United States. Congress. Senate. Committee on the.

The Industrial reorganization act. Hearings, Ninety-third Congress, first session [-Ninety-fourth Congress, first session], on S. 1167 (Volume pt. 7) online

. (page 85 of 140)
Online LibraryUnited States. Congress. Senate. Committee on theThe Industrial reorganization act. Hearings, Ninety-third Congress, first session [-Ninety-fourth Congress, first session], on S. 1167 (Volume pt. 7) → online text (page 85 of 140)
Font size
QR-code for this ebook


before us today^ — is that the industry encompasses the production
and the initial marketing of general purpose electronic digital com-
puters and/or computer systems.

By computer systems, I mean the integrated combination of the
hardware — or the machinery — the programs — the software, that is —
and the supporting services that are essential to users" understanding
of how to make the hardware and the software together do effective
work for them.

This definition of the industry that I am using today excludes
those firms that are so-called software houses which are in the business
of writing computer programs, but do not manufacture or sell equip-
ment. It excludes leasing companies, service bureaus, other entities
that purchase computers fr-om the computer manufacturers and then
n^sell them — breaking bulk in a way.

These firms buy computers, and then they lease either the computer
services for a period of years, or the use of the computers for an hour



5596

at a time, or what-have-you. The definition also excludes companies
that are in the business of making primarily communications equip-
ment or communications systems that may, incidentally, have comput-
mg equipment as parts of them. It excludes the manufacture and
marketing of military and other kinds of special purpose computing
equipment.

I make these comments not because I think that the definition of the
industry is a major issue that we should be concerned with, but simply
to inform the committee of the subject matter that I am talking about
when I comment on what I see to be the computer industry.

Briefly, I now comment about the importance of the computer in-
dustry as it's defined in this perhaps somewhat narrow way. The com-
puter industry still accounts for approximately 1 percent of all
manufacturing activity in the United States, using the so-called
value-added measure which is generally accepted as the best way
of measuring the relative importance of industries in terms of the
quantity of productive activity occurring in them. That, in itself, is a
large amount of economic activity, but I think it is an underestimate
of what the computer industry means to the U.S. economy and indeed
to the United States.

Mr. Chumbris. Mr. Miller, would you take a question at this point.
On table No. 1 of your prepared statement, you show that in 1958,
four-tenths of 1 percent of value added is in manufacturing.

Mr. Miller. Excuse me. Is that 0.3 percent ?

Mr. Chumbris. Yes. And in 1963, it was 0.7 percent. So there was
a gain there over other manufacturers. Is that right?

Mr. Miller. Right.

Mr. Chumbris. Then from 1963 to 1967, it remained steady. So it
moved exactly with the rest of the manufacturers.

Mr. Miller. Yes.

Mr. Chumbris. Is there any point that would be important to us :
Why from 1963 to 1967 it stayed even; then from 1967 to 1972 it
jumped another three-tenths of 1 percent ?

Mr. Miller. I think actually — I'm not positive, but my recollec-
tion is— that between 1963 and 1967 the Standard Industrial Clas-
sification Manual was changed : and the definition of the four-digit
industry, which is the way the Bureau of the Census collects this in-
formation and compiles it on value-added was changed so that ac-
counting machines and cash registers and other nonelectronic equip-
ment were removed from the industry that included also computer
manufacturing. And I believe that if that is correct, that change in
the definition of the industry would account for some of the reasons
why there was apparently no increase.

Certainly these numbers do not give you a very detailed or careful
explanation of the growth of the computer industry and of the output
of computers because the four-digit industry, wliich is the level on
which these statistics are calculated, is somewhat too gross to give
you a careful measure of Avliat is going on. I presented them merely
to arive you a rough idea of the facts on the computer industry.

I believe I was commenting about the imj^ortance of the computers;
not because of the economic activitv involved in the manufacture
of them or of the distribution marketing, and what-have-you, but
l)pcause of the way they are used. They are one of the products that



5597

are used pervasivel}^ in business and in Government in this country.
They are used in practically all industries.

And they are used in those, I think, most important functions of
business and Government Jiowadays, which is the making of decisions
and keeping of records.

That seems to be what our society is about, and I think it is a very
serious point. We think of businesses as making products, but what
really makes our society progress is the decisions that are made in
l)usinesses and in Government. Computers are so intimately associated
witli these processes that tliey are extremely important to the way the
country ^^•orks.

Tlie next point is that, as I see it, IB]M dominates the computer
industry. The principal point about dominance is that IBjNI is the
producer and seller of su])stantially more than half of the value of
computer systems produced and sold in this country. That is essentially
wliat I mean by dominance.

Tlie best single way I know of for measuring the shares in the
marker for computers is by k:)oking at the total outstanding stock of
geiiei-al purpose computer systems. Anotlier word for that total out-
standing stock of computer systems is the installed base of equipment.
We look at the figures to see who manufactured what parts of it and
we calculate market shares on that basis.

Those iigures are shown in my table 2 of the prepared statement.
The point 1 would make briefly is that the data in this table indicate
with very few exceptions a marlcet share of a])out TO percent for IBM,
witli the remaining oO percent fi-agmented. I think the exceptions are
tlie early years of the industry, and the company is Univac; no other
(■om}::any has e\en had a 10-percent share in the market — that is, in
the stock of installed computer systems.

I do not really think it pays to try to look at the detailed figures in
this table ; to look at 1 percent here or 1 percent there. These numbers
are compiled from publicly a^■ailable estimates about how many com-
initers of the different makes or models are in use and what the prices
are for this equipment. I am certainly not preapred to say tliat tliose
iiuml)ers are very accurate. On the other hand I would be surprised
if, when these published figures show that IBM's share is estimated at
70 i^ercent. the real or the true value based on the same concept is in
the iuMghl)oi'liood of 30 pei'cent. The numbers are not that bad. They
ai'c ballpark figures; and the point is that the ballpark is all we really
need foi- the pur]wse of this inquiry, I do not think it would make
mu''li difference to our understanding of why IBM is dominant or
Avhat the implications are or what we might like to do about it if
TB^^Fs sliare is 55 percent, or 60 percent, or 70, or 80 percent, or in
that rano-e.

If IB^Ps share was 99 I thhik we would want to modify our under-
stan.ding of what it is that makes a share that high. If it were that
high we might have to have second thoughts about how the industry
should be structured. But again, I do not think it is an^'where in that
neiirhborhood either.

Mi-. Chumbris. ]Mr. Miller, if I may interject, we have had some
charts, and some other charts have already been submitted. Your
statistics seem to be higher than what we have been getting.

40-927- 75 49



5598

I think most of them have aromicl 64 percent as IBM's share of the
market. Would that make a difference? You said whether it's GO, 65,
or 75 percent, it doesn't make any difference.

But why is it that you arrive at the figures that you have and the
others that have been submitted are closer to between 60 and 65
percent ?

Mr. Miller. Let me tell you how I arrived at the figures I have
before I explain why they might be different from others.

The publication ''Computers and Automation," I think for the
early part of this period; and later the publication called "EDF
Industry and Market Report," publishes what they call a computer-
census, which shows the number of computers of various different
models installed in the United States and the average price of those
computers.

I have used those figures. I have multiplied the number installed
by the price. Say that the figures are that there are 100 IBM 360/50's-
installed and the average price is $i0,000. I would take $40,000 times-
100, which is $1 million; and do the same for all the other models.
Then I add the figures up and calculate the shares on that basis.

As I say, different industry trade publications publish different
censuses that will show different figures. I believe, in thumbing
through Dr. Brock's testimony, that he was using figures that were
compiled in the Honeyiaell v. ^peery Rand case on which I worked for
some time.

I had not realized that those figures were now available in the public
record. Those were obtained. I telieve. by subpena of data from the
computer manufacturers; certainly, one would presume that they
more accurately represent the situation than trade press estimates of
what is going on. The trade press simply miscounted. They mail out
questionnaires. The answers they get back may not always be correct.

ISIr. Chumbris. a little later other economists will be having some
questions. The reason I was interested in your chart is because a few
days ago I raised the question of back in 1956, I believe, one of the
charts,°IBM had 42.8 percent of the world market and Univac had
51.2 percent of the world market.

IBIM had about 47 percent of the domestic market, and Univac had
45 percent of the domestic market.

That was in 1956. And then they also gave the figures for 1974 which
showed IBM went u.p from 42 to 64 percent.

Univac went way down to 8 percent. I asked that the forthcoming
witnesses, or in the future, we ought to liave a year-by-year chart show-
ing just what caused IBM to jump from 42 to 64, 65 percent, and
Univac to go the other way. What caused the other companies who are
competitors to maintain whatever percentage they had.

That's why I was interested in your year-by-year statistics. I wanted
you to be able to show that from your point of view they could tell one
story, and if somebody else has a different type of statistics, there may
be reasons for it.

That's why I asked you that particular question.

Mr. Miller. One comment I would make : In my prepared state-
ment I do discuss the early history of the computer industry.

Mr. CiiuMBRis. Yes ; I read your paper.

Mr. Miller. In regard to that point : In 1955, maj'be even early in
1956, the market was divided approximately half and half by Univac,



5599

and IB]M. If your figures refer perhaps to early 1956 or if these figures
I present are wrong because they were an early registration of a trend
that did not come until later, one would find approximately even divi-
sion between IBjM and Univac, I believe.

The only point I would make is that very soon after 1956 — if not
already in 1956 — IB^M's market share got up to souiewhere in the
neighborhood of TO percent, or well over half. Univac's had fallen to
substantially less than half, and it has been tailing off slightly since
then.

jNIr. Chumbris. Thank you very much.

Mr. Miller. The question that I think is somewhat more interesting
than whether IB^I dominates the industry, is the question of Avhy IB^SI
dominates the industry. What factors are there that can explain the
acquisition and maintenance of this substantial position in an im-
portant industry?

I identify four tj'pes of factors that may be important. Fii*st, the
advantages of large size. "VVe often call all such advantages economies,
but the fact is that a competitor that is much larger than its rivals may
either be able to produce products at a lower cost, or sell them at higher
prices, or possess otlier advantages.

Second, forces that tend to stabilize market shares and keep them
where they are. You note that those forces are somewhat in conflict
with advantages of large size. Advantages of large size will help a firm
with 40 percent of the market expand to TO percent, while pushing its
rival down, if it starts out ahead. Stabilizing forces simply stabilize
whatever pattern of market shares one gets int:o.

The third group of factors is IB]M*s special historical advantages :
The things that one looks at if one looks at the historical record and
discovers that the companies in the industry did not all start out from
the same place at the same time.

Tlie fourth group of factors is conduct factors. These are factors
that are generally adduced in explaining about industry behavior. I
will not say much about them, but I simply observed that conduct is
one of the things that explains dominance in some industries.

On the advantages of large size, the first and in many ways the
most important are what we call economies of scale, because tliose are
advantages that enable a large company to produce its products at
lower cost than smaller companies. Since one of the purposes of com-
petition is to create efficiency, economies of scale might be said to be
factors that make large companies more efficient in a useful social sense
than small ones.

Economies of scale in the computer industry might arise in several
places. In the design and the development of computei-s, the fact that
there are one-time costs involved in doing research, and so on. My own
impressions, based on discussions witli the industry and a look at the
facts I have seen, are that the cost of design, development, and research
do create some economies of large scale," but not enormous economies
of large scale. It is clearly better to have a large volume to spread fixed
costs over than a small volume. But that is not a principal factor ex-
plaining IBM's dominance.

On the question of production of computers — including also the
jH'oduction of components — the evidence is mixed, or perhaps the facts
have been mixed. Early in the history of tlie computer industry. I do'
not think there ware ver\^ lar^e economies of scale in computer produc-



5600

tion. I think, in the mid-1960's there is some evidence that the tech-
nologies developed then were amenable to economies of large scale. I
really do not laiow much about the technology currently in use. I am
told, however, by some persons I have spoken to, that they do not think
there are large economies of scale. I am told by othei-s they do think
there are large economies of scale. I really would not offer myself as an
expert on the economies of scale in present production technology.

Another area where there might be economies of scale is in the pro-
duction of the so-called systems software. Central processors are so fast
that these programs are needed merely to make the computers operate
effectively and use their time effectively. My own opinion — and it is one
that I know is not shared by e^-eryone in the industry, although I think
some people would agree — is that there really are no necessary econo-
mies of large scale in systems software.

My own impression of the way software is written is that you have
a small army of programers writing software, if you happen to be
a large company, and then discover by Parkinson's law and because
of the difficidties of communication Avith each other that they manage
to occupy all their time doing it. And the writing of software is thus
a very expensive proposition. I think the facts of the computer in-
dustry will show that the small companies — they are almost an order
of magnitude — or a factor of 10 — smaller than IBM — manage to get
their systems software written. P^ither systems software is not very
expensive to them or to IBM; or if it is expensive, the small compa-
nies manage to write their systems software for a lot less than IBM
must be spending on it, because they simply cannot afford to spend
the same absolute number of dollars as, say, 5 or 10 percent of IBM's
re^•enues that IBM might spend on systems software. They would
have no money left over for an^^thing else.

Marketing, and distribution advantages, I think, are more, im-
portant advantages of large scale. IBM's ability to cover the market,
distribute computers from a large number of points, and provide
maintenance service locally in many places, I think, does give them
an advantage.

Another area where I think there are advantages of large scale is
in what I call "external economies of large size." Alfred Marshall
pointed out that some economies of scale have to do with the size of
the firm, others have to do with the size of the industry. And there
are some economies in the computer industry having to do with the
buying and selling of excess computer time, markets for programers,
exchanging computer programs, and so on.

The only difference between the computer industry and Marshall's
external economies of scale is that because of ]:>roduct differentiation
in the computer industry these external economies depend not only
on tlie total size of the industry, but on tlie nu.mber of computers of
a particular make or a paiticular brand; because only computers of
the same brand are compatible, and these external economies can be
siiared only by firms that are using computers of the same kind. For
that I'eason any firm that is large has put its customers in the position
where they can benefit by sharing with each other these external
economies from having a popular computer system. A firm that is
small can't achieve those advantages.

The final area where I see advantages of large scale is in product
differentiation and market intearation. It is an advantao;e that IBM



5601

has a very complete product line both in hardware and software and
in service, and that they are able to assemble a package that caters
to practically anyone's needs.

You do not even have to know what you want. You simply have to
Imow that you have a problem, and IBM has a solution for it.

Most companies do not have such large packages. They are not, in
that sense, able to compete as effectively in the marketplace.

Stabilizing forces in the computer industry are next. I think the
principal one is the enormous product differentiation — the fact that
not all computers are like each other.

And as a result there is an enormous amount of brand loyalty in
tlic computer industry. I think a better way of describing that, in-
stead of saying brand'loyalty, is to say that the customers are captives
of the suppliers.

If a firm has Burroughs equipment installed, it has programs that
are written and will run only on Burroughs equipment, because high-
level languages are not always used in all computer systems or because
perfect compatibility is not achieved even in all higher level lan-
guages. There is, therefore, an enormous cost of converting a com-
puter installation from one manufacturer's system to another
manufacturer's system. When one manufacturer brings out a new
product line he tries to arrauge conversion for his own customers, but
it is nuich more difficult to con^•ert between manufacturers.

IBM's special advantages aie the third group of reasons that ex-
plain its position in the com[)uter industrj^, and these are primarily
historical.

IBJNI was the dominant firm in tlie old tabulating equipment indus-
try, which was the predecessor of the computer industry, up through
the mid-lOaO's. It had the sales force. It knew the customers for
information i)rocessing equipment. It knew what kinds of work lousi-
nesses wanted to do. It had a large lease base of tabulating equipment
and a cash flow from it. It used tabulating equipment as input and
output devices for its computer systems. And tluis, by having the
front end- — the input and output arrangements already installed in
these businesses^ — IBM, in effect, was able to tie its computers in a
l^ackage with them.

It is also said that Univac, v hich was IBM's chief competitor in
the computer industry at that time and its onl}^ substantial com-
petitor in the tabulating equipment business, was not well managed at
that period.

Tlie final group of advantages that IBM may have is its conduct
in tlie marketplace. And as I said, I do not Avish to comment any
further about conduct, but simply to note the possibility that conduct
exists. My reason is primarily a feeling that this is a matter that is
likelv to be important in the litigation and, since the concern with
this committee is Avhether there is competition and whether struc-
tural reorganization is appropriate, I think that it is reasonable for
uie to avoid tlie question of conduct here.

To sum up tlie question of these forces that accouut for all of IBM's
douiinance, my own view is that the rise of IBM to douii nance in the
computer industry is due to a combination of historical circumstances;
namely. IBM's dominant position in the tabulating industry and its
good management at that time, particularly relative to its competitors.



5602

1 do not think that IBM's initial rise to dominance was due in any sig-
nificant way to advantages of large scale.

I think IB]M's continued position of dominance is due primarily to
wliat I have called the stabilizing factors — brand loyalty and things
like that — with perhaps considerable assistance from a pattern of con-
.diict that may have reinforced those structural features that do tend
to stabilize market shares. I think there are some advantages of large
scale, as I have said. But I do not think those have had a major impact
on IBM's share of the market, because I think that what they really
explain was how IBM was able to charge prices high enough to earn
very considerable excess profits without having that market share
degraded.

If there had been no advantage of large scale then IBIM would per-
]ia|)S have been forced to choose between earning excess profits and
seeing its m.arket share competed away — as is said to have happened
to U.S. Steel in the early part of this century — or maintaining its mar-
ket share but not being able to earn excess profits.

I think, now coming to the questions that are crucial to what I under-
stand to be the purpose of these hearings, that the committee is inter-
ested in knowing whether a more competitive structure for the com-
jputer industry would be in the public interest, because that, I under-
.'Stand, is the kind of question that is raised by the Industrial Keorgani-
5;ation Act.

The general proposition in favor of structural change is that more
iirms are better than few, and that competition is more effective if the
large firms are smaller rather than if they are larger. Tliis committee
has heard from numerous experts on this proposition who are much
better qualified than I am to offer opinions and views on it, and I
cannot tee myself trying to repeat that record. I will, therefore, confine
niy comments about reorganization of the computer industry to facts
peculiar to the comjDuter industry, or facts that are relevant, pri-
marily, in the computer industry.

I^lrsL, of course, there are monopoly profits in the computer industry
-over the long period of history. IBM's rate of return on the share-
holders' equity — that is, tlie net income after taxes as a percentage of
the equity — has been in the 15- to 20-pei'cent range for at least the last
two decades, and indeed in the upper part of that range for most of
tlie period.

That is a higher rate of return than is needed to attract capital for
investment in the business. It is higher than the average rate of re-
turn on all manufacturing; and my own opinion is that the average
for all manufacturing includes alread}^ a substantial cliunk of monop-
oly profit in it.

If we look instead at profits before taxes in relation to revenues, we
see tliat over a long period of years approximately $1 of every $-1 that
IBM receives in revenues is profits before taxes. Of course, something
like half of that profit is picked up by the Government — mostly the
Federal Government — in the form of income taxes. But we still see
that most of what you are buying — -not most but a quarter of what you
are buying when you buy an TBj\I computer — is profits to the share-
holders or taxes on profits to the shareholders.

But to say that there are monopoly profits, and to say that they could
be eliminated by reorganizing the industry, is not to say that reor-
ganization is desirable.



5G03

If tlie monopoly exists because IBM is able to produce computers at



Online LibraryUnited States. Congress. Senate. Committee on theThe Industrial reorganization act. Hearings, Ninety-third Congress, first session [-Ninety-fourth Congress, first session], on S. 1167 (Volume pt. 7) → online text (page 85 of 140)