|
Article Excerpt
Abstract
Prior theoretical research has established that many software products are subject to network effects and exhibit the characteristics of two-sided markets. However, despite the importance of the software industry to the world economy, few studies have attempted to empirically examine these characteristics, or several others which theory suggests impact software price. This study develops and tests a research-grounded model of two-sided software markets that accounts for several key factors influencing software pricing, including network externalities, cross-market complementarities, standards, mindshare, and trialability. Applying the model to the context of the market for Web server software, several key findings are offered. First, a positive market share to price relationship is identified, offering support for the network externalities hypothesis even though the market examined is based on open standards. Second, the results suggest that the market under study behaves as a two-sided market in that firms a ble to capture market share for one product enjoy benefits in terms of both market share and price for the complement. Third, the positive price benefits of securing consumer mindshare, of supporting dominant standards, and from offering a trial product are demonstrated. Last, a negative price shock is also identified in the period after a well-known, free-pricing rival has entered the market. Nonetheless, network effects continued to remain significant during the period. These findings enhance our understanding of software markets, offer new techniques for examining such markets, and suggest the wisdom of allocating resources to develop advantages in the factors studied.
Keywords: Hedonic pricing, network effects, network externalities, composite goods, two-sided markets, open standards, mindshare, trialability, World Wide Web server market ISRL Categories: AD0517.01, AF1302, AM, CB0904
Introduction
Packaged software represents the third largest industry in the United States. With revenues totaling roughly $185 billion in 2001, it is one of the fastest growing sectors worldwide (Boulton 1999; Frye 2000; Rudy 2002). The dynamics of this industry are quite different from conventional markets. For example, software products are subject to network effects (2)--the idea that a product will become more valuable as its user base expands (Farrell and Saloner 1985; Oren and Smith 1981; Rohlfs 1974). Also, software products can be classified as digital or information goods with a corresponding theoretical marginal cost of zero (Negroponte 1995; Poirier 1990). The economics of zero (or negligible) marginal cost allow software vendors to leverage strategies such as subsidies, versioning, and trialability to a much greater extent than vendors of conventional goods (Bakos and Brynjolfsson 1999; Bakos et al. 1999; Shapiro and Varian 1998).
Despite the importance of software to the world economy and the notion that software markets are different from conventional markets, few studies have attempted to offer a synthesized analysis and empirical examination of the unique forces at work in this industry. Drawing heavily on the literature related to network economics, strategic complements, and software pricing, we use the Web server market as a context to present an analysis that offers the following contributions. First, we investigate the existence of network effects in markets heavily dependent upon open standards (3). Such products represent one of the fastest growing segments of the software industry (Nua 1997). Little is known, however, about the impact of network effects on such markets. Second, we investigate a network software market that can be characterized as a two-sided market. Two-sided markets exist when separate constituencies or user groups coordinate the use of different, yet complementary products (e.g., client software and serve r software) in order to derive utility (Rochert and Tirole 2001). (4) Two-sided markets represent a sizable portion of the computer industry. For example, in terms of dollar sales, client/server technologies have been classified as the largest category in the software industry (Cowen 2000). Firms producing components for two-sided markets face a key question: Should they concentrate on one component (e.g., server) and let others address the markets for requisite, complementary components (e.g., client), or should they produce both components? Third, we expand prior methods used to model software markets by considering additional factors that theory suggests may influence such markets. Specifically, we investigate the role of mindshare (or product awareness), product standards, trial versions, and the impact of a well-known, free-pricing firm entering the market.
This study presents evidence of strong two-sidedness in the Web software market. A positive relationship between server market share and server price offers evidence of network externalities; while a relationship between browser market share and server price suggests the benefit of securing market share in complementary products. Product awareness is shown to be strongly related to network effects in establishing product-pricing advantages over rivals. Findings also demonstrate the positive price benefits of supporting dominant standards and offering a trial product. Even though a negative price shock is identified in the period after a free-pricing rival has entered the market, the positive influence of the above factors continued to remain significant. For researchers, this work synthesizes existing theory and analytical techniques to develop and test a more comprehensive model for gauging the impact of factors influencing two-sided software markets. For practitioners, these findings suggest the wisdom of a llocating scarce resources to develop advantages in the influential factors identified above.
The organization of this paper is as follows: we leverage existing theory to develop a series of research hypotheses and a conceptual model related to the perceived value of Web server products. Then we detail methodology and data, and follow these with estimation results. We conclude with a summary of findings and a discussion of the contributions and limitations of this research.
Theory and Hypotheses
Foundation Model and Context
The foundation of our investigation is a well-known theoretical model of network goods (that is, products or services subject to network externalities) presented in Figure 1 (Farrell and Saloner 1986; Kauffman et al. 2000; Saloner and Shepard 1993). In this model, the value that consumers derive from a network product is a function of stand-alone benefit and network externalities. The term a denotes the stand-alone benefit that a consumer derives from the product features. The term b(N) represents the feature-independent value derived from a network of size N. Because user value increases with network size, b(N) is an increasing function of the installed base, N.
One challenge for empirical researchers is decomposing factors influencing consumer value with an awareness that some observable factors may be impossible to decompose between a and b(N). Another challenge is that the perceived value of a network product, a + b(N), is unobservable. With these challenges in mind, our approach to enhancing this theoretical model and creating the associated empirical model is similar to that employed by Brynjolfsson and Kemerer (1996). To test hypotheses, we use price (P) as a proxy for consumers' willingness to pay (W), or perceived value. This yields a formulation where
P [less than or equal to] W [less than or equal to] a + b(N)
As a result, we can derive a function of price as:
P = f(a1, a2 ..., b1(N), b2(N),...)
Using this approach, we expand the model by considering variables the literature suggests influence the perceived value of a software product. Doing so allows us to construct a richer, more-descriptive model for considering network effects and standards applicable to the unique problems of two-sided software markets.
While our study relies on research in the domains of information goods, the economics of standards, multiproduct complementarities, and software pricing, our primary context is the Web server software market. This context is referred to when developing hypotheses below. The Web server software market has been the focus of a number of recent studies (Cusumano and Yoffie 1998; Gallaugher and Wang 1999a; Gallaugher and Wang 1999b; Gogan and Applegate 1995; Quittner and Slatalla 1998; Yoffie and Cusumano 1999). Yet, despite high interest and earlier qualitative case studies and exploratory work, little rigorous empirical work has investigated this context. Web software represents the first commercially successful software supporting the evolving, ubiquitous, low-cost, global computing infrastructure. Therefore, understanding the dynamics of competition in this market and the factors critical to early market success should yield important insights for a number of current and future industries.
Network Goods and Externalities
Early attempts at modeling the behavior of information technology markets (e.g., workstations) have focused on evaluating the relationship between product features (e.g., processor speed and memory) and price (Lynch et al. 1990; Michaels 1979; Rao and Lynch 1993). However, some products or services become more valuable as more people use them. This phenomenon is referred to as positive consumption network externalities (Farrell and Saloner 1985; Katz and Shapiro 1985). Network externalities may be responsible for determining market winners in a variety of IT and electronic commerce areas including software, standards, online services, and virtual communities.
The positive relationship between value and installed base (i.e, the base of users for a particular product or service) develops from three factors: exchange, stability, and extrinsic benefits. All network goods foster some sort of exchange (e.g., content, money, programs). Therefore, each new network user adds potential value through exchange with other network members (Economides 1996). Because the number of network users is assumed to reflect long-term market stability (Katz and Shapiro 1992), consumers generally prefer firms with large installed bases. IT users are particularly concerned about being stranded with a failed and unsupported product given that many users have sunk investments in learning, files, and other resources that may greatly exceed a product's initial purchase price (Fichman and Kemerer 1993). Consumers also prefer dominant products that are more likely to attract extrinsic benefits such as supportive content, books, manuals, add-on products, and skilled labor (Parker and Van Alstyne 2 000; Shurmer 1993). In extreme cases, the lack of a rich market for these extrinsic products may lead to incomplete adoption and an inability to achieve critical mass (Moore and Benbasat 1991). Although many studies have discussed network effects within the context of proprietary standards (Church and Gandal 1992; Farrell and Saloner 1985, 1986, 1992), the implied significance of stranding concerns and extrinsic benefits are particularly noteworthy because they suggest that firms can generate network externalities even in environments that support open standards.
Examining the dynamics of network markets is critical because competition in these markets differs significantly from that in conventional markets. It can be especially difficult for newcomers to unseat an established rival in markets where the influence of network externalities is strong (Katz and Shapiro 1994). In light of the value that users place on installed base and market dominance, the importance of feature richness or quality may be diminished--thus, the "best" product or service does not always win (Rochert and Tirole 2001). Postrel (1990) found this to be the case in markets for stereo standards. In a widely cited paper, David (1985) suggested that this is the reason behind the dominance of the QWERTY keyboard format. (5)
In some cases, network externalities may also facilitate platform changes. This is manifested in the so-called bandwagon effect when consumers rush to adopt a new technology they expect will become dominant (Farrell and Saloner 1985; Katz and Shapiro 1992). Dominant firms may also be able to leverage their market share in a number of ways. These include using the installed base as a distribution channel for additional services, and distorting open standards to their own advantage (Shapiro and Varian 1998). The existence of proprietary inter-network effects provides market power by helping to eliminate close substitutes. Given the importance of establishing an early market share advantage, network market competition tends to develop early and can be particularly fierce. Such markets also tend toward monopolistic competition, with leading firms...
|
