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 93 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 93 of 140)
Font size
QR-code for this ebook


really treated as special purpose machines.

Tlie marketing cjuestion faced was of the form :

"Can you build me a program that will control this operation in
my oil refinery ? "

However, the minicomputer coidd also be pi-ogramed to perform
additional tasks. Digital equipment has lead the development of
minicomputers.

It started out marketing its macliines on a sale-only basis. This was
partly because it was a new company and had little capital aA-ailable
for financing and partly because a sales policy seemed very natural
in the type of market Digital was participating in.

Digital would usually sell the machines in parts; that is, you couhl
buy just the CPU without any systems software, or anything else, and
put the system together yourself. I think the current structure of the
minicomputer industry comes from its development as a specialized
machine.

Mr. OXeary. The suggestion has been made that competition in tlie
industry woukl be improved if a G^IAC type entity were created for
financing of the leasing of all computer systems manufactured. How
do you react to that idea ?

Dr. I^ROCK. In the sense that a leasing company can very logically
operate separate from a manufacturing company, I certainly agree
with it. If you mean would we have to, m some way by public policy,
try to get all computer leasing functions together in a single leasing
com])any, I can see no benefit in that.

We now have very effective leasing companies in a strong leasing
industry, which also has expanded into certain amounts of application
and consulting help.

So long as the computers are availal^le for sale on reasonable terms
I tliink we will see from the normal actions of the marketplace a sub-
stantial and useful leasing industry.

;Mr, OLkary. Thank you Mr. Chairman.



5649

Senator Hart. ]\Ir. Cluinibris.

Mr. CiiuMBRis. Thank you ]Mr. Chairman. Dr. Brock. I find certain
things in your paper tliat. because of legislative policy, there is a con-
llict of opinions. So. rather than getting into a debate with you, I just
wanted to note for the record, when we have an executive session that
it will be up to the Senators to determine whether your recommenda-
tions are appropriate or inappropriate for the public good. I think,
Dr. Granfield has some questions.

Mr. Granfield. I welcome fellow economists, and I am sure, as all
economists, we don't necessarily agree either because we don't under-
stand each other or because we see the world differently.

But. I think we would agree, as will everyone, Ave receive greater
efficiency witli lower prices as our ultimate goal. With the premise. Dr.
Brock, are you, or your thesis adviser, who I have been led to believe
is Dr. Ilouthakker at Harvard

Dr. Brock. Absolutely wrong.

Mr. Graxfield. Wrong?

Dr. Brock. That is what I would suspect was the reference from
Mr. Katzenbach's statement that I completed a thesis under one of the
(iovernment's advisers; Professor Houthakker is a Harvard professor.
I have had no direct contact with him. I have not even taken a course
from him. The only relation that I have had with him in a formal
sense is that he examined me on macroeconomic theory at one time, and
wage price controls.

Mr. Graxfield. AYlio was your thesis doctorate?

Dr. Brock. As I said at the beginning of the statement. Professors
Ivichard Caves and Marc Roberts.

Mr. Gr.\xfield. Are either of these gentlemen going to be involved
as witnesses in the Government's case; either you, Professor Caves, or
your other codoctorate.

Dr. Brock. Xone of us is scheduled to be witnesses at this time.

]Mr. Gr^vxfield. Thank you. I feel somewhat freer to pursue some of
my questions.

You list various entries in the computer industry. Let us assmne you
and I decided to go into the raising of racing thoroughbreds, and to
acquire capital to do so we go to bankers and present the argument
that because we are economists we know all about efficiency and we
are going to build a better horse. Do you think that we would be
granted a loan at the same rates as farmers ?

Dr. Brock. I doubt that we would be gi-anted a loan at all.

]Mr. Graxfield. T^t us assume we would be granted a loan. Do you
think it would be a higher rate of interest or lower?

Dr. Brock. If some banker would be foolish enough to grant me a
loan for racing horses it would be a higher interest rate, I would
presume — although I wouldn't necessarily comment on such a banker's
analysis. It might be lower.

Mr. Graxfield. Assuming it is higher. Why is it, do you believe,
the rate that we would be made to pay would be higher than the farmer
would be made to pay ?

Dr. Brock. Presumably, and the reason why I think we would be
turned down, is that they would not believe that we would be success-
ful in the business ?

]Mr. Graxfield. Too risky ?



6650

Dr. ijROCK. That we don't — I'll speak for myself, I don't know
about you now — that I would not have the managerial capabilities
to raise horses.

Mr. Granfield. Isn't that a barrier entry to our entering the
thoroughbred industry. No one will give us money. That is awful.

It is an obvious barrier to entry. It is discrimination.

Dr. Brock. I will certainly agi-ee to call that a barrier entry. I don't
think that has any great significance. The things in economics that we
try to examine are those that differ across industries. And one of the
things that is constant across any industry in any business is that you
must know something about the business in order to go into it. The fact
that you actually have to sell a product is also a barrier to entry. But
I don't think that has any relevance to economic analysis.

Mr. Granfield. You indicated in your analysis that there were
capital barriers to entry into the computer industr3\

Dr. Brock. Yes.

Mr. Granfield. Is there a capital barrier to entry for General
Motors to enter the computer industry ?

Dr. Brock. I would say tliere is some barrier to entry in this sense :
They presumably can raise the capital and go into it, but because of
the very large quantity of capital required in relationship to their own
capital base, they might at least think twice about entering it, or be
reluctant to because of the risk involved.

Mr. Granfield. So it is the risk. Is it riskiness that is a barrier, or
is there some wall that is there ?

Dr. Brock. I think that you are trying to trap me, but— —

Mr. Granfield. I would never want to do that. I only wanted to
bring out what exactly a barrier to entry is, because if there is a wall
there to prevent people from entering the computer industry

Dr. Brock. Let me explain what I mean by a barrier to entry. I
mean anything that raises the cost for a new company above the cost
for an established company. Now, it may not be true that there are
some barriers to entry that are insurmountable barriers to entry. If
IBM really has a production cost 10 percent lower than anybody else,
from greater technical efficiency, I would call that a barrier to entry.
It is a 10 percent barrier to entry in terms of the amount that IBM
could hold the price above the competitive profit level and still not
have entry.

But that doesn't mean that there is a wall, it is — you might say — a
hurdle of different heights for different people. Some barriers to entry
may only be a very low hurdle and some may practically be a wall.

Does that explain more clearly?

Mr. Granfield. That explains the point very well. But is it not true
for any new entrants to industry that they will probably face that kind
of barrier ? Is it somewhat of a disadvantage ? When you go into a new
industry, is there always a barrier ?

Dr. Brock. Yes; the thing that is important for economic analysis
is to determine the height of that barrier. In terms of the things that
we are talking about here, I would say yes; there is some barrier tO'
entering the minicomi')uter market. You have to get a little bit of
capital together, you have to have the technical expertise to make
machines, and in fact some people might call that a fairly significant
thing to get enough technical people together to compete with Digital



5651

Equipment, Data General, and other very highly qualified companies.

And 3'et that is a very low barrier to entry relative to the general
systems market where you have to do a lot more. So that I am talking
about relative height : It is not something that where you say there is
no barrier here and there is an insurmountable barrier here. It is
different levels of barriers.

Mr. Granfleld. What creates the height of that ; what is a critical
factor ?

Dr. Brock. Well, they all work together, but the most critical fac-
tor — and I think I would be in agreement with most of the other
witnesses that have spoken at these hearings — is the product differ-
entiation.

In particular, it is the difficulty of switching among computer sys-
tems. Because you have programs, data, operating procedures, job
control cards, and so forth, that are to one degree or another specialized
to your particular system, you must see more than the economist's in-
fuiitesimal price difference in order to move over to another supplier.

Now that is not an insurmountable barrier, unless you happen to
have everything in machine language. In many cases if you orient
yourself toward compatibility — that is, you write your programs in
COBAL and Fortran, and use ASCII coding on your data, and so
forth — it may only require a 10- to 15-percent advantage.

But that would be, I think, the very lowest price differential that
anyone would really even seriously consider moving for.

Mr. Granfield. Could I turn to the transfer cost.

Dr. Brock. Yes. Well, you could. It depends upon what you mean
by transfer cost.

Mr. Graxfield. Transfer of one system to the other.

Dr. Brock. Yes ; it is a cost of changino- systems.

Mr. Granfield. Is that transfer cost for all current members within
the industry when they approach customers without part of a system.

Dr. Brock. It exists for all current integrated systems manufac-
turers. Again, there are very diffei-ent transfer costs. If I have an
IBM ^PAi on my 360-30 and you come to me with a plug compatible
2314 and ask me to switch to your machine. I have a very low transfer
cost for that. I only have to unplug IBM's 2314 and put yours in. It
is much different if you ask me to replace my entire system with
another brand because I will have to change my programs, operating-
systems, and so forth.

Mr. Granfield. How could transfer cost be a barrier to entry to
firms already in the industry — if barrier to entry means barrier to
entry of any industry.

Dr. Brock. I don't understand your question.

Mr. Granfield How could transfer costs be a barrier to entry to a
firm that is already in that industry. How can that be a factor ge-
nerically affecting anybody ? Is that a barrier to entry ?

Dr. Brock. You could quibble over it on semantic grounds. "^^Hiat I
mean by these barriers to entry are difficulties either for a new firm
coming into the industry or for old firms gaining new market share,
which I consider entrv in a sense.

So that the. fact that Honeywell has difficulty gaining IBM cus-
tomers, or tliat IBM has difficulty gaining Honeywell customers, is^
related to the same problem.



5652

XoAY, whether you want to call it a barrier to entry or not is strictly
-a matter of semantics. I don't think it affects the reality there.

]Mr. Granfield. I reall}- thought that the use of a methodology of
barriers to entry was one which was used to explain why we didn't
see new entrants into an industry. For example, when people say,
"Why don't we see new entries into the soap industry," what are the
factors we have; what is relative here? The basic thing is brand
loyalty with respect to soap. But that loyalty doesn't affect the already
alleged firms. Do 30U understand^ If you don't understand we can
move on.

Dr. Brock. Well, I think maybe we can move on. But, let me just
say the brand loyalty, or the difficulty of switching, is a barrier to
entry to new firms. And I agree that is the normal way of using it.

And you can call this a barrier to entry or not, depending on what
you want, but it also makes it difficult to switch computers among
existing firms. So that it is difficult for a Honeywell or a RCA to in-
crease their market share, as well as other new firms coming into the
market.

Mr. Granfield. I have seen a commercial recently where a lady is
offered a can of deordorant, brand x. She is offered a carload if she
will switch deodorants to this other product, Superstick or something.
And she says she won't switch. Is that brand loyalty?

Dr. Brock. Yes ; I would say so.

]Mr. GR.VNFIELD. Is that the kind of brand loyalty professed to
customers of Honeywell ?

Dr. Brock. No. In one sense it is. but Honeywell's brand loyalty
is much stronger. The brand loyalty that we see in most consumer
products, and is traditionally associated with deodorants, soap, and
other things like that, is because you know that product. You often
don't have an absolutely determined utility function on exactly what
the product is, but you think it is good for you. So there is a fuzziness
there as to exactly what you are getting. So you are not willing to
switch to something else.

Xow the parallel to that in the computer industry is the difficulty
of measuring exact computer power. That is, you give me two different
computer systems and say without reference to a single program,
which one is better. If one is very much greater than the other, I can
tell you it is clearly better. But if they are relatively close, the kinds
of systems you would be considering in a competitive situation, I
can't tell you that without bringing in a specific program and trying
it out on them.

The measurement difficulty causes brand loyalty both in consumer
products and in the computer industry.

The second aspect, which is not there in consumer products but is
there in the computer industry, is the difficulty of switching. There
is no cost to me of trying our brand x deodorant. But, when I go to
switch complete computer systems, I actuallj^ incur a real financial
cost.

Mr. Granfield. I think that is an excellent distinction you make.
One is based upon loyalty to the product because of proven services
given to you, and you may be reluctant to change products because you
are not quite sure whether the other product will give that same
reliable level of service.



5653

The other ])oint j'oii mentioned is simply the teciinical costs. How
do we possibly determine which is operating and to what extent in
this particular market? How do we know customers don't switch from
IBM, or from Honeywell once they had Honeywell, because of the
tremendous risk involved in using another system versus the technical
cost of switching? How can we possibly sort out these two very
common complex questions so that we can make the statement that
the problem in the industry is system lock in; the technical transfer
cost of the problem rather than the customer's risk aversion to chang-
ing systems ? How can we possibly do that ?

Dr. Brock. Let me answer that in two parts. First, I am not sure-
that we have to, because I see both of those issues coming from the
same j^roblem of integrating systems selling. If you didn't have inte-
grated systems it would be much easier to judge exactly what the
system is doing for you. That is, no one has ever been able to deter-
mine the relative advantages of a Telex 5314 disk drive — I think they
call it the 5314— and an IBM 2314 disk drive, because they are desig-ned
to perform exactly the same thing and they are very similar products.

Xow the second part, assuming you don't believe that we don't have
to determine the costs separately. I think that you can determine the
relative effects of measurement difficulties and transfer costs by
analysis with the users. This is the kind of decision that computer
users are faced with all the time. They are always faced with making
a selection among computers. And they have to go through and assign
specific values to being a current system, or to the actual cost of the
transfer, and so forth.

So that it is not an insurmountable problem; you can get around
that.

Mr. Granfield. I'm just saying that I have never seen an}' analysis
of that. The risk involved to an EDP manager in changing a system,
the greatest risk is something will happen to the new system that he
feels incompetent to handle, and that is a tremendous risk because it
could mean his job.

I think that is just an internal bureaucratic view. It seems to me
I might not buy a system that may perform better rather than buy
a system that breaks down. He can be absolutely reprimanded for that;
but the sin of ommission to not buy the system that could perform
better is one that he may never l)e called upon to answer for. That
to me seems to be the greatest component of system lock in. It is not
system lock, it is risk aversion on the part of the manager. The more
complex the area the more risk aversion the manager will have.

Dr. Brock. I would fully agree with most of what you said, except
for the part of saying that that is the only thino- that is relevant. I
think that there are two parts to it. The uncertainty of exactly what
is going to happen with the new product is definitely a problem. And
that is why I would like to see the industry structui'ed along the lines
I have outlined, because it is much easier to know what is going to
h a pi^en with the new product .

]Mr. Granfield. If this total systems approach is evolved seemingly
because it is the most efficient way to sell that computer service, and
there is one firm that seems to perform the total systems function more
efficiently than the others, do you wish to change that so that we don't
have that total system offering except by firms who seem to not offer



5654

efficient!}^ as IBM? Isn't that a strange recommendation for an
economist ?

Dr. Brock. If I agreed with the two premises, then yes; it would
be a strange recommendation. I don't accept the first two premises,
that it is the most efficient method and that IBM is the most efficient
company doing it.

Mr. Granfield. You don't ?

Dr. Brock. No ; I don't.

iMr. Granfield. You don't feel that the industry is involved, that
this is the most efficient way of handling their business ?

Dr. Brock. No, I don't say they necessarily lose efficiency in this.
But, I think the reason for maintaining the entire systems market is
in order to enhance product differentiation and market control rather
than because it is more efficient.

jSIr. Granfield. How do we know that ? That is speculation.

Dr. Brock. In the sense that all economic analysis is speculation.

]Mr. Graxfield. Well, to me economic analysis is evolution in action.
Hi sorts out the less efficient ways toward more efficient ways because
it punishes inefficiency.

Dr. Brock. If you start with the assumption that all market struc-
tures are a result of efficiency sorting out the best structure, then ob-
viously it is silly to talk of ever rearranging the market structure.
That assumption is one that I don't accept, because I think that we
have too much evidence to the contrary. There are things such as the
IB]M case and a number of other industries where there is evidence
of direct market control of maintaining things that may not be di-
rectly inefficient, but you cannot say this current structure evolved in
order to produce efficiency.

Let me offer a bit of evidence. Some of this evidence is more speci-
fically documented in the chapter that I presented here.

Consider the question of computers. IBM found, at the time they
introduced the systems 360, that it was efficient to put a controller
between the channel and the peripheral device which would handle
the electronics. This helped produce the standard interface on the
channels so that you could put a number of different devices on it.

They found that to be efficient. That was done presumably for
efficiency reasons.

Once the plug compatible companies found that this standard inter-
face also helped them to replace IBM's products, then IBM started
moving the controller inside. I think it would be absolutely wrong
economic analysis to say they were doing that in order to gain more
efficiency, when you can see it was clearly a response to the competitive
attack on the product that was left exposed that way.

So that is the sort of thing I mean that is not totally without
evidence.

Mr. Granfield. In my reading of economic literature — and I am
becoming very concerned about this use of the term monopoly — I have
been led to believe that the way for a firm to achieve a monopoly is
that what they try to do is monopolize one level of vertically integrated
change of the industry. For example, the source of raw materials in
aluminum. And they are very concerned that the rest of the com-
munity remain a competitor. Yet, you are telling us — I don't know
whether you have used the term calling IBM a monopoly



5655

Dr. Brock. I don't believe I have.

]Mr. Granfield. Well others have committed that they haven't done
that. They are involved in vertically integrated parts of this business.
This is the new theory of monopoly. A new theory by literally monop-
olizing the entire market, or at least attempting to. This is redundant.
If you can monopolize one chain in the industry — it is redundant to
work like hell to do it. It is counterproductive.

Dr. Brock. No ; it is not really. IBM is a little more sophisticated
in that. It is tnie that if you monopolize one level of the vertical pro-
duction, then you can achieve a substantial amomit of monopoly rent
from that stage if the others are competitive.

But if you can also tie together several of those levels and monopolize
all of them, you can sometimes achieve a certain amount of price dis-
crimination that actually gives yon more monopoly profits.

So, I don't see any reason why they would not want to monopolize
all levels — if you want to use the word monopoly. I wouldn't apply
monopoly to the industrv^, but that is what you are

Mr. Granfield. I appreciate that. I wouldn't do that. One final ques-
tion. At the beginning of your paper you go to great effort to estab-
lisli your point of importance of various statutes. You talk about
capital barriers, you talk about economies of scale, and finally you talk
about brand loyalty, which I would correctly term as transfer cost,
of which there are two elements.

One is the changing systems risk and the other is the technical cost.

But when you come down to your final conclusion you state :

* * * Consequently, it appears that a lack of competitiveness in the main com-
puter industry is a result of integrated systems selling rather than concentration
in itself.

How does that apply to your barriers to entry argument?

Dr. Brock. I appreciate that question. What I meant by that state-
ment is that the barriers to entry are caused by the integrated systems
selling. And I tried to make that clear as I was going through what
the barriers to entry are.

In other words,' what I really meant to say at the end is that the
problems in the industry are caused by barriers to entry rather than
concentration. And that the barriers to entry are largely caused by
integrated systems selling. You don't see that? I am glad to give a
little more explanation here if you want it.

Mr. Granfield. Fine. Fine. I would appreciate it very much.

Dr. Brock. Let's take the problem first of what I term product
differentiation and you called transfer cost.

As we said, there are two elements to it. We considered the first
element the problem of the measurement difficulty — or what you
call risk aversion. If the industry is broken up into individual
functions — that is CPU's, peripheral products, maintenance and so
forth — it is much easier to measure exactly what you are getting.

If I look at one disk drive from IBM and one from Telex and they do
exactly the same thing but one is $10 cheaper, I know immediately
which one to buy.

If I look at two complete systems and I find that one is better on
its disk but worse on its tapes, and better on its operating system
"but worse on its applications software, and it is $10 cheaper, I don't
know which one to buy.



5656

But I can determine the best buy on each individual component.
I can do the same thing with the CPU's as with disks, this is a better
CPU or equal CPU and $10 cheaper, so I know which one to buy;
this maintenance service, and so forth.



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 93 of 140)