying the cheapest
companies on the basis of price to earnings, price to
cash flow, price to book has done something
like 20% a year. So the magic formula might have
just captured a little bit more of that performance. But yes, there’s
always a risk of data mining in this sort of stuff. I think that the acquirer’s
multiple is the way that value investors
think about investing. So if you think about
the way that you’re instructed in “Security
Analysis” or “The Intelligent Investor” or any of
Buffett’s letters, he says think about
buying the company, and it’s think about
buying the entire business. You’re not buying a share of it,
you’re buying the whole thing. And that’s exactly what the
acquirer’s multiple does. It says, this is what you pay
for the market capitalization, but don’t forget that you’ve got
preferred stock, debt, minority interests, off balance
sheet liabilities, underfunded pensions, other
things that you have to fund. But you get the benefit
of the cash that’s sitting in the bank,
all the net cash, and then you have
this discretion to spend the operating earnings
as they’re coming in on capex or paying down debt, or
various other things. So it’s looking
at the same thing that big acquirers
look at, and I often find that I’m in positions
that somebody else has been buying right behind me. It’s a Carl Icahn
or another activist, or a private equity firm. Something happens in them. Because when they
get very cheap, there’s a little
bit of instability. It’s not a situation that
should persist like that. There’s sort of inviting
external managers to come in and try to
rectify the situation. Data mining is always
a problem, but I think there are a
variety of reasons why this strategy
should work quite well, and it does work quite well
in different time periods and in different markets. AUDIENCE: Buffett talks
about holding stocks for the long term,
and one of the reasons that he likes it is because
of the tax consequences. Can you talk a little
bit about the difference between trying to find
companies that you can hold for a long period
of time versus rebalancing and using the deep
value strategy? TOBIAS CARLISLE: This
is one of the arguments for– this is what the franchise
investors are trying to do. They want to buy
that company that can sustain those very high
returns on invested capital, and ideally grow and compound
away, throwing off cash while they’re doing it so that
10 years or 20 years from now, you’re getting out
of it in dividends what you invested
in it, and it’s a much more valuable company. So if you think about the
process of going and finding one of those companies,
knowing that there’s that mean reversion in
return on invested capital, what would you rather buy? Do you want to buy one that has
a very high return on invested capital that might mean revert,
or are you better off finding one that’s very cheap,
that might be at a trough, that you can hold
for seven or eight or nine or ten years
that reverts up. I think you’ve got as
much chance of selecting. The challenge is
not to buy something that looks like it’s
high quality now, the challenge is to
buy something and hold something that is high quality. So we all want to
buy the same stocks. Maybe I haven’t made
that clear enough. We all want, I want
to buy, everybody wants to buy the really
high– everybody wants to own the really
high quality stock, the compounding machine that
has the very high returns on invested capital. But can you use historical
data to identify them? It’s not clear. But what is clear is
that buying the ones with higher returns
on invested capital don’t necessarily
turn into ones that have high returns on invested
capital subsequently. AUDIENCE: So one thing you
talk about in the book is, one of the reasons
this strategy works is because of activist investors
are kind of arbitraging into an alternate market where
companies are being traded, not because they want the
stock to trade high, but because they want to
take control of the company. And so activism is a
pretty important kind of catalyst to this strategy. What do you think about this
strategy working for companies where the incumbent management
is an owner operator, and they have motivations
that are completely independent of the minority
shareholder motivations, and not allowing the
activists to get in. So I’m thinking like
in countries in Asia, where a lot of the companies
are owner operator companies. They really have really
different motivations, and even though it could
be cheap on the acquirer’s multiple, what are
your perspectives on why would it work
in those markets when the founders have no
reason to let anyone in? TOBIAS CARLISLE: You
might need to look at what they’re doing with the
cash flows when they get them. Are they paying
them out, are they using them to expand the size
of their kingdom, are they investing them in lower
return businesses? I think that it definitely
does work in the US, because you have that tension
when somebody comes in. But I think if
you look in the US at the– and this is
something that I looked at, the number of those companies
that are actually taken over or have some sort
of activist event occur is actually
still quite small. It’s only about 1/20. And so for the other 19 it’s
simply that mean reversion in the business that
generates the returns. So when you’re looking at Asia,
if you’re looking at Japan, for example, where it’s
a fairly insular society, but there are some
sort of inroads being made by American
activists there. Even though the
period of time where there wasn’t a lot
of activist activity, there still were quite good
returns to those undervalued companies, even [INAUDIBLE]
across shareholdings. And I acknowledge now that they
have been very undervalued, and maybe they’re less so
now over the last few weeks. But it still has work there. I don’t think it’s
necessary for it to work, but there are always
metaphysical reasons outside the realm
of the backtesting that I can’t answer,
and you’re sort of going on faith a little bit. Perhaps if those
markets open up, too. They continue to open up and
they adopt more external type management, because it’s
a globalising world. They have to sell
outside their boundaries, so they need different
types of management. AUDIENCE: Thank you
so much for your talk. It was great. TOBIAS CARLISLE: Thank you. [APPLAUSE]

52 Replies to “Tobias Carlisle: “Deep Value Investing” | Talks at Google

  1. Its funny how the person posing the question at 57:55 (with all the jargon not withstanding) asks how this strategy (acquirer's multiple) would work in companies which have managements with "motivations that are completely independent of the minority shareholder motivations". He ends up using "countries in Asia" as an example. My question is – why go all across the Pacific and use Asian companies as examples when the company in whose campus you are currently sitting is not that bad an example for this scenario!!

  2. MCP < How long do you have to keep this stock for the magic formula to work, 7, 8, 9, 10 years from now? Will this go BK or become a duopoly? Just a thought, maybe a rich Googler can buy out this co. and possibly move on to all the other elements in the periodic table for world resource dominance. 

  3. Benjamin Graham – also known as The Dean of Wall Street and The Father of Value Investing – was a scholar and financial analyst who mentored legendary investors such as Warren Buffett, William J. Ruane, Irving Kahn and Walter J. Schloss.

    Warren Buffett once wrote a detailed article explaining how Graham's record of creating exceptional investors (such as Buffett himself) is unquestionable, and how Graham's principles are everlasting. The article is called "The Superinvestors of Graham-and-Doddsville".

    Buffett describes Graham's book – The Intelligent Investor – as "by far the best book about investing ever written" (in its preface).

    Graham's first recommended strategy – for casual investors – was to invest in Index stocks. 
    For more serious investors, Graham recommended three different categories of stocks – Defensive, Enterprising and NCAV – and 17 qualitative and quantitative rules for identifying them. 
    For advanced investors, Graham described various "special situations".

    The first requires almost no analysis, and is easily accomplished today with a good S&P500 Index fund.
    The last requires more than the average level of ability and experience. Such stocks are also not amenable to impartial algorithmic analysis, and require a case-specific approach.

    But Defensive, Enterprising and NCAV stocks can be reliably detected by today's data-mining software, and offer a great avenue for accurate automated analysis and profitable investment.

  4. IBMs price to book ratio is 10:1, but Warren Buffett still loves that price. Sorting stocks by their p/b ratios is virtually meaningless when trying to find value.

  5. 1929 of the 600 stocks 200 traded less than liquideded value. some of them was selling net cash backing—less than cash after paying all liabilities. $1 was selling for .50c with string attached.

  6. PB used to determine over value of under value. earing for most undervalue shares is falling years and most over value shares is increasing over the years

  7. PE, PB, PCF, P operating earing. undervalue slow growth stocks portfolio and overvalue high growth stocks portfolio. 3 types slowest growth, medium growth, highest growth and three type of valuation most undervaluation, average ,most overvalued—9 types has come– 3 are shown here—–

  8. glaumer portfolio has the highest rate of growth –earing, cashflow, book value, operating earning all are growing high.—-you have to pay higher multiple to acquire these—19.8 pE, 10.8 PCF higher to acquire very expensive.


  10. GLOBAL GROWTH ABOVE AVERAGE OR BELOW AVERAGE OR EXCEED THE AVERAGE—- IT DOES NOT INDICATE ANYTHING-–most company does not avoid this cycle very few can avoid this cycle


  12. buffet moved from liquidation method to franchise method. shes chandy it earning 2m on 8 capital and he valued 47m which was a discounted price.

  13. i am 85% gram and 15% phil fisher. fisher —-find out as much as possible about the quality of the business and ability to growth. shes candy reinvested only 70m to generate 1.35b income from 1972 to 2011.

  14. somebody help me out here…does he mean that buying companies whose earnings are on the decline will outperform the companies whose earnings are on the rise? how does he know that these companies wont just go bankrupt? is he expecting them to just "mean-revert" (which apparently has no reason whatsoever). I confused, to me it just sounds like: buy failing businesses and they will outperform because they will magically bounce back.

  15. I've come back to watch this talk over 10 times. Tobias Carlisle and Meb Faber are two of the best value investors out there right now. Google Meb to find his podcast

  16. This presentation should have started with the slide and statements of 35:53, then it would have been easier to follow. Information was interesting but the presentation was a bit fragmented and camera work didn't help. But thanks for the presentation.

  17. Thanks for this video! i've come back to watch it a couple of times as a form of recap. Just heard of Tobias' new book, i think he did an interesting interview with an Asian perspective too:

  18. Let me answer 51:50 question about why Buffett changed from Liquidator to deep value to franchise as I have read Buffett's biography a while back. He changed because he was afraid of confrontation. There was time when the angry workers of a liquidated company went to his office with pitchforks and police were involved and it traumatized him. He went to deep value until he met Munger.

    But I must say this talk blew my mind.

  19. Very good. Should one take into account unusual gain or loss when calculating the acquire multiple? Or is it just going to give less return?

  20. And the award of dumbest question of the talk goes to 38:45. He sure that Greenblat has enough money and does not care to make more money.

Leave a Reply

Your email address will not be published. Required fields are marked *