IF YOU BACKED
me into a corner with a revolver and demanded to know my
favorite financial benchmark, what would it be? I'm often
asked this question — albeit in a less threatening manner —
and my answer may come as a surprise. Some folks insist that
book value is the only way to find bargains. Others are
partial to screens based on earnings or growth rates. And what
about enterprise value?
Until recently I didn't have a favorite. Sure, there are
situations when some yardsticks are more appropriate than
others: It makes more sense to use book value than revenue to
screen for bargains among companies that are losing money. But
when you track their predictive power over time, all the
benchmarks I cited above are remarkably similar.
I'll explain why in what follows. But don't think you've
learned enough to stop reading. I'll also introduce a
benchmark based on cash flow from operations, which I call
P/CFO. There's persuasive evidence that it works better than
other screening yardsticks. I'll also tell you how to
calculate P/CFO and why it's so powerful. And I'll use this
new statistic to create my own table of bargains.
But first, I want to explain what those yardsticks in the
first paragraph have in common. You can't screen based on
absolute numbers — simply looking for stocks with lots of
earnings. That would reward big companies and penalize little
ones. That's why we create ratios such as price/earnings. All
price-based screens are contrarian in the same way. They
highlight companies that are cheap — relative to book value,
earnings or whatever. Wait long enough and these stocks
experience what statisticians call reversion to the mean. Low
prices move up, and high prices move down. Relating a
performance measure to price is more important than which
measure you select.
So I didn't have favorites — until I discovered P/CFO. It's
easy to calculate. Just find "Cash From Operating Activities"
in any company's statement of cash flows, and divide that into
market value (or divide per-share cash flow from operations
into the current share price). P/CFO is based on price. So it
includes what's best about traditional yardsticks. But the CFO
component also captures a key quality-of-earnings benchmark:
I discussed accruals in my August 2003 column, when I wrote
about University of Michigan accounting professor Richard
Sloan. The portfolio I selected based on his research has done
remarkably well: It's up nearly 50%, versus 20% for the
Wilshire 5000. But the technique I outlined for applying
Sloan's ideas was complicated, and my column generated lots of
mail — both from readers who wanted to learn more about
accruals and from those complaining that they were too
complicated. P/CFO is a slimmed-down alternative.
A quick refresher: In a landmark 1996 study, Sloan looked
at the noncash component of earnings — which accountants call
accruals. The largest component is usually depreciation, which
reduces earnings. Other components are more subjective. They
depend on how management values things like receivables and
inventory, and can push reported earnings up or down.
Sloan discovered a link between accruals and share prices.
Companies with high subjective accruals, i.e., lots of
management decisions that boost earnings, tend to perform
poorly. Companies with low subjective accruals tend to beat
the market. The potential profits from his strategy are
impressive: Hedged accruals portfolios (buying stocks with low
accruals and shorting stocks with high accruals) beat the
market by about 10% annually.
There are several theoretical explanations. One is that
investors simply don't understand accounting. But all accruals
are ultimately affected by what really happens. Suppose I'm an
overoptimistic executive and don't create an adequate bad-debt
reserve this year. I'll report more earnings, but I'll have to
take a hit in the future. If the market doesn't catch on,
however, my company will command a higher stock price than one
run by my conservative competitor. He booked an extra-fat
reserve and reported lower profits. Sure, he'll enjoy higher
earnings later when he collects more bills than he
anticipated. But now his shares are suffering — and might be a
| Screening Out
|These companies have made
conservative earnings assumptions, which may mean
|Toys R Us
|S&P 500 median
|* Prices as of 3/5/04.|
price divided by per-share cash flow from
Data: Reuters via NetScreen Pro from
Something more than misunderstanding could also be at work.
If executives want to manage earnings, accruals are the place
to start. Sloan's research tries to establish a connection
between high accruals and management skulduggery. Perhaps
companies with low accruals are better buys because the folks
who run them are more honest.
This is controversial stuff, and Sloan isn't without
critics. Some point out that companies with high accruals tend
to be growth stocks. If sales increase rapidly, bad-debt
reserves and inventory valuations expand too. In contrast,
low-accruals companies often grow slowly, if at all. So Sloan
may just have a different perspective on the classic
Portfolios based on this new measure of
stock value beat the market by 12.6%
Regardless of who's right, accruals analysis is a powerful
tool. But in his initial study, Sloan divided accruals by
total assets instead of price — one reason those calculations
in my earlier column were so tricky.
What happens if you relate accruals to price? A paper just
published in The Accounting Review provides the answer. Three
accounting professors (Hemang Desai at SMU, Shivaram Rajgopal
at the University of Washington and Mohan Venkatachalam at
Duke) realized that CFO — derived from the statement of cash
flows, as I indicated earlier — was a good proxy for Sloan's
accruals benchmark. Think of CFO as earnings adjusted for
So the professors created what amounts to a horse race
among screening variables. Using 25 years of data, they
measured profits from portfolios constructed using Sloan's
original accruals measure as well as conventional yardsticks
like price/book and price/ earnings. They also used price/cash
flow, calculated simply by adding depreciation to reported
earnings. This is the old-fashioned definition, and it's
different from P/CFO because it does not include accruals.
They also added the new entrant, P/CFO — their attempt to
combine conventional price-based screening with
accruals-adjusted earnings. Based on a variety of statistical
tests, P/CFO easily outperformed other screening variables.
One example: Twelve-month hedged P/CFO portfolios beat the
market by 12.6% annually. Comparable results for price/book
and price/ earnings portfolios were less than 8% over the same
I'd love to replicate those results. But even though
calculating P/CFO is easy, most screening tools don't include
this variable. So I cobbled together a version using NetScreen
Pro from Multex. As usual, I limited my screen to companies
with market values of over $500 million. I also tossed out
banks and insurance companies, where CFO calculations are
I wound up with 1,300 stocks and zeroed in on the bottom
10%, the companies with the lowest P/CFO ratios. Then I added
two unrelated tests: I wanted debt/ equity ratios of less than
1, and I wanted consensus analyst recommendations to have
improved steadily over the past three months — another factor
that predicts gains. My eight finalists are all cheap, based
on real earnings and not accounting adjustments.
The roster includes Lousiana-Pacific and Toys "R" Us, both
well-known companies dealing (successfully, I hope) with
challenges. Teekay is a beneficiary of the tight shipping
market, while shares in poultry-producer Pilgrim's Pride are
depressed because of worries about Asian bird flu. Aaron
Rents, meanwhile, meets anyone's definition of a growth stock.
And my three energy outfits aren't look-alikes: Marathon is an
integrated producer, Canada-based Talisman is big in natural
gas, and Spinnaker drills in the Gulf of Mexico.
Let me offer a final tip. If this column intrigues you,
check out StockDiagnostics.com. It's one of my
favorite Web sites and a great place to get a broader
perspective on accruals. The site features a statistic called
OPS (operational cash flow per share), which is identical to
P/CFO (per-share cash flow from operations) that I've
discussed here. There's a charge for the best material ($50
per month for screening), but much useful analysis is free.
Never miss another great SmartMoney
Magazine article — subscribe today!