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 88 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 88 of 140)
Font size
QR-code for this ebook


lating industry, getting, in effect, a head start over all of the other
competitors.

IBM has maintained its market position because of a number of
structural factors, prime among which is brand loyalty and customer
captivit^/. And it is not really necessary- — at least I do not see that
it is necessary for purposes of discussing the act — to decide whether
customer captivity is something tliat just automatically exists or wheth-
er it is something that exists because IBM made it that way. The
answer is that the industry is dominated by a single firm for historical
reasons and the domination has been maintained by a structural fea-
ture. And the next question is would the economy be better off
if that dominance were changed ? As I said, mj' answer to that question
is "Yes."

Obviously, if one does not believe, as some proponents of other argu-
ments seem to feel, that this dominance is a problem, then certainly
one does not need structural reorganization. If one wants to ascribe to
economies which one feels almost by hypothesis are worth preserving,
any of the advantages that account for dominance, then obviously the
conclusion is that one does not want to do anything about such domi-
nance as it exists.



5618

Mr. Granfield. I think wliat you are saying is that somehow an in-
dustry becomes dominated by a leading firm, that dominance is locked
in, it is inevitable and, therefore, competitive pressures will not remove
that dominance and you must do something aliout it.

Yes; I see in my counterhypothesis the following: If we see a firm
in an industry involved, dominated by one firm, that firm can only
preserve that dominance by continued superior performance.

United States Steel once dominated the steel industry. American
Tobacco once dominated the tobacco industry. In subsequent years they
lost their dominance because they no longer performed as efficiently
as their competitors. Consequently, they were punished.

NoAv, what I hear you saying is that if there is a dominant firm it
stays dominant because — my hypothesis of superior performance — it
was not prudent enough to be a superior firm and let the market erode
its dominant position.

I am very concerned about that because if these firms preserve their
dominant position, there is nothing inherent in keeping a dominant
position once j'ou get it.

We have innumerable examples where a dominant firm has lost its
position. We will enforce restructure, in essence, punish it, for the
remaining competitor. If it becomes a giant and is no longer competi-
tive, then we will not. We will not force any statutory changes upon it.

In essence, you are telling me you wish to punish superior perform-
ance. That is what you are saying; because you think it is locked in.

Mr. Miller. I am a little bit at a handicap in commenting l^ecause
I do not want to talk about IBM's conduct in a way that might be
construed as having something to do with present antitrust litigation,
but let me give one example.

IB^I's dominance is locked in, as I say and as I believe, largely by
product dift'erentiation. The effect of IBM's performance — and I am
not sure "performance" is a good word, and "conduct" is not a very
good word either — the effect of IBM's whatever it is doing that locks
this situation in is IBM's ability to build computer products that are
sufficiently different from what the other manufacturers are building,,
so that there is a substantial cost involved in switching from an IBM
product, or an IBM computer system, to somebody else's computer
system. Now, if you want to, you can characterize that as superiority,
because that particular situation enables IBM to charge high prices for
its product — prices that are sufficiently high to earn excess profits, but
that are not so high that despite the cost they induce users to switch.
If you want to call that superiority of management, I am perfectly
willing to accept the proposition that that is superior management.

I am not Avilling to accept the proposition that that kind of superior-
ity is of any benefit whatsoever to the customers of IBM or to the U.S.
public.

And I believe that that particular pattern is very important to the
computer industry, and that structural change will do something to
stop that kind of behavior.

Mr. Granfield. So you are saying that anyone who is a customer
of total systems suppliers is locked in ?

Mr. Miller. I am saying that in the computer industry, not neces-
sarily any other industry, but in the computer industry anyone who is
a customer of a total system supplier in the present environment of



5619

high product differentiation mnst pay a substantial cost penaUy for
changing to someone else.

Mr. Graxfielu. But you onl}^ single out one firm. And tliat seems to
be the firm that lias performed most effectively in terms of selling a
total systems package.

IB]\i is not the only one who does it. They seemingly do it better
tlian other firms, I guess, because customers do not switch. You singled
out one firm to end evil of a system lockout. You intend, in your orga-
nization, to allow the others to impose their costs on their customers.

Is not that mildly inconsistent ?

ISIr. ]\IiLLER. Xo, sir, it is not. The purpose of reorganization is not
to punish IBJNI. The purpose of i-eorganization is to build a competitive
structure in the industr3^ I believe I stated, and I tried to make it
clear, that I feel that the result of having a competitive structure in the
computer industry* will be to have a decrease in product diiferentiation,
because the forces of competition in that changed structural situation
will, through market pressures, induce the manufacturers to decrease
the product differentiation that they have.

I do not single out IBM for reorganization as a punitive measure.
I simply observe that if one wants to decrease the seller concentration
ratio, which I believe, from my analysis to be a key to determining
the stractural situation in the industry, it obviously will not be achieved.
by fragmenting everybody else ; it will be achieved by dissolving IBM..

Mr. Granfield. Then certainly after hearing you, I think your com-
ments indicated you have given this a great deal of thought and I cer-
tainly respect that. Perhaps an alternative solution to the problem you
])erceive in this industry, in terms of customer loss — and this bothers
me — is that we are going to determine that the customer does not know
when he is better off', we are going to tell him when he is better off.

Let us leave that aside.

j\Ir. INIiLLER. I do not think I said that we Avere going to tell them,
I simply said that I think we want a market situation in which the
cost of allowing customers to choose each time they buy a product is
reduced, that the customers do not face a substantial cost in moving
from one supplier to the other.

We are not going to tell them that they cannot continue buying from
IBM or from IBM's successors.

Mr. Graxfieij). An alernative might be to let us just mandate cer-
tain kinds of standards tliat would reduce these trends. Of course, we
do not have to break up IBM at all, we just have to mandate that all
firms have compatible hardware or software and we can define "com-
patible" in terms of transfer cost.

Mr. Miix-ER. The only thing that bothers me about that is that I have
always been from the school that believes that every time that we — the
public, the Government — start telling people how to conduct their })usi-
ness, we find that the performance is not nearly so good as the perform-
ance is when we let market forces take care of things. That is the
main ai'gument in favor of structural reorganization: AVe do it once
and we get out; and we let what happens in the marketplace ha})pen.
We don't set up some board to say, "You have got to meet these stand-
ards of compatibility; you have got to announce your product in ad-
vance ; you have got to do this, that, or the other thing in order to give
everyone else a fair chance."



5620

I just do not think it works. I do not think it has worked very well
in the industries that are presently regulated; although in some of
tliem it is undoubtedly the best solution, and in others it is not.

I certainly would not be looking for the excuse to try it in the com-
puter industry.

Mr. (irRANFiELD. Tliis is a standards regulation. That is somewhat
different than some of the regulations.

But the "we" — the Government — will do the divesting, and I am
kind of curious as to why we think we will have so much wisdom doing
that wlien 3'ou claim we will not have wisdom doing it in certain other
activities.

?i[r. Miller. I do not think we are so wise about divesting, but we
will do it once and that is all; whereas if we liave to sit there and
regulate, then vre Avill have to do it a lot more often. ^Ye get lots more
chances to foul things up if we go into the regulator}- business than if
we go into the structural reorganization business.

Mr. CiRANriELD. But every time a firm achieves a dominant position
3-0U are going to go bust them up again.

oSIr. Miller. In my reading of the historical record there have been
very few instances since 1890 or since 1911 when firms have achieved
dominant positions. I can think of something like Xerox, which has
acliieved a dominant j^osition, that is not related to some other base.

Other than that, the automobile industry concentration has in-
ci-ensed but my recolleirtion — and it is a dim one, of times in distance
past, and I am not an expert on these facts — is that the really bkq: u\-
creases in seller concentration in the U.S. economy occurred a long
lime ago. They occurred at a time when there may have been some
question about how the antitrust laws were to be enforced, or they
occurred as a result of loopholes — what I would characterize as loop-
holes— rin the merger laws prior to 1950.

I would certainly tliink it appropriate to reconsider the reliance
on structural reorganization, if you could put together a record that
shows that there was a strong tendency for concentration to develop
or increase in unconcentrated industries the American economy in the
postwar period.

]Mr. Granfield. I think we could go on for a veiT long time. I do
thank you very much.

yir. Miller. I think you for the opportunity of appearing.

Senator PTart. Thank you very much.

I anticipated we might be able to continue without a recess. After
checking with our remaining witness. Professor Brock, I am being
advised that his plane does permit him to stay awhile. I suggest a
recf ss until 2 :30.

[Whereupon, at 1 :10 p.m.. the subcommittee recessed, to reconvene
at 2 :30 p.m. this same day.]

MATERIAL RELATING TO THE TESTIMONY OF RALPH E. :MTLLER

Exhibit 1. — Prepared Statemerit of Mr. Miller

Prepared Statement op Ralph E. Miller, Economist, Washington, D.C.

My name is Ralph E. Miller, and I am an economist living in Washington. D.C.
I liave a master's degree in economies from Harvard University. My interest
in the compnter indnstry began in the mid-lfK50's. when I was still a gradnate
student at Harvard. At that time I started work on a doctoral dissertation that
was intended to be a study of the structure and performance of the computer
industry. After leaving graduate seliool. I was an Acting Assistant Professor



5621

of Economics at the Berkeley caminis of the University of California. I taui;lit
the undergracUiiite and in one year also the graduate course iu industrial organiza-
tion and public policy. While at Berkeley, I was also a consultant to the counsel
for i'ontrol Data Corporation, and I assisted them in the development and
preparation of their antitrust suit against IBM.^ I then moved to Wasliinyion,
wliere I joined a consulting firm, and one of my clients was Sperry Rand Corpora-
tion. I assisted Sperry in the antitrust aspects of their patent and antitrust liti-
gation with Honeywell." During this period, I al.so pre.sented an invited paper on
"Public Utility Aspects of Computer U.se" at one of the seminars sponsored l>y
American Telephone and Telegraph Company on Problems of Regulation and
Public T'tilities.

Most recently, I was employed for almost a year in the Antitrust Division of
the Department of Justice, where I was assigned to the IBM case.^ I left that
job to move to my present position, and I mention it only to emphasize that my
statement today will not draw in any way upon confidential information that
came to me in my work at the Justice Department. Nor shall I reveal any
proprietary data that I had access to In my other consulting activities, but my
consulting experience has of course helped to shape my views on the computer
industry.

To ensure that the record is kept straight on the sources of my information. I
have based my statement today upon my still incomplete dissertation manu-
script, which was given privately to a few persons for review at various times
prior to my entry on duty with the Justice Department. This manuscript covers
the period through about 1969, and I trust that the Committee will understand
why I have limited my detailed consideration to this time period.

Because of my past association with the Department of Justice, I shall also
refrain from comment about the pending IBM case. Specificially, I shall not
comment on whether I believe IBM is or has been in violation of the antitrust
laws since 1960 (the period of the complaint) ; nor shall I comment on remedies
appropriate in the event that the court finds for the United States in that case.

I do intend to assess the present structure of the computer industry, and to
comment on the merits of several possible ways of restructuring it, but these
views are ))ased solely on economic considerations — they are not views of what
the present law does allow or should be interpreted to allow.

Finally, I must emphasize that the views I present here are my personal views,
and they are based on my personal study of and experience with the computer
industry. I am not representing my present employer, or any past employer or
client, or any federal agency ; and my views are not necessarily those of any of
these other parties.

In this statement I first give my views on what the computer industry is. I
then present data that I believe show the industry to be dominated by IBM, and
I try to explain how IBM achieved and maintains this dominant position. 1
then" consider the question whether a more competitive structure for the com-
puter industry is likely to be in the public interest. Answering this question
affirmatively, I present and evaluate several possible ways the needed re-
structuring might be achieved. I close with a few remarks about the relevance
of this analysis in an industry changing as rapidly as is the computer industry.

THE COMPUTER INDUSTRY

One of the first questions that must be confronted in discussing the computer
industry is the question of what the computer industry encompasses. My own
view is that the industry is most clearly seen as encompassing the production
and initial marketing of general purpose electronic digital computers or com-
puter systems. Computers are the pieces of machinery and equipment them-
selves—the so-called "hardware" — while computer systems also include the
principal programs ("software") needed to make the computers operate effec-
tively, together with other essential support customarily provided by the manu-
facturers of the hardware.

The decade of the 1960"s was a period when the computer of general purpose
electronic digital computer systems was especially clear. Almost all computer
installations were confined to a single site, and the boundary between equipment
in the system and equipment not part of the system was thus clear. Each com-
puter system contained a central processing unit, where the arithmetic and
logical functions were performed (the so-called "mainframe") ; plus devices



^Control Data Corporation v. International Business Machines Corporation (D. Minn.)

'^ Honeywell Inc. v. Sperru Rand Corporation et al. (D. Minn., 4-67 Civ. 138).

3 United States v. International Business Machines Corporation (S.D. N.Y., 69 Clr. 200)^



5622

such as card readers, printers, and tape units which were used for the input,
output, and storage of data (these are the so-called "peripheral" equipment).
There has always been some peripheral equipment manufactured and sold di-
rectly by independents, but in the 19G0's the mainframe manufacturers either
made or distributed the great preponderance of the peripheral equipment, and
thus it is appropriate to represent the mainframe manufacturers as purveyors
of at least the complete hardware complement for computer systems.

During the 1960's there was also an especially clear distinction between gen-
eral purpose digital computers and other kinds of electronic computing equip-
ment. Computers built for use in weapons systems were designed to meet severe
environmental specifications set by the armed forces, and they were not com-
petitive with general purpose computers for non-military use. A clear illustra-
tion of this point is that the Department of Defense is the world's largest user
of general purpose digital computers (unless one considers the entire federal
government as a single user), in addition to being virtually the only customer
for special purpose military computers. Some lirms also manufactured com-
puters and incorporated them into special purpose systems in such fields as
communications, but the volume of business in these fields in the 1900"s was
very small relative to general purpose digital computers. The one exception is
the manufacture by Western Electric of computer-like devices for use in the
Bell System of telephone communications. Here the volume of bu^-iness may have
t>een appreciable, but the special relationship of Western Electric to the other
Bell companies kept this activity insulated from the computer industry.

Finally, the 1960's was a period when it was especially easy to distinguish be-
tween the software and services included in computer systems and those that
were not so included. The principal reason for this is that during the 1960's the
major computer manufacturers sold their systems on a "bundled" basis — that
is, they typically charged a single price based on the hardware, with the sys-
tems software and essential supporting services supplied without separate
charge. The computer system was thus the complete package offered by the
mainframe manufacturer ; and with few exceptions no other vendors supplied
parts of the system directly to users, because few users would pay an independent
vendor for something the hardware manufacturer supplied "free".

In addition to the software and services included with computer systems, there
are other types of software and sei-vices used in electronic data processing (EDP) ,
but not part of the computer industry proper. The so-called "system software"
included in a computer system encompasses only the principal program packages
needed by virtually all users for effective use of their computers. Examples are
the "compilers"' and other programs that translate from higher level languages
X3ongenial for programmers into the brutallj^ simple and elemental instructions
that computers can understand and execute ; and the programs that schedule the
internal operations of the various components of a computer system, to ensure
that they are kept as busy and productive as possible. Most computer programs
are "applications" programs that do the specific data processing work of the user,
and they are not part of the computer system as defined here. Most applications
programs are written by the users themselves. Some, especially those for the
more widely used applications such as payroll, are written and sold by so-
called "software houses" ; and some are offered by the computer manufacturers
themselves, either bundled with the system or for an extra fee.

Also in EDP but not in the computer industry inself are markets for the serv-
ices of the existing stock of computer systems. Computer leasing companies pur-
chase computers from the manufacturers and rent them to users, usually for a
term of one or more years. Owners of computers often sell the use of their com-
puter systems during the hours or even the entire working shifts that they do not
need them themselves. "Time-sharing"' vendors sell the use of "part of a com-
puter" to numerous small users, in an arrangement where each user can connect
his own terminal to the shared computer over a telephone line. Service bureaus
oft'er computer time, but they also offer "one-stop'' shopping for EDP services,
providing a sul)Stitute not only for the user's computer system, l)ut also for the
programmers, keypunch operators, and other personnel needed to use EDP. In
payroll, for example, a service bureau may take a handwritten record of the
number of hours worked by each worker during a pay period, and from it gen-
erate the payroll checks, the business accounting records, the payroll tax reports
to the government, etc. Vendors in all of these market areas are outside the com-
puter industry, because they do not produce computers or computer systems. In-
stead, they must purchase their computer systems from the computer manufac-
turers, and thus they cannot properly be said to compete in the computer manu-
facturing industry. To some extent, the computer manufacturers are themselves



5623

-engaged in these areas, often through separate corporate divisions or subsidiaries,
but this activity is still extraneous to the computer industry.''

Even with a relatively narrow focus on general purpose electronic digital com-
puter systems, the computer industry is a large and important one in the U.S.
economy. As shown in Table 1, value added by manufacturing in the computer
industry has risen from 0.4 percent of the total of all manufacturing in 1958 to
1.0 percent in 1972; and the outlook for the foreseeable future is that the com-
puter industry v?ill continue to grow more rapidly than the economy as a whole.

TABLE 1.— THE COMPUTER INDUSTRY IN RELATION TO THE U.S. ECONOMY

Value added In manufacturing

Shipments • •

of electronic Computer All

computers industry manufacturing

(millions) (billions) (billions) Percent

1958. $324 $0.6 $141.5 0.4

1963.... 880 1.3 192.1 .7

1967 1,973 1.9 262.0 .3

1972 4,362 3.6 348.0 1.0

Source: Department of Commerce, Census of Manufactures.

The economic activity occurring within the computer industry is, however, an
understatement of the importance of computers to the United States and its
economy. Computers are used prevasively throughout business and government,
and they are intimately associated with two of the vital functions of what may
be called the post-industrial society, namely decision making and record keeping.
Computers run our oil refineries and our electric utilities. They keep our business
and financial records. Without oiir computers, our society would grind quickly to
a halt.

IBM's DOMINANCE OF THE COMPUTER INDUSTRY

Because the computer industry is so important to the United States and its
economy, it is proper that the structure of this indusry receive special considera-
tion in the proposed Industrial Reorganization Act ( S. 1167 ) and in this Commit-
tee's hearings on the Act. The outstanding structure characteristic of the com-
jputer industry is its dominance by a single firm : International Business Machines
Corporation. Owing to the illustrious record of this Committee, and to the work
of the many experts who over the years have appeared here, it is not necessary
for me to explain the importance of seller concentration. I shall therefore turu
directly to the record of IBM's dominance of the computer industry, as I see it.

I find that the best single way of measuring market shares in the computer
industry is by reference to the total outstanding stock of general purpose com-
puter .'Systems, which is sometimes called the installed base. This chuice results
from con.sideration of the technical merits of the various possible measures, and
from judgments about the quality of the data available for each type of
measurement.

Shares of the installed base of computer systems are shown in Table 2. These
shares are calculated from censuses of computer installations. Each census states
how many computer systems of each model are installed on the census date, and
what the average rental price is for each model. The total value of the installed
stock of each model is the number of installations times the average monthly



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 88 of 140)