Competitive Brokerage: How Information Management Capability And Collaboration Networks Act As Substitutes

ABSTRACT

IT-based information management and collaboration networks are both important sources of competitive information. Despite anecdotal evidence, limited research examines their contemporaneous impact on firms’ ability to compete effectively. We take an information asymmetry perspective to examine the mechanisms through which the firm’s information management capability and the structure of its collaboration network influence the structure of its competition network in product markets. We argue that taking a brokerage position in competition networks has a positive influence on a firm’s performance. We then explain how the firm’s information management capability and its position in collaboration networks each have a direct positive influence on its brokerage position in the competition network. Finally, we propose that information management capability and a central position in collaboration networks act as substitutes. We empirically test our model using a longitudinal competition network, a longitudinal collaboration network, and longitudinal secondary data about firms’ information management capability, drawn from 262 firms over a 15-year period. Our findings, robust to endogeneity concerns and alternative model specifications, reveal the direct and substitution effects of information management capability and collaboration networks on competition networks. This research contributes to the information systems and strategy literatures by offering insights into how IT enables firms to design competitive strategies by facilitating analyses of competitors’ information, and how the information gained by firms as they collaborate can enhance their ability to compete.

KEYWORDS:

• Information management
• information management capability
• competition network
• collaboration network
• competitive brokerage
• collaborative centrality
• firm performance
• business value of IT
• weighted network

INTRODUCTION

Managers in the networked economy face the two contemporaneous challenges of expanding their firms’ networks of collaborators — as increasing value is added to products and services through such networks [47], and of improving their firms’ ability to manage increasing amounts of information [98] — as awareness of competitors, consumers, and the environment is essential for maintaining competitive performance. Establishing and maintaining collaborative relationships requires significant time and effort. Similarly, developing the ability to generate, acquire, tailor, and disseminate information [82] requires substantial investment in information technology (IT), as sources and levels of information increase dramatically. Ergo, a firm’s collaboration network and its approach to information management are essential determinants of competitive performance and require significant investment. Although these factors appear to be dissimilar, they may have similar effects on a firm’s ability to compete effectively, and hence the problem of allocating resources between them may be tractable.

Every competitive action requires information to be captured, used, or directed, and thus all such actions have an information management component. This information is critical as it enables firms to outperform competitors, making information management important when developing competitive strategies. In the networked economy, competition can be conceptualized as a network (consisting of a firm, its competitors, and the competitive relationships between them) rather than a series of dyadic relationships, and hence effective management of information can provide an advantage for a firm within its network of competitors. Firms such as Goldman Sachs also recognize the importance of information management, with senior executives noting that information management is at the core of their business and an essential part of operations [110]. However, the theoretical and empirical relationship between information management and competition networks has not been extensively examined.

Collaboration networks (consisting of a firm, its collaborators, and their collaborative relationships) are also an important source of competitive information. Research shows that an initial collaboration often leads to further collaborations [76], but anecdotal evidence suggests that after a collaborative relationship, a firm may compete more aggressively with other firms in product markets. For example, the pharmaceutical firms Pfizer, Novartis, Merck, Bristol-Myers, and Depomed have previously collaborated, but have subsequently used the information gained to aggressively compete in product markets [33]. The dynamics in the enterprise systems software industry have been similar [22]. This evidence of the relationship between collaboration and competition networks has not, however, been adequately examined [92]. Finally, as competitive information can be gained through both information management and collaboration networks, their coetaneous effects on competition networks merit further investigation.

We address these gaps in the literature by applying the theoretical perspective of information asymmetry to examine the mechanisms through which collaboration networks and information management influence competition networks. We also validate conceptualization of competition as a network by examining how competition networks influence firm performance. Hence, we seek to answer three research questions: 1) How does IT-enabled information management influence a firm’s competition network? 2) How does a firm’s collaboration network influence its competition network, and how does IT-enabled information management affect this relationship? 3) How does a firm’s competition network influence firm performance?

Answers to these questions are important for theory because they imply that a more comprehensive model of interfirm behavior can be generated through the interplay of collaboration relationships and information management. This model can inform extant research in information systems (IS) and strategic management that provides insights into the potential tensions between collaboration and competition in dyadic (two-firm) relationships [105]. For example, studies show that a misalignment of interests in dyadic collaborative relationships can lead to competition between partners [52], and information acquisition can be an important influence on the ability of allied firms to compete with each other [18] as asymmetric information acquisition in collaborative relationships leads to differences in the information held by allies [38]. Notwithstanding the findings of such studies, important issues remain unaddressed.

First, partners in knowledge-intensive collaborations, such as R&D alliances, may compete to acquire information [71]. Such collaborations can lead to many information acquisition opportunities, and firms may simultaneously participate in several types of collaborations (e.g., manufacturing, supply, marketing) that offer different information that is useful in competitive product markets. The composition of a firm’s collaboration network may determine how it competes in product markets [25], as the network composition defines the firm’s interactions with collaborators and the information it receives from partnerships. These collaborative relationships, how a firm manages its information, and the effects on competition in product markets [26], require further investigation.

Second, the relationship between inter-firm collaborations (such as strategic alliances) and IT has generally been analyzed in terms of IT architecture as an enabler of collaboration [107], the extent to which IT reduces coordination and transaction costs in interfirm relationships [29], and its role in improving the accuracy and efficiency of processes in supply chain relationships [27, 112]. Depending on the structure of collaboration networks, IT-enabled capabilities have been found to be both complementary and substitutive to other organizational capabilities, [26]. However, although information management is important in collaboration, competition, and integration,¹ how it influences both collaboration and competition networks, has not been extensively examined.

Third, the tension between collaboration and competition has been examined in studies of strategy and IS, in terms of the effects of partner firm characteristics, such as knowledge and learning capabilities [17, 18, 38]. However, the effects of interfirm collaboration networks on competition networks, which reflect how firms compete in product markets, have rarely been investigated. Some researchers address this by taking a network perspective, such as Polidoro, Ahuja and Mitchell [91] who examine the effects of collaboration network embeddedness on partners’ competitive incentives and the dissolution of joint ventures, and Lavie [75], who assesses the effect of multilateral competition and the opportunities for alliances on a firm’s appropriation capacity. However, studies of how competition networks can develop from collaboration networks, or how networks beget networks, are limited.

We capture the structure of a firm’s competition network by drawing on the concept of competitive brokerage, defined as the extent to which a firm is an intermediary between others in a network, resulting from the competitive relationships between firms [4]. The literature on competition networks suggests that firms with high levels of competitive brokerage act as intermediaries, or “broker” other firms that do not directly compete with one another. Thus, competitive brokerage leads to better access to market information and an ability to control the dissemination of competitive actions across the network [4]. Firms often use information and signals from multiple product markets to design their strategies and competitive activities, which in turn shape the competition network. Hence, higher competitive brokerage implies that a firm has a central position in a competition network.

We suggest that a firm can use its IT-enabled information management capability (IMC) to reduce its information asymmetry externally – with stakeholders such as competitors and customers, and the environment – and internally. A firm’s IMC reflects the extent to which internal and external information is generated (e.g., through business intelligence applications), acquired (e.g., through data mining applications), tailored (e.g., via social networking platforms), and disseminated (e.g., using enterprise systems, cloud-based applications, and websites) [64, 82]. Thus, we suggest that IMC has a positive relationship with competitive brokerage.

Collaboration networks provide a firm with market and competitor information [44, 114], which help them plan and take competitive action. A firm’s level of information access directly depends on its location in relation to the center of its collaboration network [51], and thus a higher level of centrality confers better access to information. Thus, we measure a firm’s position in the collaboration network using collaborative centrality, defined as a firm’s level of degree centrality in a network of collaborative relationships [113]. A higher level of collaborative centrality represents a greater number of connections with partners, markets, and the environment. These connections provide important information that guides how the firm interacts with its competitors in markets. Thus, collaborative centrality reflects the extent in which a firm can access information from its partners, and therefore it reduces information asymmetry² between the focal firm, its partners, customers, and the environment [3, 5]. Thus, we suggest that a higher level of collaborative centrality facilitates information spillovers from partners, enriches environmental and market information, and provides improved information regarding the competition, which together result in a reduction in asymmetry [3]. This enables successful competitive actions, which in turn enhance competitive brokerage.

We further suggest that information acquisition through collaborative centrality is bounded by a firm’s IMC, and vice-versa, as the information involved is similar. IMC and collaborative centrality each reduce the marginal effect of the other on competitive brokerage, and thus can be regarded as substitutes. We test this research framework (Figure 1) using a longitudinal dataset over fifteen years which includes data on collaboration networks, competition networks, and IMC of firms in eleven industries.

Figure 1. Research Framework

Note: (1) Positive direct effects of IMC and collaborative centrality on firm performance are well established in the literature. As the focus of this study is competitive brokerage, we do not formally hypothesize these effects

We contribute to the IS and strategic management literature by integrating the perspectives of information asymmetry and information management in examining how collaborative relationships and information management influence the structure of a firm’s competition networks. We make three significant contributions to IS research. First, we address a previously unresolved issue in the literature by finding that IMC can act as a substitute for collaborative centrality. Second, we contribute to the business value from IT literature by identifying multi-level benefits of IT that manifest beyond firm boundaries to the network level. Third, by revealing the direct effects of IMC on competitive brokerage, we draw attention to the benefits IT can have on competitive strategies – as competition networks encapsulate the entire competitor ecosystem, the relationship between IMC and competitive brokerage offers a mechanism through which firms can outperform their competition and thereby overcome Red Queen dynamics. We contribute to the literature on collaboration networks by substantiating how information gained by firms through collaborative centrality can lead to competition, which represents a different type of interaction. For managers, our findings indicate that managing information through IS can improve a firm’s competitive strategy, particularly in industries in which interfirm collaboration is low. Finally, this study can serve as a methodological guidepost as our methodology can be broadly applied to reveal the multilevel effects of IT.

THEORY

Research on Competition Networks

Prior research uncovers heterogeneity in both competition and competitive tensions between pairs of competing firms [111]. Firms are exposed to varying degrees of competitive pressure, depending on their market power relative to competitors and the importance of the market to the firm [23]. These competitive relationships and differences in strength of competition between firms collectively form a competition network [103]. Competition networks describe the competitive structure of a product market. Two firms can be linked in a competitive relationship if they participate in the same product category. Firms in the dyad bound by this competitive link, compete against each other with different levels of intensity [23], as reflected by the strength³ (weight) of the competitive link. The position of a firm in a competition network is evaluated through its competitive links and their strength, which together reflect the intensity of competition that a firm faces. Position in a competition network influences the information a firm can gather from markets [75] and how it responds to actions of its competitors. The information possessed by a firm also affects its competition network, as it influences who it competes against and the intensity of competition [111].

Pure competition networks have seldom been the focus of research. Most work has examined the influence of simultaneous cooperative and competitive, or coopetition, relationships, in which two firms compete in some domains while forming alliances in others [16]. Coopetition has mainly been studied using resource-based [23], game theory [16], and network approaches [45]. Studies addressing coopetition from a network perspective consider collaboration networks as the basis for examining competitive relationships between firms. However, this substitution may be erroneous, as the transactions within a network can directly influence its structure, value creation, capabilities, and moderating effects on firm behavior [1, 79]. Thus, collaboration networks should not be used as surrogates to understand competitive behavior. Instead, theoretical networks based purely on competitive relationships or collaborative relationships should be developed to fully understand the specific effects of competition and collaboration [104]. We follow this approach and examine the two through separate networks.

How Competitive Brokerage Influences Firm Performance

The literature suggests that “the location of firms in interfirm networks is another important element of competition … because a network approach allows consideration of the strategic benefits from optimizing not just a single relationship but the firm’s entire network of relationships” [51]. Thus, a network position approach offers a better understanding of the antecedents and consequences of heterogeneity in competition and competitive tensions, by changing the frame of reference from atomistic to relational.

We define competitive brokerage as the extent to which a firm is an intermediary between others in a network constituted of the competitive relationships between firms (i.e., a competition network) [4]. A firm is in a competitive brokerage position when its competitors do not compete directly among themselves. For example, in Figure 2, firm Z is a competitive broker and competes with firms Y, X, and W, which do not compete with one another. Firm Z competes directly with firms X, W, and Y with different levels of intensity, as depicted by the weights on the links. The differing intensity of competition, or competitive asymmetry, suggests that firms use different strategies to compete in important product categories. In addition, firm Z serves as a bridge or broker between firms in different competitor communities (i.e., firms [X, S, T, U], firms [Y, V], and firms [W, Q, R, A, B]). Competitive brokerage is not necessarily dependent on a firm’s diversity, as this position can be achieved by bridging only strategically important product categories. Thus, the bridging ties of competitive brokers extend to product categories that are strategically important for the focal firm.

Figure 2. A Simplified Competition Network with Z as a Competitive Broker

Brokerage therefore integrates competitive asymmetry and intensity and involves the evaluation of complete competition networks. Hence, we submit that it offers a comprehensive and accurate representation of bridging ties when examining competition. For example, our data (detailed later in this manuscript) reveal that Procter and Gamble (P&G) has high levels of competitive brokerage. The firm bridges several communities of competitors (e.g., dental care, laundry detergents, shaving product manufacturers) in its competition network. P&G has produced several innovative products, such as Swash, Tide Acti-Lift, and Crest 3D White, which have improved its performance relative to its competitors. Thus, we suggest that P&G’s bridging ties gave it access to valuable information, enabling it to combine knowledge gained from different network communities, and thereby innovate. In summary, competitive brokerage is a competition network concept that comprehensively captures a firm’s network position, bridging ties, diversification, multi-market position, and competitive asymmetry, relative to those of its direct and indirect competitors.

We propose that competitive brokerage positively influences firm performance due to two reasons. First, competitive brokers have better access to unique and diverse market information⁴ [4, 103]. A firm with high levels of competitive brokerage bridges competing firms, and thus can observe information from each of its competitors’ markets.⁵ Although competitors do not directly or explicitly share market information with each other, they can infer it by monitoring the competitive actions of their direct competitors [103]. Access to such unique and diverse information enables a firm to plan its competitive actions while considering the activities of its direct and indirect competitors. Thus, a firm in a competitive brokerage position can adapt to the changing environment and identify market opportunities that others may not perceive. Second, competitive brokers can disseminate competitive actions across the competition network [4]. As they have access to diverse information from various competitor groups, they can assess the environment and consider the consequences of their direct and indirect competitors’ actions [94]. Figure 2 provides an example. Firm Z is a competitive broker, and if firm Q takes a competitive action, firm Z can scan and evaluate the environment before firm W retaliates. Firm Z’s competitive brokerage position implies that it has the foresight to evaluate the advantages and disadvantages of initiating a competitive action in retaliation to firm W. If firm Z responds to firm W, firms X and Y may also initiate competitive actions, thus unleashing a competition war. Thus, brokers such as firm Z can manage the spread of competitive action in a competition network.

Firms that are not in competitive brokerage positions have limited information, which may in turn limit their view of product markets [23]. Their abilities to evaluate competitive actions may then be limited, and they may disregard important business opportunities and product market developments. For example, in Figure 2, firm A is not a competitive broker, because its direct competitors (firms B and Q) compete against each other. Firm A is less likely to have access to unique and diverse information from these competitors because firms A, B, and Q form a completely closed cluster. In such network clusters, the same firms continuously interact within the market, and thus the information they can access becomes redundant. The attention of firm A is limited to the actions of firms B and Q as this redundant information is self-reinforcing and there are cognitive limits to managerial attention [100]. Ergo, firms that are part of a close cluster of competitors are more likely to only pay attention to actions within the cluster and have similar market information. Firms that are not in a competitive brokerage position (e.g., firm A) are less likely to have information advantages in competition networks. Information is necessary for firms to sustain a competitive advantage and therefore those with informational disadvantages are unlikely to outperform others. In summary, we suggest that a competitive brokerage position in competition networks provides advantages in the form of information and the management of competitive actions, which enable a firm to make relatively more effective strategic choices. Thus, we propose the following hypothesis:

Hypothesis 1: Competitive brokerage positively influences firm performance.

Research on Information Management Capability

In the current information-rich environment, firms must generate, provide, store, distribute, tailor, and use significant volumes of internal and external information [93]. IT can facilitate information management [13] and firms can leverage that information in the design of competitive actions [26]. This ability to manage and leverage information to compete in markets is referred to as IMC [82]. Mithas, Ramasubbu, and Sambamurthy [82] define IMC as “the ability to (1) provide data and information to firms with the appropriate levels of accuracy, timeliness, reliability, security, and confidentiality; (2) provide universal connectivity and access with adequate reach and range; and (3) tailor the IS infrastructure to emerging business needs and directions.” ⁶ Research suggests that firms should focus on developing IMC through IT so they can increase their market competitiveness [31, 32]. In addition, prior research shows a relationship between IMC and organizational outcomes. For example, Mithas, Ramasubbu, and Sambamurthy [82] and Ravichandran and Lertwongsatien [93] find that IMC and information resources positively influence firm performance.

Although a positive relationship between IMC and firm performance [82] has been identified, the intermediary relationships between IMC and competition networks remain underexplored. Examining the relationships between collaborative centrality, IMC, and competitive brokerage is also theoretically and practically relevant, as both IMC and collaboration networks provide information that firms require to effectively compete.

How Information Management Capability Influences Competitive Brokerage

Competitive dynamics literature suggests that “a competitor will not be able to respond to an action unless it is aware of the action, motivated to react, and capable of responding” [25]. Thus, the position of a firm in a competition network is dependent on its access to market and environmental information, as this represents a valuable asset that reduces the information asymmetry between a firm, its competitors, partners, customers, and the environment. By possessing market information, a firm can monitor its competitors’ strategies, competitive actions, and activities. For example, market information can be used to monitor product introductions and withdrawals, market entries, alliances, and recruitment practices [77]. The collection, analysis, and interpretation of market information enables firms to analyze the competitive structure of product markets [103] and be aware of new market developments. Their view of the market may be myopic without such information, which can affect their strategic planning [55]. Environmental information enables firms to develop industry perspectives and identify opportunities [84], and anticipate and adapt to the changing environment by evaluating the competitive and institutional landscapes [102]. Without such awareness, the judgments of firms may be impaired, which can affect their competitive position.

Market and environmental information are generally public, which any firm could potentially use when planning a competitive strategy. However, cognitive limits at the organizational and individual levels imply that firms are selective and pay more attention to information that is particularly salient (at the exclusion of non-salient information) and focus on the relevant actions of their direct competitors [103]. For example, in Figure 2, firm W is more likely to collect, analyze, and interpret the salient and impactful information derived from the actions of firm R. Thus, the information that a firm has access to and focuses on is limited by its ability to acquire information. However, we propose that as IMC enables a firm to acquire information beyond that which is merely directly observable and salient, it positively influences competitive brokerage.

First, IMC helps a firm reduce the information asymmetry between the focal firm, its competitors, partners, customers, and the environment [29]. For example, a firm can use data mining and website crawling IT applications to monitor websites and capture competitors’ product and pricing information [94]. Thus, through IMC, it can obtain information about product categories and products of both its direct and indirect competitors [48, 69]. This reduces information asymmetry and aids the focal firm develop its competitive actions, such as market entry and promotions. These both enhance its competitive brokerage. In addition, such IT can help a firm to identify and record how other firms are responding to new regulations [48]. This reduces information asymmetry and enables the firm to assess competitors’ behavior and then create a response strategy to maintain, increase, or withdraw participation in product markets, improving its competitive brokerage. Blogs, social networks, and crowdsourcing sites can also enable a firm to directly interact with consumers and capture their preferences [69]. For example, customers often interact with firms in social networking sites such as Instagram and Facebook to express their concerns and support for the brand [9]. Information obtained from these sites reveals consumer sentiment and preferences, which are useful in the design of competitive actions [9, 69], leading to better competitive brokerage. Online crowdsourcing platforms facilitate the co-creation of products between customers and firms. This has become a new paradigm in the development of new products, and companies such as Coca-Cola, Nestle, and Carlsberg have been successful in developing new products and campaigns to increase their sales [72]. These systems illustrate how firms use information obtained through IMC to design competitive actions and improve their competitive brokerage [26].

Second, IMC also reduces information asymmetry within the firm by enhancing the synchronization and integration of information, by connecting and promoting interactions between multiple business units [17]. Video conferencing, asset management, and product design management systems improve a firm’s internal information sharing efforts and communication frequency⁷ [28]. These types of interaction encourage productivity and knowledge sharing, leading to accelerated product development [37]. The synchronization, integration, and sharing of information helps firms to reduce the information asymmetry between business units and departments, and thus enables multi-unit coordination when designing competitive actions. For example, to understand the requirements, preferences, and expectations of customers and markets, a marketing department can share information with the new product development department [78]. This coordination enhances the likelihood of product success, and launching successful products enhances competitive brokerage. The generation, coordination, and sharing of information within a firm also enables existing products to be modified and new products to be launched, to satisfy customer demand.

In summary, IMC reduces a firm’s information asymmetry with its competitors, partners, customers, and the environment, and within the firm by monitoring its competitors’ actions, identifying consumer needs, competitively adapting to new regulations, and sharing and synchronizing internal information. This enables a firm to better design its competitive actions; thus, IMC positively influences competitive brokerage.

Hypothesis 2: Information management capability positively influences competitive brokerage.

Research on Collaboration Networks and their Influence on Competition

The collaborative relationships between firms collectively form a collaboration network. Two firms can be linked through a collaborative relationship if they jointly participate in an alliance. Collaboration networks are a “source of competitive actions” [44], because they provide the firm with access to information⁸ and markets [51], which enables it to be aware of competitors and respond to their actions [114]. Practical examples can illustrate how firms form collaborative relationships enabling them to compete in other markets. For example, the Spanish telecom firm Telefonica SA and the Banco Bilbao Vizcaya (BBVA) collaborated to provide a wireless payment system over Telefonica’s cellular network [95], which was a new market for both firms. Similarly, T-Mobile and the digital-only bank BankMobile joined forces to introduce T-Mobile Money, a mobile-first checking account [14]. Although BankMobile was already competing in the banking industry, its collaboration with T-Mobile enabled it to compete more aggressively by introducing a new banking product. These examples illustrate how firms can leverage collaborative relationships to result in the introduction of new products. Thus, collaboration can enable a firm to compete either in new markets or more aggressively in existing markets through increased access to information, which reduces information asymmetry.

Research suggests that collaboration networks enhance firms’ access to market and environmental information [50] through two mechanisms. First, collaboration networks offer a channel for a firm to actively seek information from and transfer information to partner firms [90, 100]. Each firm in the collaboration network is thus both a recipient and transmitter of information [3]. For example, a firm can seek and receive the information required to conduct joint projects from its partners, such as adapting a product to satisfy emerging consumer preferences or taking appropriate action to meet new regulations. As collaboration between firms is strengthened, identification and trust between partners is enhanced [70]. This deeper relationship results in a more intensive exchange of information between partner firms [3]. Second, collaboration networks enable a firm to receive information about its partners’ activities [50]. Intense collaboration between firms increases the likelihood of information appropriation [87], which results in partners gaining access to high-value information about each other’s operations and responses to environmental challenges. This information is valuable as it can be used to design similar competitive actions in other markets. It also increases environmental awareness [90], such as identifying the technological developments and managerial competencies that can be applied in designing competitive actions such as introducing new products, entry into new markets, and pricing [24].

In sum, information shared within collaboration networks mitigates the information asymmetry between partners and provides the stability and knowledge required to successfully compete in markets, for example by launching new products.⁹ These actions change the structure of the competition network because competitive links become stronger, and new links may be formed.

How Collaborative Centrality Influences Competitive Brokerage

We adopt a network position approach to conceptualize the influence of collaboration networks and to examine the influence of collaborative centrality on competitive brokerage. Collaborative centrality refers to a firm’s central position in a network of collaborative relationships [113]. A higher level of collaborative centrality represents a higher number of connections with partners, which provide access to more information [3] and therefore influence firm performance [22]. Collaborative centrality reduces the information asymmetry of a firm with its competitors, partners, customers, and the environment. In comparison, collaborative brokerage¹⁰ is oriented toward the control of information [20], which does not lead to the reduction of information asymmetry.

First, collaborative centrality reduces the information asymmetry between a firm and its partners as it enables the firm to capture knowledge spillovers from its partners [81], identify their weaknesses, and reveal their intentions in markets [91]. This information can be used to plan competitive actions, which if successful can improve its competitive brokerage. Second, collaborative centrality helps reduce the information asymmetry of a firm with its competitors, customers, and the environment as it situates the firm at the confluence of multiple information flows regarding the market and the environment, which the firm can apply to its advantage, for example by developing new products [81]. In a transaction-oriented environment, these advantages enable the firm to initiate competitive actions in product markets, which can improve its competitive brokerage position. Third, greater collaborative centrality enables the firm to access information regarding the past competitive actions of other firms, thus further reducing information asymmetry. The firm can then anticipate the consequences of its own actions and determine the reasoning behind those of other firms. Thus, its competitive actions will be more effective, enabling the firm to attain a competitive brokerage position. Conversely, firms with low collaborative centrality will be at the periphery of the collaboration network, will have inadequate information to interpret the causes and consequences of their partners’ actions [45], and will therefore lack information to accurately respond. In addition, a peripheral firm will avoid attacking or responding to the competitive actions of a firm with greater collaborative centrality.

In summary, collaborative centrality increases information spillovers from partners, provides multiple information flows related to the environment and the market, and offers information advantages that increase the effectiveness of competitive actions, thus enhancing competitive brokerage.

Hypothesis 3: Collaborative centrality positively influences competitive brokerage.

Collaborative Centrality and Information Management Capability as Substitutes

The substitution effects of IT-enabled capabilities have rarely been examined [36], but some scholars suggest that IT-enabled capabilities such as IMC may function as substitutes for firm resources in competition contexts [e.g. 26]. From a statistical perspective, substitutes are characterized by a negative interaction effect and are the opposite of complements, which have a positive interaction effect [108]. Conceptually, substitutes can have one of two effects: the use of one decreases the use of the other, or the use of one decreases the marginal benefits of the other [108]. We propose that the relationship between IMC and collaborative centrality is substitutive, as the marginal effect of IMC on competitive brokerage decreases as the level of collaborative centrality increases, or the marginal effect of collaborative centrality on competitive brokerage decreases as the level of IMC increases. Thus, a significant negative interaction between IMC and collaborative centrality will suggest that they are substitutes. IMC and collaborative centrality can have a positive influence on competitive brokerage through three similar routes: 1) by reducing the information asymmetry between a focal firm, and its competitors, partners, customers, and the environment; 2) by providing information that can help plan competitive actions; and 3) by improving the internal and external coordination of information.

First, collaborative centrality enables a firm to obtain information regarding markets and the environment, thus enhancing its understanding of customers’ current and future needs [3]. This facilitates the firm to anticipate customer needs and improve its products or introduce new products. Similarly, IMC fosters direct and synchronous interactions between the firm and its customers, who provide information regarding their preferences. This direct communication increases the likelihood of product improvement and co-development [82], such as on digital crowdsourcing platforms, which can be useful for product innovation [41].¹¹ The launch of new products changes the competition network structure and enables the firm to improve its competitive brokerage. Second, collaborative centrality enables a firm to access more information and thus anticipate its competitors’ actions [50]. For example, a firm can obtain information from partners about the actions of other firms, such as their product development and market entry. This enables the firm to effectively plan competitive strategies. Conversely, a firm’s IMC can also result in it collecting publicly available information about competitors and their actions, which can enable the firm to predict the behavior of its competitors, and thus it can again effectively plan competitive strategies. For example, the firm can use data mining systems to monitor and obtain information from competitors’ websites and social network sites [48, 69], which can be analyzed through business intelligence software to predict their subsequent competitive actions. Third, collaborative centrality fosters sharing and transfer of information inside and outside of the firm,¹² which reduces fluctuations in the availability of resources and increases its bargaining power [46]. Thus, greater collaborative centrality enhances the firm’s ability to compete and provides advantages over its weaker competitors [33]. IMC plays a similar role in a firm’s internal and external information coordination¹³ [17], as it also reduces fluctuations in the availability of resources, increases internal access to non-collaborative external knowledge, and incentivizes the firm to compete by taking advantage of its competitors’ weaknesses [73]. For example, a firm can collect customer feedback about its own products and those of its competitors from social media, which it can use to identify gaps in rival offerings and improve its own products.

Thus, both IMC and collaborative centrality have similar effects, as they provide similar information regarding markets, the environment, and competitors’ actions. They also both increase information coordination activities. However, as the information that can be obtained through greater collaborative centrality or from IMC will be similar, as are their information coordination effects, these may become supererogatory and redundant. Thus, the marginal effect of IMC on competitive brokerage decreases as the level of collaborative centrality increases, and vice-versa. We therefore propose the following hypothesis:

Hypothesis 4: Collaborative centrality and information management capability are substitutes in attaining a competitive brokerage position.

METHODOLOGY

We collected archival data for this study from five sources — the Passport database, the SDC Platinum database, Compustat, Thomson Reuters Eikon, and multiple computer journals. We discuss our data collection below.

Competition Network Data Collection and Construction

To build a longitudinal firm-to-firm competition network, we extracted data of every industry and firm available in the Passport database.¹⁴ This database provides global coverage of 11 industries and is recognized by marketing and management scholars and practitioners as a reliable and accurate source of information. We obtained information of both private and publicly listed firms to build the competition network, as this enabled us to accurately illustrate interfirm competition across the 11 industries.¹⁵ The Passport database contains detailed information on market size, market share, brands, product categories, and number of products available across multiple distribution channels. The database does not contain complete data before 2004 or after 2011 for several industries, so our main analysis focuses on the 2004–2011 period. To ensure the robustness of our results, we also collected data for the 2012–2018 period and estimated our specifications using data from 2004 to 2018. Although the network constructed from this data is incomplete, we found qualitatively similar results, as shown in Tables OS.11 and OS.12 in the online supplement. To enable cross-firm comparisons, we established a clear geographic boundary and limited our data collection to the U.S. [23].

We built our firm-to-firm competition network based on firms’ annual participation in product categories. As firms compete and build market structures through products and services [35], those participating in the same product categories are considered direct competitors because their products are substitutes, offer the same functionality, and satisfy the same consumer needs [103]. Ergo, we classified two firms as competitors if they overlapped in at least one product category. Details of the competition network construction process are provided in Appendix A.

Collaboration Network Data Collection and Construction

Our competition network contains both publicly listed and privately owned companies. However, privately owned companies rarely disclose their collaborative partnerships and financial statements. Due to these data limitations, we follow previous studies and focus on the collaboration networks and financial information of publicly listed firms, which submit fillings to the U.S. Securities and Exchange Commission. Developing collaboration networks only from publicly listed firms is a well-accepted and common approach in the literature [99]. The collaboration network data of the 262 publicly listed companies in our sample were extracted from the “Strategic Alliances” section of the SDC Platinum database [99, 101], which is the most comprehensive and commonly used source of such data [33, 99]. We found 8,150 unique collaboration relationships among the 262 firms between 2004 and 2011. The details of the construction of the collaboration network are given in Appendix A.

Figure 3 presents graphical visualizations of the competition network (panel A) and the collaboration network (panel B) over one year. Other visualizations and details of the competition and collaboration networks are provided in the “Network Visualization” section of the online supplement.

Figure 3. Visualization of Competition and Collaboration Networks

Note: (1) Figure 3A is a visualization of the whole competition network for one year. (2) Figure 3B is a visualization of the whole collaboration network for one year. (3) Both visualizations were created using the Force Atlas 2 algorithm, which is a force-directed and continuous algorithm that creates symmetrical networks with minimal edge crossing

Financial Data Collection

The financial data of the firms in our sample were obtained from Compustat, Thomson Reuters Eikon, and corporate annual reports.

IMC Data Collection

Using secondary data to assess IT-enabled capabilities can alleviate concerns about common method bias in surveys [59]. Quantifying IT systems and assessing firms’ IT-enabled capabilities and investment initiatives through their news announcements is a well-established approach [97, 98]. Based on prior research, we sourced our data from a) the Factiva database, which includes more than 30,000 news sources, and b) the computer journals Computerworld, Networkworld, eWeek, eWeek Security Watch, Infoworld, and InformationWeek. These journals consistently report IT implementations and IT spending of firms in the U.S. Thus, we ensured that our data sources were comprehensive and covered firms’ IT initiatives, which also addresses any concerns about media bias.¹⁶ We eliminated duplicate and irrelevant news announcements and finally obtained 2,864 IT-related news reports for the 262 firms.

Measurements

Competitive Brokerage. This was measured as weighted betweenness centrality¹⁷ in the competition network [86], which represents the proportion of times that firm j needs firm i to reach firm k via the shortest path [15]. This measurement assumes that intermediary firms lie along the shortest paths and quantifies the number of times that these intermediary firms act as bridges or brokers. The betweenness measure also illustrates the ability of a firm to capture information and signals from multiple competitors across the competition network. Firms with higher betweenness (those that act as bridges or brokers) are located at the intersections of groups of competitors. Thus, they can capture information and signals from several product markets. Betweenness also captures the number of competitive actions conducted by each firm that pass to every other firm through a specific third firm [15]. Betweenness therefore measures the degree to which a firm can control the diffusion of competitive actions through the network. For example, a firm with a high level of betweenness can spread such actions to other parts of the network or stop their diffusion [15]. Thus, the betweenness measure is ideal for capturing the features of competitive brokerage. The calculation of Competitive Brokerage is provided in Appendix B.

Firm Performance. We used firm profits (revenues minus expenses) as an indicator of performance, as they reflect the most immediate effects of how firms compete in the markets and competition [30]. A difference between a firm’s and its competitors’ profits is often observable if it has superior performance [30]. We log-transformed firm profits to reduce the skewness of the variable.

Collaborative Centrality. Consistent with the literature [81, 113], we measured Collaborative Centrality with Freeman’s degree centrality [40],¹⁸ defined as the number of direct contacts (links in the collaboration network) of the focal firm (p_k) and represented as $C_D\left(P_k\right)={\sum }_{i=1}^{n}a\left(p_i,p_k\right)$, where $a\left(p_i,p_k\right)$ is 1 if firm $\left(p_i\right)$ and firm $p_k$ have a collaborative relationship. We applied logarithmic transformation to reduce positive skewness.

IMC. This is reflected in the information systems of a firm, which enable it to generate (e.g., business intelligence), acquire (e.g., data mining), tailor (e.g., social networking platforms and enterprise content management systems), and disseminate internal and external information (e.g., enterprise systems and websites) [82]. We meticulously coded news announcements about a firm’s information systems to create the measure for IMC. We developed a formal systematic coding protocol by adapting others from the literature [26, 59], and were thus able to objectively identify and classify firms’ IS implementations into IMC from examining the IT news articles (details regarding the protocol development and news coding processes are provided in the “IT News Coding” section of the online supplement). This approach for operationalizing IMC is consistent with studies that use secondary data on IS in the form of summative indexes of binary measures to evaluate firm capabilities [e.g., 97, 98] and operationalize IT use [e.g. 8]. Thus, following the literature, we used the number (count) of mapped information systems as a measure of IMC [59].

Control Variables

We included firm and industry level control variables in our model to account for extraneous effects (see Table OS.2 of the online supplement for the descriptive statistics and correlations). The variables included have been previously identified as potential influencers of either network positioning [106, 116] or firm performance. As firm level controls, we included the number of competitive links, diversity, multimarket contacts, number of M&A, firm size, firm age, and market share; and included industry capital intensity as an industry-level control. We theoretically justify our choice of control variables and provide the measurement details in Appendix C.

ANALYSIS AND RESULTS

To test our hypotheses, we conducted two separate analyses, with Firm Performance and Competitive Brokerage as the dependent variables. Hausman’s specification tests suggest fixed effects models for analyzing Firm Performance and random effects models for the Competitive Brokerage analysis. An analysis of Firm Performance was conducted to test Hypothesis 1. Models 1 and 2 in Table 1 indicate that Competitive Brokerage positively influences Firm Performance (β = 0.09, p ≤ 0.001), providing support for this hypothesis.¹⁹

Table 1. Influence of Competitive Brokerage on Firm Performance

	Control Model	Model 1	Model 2
Variables	Performance	Performance	Performance
Competitive Links	0.07**	0.05	0.05*
	(0.03)	(0.03)	(0.03)
Firm Size	0.24***	0.23***	0.24***
	(0.03)	(0.03)	(0.03)
Firm Age	0.02	0.02	0.02
	(0.02)	(0.02)	(0.02)
Market Share	0.06**	0.04	0.03
	(0.03)	(0.03)	(0.03)
Capital Intensity	-0.01	-0.01	-0.01
	(0.01)	(0.01)	(0.01)
CB (Residuals)	0.08***	0.07***	0.08***
	(0.02)	(0.02)	(0.02)
M&A	0.05***	0.04***	0.05***
	(0.01)	(0.01)	(0.01)
Multimarket Contact	0.02	0.01	0.01
	(0.03)	(0.03)	(0.03)
Firm Diversity	0.08***	0.09***	0.09***
	(0.03)	(0.03)	(0.03)
Collaborative Centrality		0.07***	0.10***
		(0.02)	(0.02)
IMC		0.05***	0.10***
		(0.01)	(0.01)
CB		0.09***	0.09***
		(0.02)	(0.02)
Collaborative Centrality X IMC			-0.09***
			(0.02)
R^2	0.43	0.56	0.52
N	1401	1401	1401
Note: (1) Standard errors in parentheses. (2) *** p<0.01, ** p<0.05, * p<0.10. (3) CB is competitive brokerage.

We analyzed Competitive Brokerage to test Hypotheses 2, 3, and 4. We used random effects models to analyze the unbalanced dataset, assuming that each firm has its own systematic baseline and that each intercept “is the result of a random deviation from some mean intercept” [22]. Given that not all firms in the network occupy a competitive brokerage position and some are on the outskirts of the competition network, the distribution of Competitive Brokerage was left truncated (to zero). As truncated distributions can lead to bias in ordinary random effects panel regressions, we used random effects Tobit models. Since the current network positioning may be influenced by previous positioning [106, 116], we added the prior Competitive Brokerage of year t-1 to our analysis. However, using raw values for this may confound past performance, number of competitive links, or firm size with current Competitive Brokerage. We therefore used a two-stage regression procedure to account for such possibilities. In the first stage, we used prior Competitive Brokerage (t-1) as the dependent variable and variables that are theoretically likely to affect Competitive Brokerage as independent variables (see Table D1 in Appendix D). The residual values from the first stage were used as an instrumental variable in the second stage. This ensured that none of the known significant predictors of network positioning (e.g., number of links) were repeated in the second stage regression, preserving the exogenous nature of the model [93]. It also captured the effects of unobservable variables, such as corporate espionage and ex-employees, on Competitive Brokerage.

The results of the second stage regressions of the Competitive Brokerage analysis (random effects Tobit estimates) provide broad support for Hypotheses 2, 3, and 4. Model 1 (β = 0.66, p ≤ 0.001) and Model 2 (β = 0.72, p ≤ 0.001) in Table 2 show that IMC positively influences Competitive Brokerage, supporting Hypothesis 2. Furthermore, Collaborative Centrality positively influences Competitive Brokerage (Model 1: β = 0.59, p ≤ 0.001; Model 2: β = 0.57, p ≤ 0.001) supporting Hypothesis 3. Hypothesis 4, which states that IMC and Collaborative Centrality are substitutes for Competitive Brokerage, is supported by the negative and statistically significant coefficient of the interaction between IMC and Collaborative Centrality (Table 2, Model 2: β = -0.73, p ≤ 0.05). Slope tests (Figure 4) indicate that under conditions of low IMC, an increase in Collaborative Centrality results in an increase in Competitive Brokerage. Similarly, under conditions of low Collaborative Centrality, an increase in IMC results in an increase in Competitive Brokerage. As expected in terms of substitution effects, under high levels of IMC or Collaborative Centrality, increasing Collaborative Centrality or IMC, respectively, reduces Competitive Brokerage.²⁰

Figure 4. Interaction Between Collaborative Centrality and IMC

Table 2. Influence of IMC and Collaborative Centrality on Competitive Brokerage

	Control Model	Model 1	Model 2
Variables	Competitive Brokerage	Competitive Brokerage	Competitive Brokerage
Competitive Links	1.33***	1.34***	1.34***
	(0.30)	(0.30)	(0.30)
Firm Size	0.39	-0.11	-0.12
	(0.33)	(0.34)	(0.34)
Firm Age	0.01	-0.11	-0.10
	(0.24)	(0.24)	(0.24)
Market Share	1.15***	1.03***	1.04***
	(0.27)	(0.26)	(0.26)
Capital Intensity	-0.18*	-0.21**	-0.20**
	(0.10)	(0.10)	(0.10)
CB (Residuals)	0.78***	0.71***	0.73***
	(0.17)	(0.17)	(0.17)
M&A	0.35***	0.31***	0.31***
	(0.11)	(0.11)	(0.11)
Multimarket Contact	0.97***	1.00***	0.99***
	(0.26)	(0.25)	(0.25)
Firm Diversity	3.88***	3.82***	3.80***
	(0.30)	(0.30)	(0.30)
Collaborative Centrality		0.59***	0.57***
		(0.20)	(0.20)
IMC		0.66***	0.72***
		(0.18)	(0.18)
Collaborative Centrality X IMC			-0.73**
			(0.43)
Log Likelihood	-1347.24	-1336.73	-1335.24
N	1,401	1,401	1,401
Note: (1) Standard errors in parentheses. (2) *** p<0.01, ** p<0.05, * p<0.10. (3) CB is competitive brokerage.

The coefficients and signs of the other variables are in line with previous findings, thus increasing our confidence in the results. For example, the direct effects of Collaborative Centrality (β = 0.10, p ≤ 0.001) and IMC (β = 0.10, p ≤ 0.001) on Firm Performance are positive and significant. Similarly, the control variable Firm Size (β = 0.24, p ≤ 0.001) has a positive and significant effect on Firm Performance, and the control variable Market Share (β = 1.04, p ≤ 0.001) has a significant positive effect on Competitive Brokerage. The two-stage regression procedure, one-year lagged independent variables and control variables, Hausman specification tests, extensive control variables, and various robustness tests address endogeneity issues and alternative model specifications [54] (see the section on endogeneity and reverse causality).²¹

Robustness Tests

We conducted several tests to assess the robustness of our results. First, to assess the potential collinearity between competitive links and Competitive Brokerage, we ran all of the models after excluding the Competitive Links variable and obtained similar results. Second, although our tests did not suggest the presence of heteroskedasticity, we used robust standard errors in our estimations and obtained qualitatively similar results. Third, we implemented diagnostics checks involving outliers, influential observations, and normality of residuals and found no issues or violations of assumptions. Fourth, to compensate for potential simultaneity bias and address any potential reverse causality, we tested an alternative model with two-year lags in the independent and control variables [50, 56] and obtained qualitatively similar results (see Table OS.4 in the online supplement). Finally, we used the piecemeal approach to examine the mediating effect of Competitive Brokerage on the relationships between Collaborative Centrality and Firm Performance, and IMC and Firm Performance. Competitive Brokerage partially mediates the relationship between Collaborative Centrality and Firm Performance but does not mediate the relationship between IMC and Firm Performance (see the online supplement section “Mediation Analysis” for details), which supports our theory.

Endogeneity and Reverse Causality

Although our hypothesis that Collaborative Centrality improves Competitive Brokerage is supported, a better Competitive Brokerage position may plausibly lead to Collaborative Centrality.²² Thus, a firm’s Collaborative Centrality may endogenously depend on its Competitive Brokerage and the assumed positive relationship may suffer from reverse causality. The previously described two-stage regression approach enabled us to overcome this challenge. In the first stage, we used prior Competitive Brokerage (t-1) as the dependent variable (see Table D1 in Appendix D). The residual values of the first stage were used in the second stage regression. This enabled us to be confident that none of the important predictors of prior Competitive Brokerage were repeated in the second stage regression, thus maintaining its exogenous nature. In addition, if Collaborative Centrality and Competitive Brokerage are endogenously determined, any positive relationship between them should disappear once the residuals of the first stage regression (CB (Residuals)) are added to the second stage regression (Table 2) [54]. However, the coefficient of Collaborative Centrality remains statistically significant. Nonetheless, to identify any remaining issues related to autocorrelation and endogeneity, we conducted an additional analysis using Arellano and Bond [7, 8] dynamic panel generalized method of moments estimators (see Table OS.5 of the online supplement).²³ The results are in line with our theory and models in Table 2, and Collaborative Centrality is significant after adding Competitive Brokerage to the model. Hence, our results are robust to reverse causality and endogeneity.²⁴

Supplementary Analyses

We conducted four supplementary analyses, as detailed in Appendix E, to either gain further insights or to assess the robustness of our results. Supplementary Analysis One empirically illustrates that in a competition network Competitive Brokerage and structural holes are distinct. Thus, structural holes are not an appropriate proxy for Competitive Brokerage. Supplementary Analysis Two further confirms the robustness of our results to outliers as Competitive Brokerage positively influences Firm Performance (β = 0.097, p ≤ 0.001, R² = 0.166) when Competitive Brokerage is transformed into a binary indicator. Supplementary Analysis Three highlights the empirical differences between Collaborative Centrality and Collaborative Brokerage in the context of our research model. Supplementary Analysis Four shows the robustness of our results across time periods. Specifically, we re-ran our analysis using data from 2004 to 2018 and found qualitatively similar results.

DISCUSSION

Theoretical Contributions

To the best of our knowledge, this study is the first to explore the quantitative effects of IMC in the context of competition and collaboration networks. We contribute to two areas of research, information systems - specifically the literature on information management in the context of collaboration and competition, and strategic management - specially to literature on the influence of collaboration on interfirm competition [33, 55]. Our study offers three main contributions to the IS literature. First, we contribute to the debate over whether IT-enabled capabilities help firms manage the resources and information obtained from collaboration networks [e.g., 32], or if they substitute for specific network structures in collaboration networks [e.g., 26]. Thus, whether IT-enabled capabilities complement or substitute for collaboration networks in the context of competition is still a matter of debate. We resolve this tension by suggesting that IMC can act as a substitute for collaborative centrality in its relationship with competitive brokerage. This implies that both IMC and collaborative centrality similarly decrease information asymmetry in the context of competition, which is a significant contribution to the understanding of the business value of IT. Our study highlights the importance of going beyond dyadic relationships when studying cooperation and competition and considering a firm’s ability to manage information from different types of networks. We thus offer a valuable contribution to the growing body of IS studies that explore how information management influences the relationship patterns between firms and their performance [82, 91].

Second, this study contributes to the literature on IT business value by proposing that IT has effects not only at the firm level [82, 96] but also at the competition network level. Recent studies offer a mono-level perspective and limit their theory, analyses, and implications to firm-level outcomes such as performance (e.g., [67, 98]) and innovation (e.g., [60, 61]). However, they overlook the principles of an embedded economy, which suggest that a firm is embedded in multiple networks and the firm and its networks, co-evolve. We address this gap by proposing that IT has multilevel effects that influence both firm- and network-level outcomes. By recognizing that IMC contributes to the formation of network structures, we open a new research avenue through which IS scholars can explore the effects of IT beyond the boundaries of the firm.

Third, the direct effect of IMC on competitive brokerage emphasizes the benefits of IT-enabled capabilities in competitive strategies. A competitive broker firm can be considered a “polyphagous organization” because it competes well across different markets and advances its position by feeding off information obtained through IMC. We thus contribute to the literature that applies the notion of Red Queen competition to IS phenomena [2, 109]. Competition networks proffer the opportunity to model Red Queen dynamics and our findings corroborate how IT can enable firms to “outrun” their competitors in a Red Queen world.

Our study makes four main contributions to the strategic management literature. We extend investigations into collaboration networks that suggest how previous collaborative relationships can encourage further collaboration between firms [52] by investigating whether collaboration networks can influence competition networks. We approach this question by 1) analyzing the structure of collaboration networks and assessing the influence of interconnectedness within them on competition networks; and 2) by analyzing firms’ abilities to manage information in the context of collaboration and competition. We find that in addition to the resource dependence theory paradigm of partner interactions, in which previous collaborative relationships promote more extensive collaborations between partners, specific positions and structures of collaboration networks can lead to a transition from collaboration to competition, which represents a different type of interaction. In addition, a firm’s approach to information management can alter this transition process.

Second, we enrich the literature on the tension between competition and collaboration in interfirm networks. Firms can obtain information from each other in collaborative relationships [66], but we further identify that the structure of collaboration networks and information management are key mechanisms through which information is transferred in competitive relationships. Hence, the key contribution of this research to the strategic management literature is identifying the information-related mechanisms that underlie the antecedents of competition network structure. Specifically, we make a novel contribution by reconciling the tension between competition and collaboration from an information management perspective. Finally, the notion that networks beget networks is a notable addition to this literature.

Third, we extend the literature on competition networks, as previous studies typically focus on dyadic or single industry competition [25, 42]. We examine all the competitive relationships a firm has in all the product categories it is involved in, to determine its position in a competition network. Our findings suggest that a firm’s position in the competition network, specifically its competitive brokerage position, influences its performance. Competitive brokers are more likely to capture information and signals from other firms in different product categories, which encourages them to thoroughly evaluate the consequences of their competitive actions. They are also likely to control diffusion of competitive actions and direct competitive actions to specific parts of the competition network, as they can analyze competitors’ information and signals and thus predict how other firms may respond to various actions. Competitive brokers can therefore prevent or encourage competition wars to improve their own financial performance.

Fourth, our study suggests that a relational evaluation of competition is important to further our collective understanding [25, 103]. The traditional dyadic perspective in competitive dynamics emphasizes a “one firm perspective” [23, 25], while a relational or network perspective highlights that understanding other firms, their relationships, and market signals is essential before embarking on competitive actions. We extend this argument by suggesting that firms can increase their “relational potential” by obtaining information and signals through collaboration and IT-mediated interactions (e.g., social networking platforms). These can facilitate the planning of competitive actions.

Our use of bidirectional and weighted networks to capture interfirm competition represents a significant methodological innovation. Most methods of assessing competitive asymmetry (e.g., [23, 35]) only consider a single industry or the competitive actions of a particular firm, and often exclude multi-market competition and the actions of other competitors. This limits the range of analysis and misrepresents the structure of competition. We address these shortcomings by developing a novel multi-industry, directed, weighted, and double-linked competition network. This contributes to empirical assessments of competitive asymmetry and answers multiple calls in the literature to improve objective evaluations of asymmetry in competition [35]. We also enhance the managerial understanding of competitors’ actions and reveal new market opportunities by incorporating competition across different network communities. In general, our study provides a foundation for the further development of network-based competition models for academic and managerial use.

Implications for Practice

This study provides various insights for managers. First, they should view participation in collaboration networks as a mechanism that has directly observable benefits such as access to information, but also leads a firm to an advantageous position in its competition network. Second, managers should recognize that information systems enhance a firm’s ability to manage information, such as through asset management software and web analytics, and can help the firm attain a better position in competition networks and shape their structure. Firms participating in industries with low levels of interfirm collaboration should implement such tools. Third, our study reveals that managers can obtain information through either IMC or collaborative centrality to develop their competitive strategies. Although these routes may not provide identical information, either source can help managers fulfill similar objectives as the information will be largely equivalent. For example, they can help predict a competitor’s behavior, understand customer preferences, and needs, and reduce the firm’s information asymmetry with its competitors, the environment, and within the firm.

In sum, collaboration and competition networks are essential for improving performance. A firm at the center of the collaboration network (collaborative centrality) and that develops IMC can become more important in the competition network (competitive brokerage) and thus improve its performance. A firm can choose to focus on either IMC or business collaboration, as they are substitutes, to develop competitive brokerage and create business value.

Implications for Policy

Our evaluation of competition through networks has important implications for policy making. Measuring market power has been a significant concern for policy makers and in the design of antitrust policies. The economics and industrial organization literature has mainly focused on evaluating market power at the industry level (e.g., [89]), although firms often participate in multiple industries simultaneously. Our competitive brokerage measurement assesses the overall competitive ability of a firm across multiple industries and equally considers competitors’ abilities to respond to competitive attacks. Competitive brokerage is therefore a feasible alternative for measuring market power and is particularly relevant given the success of large technology firms such as Amazon and Google that participate across multiple markets. Although competitive brokers do not typically monopolize markets, they may have the ability to effectively compete in multiple markets. They may then potentially abuse their cross-market competitive abilities and launch intense competitive action across multiple markets. Policy makers should be aware of competitive brokers and design adequate measures to prevent any adverse effects from market competition on consumer welfare, market efficiency, and profits.

Limitations

Although this study provides novel insights into the interplay between IMC, collaboration, and competition networks, three limitations merit consideration. First, we only considered specific characteristics of competition networks (competitive brokerage) and collaboration networks (collaborative centrality). Other structural characteristics of collaboration networks, such as their density, may have an influence or co-influence on the structure of competition networks. Though intriguing, we were unable to consider or control for such characteristics and offer these as possibilities for future researchers. Second, we did not delve into the specific influence different types of collaboration may have on competition networks, such as the various effects of supplier–buyer collaboration networks. Although we considered the information benefits accruing from central positions in cooperation and competition networks, firms may obtain information from other sources such as social relationships, industry associations, institutional forums, corporate espionage, or from hiring ex-employees [5]. Such data can be applied in future work to further enrich our understanding. Third, we limited our data collection to the U.S. to provide a clear geographic boundary and thus enable cross-firm comparisons [23]. Although our results should be generalizable to developing economies, recent research suggests that collaboration networks [5] and IT-enabled capabilities [67, 68] differ across economies. The replication of our work in developing economies and in GREAT (growing, rural, eastern, aspirational, transitional) domains [62, 63] can further inform scholars and practitioners about the peculiarities of IT value across contexts.

Future Research

We identify two main avenues for future research. First, this paper represents a stepping stone for the analysis of IS-related phenomena through a methodology involving bipartite and complex networks [88]. Complex networks offer the opportunity to reconsider several IS research paradigms, particularly in the analysis of relational data. For example, researchers represent collaboration and competition relationships as networks [26, 103], but most studies apply simple or unipartite networks, which only capture the existence of relationships. This simplified representation can be problematic as it does not consider the intensity or origin of the relationships and thus, measures such as centrality can be overestimated [74]. Bipartite weighted networks, which are those between two types of entities (e.g., firms and product categories in our competition network), represent the intensity of interactions via weighted links and thus address the shortcomings of the simplified network models currently used in IS research. As suggested by Guillaume and Latapy [49], “ … all complex networks have an underlying bipartite structure. This makes it possible to view their main properties as consequences of this underlying structure. This also leads to two very efficient models for the generation of complex networks having realistic properties.” Emerging technologies such as blockchain, the Internet of Things, and artificial intelligence rely on relationships between different blocks, devices, or systems, respectively. Thus, future research can use bipartite weighted networks not only to analyze interfirm relationships but also to comprehend the organizational outcomes of emerging technologies.

Second, although the positive effects of collaboration networks on competition networks are being gradually comprehended, our finding that IMC and collaborative centrality act as substitutes in the relationship with competitive brokerage may require further clarification. Whether the interactions between collaboration networks and other information-related capabilities have different effects on competition networks in certain circumstances is another avenue for future research. Similarly, in our additional analysis, we find a positive interaction effect of IMC and competitive brokerage on firm performance. Although this is beyond the scope of this manuscript, exploring such complementary relationships are exciting opportunities for future work.

Finally, we hope that our study inspires further investigations into the multilevel effects of IT and the multilevel determinants of firm performance, following our finding that IMC influences competition networks (network-level outcome) and firm performance (firm-level outcome).

CONCLUSIONS

In this study, we integrate the two parallel perspectives of firm performance in the contexts of competition networks and IT. We reconcile these streams of literature by proposing a model that includes a comprehensive competition network construct (competitive brokerage) and an IT-focused construct (IMC) in conjunction with collaborative centrality, thereby providing a more complete picture of the multilevel effects of IT and the multilevel drivers of firm performance. We offer three key takeaways for researchers and managers. First, given the complexity of competition and the prevalence of IS in most industries, future evaluations of firm performance cannot ignore the contemporaneous effects of competition networks and IT. Our results indicate that a firm may not obtain the full benefits of IT if it neglects its position in competition networks. Thus, constructs derived from competition networks can offer fertile research opportunities. Second, under certain circumstances (e.g., when firms have a high level of collaborative centrality), investing more in IT does not ultimately improve performance. However, such investment may be a viable alternative to other types of strategic investment (e.g., a firm can become a competitive broker by investing in IMC rather than collaboration). Finally, our methodological approach can encourage researchers to take a network-based perspective when examining constructs from multiple theoretical levels. Such an approach in the study of emerging phenomena can represent a new path of enquiry in IS research and management practice.

Acknowledgements

This study was partially supported by the Nanyang Technological University, Nanyang Business School, Start-up Grant (SUG/FY2016/ A.R.Mariana Giov) awarded to M.G. Andrade-Rojas. A. Kathuria would like to thank Actuate Business Consulting for partial financial support for this study. This study was also partially supported by a grant awarded to A. Kathuria by the University of Hong Kong Seed Fund for Basic Research (#201211159161) and by a fellowship awarded to A. Kathuria by the Srini Raju Centre for IT and the Networked Economy at the Indian School of Business. We acknowledge the feedback received at the 2014 Annual Meeting of the Academy of Management, where an abridged preliminary version of this study appeared in the Best Paper Proceedings and received the Best Student Paper Runner-Up Award in the Organizational Communications and Information Systems Division. We also thank participants at seminars at the University of Georgia, Temple University, and the University of Minnesota for helpful comments on previous versions of this paper. Any errors that might remain are our own.

Additional information

Notes on contributors

Mariana Giovanna Andrade-Rojas

Mariana Giovanna Andrade-Rojas is an Assistant Professor of Management Information Systems at the Terry College of Business at the University of Georgia. She holds a Ph.D. in Innovation and Information Management from the University of Hong Kong. Prior to pursuing her Ph.D. studies, she worked in consulting, marketing, and banking in different geographical regions ranging from The Americas to Asia. Dr. Andrade-Rojas’ research interests span five main areas: complex networks, IT business value, the digital economy, competitive dynamics, and data analytics. Her research has been published in such journals as Information Systems Research, Research Policy, Journal of the Association for Information Systems, and several leading conference proceedings. Her work has received several awards from premier international conferences, such as the Academy of Management Meeting and the International Conference on Information Systems.

Abhishek Kathuria

Abhishek Kathuria is an Assistant Professor of Information Systems and BAT Research Fellow at the Indian School of Business. He received his Ph.D. in Business Administration from the Goizueta Business School at Emory University. His research examines the business value of IT, focusing on innovation, digital platform strategies, and emerging economies. His work has been published in journals such as Journal of Management Information Systems, Information Systems Research, Journal of the Association for Information Systems, Communications of the Association for Information Systems, among others, and received multiple best paper nominations and awards at various academic conferences. Dr. Kathuria is an advisor to, and co-founder of multiple startups and consults on business transformation, organizational turnarounds, and IT strategy with public and private corporations in GREAT domains such as India, China, and the Middle East.

Benn R. Konsynski

Benn R. Konsynski is the George S. Craft Distinguished University Professor of Information Systems and Operations Management at the Goizueta Business School at Emory University. He holds a Ph.D. in Computer Science from Purdue University. Dr. Konsynski has held faculty positions at the University of Arizona and Harvard Business School, and served as adviser and board member of public and private corporations. He was also named Baxter Research Fellow at Harvard and Hewlett Fellow at The Carter Center. He specializes in issues of digital commerce and information technology in relationships across organizations, and has published in such diverse journals as Journal of Management Information Systems, MIS Quarterly, Communications of the ACM, Harvard Business Review, Data Communications, Decision Sciences, Information Systems Research, IEEE Transactions on Software Engineering, and others.

Notes

1. The relationships between other types of interfirm integration such as mergers and acquisitions and IT have been explored. For example, the importance of IT integration between acquirers and targets, and how IT infrastructure facilitates the development of post-merger IT integration capability, which improves post-acquisition performance, have been established [10].

2. Information asymmetry is defined as the difference between information possessed by two parties. Thus, we refer to information asymmetry as the difference between information possessed by a focal firm and other firms regarding its partners, customers, and environment.

3. In networks, links or relationships often consist of valued relationships in which the strength or intensity is recorded [102]. In a competition network, each competitive relationship or link carries a value, which is the strength of the competitive link. Further details are provided in the network construction portion of the Methodology section.

4. Market information refers to “data concerned with a firm’s current and potential external stakeholders” [77]. Some market information circulates in product markets, which is obtained by firms through market participation and observation of competitor behavior.

5. Competitive brokers have better access to market information, providing them with an advantage over firms in other positions in the network, such as firm U (see Figure 2). This advantage does not preclude non-competitive broker firms (e.g., firm U) from monitoring and actively conducting research about other firms (e.g., firm Y). However, firms have selective attention and tend to focus on the public actions of and information about their direct competitors [103]. For example, firm U is likely to focus on the public actions of firms T and X. Although not impossible, it is improbable that firm U will monitor the actions of firm Y as they are not direct competitors, and their markets are disconnected. In addition, due to selective attention and its position in the network, firm U will require increased resources to monitor firms that are not direct competitors. This already puts firm U at a disadvantage compared to those in competitive brokerage positions (e.g., firm Z). Thus, the extent to which this rare case could influence our arguments and analysis is likely to be limited.

6. Resources and capabilities are distinct drivers of firm performance. While resources are stocks of available factors of production owned or controlled by a firm [93], capabilities refer to a firm’s ability to deploy these resources to achieve a specific goal [80]. Ergo, capabilities are embedded within firms and offer a durable means of creating economic rent. In the case of IMC, a firm deploys its IT resources to provide data information, and connectivity [82]. For example, a firm can utilize IMC to leverage the IT resources of data crawling and business analytics systems to monitor competitors’ websites and use the resultant information to plan competitive actions. Detailed discussions of the relationship between IT resources and capabilities are provided by Ravichandran and Lertwongsatien [93] and Bharadwaj [11].

7. An example of an IT system that improves internal information sharing and communication frequency is that of Procter & Gamble, which installed over 40 “Video Collaboration Studios” [28], enabling employees in different locations to collaborate and socialize.

8. Although collaboration networks provide both information and resources to firms (e.g., shared facilities and intellectual property), in this study we focus on the information component. Obtaining and exchanging information are important elements in collaboration networks and essential for firms to compete.

9. Information shared within collaborative relationships can be applied outside the relationship when contractual terms do not constrain or guide the use of information-based assets [33, 66]. Firms can use such information to improve or develop their own products. In addition, the focal firm can benefit from information spillovers generated by the partnership and develop its own additional skills and knowledge to compete in markets. Information spillovers cannot be perfectly controlled by contracts, because they consist of learning that often involves new and unexpected outcomes. These outcomes may include product development in new categories, which may not be directly related to the products developed between the focal firm and its partners.

10. Firms in a collaborative brokerage position bridge “holes” in a collaboration network, implying that they bridge/connect firms that do not collaborate with one another. Collaborative brokerage enables firms to be integrated in different clusters and to mediate the knowledge transmitted between the clusters in the collaboration network [21]. A detailed explanation of the differences between collaborative brokerage and collaborative centrality is provided in Appendix E and in the “Supplementary Analysis Three” section of the online supplement.

11. For example, in 2015, Unilever launched its platform “The Unilever Foundry” to engage with innovators, designers, and customers with novel ideas for new products. This platform enables information sharing between customers and Unilever, and the firm gains knowledge about customer needs and preferences and can launch products to the market faster.

12. Internally, collaborative centrality promotes information coordination activities because awareness of resources is necessary for coordination with partners [3]. Externally, collaboration networks facilitate the exchange of information with partners.

13. Internally, IMC helps synchronize and integrate information by increasing awareness of information and of the resources developed and stocked across the firm [82]. Externally, IMC enables a firm to generate, collect, and interpret information from competitors, customers, and the environment [6].

14. Passport is an online market research tool developed by Euromonitor [57], which offers a detailed overview of product distribution and market share across multiple countries. Passport includes product information from large-scale groceries, direct sellers, local outlets, and discount stores. Regular store checks in local markets are conducted to ensure that their insights are valid and updated. In addition, numerous experts and market researchers conduct exhaustive audits and data cross-referencing to ensure the accuracy and reliability of the data.

15. The industries are beauty and personal care, alcoholic drinks, apparel, appliances, consumer electronics, consumer food service, consumer health and wellness, beverages, pet care, tissue and hygiene, and tobacco. We selected these because their commercial products satisfy consumers’ most basic daily needs, and adult consumers do not face any serious regulatory restrictions when purchasing these products.

16. We took additional steps to alleviate any possible media bias concerns due to under-reporting or over-reporting. First, we screened the firms’ websites and annual reports to collect IT announcements. Second, we included additional PR announcements and press releases that may not have been reported by the media. Third, we included the firms’ 10-K filings for any additional IT news.

17. In competition networks, weighted betweenness centrality is a more adequate measure of brokerage due to the bipartite origin of competition networks. Bipartite origin refers to how two firms have a competitive relationship if they participate in the same product categories. Opsahl, Agneessens, and Skvoretz [86] suggest that in networks of bipartite origin, such as competition networks, structural holes are over-calculated because this measure often does not consider the strength and directionality of both outgoing and incoming links.

18. We chose Freeman’s degree centrality as the measure of collaborative centrality as it has been used in other studies to assess firms’ access to information and opportunities [39]. A higher degree of centrality is also associated with better access to partners’ information [3], and degree centrality is used to assess networks as sources of external information. Finally, degree centrality is shown to be related to a steady flow of information [39], entry into new markets [58], and gathering and generating information [3].

19. We conducted additional robustness tests by assessing the impact of Competitive Brokerage on other measures of Firm Performance, such as market value, gross revenue, and sales per employee. The qualitatively similar results are shown in Table OS.7 in the online supplement.

20. For the slope tests, the sample was split into two subsamples based on the value of the first covariate (IMC or Collaborative Centrality), corresponding to low (one standard deviation below the mean) and high (one standard deviation above the mean). The effect of the second covariate (Collaborative Centrality or IMC, respectively) on Competitive Brokerage was estimated at both levels.

21. To minimize collinearity, mean-centered interaction terms were used in all of the regression models. The range of values for variance inflation factors for all models was 1.07 to 2.91, well within the conservative threshold of 3.3, suggesting that multicollinearity is not problematic.

22. There are two plausible arguments for how better Competitive Brokerage position may lead to Collaborative Centrality. First, firms in a Competitive Brokerage position have better access to information and therefore pay attention to opportunities that rivals exploit, as these actions could harm their competitive brokerage position. Such firms may try to match their competitors’ actions by forming collaborative relationships that either complement their resources or meet the competitive threat [42], resulting in higher Collaborative Centrality. Second, firms in competitive brokerage positions may face the same competitors across multiple markets. Mutual forbearance research suggests that an increase in multimarket overlap is related to an increase in cooperation between firms [101], resulting in higher Collaborative Centrality.

23. The results in Table OS.5 in the online supplement demonstrate that the lagged value of Competitive Brokerage is positive and significant (β = 0.54, p ≤ 0.001) and the main effect of Collaborative Centrality is also positive and significant (β = 0.04, p ≤ 0.05). Finally, the interaction between Collaborative Centrality and IMC is negative and statistically significant (β = -0.23, p ≤ 0.05).

24. In addition to addressing endogeneity concerns by using a two-stage procedure [54] and a dynamic panel generalized method of moments estimator [7], we ran our models with different time lags to ensure the directionality of the effects and address reverse causality [56]. The results remain qualitatively similar when using a two-year lag (see Table OS.4 in the online supplement), thus addressing directionality concerns. We conclude that Collaborative Centrality has a positive relationship with firms’ Competitive Brokerage.

Appendix A:

Network Construction

To build the competition network, we first identified the parent companies of the firms listed in Passport. We then linked the parent companies and the product categories in which these companies participated to build annual bipartite networks. We then added weights to the links between companies and product categories to represent the number of products that the company had in a specific product category (e.g., Firm D had 10 products in product category No. 1, Figure A1, A) [74]. Finally, we transformed the bipartite networks into firm-to-firm competition networks by using a one mode projection [85]. In the resulting network (Figure A1, B), a competitive relationship between two firms is represented by two directed links: one incoming and one outgoing. Each link has a weight, which represents the total number of products that a firm uses to compete against another firm across the 11 industries. For example, the competitive relationship between firm D and firm F is depicted by a double link, $D_{\to 13}^{15\leftarrow }F$. The combination of double links and weights to represent the competitive relationships between firms captures the different intensities with which firms compete against each other, which is also known as competitive asymmetry [23]. We followed the same process to create yearly networks from 2004 to 2011. The annual competition networks have an average of 1,084 companies and 37,359 competitive links. We ensured that firms in our competition networks did not have any collaborative relationships with each other. Therefore, our competition networks solely represent the competitive relationships between firms. Further details regarding the construction of the competition network are provided in the “Competition Network Construction” section of the online supplement.

Figure A1. Competition Network Construction and Visualization

Note: (1) Figure A1A is a bipartite weighted and directed network, where 1, 2, and 3 represent product categories, and D, E, F, and G represent parent firms. (2) Figure A1B is a weighted directed network, where F, D, E, and G represent firms linked in competitive relationships. Each relationship is formed by two weighted links and weights on the links reflect the intensity of competition.

The collaboration networks were constructed by creating ego networks and adjacency matrices for each firm each year. We used a 3-year moving window to deal with the often-underreported data on collaboration termination [26, 75]. The 3-year moving window is a widely used conservative approach that assumes that the average duration of a collaborative relationship is 3 years. For example, the adjacency matrix of 2011 includes collaborative relationships formed between January 1, 2009 and December 31, 2011. The matrix for 2010 includes collaborative relationships formed between January 1, 2008 and December 31, 2010. In each 3-year period, the firms in our sample maintained an average of 4.5 collaborative relationships, a minimum of 0 relationships, and a maximum of 114.

Appendix B:

Calculation of Competitive Brokerage

Weighted betweenness centrality was used to calculate Competitive Brokerage. We refined this betweenness measure to accurately represent the ability of firms to capture information and control the diffusion of competitive actions by taking into account the weights and lengths of each competitive link between two firms and identifying the shortest path between them [86]. A shortest path is defined as “a path (link) between two vertices (nodes) such that no shorter path exists …, is thus the shortest network distance between the vertices (nodes) in question” [85]. In a competition network, the nodes are firms and paths are the competitive links between them. Considering these features in the competition network, we used the short path identification algorithm proposed by Opsahl, Agneessens, and Skvoretz [86], which includes the weights and lengths of the links. The algorithm is represented by $d^{w\alpha }\left(i,j\right)=\mathrm{m}\mathrm{i}\mathrm{n}\left(1/{(w_{ih})}^{\alpha }+\dots +1/{(w_{hj})}^{\alpha }\right)$, where i and j represent firms, w is the weight of path w_ih, and α is the turning parameter used to calculate weighted betweenness centrality. As proposed by Opsahl, Agneessens, and Skvoretz [86], we used α = 0.50, such that a link’s directionality, weight, and length were equally important for identifying the shortest paths. As every competitive relationship in the network is represented by two links (an incoming link and an outgoing link, see Figure A1A), the same process was followed twice to identify the shortest paths. After we identified the shortest paths in the competition network, we used weighted betweenness centrality to calculate Competitive Brokerage as $C_b^{w\alpha }\left(i\right)=g_{jk}^{w\alpha }\left(i\right)/g_{jk}^{w\alpha }$, where $g_{jk}^{w\alpha }\left(i\right)$ is the number of weighted shortest paths that go through firm $i$, and $g_{jk}^{w\alpha }$ is the total number of weighted shortest paths between two firms. In summary, a firm with a higher value of Competitive Brokerage intersects a higher number of shortest paths with higher weights.

Appendix C:

Control Variables

Eight control variables were included in our model. First, firms that face multiple competitors are more likely to occupy a competitive brokerage position, and may participate in multiple product categories, thus benefitting from economies of scale and scope. Hence, we included a variable (Competitive Links), which equals the logarithm of the total number of outgoing competitive links that a firm has in each year. Second, a more diversified product portfolio helps firms to bridge competitive relationships because diversified firms are likely to overlap with more firms through product competition. We compute the diversity variable using the Herfindahl diversity measure [83]: $Diversit{y}_{H,it}=1-\hspace{0.17em}\sum j{M}_{ijt}^2/(\sum j{M}_{ijt}{)}^2$, where M_ijt is the proportion of products of firm i on product category j in year t. Thus, Firm Diversity represents the number of products that a firm has in a specific product category divided by the total number of products that it has in all product categories in a given year. Third, when firms encounter the same rivals in multiple markets, their competitive behavior may be different from that of single-point rivals, such that firms that face market overlaps may reduce competitive intensity among themselves [43]. Thus, we included a variable for multimarket competition (Multimarket Contact), computed as the average number of multimarket contacts with all of the focal-market rivals [43]. Fourth, firms that engage in mergers and acquisitions (M&A) may experience a rapid increase in the number of products that they have in the markets, which may influence their competitive brokerage position. Hence, we introduced the variable M&A, which is the sum of M&A deals that each firm completed in each year. Fifth, to control for size effects, we included a variable Firm Size, which is calculated as the logarithm of the number of employees. Sixth, as older firms may be more likely to bridge the relationship between multiple competitors, we used the number of calendar years since the establishment of the company as the variable Firm Age. Seventh, as market share indicates product performance in the market, and consumers may perceive market share as a signal of product quality [65], we included a variable Market Share. We adapted the following market share measurement [12]: Weighted Market Share = ${\sum }^{M}siPi$, where Msi is a firm’s market share in each of its industries i, and Pi is the proportion of the firm’s sales in the industry. Finally, we included industry-level capital intensity because firms participating in industries with distinct capital intensities are likely to rely on different mechanisms to compete in the markets, such as cost, quality, design, and features [34]. We calculated the variable for industry capital intensity (Capital Intensity) as the net book value of plant and equipment/revenues [53].

Appendix D:

First Stage Regression Results

Table D1. First Stage Regression for Prior Competitive Brokerage Position

	Prior Competitive Brokerage
Variables	Stage 1
Competitive Links	0.04***
	(0.00)
Firm Size	-0.18*
	(0.07)
R&D Expenses	0.24***
	(0.04)
Market Share	0.49***
	(0.06)
Log-Likelihood	-2490.1945
N	1401
Note: (1) Standard errors in parentheses. (2) *** p<0.01, ** p<0.05, * p<0.10.

Appendix E:

Supplementary Analyses

Supplementary Analysis One empirically illustrates the differences between Competitive Brokerage and structural holes. A structural hole exists in a competition network when there is no direct competitive link between two firms [19]. Although both Competitive Brokerage and structural holes rely on the concepts of bridging ties and lack of direct links between two firms, Competitive Brokerage captures the strength (weight) and directionality of competitive links, while structural holes mainly capture the existence or absence of links [86]. We measured structural holes by using the Burt constraint [20] measure, which assesses a firm’s lack of access to structural holes. Burt constraint [20] is C_i= (p_ij+ p_iqp_qj)², where q≠ i, j where p_j is the proportion of firm i’s competitive links invested in firm j, p_ij =z_ij/Σ_qz_iq, and the binary variable z_ij measures the absence (0) or presence (1) of a competitive link between firms i and j. Thus, we followed Zaheer and Bell [115] and calculated structural holes as 1 minus the firm’s constraint score, where constraint was non-zero and zero for all other cases. When using structural holes as the dependent variable, Collaborative Centrality and its interaction with IMC are not statistically significant (β = -0.05, p = 0.468) (Model 2 of Table OS.3 in the online supplement). In addition, structural holes have a negative and significant relationship with Firm Performance (β = -0.11, p ≤ 0.001) (Model 5 in Table OS.3). These results are consistent with previous studies (e.g., Ahuja, 2000a), which demonstrate a negative relationship of structural holes with organizational outcomes. As highlighted by Burt [20], conflicting results have been obtained, because the effects of structural holes depend on the network context and network construction. Opsahl, Agneessens, and Skvoretz [86] suggest that in networks of bipartite origin, such as competition networks, structural holes are over-calculated, as this measure does not consider the strength and directionality of the links. This could be a potential reason for the conflicting results found in the literature. In conclusion, our empirical findings support our argument that in a competition network, Competitive Brokerage and structural holes are distinct and hence structural holes are not an appropriate proxy for Competitive Brokerage. This supplementary analysis also contributes to the competition networks literature because it suggests that the strength and directionality of competitive links play an important role in reflecting competition networks. Firms do not compete against each other with the same intensity and representing this intensity in the competition network accurately reveals theoretical relationships, such as that between collaboration and competition.

In Supplementary Analysis Two, we assessed the robustness of our results to outliers by transforming Competitive Brokerage into a binary indicator: firms that are (1) and firms that are not (0) in a Competitive Brokerage position. This ensured that the most important information was retained while avoiding any effects of outliers. We lagged Competitive Brokerage by 1 year and used a panel regression with fixed effects. The results are consistent with our conceptualization and demonstrate that Competitive Brokerage positively influences Firm Performance (β = 0.097, p ≤ 0.001, R² = 0.166).

Supplementary Analysis Three highlights the empirical differences between Collaborative Centrality and Collaborative Brokerage. As shown in the online supplement, Table OS.6 Model 1 (β = 0.11, p = 0.23) and Model 2 (β = 0.12, p = 0.20), Collaborative Brokerage is not significantly related to Firm Performance and the interaction between IMC and Collaborative Brokerage is not statistically significant (β = 0.01, p = 0.42). These results demonstrate that Collaborative Centrality is the appropriate measure to represent our conceptualization of collaboration resulting in access to partners’ information. This analysis also contributes to our reference literature as the finding that Collaborative Brokerage does not influence Firm Performance addresses a major issue in collaboration networks studies. Traditional arguments suggest that collaborative brokers control the exchange of information in collaboration networks and have a competitive advantage [115]. Previous studies assume that information control is more important than information access when competing in product markets. However, our findings suggest that when competing in product markets, information access plays a more important role than information control, because it reduces the firm’s information asymmetry. Further details on this analysis are provided in the online supplement.

Supplementary Analysis Four shows the robustness of our results across time periods. We re-ran our analysis using data from 2004 to 2018 and found qualitatively similar results, as shown in Table OS.11 and OS.12 in the online supplement. The Passport database does not offer complete data of the 11 industries in our competition network before 2004 and after 2011; therefore, we limited our main analysis to the 2004–2011 period. Despite the data limitations, we collected competition data for the 2012–2018 period. Although the 2012–2018 competition networks do not extend across the same industries as those in the 2004–2011 period, from a specific industry perspective, these networks are complete because competition data are complete at the industry level. Thus, we used the 2012–2018 competition networks to evaluate the competitive brokerage positions of firms. The qualitatively similar results indicate that using the 2004–2011 period in our main analysis is neither a concern nor a limitation that affects the generalizability of our results.