Networks and Network Effects
[The text below includes excepts from Regulation by Networks, 2003 BYU L. Rev. 1179 (2003)]

Networks (in the economic sense) are combinations of facilities and rules that are characterized by network effects.  Networks appear in many forms, including: trade associations, commodity exchanges, electricity grids, Internet auction sites, peer-to-peer and business-to-business exchanges, social circles, etc.

The defining trait of a network is that  'network effects' significantly influence its structure and the behavior of its members.  Network effects are a form of economies of scale - increasing returns (or a decrease in the cost per unit) as the number of units increases.  We will first explain what are economies of scale, then explain what's unique about network effects.

Economies of Scale.  When we refer to 'economies of scale', we usually mean economies of scale on the supply-side; in other words, the more units of a product are produced (by one firm), the lower the cost per unit.  For example, writing a book may take many weeks, perhaps months, of work.  Let's put a price tag of $10,000 on that.  Once the book is written, it costs very little to print additional copies of it.  Let's assume the cost of printing another copy of the book is $10.

The more copies of the book are printed, the lower the per unit cost (total cost divided by the number of copies printed).  The total cost of producing one copy of the book is $10,010 ($10,000 + (1x$10)), and since there is only one copy produced, the cost per copy is also $10,010.  If two copies are printed, the total cost is $10,020 ($10,000 + (2x$10)), and the cost per unit is $5,010 ($10,020/2) - still on the expensive side.  If a hundred copies are printed, total cost will be $11,000 ($10,000 + (100x$10)), and the cost per copy will decrease to $110.  If 1000 copies are made, the per copy will be $20 ($20,000/1,000) - finally affordable.  The cost will continue to decline, as the big initial investment made in writing the book is divided over more copies.  The decline in cost per unit, proportionate to an increase in the quantity produced, is called (supply-side) economies of scale.

Network Effects.  While networks may enjoy economies of scale and scope in production, the unique quality of a network is economies of scale and scope in demand, referred to by economists as ‘network effects’: the value of membership in a network is enhanced by an increase in the number of other members or in the other members’ usage of the network.  An example is an internet marketplace (such as eBay).  If I want to sell an item, the probability that I find a potential buyer increases as more people use the same internet marketplace.  And as a buyer, the probability that I find a person wishing to sell the very item I seek increases as more people use the marketplace.

As we said above, in industries characterized by economies of scale in supply, firms lower their production costs by drawing more demand for their product, allowing them to produce more of the product and thus benefit from the economies of scale.  But in network industries, firms can lower their costs (or increase the consumer’s benefit from its products without increasing costs, which is equivalent to lowering costs) without having to wrest customers from their competitors - by interconnecting with competitors (i.e., joining a network with the competitor), thus allowing its customers to reap demand-side economies of scale as if its competitors’ customers were its own.

For example, the benefit to customers of a cellular telephone company increase as they are able to talk to more people through their phones.  A cellular telephone carrier could, by interconnecting with another carrier, offer its customers the added benefits of talking with the other carrier’s customers, making each carrier’s service as attractive as if one of the carriers acquired all of the other’s customers.  This characteristic of network industries creates a significant incentive for creating inter-firm networks.  It also makes membership in a large network (i.e., the ability to transact through the network) a valuable asset, and therefore the network’s ability to exclude a member may be a powerful sanction.

This does not mean that network effects inevitably result in natural monopolies.  Differences in the quality of competing network goods, or in their production costs may offset the relative advantage of the larger network.  Furthermore, production costs often exhibit at a certain point decreasing returns to scale (i.e., they rise as production increases).  This increase in cost may offset the increasing returns to scale derived from the network effect [See S. J. Liebowitz & Stephen E. Margolis, Network Effects and Externalities, in 2 The New Palgrave Dictionary of Economics and the Law 671, 672 (Peter Newman, ed. 1998)].

* * *

The Development of Understanding of Network Effects.  The concept of network effects is not at all a novel one.  Perhaps the most ancient example of recognition of the enhanced benefits that result from connectivity between entities is found in the book of Genesis, and comes directly from the mouth of God: "'Here they are, one people with a single language, and now they have started to do this [building the Tower of Babel]; henceforward nothing they have a mind to do will be beyond their reach…" [Genesis, 11:6].

Language is characterized by network effects – the benefit derived from communicating in a language increases significantly as more people are familiar with it.  Increased “membership” in this network (i.e., fluency in the language) allows communication and coordination among a larger number of people (which confers greater benefits to each of them).  Babel was destroyed by undoing its linguistic network: "’…Come, let us go down there and confuse their speech, so they will not understand what they say to one another.'  So the Lord dispersed them from there all over the earth, and they left off building the city.  That is why it is called Babel, because the Lord there made a babble of the language of all the world." [Genesis, 11:7-8].

Significantly later, economists made similar, and more refined, observations of industries in which the utility to a consumer (a member of the network) increased with the number of other consumers using the same product.  For example, Rohlfs noted network effects in communications services industries, and examined the influence of network effects on pricing and barriers to entry into the industry [Jeffrey Rohlfs, A Theory of Interdependent Demand for a Communications Service, 5 Bell J. Econ. & Man. Sci. 16 (1974)].  Such effects are considered direct network effects, as they are generated through a direct physical effect of the number of consumers on the value of a product [See S.J. Liebowitz & Stephen E. Margolis, Network Effects and Externalities, in 2 The New Palgrave Dictionary of Economics and the Law 671 (1998)].  Other scholarship identified indirect network effects – an increase in the value of a product as a result of an increase in the purchase or use of a complementary product.  For example, if more people use Excel, there are going to be more people any given user can get help from and more books and courses on how to use Excel; if more people carry MasterCards, more merchants will take MasterCards, making the cards more valuable to both cardholders and merchants [See David S. Evans & Richard Schmalensee, A Guide to the Antitrust Economics of Networks, 10 Antitrust 36 (1996)].

As such observations identified an increasing number of industries in which network effects significantly influence the behavior of firms, models simulating network environments were developed.  A key work examining competition in network environments was offered by Katz and Shapiro [Michael L. Katz & Carl Shapiro, Network Externalities, Competition and Compatibility, 75 Am. Econ. Rev. 424 (1985)].

Their model demonstrated that the presence of network effects and the need for compatibility lead to multiple equilibria, and consumers’ expectations are key in determining which equilibrium evolves.  Generally, consumers will prefer to join a network that they perceive as likely to become (or is already) the market leader.  This preference may setoff consumers’ preferences regarding the product or service itself, so that (it was argued) an inferior product that is perceived to be the market leader (perhaps because it was a first-mover into the market) will be preferred over superior but smaller competitors.

This argument led to a line of literature examining network effects as barriers to entry, and as the cause to the alleged persistence of less efficient network goods.  Another paper by Katz and Shapiro showed that the presence of network effects may lead to excessive standardization [Michael L. Katz & Carl Shapiro, Technology Adoption in the Presence of Network Externalities, 94 J. Pol. Econ. 822 (1986)].  Farrell and Saloner created a model suggesting that new technology may not be adopted even if it is superior to existing technology, because of ‘excess inertia’ caused by the presence of an installed base [Joseph Farrell & Garth Saloner, Installed Base and Compatibility: Innovation, Product Preannouncements, and Predation, 76 Am. Econ. Rev. 940 (1986)].  David offered the anecdotal example of the persistence of the “QWERTY” keyboard to argue that industries may “lock-in” to inefficient standards [Paul A. David, Clio and the Economics of QWERTY, 75 Am. Econ. Rev. 332 (1985)].

Other scholars rejected the likelihood of an inefficient lock-in.  Liebowitz & Margolis refuted the lock-in hypothesis in the “QWERTY” anecdote [S. J. Liebowitz & Stephen E. Margolis, The Fable of the Keys, 33 J. L. & Econ. 1 (1990)], as well as in another much cited anecdote, that regarding the VHS/Beta competition over the video cassette standard [S. J. Liebowitz & Stephen E. Margolis, Pathe Dependency, Lock-in and History, 11 L. Econ. & Org. 205 (1995); on these two anecdotes and a few others also see: S. J. Liebowitz & Stephen E. Margolis, Should Technology Choice be a Concern to Antitrust Policy?, 9 Harv. J. L. & Tech. 283 (1996); Michael I. Krauss, Regulation v. Markets in the Development of Standards, 3 S. Cal. Inter. L. J. 781 (1994)].  This issue is still disputed among scholars [See, e.g., William E. Cohen, Competition and Foreclosure in the Context of Installed Base and Compatability Effects, 64 Antitrust L.J. 535, 539-546 (1996)]. 

Another issue emphasized by Katz and Shapiro is decisions regarding compatibility.  Compatibility may be achieved by joint decision (e.g., coordinated acceptance of a standard), or unilaterally, by the construction of an ‘adapter’ by a single firm, to make its product compatible with another [Katz & Shapiro (1985), Supra, at pp. 434-439].  Private incentives for compatibility may differ from public incentives, possibly resulting in private action that fails to maximize social welfare from the network effects.  One strand of the literature examined the choice between unilateral and coordinated facilitation of compatibility [See, e.g., Joseph Farrell & Garth Saloner, Coordination Through Committees and Markets, 19 Rand J. Econ. 235 (1988) (comparing committee agreement on standards; unilateral declarations of standards by single firms, followed by independent decisions by other firms which standard to follow; and a hybrid system adapting features of both the committee and the market leadership mechanisms); Joseph Farrell & Garth Saloner, Converters, Compatibility, and the Control of Interfaces, 40 J. Ind. Econ. 9 (1992) (finding that when adapters do not facilitate perfect compatibility, their introduction might reduce social welfare below that in an industry without adapters at all)].  This issue is of considerable importance to antitrust scholarship, as coordinated facilitation of compatibility is usually more suspect of being used for anti-competitive ends than its unilateral counterpart, and it is therefore important to understand whether it has a redeeming advantage in increasing social welfare through exploitation of network effects [See: United States Department of Justice and Federal Trade Commission, Antitrust Guidelines for Collaborations Among Competitors; Eliason Corp. v. National Sanitation Foundation, 614 F.2d 126 (6^th Cir. 1980) (stating that alleged boycotts arising from industry self-regulation do not give rise to a Sherman Act violation absent discrimination or manifestly anticompetitive and unreasonable conduct).  But compare: Fashion Originators’ Guild of America v. FTC, 312 U.S. 457 (1941) (condemning coordinated activity aimed at preventing and punishing “style piracy” and stating that “even if copying were an acknowledged tort under the law of every state, that situation would not justify petitioners in combining together to regulate and restrain interstate commerce…”)].

Another strand of the compatibility literature examines the incentives for a decision not to be compatible with others, and the effect of such decisions on social welfare.  Compatibility is not always efficient.  Above a certain point, network effects (i.e., demand-side increasing return to scale) are usually reduced and might even reverse (e.g., due to congestion on the network).  Therefore, there may be a maximum efficient size for a network, and a refusal to allow compatibility with other systems may be designed to prevent a network from expanding beyond its efficient size.  Furthermore, facilitating compatibility has various costs (e.g., actual costs of coordination, loss of freedom to vary due to the need to remain compatible, forced disclosure of proprietary information, and facilitating anti-competitive coordination).  For this reason, a refusal to be compatible may at times increase social welfare.

However, firms may have incentives to refuse to be compatible even when compatibility would increase social welfare.  Farell and Saloner observed that in decisions whether to make two technologies compatible, when one technology is supplied by a single firm, that firm may have an incentive to make conversion costly [Farrell & Saloner (1992), Joseph Farrell & Garth Saloner, Converters, Compatibility, and the Control of Interfaces, 40 J. Indus. Econ. 9, 26-28 (1992)].  Cremer, Rey and Tirole expanded on this insight, terming this increase in the cost of compatibility (or a reduction in its quality) “degradation” [Jacques Cremer, Patrick Rey & Jean Tirole, Connectivity in The Commercial Internet, 48 J. Indus. Econ. 433 (2000)].  They refer to compatibility as “connectivity” and examine the plausibility of such a strategy and its effect on social welfare.  Aviram examines this “degradation” as a form of opportunistic behavior [Amitai Aviram, Regulation by Networks, 2003 BYU L. Rev. 1179 (2003)].

While the social welfare implications of network effects have been and continue to be examined thoroughly, less attention was given to the implications of network effects on the form of organization.  Even less attention was given, within this latter issue, to the implications of network effects on institutions that mitigate opportunism (i.e., how network effects are used to fight “garden variety” opportunism, and how institutions combat opportunism that is unique to network environments).

 

To read more about the scholarship that has addressed this field, click here.
To return to the Main Menu, click here.