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 90 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 90 of 140)
Font size
QR-code for this ebook


advance in business EDP. Tlie computer was supposed first to take over the pay-
roll computation and then add other data proces.sing work. But failure to provide
adequate programs delayed its use for a number of months after the computer
was installed, and Univac acquired a bad reputation." The problems were no
direct fault of Univac, in the sense that the hardware performed as claimed.
But if IBM offered customers help with programming, one can at least find Uni-
vac guilty of an error of omission in the battle for hardware sales.

Another disadvantage faced by Univac in the business EDP field is Remington
Rand's position as poor second to IBM in the old tabulating card equipment in-
dustry. Univac therefore had fewer estahlishsd customer relationships and a
much narrower range of sales contacts for computers. This disadvantage is high-
lighted by Univac's comparative success in scientific EDP and in salea to the
federal government : in both of these areas, IBM too was opening new ground.

A further difficulty faced by Univac in selling its large computers (the 1100
series) is failure to develop a small computer model to follow Univac I. IBM
came out \^ith the 650 in November 1954, and eventually sold over 1,000 of them.
IMauy went to customers who wanted a comparatively inexpensive trial for EDP,
and later converted to the IBM 700 series. Univac's failure to provide a full line
of equipment was thus another factor contributing to its slide from dominance.
This can perhaps be called a management error, and it is not the only one com-
mitted by Remington Rand in regard to the Univac division. Schussel feels that



^ J. Presper Eckert and John W. Mauchl.v are the builders of the ENIAC. which is
generally recognized as the flr>;t electronic digital computer. Their brief experience as
independent entrev'reneurs in the computer industry is typical of many tiny computer
engineering firms that sold out to much larger corporations.

is Remington Rand's lf».")l acquisition of Engineering Research Associates, manufac-
turers of a large scientific computer, contributed to this lead.

"The facts of this story (as told by Schussel) are disputed by John K. Swearingen
fletter to the editor, Datamation XI. No. S [1965], 1.^). who worked on the GE luoject
team when tlie Univac was installed. However, it is T'nivac's reputation, and not the
facts, that are relevant to a history of the computer industry, and there is little doubt that
the Louisville experience did give Univac a black eye.

40-927—75^^51



5630

poor management in several respects was a major factor contributing to Univac's
lack of success.^

Technological leadership — one factor that might have helped explain IBM's
dominance — is conspicuous for its absence. Instead, Univac is considered to have
been a greater contributor to the art of computing. Univac hardware was cer-
tainly the equal of, and probably superior to, IBAI's. True, IBM electromechani-
cal equipment, used primarily for input-output operations, is generally considered
to be the best in the field, but it was in the human aspect of man-machine com-
munications that IBM excelled. This excellence in the field of computer appli-
cations was manifest in the superior customer assistance provided by IBM
salesmen.

This historical sketch emphasizes IBM's entrepreneurship in achieving its
dominant position in the computer industry. But it is also striking that the only
competitor able to mount even an unsuccessful challenge was the only other firm
in the tabulating equipment industry ; and IBM and Univac emerged into the
computer age in the same relative positions they held in the tabulating industry —
with IBM dominant. More recent analysis has suggested two factors not em-
phasized by Schussel, but related to IBM's dominance of the tabulating industry,
as contributing to IBM's rise in dominance. The first is the cash fiow from IBM's
lease base of tabulating equipment, which IBM could use to finance the rapid
growth of its computer business. The second is IBM's ability to make tabulating
equipment — especially that already in the hands of its customers — serve as
input/output equipment for IBM computers.

One final historical item should also be noted : the consent decree of 1956.
Just when IBM began moving strongly into tlie nascent computer industry, the
U.S. Department of Justice instituted legal proceedings that were to have an
important impact on the new industry's structure. In January 1952, the Justice
Department filed a complaint against IBM, alleging a variety of offenses against
Sections 1 and 2 of the Sherman Act. Four years later, the case was settled out of
court by consent of the two parties, and the consent decree was filed and entered
in January 1956.^"

The complaint was directed primarily against IBM's dominance of the tab-
ulating card and tabulating card equipment industries, but the terms of the con-
sent decree were drawn to cover computers as well, and therein lies the impor-
tance of the consent decree for this study. Those provisions with significant
effects on the computer industry will now be summarized.

(1) IBM was required to offer for sale all equipment that was generally
available on lease, and the purchase prices had to have "a commercially rea-
sonable relationship" to the rental terms. Leases were limited to a term of
one year. Except for maintenance and repair, all services offered to lessees
had to be offered free to purchasers ; and maintenance and repair had to
be offered at reasonable charges.

The importance of leasing as an exclusionary practice had recently become clear
in the American Can and second Shoe Machinery cases." These provisions pre-
vented the gross abuses that had been discovered there.

(2) IBM could not prohibit the use of its equipment in a system with non-
IBM components. IBM could not tie its maintenance to the purchase of its
hardware ; nor could it require the use of IBM tabulating cards by any owner
or lessee of its equipment (not even as a condition relating to maintenance
or warranty of the equipment).



1^ Schussel also feels that IBM was never so far behind as some people have thought
It was. It participaterl heavily in the construction of the Mark I and II at Harvard, and
it also had a line of electronic calculators (including the CPC). The reason it was not yet
in the computer business is that Thomas J. Watson (St.) — like almost everyone else — ■
grossly underestimated the potential market.

i« United States v. International Business Machines Corporation, CCH 1956 Trade Cases
II 68,245 (S.D.N.Y. 1956). A summary of the complaint appears in Commerce Clearing
House, The Federal Antitrust Laws, With Summarn of Cases Instituted bii the United
States. 1890-1951 (New York: Commerce Clearing House, Inc., 1952), p. 42.S.

"U.S. V. American Can Company, 87 F. Supp. IS (N.D. Cal. 1949) ; and U.S. v. United
Shoe Machinery Corporation, 110 F. Supp. 295 (D. Mass. 195.'?). For a discussion of leasing
practices and their effects, see McKie. Tin Cans and Tin Plate, pp. 182-197, and Carl
Kaysen, United States v. United Shoe Machinery Corporation (Cambridge, Mass. : Harvard
University Press, 1956), pp. 64-73.



5631

These provisions were attempts to split the integrated dara processing market
into segments that eould he entered hy smaller, specialized firms.

(3) IBM had to offer training to indei3endent repairmen (including any
users desiring such training), and it had to offer to sell parts and subassem-
blies to owners and to independent maintenance firms.

This was apparently a special effort to break open the technology of the equip-
ment. It is noted here because scale economies in maintenance have become im-
portant in the computer industry, and, so long as maintenance is provided by
manufacturers, these scale economies are an advantage of large size in the com-
puter industry. It is therefore worth noting that the consent decree attempted to
encourage independent maintenance and repair service.

(4) IBM was required to grant non-exclusive licenses under any, some, or
all of its patents, at reasonable royalties (royalty -free, for patents reading
on tabulating cards or tabulating card equipment), to anyone applying for
such licenses. This provision covered all present patents, plus any that might
be obtained for five years.

(5) IBM was required to place all of its service bureau activities in a
separately organized, but wholly owned, subsidiary corporation (called the
Service Bureau Corporation, or SBC). SBC was not to use the IBM name or
any IBM facilities or personnel. It was required to keep separate accomits,
and to cover its costs ; and it was not to receive any favorable terms in deal-
ings with IBM, unless those terms were offered to all other service bureaus.

As a group, the remedies in the 1956 consent decree were conduct remedies
applied to the conditions in the tabulating industry as that industry was under-
stood at the time. There is only one comment that need be made about them :
they did not succeed in reducing IBM's dominance of either the tabulating in-
dustry or of the computer industry that grew out of it.

The final class of factors tliat may account for IBM's dominance of the com-
puter industry is IBM"s conduct in the marketplace. With the government's anti-
trust suit pending, it does not seem appropriate to discuss market conduct ira
detail. However, I would be remiss if I did not mention a few of the ways in whicb
market conduct may (or may not) have affected market structure : (a) the extent
of product differentiation may depend upon IBM's product policies; (b) the
practice of bundling and integration of the product line is dependent upon producte
and price policies; (c) educational discounts and, in some instances, gifts of
computers to uuiversities may have increased IBM's penetration of the imiversity
market beyond what it would have been otherwise, thus capturing new entrants;
into the ranks of EDP customers while they were still training for their jobs ;
and (d) repressive and disciplinary practices may have been used by IBM ta
shore up its position in parts of the computer market where it was threatened.

The actual situation of IBM's sustained market dominance in the computer
industry is the result of a balance among the various forces described above. My
own view is that IBM's rise to dominance is due to a combination of historical
circumstance (the dominant position in the tabulating industry) and good man-
agement, but not in any significant way to the advantages of large scale. I believe
that IBM's continued dominance can be described primarily to the stabilizing
factors, with considerable assistance from a pattern of conduct that (perhaps
not deliberately, and perhaps not even knowingly) reinforced the structural
features that tend to stabilize market shares in the pattern of dominance. Some
advantages of large scale were also present, but their impact on market share.^
has in my view been offset by a price structure for IBM that yields very large
excess profits.

A MORE COMPETITIVE STRUCTURE FOR THE COMPUTER INDUSTRY WOULD BE IN THK

PUBLIC INTEREST

The question of monopoly profits leads to the issue that is the nub of these
hearings on the proposed Industrial Reorganization Act (S. 1167) : whether a
more competitive structure for the computer industry would be in the public
interest. The argument for structural change depends first upon the proiwsitiorj
that more firms are better than fewer ; and the smaller the relatively large firms
in a market, the better. The broad applicability of this proposition lias been, sub-
ject to great debate, and I have nothing to offer the committee that it has m>t



5632

already heard from persons better able to comment than I am. Suffice it to say
tliat the weight of economic analysis seems strongly to support this proposition.

I do have some specific comments about the market performance of the com-
puter industry, and some views on how and why it will be improved by
structural change. The first aspect of performance is monopoly profits. There is
BO question that computer industry as a whole, and IBM in particular, earns
excess profits IBM's rate of return on the capital invested by the shareholders
has for many years been in the upper half of the 15-20 percent range, and this
is far above the average for all manufacturing or the rate needed to attract
new capital. In relation to sales, IBM's profit rate has stayed near 25 percent,
before taxes This means that one dollar of eveiy four received by IBM goes to
profits or to profits taxes. But the elimination of monopoly profits is no gain to
the public if it is achieved through a structural change that increases costs to
tlie level of the monopoly prices. The public benefits from lower prices, not
merelv lower profits for IBM, and the question of monopoly profits must
therefore be viewed as a matter involving cost performance and product
performance. .

One posible basis for IBM's monopoly profits is the ability to sell its products
at higher prices than its rivals charge for equivalent products. This could result
from IMB's product differentiation advantages, and from capturing the ex-
ternal economies that go to users of popular computer models. There is some
evidence that such a situation did exist at one time in the computer industry, but
the record is no so clear as one might want. If any such price advantages do
exist for IBM, their dissipation by structural change will benefit the public.

A more likely result of structural change will be an improvement in product
performance, especially in the areas of product differentiation and marketing.
My judgment is that a change in the structure of the computer industry will be
accompanied by a decrease in product differentiation and marketing activity.
There will be greater modularity and compatibility of hardware, including
peripheral equipment and related products, of software, and of services. The
expertise in systems integration, which is a major and costly ingredient of
tfae mariieting effort, will pass from the computer industry to the users of
computers- and that, I submit is where it belongs. Decreased product differ-
entiation will permit firms specializing in short product hues to flourish in
their market niches ; and it will give scope for innovation along a broad front,
using the "components' rather than the "systems" approach to technological

dianffe. „ . ■, j.^ •

Increased user expertise will lead to greater efficiency and other improve-
ments in the way computers are used. Some of the fiascos of overselling may be
avoided Special" notice should go to the problems of computer security and
computer fraud. These problems are in my view exacerbated by the ignorance
that many managements have of electronic data processing techniques. Owing to
this i«-norance the responsibility for ensuring that computers are not misused
is divided between the user (who understands his own business, but does not
know much about computers) and the purveyor of EDP services (who under-
stands computers, but may know little of the user's business). The result is that
the re'^ponsibility mav fall into the crack between these two jurisdictions. In
contrast if changes in the structure of the computer industry lead to a diffusion
of computer expertise throughout the user community, then there is hope that
business management will be better able to cope with the security and fraud
nroblems that have accompanied the information revolution. , , , ^

Ap'irt from the likely improvements in market performance, structural change
in the computer industry will likely have benefits that are more social and
BOlitical then narrowly economic. In our democratic society it is at best repugnant
Ind at worst dangerous for a large fraction of our business and government to
be dependent upon a single firm. Nor are the risks only political and social : even
as an economic proposition, it is risky for most of the nation's eggs to be in

MBain Industrial Organization, ch. xi ; Kaysen and Turner, Antitrust "P^l^c^' T>r>: I'i^f-
'tlG IndGeorseT Stftxler, "Tlie Case Against Big Business." Fortune, May 19o2. An
^cehei^^tsS of essays on this subject is in Edwin Mansfield ed.. Monopoly Power and
'pooiSc Performance (rev. ed. ; New York: W. W. Norton & Company, Inc., 1968) ; big-
BesTif doSXbv John Kenneth Galbraith ("The Economies of Technical Deyelop-
mcnf ') -and this argument is attacked by Richard Nelson. Merton J. Peck, and Edward
KalichVk ("The Concentration of Research and Development in Large Firms") and by
Jacob SchmooklPr( "Market Structure and Technological Change"). Donald F. Turner
r4he AiSust Cbief^^^^ summarizes the arguments. Also see F- M Scherer

"Statement." in Industrial Organization and Public Policy, ed. by Werner Sichel (Boston :
Houghton Mifflin Company, 19G7).



5633

one basket, especially when those "eggs" are as important to our economic diet
as computers.

A related problem is that big business almost invariably leads to "big
government," i.e. government intervention in what would otherwise be decen-
tralized decision making in a free and competitive market. Where the big'
business is perceived by the Congress as a natural monopoly, as in the case «^
American Telephone and Telegraph Company's dominance of the telephone
industry, the Congress has typically imposed economic regulation. Regulation
has not always been a good solution to the monopoly problem, but it has been
acceptable ; and in some industries it may even be the best choice among un-
palatable alternatives. But unnecessai'y monopoly also breeds government r^ -
ulation, and in a much more insidious (and therefore more dangerous) way-
Instead of a single regulatory authority, we find responsibility for economic
decisions spread among several government agencies, and the resulting mixture
of a political and administrative process is much less efficient. Automobiles arfr
the clearest example, with the auto industry caught in the midst of a three-
way tug-of-war : the safety boards want safer ears, the environmentalists wants
less poUutiun, and the energy savers want better gasoline mileage. Each tries
to impose its will by fiat, and it ends up being the responsibility of the Congress
to decide what kind of cars Americans shall drive. It lias not happened yet in
computers, but it will happen if nothing is done to change the structure of tbe
computer industry, to make it more competitive.

STRUCTUEAL REORGANIZATION OF THE COMPUTER INDUSTRY

The proposed Industrial Reorganization Act (S. 1167) prescribes structural
reorganization as the normal remedy for monoply. If it is accepted that the
computer industry is a monopoly (under the standards of the Act), and if the
structural change is in the public interest, then the only remaining question
is determining a feasible and procompetitive reorganization plan.

The first point about reorganization is that it need involve only IBM. Even
now, with Hcnieywell and Univac having absorbed all or most of the computer
operations of GE and RCA, no firm other than IBM appears to hold more than
about ten ijercent of the market, and four such firms would still fall below the
seller concentration level presumed under the proposed Act to be a monopoly.
Thus the principal purpose of reorganization is to break IBM into pieces no
larger than the size of its chief competitors.

In considering a structural reorganization of IBM, it is most convenient to
begin at the outside, with IBM's activities outside the computer industry. Other
products that are important to IBM are electric typewriters and other office
products ; magnetic disk packs, tabulating cards, and other EDP supplies : and
military electronic equipment, mostly special purpose computers. These prod-
ucts are produced respectively by IB:M's Office Products. Information Records,
and Federal Systems divisions, and these three divisions should be separated,
either singly or jointly, from the successor computer manufacturers to IBAL

The Office Products Division does its own development and marketing, as
well as production, and its Selective typewriter is the leader in its field. It
apix-ars, therefore, that this division would be viable by itself. If success in the
office products field is dependent upon integration into at least some areas of
EDP (though the separate divisional organization belies this contention),
there are numerous small electronics firms that Office Products could later
merge with to achieve this entry.

The Information Records Division should also be viable as an independent
corporation, as there are several independent and successful firms in its lin^
of commerce.

The Federal Systems Division (FSD) is concerned largelv with special pur-
pose hardware for military use. and especially with digital communications.
It also does a substantial amount of systems development work for the federal
government and also for state and local governments. In all of this work, the
division does its own marketing and iiianufacturing (except, of course, where
general purpose IBM computers are used in complete systems that it supplies).
Product design and some devehjpment work is also done at FSD's plants, bat
the division presumably relies on other IBM divisions for at least basic research.
A firm with the size and product line of FSD would presumablv have no trouble
prospering— many firms of e(iuivalent and smaller size do just that in the
military electronics field. Some care might be needed to provide the division
with a research capability, but IBMs Research Division operates several labo-



5634

ratories ; and FSD could if necessary establish its own, possibly being given some
IBM personnel as a nucleus. -r^^r,

The second layer of structural reorganization is the reduction of IBM s
integration across the various market areas in and closely related to the
computer industry itself. Four market areas are important here: small infor-
mation processing systems, peripheral equipment; electronic components; and
software and supporting services. These are considered in turn.

IBM's small information processing systems are the successor to its tabulating
equipment business. Development, U.S. manufacturing, marketing, and support
are all concentrated in the General Systems Division. This divisional structure
suggests that separation from the rest of IBM's computer business is feasible,
though information published in the late 1960's indicated that the IBM plants
manufacturing tabulating equipment also manufactured computer systems.^^
Sei)aration of the small systems business from the larger computer systems is
desirable, because it will help undermine the barriers of high product differen-
tiation and tightly integrated product lines. IBM considered some such a cleav-
age feasible when it brought out the System/3 as a product line separate from
the Svstem/360-System/370 complex. The only "economies" likely to be present
with IBM after this cleavage are the marketing advantages that result from
Ivuowing which ustio are good prosi)ects for upgrading to larger systems, and
these are private pecuniary advantages rather tliau savings of social costs.
Separation of IBM's small systems business from its computer operations will
open this group of prospective customers for all computer manufacturers, and
it is thus clearly procompetitive.

Peripheral equipment presents some different problems. IBM has within the
last several years transferred development and manufacturing responsibilities
for at least some peripherals to its new General Products Division, and this
may suggest the feasibility of separation from the rest of the computer systems
.business. However, this separation is not recommended as a key part of any
reorganization plan. One reason is that the i)roduction of peripheral equipment
occurs in the same plants as the production of main frames, and the two are so
intermingled that it would be impractical to assign some plants to the periplieral
equipment firm and others to the computer firm. Second, peripheral equipment
as a whole is not a line of commerce. The term covers a variety of different lines
of equipment related to computers, but many of these lines are not related to
each other (e.g., magnetic tape drives and cathode-ray tube terminal display
sets). There are indeed independent peripheral equipment firms, but they are
specialists — none manufacturers as long a line of peripherals as even some of
the smaller computer manufacturers. It might be feasi))le for IBM to divest



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 90 of 140)