To kezer: Is the Market Overvalued? -- EVA, IBM, and ROE.
Graham and Doddsville Revisited -- "The Intelligent Investor in the 21st Century" (4/16/98)
*********
"The underlying principles of sound investment should not alter from
decade to decade, but the application of these principles must be
adapted to significant changes in the financial mechanisms and
climate." (Benjamin Graham)
*********
kezer writes:
<< I am intrigued by the analytical process known as Economic Value
Added (EVA), which many companies (GE, KO, CPB, etc) use in their
internal planning and capital budgeting process ... are you aware of
Buffett ever referencing it in any of his orations. >>
Not directly, to my knowledge. But, see Buffett's quoting of John Maynard Keynes, below.
<< ... in its most simple form, EVA attempts to remove accounting
distortions and measure cash return on cash invested including a
charge for use of capital (both equity and debt). EVA removes items
such as amortization of goodwill and does not use full expensing of
R&D in determining cash earnings. I have found this method to be a
very sound way of comparing the results (backward looking) of mature
companies in industries with fairly stable demand. Your thoughts
would be appreciated. >>
There is a nice summary of the EVA concept at:
peterkeen.com
It is ironic that, as you accurately put it, "EVA attempts to remove
accounting distortions", since the whole purpose of accrual
accounting is to remove the distortions of cash accounting.
But, as it turns out, accrual accounting creates some distortions of
it own. A discussion of some of these distortions follows.
Accrual Accounting For The Cost Of Tangible Assets
Suppose a new business builds a single factory in its first year, and
that this factory will generate revenues for the next 30 years. For
purposes of simplification, also assume that the rules of the
financial accounting standards board (FASB), which promulgates GAAP,
would permit assigning an economic life of 30 years to the factory.
On a cash accounting basis, the entire cost of the factory would be
deducted from the company's income statement in the first year. But,
the subsequent 29 years' revenues would not be offset by any portion
of the cost of building the factory. Therefore, in the first year
there would be a large loss because of the cash expense of the
factory, and in the subsequent years there would be exaggerated
profits because the revenues in those years would not be burdened by
any of the cost of building the factory that produced these revenues.
In order to smooth out this distorted picture of periodic profit
performance, accrual accounting, if using the straight line method,
assigns 1/30 the cost of the factory to each of the 30 years in order
to "match revenues with expenses". Likewise, 1/30 the cost of the
factory would be deducted in each year from the long term assets entry
on the company's balance sheet, thereby decreasing shareholder equity
by a like amount annually.
If, subsequently, another company purchases the factory, it is a
relatively straightforward matter to continue deducting annually 1/30
of the factory's cost to match revenues with expenses in arriving at
the profit earned in a given year from the acquiring company's
investment in the factory. The same annual deduction would be made on
the acquiring company's balance sheet.
Accounting For The Cost Of Intangible Assets
But, the picture is much less clear when accounting for the
acquisition of "intangible" assets. And, in an economy where
intangibles like reputation and know-how increasingly determine sales
and profitability, the issue of how to measure the cost of this
intangible capital is increasingly relevant.
The Value Of One's Good Name
For example, when Chemical Bank purchased Chase, Chemical renamed its
own operations "Chase", because of the aura surrounding the Chase
name. Should the cost of buying the Chase name and franchise, over
and above its tangible assets, be deducted from the income statement
and balance sheet in annual increments of 1/40 over the next 40 years?
Under FASB rules, this is how "goodwill" (the price over and above
tangible asset value) is accounted for.
A factory obviously wears out in a few decades and must be replaced
with a new expenditure of capital to keep profits coming in. But,
will the name "Chase" have exhausted its value in the next 40 years --
and have to be replaced by the purchase of a new name? Or, will the
name Chase be worth more than ever in 40 years time? Our position is
that this is an empirical issue, i.e., only time will tell.
Therefore, the value of goodwill should be assumed to be constant,
unless and until future events establish otherwise.
Accounting For The Acquisition Of In-Process Research And Development
The subject becomes even more complicated, and more relevant, in
accounting for acquisitions of hi-tech companies. A controversy
surrounding one of our Dow Value Portfolio components, IBM, comes to
mind. An eminent figure in the field of accounting, Abraham Briloff,
has publicly taken IBM to task for the way it accounted for the value
of an intangible asset, "in-process research and development", when it
acquired Lotus for $3.2 billion in July 1995. (See: Barron's,
12/23/96, p. 17.)
Software companies do not have a lot in the way of tangible assets.
Most of their true economic value cannot be accounted for merely by
adding up the value of their work stations and the real estate they
occupy. In the case of Lotus, their tangible assets totalled only
$325 million. The difference between the $3.2 billion purchase price
and the $325 million in tangible assets could have been carried on
IBM's books as "goodwill". Under accounting rules, 1/40 of the cost
of this goodwill would be a charge against IBM's income and equity in
each of the next 40 years.
Apparently to avoid this result, an appraisal firm hired by IBM, the
American Appraisal Association, calculated the present value of
profits it was estimated would be derived from the software Lotus was
working on at the time it was acquired. This calculation came to $2.8
billion -- virtually the entirety of the intangible asset amount (and
about 90% of the total purchase price). This present value of $2.8
billion in future software profits, from work then in progress at
Lotus, was deemed "purchased in-process research and development".
Is R&D Like Office Supplies?
Under GAAP, in the normal course of business activity a company
deducts ongoing R&D expenses from current revenues as soon as the
money on R&D is expended, rather than waiting to match these expenses
with the revenues they generate in the future, if any.
Unlike expenditures on a factory, inventory, or advertising, it is a
matter of pure speculation as to whether or not a given expenditure on
R&D will generate revenues at all. And, it is a matter of further
speculation as to which future revenues would have to be matched with
what current R&D expenses. Also, unlike building factories, R&D
expenditures are a continuing cash expense. Therefore, there is not
the same danger as in the factory example that failure to match
expenditures on R&D with their corresponding revenues will unduly
distort the profit picture from one accounting period to the next.
To be consistent, the FASB rules require that R&D purchased
from another company be treated in the same way as any other R&D
expense. That is, the expense of purchased R&D is immediately
deducted from revenues in arriving at net income, rather than waiting
to see what, if any, revenues this R&D expense will generate in the
future.
Note that with a company's ongoing R&D expense, there is no
corresponding booking of an intangible asset on the balance sheet.
(This, in our view, is the source of the problem). Instead, R&D is
treated like salaries, office supplies, the phone bill, etc.:
Something whose value has been consumed at the time it was expended,
rather than an accreted asset to appear on the balance sheet.
Therefore, to be consistent with the FASB rule that purchased R&D be
treated the same way as would an ongoing R&D expense, purchased R&D
cannot be placed on the acquiring company's books as an asset.
The Controversy
Thus arises an anomaly with which, as Briloff points out, the FASB
itself is uncomfortable. Namely, purchased R&D gets booked on the
balance sheet as a purchased asset, but removed from the balance sheet
because it is R&D, simultaneously. The net effect is, in
Briloff's characterization of an FASB task force's comment, "if the
amount must be charged off immediately, then it has no assignable
value." In other words, for the FASB's rules on the expensing of R&D
to be applied consistently, IBM must, in effect, declare that what it
paid for Lotus' in-process R&D has no asset value.
[This begs the issue, which Briloff implicitly raises, of whether or
not the $2.8 billion should have been characterized as "purchased
in-process research and development" in the first place. To second
guess that call is beyond the scope of this article. The real issue,
in our view, is not how intangible assets should be
characterized or over what period intangible assets should be
written off, but rather whether they should be written off at
all.]
After consideration of taxes, the net present value of this in-process
R&D was put at $1.8 billion. IBM deducted this $1.8 billion of
purchased R&D from its income statement, not as an operating expense,
but as a one time charge. (After all, how often does IBM purchase R&D
from the outside?) As an R&D expense, it was simultaneously deducted
from the balance sheet, against the $3.2 billion of acquired assets,
tangible and intangible. Poof -- the $1.8 billion was gone.
Understandably, Briloff is very unhappy about this. As already noted,
the FASB is also concerned. And, as Briloff grudgingly concedes, IBM
made an effort, in the notes to its financial statements, to provide a
clearer picture of what had happened to this $1.8 billion.
Briloff Turns The Table
Then Briloff proceeds to hoist IBM and the American Appraisal
Association on their own petard. Using their present value appraisal
of $2.8 billion (pre-tax), and assuming a future payout over a 5-year
period and a discount rate of 20% per year, Briloff calculates that
earnings in the subsequent periods should be reduced by about 10% to
account for the cost of generating those revenues that came from the
in-process R&D purchased from Lotus.
One can quibble with Briloff about how long the payout will be, in
what fiscal period it will begin, and his chosen discount rate of 20%
per year. But, one cannot question his logic: The FASB rule that R&D
is an expense of current operations (rather than an investment
accruing to the balance sheet) is based upon the rationale that the
future revenues from R&D are too speculative to reliably estimate.
But, the AAA justified its appraisal upon, and IBM's accountancy firm,
Price Waterhouse, signed off on, just such a calculation of the future
benefit of Lotus' R&D then in progress.
Therefore, if purchased in-process R&D is a depleting asset
with an identifiable stream of future revenues, then the portion that
was depleted in producing revenues for a given fiscal period
must be deducted from those revenues to accurately reflect the
profits earned in that period.
Is R&D Like Inventory -- Or Like The Land The Inventory Is Stored
Upon?
Briloff's position is that this is akin to buying another company's
inventory, selling it off over a 5-year period, and recording the
revenues therefrom as profit, without offsetting those revenues with
the cost of purchasing the inventory.
But, is it? Is R&D a depletable resource like inventory or oil under
the ground? Or, is it a permanent resource like land with a good
location? Does science in its advancing discoveries discard as used
up the knowledge of the past? Or, does it build upon this knowledge?
Admittedly, the products that Lotus was working on when acquired by
IBM will be obsolete in a few years. But, will the knowledge gained
in the process be depleted of all economic value -- or will it be the
foundation for future profitable products?
Is Organizational Value A Depleting Asset?
And, as they say, actions speak louder than words. IBM spent $3.2
billion, in cash, to purchase Lotus. Did IBM really believe in
its heart of hearts that it was paying twice the then market value of
Lotus for assets 90% of which would be depleted of their economic
value in 5 years' time? Or was IBM calculating that these intangibles
would be increasing in economic value, instead of decreasing, with the
passage of time?
In fact, IBM had already spent several years and several billions of
dollars to develop an office suite for PC's. But IBM hadn't gotten
anywhere on its own. Basically, the tangible assets required for
software development are, as noted, work stations and the real estate
they occupy. The necessary intangible assets include teamwork, a
shared vision, a certain "culture", and an ability to meet deadlines,
i.e., organizational value.
Frame Of Reference
Perhaps IBM's financial statements are being examined through the
wrong end of the microscope. Lotus' spending on R&D prior to its
acquisition had already been deducted from its income statement and
balance sheet in the periods in which the spending occurred. If the
economic value of Lotus's R&D spending were depreciating, rather than
appreciating, why did a proud company like IBM feel the business
necessity to pay a 900% cash premium to tangible value to acquire
Lotus? It's surely not because IBM CEO Louis Gerstner doesn't know
the value of a buck.
Something Is Wrong
Nevertheless, Briloff is to be commended for using his estimable
reputation, knowledge, and intellect in calling to the public's
attention that something is being seriously distorted when
accounting methods developed to account for bricks and mortar are
employed to measure the value of know how and its profitable
employment. (Briloff recently brought the public's attention to what
he considers a secret reserve fund from Disney's acquisition of
ABC/Cap Cities that had been used to inflate Disney's reported
earnings in recent years.)
Are Reported Earnings Too High; Or, Is Reported Book Value Too Low?
We agree that GAAP's accounting for acquisitions and R&D obscures the
true picture. But, we think the picture may be distorted in a sense
opposite to what some have argued. Rather than artificially inflating
earnings, we think it likely that GAAP is artificially deflating book
value. If we are correct, it would mean that P/E ratios are not
inflated (at least, not from the way acquisitions are accounted for),
but instead, that return on equity (ROE = earnings/book value)
is inflated.
IBM -- Where's The Book?
The example of IBM presents an extreme instance of this. Based upon a
rough calculation using data from Value Line, it appears that during
the 10 years from 1987 to 1996, IBM generated in the neighborhood of
$100 billion of free cash flow. After making allowance for dividends
paid, share buybacks, a presumably very generous options incentive
package for key management, and assuming that all "special charges"
were actual cash expenses (many of which certainly were not), there
still should remain about $30 billion of undistributed cash -- an
average of close to $3.00 per share per year.
Yet, IBM's book value did not rise by anything like this amount. In
fact, IBM's stated book value actually fell during this period,
by $17 billion. The increase in long term debt of $6 billion (at
falling rates of interest) during this period was more than offset by
a $10 billion shrinkage in working capital. In other words, the
unaccounted for cash cannot be explained by increased debt or bloated
inventory and receivables.
Perhaps, our methods and assumptions have oversimplified the matter.
Perhaps a big chunk of FCF is "hiding" in various places on the
balance sheet under categories like "reserve against contingencies",
available for smoothing out earnings when business is slow (which is
the nub of Briloff's case against Disney's bookkeeping).
Nevertheless, given the magnitude of share buybacks in recent years,
combined with the offsetting effects of shrinking working capital and
shrinking long-term debt, IBM's cash earnings of recent years appear
legitimate, but its book value does not. Thus we find more credible
IBM's 7.7% profit margin (profits as a percentage of sales) than their
ROE (profit as a percentage of book value) of 29.0%. (Source: Value
Line.)
It's Not Just IBM
This issue is not merely restricted to IBM or Disney. We think it
goes to the heart of the matter of whether the Market's current prices
are sustainable. Much has been made of the fact that ROE is at a
level for the general Market that has never been sustained in the
past. Our view is that real ROE's are not that far out of line, but
the artificial suppression of book values makes it appear that they
are.
The organizational value of companies in general (not just high tech
companies) continues to grow as companies learn to expand output with
fewer workers, leaner inventories, and less borrowing. Yet, this
growing organizational value does not appear on a company's balance
sheet, thus exaggerating the company's reported ROE.
And, a significant portion of the profits companies are now earning is
coming from an acquisition spree, at home and abroad, of mammoth
proportions. But, the intangible organizational value of these
acquisitions is being deducted from book value, whether in a single
stroke as with IBM, or amortized over a number of years, as is the
more usual case.
Thus, a corporate buyer recognizes that an acquisition's asset values
have grown, but are not on the balance sheet. It pays up for them.
Then GAAP requires that these asset values be removed from the
acquiring company's balance sheet, at the same time that economies of
scale are increasing earnings. Thus is ROE pushed up to levels that
appear unsustainable.
Today's ROE's -- Too Much To Be True?
According to Value Line data, at the peak of the 1987 Market (2 months
before the October Crash), the price to book ratio on the Dow was 2.70
and the (accrual basis) ROE was heading toward a year end figure of
15.9%. The Dow's ROE would peak at 21.2% the following year, 1988,
then decline to 19.4% in 1989, the last full year before the
recession.
Now, as we begin the 8th year of an economic upswing, a recent issue
of Barron's (4/6/98, p. MW95) informs readers that the price to book
ratio on the Dow is an astounding 6.36, and that the (accrual basis)
1997 return on this book value is an almost as astounding 27.4%. The
1996 figure puts the Dow's ROE at 27.2% for that year. The Value Line
data gives the Dow a 1995 ROE of 25.7% and the figures for 1994 and
1993 are 20.1% and 17.8%, respectively.
The Dow has never shown such sustained levels of profitability.
In fact, heretofore there have only been 3 times since 1920 with even
a single year's ROE of 20% or more ROE: 1925, 1929, and the above
referenced 1988. Now we are to believe that there have been 4 such
years in a row, with each of the past 3 years, successively, an all
time record.
We are convinced that the American economy has never been more stable
and efficient, and therefore that higher than average Market
valuations are justified by the underlying economic
fundamentals. But, at the same time, we do not believe any
economy has ever been as stable and efficient as these figures
would have us believe.
In other words, we find the reported numbers on book value, and
therefore ROE, to be unbelievable. Yet, return on capital is the name
of the game of Capitalism. Thus, the motivation for restoring some
form of cash accounting to escape the distortions that have grown up
around accrual-based accounting.
Cash Accounting: Disaccrual World
EVA is one of the systems used to accomplish this "disaccrual" of a
company's ROE figures. It is a little too complicated for our liking.
Benjamin Graham rejected methods that gave the appearance of
exactitude to something that, at its core, cannot be precisely
measured: The future. He assumed that measurements would be
imprecise. Hence, he required that there be a wide "margin of safety"
in his investment decisions.
Sales-To-Capital-Spending And Sales-To-Debt: Proxies For ROE
For that reason, we undertook an exercise, a part of which appears in
GADR's Silicon Investor posting #1
(http://www4.techstocks.com/~wsapi/investor/reply-3556473).
We looked at overall sales, capital spending, and long-term debt for
the 25 Dow stocks that are neither retailers nor financial firms
(i.e., the Dow's 25 industrial industrials).
Sales can be viewed as a proxy for earnings. In fact, a number of
Value Investors use price-to-sales instead of price-to-earnings as a
valuation gauge, in part because sales are a less volatile figure in
the short term, and in part because they are less subject to
distortion in the long term. (See, for example, James P.
O'Shaughnessy, who has concluded, "Price-to-sales ratio is the best
value ratio to use for buying market-beating stocks." see:
oshaughnessyfunds.com
Capital spending is, of course, a direct measure of the cost of
capital in the current period.
And, long-term debt is a proxy for the total cost of capital,
aggregating that portion of assets that were "borrowed" from Mr.
Market in the past. Debt, therefore, represents an ongoing liability
against shareholders' earnings, and therefore an ongoing cost of
capital.
Thus, by comparing sales to capital spending we can get a measure of
the productivity, and indirectly the profitability, of current capital
expenses. And, by comparing sales to debt, we have a gauge of the
productivity and profitability of the total cost of capital employed.
The periods from 1987 to 1989 and from 1995 to 1997 have been periods
of peak capital spending for their respective decades. Whatever the
accuracy of absolute measures, we, like Alan Greenspan, think the
trends are relatively reliable. A 3-year average smoothes out
temporary volatility to provide a clearer picture of the overall
trend, as does the use of 25 large cap companies across a spectrum of
industries.
The 3-year average of revenues per dollar of capital spending has
risen over 23% since the peak years of the 1980's. This is a rough
measure of the gross return likely to be derived from each
additional dollar of future capital spending.
And, the 3-year average of revenues per dollar of debt has risen
almost 13% since the peak years of the 1980's. In other words, that
portion of our Dow 25's assets that has been borrowed from the capital
markets in the past is also generating a higher gross return. And,
because of the decline in inflation-adjusted interest rates, there is
perhaps 20% less interest burden on revenues per year for each dollar
of these borrowed assets.
Summary
We feel that the method currently used to account for the acquisition
of intangible assets greatly understates their real economic value.
Therefore, reported levels of ROE are significantly overstated.
EVA, among others, is a method some analysts are now using to more
accurately account for ROE. Our own, indirect, methods strongly
suggest that though ROE is not as high as reported, it is
measurably higher than its peak during the last economic recovery.
Thus, current return on capital is high enough to justify record
valuations, but not necessarily so high as to be unsustainable.
*********
Graham and Doddsville Revisited
Editor: Reynolds Russell, Registered Investment Advisor
Web Site Development/Design: ariana <brla@earthlink.net>
Consultants: Axel Gunderson, Wayne Crimi, Bernard F. O'Rourke,
Allen Wolovsky
In addition to editing *GADR*, Reynolds Russell offers investment
advisory services. His goal is to provide total returns in excess of
those produced by the S&P 500.
His investment strategy applies the principles of Value Investing
established by Benjamin Graham to the circumstances of today's economy
and securities markets.
For further information, reply via e-mail to: gadr@nyct.net
*********
For a free e-mail subscription to GADR, reply to: gadr@nyct.net
In the subject header, type: SUBSCRIBE.
*********
"There are no sure and easy paths to riches in Wall Street
or anywhere else." (Benjamin Graham)
(C) Reynolds Russell 1998. |
