Building the Value of Next-Generation Platforms: The Paradox of Diminishing Returns

Abstract

Next-generation technology provides users with new, advanced functionality that often renders the past technology obsolete, opening a window of opportunity for challengers. Major benefits can accrue to technology leaders, but for platform technologies that require complementary innovation from external complementors to create value for users, those benefits are limited by the difficulty of securing complements. The focus of this article is on the value users derive from the variety and quality of platform complements and its impact on leaders’ performance over time relative to followers. The article shows that next-generation platform leaders that build in-house complements and encourage broader participation by external complementors can enhance platform value to users at early market stages. Yet, later on, when the market has taken off, continuing to leverage these strategies will negatively affect the variety and quality of their complements, constraining their growth capacity and performance relative to followers. Leaders may thus lock themselves into suboptimal performance patterns and eventually fall behind their followers. The analysis provided herein offers new insights about what drives platform competition by highlighting the challenges platform leaders face, particularly in the growth stage of a platform’s market evolution, and the critical role played by complement quality in shaping platform competition over time.

Full text

Journal of Management
Vol. XX No. X, Month XXXX 1 –32
DOI: 10.1177/0149206316658350
© The Author(s) 2016
Reprints and permissions:
sagepub.com/journalsPermissions.nav
1
Building the Value of Next-Generation Platforms:
The Paradox of Diminishing Returns
Carmelo Cennamo
Bocconi University
Next-generation technology provides users with new, advanced functionality that often renders
the past technology obsolete, opening a window of opportunity for challengers. Major benefits
can accrue to technology leaders, but for platform technologies that require complementary
innovation from external complementors to create value for users, those benefits are limited by
the difficulty of securing complements. The focus of this article is on the value users derive from
the variety and quality of platform complements and its impact on leaders’ performance over
time relative to followers. The article shows that next-generation platform leaders that build
in-house complements and encourage broader participation by external complementors can
enhance platform value to users at early market stages. Yet, later on, when the market has taken
off, continuing to leverage these strategies will negatively affect the variety and quality of their
complements, constraining their growth capacity and performance relative to followers.
Leaders may thus lock themselves into suboptimal performance patterns and eventually fall
behind their followers. The analysis provided herein offers new insights about what drives plat-
form competition by highlighting the challenges platform leaders face, particularly in the
growth stage of a platform’s market evolution, and the critical role played by complement qual-
ity in shaping platform competition over time.
Keywords: next-generation technology; platform; platform competition; platform value; com-
plement variety; complement quality; complementors
Acknowledgments: I gratefully acknowledge Manuel Becerra, Daniel Fernandez, Andrea Fosfuri, Marco Giar-
ratana, Luis Gomez-Mejia, Torben Pedersen, Claus Rerup, Melissa Schilling, Fernando Suarez, Gianmario
Verona, Franz Wohlgezogen, and seminar participants at the London Business School 2012 Sumantra Ghoshal
Strategy Conference, Esade Business School, IE University, for their helpful feedback on earlier drafts of the article
and two anonymous reviewers and associate editor Anne Parmigiani for their valuable comments and suggestions.
All limitations are my own. I am indebted to Juan Santaló for his constant and valuable support in the early develop-
ment of this project and to the PhD Program and Strategy Department of IE Business School for generous funding
support of data collection.
Corresponding author: Carmelo Cennamo, Bocconi University, Via Roentgen 1, Milano 20136, Italy.
E-mail: carmelo.cennamo@unibocconi.it
658350JOMXXX10.1177/0149206316658350Journal of ManagementCennamo / Next-Generation Platform Value
research-article2016
by guest on July 28, 2016jom.sagepub.comDownloaded from

2 Journal of Management / Month XXXX
Introduction
Technological discontinuity often leads to a succession of technology generations in
which each one opens a window of opportunity for other firms to challenge incumbents
(Anderson & Tushman, 1990; Arthur, 1989; Chintakananda & McIntyre, 2014; Christensen,
Suarez, & Utterback, 1998). A technology’s next generation offers a “significant advance in
technical performance” (Lawless & Anderson, 1996: 1187) that may render the past technol-
ogy obsolete (e.g., Gomez, Lanzolla, & Maicas, 2016; Tushman & Anderson, 1986). Such
technological superiority can both trigger new users’ adoption by addressing the needs of
future customers and offer enough stand-alone value for current customers (early adopters)
to migrate from the past generation (see, e.g., Sheremata, 2004; Tushman & Anderson, 1986;
Xu, Venkatesh, Tam, & Hong, 2010). Thus, the “race for the market” starts anew with each
generation, and it generally favors new entrants at the expense of incumbents (Christensen &
Bower, 1996; Tushman & Anderson, 1986).
However, we know relatively little about how technology discontinuity affects technology
adoption and competition in network markets, where the technology is part of a product sys-
tem (Katz & Shapiro, 1992), so that the new technology alone may not offer sufficient addi-
tional value to encourage adoption because of too few other users or too few complementary
goods available for it (Ansari & Garud, 2009; Gupta, Jain, & Sawhney, 1999; Schilling,
2002, 2003; Suarez, 2004). Particularly for a certain class of technologies—platforms—com-
plementary products (from now on, “complements”) are an important determinant of the
technology’s value to users, as they extend its core functions (Binken & Stremersch, 2009;
Cennamo & Santaló, 2013; Claussen, Essling, & Kretschmer, 2015; Corts & Lederman,
2009; Gawer, 2014; Schilling, 2002). A technological platform is a stable set of technological
components that are shared and reused by developers of diverse complements (Baldwin &
Woodard, 2009; Gawer, 2014). Consider, for instance, smartphone platforms like Apple’s
iOS or Google’s Android and their stables of apps developers (e.g., Fingersoft or Instagram)
or gaming platforms like Microsoft’s Xbox or Sony’s PlayStation and their networks of game
producers (e.g., Electronic Arts or Ubisoft). A unique aspect of competition among these
platforms is that they cocreate value with external innovators (from now on, “complemen-
tors”) who develop platform complements. The resulting coordination problems affect the
evolution of the platform technology by influencing its value over time (Gupta et al., 1999).
Firms introducing next-generation platforms may capture “growth options” linked to net-
work effects (Chintakananda & McIntyre, 2014) through early entry; on the other hand,
because of the uncertainty inherent in the new technology and because of coordination prob-
lems with complementors, next-generation leaders may fail to establish a large enough user
base (Gupta et al., 1999) or to sustain their competitive position vis-à-vis later entrants
(Gomez et al., 2016; Suarez & Lanzolla, 2007). In this article, I ask how next-generation
leaders solve these coordination problems to generate value for users and how this solution
later affects their market performance over time.
I look at the impact of two specific strategies that have been proposed in the literature as
ways to exploit network effects to increase a platform’s value: encouraging external comple-
mentors to participate in the platform system (e.g., Boudreau, 2012; Venkatraman & Lee,
2004) and developing complements in-house (e.g., Gawer & Henderson, 2007; Hagiu &
Spulber, 2013; Schilling, 2003). In network markets, users derive value from the network of
other users of the technology (Fuentelsaz, Garrido, & Maicas, 2015; McIntyre & Subramanian,
by guest on July 28, 2016jom.sagepub.comDownloaded from

Cennamo / Next-Generation Platform Value 3
2009; see Suarez, 2004, for a review). For platform technologies, existing theory (e.g.,
Armstrong, 2006; Caillaud & Jullien, 2003; Parker & Van Alstyne, 2005; Rochet & Tirole,
2006) stresses the importance of indirect network effects: Users place more value on plat-
forms that offer a large number of complements, while complementors prefer platforms with
large user bases. Because of this self-reinforcing mechanism, platform providers that are the
first to attract a large enough user base are expected to lock competing platforms out of the
market. However, the literature has implicitly equated the value of a platform to users with
the number of its users and complements.
I extend previous work by focusing on the value users derive from the variety and quality
of complements (defined more precisely below), as opposed to their sheer quantity.
Specifically, I examine how the number and activity level of third-party complementors and
the number of in-house complements affect the variety and quality of complements of a next-
generation leader platform in relation to followers and, thus, affect the leader’s performance
over time. I argue that, in the presence of technological discontinuity, both external and
internal development of complements help the leader reduce the uncertainty that comple-
mentors encounter when transitioning to the next-generation platform and create knowledge
spillovers that increase the variety and quality of the leader’s complements at this early stage.
Yet, I argue, these strategies also generate frictions with complementors, which undermine
the leader’s capacity to expand the variety and quality of its complements in the growth stage
of the platform’s market evolution and, in turn, its performance in relation to followers. I test
the hypotheses in the context of the U.S. videogame industry over the period 1995 to 2008,
which covers three different generations of videogame consoles (i.e., platforms), with
changes in platform leadership, as well as the exit of incumbents and the entry of new com-
panies. I find general support for my arguments.
This study advances our knowledge about platform competition and technological disconti-
nuity in distinct ways. First, I present evidence that platform leaders face challenges not just in
the early stage of a next-generation technology but more critically in the growth stage, when
they can get locked into suboptimal patterns of performance. This highlights the importance of
accounting for the impact of technology and market evolution over the whole life cycle of the
platform, an area where we have limited knowledge (but see Claussen et al., 2015; Clements &
Ohashi, 2005). Second, while mainstream network effects logic stresses the increasing returns
from network size accruing to leaders (Arthur, 1989; Katz & Shapiro, 1992), I find enlarging
the network of complementors to be detrimental at later stages. Also, while some scholars have
discussed the technology leader’s difficulty in obtaining enough complements early in the tech-
nology life cycle (e.g., Ansari & Garud, 2009; Gupta et al., 1999; Schilling, 2002), I highlight
the critical role played by complement quality in shaping platform competition over time. By
removing the assumption that all complements offer comparable benefits to users and explicitly
accounting for the variety and quality of complements, I capture the heterogeneity of comple-
ments and, in turn, of competing systems. In doing so, I offer a new and more nuanced way of
studying how relative competitive positions change over time according to changes in the vari-
ety and quality of a platform’s complements. Finally, the study also presents an instance of
concurrent sourcing strategy (Parmigiani, 2007; Parmigiani & Mitchell, 2009; Puranam, Gulati,
& Bhattacharya, 2013) in the context of platform markets, documenting how the leader’s
emphasis on using outside versus inside sources of complements affects the next-generation
platform value, depending on the stage of the technology’s market evolution.
by guest on July 28, 2016jom.sagepub.comDownloaded from

4 Journal of Management / Month XXXX
Platform Systems and the Value of Complements
Products in a given technology generation share technical capabilities and characteristics
(Anderson & Tushman, 1990; Tushman & Anderson, 1986); a next-generation technology
improves on the capabilities of past generations: Digital (HDTV) versus analog (NTSC)
television, DVD versus VHS players, and 3G versus 2G are noticeable examples (e.g., Ansari
& Garud, 2009; Gomez et al., 2016; Gupta et al., 1999; Xu et al., 2010). In platform markets,
next-generation leaders must not only improve the product itself but also address adoption
challenges on both the users’ and the complementors’ sides: The platform must attract pro-
viders of complements to create value for users; yet, without a critical mass of users, comple-
mentors have limited incentives to commit to the new platform (Ansari & Garud, 2009;
Gupta et al., 1999). How to solve this “chicken-and-egg” paradox (Caillaud & Jullien, 2003)
is central in the literature on platform (two-sided) markets.
Most of the theoretical work originates from economics and focuses on pricing as the
main mechanism. Among nonpricing strategies, scholars have proposed both increasing the
number of outside complementors (e.g., Armstrong, 2006; Parker & Van Alstyne, 2008;
Rochet & Tirole, 2006) and producing complements in-house to get the spiral started (Hagiu
& Spulber, 2013; Schilling, 2003). Empirical studies have found a positive correlation
between the number of complements and the number of platform users (e.g., Clements &
Ohashi, 2005; Corts & Lederman, 2009) and have examined the impact of number of com-
plementors on market performance (e.g., Cennamo & Santaló, 2013) or on the rate of com-
plementary innovation (e.g., Boudreau, 2012; Venkatraman & Lee, 2004). For in-house
complements, most of the analysis is conceptual, though—the only exception being, to the
best of my knowledge, the qualitative study of the Intel case by Gawer and Henderson (2007),
which focuses on the conflicts this strategy can create with complementors.
Assessing the value of an individual platform requires assessing the value of the whole
system (the platform and its complements) in relation to competing systems. As I noted
above, most scholars have focused on the number of complements as the main dimension of
value (e.g., Armstrong, 2006; Boudreau, 2012; Clements & Ohashi, 2005). But some con-
sumers may care about the diversity of complements, not just their quantity (Zhu & Iansiti,
2012), or may prefer some complements to others (Binken & Stremersch, 2009; Corts &
Lederman, 2009; Panico & Cennamo, 2015). Indeed, it has been shown that high-quality
complements, generally referred to as “hits” (Corts & Lederman, 2009) or “superstars”
(Binken & Stremersch, 2009), can critically affect platform adoption. Thus, in assessing the
value of a platform to users, one should account for both variety and quality of complements
(Panico & Cennamo, 2015).
From a product-supply perspective, variety relates to different product versions along
(variation of) product attributes to meet the different needs and preferences of consumers
(Randall & Ulrich, 2001). From a user-demand perspective, variety defines individual users’
likelihood of finding a product with the specific attributes they want. Following previous
studies, I define complements’ variety at the platform aggregate level as the number and
diversity of complements available for it (e.g., Cottrell & Nault, 2004; Zhu & Iansiti, 2012).
In this sense, variety of complements defines the set of opportunities available to a platform’s
user for transacting and, thus, the likelihood of finding a product that matches his or her
preferences for particular types of complement. Depending on the kind of platform, one
could capture variety in terms of purpose (as for smartphone apps) or genre (as for games);
by guest on July 28, 2016jom.sagepub.comDownloaded from

Cennamo / Next-Generation Platform Value 5
in the empirical study below, I assess it in terms of genre. Quality is a complex, multidimen-
sional construct (Garvin, 1984). From a product-supply perspective, quality relates to reli-
ability, or how well the product conforms to specific standards (Garvin, 1984). From a
user-demand perspective, quality captures how much benefit the customer derives from con-
suming/using the product (e.g., Binken & Stremersch, 2009; Xu et al., 2010). In line with this
demand perspective is my definition of complements’ quality as the average level of con-
sumption benefits. Variety and quality reflect two distinct yet interrelated sources of platform
value. A larger number of more diverse complements increases a platform’s value by enhanc-
ing its appeal to a wider audience of consumers; a larger number of high-quality comple-
ments increases platform value by enhancing the consumption benefits from using the
platform.
This reconceptualization offers a more nuanced understanding of how indirect network
effects affect a platform’s value by distinguishing the size of the user and complementor
network from quality and variety of complements. In doing so, I also highlight how the indi-
rect network effects inherent in platform markets differ from the network effects in nonplat-
form network markets that involve standards such as 3G, word processing, or fax. In these
cases, the technology management literature typically identifies two main value dimensions:
technological superiority (e.g., Sheremata, 2004; Xu et al., 2010) and network benefits (e.g.,
Fuentelsaz et al., 2015; McIntyre & Subramaniam, 2009). These are also important dimen-
sions of value for platform technologies. Zhu and Iansiti (2012) showed that “platform qual-
ity,” the technological characteristics (e.g., processing power and speed, graphics) that make
one platform perform better than others (Gawer, 2014; Schilling, 2003), correlated to plat-
form adoption by users. Consumers also place a higher value on platforms with a larger
number of users because of the direct network benefits of interacting with other consumers
and because of the larger number of complements under the indirect network effects logic
explained above. Therefore, in this study, I control empirically for technological differences
and for indirect network effects in order to focus on the other critical dimension I propose,
platform complement benefits—that is, the benefits consumers enjoy from the variety and
quality of the set of platform complements. Table 1 compares the different value dimensions.
Next, I assess how such value emerges and evolves for platform leaders in relation to follow-
ers at early and growth stages of the market for a new-generation platform by assessing the
impact of the number and activity levels of third-party complementors and the number of
in-house complements.
Hypothesis Development
Building Platform Value: Mobilization and Growth Challenges
In network markets, it seems that pioneers exploiting technological discontinuities should
be able to initiate their increasing-return process before the competition and that the virtuous
cycle associated with network effects should then act as an isolating mechanism (Arthur,
1989; Katz & Shapiro, 1992; Lieberman & Montgomery, 1988; Suarez, 2004). However,
these (hypothetical) advantages do not accrue automatically. Capturing them requires next-
generation leaders to cope with the uncertainty of the transition from current- to next-gener-
ation technology (Chintakananda & McIntyre, 2014). Complementors have to adjust and
usually redevelop their offerings to fit the new platform (Claussen et al., 2015), at a time
by guest on July 28, 2016jom.sagepub.comDownloaded from

6 Journal of Management / Month XXXX
Table 1
Value Dimensions of Platform Technology
Value Dimensions Conceptualization Source of Value Mechanism
Platform
Technical
Quality
Intrinsic technology
attributes that enhance
the core functioning
of the technology
because of improved
services, productivity, or
interconnectivity between
layers of the system
Technical Function/
Performance
Increased technical performance
and new/improved functions/
design can help users realize
gains in productivity, ease of
use, and better services from
consumption of complementary
applications
Compatibility Interoperability with other
platforms’ complements extends
the range of use for the consumer
Platform
Complement
Benefitsa
The complements-related
benefits consumers enjoy
from using the platform
Complement
Variety
Complements of diverse type
can match a broader audience
of heterogeneous users with
different tastes—the greater the
variety of complements, the
greater the appeal of the platform
to a broad audience of users
Complement
Quality
Complements of higher-quality
increase consumers’ utility—the
more high-quality complements,
the greater the benefits to users
from using the platform and its
complements
Network Benefits The benefits consumers
enjoy from other platform
users and complementors
joining the platform
because of cross-network
externalities
Installed User Base Indirect network effects: the greater
the number of platform users, the
larger the market and potential
demand for complementors’
products and, thus, the expected
availability of complements
Direct network effects: the greater
the number of platform users, the
greater the user’s opportunities
to benefit from interacting with
other users.
aPlatform complement benefits are the focus of this article.
when it has yet to prove its technological and market effectiveness. In this early stage, both
users and complementors hold ambivalent expectations about the value of the platform and
are considering the (current and prospective) competing alternatives (Gupta et al., 1999).
Although platform providers reach out to complementors months or years before the actual
launch of the new platform to enlist their support and signal its value to prospective users,
both users and complementors may postpone their adoption decision until uncertainty is
resolved. This “excess inertia” (Katz & Shapiro, 1992) becomes a “shadow” entry barrier for
next-generation platforms (Sheremata, 2004). Thus, next-generation leaders face mobiliza-
tion challenges to attract users and complementors at this early stage.
by guest on July 28, 2016jom.sagepub.comDownloaded from

Cennamo / Next-Generation Platform Value 7
During the growth stage, when more and more users shift to next-generation platforms,
the “decisive battle” for technological dominance takes place (e.g., Anderson & Tushman,
1990; Suarez, 2004). As the next-generation technology evolves and sales take off, techno-
logical uncertainty for complementors dissolves. But the market may still be uncertain about
which platform will ultimately be the leader—or, perhaps, whether there will even be a clear
leader. In this stage, next-generation leaders face growth challenges to enhance the platform
value more quickly than competitors and attain platform leadership. While early-stage mobi-
lization challenges involve value creation (Claussen et al., 2015), growth challenges mostly
reflect complementors’ concerns over value capture (Adner & Kapoor, 2010; Ceccagnoli,
Forman, Huang, & Wu, 2012).
Encouraging Broader Participation of Third-Party Complementors
The level of third-party complementor membership—the extent a platform is supported
by a narrow-broad set of external complementors—will critically affect platform leaders’
ability to address mobilization challenges and build platform value, particularly at early
stages, by reducing the uncertainty around the next-generation technology (e.g., Garud, Jain,
& Kumaraswamy, 2002; Schilling, 2002; Suarez, 2004). Wade (1995) submits that techno-
logical uncertainty leads to bandwagon pressures, as the extent of other firms’ support influ-
ences expectations about the system’s viability. Support from a broad set of complementors
can generate momentum and legitimacy for the new technology. As Garud and colleagues
contend, “This enhanced legitimacy results in the deployment of technical and financial
resources that, in a self-fulfilling manner, generate the promised value from cooperation”
(2002: 202). Complementor membership here conceptually captures not just the number of
members but also their degree of participation in the platform system. A platform with few
active members (where most of the complements proceed from few members) and many
nonactive members will have a narrow set of third-party complementors supporting the sys-
tem, despite the large number of members (Venkatraman & Lee, 2004).
As more complementors join and participate actively in the system, they begin to compete
and therefore to experiment, producing eventually a greater variety of complements (Boudreau,
2012; Cennamo & Santaló, 2013). Research suggests that greater levels of third-party comple-
mentor membership can expand the number of complements being developed (e.g., Armstrong,
2006; Venkatraman & Lee, 2004) and accelerate the rate of complementary innovation (e.g.,
Boudreau, 2012). Also, the lead time for product introductions falls because of knowledge
spillovers (see, e.g., Augereau, Greenstein, & Rysman, 2006; Parker & Van Alstyne, 2008);
the individual knowledge developed by each complementor can be channeled by the platform
and transferred to all complementors through technical support and development program-
ming tools (Gawer, 2014; Schilling, 2003). This can help boost not just the variety but also the
quality of complements. Variety and quality will also increase in the wake of increasing com-
petitive pressure among complementors (e.g., Armstrong, 2006; Boudreau, 2012; Cennamo &
Santaló, 2013). Indeed, complementors do not just produce diverse complements in response
to competition (horizontal differentiation) but also try to differentiate their offerings on the
basis of quality (vertical differentiation): The market can function as a screening mechanism
for higher-quality complements (Cennamo & Santaló, 2013). In sum, by reducing uncertainty,
generating knowledge spillovers, and instilling competition-based incentives, greater levels of
by guest on July 28, 2016jom.sagepub.comDownloaded from

8 Journal of Management / Month XXXX
third-party complementor membership should enhance the next-generation platform’s value
at the early stages.
Hypothesis 1: In the early stage of the next-generation technology, the broader the set of external
complementors participating actively in the platform system, the more value the next-generation
platform leader will generate.
However, high levels of third-party complementor membership entail trade-offs as the
market grows and the system matures. Research suggests that greater participation by exter-
nal complementors may eventually reduce their individual innovation incentives because of
intensified rivalry and direct competition (e.g., Augereau et al., 2006; Parker & Van Alstyne,
2008). As the system environment becomes more hostile than other, less crowded systems
(Cennamo & Santaló, 2013, 2015; Venkatraman & Lee, 2004), some complementors may
supply less innovative, lower-quality complements or abandon the platform completely and
switch to competing systems (Boudreau, 2012). Complete switching is especially likely for
small, resource-constrained complementors with limited budgets for development and mar-
keting. Resource-rich complementors, too, may decide to launch their “best,” most novel
products in other platforms (Venkatraman & Lee, 2004). In my research setting, for instance,
Acclaim, developer of then-popular games such as All-Star Baseball, Alien, Double Dragon,
and BMX, decided in 1998 to stop developing games for Sega’s Saturn and move to Sony’s
PlayStation. Explaining the decision, Steven Lux, Acclaim’s marketing vice president at the
time, said, “At around $2 million or more to develop a game, you want to be certain you’ll
get some return on that investment” (Terry, 1998: 29). According to Sony’s product market-
ing director, Peter Dille, “As they [at Sega] got successful, they forgot to keep their develop-
ers and retailers happy” (Terry, 1998: 29).
Cennamo and Santaló (2013, 2015) argue that a platform system that offers limited profit-
ability for complementors ultimately becomes a “market for lemons” (Akerlof, 1970) where
a greater proportion of low-quality complements are on offer. Indeed, Venkatraman and Lee
(2004) show that complementors are more likely to launch their innovative complements on
new, less crowded platforms, even if these platforms have a smaller installed user base. As
the result is to lower complementors’ incentives to innovate, I expect greater levels of third-
party complementor membership to decrease a platform leader’s value in the growth stage.
Hypothesis 2: In the growth stage of a next-generation technology, the broader the set of external
complementors participating actively in the platform system, the less value the next-generation
platform leader will generate.
Developing Complements In-House
Building first-party complements—complements developed in-house by the platform
provider—is another way to reduce uncertainty during the early stage and induce early adop-
tion (e.g., Hagiu & Spulber, 2013; Schilling, 2003). Research on “concurrent sourcing”
(Parmigiani, 2007; Parmigiani & Mitchell, 2009) indicates that to hedge technological uncer-
tainty firms may both make their own product components and buy from external providers,
as this can help them understand the technology environment. Studies show that internal
R&D and external knowledge acquisition can be complementary (Cassiman & Veugelers,
by guest on July 28, 2016jom.sagepub.comDownloaded from

Cennamo / Next-Generation Platform Value 9
2006), particularly in markets subject to technological breakthroughs (Rothaermel & Hess,
2007). Complementarities may arise from “interrelated business activities, such as shared
production equipment and technological interdependence” (Parmigiani & Mitchell, 2009:
1065), or from sharing knowledge individually generated by internal and external suppliers
(Brusoni, Prencipe, & Pavitt, 2001; Puranam et al., 2013). Plural sourcing allows the firm to
set performance benchmarks that can generate a “virtuous cycle of continuous improvement”
across sourcing modes, particularly for novel technologies (Puranam et al., 2013: 1151).
In a platform’s early stage, when the technology is not yet fully understood and comple-
mentors face uncertainty, producing in-house complements can showcase the new platform’s
capacities and technological potential. The more evident the platform’s improvements over
older generations, the more likely it is that consumers will switch to it (Schilling, 2003;
Sheremata, 2004). By creating an initial consumer base, in-house development of comple-
ments can induce third-party complementors to expect a large future installed user base
(Hagiu & Spulber, 2013) and therefore to develop complements. Platform providers that
make their own complements can also steer complementors’ production towards higher vari-
ety by developing complements in new genres that use the platform’s new technical capabili-
ties. A case in point in my research setting is Nintendo, which with its own family-based
interactive games, such as Wii Play and Wii Sports, showcased how best to take advantage of
the new Wii console’s innovative motion-sensing controller and widen the market to so-
called casual gamers. Sega with its Sonic titles and more recently Microsoft with the Halo
series have used in-house complements to explore the enhanced graphical, audio, and techni-
cal capabilities of their new consoles in their early launch periods.
Producing complements in-house can also help increase complement quality by provid-
ing reference points for quality levels and by improving the ability to transfer critical infor-
mation to third-party producers so they can fully exploit the platform’s technical capacity
and optimize their products for it. For instance, Microsoft used the knowledge it gained
from developing popular games for its Xbox console to build a development kit worth
$10,000, which it licensed for free to producers willing to develop other games for the Xbox
(Schilling, 2003). Alexy, George, and Salter refer to this “selective revealing” of knowledge
as “a strategic mechanism to reshape the collaborative behavior of other actors in a firm’s
innovation ecosystem” (2013: 270), which—they argue—is particularly effective when the
partner faces high uncertainty. Thus, the technological uncertainty associated with the next-
generation platform should decrease, as should the barriers associated with initial learning
investments.
Hypothesis 3: In the early stage of the next-generation technology, the greater the number of in-
house complements, the more value the next-generation platform leader will generate.
However, first-party complements can create tensions with external complementors at
growth stages of the market (Garud et al., 2002; Gawer & Henderson, 2007; Zhu & Liu,
2015). Plural sourcing modes do not always reinforce each other (Puranam et al. 2013;
Rothaermel & Hess, 2007); indeed, they may be competing alternatives (Rothaermel & Hess,
2007). Research has started to investigate the optimal split between how much a firm makes
and buys (Puranam et al., 2013). Here, I argue that the evolutionary stage of the market for
the next-generation platform importantly determines the extent in-house production creates
complementarities.
by guest on July 28, 2016jom.sagepub.comDownloaded from

10 Journal of Management / Month XXXX
In deciding whether to contribute to a platform, complementors worry about whether it
can appropriate the value of their offerings (Ceccagnoli et al., 2012). A platform that exploits
in-house complements beyond the initial stage of market formation accentuates this concern
(Zhu & Liu, 2015) and may thus alienate complementors (Gawer & Henderson, 2007; Gil &
Warzynski, 2010). Particularly when the platform starts to gain technological dominance as
the market grows and technology matures, complementors will start to perceive problems
with their simultaneous cooperation and competition with the platform firm (e.g., Garud
et al., 2002; Gawer & Henderson, 2007; Zhu & Liu, 2015). Various developers, for instance,
lamented the uneven playing field for Sega’s Saturn system, suspecting Sega of favoring its
internal production studios with technical support and programming tools that were not made
generally available until months later (Kunni, 1998).
Complementors may end up locked into a platform where they confront a very powerful
competitor for their products: the platform itself. This co-opetition tension, and the ensuing
“shadow of the future” (Gulati, Puranam, & Tushman, 2012) complementors face, can
lengthen their time to market by pushing them to choose launch dates far removed from those
of the platform’s own products (Gil & Warzynski, 2010) and raise their promotional costs to
match the visibility of first-party complements. The resulting lower profitability reduces
their incentives for novel, high-quality complements. These value-capture problems can lead
to an influx of lower-quality complements and reduce the variety of complements in those
niches where the platform has its own products.
Hypothesis 4: In the growth stage of the next-generation technology, the greater the number of in-
house complements, the less value the next-generation platform leader will generate.
Data and Method
Research Setting: The U.S. Videogame Industry
I test the hypotheses in the dynamic U.S. videogame industry, a quintessential example of
platform systems, for which the importance of indirect network effects and the complementary
value of game titles has been effectively documented (e.g., Clements & Ohashi, 2005; Corts &
Lederman, 2009). Videogame consoles function as the technological platform videogame pro-
ducers use to develop their games and sell to platform users (gamers). The value of a given
console (e.g., Sony’s PlayStation) to gamers depends greatly on the games available for it. The
industry has witnessed seven technological generations since its inception in the early 1970s.
What differentiates a next-generation console from an existing one is mainly the improvement
of its underlying technological architecture, with superior operating performance resulting
from technical advancements (e.g., 32-bit vs. 16-bit central processing units), and other
improvements, such as graphics quality and speed. Here, I analyze the three most recent gen-
erations, which have seen intense competition among platform systems. This enables us to
explore heterogeneity among competing systems and capture the underlying competitive
dynamics. Also, I observe the market exit of incumbent Sega and new entry by outsiders like
Sony’s PlayStation and Microsoft’s Xbox. This allows us to explore heterogeneity in different
platform generations and account for incumbent-newcomer entry-exit dynamics.
The data set consists of monthly observations from May 1995 through June 2008 of 4,968
videogame titles and 10 consoles. I obtained these data from NPD Group, a U.S.-based
by guest on July 28, 2016jom.sagepub.comDownloaded from

Cennamo / Next-Generation Platform Value 11
market research firm that surveys the largest game retailers to compile data on the entire U.S.
market. It is the best source of data on the videogame industry, used already in prior research
(e.g., Clements & Ohashi, 2005; Corts & Lederman, 2009; Venkatraman & Lee, 2004). I
integrate this rich data set with information about the technological characteristics of con-
soles, which I collected from the manufacturers’ Web sites. Table 2 provides a stylized over-
view of the sample characteristics.
Measurement of Variables
Platform value. My analysis’ focus is on platform value stemming from the variety and
quality of platform complements. As discussed above, the literature tends to use the simple
number of complements as proxy for variety. For instance, in the same setting as mine, Clem-
ents and Ohashi (2005) and Corts and Lederman (2009) study the console-adoption decision
by consumers at time t as a function of the number of videogame titles available for a con-
sole at a given time t. However, the total number of games available at a given time does not
necessarily capture variety, as this number may reflect “more of the same” titles rather than
a diverse range. Consider two consoles, A and B, both having 50 gaming titles available but
with different distributions: For Console A, all the titles fall in either the “action” or “sport”
genres; for Console B, titles are more evenly distributed across the nine main videogame
genres (Cennamo & Santaló, 2013). While both consoles have the same number of titles,
we can expect Console B to be more valuable, since it offers a more diverse array of games,
which would appeal to a broader set of users. Measuring this variety dimension would thus
require us to account not only for the number of games but also for their typology’s diversity
(Cottrell & Nault, 2004). I capture this through the Gini-Simpson diversity index, which rep-
resents the probability that two entities (game titles, in my context) taken at random from the
population (i.e., the data set) are of different types (belong to different product genres, in my
context). More specifically, I determine complements’ variety as (1 – Ω) × log (n); where n is
the number of videogame titles released by third-party producers for a console up to a given
time t, and (1 – Ω) is the Gini-Simpson diversity index computed as 1 – ∑i pi2, pi being the
proportion of a platform’s videogame titles belonging to the ith game genre (i = 1, . . . , 9). I
take the (natural) log transformation to consider the decreasing marginal value of additional
game titles, consistent with other researchers. I assume equal weight across the genres. How-
ever, I accounted for the possibility that some genres can be more popular with users than
others by using a weighting factor reflecting a genre’s popularity (measured at the generation
level); key results do not change. Applying this formula in the example above, the values of
variety (diversity index) for the two consoles will be 1.96 (0.5) for A and 3.48 (0.89) for B. In
this case, although both consoles have 50 games, Console B is 1.8 times more valuable than
Console A in terms of complements’ variety.
I capture the quality of platform complements by counting the number of “hits” (very suc-
cessful game titles) available for a console at time t. It is problematic to unequivocally iden-
tify the specific attributes that make a game “high quality” in the eyes of customers, not least
because some of its technical aspects (graphics, speed, etc.) are conflated with the technical
characteristics of the platform (Xu et al., 2010; Zhu & Iansiti, 2012). In line with my concep-
tualization of complements’ quality is my assumption here that if a game achieves higher
sales, it must offer higher consumption benefits to customers; in other words, it is of higher
by guest on July 28, 2016jom.sagepub.comDownloaded from

12
Table 2
Sample Characteristics
Platform Provider U.S. Launch
Technological
Characteristics
Total Games
First-Party
Games (%)
Third-Party
Complementor
Membership Levels Variety Quality
Platform
Value
Installed
Base (000)
Market
Penetration
(%)CPU MHz RAM
Generation 5
Saturn Sega May 1995 32 28 4 251 31 0.8 4.33 4.97 21.52 1,382 1
N64 Nintendo September 1996 64 93.75 36 241 18 0.76 4.35 4.78 20.79 17,985 15
PlayStation Sony September 1995 32 33.87 2 882 14 0.93 5.46 5.06 27.63 30,126 22
Generation 6
Dreamcast Sega September 1999 128 200 16 250 23 0.9 4.29 4.66 19.99 4,009 4
Gamecube Nintendo November 2001 128 485 24 176 7 0.89 4.01 4.64 18.61 3,544 3
PlayStation 2 Sony October 2000 128 300 32 1,552 8 0.95 5.88 5.37 31.58 41,681 34
Xbox Microsoft November 2001 128 733 64 900 7 0.94 5.24 5.37 28.14 14,302 13
Generation 7
Xbox 360 Microsoft November 2005 512 3,200 512 289 10 0.88 4.38 4.35 19.05 10,116 9
PlayStation 3 Sony November 2006 512 3,200 256 154 13 0.87 3.8 4.07 15.47 4,668 4
Wii Nintendo November 2006 190 729 24 273 7 0.92 4.33 3.63 15.72 10,561 9
Note: For each platform technology generation, the pioneer of the next-generation is indicated in boldface. Figures for third-party complementor membership levels are
averages. CPU = central processing unit.
by guest on July 28, 2016jom.sagepub.comDownloaded from

Cennamo / Next-Generation Platform Value 13
quality relative to peer games. Following Corts and Lederman’s (2009) methodology, I deter-
mine the number of hits, log (h), by counting the number of titles whose total sales (over the
entire life span) fall above the 75th percentile of the sales distribution of all titles developed
for consoles of the same generation. As with variety, I take the log transformation of this
number. In the context of my example above, let us consider the case where Console A has
10 hits out of a total of 50 titles, while B has 30 out of 50. Console B clearly offers more value
to users: It has the same number of titles as Console A, but it has more diverse titles and they
are of higher quality. The overall resulting platform value is the interaction of the quality and
variety measures. Applied to the example above, numerically, Console B would be 2.6 times
more valuable than A (see Table A1 in the appendix).
First-party complements is the number of game titles produced by the console’s manufac-
turer at time t. In gauging the level of third-party complementor membership, I follow
Venkatraman and Lee (2004), who construct an embeddedness measure to account for how
many of a console’s game titles proceed from a narrower or broader set of game developers.
Developers’ embeddedness is computed therein as Σi(dji,t)2, where dji,t denotes the proportion
of platform j’s game titles released by developer i on date t for platform j. This index varies
between 0 and 1. Values close to 1 mean that a platform is tightly embedded, with few game
producers, which in our case represents low complementor membership. Accordingly, I take
the inverse of this index to ease interpretation. Next-generation platform leader is a dummy
variable, leader, which I set to 1 if the console is the first of its technological generation, 0
otherwise.
I control for other factors that have been shown to critically affect the supply of comple-
ments to a platform in the videogame industry (Clements & Ohashi, 2005; Corts & Lederman,
2009). Specifically, platform age—the number of months since a platform’s launch date—
provides an important proxy for life-cycle stage, which might capture complementors’
expectations about the residual useful life of the technology and affect their decisions to
produce and launch titles. Platform installed base is the cumulative unit sales of the platform
up to time t, and rivals’ installed base is the total cumulative sales for all other competing
platforms belonging to the same technology generation and active at time t. Table 3 shows
the descriptive statistics and correlations.
Statistical Methods and Analysis
I test my hypotheses of the effect of third-party complementor membership and first-party
complements on platform value of the next-generation leader by assessing the impact of their
interactions with the leader dummy, in line with recent studies on first-mover advantages
recommending testing the interaction rather than main effects (e.g., Lieberman &
Montgomery, 1998; Suarez & Lanzolla, 2007). I thus estimate the following model at early
and growth stages:
Platform valueleader first party complementsl = +
jt 1j jt 2
ββ
×-eeader
third party complementor membership first party
j
jt 3
+ ×
--
β
+ +
jt
4j
t
complements
third party complementor membership
ββ
-55j6jt7j8
tj
t
+ X+ + T +leaderββαβ ξ,
where X is a vector of control variables, α and T, platform and time fixed effects.
by guest on July 28, 2016jom.sagepub.comDownloaded from

14
Table 3
Descriptive Statistics
Variable 1 2 3 4 5 6 7 8 9 M SD Min Max
1. Platform value 1 17.63 7.23 0.26 31.62
2. Variety .71*** 1 4.97 1.26 0 7.13
3. Quality .60*** .86*** 1 4.12 0.97 0.69 5.37
4. Leader −.27*** −.16*** .12*** 1 0.33 0.47 0 1
5. Third-party complementor membership .27*** .61*** .50*** −.18*** 1 0.89 0.09 0 0.95
6. First-party complements .72*** .73*** .70*** .01 .32*** 1 44.16 32.31 1 125
7. Platform installed base .63*** .70*** .61*** −.44*** .53*** .54*** 1 8.47 1.22 4.71 10.55
8. Rivals’ installed base .24*** .25*** .36*** .03 .24*** .20*** .34*** 1 9.34 1.17 0 11.03
9. Platform age .74*** .73*** .74*** −.01 .34*** .61*** .64*** .32*** 1 29.77 20.01 1 82
Note: n = 465.
***p < .01.
by guest on July 28, 2016jom.sagepub.comDownloaded from

Cennamo / Next-Generation Platform Value 15
I estimate the model with fixed-effects, instrumental-variable estimators. Fixed effects at
the platform level allow the unobserved individual effects coming from heterogeneity across
platforms (that may arise, for instance, from differences in entry time, consumer perceptions
of quality attributable to the brand of the company sponsoring the platform, or technological
capabilities) to be correlated with the included variables. However, fixed effects assume that
differences in platform value and demand across platforms due to these unobserved factors
remain relatively stable over time. This assumption may be strong as, for instance, the per-
ception of the brand associated with a platform may change over time (e.g., Zhu & Iansiti,
2012) as the market expands and consumers become more heterogeneous. To capture these
possible changes over time at the individual platform level, beyond the effects captured by
the year and quarter dummies, I constructed a set of interaction terms between the platform
dummies and the variable reflecting the potential market for videogame consoles, that is, the
number of households with a television at time t.
I also estimate a system of equations (Greene, 2003) relating platform value to platform
market penetration to assess the impact on platform performance (details and results of this
analysis are included in the appendix).
Endogeneity and identification procedure. Because of the postulated virtuous cycle
(i.e., network effects) between complements and installed user base, the third-party comple-
mentor membership and installed base variables will likely affect, but also be affected by,
changes in the dependent variable. This creates an endogeneity problem (Greene, 2003):
They will be correlated with the error and violate the assumptions of ordinary least square
models. Following research in the field (e.g., Clements & Ohashi, 2005; Corts & Lederman,
2009), I treated the installed base and third-party complementor membership variables as
endogenous and correct for it via instrumental-variable estimators. I mainly follow Cle-
ments and Ohashi (2005) and Corts and Lederman (2009) for the identification procedure
and the exogenous variables I used to instrument these variables. The appendix explains the
details of such procedure. To reduce multicollinearity between the interaction term and its
main effect components, I standardized the components before multiplication and mean-
centered squared term variables (Smith & Sasaki, 1979). Specific analysis reveals no mul-
ticollinearity problems.
Identifying the different market stages. My research interest is in understanding the deter-
minants of platform value at the early and growth stages of a platform’s market evolution.
Figure 1 depicts the users’ adoption curve of a representative generation of videogame con-
sole technology. The classic S-shaped curve of technology evolution and the different stages
are clearly visible. For each generation, I separated the early and growth stages by taking
as the cutoff the point where the curve abruptly steepens (when sales take off; Anderson
& Tushman, 1990). This point coincides approximately with the 24th month following the
beginning of each generation. I demarcate the end of the growth stage with the inflection
point in the industry’s cumulative sales curve. This focus is consistent with Suarez and Lan-
zolla’s (2007) recommendation that scholars investigate the mechanisms of early movers’
advantage during the period from the first product introduction to the onset of maturity.
Points past this onset of maturity are thus outside this study’s present sphere of interest. In
the sample, per-month sales peaked in July 2000 for the fifth generation (the one shown in
Figure 1) and in July 2007 for the sixth generation.
by guest on July 28, 2016jom.sagepub.comDownloaded from

16 Journal of Management / Month XXXX
Results
Analysis for the system of equations reported in the appendix documents the relationship
between a console’s value and its market penetration. Results confirm the positive and
strongly significant impact of platform value on a platform’s market penetration, particularly
the quality component of platform value. This analysis also shows that the main effect of
being the leading next-generation platform is significantly negative on both platform value
and market penetration, confirming the mobilization and growth challenges faced by plat-
form leaders of a next-generation technology.
Table 4 reports estimation results for the effect of the key variables of interest. Models 4.1
and 4.4 show the coefficient estimates only for the control variables; I successively add the
key explanatory variables to test my hypotheses—Models 4.3 and 4.6 thus represent the full
model for the early and growth stages, respectively. It is worth noticing that the leader vari-
able is negatively related to platform value in the growth stage, confirming results from the
system of equations estimation. However, this variable drops in the early stage because of
multicollinearity due to limited variation across time. Because of the use of platform fixed
effects, variation in leader, and in other platform-level variables that are constant over time,
comes from variation in the set of platforms in the database at a given time (because of entry
and exit, and change across generations). Using interactions between platform dummies and
the potential market variable further reduces this variation, considering also the reduced
number of observations for the early stage.
Model 4.2 shows that the interaction term Leader × Third-Party Complementor
Membership is positive and significant in the early stage; this result is also confirmed
when including the first-party complements in Model 4.3 and, thus, confirms Hypothesis
1. However, as the second hypothesis predicts, greater complementor membership also
Figure 1
Next-Generation Platform Evolution
by guest on July 28, 2016jom.sagepub.comDownloaded from

17
Table 4
Determinants of Platform Value at Early and Growth Stage
Early stage Growth stage
Variable Hypothesis Model 4.1 Model 4.2 Model 4.3 Hypothesis Model 4.4 Model 4.5 Model 4.6
Leader × Third-Party
Complementor Membership
1 0.83*** 1.34*** 2 −22.67*** −29.99***
(0.21) (0.38) (6.06) (10.42)
Leader × First-Party
Complements
3 0.98*** 4 −3.47**
(0.34) (1.55)
Third-party complementor
membership
0.85 3.91 0.49 −29.11*** −81.95*** −112.45**
(4.04) (4.58) (6.94) (3.71) (31.73) (50.93)
First-party complements 0.09*** 0.11*** −0.00 −0.01
(0.02) (0.02) (0.01) (0.03)
Leader −17.21** −68.41** −987.18***
(9.11) (30.18) (226.58)
Platform installed base −1.18* −0.47 0.39 6.55*** 5.79*** 1.25
(0.68) (0.64) (0.64) (0.62) (0.65) 1.04
Rivals’ installed base 0.10 0.08 −0.08 −2.08*** −1.66*** −1.49**
(0.09) (0.09) (0.09) (0.40) (0.58) (0.71)
Platform age 0.16 0.18** 0.08 0.16*** 0.13** 0.32***
(0.10) (0.08) (0.11) (0.04) (0.06) (0.06)
Platform age2 −0.02*** −0.02*** −0.01*** −0.00*** −0.00** −0.00***
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
N observations 129 129 129 336 336 336
F462*** 499*** 424*** 900*** 9,651*** 636***
*p < .10.
**p < .05.
***p < .01.
by guest on July 28, 2016jom.sagepub.comDownloaded from

18 Journal of Management / Month XXXX
intensifies competition among complementors and hurts the next-generation leader dur-
ing the growth stage. The Leader × Third-Party Complementor Membership coefficient in
Models 4.5 and 4.6 is indeed negative and strongly significant, supporting Hypothesis 2.
It is noteworthy that the magnitude of this coefficient is greater in Model 4.6 than in
Model 4.3 (the positive effect in the early stage). Hypothesis 3 is also confirmed: The
interaction between leader and first-party complements is positive and strongly signifi-
cant in Model 4.3. I also find support for Hypothesis 4—Model 4.6 shows a negative and
significant impact of Leader × First-Party Complements on platform value during the
growth stage.
Decomposing the Effects: Variety and Quality
To gain further insights, I test separately the effects on the variety and quality compo-
nents of platform value. Table 5 reports these results. The first point worth mentioning is
that first-party complements’ most significant effect is on the quality component of platform
value—positive in the early stage and negative in the growth stage. The result for the main
effect of third-party complementor membership is particularly interesting; it positively
Table 5
Impact on Complement Variety and Quality
Early stage Growth stage
Variable Hypothesis Variety Quality Hypothesis Variety Quality
Leader × Third-Party
Complementor Membership
1 −0.04 0.37*** 2 −4.03*** −6.38***
(0.05) (0.09) (1.80) (1.01)
Leader × First-Party
Complements
3 −0.04 0.39*** 4 −0.18 −0.78***
(0.07) (0.12) (0.26) (0.17)
Third-party complementor
membership
3.09*** 0.05 −10.78 −25.86***
(0.56) (1.25) (8.92) (4.92)
First-party complements 0.00 0.03*** −0.00 −0.01***
(0.00) (0.01) (0.00) (0.00)
Leader −128.98*** −227.43***
(22.60) (5.02)
Platform installed base 0.03 0.49** 0.30* 0.17
(0.11) (0.21) (0.17) (0.15)
Rivals’ installed base −0.01 −0.05 −0.42*** −0.14
(0.02) (0.03) (0.12) (0.12)
Platform age 0.01 −0.02 0.05*** 0.05***
(0.02) (0.04) (0.01) (0.01)
Platform age2 −0.00*** −0.00 −0.00*** −0.00***
(0.00) (0.00) (0.00) (0.00)
N observations 129 129 336 336
F91,299*** 127*** 579*** 10,150***
*p < .10.
**p < .05.
***p < .01.
by guest on July 28, 2016jom.sagepub.comDownloaded from

Cennamo / Next-Generation Platform Value 19
affects variety in the early stage and negatively affects quality in the growth stage. However,
surprisingly, for generation leaders, in the early stage, greater complementor participation
will positively and significantly affect the quality of complements but not their variety. I
interpret this as attributable to potential time compression diseconomies (Dierickx & Cool,
1989) at the early stages associated with the learning required by developers to cope with
the new technology, which constrain their innovation rate. Individual developers might thus
focus their innovation effort on few but high-quality complements at this early stage relative
to the growth stage, by which time developers have become familiar with the next-genera-
tion technology. As for the growth stage, higher levels of third-party complementor mem-
bership negatively affect both the variety and the quality of complements, in line with my
theoretical predictions.
Robustness Analysis: Estimating a Dynamic Model of Platform Value
Because of network effects, platform value might be likely to show strong path depen-
dency from the value of past periods. Controlling for past realizations of the dependent vari-
able would require estimating a dynamic panel model. Fixed-effects estimation is not valid
in this case (Nickell, 1981), since the errors due to the individual effects would be correlated
with the lagged dependent variable, which, in its first-differenced form, would in fact become
endogenous. I thus estimate a dynamic model via a system GMM model (i.e., generalized
method of moments; Roodman, 2006). First proposed by Arellano and Bover (1995) and
fully developed by Blundell and Bond (1998), this GMM estimator efficiently instruments
the differenced variables that are not strictly exogenous with all their available lags in levels
and also allows the use of fixed effects and additional instruments (i.e., the ones I used in the
fixed-effects estimation).
Table 6 reports the results from such an estimation; Models 6.1 and 6.4 show the param-
eters’ estimate for platform value at the early and growth stages, while Models 6.2, 6.3, 6.5,
and 6.6 document the variety and quality components. Note that, apart from variety at the
early stage, the lagged dependent variable in all models significantly affects the current peri-
od’s realization. Also, the main negative effect of leader is not significant in the early stage;
this confirms that leaders, in fact, might face greater challenges during the takeoff of the
market. As for the other results, they are very much in line with those of the fixed-effects
estimation. They confirm that the positive impact of third-party complementor membership
and first-party complements on the value of the pioneering next-generation platforms is
mostly through the quality component of platform value.
Growth Challenges and Platform Value: System Saturation?
In general, these findings support my theory that for next-generation platform leaders,
strategies that leverage network effects to build value entail trade-offs and have asymmetric
impact across the different stages of the technology-market evolution. Yet they do not fully
reveal the underlying mechanism I suggest in the theory section: that next-generation lead-
ers may reach a level of competition within the platform system that prematurely saturates
the market for complements, stifling the growth of platform value. To visually appreciate
this effect, compare Figures 2 and 3, which report, respectively, median-spline plots for the
by guest on July 28, 2016jom.sagepub.comDownloaded from

20
Table 6
Determinants of Platform Value (Dynamic Model–Generalized Method of Moments)
Early stage Growth stage
Platform value Variety Quality Platform value Variety Quality
Variable Hypothesis Model 6.1 Model 6.2 Model 6.3 Hypothesis Model 6.4 Model 6.5 Model 6.6
Lagt – 1 (dependent variable) 0.58*** 0.12 0.72*** 0.95*** 0.92*** 0.93***
(0.22) (0.20) (0.27) (0.05) (0.04) (0.09)
Leader × Third-Party Complementor
Membership
1 0.29 −0.07 0.12** 2 −25.03*** −1.99*** −1.89***
(0.29) (0.06) (0.05) (5.65) (0.72) (0.62)
Leader × First-Party Complements 3 0.60* 0.09 0.60*** 4 −3.01*** 0.02 −0.28***
(0.35) (0.11) (0.14) (0.86) (0.11) (0.10)
Third-party complementor membership 1.19 4.02*** 1.48** −106.27*** −2.94 −5.97**
(3.14) (0.82) (0.72) (28.36) (3.59) (3.06)
First-party complements 0.06** −0.01 0.01** −0.06*** −0.01** −0.00*
(0.03) (0.01) (0.01) (0.02) (0.00) (0.00)
Leader −11.76 −5.01 −9.44 −91.37*** −5.37 −5.93**
(133.41) (37.07) (31.64) (27.47) (3.56) (3.03)
Platform installed base 1.86*** 0.53*** 0.38*** 3.48*** 0.47*** 0.45***
(0.60) (0.17) (0.14) (0.51) (0.07) (0.06)
Rivals’ installed base −0.41*** −0.04 −0.11*** 0.42 −0.12** 0.10**
(0.13) (0.03) (0.03) (0.39) (0.05) (0.05)
Platform age 0.02 −0.01 −0.00 0.14*** 0.02*** 0.00
(0.08) (0.03) (0.02) (0.03) (0.00) (0.00)
Platform age2 0.00 −0.00 −0.00 −0.00*** −0.00*** −0.00***
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
N observations 122 122 122 334 334 334
F513*** 246*** 327*** 28,847*** 11,137*** 92,394***
ar1 −2.59*** −2.27** −1.96** −3.01*** −2.44*** −4.18***
*p < .10.
**p < .05.
***p < .01.
by guest on July 28, 2016jom.sagepub.comDownloaded from

Cennamo / Next-Generation Platform Value 21
platform value of leaders versus followers and the number of first-party complements and
third-party complementors of leaders versus followers. Figure 3 shows that leaders of next-
generation platforms have a greater number of first-party complements and third-party
complementors than followers in the early stage, and for most of the growth stage too.
Figure 2
Platform Value of Next-Generation Leaders and Followers
Figure 3
Early Platform System Crowding
by guest on July 28, 2016jom.sagepub.comDownloaded from

22 Journal of Management / Month XXXX
Followers show not only lower levels but also a more gradual increase in the early stage and
the beginning of the growth stage (note the convex shape of both curves in Figure 3), com-
pared to the steeper and linear shape of the generation leaders’ curve.
This suggests that there is a systemic difference between next-generation platform leaders
and followers in the way they approach and address mobilization and growth challenges,
which may reflect the different intensity with which they experience these effects. As the
theory advanced here implies, next-generation platform leaders pursue mobilization strate-
gies more intensively early in the new-technology life cycle—again, probably because they
face tougher challenges. This push is associated with a sharp increase in value in the early
stage for platform leaders, as Figure 2 shows. However, in the growth stage, this effect
decreases, with platform value becoming completely flat. In sharp contrast, followers show
a sheer increase in platform value as the growth stage matures. Followers not only catch up
with generation leaders in terms of platform value but also move well ahead. This affects the
competitive standing of platform leaders, eventually muting the market-share advantage they
gained in the early stage.
Discussion
In platform markets, building value for next-generation platform technologies is more
complex than simply exploiting network effects. Leading next-generation platforms face
strong challenges not just in mobilizing complementary innovation but even more in increas-
ing value when the market takes off. My results indicate that increasing in-house comple-
ments and third-party complementor membership can help overcome inertia and mobilize
complementary innovation to generate value for users in the early stage, particularly in terms
of complement quality. However, continuing to use these strategies at later stages may not
help the platform outstrip followers; it will, in fact, lessen both variety and quality of comple-
ments. What implications do these findings have for the theory of platform competition and
technological discontinuity?
A key implication relates to the study’s main insight: that the quality of complements is an
important factor shaping platform competitive dynamics. Even if they have few users and
complements at the early stage, next-generation platform leaders can penetrate the market
through complement quality. But by aggressively trying to secure complements early on,
they can crowd the market for complements and saturate it earlier than would have happened
over the natural evolution of the new technology, with detrimental effects on the quality and
variety of complements. This represents a paradox in the light of mainstream theoretical
logic: In platform markets, where increasing returns from indirect network effects influence
platform adoption and competition, strategies for exploiting such network effects may exhibit
diminishing returns. This paradox can help explain why we observe changes in platform
leadership over time and also the coexistence of multiple platforms. Shifting the unit of
analysis from the stock of complements as a whole to the discrete components of platform
complement benefits, that is, complement quality and variety, can help explain these occur-
rences and improve our understanding of platform competitive dynamics beyond the linear
prediction that the winner will take all.
A second, related implication of the theoretical logic advanced here is that managing this
paradox would more critically affect a platform’s competitive standing than simply
by guest on July 28, 2016jom.sagepub.comDownloaded from

Cennamo / Next-Generation Platform Value 23
maximizing the numbers of users and complementors. Ultimately, the platform that can
increase its value more over time will win; and it will not necessarily be the one with the
largest network of complementors or the greatest number of complements. In fact, the nega-
tive effect of third-party complementor membership on platform value at the growth stage is
much greater than the positive effect at the early stage. Despite mainstream theory implying
that network effects, once taking hold, should effectively function as a coordination mecha-
nism, coordination problems with complementors for value cocreation still can be present,
and even more acute, at later stages of a platform market. And it is at this growth stage that
the battle for dominance gets resolved. Gaining some scale early on may give leaders a head
start, but this early advantage may quickly fade.
Mobilization and growth challenges involve two temporally interdependent tensions: one
between complement variety and quality, and one between in-house and third-party comple-
ments. Wareham, Fox, and Cano Giner (2014) speak of the need for the ecosystem to evolve
constantly—which requires a variety of innovations by external members and, thus, greater
membership—but without becoming so varied and fragmented that it loses its value to users
and complementors. Claussen and colleagues (2015) investigate the dynamic trade-offs asso-
ciated with the impact that different levels of technological sophistication have on comple-
mentors during the new technology’s life cycle. Cennamo and Santaló (2013: 1346) discuss
the importance of balancing a platform’s incentives to enlarge its complementor network and
complement portfolio against the incentives for complementors to provide novel, valuable
complements.
This may require a dynamic “orchestration” (Nambisan & Sawhney, 2011) of the plat-
form complementors’ ecosystem across the different stages. Finding the right balance
between in-house complements and externally developed ones is not an easy task. Hagiu
and Spulber (2013) suggest that the platform leader should rely on in-house complements
in the early stage, when expectations about the platform are less likely to be favorable, and
on external complementors at later stages. But recent research on plural sourcing (e.g.,
Parmigiani & Mitchel, 2009; Puranam et al., 2013) suggests that the question is how much
to use each mode at each stage. Federating complementors around the platform in the early
stage is still necessary to signal support from the developer community and, thus, generate
favorable expectations. Reputation and expectations have an important effect on users’
decision to switch from the last generation (e.g., Sheremata, 2004; Suarez, 2004). But this
study’s findings show that external sources may make little contribution to complement
variety early on, perhaps because it takes time for complementors to learn the platform’s
technological environment. Relying more heavily on internal development at this stage
would help platform leaders to compensate for this delay, and the technical knowledge cre-
ated would spill over to complementors. At later stages, though platforms may want to
keep producing complements in-house to differentiate their system from rivals (Cennamo
& Santaló, 2013), they should restrict this production to avoid discouraging external com-
plementors from producing high-quality complements. Findings suggest that platforms
should also be cautious about increasing third-party complementor membership at the
growth stage.
This, along with the study’s other insights, opens up a set of questions for further
research. For instance, how much of the performance variation among competing plat-
forms is attributable to the characteristics of complements and how much to the design
by guest on July 28, 2016jom.sagepub.comDownloaded from

24 Journal of Management / Month XXXX
and structuring of the complementor ecosystem—that is, the organization of innovation
in the ecosystem (e.g., Gawer & Henderson, 2007; Wareham et al., 2014)? Strategy schol-
ars emphasize the value of resources and innovation for competitive advantage and
accordingly focus on the strategies to obtain and deploy such resources, whereas organi-
zation scholars have highlighted the potential tensions and contradictions that the coordi-
nation mechanisms and structure of the ecosystem may impose on its players at different
stages (e.g., Garud et al., 2002; Wareham et al., 2014). The evidence presented in this
study seems to bridge these perspectives: Complements are indeed the core component of
a platform’s value, but the strategies used to obtain them in the wake of technological
discontinuity may accidentally lead to a system structure that proves ineffective as the
market evolves. Platform value and ecosystem structure coevolve via complex feedback
effects that are yet to be fully understood. Here, I do not directly assess the structure of
the ecosystem. Extending the analysis in this direction, I believe, could enrich our under-
standing not only of platform competitive dynamics by uncovering other sources of the
heterogeneity in competing systems but more broadly of value creation-capture dynamics
within and across competing ecosystems.
Another interesting topic to explore in future research relates to the study’s insight that
pioneers may be locked into suboptimal patterns of performance. Is this an unavoidable
course in platform contexts faced with technological discontinuity? Does this apply more
generally to other contexts? Which factors will be mainly responsible for this course?
Studies on technological discontinuity have focused mainly on the early stage of the next-
generation technology’s market, restricting the analysis to incumbent-challenger competi-
tive dynamics. Studies on first-mover advantage focus mainly on firms’ characteristics and
environment conditions at the time of entry, with only a few looking at how the evolution
of the competitive environment affects the duration of first-mover advantage (e.g., Franco,
Sarkar, Agarwal, & Echambadi, 2009; Suarez & Lanzolla, 2007). Suarez and Lanzolla
(2007) maintain that a first-mover advantage would be difficult to obtain under technologi-
cal discontinuities and sustained market growth because pioneers would find it hard to
foreclose market opportunities for followers—if the market were still growing, later
entrants would still find resources and room to compete successfully with first movers.
Gomez et al. (2016) provide empirical evidence of such effects, finding that, in the
European mobile communications sector, technological discontinuity and high market
growth impair first movers’ performance.
Others have pointed to a “vintage effect” (Bohlmann, Golder, & Mitra, 2002) in high-
velocity technology environments—the pioneer’s technology becomes obsolete more
quickly than that of followers, so that users may suddenly shift toward other new technolo-
gies. Platform systems may be subject to a vintage effect of a different kind, relating not so
much to the pioneer’s own technology as to its complementors’ ecosystem: earlier satura-
tion of the market for complements. The large negative effect of third-party complementor
membership on pioneers’ platform value in the growth stage underscores this point. More
generally, this poses the question of whether next-generation leaders may be ill positioned
to secure an advantage when the value of the technology is interconnected to complemen-
tary goods and services. Such yoking might limit first movers’ opportunities to preempt
critical resources, build switching costs, or quickly progress along the learning curve and
improve the product offerings’ performance. For instance, Adner and Kapoor (2010) find
by guest on July 28, 2016jom.sagepub.comDownloaded from

Cennamo / Next-Generation Platform Value 25
that in the semiconductor lithography equipment industry, innovation challenges faced by
complementors impair the potential benefits for technology leaders; the more the chal-
lenges for complementors, the stronger this (negative) effect. While they emphasize the
magnitude of challenges faced by complementors, findings from my study indicate that
equally important is the type (mobilization vs. growth) of challenge at different stages of
the life cycle and how technology leaders cope with it. Ultimately, how firms reposition
over time in the factor market of these complementary resources can critically determine
leaders’ ability to obtain first-mover advantage and the magnitude and duration of such
advantage.
This also emphasizes the importance of the strategies and capabilities that next-generation
leaders deploy in relation to followers—not just at the time of entry (e.g., Fuentelsaz, Gomez,
& Polo, 2002) but also, and perhaps more importantly, after entry (e.g., Franco et al., 2009;
Fuentelsaz et al., 2015). Recent work has called for factoring the evolution of the competitive
environment into the pioneering-performance relationship (e.g., Chintakananda & McIntyre,
2014; Gomez et al., 2016; Suarez & Lanzolla, 2007). Insights from my study also highlight
the need to look at factors at the market and technology level not as completely exogenous
but as products of firms’ competitive interactions; these interactions ultimately shape the
overall evolution of the technology and its associated market when the technology requires
complements to generate value for consumers.
Complements are valuable in contexts other than platform markets. Increasingly, advan-
tage-conferring resources such as complements reside outside the firm, in a web of special-
ized firms that constitute innovation networks or communities (Adner & Kapoor, 2010;
Garud et al., 2002; Gulati et al., 2012). They represent new forms of organizing innovation
that require rethinking interfirm relationships, as well as means of value creation and value
capture. We still know little about these issues, which represent an interesting and relevant
area for future research.
Managerial Implications
My study offers insights for managers aiming to exploit technological discontinuity to
dethrone incumbent platforms. While managers often emphasize superior performance and
other characteristics of the new technology itself as key reasons for switching to it, what they
do to build a competitive system around the technology may be more critical. Technological
prowess alone is not enough. Indeed, since a new, superior technology generally requires
new learning from complementors, it may increase mobilization challenges. When these
challenges are properly addressed, though, technological superiority may indeed be a source
of advantage to the extent that it facilitates the production of higher-quality complements,
making the whole system more valuable than competing systems. But these benefits are not
straightforward, as my findings show.
While platform providers strain to be the first to market and rapidly scale up their net-
works, managers should take care not to assume that this will inevitably increase platform
value and provide a shield against competition. Instead, they should factor in the inherent
trade-offs of such a strategy and address the tensions I document above. Their best course
may be to invest in collateral technologies, such as development tools and programming
interfaces, and in marketing activities (e.g., cobranding, product bundling) that can support
by guest on July 28, 2016jom.sagepub.comDownloaded from

26 Journal of Management / Month XXXX
complementors’ innovation, enhancing opportunities for value cocreation while reducing
tensions stemming from value-capture concerns.
Study Boundaries and Limitations
This study addressed potential endogeneity problems and biases arising from unobserv-
able platform characteristics. However, I can control only for differences that are time invari-
ant. I have an acceptable number of observations, but they are of a restricted sample of
platforms. Similarly, I could observe only three generations—but those generations do show
different companies, along with changes in platform leadership and new market entry by
outsiders. These limitations granted, the database is rich enough to allow me to explore varia-
tion among platforms in the key explanatory variables.
My measure of quality captures the essence of the theoretical construct but has the impor-
tant limitation that one can assess it only ex post. I cannot observe or determine ex ante what
product attributes make a title “high quality” in the eyes of the customer—thus, I assume that
a better selling product is a “better” product. An ideal measure would be a vector of product-
specific characteristics, such as graphics, speed, or gameplay, along which to rank each com-
plement objectively. Unfortunately, some of these characteristics (e.g., graphics, speed) also
depend on the technical characteristics of the console itself (Zhu & Iansiti, 2012). For “soft”
characteristics, such as gameplay, it would be hard to assess their value ex ante, as consumers
are the ultimate evaluators.
The empirical evidence may be industry specific. If gamers like novelty, product variety
may be more critical for platform success in this industry than in sectors where consumers
stick to a few specific applications over time. In other industries (e.g., personal computers),
users’ preference (or practical need) for continuity between different technology genera-
tions may limit the real options of next-generation leaders and the odds of success. Also, as
my findings confirm, hit games are the core component of platform value. But in other
contexts, complements’ quality may be less salient in comparison with variety. For instance,
for service-based platforms such as eBay or Groupon, variety of content may be the main
determinant of platform value, as the key role of such platforms is to “grow matched mar-
kets” (Parker & Van Alstyne, 2005: 1496). However, I speculate, quality would also play a
role by influencing customers’ expectations about the value of transactions they could con-
duct through the platform. In this context, quality can be conceived as an attribute of mer-
chants’ product offerings or as the overall quality of the transaction (correspondence
between product description and customers’ expectations, transaction terms, and the like).
Roger and Vasconcelos (2014) submit that the average quality of transactions at the system
level will contribute to the overall reputation of the platform, which in turn will make one
platform more valuable than others, and ultimately affect members’ decision to participate.
But it remains to be understood how far the quality and variety of a platform’s complements
affect its value and platform competition in these contexts. More generally, I believe that the
key insights of my study extend to other contexts, such as innovation ecosystems, in which
next-generation technology leaders need to address challenges of complementary innova-
tion. I hope insights from my analysis will stimulate additional work in this area to shed new
light on dynamic factors at industry, technology, and firm level that affect value creation and
competition.
by guest on July 28, 2016jom.sagepub.comDownloaded from

27
Appendix
Table A1
Platform Value Comparisons
Scenario 1 Scenario 2 Scenario 3 Scenario 4
Console A B A B A B A B
Number of titles 50 50 50 50 30 50 30 50
Game type Split in 2
genres
Even split across
all genres
Split in 2
genres
Even split across
all genres
Even split across
all genres
Even split across
all genres
Random
split
Random
split
Diversity 0.5 0.89 0.5 0.89 0.89 0.89 0.72 0.79
Number of hits 10 30 30 10 20 10 20 10
Variety 1.96 3.48 1.96 3.48 3.03 3.48 2.45 3.09
Quality 2.3 3.4 3.4 2.3 3.00 2.3 3.00 2.3
Platform value 4.5 11.83 6.66 8.00 9.07 8.00 7.34 7.11
by guest on July 28, 2016jom.sagepub.comDownloaded from

28 Journal of Management / Month XXXX
Platform Value and Performance: System of Equations Analysis
As described in the Method section, I validated the platform value construct by estimating a
system of equations where I model platform market penetration as a function of platform value
(platform adoption equation) and, in turn, platform value as a function of the platform installed
base (platform value equation). This is similar to the analysis performed in Corts and Lederman
(2009) to test the presence and impact of indirect network effects, the difference being in that I
use my platform value measures instead of the simple number of games they use in their analy-
sis. For the construct to be valid, it must be positively influenced by a platform’s installed base
(in the platform value equation) and affect positively and significantly a platform’s market
penetration (in the platform adoption equation). My interest being in technological discontinu-
ity, I also include the next-generation leadership variable (leader) to assess its main effect on
platform value and platform market penetration. The system of equations to estimate is
Platform valueplatform installed baserivalsins = +
jt 1jt2
ββ
ttalled base
platform age platform age lea
+ + +
jt
3jt4 jt 5
ββ β2dderjt tjt
+ T+βξ
6
Market penetrationplatform valueplatform price = +
jt 1jt2
ββ
jjt
3j
t
4jt5 jt
+
+ +
β
ββ
market share
platform age platform age 2 ++ + T
6jt7t +jt
ββηleader
where market penetration is the share of the total potential market for videogame consoles,
measured here like in Corts and Lederman as the number of households with a television
(collected from the U.S. Census Bureau Web site) minus the combined installed base of all
active platforms (i.e., cumulative past unit sales of consoles). Below I report the estimation’s
results.
Instrumental Variables Identification Procedure
I followed Corts and Lederman (2009) and used the monthly exchange rates between the
U.S. dollar and the currency of the country where the console was manufactured as an instru-
ment for platform price. The ratios offer a good proxy of console production cost and should
affect U.S. retail prices but not correlate with the error term. As an instrument for third-party
complementor membership, I used the average age of game titles active in a given month on
a console; it represents an indicator of the residual life of game titles and should affect pro-
ducers’ game introduction decisions. Also, similar to Boudreau (2012), I used the average
hourly wage of software developers (collected from the Bureau of Labor Statistics) as an
additional instrument for third-party complementor membership. Third-party complementor
membership is an endogenous variable; thus, so is the variable resulting from its interaction
with leader. Following standard econometric practice, I used the interaction between the
instruments for third-party complementor membership and the leader dummy as instrument
(Baum, Schaffer, & Stillman, 2007). In addition, I used different variables capturing compe-
tition with other platforms and variation among platform characteristics. Therefore, I included
the sum of each platform’s technical characteristics (processor speed, memory, and processor
word length in bits) divided by the sum for competing platforms, and the total number of
competing platforms belonging to the same technology generation. I also used dummies for
each platform to capture unobserved characteristics and the number of platforms from the
by guest on July 28, 2016jom.sagepub.comDownloaded from

Cennamo / Next-Generation Platform Value 29
same platform-sponsoring firm. Because of the use of these platform fixed effects, variation
in the other platform-based variables comes from variation in the set of platforms in the
database at a given time (because of entry and exit, and change across generations). As an
instrument for the installed base variables, I used the maximum age of the platforms in the
market and the cumulative age of rival platforms belonging to the same generation.
According to the Sargan-Hansen test of the joint null hypothesis of instruments validity, that
is, the instruments not be correlated with the error term, the excluded instruments were cor-
rectly excluded from the estimated equation, and I found evidence of the validity of the
instruments because Hansen’s J statistic does not reject the null prediction. As a result of
space limitations, I do not report the results for first-stage independent variable estimations
in this article.
References
Adner, R., & Kapoor, R. 2010. Value creation in innovation ecosystems: How the structure of technological inter-
dependence affects firm performance in new technology generations. Strategic Management Journal, 31: 306-
333.
Akerlof, G. A. 1970. The market for “lemons”: Quality uncertainty and the market mechanism. Quarterly Journal
of Economics, 84: 488-500.
Alexy, O., George, G., & Salter, A. J. 2013. Cui bono? The selective revealing of knowledge and its implications for
innovative activity. Academy of Management Review, 38: 270-291.
Anderson, P., & Tushman, M. 1990. Technological discontinuity and dominant designs: A cyclical model of tech-
nological change. Administrative Science Quarterly, 35: 604-633.
Table A2
Platform Value and Platform Market Penetration (System of Equations Estimation)
Platform Value Market Penetration
Variety Quality Variety × Quality Variety Quality Variety × Quality
Platform value 0.01*** 3.07*** 0.33***
(0.00) (0.78) (0.10)
Leader −239.24*** −3.28*** −34.69*** −26.96*** −23.99*** −24.48***
(34.82) (0.80) (6.82) (3.01) (3.01) (2.51)
Platform installed base 311.67*** 0.88*** 6.94***
(14.29) (0.03) (0.29)
Rivals’ installed base −21.45*** −0.06** −0.58***
(11.07) (0.03) (0.23)
Platform price −0.00* −0.01* −0.01
(0.00) (0.00) (0.00)
Market share 0.28** 0.53*** 0.59***
(0.13) (0.12) (0.12)
Platform age −9.96*** −0.01 −0.03 0.11** 0.14** 0.14***
(3.77) (0.08) (0.08) (0.05) (0.06) (0.06)
Platform age2 0.34*** −0.00 0.00*** −0.00*** −0.00*** 0.00***
(0.02) (0.00) (0.00) (0.00) (0.00) (0.00)
N observations 465 465 465 465 465 465
χ217,276*** 223,229*** 103,444*** 12,140*** 11,826*** 17,152***
*p < .10.
**p < .05.
***p < .01.
by guest on July 28, 2016jom.sagepub.comDownloaded from

30 Journal of Management / Month XXXX
Ansari, S., & Garud, R. 2009. Inter-generational transitions in socio-technical systems: The case of mobile com-
munications. Research Policy, 38: 382-392.
Arellano, M., & Bover, O. 1995. Another look at the instrumental variable estimation of error-components models.
Journal of Econometrics, 68: 29-51.
Armstrong, M. 2006. Competition in two-sided markets. RAND Journal of Economics, 37: 668-691.
Arthur, W. B. 1989. Competing technologies, increasing returns, and lock-in by historical events. Economic
Journal, 99(394): 116-131.
Augereau, A., Greenstein, S., & Rysman, M. 2006. Coordination versus differentiation in a standards war: The
adoption of 56K modems. RAND Journal of Economics, 37: 887-909.
Baldwin, C., & Woodard, J. J. 2009. The architecture of platforms: A unified view. In A. Gawer (Ed.), Platforms,
markets, and innovation: 19-44. Northampton, MA: Edward Elgar.
Baum, C. F., Schaffer, M. E., & Stillman, S. 2007. Enhanced routines for instrumental variables/generalized method
of moments estimation and testing. Stata Journal, 7: 465-506.
Binken, J. L. G., & Stremersch, S. 2009. The effect of superstar software on hardware sales in system markets.
Journal of Marketing, 73(2): 88-104.
Blundell, R., & Bond, S. 1998. Initial conditions and moment restrictions in dynamic panel data models. Journal of
Econometrics, 87: 115-143.
Bohlmann, J. D., Golder, P. N., & Mitra, D. 2002. Deconstructing the pioneer’s advantage: Examining vintage
effects and consumer valuations of quality and variety. Management Science, 48: 1175-1195.
Boudreau, K. J. 2012. Let a thousand flowers bloom? An early look at large numbers of software app developers and
patterns of innovation. Organization Science, 23: 1409-1427.
Brusoni, S., Prencipe, A., & Pavitt, K. 2001. Knowledge specialization, organizational coupling, and the boundaries
of firms: Why do firms know more than they make? Administrative Science Quarterly, 46: 597-621.
Caillaud, B., & Jullien, B. 2003. Chicken-and-egg: Competition among intermediation service providers. RAND
Journal of Economics, 34: 309-328.
Cassiman, B., & Veugelers, R. 2006. In search of complementarity in innovation strategy: Internal R&D and exter-
nal knowledge acquisition. Management Science, 52: 68-82.
Ceccagnoli, M., Forman, C., Huang, P., & Wu, D. J. 2012. Co-creation of value in a platform ecosystem: The case
of enterprise software. MIS Quarterly, 36: 263-290.
Cennamo, C., & Santaló, J. 2013. Platform competition: Strategic trade-offs in platform markets. Strategic
Management Journal, 34: 1331-1350.
Cennamo, C., & Santaló, J. 2015. How to avoid platform traps. MIT Sloan Management Review, 57(1): 11-15.
Chintakananda, A., & McIntyre, D. P. 2014. Market entry in the presence of network effects: A real option perspec-
tive. Journal of Management, 40: 1535-1557.
Christensen, C. M., & Bower, J. L. 1996. Customer power, strategic investment, and the failure of leading firms.
Strategic Management Journal, 17: 197-218.
Christensen, C. M., Suarez, F. F., & Utterback, J. M. 1998. Strategies for survival in fast-changing industries.
Management Science, 44: S207-S220.
Claussen, J., Essling, C., & Kretschmer, T. 2015. Setting technology levels in complementary goods markets.
Research Policy, 44: 328-339.
Clements, M. T., & Ohashi, H. 2005. Indirect network effects and the product cycle: Video games in the U.S.,
1994–2002. The Journal of Industrial Economics, 53: 515-542.
Corts, K. S., & Lederman, M. 2009. Software exclusivity and the scope of indirect network effects in the U.S. home
video game market. International Journal of Industrial Organization, 27: 121-136.
Cottrell, T., & Nault, B. R. 2004. Product variety and firm survival in the microcomputer software industry. Strategic
Management Journal, 25: 1005-1025.
Dierickx, I., & Cool, K. 1989. Asset stock accumulation and sustainability of competitive advantage. Management
Science, 35: 1504-1511.
Franco, A. M., Sarkar, M. B., Agarwal, R., & Echambadi, R. 2009. Swift and smart: The moderating effects of tech-
nological capabilities on the market pioneering-firm survival relationship. Management Science, 55: 1842-1860.
Fuentelsaz, L. E., Garrido, J., & Maicas, P. 2015. A strategic approach to network value in network industries.
Journal of Management, 41: 864-892.
Fuentelsaz, L., Gomez, J., & Polo, Y. 2002. Followers’ entry timing: Evidence from the Spanish banking sector after
deregulation. Strategic Management Journal, 23: 245-264.
by guest on July 28, 2016jom.sagepub.comDownloaded from

Cennamo / Next-Generation Platform Value 31
Garud, R., Jain, S., & Kumaraswamy, A. 2002. Institutional entrepreneurship in the sponsorship of common tech-
nological standards: The case of Sun Microsystems and Java. Academy of Management Journal, 42: 196-214.
Garvin, D. A. 1984. What does “product quality” really mean? MIT Sloan Management Review, 26(1): 25-43.
Gawer, A. 2014. Bridging differing perspectives on technological platforms: Toward an integrative framework.
Research Policy, 43: 1239-1249.
Gawer, A., & Henderson, R. M. 2007. Platform owner entry and innovation in complementary markets: Evidence
from Intel. Journal of Economics & Management Strategy, 16: 1-34.
Gil, R., & Warzynski, F. 2010. Vertical integration, exclusivity and game sales performance in the US video game
industry. Working paper no. 10-06, NET Institute, Daytona Beach, FL. http://ssrn.com/abstract=1694311.
Accessed June 22, 2016.
Gomez, J., Lanzolla, G., & Maicas, J. P. 2016. The role of industry dynamics in the persistence of first mover advan-
tages. Long Range Planning, 49: 265-281.
Greene, W. H. 2003. Econometric analysis (5th ed.). Upper Saddle River, NJ: Prentice Hall.
Gulati, R., Puranam, P., & Tushman, P. 2012. Meta-organization design: Rethinking design in interorganizational
and community contexts. Strategic Management Journal, 33: 571-586.
Gupta, S., Jain, D. C., & Sawhney, M. S. 1999. Modeling the evolution of markets with indirect network externali-
ties: An application to digital television. Marketing Science, 18: 396-416.
Hagiu, A., & Spulber, D. 2013. First-party content and coordination in two-sided markets. Management Science,
59: 933-949.
Katz, M. L., & Shapiro, C. 1992. Product introduction with network externalities. The Journal of Industrial
Economics, 40: 55-83.
Kunni, J. 1998. The games Sony plays. BusinessWeek, June 15: 128.
Lawless, M. W., & Anderson, P. C. 1996. Generational technological change: Effects of innovation and local rivalry
on performance. Academy of Management Journal, 39: 1185-1217.
Lieberman, M. B., & Montgomery, D. B. 1988. First-mover advantages. Strategic Management Journal, 9: 41-58.
Lieberman, M. B., & Montgomery, D. B. 1998. First-mover (dis)advantages: Retrospective and link with the
resource-based view. Strategic Management Journal, 19: 1111-1125.
McIntyre, D. P., & Subramaniam, M. 2009. Strategy in network industries: A review and research agenda. Journal
of Management, 35: 1494-1517.
Nambisan, S., & Sawhney, M. 2011. Orchestration process in network-centric innovation: Evidence from the field.
Academy of Management Perspectives, 25: 40-57.
Nickell, S. 1981. Biases in dynamic models with fixed effects. Econometrica, 49: 1417-1426.
Panico, C., & Cennamo, C. 2015. What drives a platform’s strategy? Usage, membership, and competition effects.
Academy of Management Proceedings, 2015(1): 15942.
Parker, G. G., & Van Alstyne, M. W. 2005. Two-sided network effects: A theory of information product design.
Management Science, 51: 1494-1504.
Parker, G. G., & Van Alstyne, M. W. 2008. Innovation, openness and platform control. Working paper, Tulane
University, New Orleans.
Parmigiani, A. 2007. Why do firms both make and buy? An investigation of concurrent sourcing. Strategic
Management Journal, 28: 285-311.
Parmigiani, A., & Mitchell, W. 2009. Complementarity, capabilities, and the boundary of the firm: The impact
of within-firm and interfirm expertise on concurrent sourcing of complementary components. Strategic
Management Journal, 30: 1065-1091.
Puranam, P., Gulati, R., & Bhattacharya, S. 2013. How much to make and how much to buy? Explaining optimal
plural sourcing strategies. Strategic Management Journal, 34: 1145-1161.
Randall, T., & Ulrich, K. 2001. Product variety, supply chain structure, and firm performance: Analysis of the U.S.
bicycle industry. Management Science, 47: 1588-1604.
Rochet, J.-C., & Tirole, J. 2006. Two-sided markets: A progress report. RAND Journal of Economics, 37: 645-667.
Roger, G., & Vasconcelos, L. 2014. Platform pricing structure and moral hazard. Journal of Economics &
Management Strategy, 23: 527-547.
Roodman, D. 2006. How to do xtabond2: An introduction to “difference” and “system” GMM in Stata. Working
paper no. 103, Center for Global Development, Washington, DC.
Rothaermel, F. T., & Hess, A. 2007. Building dynamic capabilities: Innovation driven by individual-, firm-, and
network-level effects. Organization Science, 18: 898-921.
by guest on July 28, 2016jom.sagepub.comDownloaded from

32 Journal of Management / Month XXXX
Schilling, M. A. 2002. Technology success and failure in winner-take-all markets: The impact of learning orienta-
tion, timing, and network externalities. Academy of Management Journal, 45: 387-398.
Schilling, M. A. 2003. Technological leapfrogging: Lessons from the U.S. video game console industry. California
Management Review, 45(3): 6-32.
Sheremata, W. A. 2004. Competing through innovation in network markets: Strategies for challengers. Academy of
Management Review, 29: 359-377.
Smith, K., & Sasaki, M. S. 1979. Decreasing multicollinearity: A method for models with multiplicative functions.
Sociological Methods and Research, 8: 35-56.
Suarez, F. F. 2004. Battles for technological dominance: An integrative framework. Research Policy, 33: 271-286.
Suarez, F. F., & Lanzolla, G. 2007. The role of environmental dynamics in building a first mover advantage theory.
Academy of Management Review, 32: 377-392.
Terry, L. 1998. Looking for a Sonic boom. Brandweek, 39(9): 26-30.
Tushman, M. L., & Anderson, P. 1986. Technological discontinuities and organizational environments.
Administrative Science Quarterly, 31: 439-465.
Venkatraman, N., & Lee, C. H. 2004. Preferential linkage and network evolution: A conceptual model and empirical
test in the U.S. video game sector. Academy of Management Journal, 47: 876-892.
Wade, J. 1995. Dynamics of organizational communities and technological bandwagons: An empirical investigation
of community evolution in the microprocessor market. Strategic Management Journal, 16: 111-133.
Wareham, J., Fox, P. B., & Cano Giner, J. L. 2014. Technology ecosystem governance. Organization Science, 25:
1195-1215.
Xu, X., Venkatesh, V., Tam, K. Y., & Hong, S.-J. 2010. Model of migration and use of platforms: Role of hierarchy,
current generation, and complementarities in consumer settings. Management Science, 56: 1304-1323.
Zhu, F., & Iansiti, M. 2012. Entry into platform-based markets. Strategic Management Journal, 33: 88-106.
Zhu, F., & Liu, Q. 2015. Competing with complementors: An empirical look at Amazon.com. Working paper no.
15-044, Harvard Business School, Cambridge, MA.
by guest on July 28, 2016jom.sagepub.comDownloaded from