Here's the article. (I think it's okay to post now that it's "old.")
<<<
Pooling and Fooling
A critic draws a bead on Cisco's accounting practices
By Abraham J. Briloff
Cisco Systems might well be dubbed Wall Street's New Economy Poster Child.
No better illustration of investors' vast esteem is that they've graced the
company with a cool $400 billion (it had been as high as $500 billion)
market capitalization. However, Cisco might also just as well be designated
as the New Economy Creative Accounting Exemplar. It merits that sobriquet
thanks to its ability to exploit some of the more questionable, even
dubious, accounting concepts -- all of which it should be noted are
presently enshrined in the Good Book of GAAP (or, in formal financial
parlance, Generally Accepted Accounting Principles).
Indisputably high tech and the leading supplier of networking equipment for
the Internet, Cisco Systems describes itself as providing networking
solutions that connect computing devices and computer networks, allowing
information to be accessed or transferred without regard to differences in
time, place or type of computer system. The leading supplier of networking
equipment for the Internet, it boasts a truly global reach, selling its
products in over 100 countries.
But it is less its vaunted technological prowess than Cisco's inordinate
addiction to arguable notions of accounting for business combinations, as
well as for stock options, and its occasionally obscure disclosure standards
that especially intrigue me, as a critical observer of corporate accounting
and accountability.
Cisco's accountings for its fiscal years ended July 1999 and 2000 furnish
vivid demonstration of the causes of my concern. Let us start with
pooling-of-interest accounting. Just to refresh your memory, under the
pooling method of accounting for a business combination, if Company A
acquires Company B paying, say, $100 million in stock, it would show as its
cost a mere $10 million, assuming that was the amount listed on Company B's
books as its shareholders' equity.
In consequence, $90 million of costs actually incurred by A will never pass
through A's income statements. Company A thus will be able to realize $90
million of revenues derived from the acquired properties without those
revenues being burdened with even $1 of cost. And, a safe bet is, A's
earnings will be correspondingly engorged.
Nor does it matter one whit whether that $90 million cost suppression is
related to real estate, plant and equipment, inventories, intellectual
properties (copyrights, patents or other intangibles) or a pool of Nobel
Laureates. The sole aim of the exercise is to add to the glory of A's bottom
line.
Pooling, make no mistake about it, is bad stuff. Three decades after
Barron's first attacked the practice in an article aptly entitled "Dirty
Pooling," the Financial Accounting Standards Board (FASB) promulgated its
Business Combinations Exposure Draft, which would effectively outlaw
pooling. Alas, egged on by legions of lobbyists, both houses of Congress
held extensive hearings this past spring on the pooling controversy and, no
surprise, most of the fire was directed against the FASB.
The pro-pooling chorus sang the virtues of the practice. Their theme was
that the New Economy vitally depends on pooling accounting and its end would
put our unprecedented prosperity in dire jeopardy. Of course, if indeed that
absurd claim were true, then our prosperity rests on quicksand.
To see such outlandish twaddle for what it really is, our chosen
representatives need look no further than Cisco. In the company's fiscal
year ended July 31, 1999, it made three acquisitions, all of which were
accounted for as poolings of interest. Of the three, Cisco deemed the
historical operations of two not material to the company's consolidated
operations on either an individual or aggregate basis, a status that
relieved Cisco of the need to restate its prior period results to reflect
the revenues and deficit of this "not material" pair.
There's no blinking the fact, though, that for these supposedly
insignificant acquisitions, Cisco parted with 16 million shares worth
roughly $400 million. For that $400 million, Cisco booked a cost of $45
million, net of a $70 million charge to retained earnings, implying that
throughout their pre-merger life the two had accumulated losses of $70
million.
The really big deal in fiscal '99 was the takeover in June of GeoTel
Communications. For the latter, Cisco forked over 68 million shares, worth
some $2 billion. How much of this $2 billion of cost found its way on to
Cisco's books? To find out, one has to burrow into the company's financials,
which are not a model of transparency.
The rules require that prior-period statements must be restated to reflect
any "material" acquisitions for which pooling was used. By juxtaposing the
fiscal '98 balance sheet in that year's annual with the fiscal '98 balance
sheet in the fiscal '99 annual, we can come up with a reasonable answer to
the question posed above.
In a nutshell, the $2 billion cost of the acquisition surfaced in Cisco's
accounts as a mere $41 million! A zero addition to retained earnings clearly
implies that GeoTel was bereft of earnings. So, all due thanks to the
legerdemain of accounting, Cisco had acquired $2 billion of value capable of
being insinuated into its bottom line, with hardly a penny of related cost.
Cisco grew far more audacious in fiscal 2000, ended July, snapping up no
fewer than 12 companies in exchange for stock worth a total of $16 billion.
Five of the dozen were deemed immaterial; hence, they were not included in
the pooling restatement process. Cisco paid $1.2 billion for the "immaterial
five," a cost that showed up in Cisco's books as a mere $1 million (of the
purchase price, only $75 million went into the company's capital stock
account, offset by a $74 million deduction of retained earnings).
Table A lists the seven pooling acquisitions that did require restatement.
By my reckoning, the cost of those seven, for which Cisco paid an aggregate
of over $14.8 billion, was booked at only $133 million ($207 million was
added to the capital stock account, offset by a $74 million deduction from
retained earnings, representing the accumulated losses of the seven acquired
companies).
Cerent merits a harder look and not only because of the megabucks involved
in the takeover. In less than two years it enjoyed a magical transformation
from sickly worm into glorious butterfly.
A form S-1 was filed with the SEC in July 1999, as a prelude to a Cerent
IPO. The underwriting, however, was aborted with the announcement of the
Cisco merger. Happily, though, we have the prospectus, and it furnishes some
remarkable insights. Incorporated on January 27, 1997, as Fiberlane
Communications, it subsequently changed its name to the more beguiling
Cerent. Its business is the development and delivery of a multi-service
optical transport platform, designed to reduce network operating costs and
boost the efficiency of bandwidth delivery within transport networks.
Sales of its sole product, the Cerent 454, began in December 1998. Targeted
markets include emerging interexchange, competitive local exchange and
independent carriers, and cable companies.
On page 54 of the prospectus, we learn that Cerent's principal stockholder
was venture capitalist Kleiner Perkins Caufield & Byers, with a 30.8%
beneficially owned stake. But we also learn that Cisco itself owned 9% of
the fledgling. Hence a plausible inference is that the $6.9 billion paid by
Cisco was for the remaining 91%, so that Cerent actually was valued at
around $7.5 billion.
Before entering the Cisco fold, Cerent had a brief but interesting history.
In its first year of operations, it generated no revenues and suffered a
loss of $7.9 million. In 1998, it rang up revenues of $220,000, on which it
lost $22.5 million. In its last phase as an independent company, the
five-plus months ended June 1999, Cerent had revenues of not quite $10
million and a loss of over $29 million.
From mammoth to minuscule
And it was for this company, whose rapid growth in revenues was accompanied
by more and more red ink, that Cisco shelled out $6.9 billion worth of
stock. Which demonstrates the great virtues -- if any demonstration be
needed -- of having a very richly priced stock to use as acquisitive
currency and of pooling, which reduces that mammoth sum to minuscule
proportions when it comes to accounting for costs.
In June of this year, Cisco performed a Cerent encore, acquiring ArrowPoint
Communications for $5.7 billion in stock. ArrowPoint, unlike Cerent, did get
off an IPO in an early spring, selling 5.75 million shares at $34 each,
endowing it with a market cap of $1.5 billion, or some $4.2 billion less
than Cisco paid for the company barely two months later.
Of the $5.7 billion cost of the acquisition, it booked about $40 million.
Again, the remaining billions of the cost simply vanished, so far as Cisco's
books are concerned.
By my reckoning, in the two fiscal years ended July 2000, Cisco has
suppressed a grand total of $18.2 billion of costs by using pooling in
accounting for its acquisitions. Even in today's wondrous financial world,
when billions are commonplace, $18 billion of costs not taken is
mindboggling. Manifestly, the handmaiden of pooling is fooling.
But pooling is not the only accounting device that Wall Street's favorite
company uses to enhance its operating results. Another, equally egregious,
involves stock options and the way Cisco accounts for them.
How to account for options has been the subject of agonizing reappraisals in
board rooms, among scholars in academe, at the FASB and even in Congress.
The crucial questions are: Can options be valued and, if so, should they be
entered into a company's accounts and when? Further, if they were to be
recorded, should it be as a cost of doing business or merely a capital
transaction?
The correct answer to the first question is yes, they should be entered into
accounts and, to the second, as a cost of doing business. Let me elaborate,
using Cisco as a prime exhibit.
In the statement of shareholders' equity in Cisco's 1999 10K, there's an
entry described as "tax benefit from employee stock option plans." This item
added $837 million to the capital stock and additional paid-in capital and
shareholders' equity columns. The implications of that apparently innocuous
entry are, in fact, far from innocuous.
When the employee exercises his or her options, the resultant gain is deemed
to be compensatory income, i.e., salaries or wages to the employee and,
accordingly, subject to tax. Correspondingly -- and this is the critical
side of the relationship -- Cisco was presumed to have paid wages or
salaries equal to the income earned by the employee and thus the company is
entitled to a tax deduction (all spelled out in Section 83 of the Internal
Revenue Code).
Now then, that $837 million tax benefit means that at an assumed 33% tax
rate, the related deduction for Cisco's tax return would have been $2.5
billion in the fiscal year ended July 31, 1999. If $2.5 billion is a cost
for tax purposes, logic dictates that it is also a cost for determining
Cisco's operating results.
More specifically, for fiscal '99, Cisco's pretax income should be reduced
by $2.5 billion; its income tax cost would be cut by $837 million. Net
income, accordingly, would be slashed by a whopping $1.6 billion, or by
nearly 80% from the reported figure of $2.02 billion, to $423 million.
The impact of options on Cisco's fiscal 2000 results was even more
pronounced and even more stunning. According to the 10-K (footnote 11), the
tax benefit derived for the exercise of options amounted to $3.077 billion.
At the assumed 33% tax rate, that amount translates into over $9 billion of
salaries.
Especially noteworthy is that fully $2.147 billion of that $3 billion-plus
was generated during the final fiscal quarter, the May-July time span.
Clearly, as Cisco's share price dropped, options holders made a mad dash to
cash in their chips, in the process triggering roughly $6 billion of imputed
salaries and wages.
How should that humongous full-year figure of $9 billion of imputed wages
and salaries be factored into fiscal 2000 operating results? Let's assume
only $310 million of tax benefit is "normal" for the fourth quarter (the
average of the first three quarters of the year) rather than the actual
total, hugely swollen by the period's extraordinary stampede to sell. That
would make the "normalized" tax benefit from exercised options for fiscal
2000 a not exactly modest $1.246 billion, implying an addition to the year's
operating expenses of $3.7 billion and an after-tax reduction of the bottom
line by $2.5 billion.
Put another way, if Cisco had treated the exercise of options as they should
be treated -- that is, as a charge to income -- the company would have
reported not the $2.1 billion in earnings it did report, but a loss of $363
million (excluding $531 million of net gains on minority interests).
My restatement of Cisco's income to give due allowance to the cost of
options is not a capricious exercise. For it's squarely in accord with
underlying accounting precepts, especially Statement of Financial Accounting
Standards No. 5, "Accounting for Contingencies." The statement, promulgated
over a quarter-century ago, holds that a loss should be accrued as a charge
to income when "it is probable ... a liability had been incurred" and "the
amount of loss can be reasonably estimated." In other words, the charge to
income from stock options kicks in when those options are exercised.
The result of Cisco's accounting aggressiveness, both in its energetic use
of pooling and its treatment of exercised options, then, has been to
enormously inflate reported earnings. And enormously inflated earnings have
played no small role in elevating the company -- and its stock -- to the
pinnacle of investor esteem.
----------------------------------------------------------------------------
----
ABRAHAM J. BRILOFF, a CPA and frequent contributor to Barron's over the past 30 years, is a distinguished professor emeritus at Baruch College in New York. |
