John Irons's Blog


Economic News, Data and Analysis

The Economics of Networks: Interview with N. Economides

Following up on my past feature on Network Externalities
(feature: 3/1/98), I have put together this series of Q&A’s with
Professor Nicholas Economides. This interview is broken into three parts.
The first discusses networks and network externalities in general. The
second part looks at networked industries and industry structure. The third
part looks at policy and the microsoft case in light of the economics of


Nicholas Economides is Professor of Economics at
the Stern School of Business, New York University. He earned his Ph.D.
from University of California at Berkeley and previously taught at Columbia
University and Stanford University. 

Economides has published extensively on networks, network pricing, market
structure and other topics. He recently edited a special issue of the International
Journal of Industrial Organization
on Network Economics and is currently
working on a book entitled The Impact of the 1996 Telecommunications Act
to be published by the American Enterprise Institute and M.I.T. Press (1998). 

He also maintains web sites on the Economics
of Networks
and on the current Microsoft
Legal Battle


On networks in general


Q:  First it would be nice to clarify a couple of definitions. What
are the defining characteristics of a network?


Networks are composed of links that connect nodes. It is inherent in
the structure of a network that many components of a network are required
for the provision of a typical service. Thus, network components are complementary
to each other. Typically a number of components are required to produce
a service. These components may be produced by different networks. For
example, a telephone call from New York City to San Francisco starts over
the circuits of New York Telephone (Bell Atlantic) until it reaches a local
switch in New York City, wherefrom it is picked up and carried by a long
distance company (such as AT&T) to a similar switch of Pacific Bell
(SBC) in San Francisco. Then Pacific Bell carries it to the San Francisco
subscriber. So, the call requires the use of the circuits of New York Telephone,
AT&T, and PacBell, and (at least) two switches. These are all complementary
components of the network. See the example of the “information

Typical examples of networks are telecommunications networks (fixed
and wireless), fax networks, credit card networks, ATM networks, railroads,
electricity networks, etc. Network features, including the existence of
“network externalities” (see below), are discussed in more detail in “The
Economics of Networks”


Q:  And what are network externalities?


Networks exhibit positive consumption and production externalities.
A positive consumption externality (or “network externality”) signifies
the fact that the value of a unit of the good increases with the expected
number of units to be sold.


Q:  How do network externalities arise? Please give examples.


Network externalities can arise directly or indirectly. For example,
in a typical telecommunications network with n subscribers, the
addition of the (n+1)th subscriber allows for 2n
additional types of calls to be made (from the new subscriber to all old
subscribers and vice versa). Clearly, the number of additional calls (2n)
that become possible when the (n+1)th customer is added
increases with the size of the network n. If each of the calls has
a fixed value, it is clear that the value of adding one more customer increases
as the network size (n) increases. This is an example of a direct
network externality.

Network externalities arise directly in telecommunications and fax networks
as well as in the internet, financial exchange (stock market), credit card,
and ATM networks. Network externalities also arise in “virtual networks”
where two complementary components are required to make a valuable good
or service. Typical examples of virtual networks are the combination of
compatible software and hardware, or the combination of a computer operating
system (“OS”) and applications software that run under the OS. In these
cases, the value of an operating system is higher when there is a large
number of applications that can run under it. Conversely, an application
has higher value if it runs on a widely-accepted OS. The combination of
these effects makes the value that a consumer receives from an OS higher
when the sales of the OS are higher. Similar reasoning shows the existence
of network externalities in any good that requires a distribution or a
repair network, or, more generally, requires a complementary good.



Q:  It seems that many industries exhibit network externalities. Where
are network externalities crucial? 


Some products have no value if there is no network. For example, a
fax machine is useless unless there exists another fax machine to communicate
with. In contrast, a computer program, say WordPerfect, has value even
if no one else bought it. Its value increases with the number of users
(because this enhances the availability of in-house technical support and
the availability of secretaries trained in WordPerfect), so there are network
externalities, but this may not be the most important determining factor
of the value of the program.



Q:  Do you think that the importance of networked goods and network
industries in the economy is growing?


Network industries and network goods are a very significant part of
the US and world economy since they encompass key industries, including
telecommunications, computers, electricity, and transportation. These industries,
especially computers and telecommunications are growing very fast. They
are driven by the steep reductions in the costs of integrated circuits,
the progressive digitization of many functions, and the wider use of programmable
multi-purpose devices. Network industries are expected to continue growing
at a faster pace than the US economy.



Q:  How will the workings of the economy begin to change in response
to the increasing expansion of network industries?


Some of the changes are already here. The World Wide Web has completely
transformed the information industry. Information retrieval and information
gathering and mining have become much easier. Multiparty communication,
chat, conferencing etc., although feasible earlier, became widely available
through the web. From a social point of view, the World Wide Web is especially
useful in bringing together individuals around the world that share a common
interest. The WWW also facilitates trade. Although electronic commerce
is still at its infancy, distribution of books, music CDs, software, as
well as securities trading have taken off with significant success. Interactive
selling is just starting. Customers already have the ability of get live
simultaneous offers on automobile loans by different firms on the web.
There is an increasing competitive pressure for better price and higher

As more information becomes available to the average household, the
deciphering of what is important and relevant becomes increasingly difficult.
The need for services that make selections for users (or, alternatively,
facilitate users’ search) is ever more important. I expect that some of
the most profitable businesses on the web will be those that guide customers
to information. There is much to be done in this direction. Despite the
good efforts of Yahoo, there are still no authoritative “yellow pages”
on the web.

Changes in network industries require more flexibility by industry.
The response of some traditional industry players to change is sometimes
far from in tune with the times. For example, the demand for telecommunications
services is currently much higher than ever envisioned, partly because
of internet use. But as more households demand a second or third telephone
line, traditional local telephone companies (for example, Pacific Bell
in California) complain that callers stay too long on line. Companies have
to adjust to the times!

In the telecommunications sector, the Telecommunications Act of 1996
was been a brave attempt by Congress to bring the legal framework of the
telecommunications industry closer to a natural competitive environment
as shaped by rapid technological change. A key provision of the 1996 Act
was the facilitation of entry in the local market, which is presently dominated
by the local telephone companies that emerged from the 1984 breakup of
AT&T, and by GTE. Each of these companies is a monopolist in its local
market. Unfortunately, over two years after the passing of the 1996 Act,
armies of lawyers of the incumbent monopolists have fought against entry
in the local market, and very little entry has been realized, to the great
detriment of consumers.

On Networked Industries and Market Structure


Q:  Is seems that when significant network externalities exist, there
may be a natural trend towards monopoly or imperfect competition. What
factors influence the path of market structure in network industries?


In all economic activity history plays some role. For example, the
AT&T trademark assures consumers of a certain level of competence and
customer service. Similarly, when two operating systems of the same functionality,
quality, and efficiency are offered at the same price, consumers
will tend to buy the OS for which there are more applications. Microsoft
and other OS sellers spend a significant amount of money and effort to
make sure that independent software companies write applications for their
operating system. An OS seller may also keep its price low to guarantee
wide acceptance of its OS so that independent software companies anticipate
a large market and write software for this OS. The market outcome depends
on historical market share, but most importantly, it depends on the pricing
strategies of the competitors and the availability of software compatible
with each OS. The winner is often the company that is willing to sell at
a low price for a long time to gain market share and reap the externalities.
The loser is typically the firm that sells at a high price, receives short
term profits, but loses in the long run. Being first is useful, but persistence
and the correct pricing policy are more important.

In summary, we have the paradoxical situation where the winning firm
(that has high market share) is at the same time the company that sells
at a low price. Sacrifices in price pay off in higher market share, and
market share in more valuable when the industry has network externalities,
since higher sales signify higher value. In this process, consumers benefit
from the low market price.

I have shown in theoretical models (see “Compatibility
and Market Structure for Network Goods”
) that monopoly (or a very concentrated
industry) can be a natural free-entry equilibrium in a market where there
are very strong externalities. In such a market, at a natural equilibrium,
the leading firm can have three times the market share than the second
firm, the second largest firm can have three times the market share of
the third firm, and so on. Moreover, sometimes it is better for society
to have a more concentrated industry because then society realizes the
benefits of high network externalities of the leading platform.



Q:  From a welfare maximizing point of view, should we prefer one market
structure over another? 


My paper “Compatibility
and Market Structure for Network Goods”
shows that sometimes a more
concentrated market results in higher social welfare than a less concentrated
market. A judgement has to be based on the very specific features of a



Q:  It seems difficult to forecast the financial success of a producer
of a good subject to network externalities. Do network externalities imply
a “winner take all” industry?


A network consists of many complementary components. The use of a number
of components is required to produce a final service. Often the components
of the network have their own markets. A firm can be most successful if
it can monopolize the market for one crucial component while markets for
complementary components are competitive. The failure of analysts to distinguish
between promising and not promising ventures arises from a failure to adequately
define the market for the final service or from a failure to identify the
extent of market power in each of the markets for the components.



Q:  You have, in past research, focused on interesting issues raised
by networks such as technical compatibility and standards, interconnection,
pricing, etc. In what direction is your current and future research headed?


I am currently doing research in telecommunications and on network
industries in general. The latest papers are on the “Economics
of Networks”
web site.

On Policy


Q:  What challenges do network industries raise for public policy? 


Network industries have features I discussed above that should be taken
into account in giving public policy advice. Also, the economic analysis
of networks is a relatively new field in economics, and I do not think
economists have an all-encompassing analysis of the field.

The goal of public policy should be to maximize economic efficiency.
Economic efficiency has three components: (i) allocative efficiency; (ii)
productive efficiency; and (iii) dynamic efficiency. To maximize allocative
efficiency, goods must be sold at prices close to cost. To maximize productive
efficiency, goods should be produced in the most efficient way possible.
To maximize dynamic efficiency, the industry must maximize innovation and
growth. In general, these objectives may be in conflict, and there may
not exist a policy that maximizes all of them.

Like for any other industry, there is a need of a deep understanding
of the particular network industry before offering policy advice or imposing
rules and regulations. This is especially important because many network
industries, including the computer industry, exhibit high growth. Elementary
or static models of economic analysis may miss the essence of the market
and lead to social losses.



Q:  How active should the government be in following these policies? 


The government should intervene when there are clear violations of
the antitrust law. The government should also be relatively cautious, taking
into account that economists do not yet have an all-encompassing analysis
of the field.


On Microsoft


Q:  You have been following the details of the Microsoft
on part of your website. Do you have any predictions you’d like
to share?


A full answer to your question will take many pages of analysis. I
will only say briefly (and enigmatically) that I do not expect the outcome
of the present case against Microsoft to result in any significant change
in the market structure in the OS market and other software markets.



Q:  You argue in a recent paper that a monopolist has incentives to
raise the costs of rivals in complementary markets and that Microsoft may
have done this by bundling IE4 with Windows operating systems. Can you
explain the motivation for this action and the social welfare consequences?


A firm that monopolizes an essential input and also produces a complementary
good has an incentive to raise the costs of rivals in the complementary
good market. For example, a local telephone company, such as Bell Atlantic
has an incentive to raise the costs of a long distance company, such as
AT&T, once Bell Atlantic is allowed to provide long distance services.
Microsoft has similar incentives, i.e., to raise the costs of applications
rivals. But there is also a significant difference between the case of
a local telephone company and Microsoft in the extent of their ability
to increase rivals’ costs. There are presently no alternatives to telephone
access at home except through the local telephone company, which can raise
the costs of long distance or local rivals to the extent it wants. On the
other hand, Microsoft has a long run commitment to an open system. When
independent software providers write applications for Windows, the value
of Microsoft’s operating system increases. Thus, Microsoft discloses hooks
(APIs) that allow the independent programs to run under Windows. Thus,
there are market substitutes for key Windows OS functions. For example,
Symantec sells a replacement for Windows Explorer and Desktop. Because
of the existence of such substitutes, the extent to which Microsoft can
increase browser rivals’ costs is limited.



Q:  The Microsoft case has received a considerable amount of attention
in the press recently. Are there any common misconceptions about the economics
of the Microsoft case?


A common misconception is that “path dependence” and “history” determines
everything. Even when history plays a big role, companies have the opportunity
to determine their fate by using pricing policies. If everything were determined
by path dependence and history, Apple would be where Microsoft is today,
in terms of market share in the OS system market. Apple made huge and repeated
errors in its decisions on pricing and compatibility, resulting in its
present marginal position in the industry.

A second type of common misconceptions arises from isolating a particular
fact or a particular statistic from the OS market without checking for
consistency with other facts. Economic behavior has to be consistent and
make sense in the context of the present market structure and the objectives
of the players. Journalists expect consistency in mathematics and physics
but hardly require it in economics. For example, it is not sufficient to
note the large market share of Windows. One has to also offer an explanation
for the relatively low price of Windows, and this explanation has to be
consistent with the observed high market share of Windows.

A third type of misconception arises out of lack of understanding of
basic economics and basic antitrust law. For example, the fact that the
Stern School of Business has a monopoly (100% market share) in Stern MBA
degrees hardly gives Stern monopoly power in the MBA market in the eyes
of economists or in terms of antitrust law. Why? Because there are close
substitutes, i.e., MBA degrees from Columbia, Harvard, MIT, etc. Economic
and legal determination of anti-competitive conditions and actions requires
a deeper examination of the industry.

More links

Filed under: Microeconomics

Network Externalities and the “New Economy”

A central idea in much of the New Economy thinking is the idea that
certain new products being produced today increase in value as more people
use them. A familiar example is the internet. If you are the only one with
an internet connection, it isn’t worth very much since there are no others
to email with or any web sites to visit. As the network expands, and friends,
family and get connected to the internet, the connection becomes
more and more valuable. These effects have been termed “network externalities”,
and they may have important effects on the workings of the economy.Cable

The term “externalities” refers to the indirect effects that one user
has on the other users of the same network. For example, when you link
up to the internet, you are indirectly increasing the value everyone else’s
connection since they can now send you email.

This idea is also linked to a related idea typically called “path dependence”.
If the value of a product in question grows with usage, then it pays to
be big. If a product, for any reason whatsoever – including historical
accident – gets enough users early on, then the product may be favored
over others of higher “innate” quality just because it was there first
and has an established user base.

The typical (but misleading*) example here is
the competition between VHS and Beta VCR’s. As VHS initially gained a majority
of users, it became a more valuable product simply because others used
it (for example you were more likely to be able to trade tapes with friends
if you had a VHS). Path dependence is then the idea that the current dominance
of VHS, over a technologically superior Beta format, is a result of a simple
historical accident. If BETA had an initial lead it too could have taken
over the market.

Similar examples could be made with other high-tech products like Microsoft’s
Windows operating system. The role of network externalities is also playing
a central role in the war over internet browsers and modem protocols –
the players in the market know that whoever gets biggest fastest will probably
end up dominating the market.

Is this a really a new feature of the Economy?

Not really.

There have always been networks and network externalities associated
with products in the economy. The telephone network as well as the railroad
and interstate “networks”, to take three large examples, are not exactly

The important question is whether or not the new economy is producing
relatively more of these products than in the past. Casual observation
suggests that they may be taking on a more important role, since high-tech
industries often display some of the characteristics needed to generate
network externalities.

What do the changes mean for the economy?

As the economy shifts to the production of network related products
we must also ask ourselves what this implies about the working of the economy
and what kind of policy the government should pursue.

The first important issue to consider is market structure. If network
externalities imply that bigger is better, then more and more “natural
monopolies” will arise. Consider the case of Microsoft’s dominance of the
operating system market. As MS Windows took over the OS market, in part
due to the presence of network externalities, Microsoft became virtually
a monopoly producer of computer operating systems.

We are used to thinking that monopolies are bad – since they often have
the ability to raise prices and extract monopoly rents from consumers which
may then lead to inefficient levels of production. However, in the case
of a “natural” monopoly, we would like to have only one producer since
this is the best way to efficiently produce the product. One way to deal
with this problem is to have the government run things – by outright ownership
or heavy regulation – as in the case of the post office or the phone and
cable industries pre- deregulation.

If government ownership or heavy regulation isn’t a practical option
then what? Answering this question will be a major policy challenge for
the new economy.

Further Readings:


* The VHS-BETA example is perhaps the second
most common story used to illustrate path dependence (the first most common
being the QWERTY keyboard). It is slightly misleading in that the two technologies
were not exactly the same except for a historical “accident” or two. First,
despite slightly lower picture quality, VHS had a longer running time of
2 hours meaning that it could comfortably hold most movies on one tape
– so the choice of VHS may be more than just a historical accident. Also,
other factors such as pricing policies and licensing strategies may play
important roles in determining the ultimate winner. See Economides’ Letter
to the Wall Street Journal on Path
for more.

Filed under: Microeconomics