16.4 Network Externalities
In Chapter 13 we explained that a network externality exists when the value of a good or service to an individual is greater when a large number of other people also use the good or service. Although network externalities are common in technology-
Unlike positive and negative externalities, network externalities have no inherently favourable or adverse effect on society. What they share, rather, is the existence of an external effect from one person’s actions.
Network externalities play a key role both in the modern economy and in a number of policy controversies. Here we will examine more closely where and how network externalities occur and then at some of the regulatory issues they raise.
Types of Network Externalities
For all network externalities, the value of the good or service is derived entirely from its ability to link many people possessing the same good or service. As a result, the marginal benefit of the good or service to any one individual depends on the number of other individuals who use it.
Although most network externalities involve methods of communication—the Internet, telephones, fax machines, and so on—they can exist when other users are not strictly necessary for the use of a good, as long as they enhance its usefulness. For example, in the early days of railroad development, a railroad from Montreal to Winnipeg would have had considerable value all by itself, as would have a railroad from Vancouver to Winnipeg. However, each line was worth more given the existence of the other, because once both were in place, goods could be shipped via Winnipeg between Montreal and Vancouver. In the modern world, a scheduled flight between two airports becomes more valuable if one or both of those airports is a hub with connections to other places.
Even this kind of direct link need not be necessary to create important network externalities. Any way in which other people’s consumption of a good or service increases your own marginal benefit from consumption of that good or service can give rise to network effects.
Recall that the classic case of indirect network externalities is computer operating systems and that most personal computers around the world run on Windows by Microsoft.
Why does Windows dominate over other operating systems such as Apple’s OSX or Linux? Is a personal computer running Windows useful only to the extent that other people possess the same good? Not in a direct sense; there isn’t a literal network issue making Windows the preferred system.
The dominance of Windows is self-reinforcing for at least two indirect reasons. First, it is easier for a Windows user to get help and advice from other computer users than for someone using a less popular system. Second, Windows attracts more attention from software developers, so more programs run on Windows than on any other operating system.
Network externalities in this broad sense occur for many goods and services. Even your choice of a car is influenced by a form of network externalities. Most people would be reluctant to switch to a car that runs on natural gas because fueling the car would be difficult: very few gas stations offer natural gas. And the reason service stations do not offer natural gas is, of course, that few people drive anything other than gasoline-powered cars. Or to take a less drastic example, people who live in small towns are reluctant to drive an unusual imported vehicle: where would they find a mechanic who knows how to fix it? So the circularity that makes one person choose Windows because everyone else uses Windows also applies to non-high-tech goods like cars.
When a good or service is subject to a network externality, it exhibits positive feedback: if large numbers of people buy it, other people become more likely to buy it, too. If people don’t buy the good or service, others become less likely to buy it. So both success and failure tend to be self-reinforcing. This leads to a kind of “chicken-versus-egg problem”: if each person places a positive value on a product based on whether another person owns it, how do you get anyone to buy it in the first place? Producers of products that are subject to network externalities are aware of this problem, understanding that of two competing products, it’s the one with the largest network—not necessarily the one that’s the better product—that will win in the end. That is, the product with the largest network will eventually dominate the market, and competing products will eventually disappear.
A good is subject to positive feedback when success breeds greater success and failure breeds further failure.
One way to gain an advantage at the early stages of this kind of market is to sell the product cheaply, perhaps at a loss, in order to increase the size of the network. So we often see companies introducing new high-technology products at a price well below production costs. For example, during the 1990s, the two main competitors in the market for Internet browser software, Netscape Navigator and Microsoft Internet Explorer, both offered their products for free. And even today, many cellphone companies offer free handsets to attract consumers to their wireless network.
Finally, network externalities present special challenges for antitrust regulators because competition and antitrust laws do not, strictly speaking, forbid monopoly. Rather, they only prohibit “monopolization”—efforts to create a monopoly. If you just happen to end up ruling an industry, that’s okay, but if you take actions designed to drive out competition, that’s not okay. So we could argue that monopolies in goods with network externalities, because they occur naturally, should not pose legal problems.
Unfortunately, it isn’t that simple. Firms investing in new technologies are clearly trying to establish monopoly positions. Furthermore, in the face of positive feedback, firms have an incentive to engage in aggressive strategies to push their goods in order to increase their network size and tip the market in their direction. So what is the dividing line between legal and illegal actions?
At this point, the rules are somewhat in flux. In the Microsoft antitrust case, described in the following Economics in Action, reasonable economists and legal experts disagreed sharply both about whether the company had broken the law by pursuing a monopoly position and about whether the company should be broken up to diminish its ability to tip new markets in its favour.
THE MICROSOFT CASE
In 2000 the U.S. Justice Department took on Microsoft in one of the most watched antitrust cases in history. By that time, Microsoft had become the world’s most valuable corporation, and its founder, Bill Gates, was the world’s richest man. What the U.S. government sought was nothing less than the breakup of the company.
The case involved almost all of the issues raised by goods with network externalities. Microsoft was, by any reasonable definition, a monopoly: leaving aside the niches of Apple customers and Linux users, just about all personal computers ran the Windows operating system. The key fact sustaining the Windows system was the force of a network externality: people used Windows because other people used Windows.
The U.S. government did not, however, challenge the Windows monopoly itself (although some economists urged it to). Most experts agreed that monopoly per se is a natural thing in such industries and should not be prevented. What the U.S. government claimed, however, was that Microsoft had used its monopoly position in operating systems to give its other products an advantage over competitors. For example, by including Internet Explorer as part of the Windows system, it was alleged, Microsoft was giving itself an unfair advantage over its rival Netscape in the browser software market.
Why was this considered harmful? The U.S. government argued both that monopolies were being created unnecessarily and that Microsoft was discouraging innovation. Potential innovators in software, the U.S. government claimed, were unwilling to invest large sums out of fear that Microsoft would use its control of the operating system to take away any market competitors might win: Microsoft would produce a competing product that would then be sold as a bundle with the Windows operating system. For its part, Microsoft argued that by setting the precedent that companies would be punished for success, the U.S. government was the real opponent of innovation—innovation that had benefitted customers with lower prices and increasingly sophisticated products.
At first the case went against Microsoft, when a judge ordered the company split in two—into an operating-system company and a company selling the firm’s other products. But this judgment was overturned on appeal. In November 2001, the U.S. government reached a settlement with Microsoft in which the company agreed to provide other companies with the technology to develop products that interacted seamlessly with Microsoft’s software, thus removing the company’s special advantage acquired through bundling its products.
Competitors complained bitterly that this settlement had far too many loopholes and that Microsoft’s ability to exploit its monopoly position would remain. And by early 2004, the U.S. government agreed: antitrust lawyers from the Justice Department reported to the judge who negotiated the original settlement that they were increasingly uneasy about the plan’s ability to spur competition. However, in mid-2004 an appeals court upheld the 2001 settlement, and in November 2007, Microsoft’s obligations under the original settlement expired.
Quick Review
Network externalities arise when the value of a good increases when a large number of other people also use the good. They are prevalent in communications, transportation, and high-technology industries.
Goods with network externalities exhibit positive feedback: success breeds further success, and failure breeds further failure. The good with the largest network eventually dominates the market, and rival goods disappear. As a result, in early stages of the market, firms have an incentive to take aggressive actions, such as lowering price below production cost, to enlarge the size of their good’s network.
Goods with network externalities pose special problems for antitrust regulators because they tend toward monopoly. It can be difficult to distinguish what is a natural growth of the network and what is an illegal monopolization effort by the producer.
Check Your Understanding 16-4
CHECK YOUR UNDERSTANDING 16-4
Question 16.7
For each of the following goods, explain the nature of the network externality present.
Appliances using a particular voltage, such as 110 volts versus 220 volts
8½-by-11-inch paper versus 8-by-12½-inch paper
The voltage of an appliance must be consistent with the voltage of the electrical outlet it is plugged into. Consumers will want to have 110-volt appliances when houses are wired for 110-volt outlets, and builders will want to install 110-volt outlets when most prospective homeowners use 110-volt appliances. So a network externality arises because a consumer will want to use appliances that operate with the same voltage as the appliances used by most other consumers.
Printers, copy machines, fax machines, and so on are designed for specific paper sizes. Consumers will want to purchase paper of a size that can be used in these machines, and machine manufacturers will want to manufacture their machines for the size of paper that most consumers use. So a network externality arises because a consumer will want to use the size of paper used by most other consumers—namely, 8½-by-11-inch paper rather than 8-by-12½-inch paper.
Question 16.8
Suppose there are two competing companies in an industry that has a network externality. Explain why it is likely that the company able to sustain the largest initial losses will eventually dominate the market.
Of the two competing companies, the company able to achieve the higher number of sales is likely to dominate the market. In a market with a network externality, new consumers will base their buying decisions on the number of existing consumers of a specific product. In other words, the more consumers a company can attract initially, the more consumers will choose to buy that company’s product; therefore, the good exhibits positive feedback. So it is important for a company to make a large number of sales early on. It can do this by pricing its good cheaply and taking a loss on each unit sold. The company that can best afford to subsidize a large number of sales early on is likely to be the winner of this competition.
A Tale of Two Research Clusters
Silicon Valley in California and Route 128 in Massachusetts are the pre-eminent high-tech clusters in the world. Silicon Valley dates back to the early 1930s, when Stanford University encouraged its electrical engineering graduates to stay in the area and start companies.
In the early 1950s Stanford created the Stanford Industrial Park, leasing university land to high-tech companies that worked closely with its engineering school. In the mid-1950s, defense contractors such as Lockheed brought dollars to the area. By the late 1960s, a critical mass of such talent had accumulated. For example, in 1968, eight young engineers left their employer over a disagreement; over the next 20 years, they founded 65 new companies, including Intel Corporation, which later created the microprocessor chip, the brain of personal computers.
This pattern repeated: one researcher estimated that in small and medium-sized firms, 35% of the workforce would, on average, turn over in a year. Silicon Valley became a fertile location for startups, with dozens sprouting every year—everything from firms specializing in hardware and software to network firms like eBay, Facebook, and Google. It also became home to investors who specialize in financing new high-tech companies. Silicon Valley’s compact geographical location allowed people to form close social and research bonds even while working for rival firms.
On the other side of the country, a high-tech cluster known as Route 128 lies on a stretch of highway surrounding Boston and Cambridge. It owes its beginnings to the Massachusetts Institute of Technology (MIT), the top engineering university in the world, as well as funding from the U.S. military, NASA, and the National Science Foundation. In the 1950s Route 128 dominated Silicon Valley, with three times the employment.
But early on, Route 128 differed from Silicon Valley in significant ways. Geographically, Route 128 was more spread out than Silicon Valley. Its firms were larger, reflecting the needs of defence contractors during the Cold War. And MIT extended little help to Route 128 firms.
Another major difference between the two clusters lay in how firms were organized. Route 128 firms tended to be “vertically integrated,” combining the entire chain of production from research to design to production in the same firm. Silicon Valley firms focused exclusively on research and design, contracting production out to specialized firms that achieved economies of scale. In contrast to the fluidity of employees and ideas across companies in Silicon Valley, Route 128 firms emphasized a commitment to lifetime employment and closely guarded their innovations to remain competitive.
The 1970s and 1980s were harsh for Route 128. Military spending dried up, and it lost its edge in minicomputers when Apollo Computers lost its pre-eminence to an aggressive Silicon Valley firm, Sun Microsystems. By 1980, electronics employment in Silicon Valley was three times that of Route 128. Over time, Route 128 ceded the advantage to Silicon Valley in electronics and networking. Today its niche is in biotechnology, genetics, materials engineering, and finance.
QUESTIONS FOR THOUGHT
Question 16.9
What positive externalities were common to both Silicon Valley and Route 128? What positive externalities were not common to both? Explain.
Question 16.10
What factors made Silicon Valley such a fertile place for startups? How did these factors interact with one another? What inhibited startups in Route 128?
Question 16.11
In hindsight, what could Apollo Computers have done to maintain its advantage in minicomputers? What does this tell you generally about research clusters?
