Graduate School of Business The University of Texas at Austin

To appear in the Brazilian Electronic Journal of Economics (the inaugural issue, December, 1997).

Choi, Stahl and Whinston are co-authors of the book "The Economics of Electronic Commerce" (Macmillan Technical Publishing 1997) which discusses the economics of the digital marketplace and policy options in more detail. The authors are affiliated with the Center for Research in Electronic Commerce (http://cism.bus.utexas.edu) at the University of Texas at Austin.

Abstract

Static theories of firms often fail to explain the tendency toward standardization and de facto monopolization by software and digital product firms. This article examines arguments favored by new economics of information: no market power, increasing returns, and network effects. We evaluate whether these concepts can explain the nature of Microsoft's monopoly market power or provide any rationale for the argument that the de facto monopolization creates no inefficiency.

The cardinal sin of a monopolist, according to the economic theory of firms, is its restricting output, and by doing so, raising the price above the marginal cost of production. Society's resources could be better used if price equals marginal cost and the output is expanded. If Microsoft is a monopolist in this traditional sense of economics, then it must be artificially restricting its output in order to raise price and, in that case, it should be encouraged to expand its output. Is Microsoft restricting its output? Is Microsoft charging monopoly prices? To the contrary, Microsoft seems to be selling too many of its products, sometimes giving them out for free.

Standard theories of firms often fail to explain peculiar behaviors of Microsoft and their effects on competition and market efficiency. Free market advocates and efficiency-minded economists argue that firms producing digital products such as computer software behave differently from non-digital product producers. In the digital marketplace, the simple fact that there is one dominant firm does not suggest that the firm has monopoly market power or that the market is inefficient. Three arguments can be made to defend Microsoft's contentious position as the dominant player in the computer industry: (1) Microsoft may have no monopoly power even when it is a de facto monopolist. (2) An efficient software market may indeed support only one firm due to scale economies and the digital product's falling marginal cost. (3) Network externalities in software tend to favor one firm over the others. We analyze each of these arguments in the hope of highlighting

No Monopoly Market Power

The first defense postulates that the monopolistic output restriction occurs after the firm achieves market power. Microsoft, on the other hand, still faces competition despite its dominance-roughly 80% of operating systems for desktop computers. Consequently, Microsoft is still expanding its output, and its success does not have the problem associated with the monopoly.

The monopoly problem is straightforward. The firm in Figure 1 is assumed to have monopoly market power and can restrict output and raise price only because it does not face competition. To maximize its profit, the firm operates at the output level when its marginal cost is equal to marginal revenue (Qm). At that output level, market price is determined by the demand (Pm) which is greater than marginal cost (MC). The firm's profit is (Pm - MC) Qm > 0. At Qm level of production, consumers' willingness to pay, Pm, is greater than the cost of producing the product (MC). Thus society will be better off if more resources are allocated into this industry leading to a greater output level at Qc and lower price Pc.

Unlike the firm in Figure 1, the proposition is that Microsoft's market behaviors do not exhibit any sign of either restricting output or raising price. Despite its market share, Microsoft faces fierce competition from Apple's Mac Os, IBM's OS and UNIX-based operating systems as well as new threats from Web browsers and network computers based on Java language. There is scant evidence that Microsoft is in a position to raise prices because of present and future competition.

However, Microsoft's market power should be determined in the PC clone, not the operating system (OS) software, market. An OS is useless without a platform to run on. For that matter, a computer is also useless if it does not have an OS. This hardware-software relationship exists in a variety of products: video and audio players, TVs and radios. These so-called complementary products differ significantly from conventional products with complementarity such as "coffees and sugar." Unlike the latter, hardware-software products must have each other to be useful. There will be no market for video tape players if we have no video tapes; no useful TVs if there are no networks providing programming. In this sense, Microsoft as an OS vendor is a player in the PC clone market.

It may seem that there is fiercer competition in the PC market where no one firm commands more than 25% of the market. Because of this, computer prices are assumed to be efficient. That might be true in terms of hardware prices. However, PC buyers must also buy an operating system such as Windows 95, which commands a monopoly price. If this price were lower, there would be more personal computers being purchased. If competing operating systems are available to PC manufacturers at lower price, PC prices will be lower. Consequently, monopoly problems are found in the PC/OS market where the level of output may be well below the socially efficient level. Besides OS, the components of a PC include microprocessors dominated by Intel and peripheral equipment and utility programs. Despite cost savings in many components of a PC, its price is still prohibitively high for the majority of potential users. The reason should be found in the monopoly market power of Microsoft and Intel.

The Economies of Scale

The second defense argues that, unlike the firm with increasing marginal cost (depicted in Figure 1), computer software generally has a decreasing marginal or average cost because of its high fixed cost and extremely low duplicating (variable) cost. In other words, the average cost of production decreases as more and more products are sold. An efficient market result in such a case is to allow one firm to produce all necessary output in order to achieve the maximum economy of scale. Competing firms duplicate fixed costs unnecessarily and result in waste. Therefore, Microsoft's dominance in operating system software is an efficient result due to the characteristics of the software industry.

Economies of scale refer to gains or losses in production cost as output is increased. Typically, the total production costs first decrease because fixed costs such as buildings and management expenses remain constant, and then increase as inefficiencies kick in when the output level goes beyond the optimal level, resulting in the typical U-shaped average cost curve. For many digital products, their costs seem to be dominated by the increasing return in all output levels, a pattern observed in many natural monopolies.

The average cost decreases since there is a high fixed cost of developing software but no or constant variable (reproduction) cost. It is commonly asserted that computer software costs nothing or little to duplicate and development costs account for the majority of its production cost. Consequently, the average cost of a computer program will go down as more copies are sold.

When a product has decreasing average cost or an increasing economy of scale, a competitive market often fails to achieve an efficient solution. That is, competition implies duplicated fixed costs, and no firm could recoup its fixed cost unless the market price equals the average cost. Without such guarantee, there will be no firms producing the product. A primary example is the telecommunication industry where infrastructure costs are so overwhelming that it is often more efficient to allow one firm which is guaranteed to recoup its investments in return for its subjecting to regulatory oversights. If computer software has an increasing economy of scale, an efficient market solution tends to favor one dominant firm. Therefore, Microsoft's dominance in the OS market may well be efficient.

There are two problems with this argument. First, computer software, and most digital products for that matter, may not have decreasing average costs. Although duplicating costs may be relatively small compared to initial development costs, duplication costs are not the only variable costs for most products. For example, many physical products also have low variable costs, e.g. cereals, shoes, etc. For each copy sold, there may be other substantial costs such as copyright payments and enforcement costs, customer support expenses, and management and accounting costs. To expand in the market, new features may be needed, and offering interoperability for new and old products may impose further costs (Choi, Stahl and Whinston 1997). As a result, it is still an empirical question whether digital products do or do not exhibit decreasing average costs. If not, the economy of scale will not explain away the monopoly problem of Microsoft.

The second problem is that the economy of scale is not so relevant when products are extremely heterogeneous. The economy of scale simply tells us that having one producer is more efficient if we consider all varieties as being essentially the same. For differentiated products catering to different segments of consumers, multiple products are desired. For application software such as word processing and database programs, product differentiation means having different features, a highly desirable aspect. These products may well be produced by one producer. However, the costs associated with differentiation and customization, which may be increasing with variety, are more relevant than duplicating costs, which may be decreasing with output level.

It is often presumed that the scale economy in computer software, rather than monopolization, explains the dominance of Microsoft. However, the increasing economy of scale is limited to the simple economics of mass reproduction, which may constitute only a small part of software production and marketing processes. More empirical and theoretical studies that consider other variables such as product choices and usage are needed before we attribute market dominance to economies of scale.

A more problematic development in this new information economics is the indiscriminate use of the term "increasing returns." For example, consider this statement: "Operating systems show increasing returns: if one system gets ahead, it attracts further software developers and hardware manufacturers to adopt it, which helps it get further ahead." (Arther 1996). What this statement alludes to is the network effect, which is often mixed with interoperability and standardization to explain why one product may naturally dominate its market. Increasing returns, network effects, and the interoperability together seem to produce a monopolized but efficient market. All of these concepts indicate some kind of incentives that drive the market toward a single dominant product whereby consumers benefit from its dominance. The vague impression is that somehow Microsoft's market position is due to the nature of products, not its behaviors which might be anticompetitive.

Network Externality and Interoperability

The third defense for Microsoft is found in the network externality by which consumers benefit from having one standard product. With network externality, the value of a product goes up as more people have the same product: the more the merrier. Microsoft's dominance is simply a manifestation of the network externality which relentlessly drives computer software to standardization.

An externality is an effect on costs or benefits that is not accounted for by market mechanisms such as price. For example, there is no market mechanism to require a neighbor to pay for such benefit even if the neighbor gets some benefit from the tree you plant. In this sense, an externality distorts the resource allocation process and creates market inefficiency. A network externality is an externality related to the number of users (or networks) for a group of products. A typical example is a telephone network where consumption benefits increase as more people join the network (positive network externality). A negative network externality exists when more users result in congestion, thereby diminishing the amount of total benefits.

However, an externality is no longer an externality if a market price already reflects the price of an external benefit or loss. For example, a computer operating system may have a positive externality in that its value increases as there are more people using the same product. This can be represented by an upward sloping benefit schedule for consumers (Liebowitz and Margolis 1995). In Figure 2, two lines represent the level of consumer benefits or willingness to pay (WTP) with respect to the number of total users. The lower line shows the average willingness to pay without externality. It shifts upward when we consider externality (with more benefits as the number of users increases). The optimal number of product is Q1 at P1 if the firm can charge the benefit from network externality. If not, the output is reduced to Q2 with a lower price P2. At P2, the marginal cost is below the true consumers' willingness to pay, implying that the product is underproduced.

If the firm can operate at (Q1, P1), the market does not have an externality problem. On the other hand, if the firm is forced to operate at (Q2, P2), the market is characterized by a network externality. The issue is then whether Microsoft can price its OS software to extract the increased benefit enjoyed by its customers. Unlike other public goods, software manufacturers can and do raise their prices for that purpose. In this sense, the network externality is fully absorbed by the market price, and as a result there is no externality problem.

Still, the value of the product is increased simply because there are more users. This effect is termed as a "network effect" to distinguish it from the problematic network externality. We can certainly attribute Microsoft's dominance in OS to network effects. However, the nature of network effects may be sufficiently different from telephone service and raise different sets of questions.

Network effects may be direct effects as in the case of telephone, where the issue is whether competing products can be used together (a horizontal interoperability). There are also indirect network effects commonly found in hardware-software platforms in computer, video and audio, and computer games industries, where the issue is whether a complementary product can be used with competing products (a vertical interoperability). Numerous studies have shown that the competition among upstream products (e.g. VHS or Beta video players) critically depends on how many downstream products (video tapes) there are (Katz and Shapiro 1985; Chou and Shy 1990; Church and Gandal 1992).

The Intel-Windows machines have become the de facto standard in desktop computing just as the VHS video format chosen by the market over the Beta format. However, unlike the video market or most other hardware-software products, the vertical interoperability issue continues to affect market efficiency in the computer industry. For example, OS vendors can easily modify their OS to influence application program manufacturers. When the de facto monopolist also sells application programs, its dominance in OS may give unfair advantage over its competitors in application software.

Vertically integrated hardware-software firms are commonly observed. For example, audio equipment manufacturers such as Sony are selling musical CDs. However, Sony CDs have no inherent advantage over non-Sony CDs in terms of operating (being played) in a Sony-produced CD player. In the computer industry, the Internet browser competition illustrates the problem with vertical integration.

It is tempting to attribute Microsoft's success with application programs to network effects, scale economies or interoperability. However, an undue monopoly market power may be evident if one firm with a dominant product tries to extend its advantage in other markets by unfairly raising the cost of entry to the application market or limiting access to the OS market. Network effects may have favored Windows over Mac OS or UNIX, but Microsoft's dominance in application programs cannot be explained away by network effects.

Implications

The proper question to ask is whether Microsoft's monopolistic position has any ill effect on market efficiency and consumer welfare. Arguments based on the peculiarities of computer and digital products tend to regard one firm's dominance as natural and efficient. The war of Web browsers has resulted in lowered prices for both Internet Explorers and Netscape Navigator, a positive proof that the industry is far from inefficient. Even if Microsoft becomes the only player in all computer-related markets, some would argue that new and better products will ultimately appear and become dominant. Static models of monopoly no longer seem to apply to this new world of electronic economy.

The purpose of this article was to point out some incongruities in the new economics of the digital economy. Contrary to the belief that the computer industry and the digital economy may differ from physical product markets, there may not be many fundamental differences. Output restriction and price increase are observed if we place Microsoft's OS market within PC-cloning business. Increasing returns may be fallacious. Network effects are indiscriminately applied. A very permissive view holds that Microsoft may do whatever it chooses with Windows OS as its owner. But can it refuse to license Windows OS if a PC vendor wants to install both Windows and its competing OS? Can Microsoft freely change its OS so as to make it difficult for consumers to use Netscape's browser? These questions are more mundane and familiar to economists and antitrust lawyers than the so-called new economics of information (i.e. network externalities, increasing returns and interoperability).

References

Arther, W.B., 1996. "Increasing returns and the new world of business." Harvard Business Review, July-August, 1996, pp. 100-109.

Choi, S.-Y., D. Stahl and A.B. Whinston, 1997. The Economics of Electronic Commerce. Indianapolis, Indiana: Macmillan Technical Publishing.

Chou, C.-F. and O. Shy, 1990. "Network effects without network externalities." International Journal of Industrial Organization, 8: 259-270.

Church, J. and N. Gandal, 1992. "Network effects, software provision, and standardization." Journal of Industrial Economics, 40: 85-103.

Katz, M.L. and C. Shapiro, 1985. "Network externalities, competition, and compatibility." American Economic Review, 75(3): 424-440.

Liebowitz, S.J. and S. E. Margolis, 1995. "Are network externalities a new source of market failure?" Research in Law and Economics, 17:1-22.