Hello Fred, John F.:
First off, I didn't see the CNBC segment, so I don't have the first-hand
ability to comment on the veracity of the information presented therein, as
do you. But it raises the issue of: should Intel's market valuation be
penalized because of the necessity of spending billions of dollars on R&D
just in order to maintain market share? For purposes of discussion, I am
going to lump in capital spending with the R&D costs. From what I can tell
from the Intel annual report, depreciation, which is how capital spending
is eventually accounted for, shows up in the income statement in the R&D
and cost-of-sales, so I think this is fair.
In contrast to you, John F., I continue to be very very surprised that the
issue of capital expenditures finds its way on to these boards so rarely,
with comments from so few individuals. [You will find me briefly whining
about quality of earnings on post #1146.]
Let me ask you, how should one value a share of Intel? Let me be a
Graham-ite for a minute and argue that you should value it the same as
would a private owner of a business. Let me further argue that I, as a
private owner, would value Intel as a going concern by the sum of how much
cash can be extracted over its lifetime. Or rather, the present value of
the extractable-cash. Earnings can give a very misleasing picture of this
extractable cash, aka free cash flow to equity (FCF). FCF is simply
earnings plus depreciation minus capital expenditures. Rationale is that
depreciation is a non-cash charge, so doesn't affect actual cash flow,
while capital expenditures are not available as wealth for the owner.
Particularly in the case of semi companies, money spent on R&D, capital
expenditures, etc., is pretty much money down the rathole, because
processes and technologies change so fast. The point is that a company
with high expenditures and depreciation less than expenditures is not
generating very much cash to equity, so the earnings give a misleading
picture of the worth of the company. (Conversely, a company with capital
spending behind it, but still heavily depreciating, may be worth much more
than earnings suggest. I think pipelines are the classic example.)
Your argument that Intel's heavy investments in R&D, etc., are to credit
for its business success is a valid one, of course. But here is the
important point: The subtlety is that, *all other things being equal*, a
business which does not need heavy R&D and capital expenditures is more
valuable. Why? Simply because, if I don't need heavy expenditures to
maintain earning power, I can extract more cash from the business (or,
equivalently, have more free cash to plow back into the business to grow
future earnings power).
So let's look at Intel a little more closely. I am pulling numbers from
the annual report. Strictly speaking, it is not fair to do a "snapshot"
like this, but I think the numbers are typical enough to illustrate my
point.
1995:
Net income 3,566m 4.04/sh
+ Depreciation 1,371m 1.56/sh
- capital spending 3,550m 4.02/sh
------------------------
= free-cash-to-equity 1387m 1.57/sh
FCF is a lot less than reported earnings. Someday the two will catch up,
but as long as Intel is growing its investments in this capital-intensive
industry, depreciation is going to lag expenditures, and earnings will
overstate the true value of the company. And so Intel should not command a
very high P/E, because the "E" part has, to quote a famous investor, a
"mirage-like quality". Now you see it, now you don't.
To be fair to Intel, if you compare with many other semi companies, you
will see Intel is superior in terms of the FCF/E criterion. On average
over the past five years, I think something like 40-50% of earnings have
been available as FCF. I could say some nice things here about how they
are trying to "recycle" plant expenditures (see the slides from the analyst
meeting John Hull gave a pointer to) but this is already a long post. In
comparision, there are many semi companies that have over 80% of earnings
consumed by capital investment, and in fact some that have nearly negative
FCF--i.e., as a going concern, they are consuming "real" stockholder equity
(or going into debt, or both).
BUT, it is possible to find companies that are superior to Intel by this
criterion. Nearly all of Xilinx's earnings are available as FCF. Lattice,
Linear Tech. and Novellus also generate a good percentage of earnings as
FCF.
If you want a case study in the true cost of capital spending to investors,
look at Micron. This is a company in trouble because they can't generate
enough cash to remain competitive. Even during the boom years capital
spending requirements were eating something like 70% of cash flow.
John Fowler argues that consumer products companies that have to spend
heavily on advertising have "the same problem". Without further
elaboration, I don't believe this is a valid argument one way or the other.
If Brand X has to spend 90% of cash flow on advertising to maintain market
share, sure, it has a real problem, and I would not value such an
enterprise much above tangible asset value. But if same enterprise has
strong franchise presence, little capital expenditure needs, and spends 10%
of cash flow on advertising, that is a very valuable business.
Comments solicited. After all, I'm just a silly engineer with no business
sense.
--joel |
