Network effect


Author Definition



A network effect exists when the value of a product or service increases as the number of other people using the same product or service increases. It has emerged as one of the primary sources of value in the digital economy and the leading theoretical basis for the competition law cases against large tech firms, beginning with the cases against Microsoft.



Although the concept was invoked as far back as the Annual Reports AT&T issued for 1901 and 1908, the modern study of network effects began with Jeffrey Rohlfs’s 1974 article in the Bell Journal of Economics and Management Science. The literature expanded greatly during the 1980s after a seminal series of publications by Joseph Farrell and Garth Saloner and Michael Katz and Carl Shapiro. It is also deeply linked to the literature on path dependence pioneered by W. Brian Arthur and Paul David. Networks effects are often theoretically grounded in Metcalfe’s Law, which recognizes that the number of pairwise connections increases quadratically with the number of nodes. Network effects also formed the basis for the government’s landmark cases against Microsoft resolved by the courts in 1998 and 2001.

Network effects may be direct or indirect. Direct network effects arise when the value is determined by the number of people using the same product. Indirect effects arise when two or more groups of people use the network to exchange value.

Network effects are often said to give rise to two types of market failure. First, the fact that a consumer’s decision to join a network increases the value of those who are already part of the network is sometimes said to create a positive network externality that creates benefits for other consumers who have already joined the network that the adopter does not internalize. This can cause consumers faced with a choice between two network technologies to postpone choosing between them until other consumers have acted, which can cause a delay in the adoption of a superior technology until the market reaches critical mass. Even after a technology has passed this tipping point, the inability to internalize all of the benefits of an adoption decision may lead marginal consumers to decline to adopt the technology even when doing so would be socially beneficially.

Second, networks that have passed this tipping point can become locked in even after a superior alternative has emerged until adoption of that new technology passes its own tipping point. The greater value enjoyed by the largest network can theoretically become a source of market power. In the limit, network effects can create a winner-take-all dynamic in which the higher value associated with the largest firm allows it to capture the entire market. These concerns have led some to invoke antitrust or regulation to impose some form of mandatory interoperability.

A close reading of this literature reveals that these inferences are too simplistic. The seminal analysis by Joseph Farrell and Garth Saloner points out that the decision to adopt a new network technology creates two separate externalities pushing in opposite directions. At the same time an adoption decision increases the value of the new network, it simultaneously decreases the value of the old network that is being abandoned. This means that whether a market with network effects will exhibit excess friction or excess momentum depends on which of these externalities dominates and cannot be determined a priori.

The theoretical ambiguity has prompted the creation of an empirical literature attempting to identify examples of markets subject to network effects. Prominent examples of supposedly inferior technologies that have become locked in include the QWERTY keyboard and the VHS videotape format, among others. A vibrant literature has arisen challenging each of these examples.

In addition, a number of common market features can solve many of these supposed problems associated with network effects. For example, giving discounts to newcomers can allow them to internalize the benefits that they provide to existing members of a network. Heterogeneity of preferences can also eliminate any tendency for everyone to join a single network, as those confronted with a choice between being part of a larger network built around a disfavoured standard may choose to join a smaller network built around the standard they prefer more. In addition, markets undergoing rapid growth are not susceptible to lock in, because the number of customers currently controlled by any one actor matters little when there are even more new customers joining the market for whose business other actors can compete. For similar reasons, a large customer can unstick a tipped market by allowing a new technology to achieve critical mass. Network effects also lead to market failure only under certain market structures: in markets with at least three nondominant players, the equilibrium outcome is for all of the networks to interconnect. Furthermore, as Jeffrey Rohlfs has noted, diminishing marginal returns to scale are inevitable. Lastly, the existence of gateways between networks or technologies that allow consumers to join multiple networks simultaneously (known as multihoming) eliminate the winner-take-all aspects of network effects by allowing consumers to be members of multiple networks simultaneously.

One particularly interesting application of network effects arises when a network consists of two different types of actors and the value to one type of actor is determined by the number of the other type. This can create two-sided markets, in which the price charged to one type of actor is determined by the revenue generated by the other type. In extreme cases, it can be both profitable and efficient for one side to subsidize the other side. Economic theory concludes that proper analyses of two-sided markets require taking both sides of the market into account, a conclusion that the Supreme Court of the U.S. adopted in Ohio v. American Express Co.

The complexity of the issues surrounding network effects reveals the truth of the Microsoft court’s observation that “[s]imply invoking the phrase ‘network effects’ without pointing to more evidence does not suffice to carry plaintiffs’ burden.” A proper determination of the impact of network effects requires an empirical evaluation of each particular circumstance in which problems are asserted to arise.


Case references

Ohio v. American Express Co., 138 S. Ct. 2274 (2018)

United States v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001) (en banc) (per curiam)

United States v. Microsoft Corp., 147 F.3d 935 (D.C. Cir. 1998)



Bob Metcalfe, ‘Metcalfe’s Law: A Network Becomes More Valuable as It Reaches More Users’ (1995) InfoWorld, Oct. 2, p. 53.

Christopher S. Yoo, ‘Network Effects in Action’, in Douglas H. Ginsburg and Joshua D. Wright (eds.), GAI Report on the Digital Economy 159–191 (Global Antitrust Institute 2020).

Jeffrey H. Rohlfs, Bandwagon Effects in High-Technology Industries (MIT Press 2001).

Joseph Farrell and Garth Saloner, ‘Installed Base and Compatibility: Innovation, Product Preannouncements, and Predation’ (1986) 76 American Economic Review 940.

Michael L. Katz and Carl Shapiro, ‘Product Introduction with Network Externalities’ (1992) 40 Journal of Industrial Economics 55.


  • University of Pennsylvania Law School (Philadelphia)


Christopher Yoo, Network Effect, Global Dictionary of Competition Law, Concurrences, Art. N° 12232

Visites 1785

Publisher Concurrences

Date 1 January 1900

Number of pages 500


Institution Definition

Network effects arise when a good is more valuable to a user the more users adopt the same good or compatible ones. Economists refer to this phenomenon as a network externality because when additional consumers join the network of current consumers they have a beneficial "external" impact on the consumers who are already part of the network.

© European Commission

a b c d e f g h i j l m n o p r s t u v w