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 101 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 101 of 140)
Font size
QR-code for this ebook


criterion of economic performance when faced with a choice between objectives.
Generally, this means that objectives 2 and 3 have been emphasized at the ex-
pense of objective 1 because the economic performance gains from a small de-
crease in barriers to entry are not great enough to outweigh the economic costs
of substantial disruption. Because of the critical nature of computers in the
economy, significant user disruption could have consequences well beyond the
computer industry itself.

All of IBM's maintenance operations should be made into a single independent
company, referred to here as Maintenance Company. All of IBM's current
Field Engineering personnel, including managers and supervisors, would become
part of the new company. IBM's current inventory of spare parts for maintenance
purposes would be turned over to the new company. The Maintenance Company
would be given non-exclusive rights to the current maintenance manuals and
wiring diagrams for IBM equipment. Copies would also be made available to any
interested party at reproduction cost. All maintenance contracts currentl.v in
force with IBIM would go to the Maintenance Company. Maintenance for IBM
machines currently on rent would be assigned to the IMaintenance Company,
initially at current IBM rates, unless the customer chose to make alternative
arrangements. Where maintenance personnel now share office quarters with other
IBM services, the Maintenance Company could continue in the same quarters by
paying a commerically reasonable rent to the other company involved for a transi-
tional period, but the company should be required to relocate its personnel within
a reasonable time to assure independence from other IBM operations.

All of IBM's peripheral equipment operations should be separated into an
independent company, here designated as Peripheral Company. All manufac-
turing facilities for producing discs, tapes, printers, card readers, terminals, or
nther types of input-output equipment would go to the Periphei-al Company.
The company would also receive facilities related to non-integrated controllers,
but not for integrated controllers, memory, or CPU's. Research laboratories
related to peripheral equipment would also go to the Peripheral Company.
Where project,'* related to both periplierals and CPU's occur in the same labora-
tory, the laboratory should go to the predominant activity, with provision for
transferring personnel among the laboratories as necessary. All IBM patents
related to peripheral equipment would become the property of the Peripheral
Company. The company would receive all peripheral equipment owned by IBM
and not currently installed with a customer, but not the inventory of peripherals
on rent. All orders placed for IBM peripheral equipment would automatically
become orders for eqiiipment from the Periplieral Company, unless the cus-
tomer changed the order. The Peripheral Company would not be assigned IBM
sales or maintenance personnel, but all other personnel associated with pe-
ripheral equipment, including production, research, and management personnel
would be assigned to the new company. The company would receive non-
exclusive rights to manuals and other documentation of the interface between
peripheral products and CPU's for currently delivered products and future
products for which IBM has already begun peripheral development work. Copies
of tlie documentation received would also be made available to any inte'rested
party at reproduction cost. Future specifications could only be released to the
Peripheral Company at the same tiine and under the same tenus as they were
released to all intere.?>ted companies.

The third company, ^Marketing Company, would be IB]M to most customers.
It would receive the entire stock of IB:\I equipment installed on rent. Rental
contracts would be split into a maintenance comi)onent, assigned to Maintenance
Company, and an equipment rental component, assigned to Marketing Company.
Marketing Company would receive IBM's library of application programs, in-
cluding Type I, II, III, and IV programs-, and all programs classified as Program
Products except language compilers. It would not receive System Control Pro-
gramming and language compilers. It would receive complete rights to all man-
uals and documentation relating to its programs. It woiild receive nonexclusive
rights to the releases of System Control Programming and language compilers
which are currently in."vtalled without separate rental charge, together with
40-92T — 75 55



5694

their associated manuals and documentation. Otlier interested iiarties would
also receive rights to those programs and manuals at reproductiim costs. Market-
ing Company would not receive any rights to programs not assigned to it which
are currently under separate charge.

ilarketing Company would talve over all of IBM's branch offices and the
personnel as.sociated with them except for maintenance personnel. All Systems
Engineers, salesmen, data processing education personnel, and other people in
the IBM organization concerned with sales and customer service would go to
the Marketing Company. All IB^I programm(>rs assigned to programs of the
type taken over by Marketing Company would go to that company. As with
research personnel, care and flexibility woubl lie neces^sary to separate systems
programmers from applications programmers where both types currently work
in the same facility. However, if properly done, it should cause minimal dis-
ruption and hardship. The Marketing Company would receive no hardware
production or research facilities or patents. It would contract for maintenance
fr>ervices and hardware from other companies.

The fourth company. CPU Company, would receive the remaining IBM com-
]tuter industry assets and personnel. This includes the plants and personnel
associated with manufacturing compcments, memories, and CPU"s. and the re-
lated lalioratories and research personnel. CPU Company would receive all IBM
]iatents associated with components, memories, and CPU's. The company would
receive all IBM System Control Programming and language compilers, and the
IBM personnel assigned to program upgrades and program maintenance, or new
programs of that type. As with the Peripheral Company, the CPU Company
would receive all CPU'.s>, memories, or components currently owned by IB^I
but not installed with a customer. Outstanding orders for IBM CPU'.s and
memories would automatically become orders to the CPU ccmipany.

The division of IBM's stock could be handled simply by giving each share-
holder one share in each of the four comiianies for each sliare of IBM held.
I'ast problems with this method in antitrust dissolutions have been due to a
single dominant shiireholder who retained control over the new companies.
In IBM's case, the stock is widf^ly dispersed with no doininant shareholders,
so that no joint control over the four new companies could be exercised hy
stockholders owning sliares in all. IBM's ctirrent management should be divided
among the fotir new companies so as to minimize disruption. Once the manage-
ment was divided, a ne»v l)oard of directors could be cho.sen for each company,
subject only to the reiptirement that no person sit on the board of more than one
of the new companies.

A few temporary conduct restrictions would be necessary to instire competition.
Generally stated, the condition should be that the four companies could not give
favored treatment of any kind to each other. Specifically, this means that the CPU
Comi)any could not relea.se either interface specifications or maintenance infor-
mation to the other IBM companies without making it available to any other
company on the same terms. The CPU Compan.v could not provide parts on a dis-
criminatory basis. It could not refuse to deal or discriminate among customers in
prices or terms of .service unless the discrimination were price justified. The main-
tenance company could not sign an exclusive dealing arrangement with any of the
other companies that prohiliited it from maintaining non-IBM e(piipnient. Simi-
larly, the marketing company cotild n.of make any arrangements with the other
IBM companies which restricted its freedom to market non-IBM computers or
which restricted the freedom of Peripheral Company or CPU Company to sell to
other marketing or leasing companies, or directl.v to customers. Because the four
companies would each be dominant in their segment of tlie market at first, tliese
conduct restrictions would generally hold anyway under current interpretations
of the existing antitrust laws, but it could be advantageous to make the restric-
tions explicit to speed enforcement if the need sliould arise.

Some of IBM's operations (such as the Office Products Division which nmkes
ty))ewriters, dictating eciuipment, and copiers, and Scienc-e Research Associates
which produces educational materials) are not a i)art of the computer industry.
The non-computer ojierations could be either made a separate company or assigned
to any one of the four comjianies without .substantial effect on competition in
the computer industry. IBM's foreign operations have not been explicitly con-
sidered in this analysis. Further study would be reipiired to make a recommen-
dation for IBM's foreign assets.

The Maintenance Company would dominate the maintenance industry in the be-
ginning. However, it would have only minor economies of scale and some advan-



5695

t:\'j;e ill repiitiilidii as barriers to entry. Tlie basit- barrier to iiiaintenaiice entry,
IBM control over information and parts distribution. \YOuld be eliminated. The
Maintenance Company would write contracts with individnal users, hardware
manufacturers, or leasing or marketing companies. Although it would probably
only maintain IB^I e(juipiuent at lirst. self interest wouid cause the company lo
expand into the maintenance of non-IBM machines in the absence of collusion or
restrictive agreements among the former IBM companies. We could expect Main-
tenance Company to remain dominant in sparsely populated areas but to face
strong C(mipetition in major metropolitan centers, where population density prac-
tically eliminates economies of scale. Assuming Maintenance Company gave good
service and retained low enough prices to avoid being driven out of business, it
would probalily retain some advantage over independents simply through size and
visil)ility. but this would be a small enough advantage that it would not cause i)oor
performance.

The separation of ^laintenance Company is also designed to reduce barriers to
entry in segments other than maintenance by providing a relialile maintenance
service for the equipment manufactured by the new entrant. Some inedpendent
maintenance companies currently exist (such as Comma Cor] >o ration), but an
expansion of tlie independents as well as the willingness of the Maintenance Com-
pany to provide service to non-IBM machines is necessary in order to free pros-
l)ective entrants from the necessity of setting up their own maintenance services.
With maintenance separate. IBil C(»uld also lose control over machines it had
sold rather than rented. Currentl.v, IBM can control additions or modifications
to IBM nuK'hiiies which the customer owns l)y the threat to cut off maintenance
if tlie modification is harmful to IBM"s competitive pttsition."

Very little disruption to either customers or IBM oiierations should be caused
by separating tlie maintenance organization, so the reduction in l>arriers to entry
should result in a clear gain in economic efticienc.v. More definite competition
cimld be brought to the maintenance industry by separating Maintenance Com-
pany into several companies. l)ut the economic gain from such a move would prob-
ably not be very significant. If such a move were desiraltle, the Iiest method would
be to divide Maintenance Conii)any reg-ionally, and then trust to expansion to
jirovide <-ompetition among tlie resulting companies.

The current barriers to entry in the peripheral industry are due to IBM control
of the interface specifications between CPU's and peripherals and IBM's ability
to vary the profit margins on peripherals and CPT""s to keep the same systems
]irofit while reducing the profitability of peripheral production. Both of these
would l)e eliminated l»y separating peripherals into a separate company. Tlie
CPIT company would have to release interface specifications so that iieripli-
erals could be made and would be required to release them without favoring a
particular ctmipany. The Peripheral Company would have to charge a profitaI)le
]irice because it would have no other products to sell. The Peripheral Company
\vnuld retain a minor advantage over other companies through economies of scale
and reputation.

Although the (TT company would probalily continue to set interface standards
in the beginning, the separation of peripherals would be a strong impetus to
standMrdized interfaces lietween CPT''s and peripherals. A standardized interface
would be of great advantage to the user because it would allow him to select from
a wide range of CPU's and peripherals and put together the combination that best
fit his needs. With the current market structure, IBM has an incentive to
strengthen the tie between peripherals and CPU's in order to protect the exposed
iseripheral market. m:iking it unlikely that voluntary standardization efforts will
succeed. With the new market structure, all companies involved would find it
to their advantage to have a standardized interface, greatly increasing its
chances for success. The peripheral companies would desire it i>e<ause it would
eliminate the need to redesign their pr(tducts for every new CPU. and the CPU
companies would welcome a staudai-dized interface because it would eliminate
cdiicern over whether or not periphei-al products would be produced wJiich could
be used with their CPT'.

The Peripheral Company would have no sales or maintenance personnel at
first. It could sell its products thi-ongh the :Marketiiig Company, or other leasing
companies, and could develop a sales organization. A possibh^ v.-iriation of tlie
plan recommended here would be to give the Peripheral Company some siiare

1'' Sep E. Drakp T.iindell. .Jr.. "IinlepfndfMit Mpiiiorv M.ikprs Wcathcrcil Maintenanoe
Storm." CompntfT W'drld VI (Ma.v 24. liiTi'i. ]>. 40. for a report on IBM s thrpat to stop
niainteuauce on IBM machines to which over-sized memories were attached.



5696

of IBM's marketing personnel. However, it appears that there would be less
possibility for user disruption by keeping the current marketing orgamization in-
tact. The Peripheral Company would be strong enough in the market to survive
any temporary disadvantage relative to other peripheral companies that oc-
curred from its lack of an integrated sales and maintenance orgamization. No
restrictions would be placed on Peripheral Company's freedom to develop its own
sales and/or maintenance organization if it so desired.

The separation of Peripheral Company is designed to reduce barriers to
entry both in the peripherals segment and in the CPU segment. Currently,
there are a large number of independent peripheral makers who could provide
peripheral equipment to a new CPU entrant, but IBM's recent policies have left
the independents in a precarious financial and competitive position. With a large
independent Peripheral Company, we can be confident that companies will be
available which can supply a wide variety of peripherals to be used with any
CPU that comes on the market.

Software is an inherently difficult problem in any proposal to make the in-
dustry more competitive because of its intrinsic economies of scale. The free
market competitive system simply does not function efficiently when large econ-
omies of scale are present because it is impossible to have enough companies to
provide competition without losing technical efficiency. The traditional solution
ill the U.S. economy has been to grant such industries monopolies and put
them under regulation (electric power companies, pipelines, telephone com-
panies, etc.). If we followed this tx-adition. we would suggest putting all soft-
ware production under a single company and setting up a commission to regulate
its price. The author has rejected this solution for the computer industry because
of the observed stifling effect of government regulation on innovation and because
of the great variations that are possible in the quality of software. Although free
competition in software does lead to some wasteful duplication of effort, it seems
more effective in meeting the needs of computer users than a single regulated
software firm. The ability of thousands of small software firms to profitably
compete with the few large ones suggests that effective economies of scale in
software compete with the few large ones suggests that effective economies of
scale in software are not so great as the theoretical analysis suggests. In par-
ticular, economies of scale are far more limited in application programs than iin
systems control programs because of the great variety of users needs.

The difference in economies of scale between application programs and systems
control programs was the basis for the division of programming capability be-
tween Marketing Company and CPU Company. Competitive marketing com-
panies can write application software to supply the same functions as that pro-
vided by Marketing Company if needed, or other types of application software
which best meets the needs of their customers. However, all customers must have
the use of the system control programs in order to use the machine. Conse-
quently, if Marketing Company writes the control programs, it could effectively
force all customers to buy the control programs from it or cause the alternative
firms to take on the expensive and wasteful task of duplicating the system con-
trol programming.

Tlie Marketing Company would have some protection from competition because
of its reputation, software, and extensive consulting network. However, to break
those up enough to eliminate barriers to entry would cause considerable user
disruption. So long as the Marketing Company is kept intact, the user should see
little harm from the reorganization. Users who felt they did not need extensive
advice could make their own arrangements for various pieces of hardware and
maintenance service. Those who felt less confident of their ability to evaluate all
ef the potential equipment could arrange with Marketing Company or similar
competitive companies to find the best combination of equipment for them and
provide whatever help was needed in the way of programming and design. No
restrictions would be put on JIarketing Company's freedom to price its services.

A strong case could be made for forming several mnrketing companies from
the current IBM organization instead of only one. If INIarketing Company were
divided according to the current IB]M regions, little user disruption would result
because the local organizations would stay intact. At first little difference would
exist between a single marketing company and regional marketing companies, but
competition would increase as the regional companies expanded out of their
home regions. The primary reason for regional division would be to be sure that
MarlvCting Company did not make any official or understood restrictive agree-
ments with the CPU Company. If Marketing Company purchases from CPU
Company on an equal basis witli other firms, and also makes a sincere effort



5697

to provide the best equipmeut for its customers from all possible sources and
not just the IBM companies, then competition and economic performance will be
satisfactory with a single Marketing Company. If, on the other hand, Marketing
Company only offers IBM equipment to its customers, and CPU Company only
sells to Marketing Company, the performance will be unsatisfactory. Such a
restrictive agreement would be far harder to reach with many marketing com-
panies than with a single one. The disadvantage of further dividing Marketing
Company is that the efficiency of some of the current application programmer
groups could be disrupted by dividing them. There would also be some loss in
ability to shift highly specialized personnel to various areas of the country
when needed to solve unusual problems. The author does not have the informa-
tion available to distinguish between the relative benefits of a single Market-
ing Company with a prohibition of restrictive practices and of a set of regional
marketing companies with no restrictions on their conduct. In either case, the
marketing division would retain some market power but not enough to dominate
the industry.

As with the separation of peripherals and leasing, the separation of marketing
is designed to reduce barriers to entry in other segments than marketing. Because
Marketing Company would provide leasing services and customer contacts, it
would ease the problems of a company trying to enter any segment of the
computer business. One of the Marketing Company's functions would be that
of a professional evaluator of available equipmeut. It would be easier for a
iiew company with superior equipment to convince a professional evaluator
of that fact than to convince every customer. A firm with special capabilities
in a narrow segment of the computer industry could sell its products through
Marketing Company or its competitors without the necessity of setting up a
complete customer service organization and integrated product line.

CPU Company would have the most extensive economies of scale because
of its production of systems software. Because there are few if any economies
of scale in the actual production of hardware, it would be desirable to separate
CPU production from control program production in order to allow competitive
CPU makers the option of purchasing a control program. However, such a
separation appears to be technologically infeasible. The design of the CPU and
the control program are closely intertwined. Many functions can be provided
by either hardware or software. A CPU Company which could not produce its
(nvn control programs would be forced to design new CPU's so that they would
be iisable with the existing control programs in order to be sure that a control
program would be available when needed. Similarly, the control program writers
would be forced to write programs according to the specifications of existing
computers. Inefficiency and a slowing of technical progress could be expected
from separating CPU and control program operations.

If the companies are not separated, there does not appear to be a way to
guarantee alternate CPU makers access to the CPU Company control programs
without extensive regulation. A simple requirement that the CPU and Control
Program be priced and sold separately without making one conditional upon the
other would not be sufficient. The CPU Company could put most of the price
on the control program (which is fairly protected from competition due to
economies of scale) and a lower price on the actual CPU which would have no
protection from competition. This problem could be solved by requiring the
company to make the same profit rate on programming and CPU manufacturing,
but that would restrict the company's ability to respond to market forces and
require regular supervision. It seems better to put no restrictions on the pricing
policies of CPU Company (other than the exclusion of restrictive agreements
with the other IBM companies mentioned above) and accept the necessity for
a competitive CPU maker to produce its own operating system. This restricts
the freedom of very small firms to enter but is not an insurmountable obstacle
to medium size firms as evidenced by the success of IBM's competitors in
producing effective and innovative system control programs.

Although barriers to entry in the CPU segment of the business would still
exist, they would be greatly reduced because of the existence of other companies
to provide maintenance, leasing, marketing, and peripherals. The necessity
to build an integrated organization with its consequent huge capital costs and
long period of losses would be gone, and the new entrant would be faced with
the basic problem of providing a needed product more cheaply than the existing
companies. The scope for exploitation of technical breakthroughs by a small
company would be greatly expanded because it would not have to provide all
the services needed by the customer itself. The existence of alternate CPU



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 101 of 140)