A Classroom Demonstration for Teaching Network Effects
James Sawler. Journal of Economic Education. Washington: Spring 2007. Vol. 38, Iss. 2; pg. 153, 7 pgs
The introduction of the concept of network effects is useful at the principles level to facilitate discussions of the determinants of monopoly, the need for standards in high-tech industries, and the general complexity of real-world competition. The author describes a demonstration and an extension that help students understand how consumers make choices in markets where network effects are prominent and how these choices affect market outcomes. He then provides an outline for a classroom discussion of the results.
Although the analysis of network effects is relatively new, the topic has begun to find its place in principles textbooks (Colander, Sephton, and Richter 2003, 313-15; Lipsey and Ragan 2003, 40). The introduction of network effects is useful at this level to facilitate discussions on the determinants of monopoly, the need to establish standards in high-tech industries, and the general complexity of realworld competition.1 It can be a challenging topic for principles students; however, a demonstration such as the one described here can provide students with firsthand knowledge of how consumers make decisions in the face of network effects, and how these decisions can affect market outcomes.2
A network effect is the change in benefit (or cost) that an agent derives from a good when the number of other agents consuming that good changes.3 As an example of a (positive) network effect, consider the fax machine. As the number of people owning fax machines increases, so does the value of owning a fax machine-the more people who own fax machines, the more people with whom an individual can send and receive faxes.4
There are two general sources of network effects-direct and indirect (Liebowitz and Margolis 1998). Direct network effects are those generated through the direct outcome of the number of purchasers on the value of the product to consumers. The fax machine is an example of a direct network effect. Indirect network effects arise as complementary products become more readily available as the size of the network increases. Consider the video game console industry. Software developers are more likely to produce games for a console that has a large number of users. In turn, greater access to games increases the value of owning that particular console.
Network effects are related to the concept of path dependence. Path dependence occurs when "a minor advantage or seemingly inconsequential lead for some technology, product or standard can have important and irreversible influences on the ultimate market allocation of resources, even in a world characterized by voluntary decisions and individually maximizing behavior (Liebowitz and Margolis 1995, 205)." Path dependence resulting from network effects is significant because it can lead to the formation of monopolies and because it can affect the choice of standards.
The presence of positive network effects can create first-mover advantages that can lead to the formation of monopolies. Imagine an industry where there are two firms selling competing but incompatible technologies. If each technology bestows a network effect on its users, then the firm with the larger market share will have an advantage (consumers will derive more value from that firm's product by being connected to a larger network). Providing that production costs do not exhibit decreasing returns that overwhelm the network effects, ceteris paribus, consumers will continue to purchase the first-mover's product and a monopoly will result.5
Network effects also play a role in the choice of the correct standard. It is possible for an inferior technology to become its industry standard if that technology has a head start that creates a large network. If the network effects of the inferior technology exceed the extra benefits provided by the superior technology, then consumers will continue to purchase the inferior technology, and it will become the industry standard. However, the existence of a positive network effect does not guarantee the adoption of an inferior standard. Although possible in theory, there is little empirical evidence to support the belief that network effects lead to the type of market failure in which inferior technologies are adopted (Liebowitz and Margolis 1994).
Network effects can affect the rate of adoption of new technologies. Given that network effects can lead to a monopoly, consumers might be hesitant to adopt a new technology lest a competing technology becomes the standard (no one wants to be left holding a Beta video cassette recorder).
A DESCRIPTION OF THE DEMONSTRATION
At the beginning of class, the instructor announces that there will be a brief pop quiz later in the class. However, there is a twist; during the quiz, students will be able to discuss the solutions with certain other students in the class. They will be able to communicate only with students who choose the same "communicator" as they do. The communicators are simply cards of two (or more) different colors, say blue or green. Anyone choosing a blue card will be able to discuss solutions to the quiz with anyone also possessing a blue card, but not with anyone who has chosen a green card, and vice versa. The instructor then approaches each student sequentially, and allows him or her to make a choice. The other students can hear the choice made.
Students quickly realize that the rational decision is to select the color that the majority of the students preceding them have chosen. To date, four trials have been conducted; in each trial thus far, all students have selected the same color.
The following are suggested questions for discussion:
Why did you all choose the same color? Students have responded that it did not make sense to choose a color that no one before them had chosen. If they had taken a different color, they would not have been able to discuss the solutions with anyone who chose before they did. There also would have been no guarantee that anyone after them would choose their color.
This is a good time to introduce the term, network effect. A network effect exists when the value of a product depends on the number of people who purchase or are expected to purchase a similar product. In this case, each student's choice of the color blue has increased the value of everyone else's choice of the color blue by expanding the network of students with whom everyone could communicate.
Can you provide examples of technologies that bestow network effects? Common examples include fax machines, DVD players, computer operating systems, video game consoles, and so forth. For each example, the instructor should lead a discussion of the sources of the network effects. It is helpful at this time to distinguish between the direct and indirect sources of network effects.
Returning to our experiment, suppose that each of the colors was a product owned by a separate company. What would be the end result of this process? Students have recognized that the company owning the color blue would have a monopoly.
Are there any characteristics of blue that made it a superior product to green? Did it cost less? These questions help facilitate a discussion of how network effects can lead to path dependence and the formation of a monopoly, even if there are no discernible differences in the quality or price of the product. The only reason blue became the monopoly product was because it was chosen first and, thus, developed a network before green had the chance. Microsoft Windows provides a useful example (it is important to note here that there are other probable determinants of Microsoft's monopoly position).
This is a good opportunity for the instructor to discuss how it is possible for an inferior technology to become its industry standard when it benefits from sizable positive network effects.6 Some commonly cited examples include: IBM DOS vs. Apple Macintosh, Beta vs. VHS, and the QWERTY keyboard. Despite these anecdotal examples, the instructor should discuss how the existence of network effects does not necessarily lead to the adoption of inferior technologies and that there is little empirical evidence supporting the belief that network effects lead to market failure (Liebowitz and Margolis 1994, 146-49).7 Network effects can be overcome when consumer preferences are stronger than the network effects, so that an inferior technology that has a head start may be overtaken by a preferred, superior technology. This process allows new technologies to replace established networks (e.g., DVD's replace VHS tapes). When preferences are substantially differentiated, a market that exhibits network effects may be able to support more than one competing product. The video game console industry provides an example.
Given the color you chose, what choice did you expect the students who selected after you to make? Students have responded that they expected those students to make the same choice that they did. This question enables a discussion of how expectations play an important role in determining the extent of network effects and path dependence. The value of the product depends not only on the number who are currently connected to the network but also on the expected size of the network in the future.8
AN EXTENSION: THE IMPACT OF NETWORK EFFECTS ON THE ADOPTION OF NEW TECHNOLOGIES
Upon completion of the discussion, instructors may wish to conduct the following extension. If this option is chosen, the class should be split in two, with one-half participating in the original demonstration and the other in this extension, which can be used to demonstrate the effect that network effects can have on the adoption of new technologies.
The instructor announces that the pop-quiz will consist of 10 questions. As with the original experiment, students are told they will be able to communicate with other students providing they choose the same colored card.9 This time, there are a few differences. Rather than choosing a card directly, students will have 15 seconds to write down the color they wish to choose. Students will not be permitted to discuss their choices with their classmates. If they wish, they can delay their decision and make their choice during a second (or third) round of 15 seconds. However, there is a cost to choosing this option; for each round of 15 seconds that they delay their choice, they will not be able to cooperate on one of the questions. For example, if they delay their choice until the third round, they will not be able to communicate with anyone on the first and second questions. At the end of each round of 15 seconds, students who have made a choice are given the colored card they have chosen, and they cannot change their decision. The process is continued until all students have chosen a color.
Students recognize that there is a tradeoff. If they make their decision too early, they risk choosing a color that few of their classmates ultimately choose. On the other hand, for each period that they delay, they will not be able to discuss the solution to one of the questions with any of their classmates. In each trial thus far, fewer than 20 percent of the students chose a color during the first round, typically rising to just above 50 percent by the second round. By the third or fourth rounds, almost all students had chosen a color.
For those of you who did not choose a color in the first round, why did you delay your choice? Students have replied that they thought it was better to wait to see what color(s) their classmates chose before making a decision. No one wanted to end up with a color that no one else had chosen.
What was the cost of delaying your decision? Students replied that it was not being able to communicate on the first question. These questions help facilitate a discussion on the rate of the adoption of new technologies in the face of network effects. The instructor should ask students to consider a new market with competing but incompatible technologies. Just as the students did in the demonstration, many consumers will delay purchasing one of the new technologies and will wait to see which one might become the industry standard. However, this delay comes at a cost-the lost utility consumers could have had from using the technology during the delay.
Given the lost utility arising from the delay in adoption of the new technologies, is the market outcome necessarily optimal? This question facilitates a discussion of the need to create standards in markets that exhibit network effects. In some cases, the market outcome is not optimal. Not only do consumers lose utility from delaying their purchases, but producers of complementary products (i.e., software developers) may delay developing new products until a clear standard emerges. This reduces utility for those consumers who have adopted one of the new technologies.
The welfare loss associated with the delayed adoption of products in markets with network effects can be reduced if a standard emerges quickly. This can be accomplished if the firm promoting a particular technology is successful in convincing consumers that its technology will emerge as the standard. Some of the strategies employed to this effect include selling the product at a very low price (or even giving it away) to develop a large installed base, encouraging the development of complementaries to realize indirect network effects, offering a free version of a component of the technology that complements the full version (e.g., Adobe's free versions of Adobe Reader that complement Adobe Acrobat).10 Not all standards emerge as the direct result of noncooperative competitive strategy. Sometimes firms will license their technology to competitors. At other times, competitors in markets with network effects will work together to develop jointly or to agree on standards. Also, particularly when the industry players cannot come to an agreement, governments may intervene to help establish standards.
The two classroom demonstrations described here have proven effective in introducing the concept of network effects to principles of economics students. They have also been employed successfully in introductory industrial organization classes to facilitate discussions on the difficulties of antitrust enforcement in high-tech industries. Not only are these demonstrations effective, but they are also not overly time consuming-a genuine concern in principles classes where there is such a wide variety of topics to be discussed. Each demonstration can easily be performed, with instructions, within 5 to 10 minutes. This leaves ample time for analysis and discussion.
1. For a general introduction to the concept of network effects, the instructor should refer to Liebowitz and Margolis (1998). Arthur (1996) provides a clear, nontechnical introduction that is accessable to principles students.
2. An alternative classroom demonstration for teaching network effects has been presented by Ruebeck et al. (2003). This demonstration relies on the construction of an artificial utility function that, particularly at the principles level, can leave students wondering whether network effects are a real phenomenon. The demonstration presented here requires no such artificial construct; rather, students make decisions that directly influence their own utilities.
3. This concept is often described as a network externality. However, the term externality carries a connotation of market failure. Given that the circumstance where the net benefit of an action is affected by the number of agents taking similar actions does not necessarily result in market failure, Liebowitz and Margolis (1994) recommended the adoption of the term network effect to describe this phenomenon. This terminology is adopted for this article.
4. An example of a negative network effect is the congestion caused by automobiles. The more people who take their cars out on the highway, the greater the cost imposed on each driver by the extra congestion. Other examples include fashion and collectible items. The more people who purchase a particular collectable, the more common and, therefore, the less valuable it becomes.
5. Of course, ceteris is not always paribus. Markets where network effects are prominent may support more than one competing product if the products can be significantly differentiated to overcome the network effects. The video game console industry is an example; the market currently supports three networks (Sony Playstation, Microsoft X-box, and Nintendo Game Cube). Another example exists in personal computers, where Apple continues to survive as an alternative to Windows.
6. Note that the adoption of an inferior technology is not addressed directly by the classroom demonstration in which, presumably, neither color is inferior to the other. One referee has suggested having students choose between items for which they have intrinsic preferences, say between apples and oranges. Given the presence of the network effect, there will be students who select apples even though they prefer oranges. Although this still does not imply market failure, it indicates that network effects can lead to choices that would not otherwise be made.
7. The QWERTY keyboard is often cited as an example of how network effects can lead to market failure. It is claimed that the QWERTY keyboard is inferior to the competing Dvorak keyboard, which has been designed to allow greater speed. Recent studies, however, have found that there is practically no difference between the typing speeds of the two designs. Another common example is the claim that the VHS videotaping format was inferior to the Beta format, specifically that Beta had a superior picture quality. Technical experts, however, have concluded that this was not the case. Furthermore, VHS had an advantage over Beta in that it allowed a longer playback. Liebowitz and Margolis (1994) discussed these findings in detail.
8. The use of expectations here is helpful to avoid conflict with the law of demand. In the presence of network effects, the willingness to pay for a product increases with the expected size of the network, so there is a positive relationship between price and expected quantity. However, the quantity demanded still decreases with an increase in price, preserving the law of demand. Implications of network effects for the law of demand are probably best avoided at the principles level.
9. For this demonstration, I recommend giving students a choice of four cards (or more depending on class size) rather than two. This increases the likelihood that students, if they choose a particular color, will finish a round having no one or only a few with whom to communicate on the quiz. Thus, it produces the desired result of having some students delay their choice.
10. For more detailed discussions of strategy in the presence of network effects, refer to Sun, Xie, and Cao (2004), Katz and Shapiro (1994), and Besen and Farrell (1994).
Arthur, B. 1996. Increasing returns and the new world of business. Harvard Business Review 74 (July/Aug): 100-10.
Besen, S. M., and J. Farrell. 1994. Choosing how to compete: Strategies and tactics in standardization. Journal of Economic Perspectives 8 (2): 117-31.
Colander, D. C., P. S. Sephton, and C. Richter. 2003. Microeconomics. 2nd Canadian ed. 313-15. Toronto, ON: McGraw-Hill Ryerson.
Katz, M. L., and C. Shapiro. 1994. Systems competition and network effects. Journal of Economic Perspectives 8 (2): 93-115.
Liebowitz, S. J., and S. E. Margolis. 1994. Network externality: An uncommon tragedy. Journal of Economic Perspectives 8 (2): 133-50.
_____. 1995. Path dependence lock-in and history. Journal of Law, Economics, and Organization 11 (1): 205-26.
_____. 1998. Network effects and externalities. New Palgrave dictionary of economics and the law. London: MacMillan Reference Limited.
Lipsey, R. G., and C. T. S. Ragan. 2003. Microeconomics, 10th Canadian ed.: 40. Toronto, ON: Addison Wesley.
Ruebeck, C. S., S. Stafford, N. Tynan, W. Alpert, G. Ball, and B. Butkevich. 2003. Network effects and standardization: A classroom demonstration. Southern Economic Journal 68 (4): 1000-08.
Sun, B., J. Xie, and H. H. Cao. 2004. Product strategy for innovators in markets with network effects. Marketing Science 23 (2): 243-54.