Download Evaluating network externalities

169
Suomi, Reima (2006) Evaluating Network Externalities.
In Remenyi, Dan –Brown, Ann (editors)
Proceedings of the 13th European Conference on Information Technology Evaluation (ECITE), Genova,
Italy 27-29.9.2006, 437-442. ISBN 978-1-905305-32-2 (print), 978-1-905305-34-6 (cd).
EVALUATING NETWORK EXTERNALITIES
Reima Suomi
Turku School of Economics
Abstract
Networks externalties, both positive and negative, are known to be part of every network.
Positive network externalities are the main reason for building networks, and a primary
source of wealth for the modern society, often actually called the networked society.
Network externalities stemming from telecommunication networks are a primary source
of wealth in the information society.
This article discusses what kind of networks we have in the society, what network
externalities are, and in which varieties they exist. Further the article draws some initial
ideas on how network externalities could be measured. The article identifies four major
problems in network externality measurement: the stakeholder problem, the identification
problem, the arbitrator problem and the measure problem.
Introduction
Network externality has been defined as a change in the benefit, or surplus, that an agent
derives from a good when the number of other agents consuming the same kind of good
changes (Liebowitz & Margolis).
According to a recent research, network externalities are on of the 10 key forces that
drive information technology to the society (Andal-Ancion, 2003). For economists, the
theory of network externalities, or network externalities, or standardization, has wide
applicability. Indeed, it has fundamental importance for competition policy, regulation,
business strategy, intellectual property, and technical change in a wide range of
industries; developments in these industries cannot be fully understood without an
understanding of network externalities (Besen).
Network effect are a popular and important theoretical concept, yet very much neglected
by the information systems research community. Because of this, there remains a risk
that their operationalization of network externalities in the field of information networks
is not conducted properly. This article discusses what different methods there could be to
operationalize, measure and evaluate network externalities.
This article comes up with four fundamental problems that are met with when network
externalities are studied: the stakeholder problem, the identification problem, the
arbitrator problem and the measure problem.
Networks externalties, both positive and negative, are known to be part of every network.
Positive network externalities are the main reason for building networks, and a primary
source of wealth for the modern society, often actually called the networked society
(Castells, 1996; Stalder, 1998). Network externalities stemming from telecommunication
networks are a primary source of wealth in the information society. Taking this seriously,
it is astonishing how little effort information system researchers have devoted to
understanding network externalities. One reason might be that network externalities have
traditionally been a playfield of economists, and information system researchers have felt
unease at the field. This, however, needs not to be the case. Aside the rather theoretical
discussion on network externalities the economists run, a more practical and operative
approach to the issue is needed. This is aimed at in this article, actually just presenting
some preliminary and initial thoughts on a research agenda to be run over the coming
years.
According to a recent research, network externalities are on of the 10 key forces that
drive information technology to the society (Andal-Ancion, 2003). For economists, the
theory of network externalities, or network externalities, or standardization, has wide
applicability. Indeed, it has fundamental importance for competition policy, regulation,
business strategy, intellectual property, and technical change in a wide range of
industries; developments in these industries cannot be fully understood without an
understanding of network externalities (Besen).
Network effect are a popular and important theoretical concept. However, there remains
a risk that their operationalization is too difficult to perform. This article discusses what
different methods there could be to operationalize, measure and evaluate network
externalities.
What are networks?
Networks exist in many varieties and forms. The microstructure of the underlying
network of connections (physical or virtual) can influence the strength of network
externalities (Sundararajan, 2003). Common for all networks is the existence of two or
more network nodes, and some kind of relationship between them. Networks can be
physical or abstract, and actually all networks exhibit some kind of abstract
characteristics, and a network having physical components can be seen as a special case.
More specifically, the abstract/physical characteristics can materialize both in the nodes
and in the relationship/connections between the nodes.
Figure 1
Types of networks
Typical physical networks are those of telecommunications, including the Internet,
mobile phone networks, local area networks, Plain Old Telephone (POTS) networks, etc.
Many of networks have to do with the society infrastructure, such as railway, electricity,
water, sewer or road networks.
The relationship between the network nodes need not to be physical. We so speak of
airport networks, bank networks, retail outlet networks, hotel networks etc. These
networks exist and are valuable as such, even though they are usually connected by
several physical networks such as telecommunications or roads.
Sometimes the relationships between the nodes are more important than the nodes.
Actually, the network nodes can sometimes be hard to see. Take the example of a road
network. Actually it is hard to see any specific nodes that would act as junctions of the
connections. The roads are the connections itself, and they connect cities etc. However,
we do not speak of a city network but of a road network. One way to look at the issue is
to understand each road (usually with an assigned number in the national road system) as
a node in the network. In that case, the junctions come in as connections. This example
shows how difficult the terminology is.
Abstract networks are many. A good example is the art of formulating astrons. No real
connection of any kind exists between the stars in the astron, but an abstract network is
formulated. In this case, the nodes themselves are still physical. A mother of all network
with abstract nodes and abstract connections between them is that of the holy trinity.
Networks can have well-defined or vague boundaries. An example of a well-defined
network should at lest in theory be an intranet or local area network of a company, that is
separated from the rest of the telecommunication network by a firewall. The network
administrator should be able to exactly say what are the nodes of the network. A vague
network is for example that of an academic community, say information system
researchers. Both the potential network nodes as external evaluators can have different
ideas on who or what is a part of the network and who or what not.
All networks are dynamic at the end, also change their nodes and relationships between
them. For some limited period, a network can however be seen as static.
A summary of our network definition is in Table 1.
Table 1
Characteristics of network
• Consist of nodes and relationships/connections between them
• All networks exhibit abstract characteristics, some have even physical nodes
and/or connections between them
• Network boundaries can be vague or well-defined
• All networks are dynamic at the end, but can look like static for a certain
period
________________________________________________________________________
Individuals or organizations (units) can have different roles as it comes to networks. The
most clear role is that of being a member, node, of a network. Other typical roles are
those of owner of a network, administrator of a network, or customer of a network.
Depending on the situation, might be that a network has no customers, owners or
administrators. A unit can be simultaneously in several roles in a network, even
simultaneously a network node, owner, administrator and a customer.
Needless to say, each unit is a member of several networks, sometimes even
unconsciously, and each unit has several roles to several networks.
What are network externalities?
Network externality has been defined as a change in the benefit, or surplus, that an agent
derives from a good when the number of other agents consuming the same kind of good
changes. (Liebowitz & Margolis). The roots of the network effect research are in the
marketing discipline, where it was understood that the success of a product or service is a
phenomenon strengthening itself. The phenomenon was called the bandwagon effect by
which was meant “the extent to which the demand for a commodity is increased due to
the fact that others are also consuming the same commodity. It represents the desire of
people to purchase a commodity in order to get into ‘the swim of things’; in order to
conform with the people they wish to be associated with; in order to be fashionable or
stylish; or, in order to appear to be ‘one of the boys.”(Leibenstein, 1950) Still today, the
network effect is often connected the act of buying and selling, and not the the act of
consuming, as above: “A positive consumption externality (or network externality)
signifies the fact that the value of a unit of the good increases with the number of units
sold” (Economides, 1996). Another definition stressing buying is that of: “Network
externalities arise when a consumer values compatibility–often stemming from ability to
take advantage of the same complements–with other consumers, creating economies of
scope between different consumers’purchases”(Farrell & Klemperer, 2006).
One should make a difference between network effect and network externality. Network
externalities should not properly be called network externalities unless the participants in
the market fail to internalize these externalities (Liebowitz & Margolis). An externality
is the effect of a transaction between two parties on a third party who is not involved in
the carrying out of that transaction. Internalizing an effect means that it is no more
directed towards a third party.
Network externalities can be direct or indirect, and positive or negative, we arrive at
Figure 2.
Figure 2
Types of network externalities
Direct network externalities exist when an increase in the size of a network increases the
number of others with whom one can “communicate” directly. Indirect network
externalities exist when an increase in the size of a network expands the range of
complementary products available to the members of the network (Besen).
Network externalities can be positive or negative. A typical negative network effect is a
traffic jam. All too often network externalities are understood just as posivite. The same
phenomenon can be both positive and negative, depending on the role of the observer.
To take an example, to a railway operator having a lot of customers is a good thing (more
revenue), but for the customer the same situation can mean congestion, also an negative
effect.
The enchantment of network externalities is that they often come out as surprise and as a
byproduct that was not calculated or foreseen in any way.
How to measure networks
Networks as such are already complicated objects for measurement, not to speak of their.
Typical measures of networks are geographical coverage (materializing in concepts such
as WAN, MAN, LAN), or the length of a network, say for example the length of a road
network in a country. A basic indicator is the number of active and passive nodes in the
network, even though for example in a telecommunication network it is actually difficult
to say which node is a passive and which one an active one.
An active
telecommunication network node is expected to have some decision capacity, usually
programmed to the device.
A group itself are the different indicators showing the output or performance of the
network. A telephone network might be evaluated through the number of calls taken in a
certain period of time, or a liquid pipeline through the amount of liquid it transfers. A
more complicated measure would take into account the performance. A telephone
network might be evaluated based on the phone minutes in a certain period of time.
More complicated, say an optic cable can be of very different capacity. A teleoperator
might report the amount of shield kilometers dig to the round, not taking into account
what kind of cable was actually laid. In the railway industry, train transport performance
is measured in axis kilometers: a train with 100 axes running 50 kilometres would
perform the same as a train with 50 axes running 100 kilometres.
The age of a network is too a complicated concept. It is typical to refer to the highest
documented node age of the network., take an example from the London underground:
“London Underground was formed in 1985, but its history dates back to 1863 when the
world's first underground railway opened in London” (London underground. History.,
2006). However, counting some average age for a network can be a haunting task,
especially if the relative weight (what might that be) of each node has to be taken into
account.
To summarize, measuring networks is difficult. Overall measures that might allow for
comparing different types of networks are rare if any. The number of network nodes is
the only actual comparable figure, but even it is sometimes had to define.
How to measure network externalities?
We finally arrive to the crucial and actual question of this paper, how to measure network
externalities. Unfortunately, in this paper we are not yet ready to deliver an answer.
What we can do is identifying four kinds of problems:
• The stakeholder problem
• The identification problem
• The arbitrator problem
• The measure problem
By our definition, the network externalities should be materialized to an external, third
party, of the transaction taking place in the network. Already identifying who are the
third parties can be a hard task. If we for example pollute, it is hard to find out who
exactly will suffer from that transaction. This we call the stakeholder problem.
The second problem is to identify the available network externalities, analyzing the issue
from the viewpoint of the different stakeholders identified first. As already discussed,
network externalities often materialize themselves in an unexpected way and first after a
long maturing time. Most probably the direct negative externalities are to be identified
first by the customers. The network owners and operators, also those trying to gain
benefit from the networks, are the first to identify positive network externalities, both
direct and indirect. The indirect negative externalities might be the most difficult to
identify. The problem of identifying network externalities is here called the identification
problem.
The third problem we take up is called “arbitrator” problem. The externalities are not
always direct, but materialize themselves over complicated value chains. Understanding
these value chains, especially in the case of the indirect network externalities, is a major
problem.
We have to measure the network externalities in some scale. Network externalities
discipline has its roots in economics, and so a natural conclusion would be to measure
everything in money. In reality, this can hardly be done. For example, in the case of the
popular example pollution, what is the economic value of an unpolluted environment.
Another possibility of measuring network externalities would be that of giving them
assessed values by the stakeholders on different scales. For example, through pairwise
comparisons stakeholders could provide some ordinal scale on which network
externalties can be ordered. One further possibility is to measure the importance of
different network externalities on a likert-like scale in the case of different stakeholders.
One of the lessons is that mental evaluations of network externalties can direct network
behaviour more than values presented in money – values which anyway can be arbitrary
and artificial.
Most obviously network externalities materialize themselves to different stakeholders in
different scales, and thus different measures are needed. Mastering this mess we call the
measure problem.
To come to some kind of answer to the question posed in the beginning of this table, the
process as depicted in Figure 3 is needed in the evaluation of network externalities.
Figure 3
The process of network externalities evaluation
Further research of mine, hopefully of many others too, will further elaborate this
process.
An example
In order to give an example of the network externalities evaluation, let us discuss the case
of a local area network, which by definition is a limited and small network.
Stakeholder identification
The stakeholders of the network are at least three: the users, the network (operative)
administrator, and the owner of the network (the one paying the final bills).
Value chain analysis
The value chain of externalities materialization can already get complicated. A
simplified value chain could be that of delivering the infrastructure, delivering resources
using the infrastructure, and sharing resources through the infrastructure. Each step in the
value chain adds value, and the last, third activity will produce network externalities, also
added value through utilizing the network.
Network externalities identification
When utilizing the network, users get more resources to use when the network grows and
network externalities grow. The bigger the network, the more specialized processing
nodes and information resources will be in the network. For the network, the network
externalities might be negative, as more work is to be conducted in the network. For the
network owner, more activity in the network should mean more return on the investment.
One must remember that some other network of the organization might suffer because of
the powerful local area network. A clear example is that many organizations have got rid
of traditional telephone networks because of modern computer networks or/and because
of mobile phone networks. For stakeholder, what actually counts is the total balance
when all the used networks are taken into account.
Measurement of externalities
How the network externalties materialize to each stakeholder depend a lot on the
charging arrangements designed for the network. For example, the owner of the network
might get some revenue for each new user, as well as the network operator. Most likely
the revenue flow of the extra users will be bigger than the marginal cost of each new user.
Through the network, final users might avoid acquiring the needed resources themselves,
and would thus save money. For example, buying an own color laser printer could be
avoided because of the positive network externalities provided by the fact that one laser
can serve a big group of users.
Conclusions
Benefits and costs caused by information systems are hard to measure, even evaluate.
The problem is further complicated when network externalities are taken into account.
No wonder most researchers at the information systems research discipline have ignored
the topic of network externalities. This means that information system researchers would
ignore the mechanisms running at the very heart of the information society.
Network externalities can be a complicated concept, and with wrong kind of approach
they remain an untouched white area even in the future. This however needs and can not
be the case. In order to really understand the inner functions of the information –
network - society, network externalities must be studied. What is needed is an operative,
simple, sometimes even childish approach, where a genuine will exists to see network
externalities in work around us. We can see them if we want and give them names, but
we can as well ignore them. A topic too needing attention is the comparison of different
networks. We information system researchers might shrink the idea that there is
something in common say with the GSM network and the sewer network, but most surely
there is.
The first step in the process of mastering the problem is identifying and acknowledging
its size and severity. This article has proceeded with this in mind. The very basic
concepts of networks and network externalities have been elaborated, and they already
have been found as complicated.
The haunting task of evaluating network externalities remains largely untouched in this
article. However, a good start is that of understanding and acknowledging the basic
problems: the stakeholder problem, the identification problem, the arbitrator problem and
the measure problem.
References
Andal-Ancion, A. (2003). The digital transformation of traditional business. MIT Sloan
Management Review(Summer), 35-41.
Besen, S. M. Innovation, Competition, and the Theory of Network Externalities: Charles
River Associates.
Castells, M. (1996). The rise of the network society. Malden, Massachusetts: Blackwell
Publishers Inc.
Economides, N. (1996). The economics of networks. International Journal of Industrial
Organization, 14(2).
Farrell, J., & Klemperer, P. (2006). Network Effects and Switching Costs (forthcoming).
In S. N. Durlauf & L. E. Blume (Eds.), the new New Palgrave Dictionary of
Economics: Palgrave Macmillan.
Leibenstein, H. (1950). Bandwagon, Snob, and Veblen Effects in the Theory of
Consumers’Demand. The Quarterly Journal of Economics(May).
Liebowitz, J., & Margolis, S. E. Network externalities (Effects).
London underground. History. (2006). Retrieved 20.6.2006, 2006, from
http://www.tfl.gov.uk/tube/company/history/
Stalder, F. (1998). The rise of the network society, the information age: Economy, society
and culture, vol I. Information Society, 14(4), 301-308.
Sundararajan, A. (2003). Network effects. Retrieved 20.6.2006, 2006, from
http://oz.stern.nyu.edu/io/