Do Network Effects Raise Consumer Interest? The Mediating Role of
Consumer Interest in Explaining Platform Success
Name: Alco van Spengen Student number: 2015285
University of Groningen Faculty of Economics and Business MSc Strategic Innovation Management
Supervisor: Thijs Broekhuizen Co-assessor: Isabel Estrada
Master Thesis
2
Abstract
This study investigates the influence of network effects on consumer interest and platform performance in two-sided markets. Prior literature on network effects has argued that platform performance largely depends on the size of the installed base (direct network effects) and the availability and quality of complementary products (indirect network effects). In the context of the 8th generation of video game consoles, this study investigates whether direct network effects and
indirect network effects lead to better platform performance directly, or through consumer interest indirectly. The findings support that consumer interest acts as a mediator of the positive relationships between both direct and indirect network effects and platform performance. This study shows that both direct and indirect effects are important to raise consumer interest. Both quality and availability of complementary products are important to raise consumer interest in two-sided markets, as well as the installed base of users. The findings show that direct network effects are not as strong as indirect network effects. Although, direct network effects still seem to matter, platform providers should only increase their installed base up to a certain size before switching their attention to indirect network effects to optimize platform performance. Furthermore, quality seems to be more important than availability of complementary products; superstars in particular have strong positive indirect effects on platform performance, via raising consumer interest.
Keywords: two-sided markets, network effects, platforms, video game industry, consumer interest,
installed base, availability of complementary products, quality of complementary products
3
Table of Content
1. Introduction ... 4
2. Literature Review ... 5
2.1 Two-sided markets, Platforms & Network effects ... 5
2.2 Traditional pipelines versus platforms ... 6
2.3 Increasing returns to scale ... 6
2.4 Conceptual Framework: Determinants of Platform Performance ... 7
2.5 Consumer Interest ... 8
2.6 Network effects and platform performance ... 9
2.7 Direct Network Effects: Installed base ... 10
2.8 Indirect Network Effects ... 10
2.8.1 Availability of complementary products ... 11
2.8.2 Quality of complementary products ... 13
3. Methodology ... 14
3.1 Sample Statistics ... 14
3.2 Measures and Data Collection ... 15
3.2.1 Dependent variables ... 15
3.2.2 Independent variables ... 16
3.2.3 Control variables ... 17
3.3 Analysis ... 17
4. Results ... 18
4.1 Descriptive Statistics and Correlation Matrix ... 18
4.2 Direct effect of consumer interest on platform performance ... 19
4.3 Direct effects on consumer interest ... 19
4.4 Mediation effects ... 20
5. Discussion & Conclusion... 21
5.1 Discussion ... 21
5.2 Conclusion & Managerial Insights ... 22
5.3 Limitations and Future research ... 23
6. References ... 23
4
1. Introduction
Many markets in today’s economy are organized around platforms: products and services that bring together different groups of users in two-sided markets (Cennamo & Santalo, 2013; Eisenmann, Parker & Van Alstyne, 2006). Platforms are disrupting the traditional business landscape by displacing some of the world’s biggest firms and transforming traditional value chains, consumer behavior and the structure of major industries (Parker, Van Alstyne & Choudary, 2016). Companies like Facebook, PayPal, Apple, Amazon and Uber have revolutionized entire industries by harnessing a single phenomenon: the platform business model. Platforms apply technology to link producers and consumers in a multi-sided marketplace, unlocking hidden resources and creating new forms of value with the help of network effects (Parker, Van Alstyne & Choudary, 2016). Platforms provide infrastructure and rules that facilitate transactions and value creation activities (Cennamo & Santalo, 2013; Eisenmann, Parker & Van Alstyne, 2006; Evans, 2003; Gawer, 2010; Rochet & Tirole, 2006). Platforms that serve two-sided (or more generally multi-sided) markets are not a new phenomenon and have existed for years (Armstrong, 2006; Caillaud & Jullien, 2003; Hagiu, 2005, 2009; Rochet & Tirole, 2003, 2006). They can be found in many industries, such as banking, media and software. Some examples of two-sided markets are shopping malls linking shoppers and merchants; PC operating systems linking software developers with computer users; newspapers linking subscribers and advertisers; DVD’s linking consumers with studios; recruitment services linking job seekers with recruiters; online websites such as Amazon or eBay linking buyers and sellers.
The biggest change in this century is that information technology, and in particular the internet, has greatly reduced the need to own physical infrastructure and assets, which has increased the advent of platforms. IT makes building and scaling up platforms much simpler and cheaper, allows effortless participation that strengthens network effects and enhances the ability to capture, analyze and exchange huge amounts of data that increase the value of the platform to all parties involved (Van Alstyne, Parker & Choudary, 2016).
The hallmark of two-sided markets is the concept of network effects. Platforms with a large number of users are more valued by consumers, because of direct links with other consumers (direct network effects) or because they anticipate that platforms with more users (larger installed base of users) will also offer a wider number and variety of complementary products and services (indirect network effects) (Cennamo & Santalo, 2013; Evans, 2003; Rochet & Tirole, 2003, 2006). The value of the platform increases when the utility of the platform increases due to a greater number of adopters. As a result, platforms that are expected to be popular, have widely available components and will be more popular for that very reason (Katz & Shapiro, 1994). When successful, platforms stimulate a virtuous cycle that reinforces itself: more demand from one user group stimulates more demand from the other (Eisenmann, Parker & Van Alstyne, 2006; Hill, 1997; Schilling, 2002, 2003). For example, the more video games developers create for a video game console, the more players will be willing to buy the video game console. Meanwhile, the more players use the video game console, the more video game developers are willing to pay licensing fees to the providers of these video game platforms.
5
appear to have a direct positive effect on sales of the platform. These studies presume that the higher utility of the platform (instigated by the direct and indirect network effects) positively affects sales. However, literature has not yet examined the role of consumer interest in two-sided platform markets. Literature on consumer behavior suggests that consumer interest leads to an increase in sales (e.g. Klein, 1998; Kulkarni, Kannan & Moe, 2012; Shim et al., 2001). Consumer interest has never been studied before in a conceptual framework of network effects and their impact on platform performance in the context of two-sided markets. Therefore, this study attempts to fill that void and complement existing literature by exploring the phenomenon of consumer interest in relation to network effects and platform performance. The existing literature has failed to consider consumer interest as a mediating factor, hence the objective of this study is to propose consumer interest as a mediator between both direct and indirect network effects and platform performance in two-sided markets, while accounting for important control variables such as platform age and seasonality. This study collects data on the 8th generation of video game consoles to examine the
factors that affect consumer interest and platform performance in the video game console market. Managerially, this study provides some relevant insights to platform providers by guiding them in their decisions about which network effects to stimulate and how, so that they can optimize platform performance. The current study draws upon recent theoretical work to create a conceptual framework. Several hypotheses are developed about the direct network effects (installed base) and indirect network effects (complementary products) and tested in the video game industry.
The thesis is structured as follows. First the study begins by defining terms and specifying the theoretical background for the model and hypotheses. After discussing the methodology, the study presents the results. Subsequently, a discussion of the results and a conclusion follow. The thesis ends with research limitations and suggestions for future research.
2. Literature Review
2.1 Two-sided markets, Platforms & Network effects
Platforms are products or services that bring together different groups of users in two-sided markets (Cennamo & Santalo, 2013; Eisenmann, Parker & Van Alstyne, 2006). Two-sided markets are roughly defined as markets in which a platform enables interactions between two distinct groups of users and try to get the two sides “on board” by appropriately charging each side (Rochet & Tirole, 2003, 2006). These platforms provide infrastructure and rules that facilitate transactions and value creation activities (Cennamo & Santalo, 2013; Eisenmann, Parker & Van Alstyne, 2006; Evans, 2003; Gawer, 2010; Rochet & Tirole, 2006).
6
A related stream of research focuses on two-sided markets and uses related but distinct type of network effects to explain the value of a platform: same-side network effects and cross-side network effects. Same-side network effects occur when an increase in the number of users, attracts more users on the same side. For instance, the more people own a telephone, the more valuable the telephone is to each user, which makes it more attractive for people to use the telephone network. Similarly, online gamers benefit from the participation of other gamers on the same platform, so that they can play together with friends or other fellow gamers. However, these network effects can sometimes be negative. For instance, when there is competition in an online auction market. Cross-side effects refer to the value or utility of the platform that is generated by the users on the other side of the network. An increase in the number of adopters on one side will make the platform more attractive for users on the other side of the platform. For example, video game developers will incentivized to develop games only for the most popular platforms that have a critical mass of players. This is because developers need a customer base that is big enough to recover their upfront programming costs. On the other hand, players prefer platforms with a greater variety of offered games (Eisenmann, Parker & Van Alstyne, 2006; Venkatraman & Lee, 2004). When successful, platforms stimulate a virtuous cycle: more demand from one user group stimulates more demand from the other (Hill, 1997; Eisenmann, Parker & Van Alstyne, 2006; Schilling, 2002, 2003). The value of a platform grows as the platform matches demand from both user sides, enabling network effects and increasing returns to scale for platform providers.
2.2 Traditional pipelines versus platforms
Traditional pipeline businesses typically grow in one of two ways: Through vertical integration, buying upstream suppliers or downstream distributors or by widening the pipeline, creating new products and brands through horizontal integration (Parker, Van Alstyne & Choudary, 2016). By connecting products and services in a network of users, firms can create additional value with platforms through information and interactions, which together can lead to a competitive advantage. Apple has been one of the pioneering firms of the platform business model. They had great success in building a platform business within a conventional product firm, by leveraging the new rules of strategy they give rise to. While plenty of traditional pipeline businesses are still highly competitive, they usually cannot compete with platforms in the same market (Van Alstyne, Parker & Choudary, 2016). This is why huge incumbent firms like Walmart or Nike are rushing to incorporate platforms into their business models. To understand how the rise of platforms transforms competition, it is important to examine how platforms differ in a fundamental way from traditional pipeline businesses that have dominated many industries for decades. In the traditional value chain, value moves from the left to the right in a linear series of activities (Eisenmann, Parker & Van Alstyne, 2006; Van Alstyne, Parker & Choudary, 2016). Inputs at one end of the value chain go through multiple steps that transform them into an output that has more value. To the left of the firm is cost and to the right of the firm is revenue. In two-sided networks, cost and revenue are both to the left and right, because the platform has a distinct group of users on each side. The platform incurs cost in serving both groups of users but can also collect revenue from each user group (Eisenmann, Parker & Van Alstyne, 2006).
2.3 Increasing returns to scale
7
find the firms’ value proposition appealing. The promise of increasing returns to scale can lead to fierce competition (Eisenmann, Parker & Van Alstyne, 2006). This is why mature two-sided network industries are usually dominated by a few large platforms, such as in the video game industry. Sometimes, a single firm emerges and takes all of the market. The literature generally predicts a winner-take-all outcome, where the platform with the largest installed base of users will tip the market in its favor (Arthur, 1989, 1994; Cennamo & Santalo, 2013; Besen & Farell, 1994; Caillaud & Jullien, 2003; Gawer & Cusumano, 2008; Katz & Shapiro, 1994; Shapiro & Varian, 1999). The winning platform is not always the most advanced or the best technology, because network effects can effectively lock-out other platforms even if other platforms are superior in functionality or value (Arthur, 1994; Lee, O'Neal, Pruett, & Thomas, 1995; Schilling, 2002, 2003).
The unconditional logic of the winner-take-all approach has been questioned recently by several studies, because multiple platform systems may coexist. Asymmetrical network effects, local network effects (Eocman, Jeho & Jongseok, 2006; Shankar & Bayus, 2003; Suarez, 2005), modest costs of adopting multiple platforms (Eisenmann, 2007; Eisenmann, Parker & Van Alstyne, 2006) or differentiated consumer preferences (Armstrong & Wright, 2007; Eisenmann, Parker & Van Alstyne, 2006) may lead to the existence of multiple platforms in the same market. In the case of the video game console industry, only a few large players compete. The video game industry is not bound to be served by a single platform. Although multi-homing costs are high and network effects are positive and strong, the users seem to have a strong preference for special features. This causes the competition between rivals to be fierce but this makes it possible for the current competing platforms to coexist. Microsoft, Sony and Nintendo have shared the market for quite some time now (since the 6th generation of video game consoles).
Because of the increasing returns to scale, literature on platform competition emphasizes the importance of rapidly expanding networks of platform users and complementary products to capture entire markets (Cennamo & Santalo, 2013). A winner-take-all paradigm in multisided markets suggests that platform providers should embrace aggressive strategies to expand both their installed base of users and their catalog of complementary products so that benefits on each side of the markets are mutually reinforcing (Cennamo & Santalo, 2013). A get-big-fast strategy stimulates platform providers to rapidly acquire and grow their installed base of users, lock-in those users by ensuring there are considerable switching costs involved and undermining the ability of competitors to do the same (Eocman, Jeho & Jongseok, 2006; Cennamo & Santalo, 2013).
2.4 Conceptual Framework: Determinants of Platform Performance
8
consumer interest, which in turn increases platform performance in two-sided markets. The concepts illustrated in the conceptual framework will now be discussed.
Figure 1. Conceptual framework: Determinants of platform performance
2.5 Consumer Interest
Nelson (1970, 1974) classified products into search and experience goods according to the ability of consumers to gather product quality information before purchase. The internet has enabled consumers to search for experience goods online by sharing experiences with other consumers and gathering information otherwise hard to obtain offline (Alba et al., 1997; Klein, 1998; Peterson,
Direct Network Effects
Indirect Network Effects
Superstars Released Superstars Active
Superstars Total Average Quality Top 25 Games Relative Quality Top 25 Games
Games Released
9
Balasubramanian & Bronnenberg, 1997). This has made it easier to gather information on experience products, such as video games in the video game industry. Literature on consumer behavior points out that consumer interest has a positive effect on sales. For example, several studies based on Ajzen’s Theory of Planned behavior (1985, 1991), suggest that online search behavior leads to the intention to purchase a product, which in turn influences the chance of actually purchasing the product (Shim et al., 2001; Klein, 1998). Kulkarni, Kannan & Moe (2012) find that online search behavior, which is a valuable measure and indicator of consumer interest, is a significant predictor of product sales in the motion picture industry. Online search behavior is likely to take place when consumers seek knowledge on specific attributes of a product or when they seek to compare the product with other products (Huang, Lurie & Mitra, 2009). Consumers searching for information before the launch or release of a product are usually driven by the need for general information about the product, such as features, press releases, etc. Once the product is launched or released, emphasis of the search for information will shift to information that is more dedicated towards consumption, such as the availability of the product, the price, the quality of the product (reading online reviews), quality of complementary products, etc. (Kulkarni, Kannan & Moe, 2012). The information will create the option to actually buy the product or service. The literature on consumer behavior confirms that online search behavior is a decent measure for the degree of consumer interest (Huang, Strijnev & Ratchford, 2008; Kulkarni, Kannan & Moe 2012). Therefore, this study defines consumer interest as: The degree of interest that consumers show in a product by actively searching for it online. The intention to search for product information online acts as a central mechanism through which consumers reach the next level of decision-making. Searching for information can lead to acceleration of the buying process as consumers actively gather information on a product that they are interested in, which may lead to the intention to purchase the product (Kulkarni, Kannan & Moe, 2012; Shim et al., 2001). While searching online, consumers will be confronted with online store offerings of the platform, which provoke consumer interest and increase the likelihood of adoption. Therefore, consumer interest is expected to have a positive direct relationship with platform performance. The greater the number of consumers that search actively online for a platform, the greater platform performance.
H1: A positive direct relationship exists between consumer interest and platform performance.
2.6 Network effects and platform performance
10 2.7 Direct Network Effects: Installed base
Prior literature on two-sided markets emphasizes that growth in the installed user base is one of the main mechanisms that drives a platform’s value and market share (Armstrong, 2006; Cailloud & Jullie, 2003; Evans, 2003; Hagiu, 2005; Rochet & Tirole, 2003, 2006). Over the last decade, several studies have found empirical support for installed base effects on technology adoption (e.g. Cottrell & Koput, 1998; Shurmer, 1993; Wade, 1995). The installed base of users plays a major role in user adoption. Thus, if the installed base of a platform is not big enough in comparison with rival platforms, this may result in technological lockout (Schilling, 2002, 2003; Brynjolfsson, 1996; Cottrell & Koput, 1998; Katz & Shapiro, 1986; Ohashi, 2003; Shurmer, 1993; Wade, 1995). A large installed base can extend the range of a network, increasing the value of the user's training in the particular technology, which attracts more developers of complementary technologies, therefore increasing the options available to users (Ghoi, 1994; Katz & Shapiro, 1986; Schilling, 2002). Shankar & Bayus (2003) conclude that a firms’ customer network is an important strategic asset in network markets. The benefits of selling products on a popular platform with a large installed base usually outweighs the negative effects of competition (Cennamo & Santalo, 2013). A larger installed base of users increases the opportunity for its users to exchange value and lowers transaction costs and searching costs by stimulating the production of software. This in turn, attracts a greater number of consumers. This way, platform providers can create value by making use of the increasing returns to scale, increasing transaction opportunities and lowering searching and transaction costs (Cennamo & Santalo, 2013; Armstrong, 2006; Armstrong & Wright, 2007; Boudreau & Lakhani, 2009). The literature on network effects points out that the size of the installed base of users and the availability of complementary products reinforce each other. Products with a large installed base are likely to attract more developers of complementary products. Since the availability of complementary products influences the choice of users among competing platforms, the availability of complementary products in turn affects the size of the installed base. As a result, a self-reinforcing cycle arises, whereby the installed base increases the development of complementary products and the availability of complementary products attracts a larger installed base (Hill, 1997; Schilling, 2002). The same snowball effect can happen on the same side of the platform, when the increase in adoption will stimulate growth in adoption in the same user group. For example, when additional gamers on a gaming platform increase the value of the platform to other gamers, because they can play together, swap games, share game experiences, etc. The increase in value due to growth of the installed base will increase the consumer interest in the platform. Therefore, the installed base of users is expected to have a positive indirect effect on platform performance, via raising consumer interest. The larger the installed base of users, the greater the number of consumers that will search actively online, which in turn increases platform performance.
H2: A positive indirect relationship exists between the cumulative number of adopters (installed base) and platform performance, which is fully mediated by consumer interest.
2.8 Indirect Network Effects
11
Ohashi, 2005; Corts & Lederman, 2009). Prior literature on network effects proved that indirect network effects, such as the availability of complementary products as well as the quality of complementary products, affect the hardware sales of a new product. The expected utility of the primary product, and thus its sales, increases as more complements become available (Stremersch et al., 2007) or as the quality of complementary products increases (Binken & Stremersch, 2009).
2.8.1 Availability of complementary products
Research in economics has argued that hardware sales mainly depend on the availability of related complementary products (software) (Farrell & Saloner, 1986; Gupta, Jain & Sawhney, 1999; Katz & Shapiro, 1994). The importance of software has become a major concern for hardware firms (Binken & Stremersch, 2009). Most empirical evidence in the literature on indirect network effects has focused on the dimension of availability of complementary products (Binken & Stremersch, 2009; Stremersch et al. 2007). The complementary products usually have a major impact on the customer’s choice between competing technologies (Farrell & Saloner, 1986; Schilling, 2002; Katz & Shapiro, 1986). A technology for which the availability of complementary products is worse than that of competing technologies is less likely to be adopted by customers, such as in the case of Sony’s Beta video standard (Ohashi, 2003; Schilling, 2002). Therefore, the amount of available software is of critical importance to hardware sales growth (Church & Gandal, 1992; Gupta, Jain & Sahwney, 1999; Katz & Shapiro, 1986; 1994). A platform mediates the relationship between the installed base of users and the complementary products. Strategies aimed at managing complementary products can be even more critical than those designed to attract a large installed base of users (Gawer, 2010; Yoffie & Kwak, 2006). Ensuring that a platform offers a larger variety of complementary products than its rivals is one of the main mechanisms for increasing the installed base of users (Clements & Ohashi, 2005; Schilling, 2002). A large array of complementary products attracts a large installed base of users, because consumers prefer platforms with greater variety of complementary products. Greater variety of complementary products to choose from leads to more consumer interest. A conventional way of measuring the availability of complementary products in the video game industry is by analyzing the effect of the number of games released on each platform. This study measures the number of games in two ways. First, the number of games released in a certain week is expected to have a positive indirect effect on platform performance, via raising consumer interest. If platform providers ensure a steady flow of game releases, this can raise consumer interest because consumers will be continuously confronted with these new games and some consumers may prefer only the most ‘cutting edge’ or recent games. If one platform releases more games than another, consumers may choose the platform that releases the most new games. Therefore, the higher the number of games released in a certain week, the greater the number of consumers that will search actively online, which in turn increases platform performance.
H3a: A positive indirect relationship exists between the number of games released on a platform in a certain week and platform performance, which is fully mediated by consumer interest in that platform.
12
H3b: A positive indirect relationship exists between the cumulative number of games released on a platform and platform performance, which is fully mediated by consumer interest in that platform.
One of the ways platform providers can enhance the availability of complementary products, while denying these complementary products to its rivals, is through the introduction of exclusive games (e.g. developed for and sold only on the focal platform). This enhances the variety of software customers can choose from, which enhances the availability of complementary products relative to other platforms (Binken & Stremersch, 2009; Cennamo & Santalo, 2009). The primary objective of exclusivity is to provide users with high quality software that cannot be obtained on rival platforms (e.g. Lee, 2007; Mantena, Sankaranarayanan & Viswanathan, 2008). Platform providers can use exclusivity contracts as a strategy to ensure that certain games are only released on their platform to prevent rival platforms from obtaining valuable products (Lee, 2007; Mantena, Sankaranarayanan & Viswanathan, 2008). This approach can enhance a platforms competitive position, because customers with strong preferences for specific products have no choice but to adopt the platform that controls them. Therefore, exclusive game are expected to increase the sales of hardware. By securing exclusivity agreements, and thereby denying access to rivals, a platform might also reduce its competitor’s effective participation in the content and consumer markets. (Armstrong & Wright, 2007; Mantena, Sankaranarayanan & Viswanathan, 2008; Shapiro, 1999). By developing contracts with developers of software, platform providers can develop close relationships with those producers over time. This might enable enhanced coordination over the timing of releases and on the type and quality of content (Cennamo & Santalo, 2013; Yoffie & Kwak, 2006; Stennek, 2007). Exclusivity is a valuable commodity that can create competitive advantage (Shapiro, 1999). Exclusivity increases the variety of complementary products of a platform, while limiting the potential variety of its competitors, which increases the consumer interest in the platform. This study analyzes the effect of the exclusivity of complementary products on each platform in two ways. The first way is by measuring the degree of exclusivity on certain points in time. Exclusive games are often used as a way to increase sales at certain points in time (for example in the holiday season), because they can attract a lot of attention from consumers. Therefore, it is important to examine whether the effect of timed exclusives on consumer interest and platform performance is relevant. The number of exclusive games released on a platform in a certain week is expected to have a positive indirect effect on platform performance, via raising consumer interest. The higher the number of exclusive games released in a certain week, the greater the number of consumers that will search actively online, which in turn will increase platform performance.
H3c: A positive indirect relationship exists between the number of exclusive games released on a platform in a certain week and platform performance, which is fully mediated by consumer interest in that platform.
13
H3d: A positive indirect relationship exists between the cumulative number of exclusive games released on a platform and platform performance, which is fully mediated by consumer interest in that platform.
2.8.2 Quality of complementary products
Until recently, literature only acknowledged the availability of complementary products as a significant indirect network effect affecting platform performance (Stremersch et al. 2007). However, literature has shifted to a model in which the quality of complementary products is at least as important (Binken & Stremersch, 2009). For example, several studies suggest that high quality software titles may have a disproportionately large effect on hardware sales. (Binken & Stremersch, 2009; Frels, Shervani, & Srivastava, 2003; Shapiro & Varian 1998; Williams, 2002). Binken & Stremersch (2009) study the effect of exceptionally high software quality (often referred to as superstars) on hardware sales and find that the quality of superstars plays a vital role in the sales of hardware in two sided markets. These products often sell extremely well in comparison to competing products. Quality is strictly ranked and determined by the end users; only a few games are able to score very high in terms of product evaluations. Superstars are unique and have superior attributes that demand a disproportionately large compensation (Binken & Stremersch, 2009). Industries characterized by superstars, such as the music and gaming industry, show increasing returns to quality, because of the scarcity of high quality products (Binken & Stremersch, 2009; Rosen, 1981). Binken and Stremersch (2009) compare the commonly used indirect network effect of the availability of complementary products against the quality of the complementary products, and find that quality is more important than availability of complementary products (although availability still matters). They conclude that the availability of complementary products alone as a proxy for the indirect network effects, depicts an incomplete picture, because it does not account for increasing returns to quality (Binken & Stremersch, 2009). This study uses superstars as a measure of quality, by analyzing the number of superstars released in a certain week, the number of active superstars and the cumulative number of superstars released. Superstars evoke consumer interest, because they are of exceptionally high quality. This study expects the number of superstars released in a certain week to have a positive indirect effect on platform performance via consumer interest. The higher the number of superstars released in a certain week, the higher the consumer interest, which in turn increases platform performance.
H4a: A positive indirect relationship exists between the number of superstars released in a certain week and platform performance, which is fully mediated by consumer interest.
The number of active superstars indicates the lagged time effect of a superstar. Binken & Stremersch (2009) show that a superstar has a significant effect on hardware sales for the first five months after introduction. This study uses the same method to analyze the lagged time effect of superstars on consumer interest. It is expected that recently introduced superstars raise more consumer interest for a platform for a certain period, because they are new and of high quality. Thus, the number of active superstars is expected to have a positive indirect effect on platform performance, via raising consumer interest. The higher the number of active superstars, the greater the number of consumers that will search actively online, which in turn increases platform performance.
H4b: A positive indirect relationship exists between the number of active superstars released in a certain week and platform performance, which is fully mediated by consumer interest.
14
complementary products on a platform. Even if superstars become older, they could still have an important role of keeping overall quality ratings high. Therefore, the cumulative number of superstars is expected to have a positive indirect relationship with platform performance, via raising consumer interest. The higher the cumulative number of superstars, the greater the number of consumers that will search actively online, which in turn increases platform performance.
H4c: A positive indirect relationship exists between the cumulative number of superstars released and platform performance, which is fully mediated by consumer interest.
The quality of complementary products on a platform can also be examined by looking at the average quality of available games. Consumers may be more interested by the overall quality of games instead of more variety to choose from. Most consumers search for information about complementary products before they buy the product. The gaming industry is characterized by high switching costs, which may stimulate consumers to check the overall quality of games before they commit to a certain platform. Therefore, this study argues that higher overall quality of complementary products leads to more interest in a platform. The average game quality is expected to have a positive indirect relationship with platform performance, via raising consumer interest. The higher the average quality of games released, the greater the number of consumers that will search actively online, which in turn increases platform sales.
H4d: A positive indirect relationship exists between the average game quality of the top 25 games released and platform performance, which is fully mediated by consumer interest.
Finally, another relevant factor is the relative quality of supply of complementary products. If one platform has high quality games, it does not automatically lead to more interest in the platform if competitors offer the same or even higher quality. Competition may play an important role when consumers choose their platform, as they will not only compare specifications and quality of the hardware, but also compare overall quality of complementary software. It is expected that higher overall quality of games relative to other platforms will lead to more consumer interest, which in turn will increase platform performance. The higher the relative average quality of games on a platform, the greater the number of consumers that will search actively online, which in turn increases platform performance.
H4E: A positive indirect relationship exists between the relative game quality of the top 25 games released and platform performance, which is fully mediated by consumer interest.
3. Methodology
3.1 Sample StatisticsThis study is based upon data on the 8th generation of video game consoles: Xbox 360, PlayStation 4
and Wii U. The database used for this study was created by gathering data on three types of data: search behavior data, game related data, and sales data of the consoles. Data were gathered from a variety of sources that were available online. The database contains weekly observations from the introduction of the first video game console in November 2012 until April 2016. The Wii U was the first video game console to be released in November 2012, almost an entire year ahead of competition. Therefore the Wii U had the most observations: 175 weeks. PlayStation 4 and Xbox 360 were both released a year later within a week of each other, with respectively 124 and 123 weekly observations. This study focuses on the time period in which all three platforms were active and competing from November 2013 until April 2016. The lack of competition in the beginning of the 8th
15
create a fair comparison of data (ceteris paribus), this study therefore only used data on the period that all three gaming platforms were active and competing. The final data set available for analysis thus consists of 369 weekly observations, 123 weekly observations for each of the platforms. Data available for these platforms includes the number of games released, the number of exclusive games released, the number of superstar games released, the number of superstars active, data on search behavior on the Google search website, the average game quality and finally the number of sales for each platform.
The video game industry is an excellent industry to measure platform dynamics and network effects in two-sided markets. The video game industry has been chosen as the subject of this study, because reliable data is readily available. There are several sources of game related information available online. Furthermore, other studies have already documented the video game industry extensively (e.g. Binken & Stremersch, 2009; Cennamo & Santalo, 2013; Clements & Ohashi, 2005; Corts & Lederman, 2009; Shankar & Bayus, 2003; Venkatraman & Lee, 2004). Additionally, because multiple platforms exist in the industry, the difference between how platforms manage network effects can be observed more easily. Finally, the impact of indirect network effects is easy to measure in this industry, because the hardware (video game console) has little value for consumers without the software (video games).
3.2 Measures and Data Collection
3.2.1 Dependent variables
This study uses weekly sales as the dependent variable. This measure is considered to be the most important proxy for platform performance. Weekly sales data were gathered from www.vgchartz.com, a website that tracks sales on games and gaming consoles. The data on this website is regularly checked against manufacturer shipments and data released from other tracking firms to ensure accuracy.
16 3.2.2 Independent variables
Installed base of platform
The installed base is measured by the cumulative number of sales in week t. The higher the total number of sales, the bigger the installed base of users.
Availability of complementary products
To test whether the availability of games has any significant contribution to the performance of a platform, the current study measures the number of games released in week t, the total number of games released, the number of exclusive games released in week t and finally the total number of exclusives released.
Quality of complementary products
This variable represents the overall quality of the top 25 games for a platform. To measure the quality of these games, this study evaluates the number of active superstars released and active, the total amount of superstars released, the average game quality of the top 25 games and the relative game quality of the top 25 games. The relative game quality is measured with index numbers, computed by dividing the quality of the top 25 games by the average quality of the top 25 games of the other two competing platforms.
Video games are experience goods, which cannot be judged easily by consumers before consumption (Nelson 1970, 1974). Consumers commonly seek information about the quality when purchasing new products or services. Thanks to the growing popularity of the internet in the last decades, online consumer and expert reviews have become an important source for consumers seeking to determine the quality of products and services (Chen & Xie, 2005; Zhu & Zhang, 2010; Chen, Liu & Zhang, 2012). Since the very beginning of the video game industry, there have been reviews and scores on games. These have been useful tools to help with buying decisions for consumers (Everiss, 2008). This study acquired data on the quality of games from www.metacritic.com, a website that helps consumers make an informed decision about how to spend their time and money on entertainment. Metacritic's proprietary Metascore is based on a weighted average (more respected reviewers have more influence) of the most respected critics writing reviews online and in print (Metacritic, 2016). Metacritic has great influence in the entertainment industry (Greenwood-Ericksen, Poorman & Papp, 2013). When Metacritic came along in 2001, it changed the entire industry (Everiss, 2008). Because of the high acceptance of a direct relationship between scores and sales, some industry figures and studios have taken the logical step pf connecting scores to studio and employee valuation and have implemented policies to support and incentivize high scoring games (Greenwood-Ericksen, Poorman & Papp, 2013). The fact that even a firm’s stock price can go down as a result of a bad score for a new game proves that Metacritic has become the benchmark of the entire industry (Everiss, 2008). This study uses Metacritic as the main source to determine the quality of games, because it is one of the most comprehensive sources of game reviews available online (Chen, Liu & Zhang, 2012; Greenwood-Ericksen, Poorman & Papp, 2013). The reviews from Metacritic have been increasingly used in recent research (e.g. Chen, Liu & Zhang, 2012; Elberse & Anand, 2007; Huang, Strijnev & Ratchford, 2008; Sun, 2009; Wiles & Danielova, 2009).
17
identified a threshold for the quality of games at which point there are increasing returns to quality. To be considered a superstar, a software title must have a quality rating of 90 or above. A quality threshold of 90 identifies only the very best software titles as superstars, which have a significant effect on sales. This study will use the same measures to identify superstars and their effects. This threshold identifies 25 games of approximately 1259 software titles for the 8th generation of video
game consoles as superstars. 3.2.3 Control variables
Just like software, hardware also has as limited life expectancy due to continuous innovation and changing consumer preferences. Consumers lose their interest as the platform ages. The hardware becomes less attractive over time, because it becomes less “cutting edge” (Binken & Stremersch, 2009). Therefore, this study expects platform age to negatively affect the hardware sales of the platform. Platform age is a common control variable for the sales of hardware (Basu, Mazumdar & Raj, 2003; Binken & Stremersch, 2009; Brynjolfson & Kemerer, 1996; Shy, 2001; Stremersch, 2007). This study measures the age of a platform as the number of weeks that have passed since the platform's release.
This study also uses a dummy variable named ‘Holiday’ to control for the abnormal buying spree or seasonality of the sales in the holiday period in November and December (Christmas & New Year shopping effect). This control variable is also used in other studies (e.g. Binken & Stremersch, 2009). The dummy variable identifies a week as holiday (1) or not (0). The holiday effect is expected to affect the hardware sales positively.
Table 1. Variables and measures
Variable Measure
Platform Performance Weekly Sales
Installed Base Sales Total
Availability Games Released
Games Total Exclusives Released Exclusives Total
Quality Top 25 Games
Index Number Top 25 Superstars Released Superstars Active Superstars Total
Consumer Interest Google Trends Data
3.3 Analysis
18
4. Results
The results section is divided into three parts. The first part presents the descriptive statistics and the correlation matrix. The second part analyses the interaction effects of consumer interest, the direct network effects (installed base) and the indirect network effects (complementary products). The final section analyses the mediation effect of consumer interest in the relationship between all the independent variables and platform performance, including a Sobel test to measure the significance of the indirect effects.
4.1 Descriptive Statistics and Correlation Matrix
Table 2 provides an overview of the descriptive statistics of the variables used in this study. It contains the number of observations, mean, standard deviation, minimum and maximum of the data.
Table 2. Descriptive statistics
Variable N Mean Std. Deviation Minimum Maximum
Platform (ID) 369 2 .82 1 3
Week (t) 369 114 35.55 53 175
Holiday (HOLIDAY) 369 0.20 0.40 0 1
Platform age (PLATAGE) 369 79.67 43.07 1 175
Weekly Sales (SALES) 369 185931 2111061.1 26550 1607241
Total Sales (SALESTOTAL) 369 12506068 8445799 1104585 39775365
Games Released (GAMESINT) 369 3.07 2.57 0 23
Games Total (GAMESTOTAL) 369 206.42 125.1 19 484
Exclusives Released (EXCLINT) 369 .79 1.24 0 11
Exclusives Total (EXCLTOTAL) 369 60.41 60 7 225
Superstars Released (SUPINT) 369 .06 .24 0 1
Superstars Total (SUPTOTAL) 369 4.86 3.29 0 13
Superstars Active (SUPACT) 369 1.30 1.06 0 4
Top 25 Games (TOP25) 369 85.14 4.82 66.29 89.84
Index Number Top 25 Games (INDEX25) 369 100.10 5.61 81.22 121.11
Consumer Interest (INTEREST) 369 30.54 23.21 5 100
Table 3 presents the correlation among the variables used in this study. Most correlations were according to expectation. However, one of the results that stands out is the negative correlation between exclusivity and consumer interest. Both the number of exclusive games released in a certain week as well as the cumulative number of exclusive games are negatively correlated with the degree of consumer interest. One would expect these correlations to be positive: The higher the degree of exclusivity, the greater the consumer interest in a platform.
Table 3. Correlation matrix Variable Platform
Age
19 Exclusives Total 0.85*** 0.01 0.02 0.01† 0.69*** 0.39*** 1.00 Superstars Released -0.47 -0.01 0.05 0.23*** -0.02 0.07 -0.07 1.00 Superstars Total 0.67*** 0.02 0.86*** 0.36*** 0.85*** 0.17** 0.42*** 0.08 1.00 Superstars Active -0.07 0.12* 0.43*** 0.21*** 0.10* -0.05 -0.24*** 0.24*** 0.47*** 1.00 Consumer Interest -0.26*** 0.37*** 0.62*** 0.18** 0.07 -0.23*** -0.47*** 0.05 0.39*** 0.48*** 1.00 Top 25 Games 0.73*** -0.10† 0.62*** 0.27*** 0.75*** 0.11* 0.39*** 0.06 0.71*** 0.33*** 0.07 1.00 Index Top 25 Games 0.20*** 0.01 0.15** 0.02 0.17** 0.09† 0.08 0.07 0.28*** 0.42*** -0.03 0.58*** 1.00 Significance level: † p < .10, * p < .05, ** p < 0.01, *** p < .001
4.2 Direct effect of consumer interest on platform performance
Table 4 presents the results of the regression analysis of model 1 and 2. Model 1 shows the results of the control variables platform age and holiday on platform performance. These control variables together explain a decent variance in platform performance (R² = 0.42). Platform age has a significant negative effect on platform sales (B = 123.57, p < 0.01), which becomes insignificant when introducing consumer interest in model 2 (p > 0.10). The holiday effect has a significant positive effect on platform sales (both p’s < 0.001). Model 2 added consumer interest as an independent variable to test the effect consumer interest has on platform performance. The variance explained improves to (R² = 0.70). Consumer interest has a significant positive effect on platform performance (B = 5327.50, p < 0.001). Therefore, Hypothesis 1 is accepted.
Table 4. Regression analysis with consumer interest as independent variable
Factors Variable Model 1 Model 2
Control Platform Age -576.92 (0.003) ** 123.57 (0.398)
Holiday 332785.5 (0.000) *** 222136 (0.000) ***
Consumer Interest Consumer Interest 5327.50 (0.000) ***
R² 0.42 0.70
F 133.64 281.76
Observations 369 369
Significance level between brackets: † p < .10, * p < .05, ** p < 0.01, *** p < .001
4.3 Direct effects on consumer interest
20
Table 5. Regression analysis with Consumer Interest as dependent variable
Factors Variable Model 3 Model 4 Model 5
Control Platform Age -1.31 (0.000) *** -0.33 (0.000) *** -1.12 (0.000)***
Holiday 20.77 (0.000) *** 20.58 (0.000) *** 17.98 (0.000)***
Installed Base Sales Total 0.01 (0.000) *** 0.01 (0.882)
Availability Games Released -0.18 (0.480)
Games Total 0.26 (0.000) ***
Exclusives Released -0.58 (0.257)
Exclusives Total 0.01 (0.806)
Quality Superstars Released -3.13 (0.109)
Superstars Total 2.30 (0.002) **
Superstars Active -1.46 (0.065) †
Top 25 Games 2.13 (0.000) ***
Index Number Top 25 Games -0.66 (0.000) ***
R² (non-adjusted) 0.19 0.84 0.87
F 44.29 660.37 207.85
Observations 369 369 369
Significance level between brackets: † p < .10, * p < .05, ** p < 0.01, *** p < .001
4.4 Mediation effects
Table 6 presents the regression analysis to test for the mediation effect of consumer interest according to Baron and Kenny’s (1986) test. The last column states the type of mediation for every variable. The number of games released, the total number of superstars released and the number of active superstars seem to be fully mediated. The total number of sales and the total number of exclusives appear to be partially mediated by consumer interest. The other variables are not mediated at all. This study also conducted a Sobel test to investigate whether the indirect effects are significant. The Sobel tests indicated that all indirect effects are significant.
Table 6. Regression equation tests for consumer interest mediation Independent Variable Equation 1
X Y Equation 2 X M Equation 3 X & M Y Mediation Effect Independent Variable Independent Variable Independent Variable Mediator
Sales Total 0.01 (0.000) *** 0.01 (0.000) *** -0.01 (0.001) *** 7571.25 (0.000) *** Partial
Games Released 12971.33 (0.002) ** 1.61 (0.001) *** 2284.98 (0.440) 6642.97 (0.000) *** Full
Games Total 123.07 (0.162) 0.01 (0.183) 37.093 (0.536) 6674.07 (0.000) *** No
Exclusives Released -2614.20 (0.769) -4.25 (0.000) *** 27244.75 (0.000) *** 7018.89 (0.000) *** No
Exclusives Total -846.37 (0.000) *** -0.18 (0.000) *** 461.59 (0.001) *** 7244.73 (0.000) *** Partial
Superstars Released 10567.78 (.820) 5.08 (0.320) -23463.40 (0.457) 6700.39 (0.000) *** No
Superstars Total 16348.53 (0.000) *** 2.72 (0.000) *** -2180.00 (0.376) 6807.31 (0.000) *** Full
Superstars Active 66409.83 (0.000) *** 10.52 (0.000) *** -5146.69 (0.522) 6801.35 (0.000) *** Full
Top 25 Games 1126.03 (0.622) 0.31 (0.213) -970.12 (0.533) 6701.04 (0.000) *** No
Index Number Top 25 Games -299.01 (0.879) -0.11 (0.614) 429.12 (0.747) 6690.68 (0.000) *** No
Significance level between brackets: † p < .10, * p < .05, ** p < 0.01, *** p < .001
21
Table 6 shows that the number of games released is fully mediated by consumer interest when tested in isolation. However, table 5 and table 7 show that the number of games released has no significant effect on either consumer interest (model 5) or platform performance (model 8 & 9). H3a is rejected, because it has no significant effect when measured in conjunction with other indirect network effects.
Table 6 shows that the total number of games released, the number of exclusives released, the number of superstars released, the average quality of the top 25 games and finally the relative quality of the top 25 games are not mediated at all. H3b, H3c. H4a, H4d and h4e are rejected. Table 6 shows that the total number of exclusives released is partially mediated by consumer interest when tested in isolation. However, this variable has no significant effect on consumer interest in conjunction with other indirect network effects in table 5 (model 5) or on platform sales in table 7 (model 8 & 9). Therefore, H3d is rejected.
Table 6 shows that the number of active superstars is fully mediated by consumer interest when tested in isolation and it also shows a significant positive effect on consumer interest. However, table 5 indicates a weak significant negative effect on consumer interest when tested in conjunction with other indirect network effects (model 5). Table 7 shows that there is no significant effect on platform performance (model 8 & 9). Therefore, H4b is partially supported.
Table 6 shows that the total number of superstars is fully mediated by consumer interest when tested in isolation. Table 5 indicates a significant positive effect on consumer interest (model 5) and table 7 shows no significant effect on platform performance (model 8 & 9). Therefore, H4c is fully supported.
5. Discussion & Conclusion
This study investigated the impact of direct and indirect network effects on platform performance and analyzed the mediating role of consumer interest in the video game industry. The following sections present the discussion, conclusion, managerial implications and recommendations for future research.
5.1 Discussion
22
of games increase consumer interest in the platform as well. The direct network effect of the installed base becomes insignificant on both consumer interest and on platform performance when indirect network effects are included in the models. These results suggest that the indirect network effects have a stronger impact on consumer interest and platform performance and overrule the direct effect of the installed base. Although the direct network effect of the installed base is present and should be accounted for when managing platforms, it is not the primary concern for platform providers that want to increase consumer interest and platform performance. These results are in line with the findings of Shankar & Bayus (2003), they argue that the installed base of a platform is not the most important aspect, as firms with a smaller customer network can certainly outcompete and overtake the sales of a firm with a larger network size.
5.2 Conclusion & Managerial Insights
In an attempt to help platform providers and policy makers, this study investigated the role of direct and indirect network effects in two-sided markets and the importance of consumer interest as a mediating factor. A conceptual framework is presented that integrates both direct network effects (installed base) as well as indirect network effects (quality & availability of complementary products) to measure the impact on platform performance through consumer interest. This study has found evidence that consumer interest plays an important role as a mediator in the video game industry. Previous research has failed to consider consumer interest as a mediating factor when studying network effects. The findings of this study imply that indirect network effects are more important than direct network effects. Scholars need to take into account the mediating factor of consumer interest as well as the importance of indirect network effects. This is in contrast to the findings of Stremersch, Tellis, Franses & Binken (2007), they argue that the indirect network effects are not as important as expected on the basis of prior literature. However, the current study has found evidence that both the availability and quality of complementary products are crucial to ensure platform sales. Although most research on indirect network effects has focused on the availability of complementary products, this study has found evidence in line with the findings of Binken & Stremersch (2009) that the quality of complementary products is at least as important as the availability of complementary products to increase platform performance. This study has made it clear that quality of complementary products can no longer be ignored by scholars and platform providers alike. The findings of this study suggest that the quality of complementary products is even more important than the availability of complementary products to optimize consumer interest. Superstars are considered to be an important tool to increase the quality of supply and to increase consumer interest. However, besides quality, the availability of the complementary products still seems to matter. By increasing the total supply of complementary products, platform providers can have a significant impact on both consumer interest as well as platform performance.
23
increase consumer interest, thereby indirectly stimulating platform performance. Firms can ensure that they always have superstars active, by regulating the pace of release of superstars. Platform providers should carefully analyze which strategies are compatible to optimize platform performance. Conflicting strategies can arise when platform providers use multiple strategies at the same time, which can actually decrease platform performance (Cennamo & Santalo, 2013). While this study used the 8th generation of video game console to illustrate the conceptual framework
presented in this paper, the approach is generalizable to other two-sided markets to measure the impact of network effects on platform performance. For instance, search behavior can also be used as an indicator of consumer interest for other type of products, such as books or music. Superstars are not restricted to the video game industry, as superstars can be found in many industries that are also characterized by two-sided markets. The conceptual framework presented in this study can be used as a guide to stimulate platform performance. The results also lead to an important implication for policy makers. The mediating role of consumer interest shows that the perverse effects often associated with network effects are not present in the video game industry, because consumers in the video game industry are actually more enthusiastic about the product, and are not always forced to buy the platform with the highest market share or largest installed base.
5.3 Limitations and Future research
This study has several limitations, which could be addressed by further research. First, this study only examined one two-sided market: the market for video game consoles. Future research can focus on other type of markets as well, to see whether consumer interest also has an impact on those markets. Future research can also study multiple industries at the same time, so that differences between strategies can be observed more easily. Furthermore, this study did not focus on technological characteristics of the hardware. For instance, there is a clear difference between the Wii U and the other two consoles in technology. Future research should incorporate these differences in technology. Another limitation of the current study is that the quality was measured in its entirety: the average quality of the top 25 games may be measured over too large of a period, which can lead to little variety in data. Also, no data on previous platforms was gathered. Video gaming consoles can have a large effect on next-generation video game consoles, because of considerable switching costs (old games can usually be reused only on next-generation consoles developed by the same firms). Another limitation of this study is that platform performance was measured with weekly sales. Other measures of success or a combination of factors, such as market share or customer satisfaction may be a better proxy for platform performance. Future research could examine which proxy is best to measure the performance of platforms. Although the price of a platform can have a significant effect on hardware sales, the price of the consoles was not used as a control variable in this study, simply because there was no obtainable data on price-adjustments and promotion bundles globally. Finally, search behavior was measured by Google Trends data. Although online search behavior reflects consumer interest in a decent way, it does not measure consumer interest expressed in other ways, like word of mouth, third party reports or manufacturer print and web sources.
6. References
Ajzen, I. 1985. From intention to action: A theory of planned behavior. In J. Kuhl & J. Beckman (Eds.),
Action Control: From Cognitions to Behaviors. New York, NY: Springer.
24
Alba, J., Lynch, J., Weitz, B., Janiszewski, C., Lutz, R., Sawyer A., & Wood, S. 1997. Interactive Home Shopping: Consumer, Retailer, and Manufacturer Incentives to Participate in Electronic Marketplaces. Journal of Marketing, 61(July): 38–53.
Armstrong, M. 2006. Competition in two-sided markets. RAND Journal of Economics, 37(3):668-691. Armstrong, M., & Wright, J. 2007. Two-sided market, competitive bottle-necks and exclusive
contracts. Economic Theory, 32:353-380.
Arthur, W.B. 1989. Competing technologies, increasing returns, and lock-in by historical events.
Economic Journal, 99: 116-131.
Arthur, W.B. 1994. Increasing returns and path dependence in the economy. Ann Arbor: The university of Michigan Press.
Baron, R.M., & Kenny, D.A. 1986. The moderator-mediator variable distinction in social psychological research: Conceptual, strategic and statistical considerations. Journal of Personality and
Social Psychology, 51, 1173-1182.
Besen, S.M., & Farell, J. 1994. Choosing how to compete: strategies and tactics in standardization.
Journal of Economic Perspectives, 8(2): 117-131.
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.
Boudreau, K., & Lakhani, K. 2009. How to manage outside innovation. MIT Sloan Management
Review, 50(4): 69-76.
Brynjolfsson, E. 1996. The contribution of information technology to consumer welfare. Information
Systems Research, 7(3): 281-300.
Caillaud, B., & Jullien, B. 2003. Chicken & Egg: Competition Among Intermediation Service Providers,” RAND Journal of Economics, 34(2): 309–328.
Carrière-Swallow, Y., & Labbé, F. 2013. Nowcasting with Google Trends in an Emerging Market. Journal of Forecasting, 32(4): 289-298.
Chen, Y., Liu, Y., & Zhang, J. 2012. When do third-party product reviews affect firm value and what can firms do? The Case of Media Critics and Professional Movie Reviews. Journal of
Marketing, 76(2): 116-134.
Chen, Y., & Xie, J. 2005. Third-party product review and firm marketing strategy. Marketing Science, 24(2): 218–40.
Choi, H., & Varian, H. 2012. Predicting the Present with Google Trends. Economic Record, 88(Special Issue): 2-9.
Church, J., & Neil, G. 1992. Network effects, software provision, and standardization. Journal of
industrial economics, 40(1): 85-103.
25
Clements, M.T., & Ohashi, H. 2005. Indirect network effects and the product cycle: video games in the U.S., 1994-2002. Journal of Industrial Economics, 53(4): 515-542.
Coakly, M., & Song, D. 2015. Using Google Trends to Predict Retail Sales. PricewaterhouseCoopers. 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.
Eisenmann, T., Parker, G.G., & Van Alstyne, M.W. 2006. Strategies for Two-Sided Markets. Harvard Business Review, 84(10): 92-101.
Elberse, A., & Anand, B. 2007. The effectiveness of prerelease advertising for motion pictures: An empirical Investigation using a simulated market. Information Economics and Policy, 19(3/4): 319–43.
Eocman, L., Jeho, L., & Jongseok, L. 2006. Reconsideration of the winner-take-all hypothesis: complex networks and local bias. Managament Science, 52(12): 1838-1848.
Evans, D.S. 2003. Some empirical evidence of multi-sided platform industries. Review of Network
Economics, 2: 191-209.
Everiss, B. 2008. Metacritic has changed the games industry. Available at:
http://www.bruceongames.com/2008/06/04/metacritic-has-changed-the-games-industry/ [Accessed: 17 June 2016].
Farrell, J., & Saloner, G. 1986. Installed base and compatibility: Innovation, product preannouncements, and predation. American Economic Review, 76(5): 940-955.
Frels, J.K., Sharvani, T., & Srivastava, R.K. 2003. The integrated networks model: explaining resource allocations in network markets. Journal of Marketing, 67(January): 29-45.
Gandal, N. 2002. Compatibility, Standardization, and Network Effects. Oxford Review of Economic
Policy, 18(1): 80–91.
Gawer, A. 2010. Platforms, Markets and Innovation. Northampton, MA: Edward Elgar. Gawer, A., & Cusumano, M.A. 2008. How companies become platform leaders. MIT Sloan
Management Review, 49(2): 28-35.
Greenwood-Erickson, A., & Poorman, S.R., Papp, R. 2008. On the validity of Metacritic in Assessing Game Value. Eludamos. Journal for Computer Game Culture, 7(1): 101-127.
Gupta, S., Jain, D.C., & Sawhney, M.S. 1999. Modeling the Evolution of Markets with Indirect Network Externalities: An Application to Digital Television. Marketing Science, 18(3): 396-416.
Hagiu, A. 2005. Pricing and commitment by two-sided platforms. RAND Journal of Economics, 37: 720-737.
Hagiu, A. 2009. Two-sided platforms: product variety and pricing structures. Journal of Economics &
Management Strategy, 18: 1011-1043.
