NET Institute* 
 
www.NETinst.org 
 
 
 
Working Paper #04-01 
 
October 2004 
 
Indirect Network Effects and the Product Cycle: 
Video Games in the U.S., 1994-2002 
 
Matthew T. Clements & Hiroshi Ohashi 
 
University of Texas & University of Tokyo 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
* The Networks, Electronic Commerce, and Telecommunications (?NET?) Institute, 
http://www.NETinst.org, is a non-profit institution devoted to research on network 
industries, electronic commerce, telecommunications, the Internet, ?virtual networks? 
comprised of computers that share the same technical standard or operating system, and 
on network issues in general. 
Indirect Network Effects and the Product Cycle:
 Video Games in the U.S., 1994-2002
 ?
 Matthew T. Clements
 ?
 Hiroshi Ohashi
 ?
 September 2004
 Abstract
 This paper examines the importance of indirect network effects in the U.S. video game market be-
 tween 1994 and 2002. The diffusion of game systems is analyzed by the interaction between console
 adoption decisions and software supply decisions. Estimation results suggest that introductory pricing is
 an effective practice at the beginning of the product cycle, and expanding software variety becomes more
 effective later. The paper also finds a degree of inertia in the software market that does not exist in the
 hardware market. This observation implies that software providers continue to exploit the installed base
 of hardware users after hardware demand has slowed.
 Keywords: indirect network effects; penetration pricing; software variety.
 JEL classifications: C23; L68; M21.
 ?
 We have benefited comments from Paul Chwelos, Adam Copeland, Avi Goldfarb, and seminar and conference participants
 at Hitotsubashi University, the 2004 International Industrial Organization Conference, University of Tokyo and the Winter 2003
 American Economic Association meetings. Financial support from the NET Institute (http://www.NETinst.org) is gratefully
 acknowledged.
 ?
 Department of Economics, University of Texas, 1 University Station #C3100, Austin, TX 78712-0301. Phone: (512)
 475-8524. Fax: (512) 471-3510. clements@eco.utexas.edu.
 ?
 Department of Economics, University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo, 113-0033. Phone: +81-3-5841-5511. Fax:
 +81-3-5841-5521. ohashi@e.u-tokyo.ac.jp.
 1
1 Introduction
 Many high-tech products exhibit network effects, wherein the value of the product to an individual increases
 with the total number of users. Often these effects operate indirectly, through the market for a complementary
 good. For example, the value of a CD player depends on the variety of CDs available, and this variety
 increases as the total number of owners of CD players increases. Other examples include DVD players and
 discs and computer hardware and software. In this paper, we estimate indirect network effects in the market
 for video game systems. A system consists of a video game console (hardware) and game titles (software).
 The console itself does not have any value apart from facilitating the use of software. Other factors such as
 console price and quality being equal, a consumer would prefer to buy the console that offers a wider variety
 in game titles.
 Understanding indirect network effects is crucial for understanding why products like these succeed or fail.
 Moreover, since high-tech products tend to have short product cycles, it is also important to understand how
 the implications of indirect network effects differ over the course of the product cycle. Penetration pricing
 is often mentioned as a useful strategy in these kinds of markets.
 1
 By offering a low introductory price, a
 firm selling hardware can build up an installed base of consumers, which will lead to more software provision
 and a higher willingness to pay for hardware later in the cycle. It is also regarded as crucial for platform
 providers to have a broad selection of software available in order to promote console sales and raise royalty
 revenues.
 The purpose of this paper is to measure the effects of software variety and hardware price throughout
 the evolution of a network industry. The modern U.S. video game market provides an ideal opportunity
 to study this issue for two reasons: (a) the presence of indirect network effect is apparent; (b) because of
 the short product cycle and intense inter- and intra-generational rivalry, we observe multiple incompatible
 console systems in the market, providing us with sufficient data variation for analysis. To investigate the
 effectiveness of the business strategies, we must investigate the causal relationship between the hardware
 installed base and software title variety. Both the installed base and software variety are, in the end,
 endogenously determined as market outcomes. In order to address the endogeneity problem, we explicitly
 characterize the indirect network effect as an interaction between console purchases made by consumers and
 software supply chosen by game providers.
 To date, there has been only a handful of empirical papers studying indirect network effects. Among
 them, some estimate network effects only from the installed base of consumers. These include Bayus and
 Shankar (2003), Ohashi (2003), and Park (2002). These papers essentially model indirect network effects
 as though they were direct?i.e., consumers benefit directly from the existence of other consumers,
 2
 rather
 1
 See, for example, Shapiro and Varian (1999).
 2
 The direct network effect model is most appropriate for something like a telephone network. As more consumers use
 telephones, the value of the telephone to an individual consumer increases because it is possible to call more people. It is as if
 2
than indirectly through the market for a complementary good. The work deals explicitly with such markets
 includes Gandal, Kende, and Rob (2000), which focuses on the compact disc player market in order to explain
 the diffusion process of a single technology with network effects; Dranove and Gandal (2003), which estimates
 indirect network effects of DVD and Divx players; and Nair et al (2004) on personal digital assistants (PDA?s).
 These papers analyze a one-shot standards war, not a situation like ours in which technologies are evolving
 and one standard is dominant for some period of time but is eventually overtaken by a superior standard.
 This unique feature of the video game market helps us look at the evolution of console market shares and
 software availability for multiple technologies. Following an approach of Nair et al (2004), our empirical
 model draws extensively from the results of the theoretical work of Church and Gandal (1992; 1993) and
 the extension in Park (2002). Our paper contains three differences from Nair et al (2004) in identification
 strategy. The first difference is due to the nature of the market under study. While Nair et al (2004) analyzes
 the developing phase of the product cycle, our study period covers both developing and declining stages of
 consoles. Thus we need to account for this nonlinearity of product evolution. The second difference is that
 we do not use observed product characteristics as instruments to control for unobserved characteristics. In
 the video game market, the assumption may not hold in that observed characteristics may be positively
 correlated with brand image or other attributes for which we do not have data. The last difference is that
 we create cost-side instruments by using the fact that U.S. game consoles are the same as Japanese consoles.
 These instruments are similar to those proposed by Hausman (1996), but likely to be free from a criticism
 of Bresnahan (1996).
 We find that lowering price is particularly effective near the beginning of the product cycle: Demand
 for hardware is particularly elastic with respect to price at the beginning of the cycle.
 3
 Furthermore, we
 find that the elasticity of demand for hardware with respect to the available variety of software is relatively
 low at the beginning, and higher in the middle of the product cycle. Thus, while it is generally regarded as
 crucial to have some software available in order to launch hardware successfully, we find that on the margin
 an additional title does not have nearly as much effect on hardware demand at the beginning of the cycle as
 it does later. At the end of the cycle, when a hardware standard is becoming out-of-date relative to newer
 competitors, the elasticity of hardware demand with respect to both price and software variety is low.
 We also uncover a degree of inertia in the software market that does not exist in the hardware market.
 As a console becomes obsolete, both the installed base and software variety decrease. By characterizing
 the hardware and software decisions explicitly, we obtain an additional insight that growth of the hardware
 installed base diminishes first, and software provision slows down only after a lag. This finding implies that
 software providers continue to exploit the installed base of consumers after hardware demand has slowed.
 The organization of the paper is as follows. Section 2 describes important features of the U.S. video
 the quality of the telephone is increasing in the number of consumers.
 3
 Even if hardware is priced most aggressively near its introduction, the price may be highest then because the marginal
 production cost is much higher than later in the product cycle.
 3
game market and gives descriptive statistics from our data set. Section 3 presents the model used to
 analyze the indirect network effect. The model characterizes two economic activities in the U.S. video game
 market: hardware adoption and software provision. Section 4 describes the data and instruments used in
 the estimation. In the construction of the instruments, we use the fact that all the game systems in the
 data were manufactured in Japan during the period. Section 5 discusses the estimation results. Using these
 results, Section 6 describes the role of indirect network effects in the platform competition in the period from
 1994 to 2002. Section 7 concludes. A technical appendix follows.
 2 The modern U.S. video game market
 The U.S. market for home video game systems has grown enormously in recent years. In the period of our
 study, console sales more than doubled, from 6 million units in 1994 to 13.1 million in 2001. Total revenues
 for the industry were $9.4 billion in 2001, larger than total box-office revenues in the movie industry ($8.4
 billion in 2001).
 4
 Table 1 shows market structure in the U.S. video game market during the period from
 1994 to 2002 (because of data availability, the last year of our sample includes data only for the first quarter
 of the year. We discuss the data sources in Section 4.1).
 A video game system consists of hardware (the video game console) and software (game titles). Games
 are produced on cartridges or discs for use with the console. Hardware firms (like Nintendo) design and
 manufacture hardware and charge licensing fees to firms producing software; we will also refer to hardware
 firms as platform providers. Hardware producers generally produce some of their own software, and many
 independent firms produce software for one or more consoles. For the leading consoles, the vast majority of
 titles are produced by independent software publishers.
 In Table 1, we present eight major game systems in order of the total units sold in the sample of over
 seven years. Figure 1 is a simple way to verify the presence of indirect network effects. The figure plots
 yearly pairs of installed base and software variety for five major consoles in the period from 1994 to 2002.
 Installed base is represented as the number of cumulative console units sold up to a given time, and software
 variety is the number of game titles that receive sales in the market. In any given year, we calculate a share
 by console type for each of the variables. Generally, the size of the installed base of hardware users and the
 amount of software variety available are positively correlated for any given technology. As a console increases
 in popularity, both variables increase; as a console becomes out-of-date and is overtaken by competition,
 both variables decrease.
 The significant market growth in the U.S. video game market was accompanied by considerable upgrading
 of console quality, leading to a rapid turnover of systems. At the broadest level, three technical factors
 determine the quality of the systems as presented in Table 1: instruction word length (in bits) of either the
 4
 ?Recession? Don?t Tell the Video Game Industry,? New York Times, May 24, 2002.
 4
central processor (CPU) or graphics processor (GPU), clock speed (in MHz), and the amount of RAM (in
 mega bytes). The instruction word length is a measure of the maximum complexity of any single command
 sent to the processor, clock speed measures the number of such instructions that can be processed per second,
 and RAM provides temporary storage of information as a game is being played. The earliest machine in our
 sample was the Nintendo Entertainment System (NES), introduced in January 1986. The NES was an 8-bit
 console that ran at 4 MHz and had up to 8 kilobytes of RAM. The technical limitations of early systems
 restricted on-screen objects to two dimensions with a narrow range of colors. These technical characteristics
 were upgraded considerably in later consoles: Comparison of the NES console with the Sony PlayStation 2
 (PS2), introduced in October 2002, tells us that instruction word length increased by 16 times; clock speed
 by 164 times; and lastly, RAM increased by 16,000 times! The latest game systems can create more realistic
 sounds and improved graphics with faster and more complicated play.
 Since the late 1980s, game makers have introduced new game systems approximately every five years
 to satisfy the needs of consumers who look for more powerful games to play. The considerable quality
 upgrading leads to frequent console turnover, along with significant market growth. Table 1 indicates that
 market growth was also stimulated by aggressive pricing by console providers: For the first three years of
 the console introductions, the average price cut was about 28.4% per year, whereas the price drop for older
 consoles was modest at 7.5%. The console prices in general continued to drop even into the period when the
 consoles sales were in decline. The estimation of console demand must take account of this rise and fall of
 popularity in the console lifecycle.
 For any given year in the sample period, there have generally been two dominant consoles and a few fringe
 players. At the beginning of the sample period, the Sega Genesis and the Super Nintendo Entertainment
 System (SNES) dominated the console market (see units market share in Table 1). They were quickly
 replaced by the Sony PlayStation (PS) and Nintendo N64. By the end of the sample period, PS2 sales were
 growing fast (to date, the PS2 is the leading console and has sold approximately 60 million units worldwide
 5
 ). All the game systems in Table 1 were originally developed in Japan
 6
 and sometimes sold under different
 names there.
 7
 We use this fact to construct instruments to control for endogeneity of some variables in
 Section 4.2.
 Table 1 also presents information on the software market. The third row for each platform (% software
 variety) indicates the share in terms of the number of game titles sold by year. The total number of game
 titles is provided at the bottom row in the table. Software publishers provide finance for game development,
 manage relations with hardware providers, and perform packaging and marketing for game titles. Marketing
 5
 ?Playing Mogul,? New York Times, December 21, 2003.
 6
 During our sample period, American-made consoles were not strong competitors. The 3DO system, introduced in 1993,
 never captured more than 2% of the market. Microsoft?s Xbox was introduced in November 2001 and was not well established
 by the end of the sample period.
 7
 For those systems that have different names, we list Japanese names as follows with corresponding English names in
 parenthesis: Nintendo Famicon (NES); Super Famicon (SNES); and Sega Mega-Drive (Genesis).
 5
of game titles entails extensive advertising and promotion at trade shows, such as the Consumer Electronic
 Show and the Electronic Entertainment Expo. A software publisher may either develop games in-house or
 subcontract game development to independent developers. Platform providers also publish some software
 titles themselves, but these ?first-party? titles comprise a modest share of the software variety available for
 their own consoles (see %variety offered by platform provider in the table). A simple calculation from Table
 1 shows that the software share provided by platform providers starts with an average of 27.7% in the year
 of a console?s introduction, immediately declines to 21.5% in the following year, but hits another high of
 26.6% six years after the console release. From this point, the share declines. Some titles are available on
 multiple platforms; however, this is true for only 17% of the titles in our sample. Converting a game from
 one system to another has required additional development time and cost, and contractual agreements with
 platform providers sometimes require exclusivity to one game system.
 An independent publisher pays a royalty fee to a platform provider for every unit of a game title sold.
 Software licensing fees are the primary source of revenue for hardware producers. Although data on hardware
 cost are not available, it is widely speculated that all of the major consoles have been sold at a price near
 marginal cost. According to Brandenburger (1995), there is good reason why it is in the interest of a hardware
 provider to keep the price of the hardware itself low and profit through software sales instead.
 8
 When deciding
 whether to buy a console, consumers face uncertainty about the quality of the game experience they will be
 getting and about future software prices. A low hardware price signals the platform provider?s confidence
 that the consumer will want to buy games. There is also a holdup problem: Once a consumer buys a console,
 he is captive to that platform to some extent and can be induced to pay a lot for games. Knowing this,
 consumers are willing to pay less for the hardware. Although we do not explicitly model uncertainty, our
 estimation results are consistent with the theoretical predictions described above. The results discussed in
 Section 5 imply that lowering hardware prices is effective, especially early in the product cycle.
 3 A model of indirect network effects
 This section describes the estimation model we use to analyze indirect network effects in the U.S. video
 game market in the period from 1994 to 2002. Based on the descriptive statistics illustrated in Figure 1, this
 section seeks to establish the causal relationship between the hardware and software markets. The model
 comprises two main components, hardware adoption and software provision, and the indirect network effect
 is characterized by the interaction of these two components. We use a canonical model often used in the
 8
 The story of the 3DO Multiplayer reinforces this view. The company owned the rights to the most technologically advanced
 console on the market at the time. However, any firm could produce a Multiplayer. As with other platforms, software producers
 had to pay a royalty to 3DO. This royalty was unusually low ($3 per unit, as compared to approximately 5 times this for SNES).
 The hope was that the low royalties would foster a large variety of software, and that consumers would buy the console because
 of this. However, since hardware producers could not subsidize hardware production with profits from software royalties, the
 price of the hardware was high (two to three times the price of other consoles on the market at the same time). Even though
 the quality of the console was undisputed, consumers were unwilling to pay the high price of hardware.
 6
literature to describe the hardware and software markets. Our empirical model of indirect network effects is
 similar to that of Nair et al (2004), and thus we refer the mathematical derivation of our model to Nair et al
 (2004) and the unpublished appendix (available from the authors). We first describe hardware adoption and
 then turn to an empirical model of software entry. Section 4 addresses the endogeneity issue and introduces
 instruments used in the estimation. Section 5 discusses the estimation results for the model presented in
 this section.
 3.1 Hardware adoption
 Following the theoretical work on indirect network effects including Church and Gandal (1992; 1993) and
 Chou and Shy (1990), our model is based on consumer preferences for hardware and software. As discussed
 above, a video game system consists of a console technology and compatible game titles. Since a console itself
 has no stand-alone benefit, a consumer who purchases a console must purchase game software written for
 that system. We capture this aspect of preferences by using a symmetric constant elasticity of substitution
 (CES) utility function. This specification assumes that a consumer values all available game titles equally.
 Though tractable, this specification is not entirely consistent with the U.S. video game market. According
 to Coughlan (2001), only a handful of game titles shared a majority of the industry revenues: The top five
 percent of the titles made more than 50 percent of the software industry revenue in our sample period.
 Furthermore, more than 50 percent of the revenues for a particular game title were typically made in the
 first year after the game release. It is, however, difficult to extend this CES specification to incorporate
 heterogeneity of game titles. We instead use a measure of heterogeneity in game titles as an instrument to
 achieve identification in Section 4. Hardware adoption in our model is thus assumed to depend only on the
 number of game titles provided (where the quality of each title is assumed to be the average quality of all
 titles), the price of games, and the price and other characteristics of consoles. Since we do not intend to
 make contributions to the theoretical work on indirect network effects, but rather are interested in testing
 an implication of the model often used in the literature, we leave the derivation of the underlying model
 setup to the unpublished appendix.
 We use the television household as the purchasing entity, where each household has a unit demand for a
 video game system.
 9
 Video games are normally played by individuals whose ages range from 10 to 30 years
 old. Demographic data are, however, not available in our data set. Using an implication of the theoretical
 9
 In this paper, we use data for console games only, and all consoles require the use of a television as a monitor for game
 play. There is also a significant market for video games that can be played on a personal computer. However, this is commonly
 regarded as a significantly different market by those in the industry, since console games are generally played in the living
 room rather than at a desk and thus are more likely to be regarded as entertainment. Certain genres of games, most notably
 educational, are more popular in the PC format than any console format. Also, because there are no security measures built
 into PC hardware, piracy is more of a problem for PC games than for console games. The number of titles available in the
 PC format at any given time has generally been large, but the total sales volume is relatively small (less than 30% of the total
 market in 2001) and declining. It would be very difficult to incorporate PC data because of the inherent problems in tracking
 PC sales and imputing some percentage of PC use to game play.
 7
result, we assume that a representative household maximizes the following utility function at time t by
 choosing console j among J
 t
 +1 alternatives, one of which is the option of not purchasing a console:
 u
 jt
 = ?
 0
 + x
 j
 ?
 x
 + ?
 p
 p
 jt
 + ?h (N
 gt
 ) + ?
 jt
 + ?
 jt
 , (1)
 where u
 jt
 is a representative household?s utility from consuming console j that belongs to format g. Generally,
 g and j are the same. We use different indices to account for the backward compatibility of the PS2. Since
 the PS2 can be used to play PS games, but not vice versa, the PS2 format includes both the PS and PS2
 consoles, whereas the PS format includes only the PS console. Let p
 jt
 be the price of console j at time t
 (adjusted by the CPI), and ?
 0
 contains a constant term and other control dummies discussed in Sections 4
 and 5.
 We have data on three observed characteristics in Table 1: data width, clock speed and RAM, and denote
 console j?s observed characteristics by a vector x
 j
 . Utility from these observed qualities is, however, realized
 only through the presence of software titles: The quality is constrained by the console technology, x
 j
 , for
 some games but not others. Thus the vector of coefficients, ?
 x
 , would change over time with consumers?
 perception of the game quality. Since the quality of game software is not observable, x
 j
 ?
 x
 captures the
 average benefit from the console technology, and the deviation from the average is captured by an error, ?
 jt
 ,
 where E
 (
 ?
 jt
 )
 = 0. The unobserved error also reflects important factors that lead consumers to purchase
 a particular console that are not present in the data. A process of building console image, perhaps partly
 stimulated by advertising, may be one example of such a factor. In an effort to control for the time-varying
 consumer tastes, we include console dummies and allow for them to change over time. Note that console
 dummy variables substitute for the use of x
 j
 ?
 x
 because x
 j
 does not change within a console. Section 5
 explains the estimation method in detail.
 The indirect network effect is captured by h (N
 gt
 ). We use a Box-Cox transformation for the number of
 game titles, h (N
 gt
 ) =
 (
 N
 ?
 gt
 ? 1
 )
 /?, where ? is to be estimated. This transformation allows for linear (when
 ? = 1) and logarithmic (when ? = 0) specifications. We estimate only an indirect network effect, not a direct
 effect. There would be a direct network effect in the video game market if consumer utility, and thus console
 demand, depended on the number of consumers who own the same console. This would be the case if, for
 example, console users derived value from borrowing games from other users of the same console. Such an
 effect may be present in a local region, but with the country-level data at hand, we believe the indirect effect
 to be of far more significance.
 We impose assumptions on ?
 jt
 that generates the following standard nested logit model: On the first
 node, a TV household that does not have a game system decides whether or not to purchase a game console.
 If the household decides to buy, it makes a console choice on the second node. Following Berry (1994), a
 8
linear regression model for this two-stage logit is derived as follows:
 10
 ln (s
 jt
 )? ln (s
 0t
 ) = ?
 0
 + x
 j
 ?
 x
 + ?
 p
 p
 jt
 + ?h (N
 gt
 ) + ? ln
 (
 s
 j|B
 t
 =1
 )
 + ?
 jt
 (2)
 ? ?
 jt
 + ? ln
 (
 s
 jt|B
 t
 =1
 )
 + ?h (N
 gt
 ) ,
 where s
 jt
 is the share of the hardware market captured by console j ? J
 t
 during period t, and s
 jt|B
 t
 =1
 is
 the console j?s market share given that consumers decide to purchase video games at period t (i.e., B
 t
 takes
 1 when purchase is made); thus s
 jt|B
 t
 =1
 equals s
 jt
 / (1? s
 0t
 ), where s
 0t
 is the market share of the outside
 option at time t (Thus, s
 0
 +
 ?
 s
 j
 = 1). The mean utility of the outside option is assumed to be zero.
 Otherwise it should be incorporated in the constant term of the demand equation. This assumption is the
 standard treatment in the literature, and does not affect the estimates of own- and cross-price elasticities.
 We estimate the above model in Section 5. We turn now to the estimation model of software entry.
 3.2 Software entry
 We describe the determination of variety in game titles. When more consumers buy a particular console,
 software firms have more incentive to produce games designed for that console. We assume that there
 are many software firms that can potentially produce game titles for any particular console. According to
 Coughlan (2001), software firms normally publish more than one game title for a particular console. For
 example, Electronic Arts, the largest software publisher, published nearly 6.3 percent of the overall game
 titles during our sample period. To simplify the estimation model, however, we assume a single-product
 software firm provides its game title to a console j ? J
 t
 , where J
 t
 is the number of consoles available at
 t. Software production exhibits increasing returns to scale and free entry. Those consumers who purchase
 game titles already own a console. The market size for the software is thus the size of the installed base,
 IB
 gt
 . We use the index g to account for the backward compatibility of the PS2, already mentioned in the
 previous section. Each consumer in the installed base of console j has a CES demand for software s. Under
 the above assumptions, the symmetric Bertrand equilibrium determines the degree of available variety in
 game titles as:
 N
 jt
 = A
 j
 ? (IB
 gt
 )
 ?
 .
 Church and Gandal (1992; 1993), Chou and Shy (1990), and Nair et al (2004) use the same assumptions
 on the software market listed above to derive the equilibrium degree of software variety. The derivation is
 10
 The derivation is available in the unpublished appendix.
 9
also available in our unpublished appendix. We thus use the following reduced-form empirical model:
 ln (N
 jt
 ) = ?
 j
 + ? ln (IB
 gt
 ) + ?
 jt
 , (3)
 where ?
 jt
 is a mean-zero error. The model includes a console fixed effect, ?
 j
 . We adopt the usual definition
 of the installed base, in that IB
 gt
 , is the cumulative sum of console sales up to the time t ? 1. Thus, by
 definition, the size of the installed base never declines throughout the console lifetime. Other factors being
 equal, an older console is usually less attractive for game providers to supply titles, since such a console
 embodies older technology. Thus, the sensitivity of the installed base to the variety in titles, represented by
 ?, may be different for an old console as opposed to a new one. In order to consider this vintage effect, we
 include the hardware-age variable directly in equation (3), and allow it to interact with the installed base
 variable. This age variable counts the number of years after the console release. Accounting for the vintage
 effect is important in our analysis, because the data cover the initial as well as final stages of lifecycle for
 some consoles.
 The rest of the paper analyzes the two equations (2) and (3). As is common with models of network
 effects, our model has multiple equilibria, and we discuss this issue in the appendix. This paper essentially
 assumes that the data and estimation result correspond to the stable equilibrium.
 Before we turn to the estimation method, it is useful to discuss how the model presented above identifies
 elasticities in the market. We address two issues, hardware supply and dynamics. To estimate the price and
 variety elasticities of hardware adoption, it is more efficient to jointly estimate hardware demand in addition
 to supply equation by using cross-equation restrictions imposed by an imperfect-competition model. If the
 supply equation is misspecified, however, the resulting estimates would not be efficient, but even worse, not
 consistent. Under dramatic changes in the market with uncertain product life and lack of appropriate cost
 proxies, it is difficult to specify a hardware supply model for video game consoles. Since we are interested
 in obtaining consistent estimates of the demand model, this paper estimates only the hardware adoption
 equation.
 The model presented in this section is a static model. Since a video game console is durable, it may
 be more appropriate to use a dynamic model to describe the market. The major issue of dynamics in the
 video game market concerns the timing of both hardware adoption and software entry. In the hardware
 market, consumers decide to purchase a console based on their expectations of the future popularity of the
 console. We can think of two types of console buyers: (i) those who did not purchase a video game before;
 and (ii) those who own older game systems. At each point in time, the first consumer type compares the
 net benefit of purchasing a game system to the value of outside option, and the second type compares it to
 the net benefit of sticking with the older system. While the dimensionality of the problem makes it very
 difficult to model dynamic behavior that fully accounts for this trade-off, we try to capture it in the empirical
 10
implementation. We include console-time interaction dummies that proxy for console-specific events affecting
 expectations. To account for the type (ii) consumers, we allow for the installed base to depreciate, so that
 the outside market share changes with the flow of returning consumers. Although we cannot uncover the
 underlying decision-making process in the choice of purchase timing, we believe that our reduced-form static
 treatment still gives us consistent estimates of hardware adoption process.
 Timing of product launch in the software market is another issue. Forward-looking software publishers
 base their entry decisions on expectations about future growth of console sales. For example, if uncertainty
 exists as to the future profitability of a console, a publisher may wait before introducing its software com-
 patible with the console, or may wish to supply another console. Although modeling the product launch
 decision is beyond the scope of the paper, we try to capture this feature by including console dummies that
 proxy for differences in profitability across console types that influence expectations, and also by including
 console age interacted with installed base to capture publishers? anticipation of the console market size at
 a given point of the console lifecycle. While we believe that estimation results based on (3) still hold in a
 dynamic setting, we need to interpret the results with caution.
 4 Data and Estimation
 4.1 Data
 Our data on console sales and the number of available game titles come from the NPD Group, a market
 research firm. NPD Group collects data from approximately two dozen of the largest game retailers in the
 United States. These retailers account for approximately 65% of the U.S. market; from this data, NPD
 formulates estimates of figures for the entire U.S. market. These estimates do not take into account sales to
 rental outlets such as Blockbuster.
 We have monthly data for the period from January 1994 to March 2002. We excluded the two latest
 consoles, the Nintendo Game Cube and the Microsoft Xbox, due to small sample sizes (both of these consoles
 were introduced late in 2001). It is important to use monthly rather than annual data because of the short
 life cycle of hardware, and the even shorter life cycle of software titles (an individual title has positive sales
 for approximately 30 months, on average).
 For game consoles, we have retail revenues and retail quantities sold, broken down by console. We
 calculated the average retail price of a console from the data of revenues and quantities. We use the consumer
 price index (all urban consumers: all items less food and energy) to adjust the nominal resale price. The
 data on technical characteristics of the various consoles, noted in Section 2, were collected from console
 manufacturers? websites.
 11
 For game titles, we know when an individual title receives sales, broken down
 11
 See www.nintendo.com, www.playstation.com, and www.sega.com.
 11
by console. In addition, game titles are categorized by publisher (the firm that markets the title; publishers
 may develop games themselves or contract with independent game developers).
 We define the potential video game market as the number of people who had a TV but did not have a
 video game system prior to their purchase. The number of U.S. households with at least one television set
 in the study period comes from the Census Bureau?s 2003 Statistical Abstract of the United States. The
 size of the installed base by console and by month is obtained by the cumulative console sales up to the
 previous month. The installed base at the beginning of 1994 is obtained from Bayus and Shankar (2003,
 Table 1), which reports the installed base of NES (25.7), SNES (4.8), and Genesis (7.6) in 1994. The number
 in a parenthesis is in million units. Although the numbers are somewhat inconsistent with those reported
 in Brandenburger (1995b) for SNES (8.5) and Genesis (10.6), we use the data from the former. Use of
 the Brandenburger (1995b) data does not effectively change the results reported in the subsequent sections.
 Demand for upgrading (i.e., switching from old to new consoles) may be a concern due to rapid quality
 improvement. It is, however, impossible to read the magnitude of upgrading demand from the dataset. To
 check the significance of our concern on the presence of upgrading demand, Section 5 introduces depreciation
 in the console installed base.
 4.2 Instruments
 This subsection addresses identification issues in the base estimation model of hardware adoption and soft-
 ware entry. The estimation models are:
 ln (s
 jt
 )? ln (s
 0t
 ) = ?
 0
 + x
 j
 ?
 x
 + ?
 p
 p
 jt
 + ?h (N
 gt
 ) + ? ln
 (
 s
 j|B
 t
 =1
 )
 + ?
 jt
 (2)
 ln (N
 jt
 ) = ?
 j
 + ?
 1
 ln (IB
 gt
 ) + ?
 2
 [h_age] + ?
 3
 [ln (IB
 gt
 ) ? h_age] + ?
 jt
 (4)
 ? ?
 j
 + k (IB
 gt
 , h_age) ? + ?
 jt
 ,
 The variables are defined in Section 3. We first discuss the estimation of (2) and then turn to the estimation
 of (4).
 Hardware adoption Much of the previous literature makes the identifying assumption that x
 j
 and ?
 jt
 are not correlated with one another. Although it helps greatly by reducing the number of instruments needed
 in the estimation, this assumption may not be accurate in that observed characteristics could be positively
 correlated with brand image or other attributes for which we do not have data. Because of this concern,
 we use console dummy variables to control for unobserved attributes. Section 3.1 discusses the possibility
 that brand images and consumers? perception of observed quality could change over time. To account for
 this concern, we include different console dummies by year, along with year fixed effects in the estimation.
 12
Although our data are of monthly frequency as described in Section 4.1, we could not obtain meaningful
 estimates by including monthly dummies due to the lack of cross-sectional variation with only a few consoles
 in the market.
 Even after controlling for brand and time dummies, the deviation from the mean in some variables may
 still be correlated with the mean-deviation of ?
 jt
 . We are concerned that three variables are correlated with
 console j?s error, ?
 jt
 : they are within-group share, price and software variety. An obvious variable that
 is plausibly correlated with ?
 jt
 is ln
 (
 s
 jt|B
 t
 =1
 )
 , since s
 jt|B
 t
 =1
 contains part of the dependent variable, s
 jt
 .
 Console prices, p
 jt
 , may be endogenous, because if ?
 jt
 is correctly perceived by consumers and suppliers in
 the market, a console with a better image may induce higher willingness to pay, and thus sellers may be able
 to charge higher prices in an oligopolistic market. The last endogenous variable in the hardware adoption
 is the variety in game titles, N
 gt
 . This concern comes from the interaction with the software entry model
 (4), and the autocorrelation on ?
 jt
 . An increase in console demand at t ? 1, because of the change in the
 unobserved error, would inflate the installed base at t, leading to an expansion of the variety. Thus ?
 jt
 and
 N
 gt
 are positively correlated with each other in the presence of the autocorrelation in ?
 jt
 .
 In order to control for the endogeneity in console price, we employ two kinds of instruments from the cost
 side. The two instruments are constructed by using the fact that all the consoles in the data were imported
 from Japan. One instrument is monthly exchange rates between the Japanese Yen and the U.S. dollar (the
 data are from International Financial Statistics, 2002). Since most of the manufacturing processes occurred
 in Japan during the period, the U.S. retail price of a console should have been affected by exchange rates
 between Japan and the United States. We use a lag of one year for the exchange rate, because the console
 introduction date in the U.S. was usually one year behind that of Japan. Note, however, that this instrument
 is an industry aggregate, and does not vary by console type. The use of this instrument thus only helps
 identify the hardware demand through the variations of the instrument over time.
 The other cost-side instrument is the console retail price in Japan. The data are from various semi-
 weekly issues of the Famicon bulletin (in the period from January 1992 to December 1998) and from Nikkei
 Newspaper (from June 1996 to March 2001). We cross-checked the overlapped period to find that the price
 levels are similar across the two sources. We again take a lag of one year for the console prices in Japan, in
 view of the difference in the release dates between Japan and the United States. Since almost all consoles in
 the data were manufactured and sold in Japan, the Japanese console price would contain cost shocks, as well
 as effects of consumers? tastes for unobserved quality in Japan. Thus, if Japanese gamers? tastes differ from
 American tastes, the Japanese console retail prices serve as a cost-side instrument for retail prices in the U.S.
 market. Some evidence suggests that such a difference in tastes does exist.
 12
 The identifying assumption
 made here is reminiscent to that made in Hausman (1996) in his study of the ready-to-eat cereal industry.
 The identification condition is that all demand shocks are uncorrelated across cities in the country. While
 12
 ?New Riddle for Xbox: Will it Play in Japan?? New York Times, February 18, 2002.
 13
some papers (for example, Bresnahan, 1996) criticize this identifying assumption because of the importance
 of national advertising and fads for some products, our assumption of uncorrelated demand shocks across
 countries may be more reasonable, especially for the countries where cultural traits are very different.
 Because of our unique data source detailed in Section 4.1, we have instruments available from software
 side: the monthly average age of software titles provided to a console. Popular titles tend to stay in the
 market longer and attract more consumers to the associated console. Thus we expect that a console with
 more older-game titles achieve a larger within market share, ln
 (
 s
 jt|B
 t
 =1
 )
 . Note that the monthly average
 age of software titles does not simply represent a time trend by console, due to significant entry and exit
 of game titles. We also use console age as an instrument. This instrument is measured by the number of
 months that passed since the console introduction. The squared and interaction terms of the instruments
 mentioned above are also used.
 Software entry Our concern here is endogeneity of the installed base in (4). If software entry associated
 with console j is accelerated due to ?
 jt?1
 , an unobserved shock in the software market at t? 1, this shock
 would induce new console adoption and boost the share of the console, s
 jt?1
 , and hence the installed base
 in the next period, IB
 gt
 . Thus if ?
 jt
 is autocorrelated with ?
 jt?1
 , endogeneity in IB
 gt
 arises. We use
 as an instrument the monthly average age of software titles provided to a console, the same instrument
 used in hardware adoption. We can think of cases where the average software age correlates with the entry
 error. If potential entrants perceive the presence of many older titles as a sign of a profitable opportunity,
 the instrument would be positively correlated with ?
 jt
 . On the other hand, if they see it as a result of
 tough competition (i.e., that young titles cannot survive in a market), the instrument would be negatively
 correlated with the error. Thus the direction of the bias by use of this instrument, if it exists, could go
 either way. We thus rely on the statistical test of overidentifying restrictions to check if the instruments are
 orthogonal to the error. Section 5 reports that the test would not reject the orthogonality condition. We
 also include in the set of instruments console age, squared console age, and the interaction of hardware and
 software ages.
 5 Estimation results
 This section presents estimation results of the hardware adoption and software entry equations discussed in
 the previous section. We first estimate the equations independently using a two-stage least squared (2SLS)
 method, and then estimate them jointly. Important statistics in the hardware market are presented in Table
 1, and definitions of variables and summary statistics are in Table 4.
 It is known that the 2SLS method can produce severe inconsistency, if the instruments are weak. We
 thus check the explanatory power of instruments, conditional on the included exogenous variables in the
 14
first stage of the 2SLS method. We obtain a F-statistic for each of the endogenous variables discussed in
 Section 4.2. To conserve space, Table 2 reports the average value of the F-statistics. We find, however, that
 all the instruments used in this paper are not weak at the 95-percent confidence level of F statistics. The
 estimated coefficients in the table are obtained by regressing the dependent variable onto the exogenous and
 fitted values of endogenous variables.
 Hardware adoption Table 2 shows three estimation results from hardware equation. The first specifi-
 cation (H1) controls only for time and console effects, and does not include the interaction of the two. We
 use the instruments introduced in Section 4.2 to control for endogeneity in console price, software variety,
 and within-group market share. Since we have more instruments than we need to identify an equation, we
 can test whether the additional instruments are uncorrelated with the error by using the J-statistic (i.e.,
 the statistic for overidentifying restrictions). We find that the model (H1) does not fit well: Although the
 average F-statistics indicate that the instruments are not weak, the J-statistic rejects the hypothesis that the
 instruments are orthogonal to the error. In order to check for the presence of autocorrelated errors and the
 resulting endogeneity problem for software variety addressed in Section 3.2, we supplement the estimation
 with a test on whether the residuals are autocorrelated. We construct the AR(1) coefficient in the table
 by first estimating a coefficient of the lagged residual for each console, and then aggregating them by using
 console market share as a weight. The aggregated coefficient is found to be at modest level at 0.45, indicating
 a need to control for software variety.
 13
 The estimated coefficients on price and the network effect are not
 significantly different from zero. The parameter in the Box-Cox transformation, ?, is estimated at 0.99,
 rejecting the hypothesis of a logarithm specification on N
 gt
 .
 One possible explanation of the insufficient fit in (H1) is that the present model does not account for
 the dynamic nature of the industry: Consumers? attention to game consoles was presumably stimulated over
 time by the introduction of new game titles. It is thus unrealistic to think that consumers? perception of
 console quality (both observed and unobserved), represented by console dummies, is constant across time.
 An ideal estimation should rather allow for consumers? preference over consoles to evolve with time. To
 respond to this concern, we include different console dummies by year.
 14
 The underlying assumption here is
 that unobserved console attributes differ by year, and the derivation from the time-varying console dummies
 are obtained as a regression error. For all the specifications with the console-year interaction dummies, the
 parameter in the Box-Cox transformation is not estimated precisely, and we cannot reject the hypothesis that
 ? = 1. We therefore use a linear form in the software variety for the remaining hardware specifications.
 15
 13
 Taking a simple average in the aggregation of the AR(1) coefficients does not change this result much.
 14
 We include the console-year interaction dummies when the console received more than one million units of sales for the
 year. This method makes 27 interaction dummies.
 15
 The estimation result on ? does not preclude a logarithmic specification on software variety in these models; however, the
 linear specification fits better than the logarithmic specification. We also experimented with a power function of N . Again we
 found the linear specification to have a better fit to the data.
 15
The estimation result is under (H2). The J-statistic reports that the model now fits far better than the
 previous model. The value of autocorrelation coefficient does not change much from that reported in (H1),
 and thus we continue to treat software variety as an endogenous variable. Table 3 shows the yearly demand
 elasticity with respect to price by console (E
 p
 in the first row for each console), and its standard error.
 16
 The elasticity is estimated at -1.07 on average.
 17
 Though the elasticity values differ substantially across
 consoles, Table 3 documents that the console demand becomes less price elastic with the age of console. The
 elasticity in the first year of introduction was on average estimated at -1.92, and it increased with console
 age until the value reaches -0.52 when the console had been in the market for seven years.
 18
 Table 3 also shows the elasticity of demand with respect to software variety (E
 s
 in the third row for each
 console), and its standard error.
 19
 While the demand is found to be elastic at 1.89 on average, the elasticity
 values vary a lot across the consoles, from a high point of 5.51 for PS2 down to 0.75 for Saturn.
 In a market with strong indirect network effects, it is crucial to make sure that a new game system
 is widely adopted. Two ways a platform provider can do this are lowering the price of hardware and
 encouraging software entry. One interesting question is to measure the relative effectiveness of these two
 strategies. Following the idea of Gandal, Kende and Rob (2000), we calculate a ratio of E
 p
 and E
 s
 . This
 ratio measures the effect of console price equivalent to a one-percent increase in software variety (in absolute
 value). The result is in Table 3 (under ?E
 s
 /E
 p
 ). The ratios suggest that, as far as consumers are concerned,
 a 1% increase in game titles is equivalent to a 2.3% of price cut in game titles in the market, aggregating
 across years and consoles. In general, the ratio starts low with the introduction of a new console, increases
 to as large as 2.80 (for PS and Genesis), and eventually declines as the console retires from the market.
 Section 4.2 discusses that, without regard for the endogeneity, both the price, variety, and within-group
 share coefficients would be biased upward. In order to check the severity of the endogeneity concern, we
 estimate the model (H2) with the assumption that the explanatory variables are exogenous. The result with
 the exogenous variables is under (H3). The comparison with the result in (H2) points to the elimination
 of the endogeneity biases, although the differences of the OLS and 2SLS estimates of price and variety
 coefficients are not significantly different from zero . The OLS estimate on price (-0.50) is 30% higher than
 the 2SLS estimate (-0.71), but the 2SLS yields an estimate on software variety close to the corresponding
 OLS estimate (0.43). We use the result in (H2) as the base estimate for hardware adoption in the subsequent
 sections.
 Lastly we estimate the model under the assumption that the console installed base depreciates at an
 16
 The elasticity for console j at time t is
 (
 ?
 p
 / (1? ?)
 )
 p
 jt
 [
 1? ?s
 jt|B
 t
 =1
 ? (1? ?) s
 j
 ]
 . We obtain the standard error by
 using a delta method. The mathematical derivations are available upon request.
 17
 The yearly elasticity (for both price and software variety) is calculated as follows: we first obtain elasticities by console and
 by month, and then aggregate them by taking a simple average.
 18
 Only four consoles survived for seven years within the sample period: PS, Genesis, Saturn and SNES.
 19
 The elasticity for console j at time t is (?/ (1? ?))N
 jt
 [
 1? ?s
 jt|B
 t
 =1
 ? (1? ?) s
 j
 ]
 . We obtain the standard error by using
 a delta method. The mathematical derivations are available upon request.
 16
annual rate of 5 percent (The next subsection details how we incorporate the depreciation). This assumptions
 allows those consumers who own older consoles to purchase another game systems. The outside market share
 thus changes with the flow of returning consumers. The estimation result is very close to those of (H2),
 indicating that the results in (H2) are robust to the size of outside market share (The results are available
 in the unpublished appendix).
 Software entry We now turn to results of estimating the software entry model, (4). The estimation
 results are under (S1) - (S4). The specification (S1) leaves out the hardware-age variable from (4), by
 assuming that ?
 2
 = ?
 3
 = 0. Although the installed base coefficient is found significantly positive, the
 model (S1) does not fit well, and we could reject the hypothesis that the instruments are orthogonal to the
 error. One possible reason for this insufficient fit is that game providers, when supplying titles, may respond
 differently by the vintage of consoles. The specification (S2) thus include hardware age as an explanatory
 variable. The J-statistic shows that the model now fits moderately well with the instruments. The high and
 significant average autocorrelation coefficient in ? (0.92) indicates the need to use instruments for installed
 base. Comparison with the OLS result in (S3) shows that the instruments successfully control for the bias
 in the installed base coefficient, from 1.47 (OLS) to 2.94 (2SLS). The direction of bias in the coefficient is
 difficult to predict, because we also include hardware age variable in the estimation: Hardware age traces a
 trend of installed base, and the two variables are correlated at the level of 0.77. The F-statistic indicates
 that the instruments are not weak. Holding the hardware age being constant at the mean (3.6 years), a
 1% increase in the installed base expands the software variety on average by 4.52%. The result also shows
 that, holding the installed base size at the mean value, an older console would be less attractive for software
 providers to launch game titles: Console with additional one year in the market would lose 2.6 percent of
 software titles. Table 3 indicates that the elasticities of software variety with respect to the installed base
 (under Entry Elasticity wrt IB) are estimated to be 5.05 on average. Based on the estimation results in
 Tables 2 and 3, we discuss implications of network effects in the U.S. video game market in the next section.
 The last software specification (S4) employs a different set of the installed base to test the robustness
 of the base result. The specification (S2) assumes that the installed base size equals the sum of past sales
 (IB
 gt
 =
 ?
 s=t?1
 s=t0
 q
 gs
 , where q
 gs
 is the sales of format g at time t, where t0 is the console introduction
 date). Under (S4), we consider that the installed base depreciates at an annual rate of 5 percent (that
 is, the definition of the installed base of format g becomes
 ?
 s=t?1
 s=t0
 ?
 (s?t0)
 q
 gs
 , where ? is a depreciation
 rate).
 20
 The coefficients of installed base and its interaction with hardware age are smaller than those of
 (S2). Nevertheless, the main results in Section 6 qualitatively hold with this specification.
 20
 Since the data are of monthly frequency, we set ? as 0.9957 (= exp (log (0.95)/12)).
 17
Joint Estimation The specification (J) in Table 2 reports the joint estimation results from (2) and (4).
 Although we do not have cross-equation restrictions, the joint estimation produces more efficient estimates,
 when the errors from the hardware and software equations are correlated. We calculate the generalized
 method of moment (GMM) estimators, with an optimal weighting matrix being constructed by the 2SLS
 residuals from the specifications (H2) and (S2). We find that the obtained estimates are almost the same as
 those in (H2) and (S2), but their standard errors are somewhat reduced for the hardware equation, and does
 not change much for the software. Nevertheless, the values of standard errors calculated by the estimates in
 (J) are similar to those in Table 3 we discussed in the previous sections.
 6 Implications of the indirect network effect
 This section describes how the indirect network effect identified in the previous section plays a role in video
 game system competition in the period from 1994 to 2002. To analyze the relative strength of each console,
 we take a deviation in (2) and (4) from the averages to obtain the following equations:
 ? ln (s
 jt
 ) = ??N
 gt
 +?
 [
 ? ln
 (
 s
 jt|B
 t
 =1
 )
 + ?
 jt
 ]
 , (5)
 ? ln (N
 jt
 ) = ? [k (IB
 gt
 , h_age
 jt
 )]? +
 [
 ??
 j
 +??
 jt
 ]
 . (6)
 where g = j except that, for PS2, g is the sum of PS and PS2 because of the backward compatibility. We
 define k (IB
 gt
 , h_age) in (4), and ?y
 jt
 ? y
 jt
 ?y
 t
 , where an upper bar on a variable indicates the average of
 the variable across consoles available in the market in a given time t. The deviation in the console market
 shares and software provision can be decomposed into the network effect (the first term) and the non-network
 effect (the second term). We use the estimates under (J) presented in Table 2 to explore the importance of
 the network effect in explaining the market outcomes, relative to the industry average (i.e., the left hand
 side of the equations).
 We first discuss implications from console adoption (5). Figure 2 presents the relationship between the
 relative market shares and the difference in network effect for five selected consoles.
 21
 The figure illustrates
 that PS performed better than the average in console sales (i.e., the deviation in the market share is above
 zero), while the effect of software variety is stronger than the average only in 1997 and afterwards. On
 the other hand, Saturn?s performance was always below the average. The figure confirms that the software
 variety predicts well the changes in the relative strength of console market share: When more game titles
 enter the console market relative to the industry average, the consoles sell better than the average.
 In Figure 2, besides a fairly strong positive correlation, we also see a generally clockwise pattern in the
 21
 A figure for the other consoles is available from the authors.
 18
change of deviation of market share (i.e., ? ln (s
 jt
 )) versus deviation of software variety (i.e., ??N
 gt
 ). Since
 ? is positive, the observation of clockwise pattern was not affected by the estimation results obtained in
 the previous section. Consider the data points for the Sega Saturn for 1997 and 1998. From 1997 to 1998,
 the market share deviation dropped, but the software variety deviation did not change much. That is, the
 relative market share dropped, and this was followed by a drop in the relative amount of software variety.
 We take this to be an indication of inertia in the software market. Even after the growth of the installed
 base has slowed down, software publishers continue to develop new titles in order to reap profits from the
 established installed base. At some point, however, new software development tapers off, causing a further
 decrease in relative market share (i.e., a decline in the growth rate of the installed base). In the declining
 stage of the product cycle, market shares are more sensitive to the network effect than in the growth stage.
 22
 Figure 3 presents the deviation in the installed base from the average (i.e., [k (IB
 gt
 , h_age
 jt
 )]?) versus
 the deviation in software variety (i.e., ? ln (N
 jt
 )). Since hardware age essentially traces a trend of the size of
 installed base (and the correlation coefficient between the two variables is 0.77), it is difficult to separate the
 two variables and discuss the impact of the installed base. We thus take the first three explanatory variables
 in (4) as k (IB
 gt
 , h_age), and construct the installed base index in Figure 3.
 Figure 3 shows a positive correlation between the two. In contrast to Figure 2, however, the trajectory
 for each console is generally counterclockwise. This trajectory pattern does not depend on the values of
 ?.
 23
 Again considering the data points for the Saturn from 1997 to 1998, the deviation in the installed base
 changes significantly but the deviation in software variety does not. As a console ages, superior technology
 emerges, and growth in the installed base of the technologically inferior platform begins to decline. The
 relative number of software titles also declines, but only after a lag. Looking at the trajectories for different
 consoles, we see different rates at which the deviation in software titles declines relative to the deviation in
 installed base; but different consoles tend to follow the same counterclockwise pattern.
 The elasticity results from the previous section further illustrate the U.S. video game market during
 our study period. As discussed in Section 2, platform providers profit primarily through software royalties.
 They can only do this if they establish both sides of the software market: i.e., establish an installed base of
 customers, which then induces software entry and provides the ultimate source from which royalties will be
 drawn. Once the feedback process is under way, the consumer base and software variety build upon each
 other. To get the process started, however, it is particularly effective for a hardware producer to attract
 consumers through price. The price elasticity of demand for hardware by vintage points to the effectiveness
 of penetration pricing. In fact, as we describe in Section 2, console providers priced aggressively in the first
 few years of console introduction: Figure 2 illustrates that the price cut was on average 28% annually in the
 first three years of console introduction, while the price drop became more modest at 7.45% when console is
 22
 An exception is a small increase in the market share for the Sega Genesis in 1999. This is probably due to the consumer
 response to a large price cut. Sega cut the price of the Genesis by more than half in 1999 (see Table 1).
 23
 The feature in Figure 3 still holds when we replace ?g (IB
 gt
 , h_age
 jt
 ) ? with ?
 1
 ?IB
 gt
 .
 19
in the market for four years or longer. The price elasticity declines throughout the product cycle, indicating
 that price cutting is less effective as a console ages (Although we observe hardware prices declining over time,
 costs are certainly declining. Penetration pricing is thus most reasonably interpreted as an increasing price-
 cost margin. Holding the degree of competition fixed in the hardware market, an decrease in price elasticity
 implies an increasing price-cost margin.
 24
 It is likely that console producers have followed a strategy of
 penetration pricing in this sense).
 On the other hand, the elasticity of demand for hardware with respect to software variety is relatively
 low at the beginning of the product cycle, increases to a peak in the middle of the cycle, and then declines.
 This suggests that, while a low price is necessary to start the adoption process, software variety is necessary
 to continue adoption of the console. It is not obvious why the elasticity with respect to software variety
 is low at the beginning of the product cycle. The industry wisdom seems to be that software provision is
 crucial for the establishment of a console. This is a primary reason why hardware firms develop their own
 game titles: They want to ensure the supply of enough high-quality games to start the adoption process.
 However, our elasticity results indicate that an additional software title has little effect on software adoption
 early in the cycle. We could speculate that it is necessary to have a set of games to draw early adopters, but
 that there is little marginal impact beyond this critical level.
 25
 Later in the product cycle, as the console
 becomes more mainstream, the variety of software expands greatly, and the impact of each additional title
 is greater than before. Considering the incentives of a hardware producer, the best strategy in the middle
 of the product cycle is to encourage software entry directly, perhaps by lowering royalties or relaxing other
 restrictions on the acceptance of new titles.
 26
 Near the end of the cycle, when a platform is in decline, additional software has less effect on demand.
 This could be due to the fact that by then there is already a large set of software associated with the platform,
 so each additional title is not worth as much to consumers. At this point, the network effect becomes less
 important: Increases in software variety have less of an effect on hardware demand. Because of competition
 from newer consoles, there are not many new adopters. It is in the interest of the platform provider to
 capture as much surplus as possible from the established installed base.
 We have examined a market in which indirect network effects are crucial to the persistence of a technology:
 Without game software, video game hardware is useless. Other notable markets have this same characteristic:
 PCs and software, CD players and CDs, DVD players and DVDs, and probably more to come in the future.
 24
 The degree of competition, measured by the annual Herfindale index from 1994 to 2002 is calculated as 0.47, 0.34, 0.22,
 0.40, 0.48, 0.29, 0.28, and 0.38 (the values are divided by 10,000). This magnitude of changes in the index is swamped by the
 magnitude of those in the price elasticity. Although this inference is based on the homogenous Cournot competition, it is hard
 to believe, from our reading trade press, that the hardware market became concentrated by a similar magnitude of the elasticity
 changes during the study period.
 25
 To get an idea of what this critical level is, we would need to compare adoption patterns of successful and unsuccessful
 consoles; i.e. consoles that never quite caught on give us an indication of how much software provision is necessary to launch a
 console.
 26
 This point perhaps explains why 3DO did not succeed in the market: 3DO expended much of their attention to providing
 game titles in the early stage of product cycle, rather than to penetrating the console market.
 20
It would be reasonable to guess that the product cycle is similar in all of these markets, and thus that the
 diffusion strategies discussed here would be useful in these markets also.
 7 Conclusion
 Network effects and positive feedback loops have received a great deal of attention, academically and other-
 wise. In a market with network effects, competition among multiple incompatible systems is intense, because
 a small, initial advantage confers a larger advantage in the future. Many theoretical papers suggest various
 competitive strategies in a market with strong indirect network effects, but little work has been done on
 what strategies are most effective in each phase of the product cycle. To tackle this problem, this paper
 analyzes two sides of the U.S. video game market, hardware adoption by consumers and software provision
 by game makers, and estimates the elasticities of adoption with respect to console price and software variety.
 We find that the relative size of the elasticities of hardware demand differs over the product cycle: When
 a console is introduced, hardware demand is quite elastic with respect to price, but much less elastic with
 respect to software variety. As the console becomes mature, the price elasticity declines substantially, but
 the elasticity with respect to software variety increases substantially. The estimation results suggest that,
 while a sufficiently large set of software may be necessary to launch a system, a platform provider should use
 penetration pricing to encourage adoption at the outset (i.e., a lower price-cost margin). Once the platform
 provider succeeds in establishing an installed base, it can expand the installed base, and thus the profitability
 of the platform, by encouraging software entry. A wider variety of software is crucial for attracting later
 adopters to the platform.
 An important direction for future research is to characterize the incentives of platform providers more
 precisely. By explicitly incorporating hardware supply into our framework, we can expand upon the inferences
 drawn in this paper.
 A Comment on Multiple Equilibria
 A common implication of models of network effects is the existence of multiple equilibria. Generally, there
 is an equilibrium in which no consumers buy hardware and no software firms enter. This degenerate equi-
 librium is eliminated from our model because of the use of logarithm specifications in (2) and (4). With the
 assumption of linear h (N) in (2) (we find a better fit with this specification; see Section 5), the model has
 at most two equilibria; it always has one stable equilibrium, and the other equilibrium, if exists, is unstable.
 Substituting N
 jt
 in (4) with the right-hand side of (2) yields:
 ln
 (
 s
 jt
 1?
 ?
 i
 s
 it
 )
 = ?
 jt
 +B ? ?
 (
 t?1
 ?
 q=1
 MS
 q
 ? s
 jq
 )
 ?
 , (7)
 21
where B ? exp
 (
 ?
 j
 + ?
 jt
 )
 and MS
 q
 is the potential market size for video game consoles at time q. In a
 steady state, the left-hand size of (7) is monotonically increasing in s
 j
 , and the right-hand side is either a U
 shape (if ? > 1), or an inverse U shape (if ? < 1) with respect to s
 j
 . The stable steady-state equilibrium is
 where the left-hand side of (7) intersects with the right-hand side from the above. We assume that the data
 and estimation results correspond to the stable equilibrium.
 References
 [1] Bayus, B.L., and V. Shankar (2003): ?Network Effects and Competition: An Empirical Analysis of the
 Home Video Game Industry,? Strategic Management Journal, 24, 375-384.
 [2] Berry, S (1994): Estimating discrete-choice models of product differentiation. Rand Journal of Eco-
 nomics, 25, 242-262.
 [3] Brandenburger, A. (1995a): ?Power Play (A): Nintendo in 8-bit Video Games,? Harvard Business School
 Case 9-795-102.
 [4] Brandenburger, A. (1995b): ?Power Play (B): Sega in 16-bit Video Games,? Harvard Business School
 Case 9-795-103.
 [5] Bresnahan, T.F. (1996): ?Comment,? to Chapter 5 in The Economics of New Goods, Studies in Income
 and Wealth, 58, ed. by T. Bresnahan and R. Gordon. Chicago. NBER.
 [6] Bureau of Labor Statistics (2003): Consumer Price Index: all urban consumers, all items less food and
 energy, available at www.bls.gov.
 [7] Census Bureau (2003): Statistical Abstract of the United States
 [8] Chou, C., and O. Shy (1990): ?Network Effects without Network Externalities,? International Journal
 of Industrial Organization, 8, 259-270.
 [9] Church, J., and N. Gandal (1992): ?Network Effects, Software Provision, and Standardization,? Journal
 of Industrial Economics, 40, 85-103.
 [10] Church, J., and N. Gandal (1993): ?Complementary Network Externalities and Technological Adop-
 tion,? International Journal of Industrial Organization, 11, 239-60.
 [11] Coughlan, P.J. (2001): ?Note on Home Video Game Technology and Industry Structure,? Harvard
 Business School Case 9-700-107.
 22
[12] Dranove, D. and N. Gandal. (2003): ?The DVD vs. DIVX Standard War: Empirical Evidence of Network
 Effects and Preannouncement Effects,? Journal of Economics and Management Strategy, 12 (3), 363-
 386.
 [13] Gandal, N., M. Kende, and R. Rob (2000): ?The Dynamics of Technological Adoption in Hard-
 ware/Software Systems: The Case of Compact Disc Players,? RAND Journal of Economics, 31, 43-61.
 [14] Hausman, J.A. (1996): ?Valuation of New Goods under Perfect and Imperfect Competition,? in The
 Economics of New Goods, Studies in Income and Wealth, 58, ed. by T. Bresnahan and R. Gordon.
 Chicago. NBER.
 [15] International Monetary Fund (2002): International Financial Statistics Yearbook.
 [16] Nair, H., P. Chintagunta, and J.-P. Dub? (2004): ?Empirical Analysis of Indirect Network Effects in the
 Market for Personal Digital Assistants,? Quantitative Marketing and Economics, 2, 23-58.
 [17] Ohashi, H. (2003): ?The Role of Network Effects in the U.S. VCR Market, 1978-1986,? Journal of
 Economics and Management Strategy, 12(4), 447-496.
 [18] Park, S. (2002): ?Network Benefit Function in the Presence of Indirect Network Externalities,? Depart-
 ment of Economics, SUNY-Stony Brook.
 [19] Shapiro, C. and H. R. Varian (1999): Information Rules: A Strategic Guide to the Network Economy,
 Cambridge, MA: Harvard Business School Press.
 23
TABLE 1
 U.S. Video GameMarket
 1994 - 2002 Q1
 Platform Types Introduction Platform Main console characteristics
 (format) Year Provider CPU bits MHZ RAM (M bytes) 1994 1995 1996 1997 1998 1999 2000 2001 2002 Q1
 PlayStation September 1995 Sony 32 33.87 2 % Console units sold 11.15 28.83 49.82 61.38 55.06 39.17 16.89 12.87
 (CD-ROM) Mean Console Price (USD) 301.67 235.15 158.03 138.79 117.84 99.59 99.63 109.31
 % software variety 1.17 9.31 20.65 32.20 43.54 51.94 53.18 51.61
 %variety offerd by platform provider 25.00 19.04 19.48 20.45 20.18 18.20 16.26 15.08
 N64 September 1996 Nintendo 64 93.75 4 % Console units sold 24.99 38.69 31.27 28.98 30.59 7.51 1.54
 (Cartridge) Mean Console Price (USD) 199.61 159.33 138.06 121.92 105.23 90.09 84.42
 % software variety 0.20 1.68 5.62 11.34 15.41 15.12 14.65
 %variety offerd by platform provider 80.00 42.11 24.16 17.22 18.99 20.03 20.74
 Genesis September 1989 Sega 16 7.60 0.072 % Console units sold 57.87 42.68 18.56 4.12 5.31 3.53 0.67 0.01
 (CD-ROM) Mean Console Price (USD) 117.59 113.92 94.13 73.46 46.87 22.84 19.90 19.23
 % software variety 42.02 45.86 42.08 34.46 25.52 17.98 10.30 5.55 3.60
 %variety offerd by platform provider 28.63 28.89 29.88 32.37 34.94 33.20 26.56 21.19 19.40
 PlayStation 2 October 2000 Sony 128 294.91 32 % Console units sold 13.41 47.00 56.38
 (DVD-ROM) Mean Console Price (USD) 297.47 304.11 298.33
 % software variety 0.69 7.04 12.75
 %variety offerd by platform provider 2.22 5.90 8.37
 Super Nintendo September 1991 Nintendo 16 3.6 0.128 % Console units sold 36.43 37.71 15.88 5.11 1.61 0.12 0.01 0.0001
 Entertainment System Mean Console Price (USD) 115.01 121.19 121.99 94.94 75.47 60.90 52.87 53.40
 (Cartridge) % software variety 32.15 37.40 34.96 26.68 20.23 14.04 8.66 5.04 3.33
 %variety offerd by platform provider 8.47 8.92 10.36 13.18 16.67 21.03 25.44 24.55 17.46
 Dreamcast September 1999 Sega 128 200 16 % Console units sold 12.27 16.14 8.76 0.05
 (CD-ROM) Mean Console Price (USD) 198.34 182.79 93.83 50.25
 % software variety 1.16 7.39 12.07 13.49
 %variety offerd by platform provider 22.06 19.59 20.15 22.09
 Saturn May 1995 Sega 32 28 4 % Console units sold 4.93 10.49 2.15 0.44 0.05 0.01 0.000
 (CD-ROM) Mean Console Price (USD) 369.58 233.98 172.71 77.86 40.30 31.65 36.14
 % software variety 1.18 7.71 14.03 15.50 11.81 5.57 2.01 0.58
 %variety offerd by platform provider 43.75 31.45 27.67 30.07 33.57 35.64 28.76 0.00
 Nintendo Entertainment January 1986 Nintendo 8 1.8 0.002 % Console units sold 4.75 2.25 0.66 0.07 0.001
 System Mean Console Price (USD) 55.57 54.38 49.51 43.08 20.16
 (Cartridge) % software variety 25.83 14.40 5.75 2.50 0.94
 %variety offerd by platform provider 11.50 14.92 25.60 34.03 35.29
 Industry console sales (M units) 5.65 4.61 7.09 11.60 12.41 12.21 8.11 13.16 1.37
 Total No. Variety 1234 1436 1480 1518 1494 1514 1678 1945 473
 The platforms are in order of the total units sales in the period of 1994-2002. The eight platforms covered 99.4 % of the U.S. home video game market. The data of 2002 are up to the first quarter.
 % Console units sold is the console market share in the industry at a given year. Thus one can obtain console sales units by multiplying % console units sole by Industry console sales.
 % software variety is the share in the total number of softwrare titles available in the market at a given year. The number of software titles for a console is obtained by multiplying %software
 variety by Total No. variety listed at the bottom.
TABLE 2
 Estimation Results on
 Hardware Adotpion (2) and Software Entry (4)
 Variables Est. Std. Est. Std. Est. Std. Est. Std. Est. Std. Est. Std. Est. Std. Est. Std.
 Hardware:
 Constant -13.26 ** 0.40 -16.40 ** 1.87 -16.37 ** 0.96 -16.40 ** 1.56
 Price -0.05 0.14 -0.71 ** 0.25 -0.50 ** 0.15 -0.71 ** 0.22
 Number of Game Titles 0.11 0.09 0.41 * 0.17 0.43 ** 0.09 0.41 ** 0.13
 Within Group Share 0.79 ** 0.04 0.35 ** 0.09 0.60 ** 0.03 0.35 ** 0.06
 ? 0.99 * 0.44
 Software:
 ln(IB) 2.19 ** 0.10 2.94 ** 0.20 1.47 ** 0.04 2.30 ** 0.12 2.94 ** 0.21
 ln(IB)*Hardware Age 0.44 ** 0.12 -0.01 0.01 0.38 ** 0.09 0.44 ** 0.12
 Hardware Age -7.58 ** 2.05 -0.24 0.23 -6.40 ** 1.53 -7.58 ** 2.08
 Time Dummies
 Console Dummies
 Console Dummies by Year
 No. Observations
 R-squared
 1st stage F stats
 J statistics (D.F.)
 AR(1) Coefficient
 * Significance at the 95-percent confidence level.
 ** Significance at the 99-percent confidence level.
 The dependent variable for the hardware adoption is the logarithm of console market share minus the logarithm of the outside share. The console  market share is defined as the fraction
 of the TV households that do not have game systems by a given time. The hardware equation includes year dummies, console dummies, and interactions of console and year dummies (except for (H1)),
 which are not reported here. The instruments for hardware adoption are exchange rate (USD/JY), console prices in Japan (CPI adjusted), hardware age (the months passed since the
 console introduction), monthly average software age. The squared and interaction terms of the instruments are also used. The number of game titles are divided by 100 for the presentation
 purpose. 
The dependent variable for the software entry is the logarithm of the number of game titles provided to a console. The software equation includes year and console dummies, which
 are not reported in this table. The instrument used for software entry are monthly averaged software age, and monthly hardware age. The squared and interaction terms of the two age
 variables are also used. Heteroskedasticity-robust standard errors are used in the table. The estimates of the dummies not reported here are available in the unpublished appendix.
 Y Y
 Y Y
 Y
 Y
 Y
 Y-
 -- - -
 Y
 Y
 Y
 -
 -
 Y
 Y
 Y
 - -- - - -
 -
 -
 -
 -
 -
 -
 -
 -
 -
 -
 -
 -
 - -
 -
 -
 0.87
 -
 -
 -
 13.70 (9) - 0.35 (2) - 0.66 (2) 14.05 (11)
 ( J )
 SLS
 1.71E+04
 1055
 -
 Y
 562 562
 -
 0.43 **
 8.71E+04
 0.92 **
 0.96
 0.97 **
 -
 0.92 **
 493 493
 - -
 1.40E+03
 0.45 ** 0.48 **
 1.40E+03
 62.38 ** (8)
 (S 4)
 SLS
 Box-Cox on N Base Base 5% Depreciation on IB
 (S 2)
 SLS
 (S 3)
 OLS
 (H 2)
 2SLS
 (H 3)
 OLS
 (S 1)
 SLS
 (H 1)
 2SLS
 493
 7.97E+04
 562
 -
 -
 -
 -
 -
 No hard age
 -
 -
 -
 -
 -
 Y
 Y
 0.91 **
 -
 -
 562
 -
 4.52E+04
 20.39 ** (3)
TABLE 3
 Elasticities of Hardware Adoption (2) and Software Entry (4)
 Platforms Demand Elasticities 1994 1995 1996 1997 1998 1999 2000 2001 2002 Average
 with respect to:
 PlayStation Price (Ep) -2.92 -2.15 -1.33 -1.06 -0.90 -0.79 -0.83 -0.92 -1.37
       std. error (Ep) 0.60 0.77 0.47 0.37 0.32 0.28 0.30 0.29 0.43
 Software variety (Es) 0.17 0.67 1.55 2.25 3.16 4.50 5.86 5.56 2.97
       std. error (Es) 0.75 0.45 0.72 0.97 1.34 1.90 2.47 1.04 1.20
 - Es / Ep 0.06 0.31 1.17 2.12 3.51 5.67 7.02 6.02 3.24
 Entry Elasticity wrt IB 3.02 3.29 3.72 4.16 4.59 5.03 5.46 5.72 4.37
 PlayStation2 Price (Ep) -2.47 -2.21 -1.93 -2.20
       std. error (Ep) 0.44 0.78 0.34 0.52
 Software variety (Es) 5.55 5.66 5.31 5.51
       std. error (Es) 2.10 2.43 1.02 1.85
 - Es / Ep 2.24 2.57 2.75 2.52
 Entry Elasticity wrt IB 3.00 3.25 3.51 3.25
 Genesis Price (Ep) -1.01 -1.01 -0.88 -0.72 -0.45 -0.22 -0.18 -0.17 -0.58
       std. error (Ep) 0.36 0.36 0.32 0.26 0.17 0.08 0.07 0.17 0.22
 Software variety (Es) 2.43 3.31 3.40 3.03 2.20 1.53 1.03 0.66 2.20
       std. error (Es) 1.04 1.41 1.46 1.30 0.94 0.65 0.44 0.66 0.99
 - Es / Ep 2.39 3.28 3.87 4.22 4.83 7.09 5.56 3.77 4.38
 Entry Elasticity wrt IB 5.03 5.46 5.90 6.33 6.77 7.21 7.64 8.08 6.55
 Saturn Price (Ep) -3.77 -2.32 -1.70 -0.76 -0.39 -0.30 -0.33 -1.37
       std. error (Ep) 1.04 0.85 0.62 0.28 0.14 0.10 0.19 0.46
 Software variety (Es) 0.11 0.60 1.24 1.39 1.05 0.55 0.28 0.75
       std. error (Es) 0.84 0.49 0.67 0.64 0.46 0.27 0.60 0.57
 - Es / Ep 0.03 0.26 0.73 1.82 2.71 1.85 0.87 1.18
 Entry Elasticity wrt IB 3.09 3.43 3.87 4.30 4.74 5.17 5.61 4.32
 Dreamcast Price (Ep) -1.75 -1.61 -0.81 -0.45 -1.16
       std. error (Ep) 0.30 0.59 0.30 0.26 0.36
 Software variety (Es) 0.19 0.64 1.36 1.56 0.94
       std. error (Es) 0.36 0.41 0.62 0.76 0.54
 - Es / Ep 0.11 0.40 1.67 3.45 1.41
 Entry Elasticity wrt IB 3.02 3.29 3.72 3.98 3.50
 Nintendo Entertainment Price (Ep) -0.59 -0.57 -0.50 -0.43 -0.20 -0.46
 System       std. error (Ep) 0.22 0.21 0.18 0.16 0.18 0.19
 Software variety (Es) 1.86 1.19 0.50 0.20 0.08 0.77
       std. error (Es) 0.81 0.54 0.25 0.13 0.43 0.43
 - Es / Ep 3.16 2.11 1.00 0.48 0.39 1.43
 Entry Elasticity wrt IB 6.62 7.06 7.50 7.93 8.37 7.50
 Super Nintendo Price (Ep) -1.08 -1.09 -1.14 -0.92 -0.73 -0.59 -0.49 -0.49 -0.82
 Entertainment System       std. error (Ep) 0.39 0.39 0.41 0.34 0.27 0.22 0.18 0.26 0.31
 Software variety (Es) 2.03 2.74 2.84 2.35 1.79 1.25 0.87 0.71 1.82
       std. error (Es) 0.90 1.19 1.25 1.04 0.79 0.56 0.40 0.58 0.84
 - Es / Ep 1.88 2.52 2.49 2.54 2.43 2.13 1.77 1.45 2.15
 Entry Elasticity wrt IB 4.16 4.59 5.03 5.46 5.90 6.33 6.77 7.21 5.68
 N64 Price (Ep) -1.74 -1.37 -1.21 -1.04 -0.85 -0.78 -0.76 -1.11
       std. error (Ep) 0.48 0.49 0.44 0.37 0.30 0.29 0.28 0.38
 Software variety (Es) 0.02 0.11 0.42 0.94 1.37 1.73 1.67 0.90
       std. error (Es) 0.89 0.16 0.27 0.46 0.61 0.77 0.77 0.56
 - Es / Ep 0.01 0.08 0.34 0.90 1.61 2.21 2.21 1.05
 Entry Elasticity wrt IB 3.02 3.29 3.72 4.16 4.59 5.03 5.28 4.16
 Average Price -0.89 -1.87 -1.46 -1.08 -0.74 -0.81 -0.96 -0.80 -1.02 -1.07
       std. error (Ep) 0.32 0.52 0.50 0.39 0.28 0.24 0.28 0.33 0.29 0.35
 Software variety (Es) 2.10 1.50 1.34 1.42 1.35 1.35 2.07 2.32 3.53 1.89
       std. error (Es) 0.92 0.95 0.80 0.67 0.67 0.64 0.88 1.16 0.90 0.84
 - Es / Ep 2.48 1.60 1.32 1.54 1.99 2.74 2.73 2.79 3.61 2.31
 Entry Elasticity wrt IB 5.27 4.64 4.69 5.10 5.54 5.01 5.07 5.48 4.62 5.05
 Note:
 The elasticities are calculated based on estimates from (H2) and (S2) in Table 2.
TABLE 4
 Definitions and Summary Statistics for the Variables
 Descriptions Mean Std. Error Min Max
 console sales (quantity units in thousand) by month 148.16 285.35 0.00 2795.16
 CPI-deflated console price in the United States (in January 1978 U.S. dollars = 100) 1.01 0.69 0.13 3.84
 The number of game titles for a system 320 272 2 1244
 Installed base by format (in million households) 12.88 9.26 0.03 35.78
 Age of console system (in year) 6.01 4.10 0.08 16.08
 CPU / GPU (in bits) 51.6 47.5 8.0 128.0
 Clock speed (in MHz) 83.3 101.5 3.6 295.0
 RAM (in mega bytes) 14.0 14.5 0.1 36.0
 Current nominal exchange rate of $US/Japanese Yen 111.45 11.83 84 145
 CPI-deflated console price in Japan (in Yen) 16560 9562 1513.3 42500
 Average age of software titles by console system (months) 26.40 20.96 0.67 83.91
 Average lifetime of software titles by console system (months) 51.63 16.85 8.41 85.79
 Sample means of year dummies Sample means of console dummies
 1994 0.07 PS 0.16
 1995 0.10 PS2 0.03
 1996 0.13 Genesis 0.19
 1997 0.15 Saturn 0.14
 1998 0.13 DreamCast 0.06
 1999 0.13 NES 0.11
 2000 0.15 SNES 0.17
 2001 0.13 N64 0.13
 2002 0.02
 Sample size: 1055
FIGURE 1
 Rivalry in the Major Five U.S. Video Game Systems,
 1994-2002
 PS
 N64
 Genesis
 PS2
 NES
 1998
 2001
 2002
 2001
 1998
 1998
 1998
 1995
 1994
 0
 10
 20
 30
 40
 50
 60
 0 10 20 30 40 50 60 70
 Share of Installed Base (%)
 S
 h
 a
 r
 e
  
o
 f
  
s
 o
 f
 t
 w
 a
 r
 e
  
v
 a
 r
 i
 e
 t
 y
  
(
 %
 )
FIGURE 2
 Network Effects in the Hardware Market (5)
 Selected Consoles
 02
 97
 95
 NES
 Saturn
 Genesis
 PS2
 PS
 96
 01
 95
 9801
 94
 01
 -8.0
 4.0
 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8
 (
 )
 j
 t
 s
 l
 n
 ?
 gtN??
FIGURE 3
  Indirect Network Effects in the Software Market (6)
 for Selected Consoles
 PS
 PS2
 Genesis
 NES
 02
 00
 01
 3.0
 96
 98
 00
 95
 0295
 Saturn
 97
 00
 99
 -5 -4 -3 -2 -1 0 1 2 3
 (
 )
 j
 t
 N
 l
 n
 ?
 ( )?jtgt agehIBk _,?