University Repository An empirical investigation of Network-Oriented Behaviors in Business-to-Business Markets

Thisstudy isconcerned with theextent to which network-orientedbehaviors directlyand/or indirectly affect ﬁ rm performance. It argues that a ﬁ rm's interaction behaviors in relation to an embedded network structure are key mechanisms that facilitate the development of important organizational capabilities in dealing with business partners. Such network-oriented behaviors, which are aimed at affecting the position of a company in the network, are consequently important drivers of ﬁ rm performance, rather than the network structure alone. We develop a conceptual model that captures network-oriented behaviors as a driving force of ﬁ rm performance in relation to three other key organizational behaviors, i.e., customer-oriented, competitor-oriented and relationship-oriented behaviors. We test the hypothesized model using a dataset of 354 responses collected via an on-line questionnaire from UK managers, whose organizations operate in business-to-business markets in either the manufacturing or services sectors. This study provides four key ﬁ ndings. First, a ﬁ rm's network-oriented behaviors positively affect the development of customer-oriented and competitor-oriented behaviors. Secondly, they also foster relationship coordination with its important business partners within the network. Thirdly, the effective management of the ﬁ rm's portfolio of relationships is found to mediate the positive impact of network-oriented behaviors on ﬁ rm pro ﬁ tability. Lastly, closenessto end-users ampli ﬁ es the positive effect of network-oriented


Introduction
From a focal firm's perspective, its business relationships are some of the most important sources of competitive advantage. They provide combinations of resources embedded in these relationships, which are unique and difficult to imitate by the competition (Gulati, Nohria, & Zaheer, 2000;Spector, 2006;Zaefarian, Henneberg, & Naudé, 2011). This has an important implication for firms operating in business-tobusiness markets, since they need to develop strategies for collaborating with both their customers and suppliers within the business network (Day, 2000). However, overly relying on established relationships and overlooking the critical aspect of introducing new relationships might lead to a lack of novel information and the resources needed for innovation success (Uzzi, 1996(Uzzi, , 1997. Therefore, a firm's ability to change the formation of its relationship portfolio in response to changes in the wider business network has strategic implications for its performance (Baum, Cowan, & Jonard, 2014;Cui & O'Connor, 2012;Gulati et al., 2000).
It is evident that business networks have a profound impact on firms' performance (Jack, 2005;Uzzi, 1996). Although the causal link between a focal firm's network position in the context of its portfolio of business relationships, and its performance, has been researched from a structural perspective, empirical evidence on this link with regard to behavioral issues is still missing (Baum et al., 2014). Salancik (1995) suggests that the fact that relationships and interactions are taken as given in network analyses might have contributed to the lack of behavioral research. Based on the resource dependence theory (Pfeffer & Salancik, 1978), we argue that there is a need to study this pivotal causal relationship from a firm's behavioral perspective. Firms have the ability to proactively seek the requisite resources through which they can potentially change their relationship portfolio, and with it, their position in the network, by managing their interactions and business relationships (Johanson & Mattsson, 1992;Salancik, 1995;Stevenson & Greenberg, 2000). On the other hand, firms' behaviors are also shaped by their web of relationships, which constitute the network structure (Granovetter, 1985;Rivera, Soderstrom, & Uzzi, 2010). Firms embedded in the network are all assumed to be "perceiving and opportunity-seeking actors" (Kilduff & Krackhardt, 1994, p. 88) in the sense that their actions are based on their perception of their surroundings and their intention to sense and seize opportunities afforded by the network. However, the way in which a firm responds to other actors changes the dynamics of the network (Kilduff & Krackhardt, 1994;Porac, Thomas, & Baden-Fuller, 2011). Firms attempt to shape their networked environment by changing the pattern of their interactions with their counterparts in order to grasp the network dynamics and further capitalize on these dynamics based on their understanding of the network (Andersson & Mattsson, 2010). The bilateral influences between a focal firm and its business network are an ongoing interactive process, manifested in the interactions between the firm and its counterparts, which are either directly or indirectly connected to it (Håkansson & Ford, 2002).
In this context it is important to consider that from a strategic perspective, firms interact differently within their business relationships, in that they have different behavioral options open to them. They can actively shape the network through strong-or weak-tie relationships based on the anticipated business outcomes (Thornton, Henneberg, & Naudé, 2013). However, the resulting interaction behaviors do not necessarily contribute to firm performance directly, as the outcomes of such acts cannot be foreseen (Ford, Gadde, Håkansson, & Snehota, 2003;Thornton et al., 2013). In addition, firms can reactively sense network dynamics, which can be seen as part of a firm's ability to respond to the network. This set of sensing behaviors relate to learning from, and utilizing the environment in which its important counterparts are embedded (Ford & Mouzas, 2013).
Following this argument, we infer that a firm's interaction behaviors in relation to an embedding network structure are key mechanisms that facilitate the development of important organizational capabilities in dealing with its business partners. Such network-oriented behaviors (Thornton et al., 2013) are consequently important drivers of firm performance, rather than the network structure alone (Salancik, 1995). This proposition provides the starting point and research objective for our study: it is concerned with the extent to which network-oriented behaviors directly or indirectly affect firm performance. Building on the existing literature of network theory and business-to-business marketing, this research contributes to the literature in two ways. First, it conceptualizes and validates a nomological model in which networkoriented behaviors are hypothesized as the drivers of other important firm behaviors toward their important counterparts in the network, such as customer-oriented, competitor-oriented and relationshiporiented behaviors. This is important as it enhances our understanding of how different organizational behaviors oriented toward different aims are interacting (Day, 1994). Secondly, this study establishes the role of firms' network-oriented behaviors in driving firm performance from a behavioral perspective. This is important as it directly provides managerial guidance about which behaviors in response to the wider business network firms should focus on in order to optimally sense the network dynamics and seize the opportunities (Gulati et al., 2000).
This study aims to provide a conceptual model that outlines how a firm can utilize network-oriented behaviors to understand its customers and competitors, and coordinate with its important business partners within the network. This framework provides an explanation as to how these strategic behaviors contribute to firm performance, either directly or indirectly. The argument will develop as follows. First, through a concise review of the relevant literature, we develop a conceptual model that captures network-oriented behaviors as a driving force of firm performance in relation to three other key organizational behaviors (i.e., customer-oriented, competitor-oriented and relationship-oriented behaviors). Secondly, we outline our empirical research design and test the hypothesized model using a dataset of 354 responses collected from UK managers, whose organizations operate in business-tobusiness markets in either the manufacturing or services sectors. Lastly, we conclude with a discussion of theoretical and practical implications of the study, acknowledge the limitations, and provide directions for future research.

Business interactions and network-oriented behaviors
Firms are unavoidably embedded in business networks (Ford et al., 2003). The general consensus in the network literature is that networks have some properties that allow firms to achieve certain economic outcomes, such as faster knowledge transfer and more effective resource utilization (Achrol & Kotler, 1999;Granovetter, 2005;Jack, 2005). This is done by mobilizing other actors, such as customers or suppliers, in the network (Mouzas & Naudé, 2007;Zaefarian et al., 2011). Furthermore, such mobilizing activities, i.e., interacting and building relationships with business partners, are linked to specific behaviors by a firm and thus economic outcomes are influenced by the way in which firms interact with others (Granovetter, 1985). We conceptualize such interaction behaviors as network-oriented behaviors. They are derived from the need of a firm to sense its position in the network (i.e., the opportunities and threats associated with its direct and indirect business relationships) and seize the opportunities derived from this position accordingly (Thornton et al., 2013). Thorelli (1986) suggests that one of the key issues related to such 'networking' is the way in which a firm positions itself in the network by changing its portfolio of relationships. Therefore, it can be assumed that firms' ability to maneuver themselves in the networks differ, depending on how well they use network-oriented behaviors, and that such differences will help generate insights related to firm performance differences (Zaheer & Bell, 2005).
Most studies in the existing network literature adopt structural network measures, such as the centrality and density of a focal firm's network, to understand firm performance differences (e.g., Hagedoorn, Roijakkers, & Van Kranenburg, 2006;Hendry & Brown, 2006;Zaheer & Bell, 2005). Empirical research that focuses on firms' behaviors toward their networks is still scarce (Ford & Mouzas, 2013). There are, however, some studies that discuss network-oriented behaviors, e.g., under the name of organizational networking (Thornton et al., 2013), network competence (Ritter, 1999;Ritter & Gemünden, 2003) or business networking (Ford & Mouzas, 2013;Ford et al., 2003). These studies broadly adopt an Industrial Network Approach, which allows researchers to understand how firms interact with others in order to cope with organizational problems at hand by utilizing 'external' resources and reconfiguring the combination of them (Håkansson, 1982;Håkansson & Snehota, 1989). Such behaviors have also been conceptualized as actions taken by a firm to change the formation of its network in favor of its business aims (Smith & Laage-Hellman, 1992). Initiating, maintaining and terminating relationships as part of a portfolio approach have been identified as important capabilities that enable firms to effectively form a pool of accessible resources that are embedded in their relationship portfolio (Cui & O'Connor, 2012;Mitrega, Forkmann, Ramos, & Henneberg, 2012;Zaefarian et al., 2011). In this context Thornton et al. (2013) conceptualize organizational networking as four sets of anticipated outcome-driven behaviors, specifically information acquisition, opportunity enabling, strong-tie resource mobilization and weak-tie resource mobilization. Each of these dimensions reflects manifested behaviors, which capture a distinct way in which firms utilize their relationships in an attempt to achieve their anticipated goals. First, information acquisition refers to a firm's tendency to use both strong-tie and weak-tie relationships in order to obtain desired information for making informed decisions. Secondly, opportunity enabling relates to a firm's conscious acts to sense the opportunities by strategically interacting with relevant parties in its network. Thirdly, strong-tie resource mobilization is utilized by a firm to adjust, transfer and pool resources across various established relationships in order to address certain firm challenges. Finally, weak-tie resource mobilization refers to the ability to mobilize resources that are linked to firms' less established relationships.
This conceptualization is in line with our research objective of conceptualizing the way in which firms interact with their embedding network. We therefore use the four behavioral dimensions by Thornton et al. (2013) to conceptualize network-oriented behaviors related to a focal firm's business relationships, be they direct or indirect.

Nomological model development
The extant literature has established the role of different organizational behaviors for driving firm performance; in this context marketoriented behaviors have been identified as key drivers of favorable firm performance (Jaworski & Kohli, 1993;Narver & Slater, 1990). Crucial aspects of market-oriented behavior are a customer and a competitor orientation. However, empirical evidence also suggests that these behaviors are moderated by contextual factors (e.g., Cadogan, Kuivalainen, & Sundqvist, 2009;Ellis, 2006;Kirca, Jayachandran, & Bearden, 2005), or have no influence on firm performance (e.g., Grewal & Tansuhaj, 2001). A longitudinal study by Kumar, Jones, Venkatesan, and Leone (2011) provides evidence that market oriented-behaviors have recently become a prerequisite, rather than a competitive advantage, for any firm to compete in the market place. Furthermore, being market-oriented could be a mere 'self-portrait' rather than a true representation of a market-centric approach (Deshpandé, Farley, & Webster, 1993).
Besides market-oriented behaviors, relationship-oriented behaviors are also seen as key drivers of firm performance in the business-tobusiness marketing literature (e.g., Morgan & Hunt, 1994;Palmatier, Dant, & Grewal, 2007;Palmatier, Scheer, Evans, & Arnold, 2008). This is backed by a stream of research on business relationships based on the Industrial Network Approach which focuses on the interconnectedness of the business relationships within which a firm is embedded (e.g., Ritter & Gemünden, 2003;Walter, Auer, & Ritter, 2006). Similarly, in the literature on strategic management, the main focus has moved from the resource-based view to a relationship-centric approach of business relationship management. However, in addition to this movement from a monadic firm-centered view to a dyadic relationshipcentered perspective, Vargo and Lusch (2011) as well as Anderson, Håkansson, and Johanson (1994) argue that dyadic business relationships need to be placed in a wider context. Without an understanding of the important direct and indirect actors surrounding the focal firm, i.e., the network context, the understanding of firms' market-and relationship-oriented behaviors is only limited.
We therefore hypothesize that network-oriented behaviors are the antecedents of market-and relationship-oriented behaviors. Firms are able to understand the key players in a broader context through network-oriented behaviors. The sensing and seizing network-oriented behaviors, as described by Thornton et al. (2013), allow a focal firm to make more informed decisions in relation to customers and competitors (Thorelli, 1986). This issue relates to the fact that when managing relationships successfully in a dyadic sense, the interconnectedness of all these relationships as part of a focal firm's relationship portfolio needs to be managed in a holistic way (Hoffmann, 2007;Roseira, Brito, & Henneberg, 2010;Vargo & Lusch, 2011). Network-oriented behaviors facilitate such coordination of different relationships as part of a relationship portfolio approach, based on an understanding of the available resources and opportunities (e.g., opportunities for synergies) that lie beyond the direct dyadic relationships of a firm (McEvily & Zaheer, 1999;Rowley, 1997).
However, these effects of network-oriented behaviors are likely amplified in a fast moving technological environment (Achrol & Kotler, 1999), e.g., the level of technological turbulence has become an important contextual factor in affecting the effectiveness of firms' endeavors to respond to markets (Jaworski & Kohli, 1993;Kumar et al., 2011). On the other hand, the effectiveness of network-oriented behaviors is also hypothesized to be affected by how well a focal firm interacts with the end users of a firm's offerings. Firms that are able to interact and/or understand their indirect final customers might utilize their network-oriented behaviors in a more effective way (Henneberg, Mouzas, & Naudé, 2009). Following the aforementioned key areas of research, we derive a nomological model, as depicted in Fig. 1, based on three key themes: (1) network-oriented behaviors as a driver of market-oriented and relationship-oriented behaviors, (2) the effects of market-oriented and relationship-oriented behaviors on firm performance, and (3) the direct role of network-oriented behaviors on firm performance. Technological turbulence as well as closeness to endusers are included as important moderating constructs on the effect of network-oriented behaviors on firm performance.

Network-oriented behaviors as a driver of market-and relationship-oriented behaviors
Network-oriented behaviors are defined as "activities/routines/ practices, which enable firms to make sense of and capitalize on their networks of direct and indirect relationships" (Thornton et al., 2013(Thornton et al., , p. 1155. Following this conceptualization, we define the construct of network-oriented behaviors as a set of behaviors using direct and indirect relationships, which include both strong-tie and weak-tie relationships, in order to achieve four different anticipated outcomes. In other words, it is the combination of these four dimensions of networking, i.e., information acquisition, opportunity enabling, strong-tie resource mobilization and weak-tie resource mobilization, which represent a firm's network-oriented behaviors. It is important to note that these behaviors have to be understood as a systemic whole, e.g., the utilization of strong-tie and weak-tie relationships are complementary (Tiwana, 2008). In his empirical study, Uzzi (1996) suggests that the balanced use of strong-tie and weak-tie relationships can minimize a firm's failure rate, which means that the use of these two types of relationships need to be examined in combination rather than in isolation. Following a similar argument, Tiwana (2008) argues that it is imperative to go beyond the dyadic relationships of a firm and consider the portfolio of both strong-tie and weak-tie relationships, which resemble its network structure, particularly when assessing firm performance.
Market-oriented behaviors are derived from the behavioral market orientation concept developed by Narver and Slater (1990), which includes customer orientation, competitor orientation and inter-functional coordination. Most studies have treated these three components as the manifestations of a firm's market orientation based on a reflective measurement model. This way of conceptualizing market orientation is not without criticisms (e.g., Cadogan, Souchon, & Procter, 2008) as it implies that the three components of market-oriented behaviors are interchangeable and replaceable. Therefore, the uniqueness of these three components is not accounted for. This, in our view, neglects the important and distinct implications of how a firm's customer and competitor-oriented behaviors affect, and are affected by, other constructs, when placed within a nomological model.
We focus on customer and competitor-oriented behaviors as part of a market orientation since these capture a firm's market-oriented behaviors on the basis that customers and competitors are two key players in a firm's network environment (Mattsson, 1997;Möller & Halinen, 1999). We are particularly interested in how network-oriented behaviors affect firm's behaviors toward these two network actors. First, customer-oriented behaviors refer to firms' tendencies to continuously create superior value for their customers based on a sufficient understanding of customers' business environments (Narver & Slater, 1990). Secondly, competitor-oriented behaviors refer to firms' tendencies to continuously seek to sense competitor actions and respond to them timely and appropriately (Narver & Slater, 1990). The means by which a firm generates the necessary information and mobilizes certain resources (e.g., knowledge) in order to appropriately recognize, address and respond to customer needs and competitor actions have not been well articulated in the literature. Kohli and Jaworski (1990) merely describe in their seminal work that "a customer focus involves obtaining information from customers about their needs and preferences […] it goes far beyond customer research" (p. 3). They also stress that the necessary information for understanding customers' current and future needs to include "exogenous market factors (e.g., competition, regulation)" (p. 3). In essence, behaviors aimed at going beyond direct interaction partners, such as network-oriented behaviors, are driving the process of recognizing, addressing and fulfilling customer needs.
Through its unique network position a firm can potentially obtain useful, possibly critical, information by interacting with its web of different types of relationships, be they well established or newly formed (Uzzi, 1996). Such network-oriented behaviors comprise different aspects: strong-tie relationships foster effective tacit or complex information transfer and resource mobilization across relationships, because of their established trust mechanism. On the other hand, weak-tie relationships provide a bridge that links novel information and resources to the focal firm. The combination of these two types of relationships to a certain extent would increase a firm's success rate (Uzzi, 1996), since it can better understand its customers as well as its competitors by means of seeking information dispersed in the network that help 'contextualize' the identified issues at hand. We therefore hypothesize that: H1. A firm's network-oriented behaviors positively affect its customeroriented behaviors.

H2. A firm's network-oriented behaviors positively affect its competitororiented behaviors.
Relationship-oriented behaviors refer to a firm's activities to coordinate with its counterparts based on involved parties' mutual goals (Walter et al., 2006). Relationship-oriented behaviors are not specific to customers; rather they are also intended for suppliers, as well as other relevant business partners. The resource mobilization within a confined set of established relationships will allow firms to more effectively coordinate with each partner due to the level of trust and relational norms that have been established (Wuyts, Dutta, & Stremersch, 2004;Zaefarian et al., 2011). In addition, the understanding of the resource constellations surrounding the focal firm and the ability to mobilize resources via network-oriented behaviors allows the firm to utilize the available resources pooled from its relationship portfolio (Håkansson & Ford, 2002;Mouzas & Naudé, 2007). This has two strategic implications. First, firms are able to reconfigure the combinations of the available resources based on its goals (Hoffmann, 2007;Roseira et al., 2010). Secondly, they can identify other desirable resources that are not currently within the direct reach of the firms (Hoffmann, 2007). The contextual understanding of the resources embedded in the network and the ability to configure/reconfigure those resources through the use of network-oriented behaviors will allow a firm to make decisions on whether or not to adjust the levels of relationship investments with each existing business partner, given their mutual goals. We therefore hypothesize:

The effects of market-and relationship-oriented behaviors
The effects of market-oriented and relationship-oriented behaviors have been well established and documented through empirical evidence in the literature (for a summary see Liao, Chang, Wu, & Katrichis, 2011). Therefore, we treat the resulting hypotheses as an integral part of the overall nomological model, which helps to conceptualize the impact of different firm behaviors on firm performance. Firm performance in this study refers to two different measures, namely firm profitability and relational performance, the latter of which is particularly applicable for business-to-business studies (Bhappu & Schultze, 2006). Financial performance is based on a firm's assessment of its profitability compared to its competition, in line with Venkatraman (1989). Relational performance refers to the overall effectiveness of a firm's relationship portfolio (Johnson, Sohi, & Grewal, 2004).
Being customer-oriented has been the pivot of the argument as to why businesses exist in that "to satisfy the customer is the mission and purpose of every business" (Drucker, 1973, p. 79). Being customeroriented allows firms to more effectively deal with other important business partners in order to satisfy customers' need (Smirnova, Naudé, Henneberg, Mouzas, & Kouchtch, 2011). Customer-orientated behaviors therefore inform firms' relationship coordinating activities that aim at better satisfying those needs. This is done, for example, through activities of demand chain integration (Jüttner et al., 2010). Hence:

H4.
A firm's customer-oriented behaviors positively affect its relationship-oriented behaviors.
In addition, customer-orientated behaviors help firms to become more aware of competition since satisfying customers require the identification of a competitive advantage over major competitive offerings and also relate to how firms react to competitor's activities (Narver & Slater, 1990). Our next hypothesis is therefore:

H5.
A firm's customer-oriented behaviors positively affect its competitor-oriented behaviors.
The majority of the extant literature shows that more market-oriented firms perform better in their financial outcomes (Greenley, 1995;Jaworski & Kohli, 1993;Narver & Slater, 1990) and innovation success (Deshpandé et al., 1993;Narver, Slater, & MacLachlan, 2004). However, there is no specific empirical evidence to support the notion that both customer-and competitor-oriented behaviors independently lead to a superior profitability due to the fact that these two constructs are often conflated within the overarching market orientation construct. For example, Deshpandé et al. (1993) base their conceptualization of 'customer orientation' on the combination of Narver and Slater (1990) 'customer orientation' and 'competitor orientation', and provide evidence of a positive effect on firm performance. We can therefore hypothesize that: H6. A firm's customer-oriented behaviors positively affect its profitability.

H7.
A firm's competitor-oriented behaviors positively affect its profitability.
Ample evidence in the literature of business-to-business marketing suggests that effective relationship management allows firms to achieve favorable relational outcomes, such as customer trust  and cooperation (Morgan & Hunt, 1994), as well as beneficial financial outcomes, such as customer value capture (Palmatier, 2008) and business performance (Smirnova et al., 2011). In addition, Johnson et al. (2004) suggest that a firm's ability to manage relationship activities and initiate cooperation with business partners increases the effectiveness of the overall relationship portfolio. Given the evidence in the literature, we hypothesize that firms' abilities to coordinate their business relationships allow them to develop a holistic view of their relationship portfolio and facilitate the optimized use of the resources within these portfolios, hence: H8. A firm's relationship-oriented behaviors positively affect its relationship portfolio effectiveness.

The role of network-oriented behavior on firm performance
The existing literature has yet to provide evidence on whether firms' networking efforts can produce certain desired outcomes. Ford et al. (2003) contend that firms' attempts to change their network position cannot be linked directly to any intended outcomes, as the affected business interactions are dynamic and fluid as well as complex. However, although a firm's network-oriented behaviors might not result in direct contributions to its profitability, it is plausible to infer that a firm's strategic activities within its network, which are its network-oriented behaviors, help the overall effectiveness of its relationship portfolio (Hoffmann, 2007). Particularly, resource mobilization across various relationships fosters the effectiveness of its relationship exchanges based on its 'network horizon', that is the firm's vision through which it grasps the dynamics in the network (Holmen & Pedersen, 2003). Here, for a firm's relationship portfolio to be effective, the sensing and seizing aspects of network-oriented behaviors allow the firm to effectively utilize the pooled resources that are embedded in this portfolio (Thornton et al., 2013). We therefore hypothesize: H9. A firm's network-oriented behaviors positively affect its relationship portfolio effectiveness.
It is widely recognized that the ability to manage business relationships effectively is a key driver of a superior firm performance (Morgan & Hunt, 1994;Palmatier, 2008;Palmatier et al., 2008). These business relationships provide a firm with unique access to information, resources and opportunities that are crucial for firm success (Burt, 2000;Zaefarian et al., 2011). This uniqueness of resources embedded within the specific constellation of a relationship portfolio facilitates leveraging and utilizing different configurations of these resources (Zaheer & Bell, 2005). These competitive advantages created from a firm's effective relationship portfolio lead to a higher likelihood for a firm to strategically succeed (Gulati et al., 2000). The next hypothesis is therefore: H10. A firm's relationship portfolio effectiveness positively affects its profitability.

Moderation effects
Based on the nomological model in Fig. 1, two moderating factors are included. We focus on the innovative aspects of the model, which are the effects of network-oriented behaviors (while other possible moderation effect, such as those affecting market, competitor, or relationshiporiented behaviors are not included for reasons of parsimony). If a firm's network-oriented behaviors can help generate an effective relationship portfolio, the question arises as to under what conditions these network-oriented behaviors are more or less likely to be successful in facilitating an effective relationship portfolio. We hypothesize one contextual factor and one firm-specific factor to amplify the positive effect of network-oriented behaviors on relational performance.
Network-oriented behaviors are said to be crucial in specific contexts, for example a highly volatile environment in relation to technological developments (Möller & Halinen, 1999;Mouzas & Naudé, 2007). When technologies change rapidly within a firm's environment, its ability to sense these dynamics and seize the opportunities will better enable the firm to utilize its existing relationship portfolio, and possibly change the formation of the portfolio in order to effectively compete in such a dynamic environment (Hagedoorn et al., 2006). We hypothesize that: H11. The higher the technological turbulence, the stronger the positive effect of network-oriented behaviors on relationship portfolio effectiveness.
Firms operating in business markets are commonly not in direct contact with the end users of the end product/service offerings to which they contribute, as their offerings (e.g., a component) may be only one part of the final offering to the final customer. Alternatively, in the case of equipment, their offerings might only help transform resources into a final offering. A firm could be very 'far' away from the end users, if it is located at a more upstream network position. The difference in firm position has an important implication related to how much insight it can gather through networking from its direct business partners about aspects further afield in the network, such as final customer preferences (Rowley, 1997;Wu, 2008). This issue of 'closeness' to the final customer relates to how closely a firm is located in the network in relation to the end users of the offering on the one hand, but also how much, or how easily a firm interacts with the end users. However, there exist instances where it is not possible for a firm to interact with the end users due to contractual constraints with its direct customers.
We postulate that when firms are able to interact with the end users of their offerings, or are close to the end users, their network-oriented behaviors will be more effective in affecting their relationship portfolio effectiveness. Under such conditions, firms are likely to utilize their network-oriented behaviors strategically to more effectively utilize the relationship portfolio, hence: H12. The closer a firm is to its end users, the stronger the positive effect of network-oriented behaviors on relationship portfolio effectiveness.

Research design
We chose UK services and manufacturing firms as the research context to study their behaviors toward their network, their relationships, and their customers and competitors. Given the fact that these firms are facing intensifying global competition from developing countries, such as China, the ability to leverage and mobilize resources in their networks becomes critical (Achrol & Kotler, 1999). This context is therefore well suited to the present study. In the following sections, we will detail the process of data collection, measurement instrument development, and the validation of the measurement model. Data analysis was carried out using a combination of SPSS (ver. 20.0) and Mplus (ver. 7.11).

Sampling and data collection
We conducted a web-based survey using Qualtrics, an integrated platform for survey design and data collection. A panel database of managers working across a wide range of industries in the UK was utilized as a sampling frame for this study. Research invitations were sent to the potential respondents in the sampling frame in four batches between July and August 2013. We purposefully utilized three filtering questions to select suitable respondents from the sampling frame. The respondents were allowed to participate in this study if (1) their companies operated predominately in business-to-business markets, (2) their self-rated knowledgeability about the business relationships of the companies they represented was equal to or more than 4 out of a scale of 1 (poor) to 7 (excellent), and (3) their companies are within either the services or manufacturing sectors. Out of 6715 potential respondents contacted, a total of 1379 possible respondents were eligible for taking part in this study. After deleting the screen-out and incomplete responses, the survey resulted in 413 completed responses, a response rate of 29.9%. However, to ensure the quality of the dataset, we further eliminated responses completed in less than 5 min, which yielded 354 valid responses for the subsequent analyses. The threshold of 5 min was decided as the cut-off point of a 'valid' response based on a pre-test which showed that faster results indicated 'pattern responses' (Fricker, Galesic, Tourangeau, & Ting, 2005).
Next, we used an extrapolation method (e.g., comparing early and late responses) to assess possible non-response bias in the data (Armstrong & Overton, 1977). We use the first and the fourth quartiles (25% and 75%), based on the subtraction of the recorded time of survey completion and invitation sent, as early and late respondents. The late respondents are assumed to approximate non-respondents. Both key respondent profile variables (company size, job position and years in the current job position) and the main construct variables (networkoriented, customer-oriented, competitor-oriented and relationshiporiented behaviors) were compared across early and late response groups, using a series of independent t-tests to compare means of continuous variables and χ 2 difference tests for categorical variables. The results of these tests show no significant difference across these two groups, which lead us to conclude that nonresponse bias is not a concern for our data. We additionally compared known socio-demographic information for respondents and non-respondents (including firm characteristics) and did not find significant differences. Table 1 summarizes the profile of the respondents and their organizations. A total of 68.4% of the 354 respondents came from service industries, while 31.6% of them worked in the manufacturing sector. In terms of their organizational size, 29.1% and 27.7% of them, respectively, were classified as small and medium businesses, while large businesses accounted for 43.2%. With regard to respondent characteristics, almost half of the respondents were at a position of middle to top management (44.6%), followed by owner or joint-owner (18.9%), managing director (15.0%) and other top-level directors (13.3%). In addition, just over half of the respondents had more than 10 years of managerial experience (53.4%), while 20.6% and 26.0% of them had 0-5 years and 6-10 years experience, respectively.

Construct measurements
We adapted most measures from the existing literature as prior studies provide satisfactory reliability and validity results. A seven-point Likert scale, labeled at the two endpoints, 1 = 'completely disagree' and 7 = 'completely agree', was used for most multi-item measures that reflect underlying constructs, unless otherwise stated. The full list of measures can be found in Appendix A.
Network-oriented behaviors is a second-order formative construct, measured by four reflective first-order constructs based on the empirical study of Thornton, Henneberg, and Naudé (2014). The four key components are information acquisition, opportunity enabling, strong-tie resource mobilization and weak-tie resource mobilization, which are measured by four reflective indicators respectively. These four components with reflective indicators will be modeled to form the overarching organizational networking behaviors as a second order formative construct. Based on the empirical results, the measurement model produces a good fit (RMSEA = 0.041, SRMR = 0.048, NFI = 0.98, CFI = 0.99, IFI = 0.99, GFI = 0.93, AGFI = 0.91) and shows that all four components (standardized coefficients are 0.29***, 0.28***, 0.42*** and 0.18**, respectively) are significant contributors to the secondorder formative construct through the validation process of a multiple indicators and multiple causes (MIMIC) model (Diamantopoulos & Winklhofer, 2001). In the overall structural model the four first-order reflective constructs will be treated as four single indicators that form the network-oriented behaviors by taking an average score for each of the four constructs.
Customer-oriented behaviors and competitor-oriented behaviors are adapted from Narver and Slater (1990). These two constructs are part of their market orientation operationalization in relation to the two key counterparts in a focal firm's business network. The components of customer orientation and competitor orientation, plus a third aspect of inter-functional coordination, each measured by reflective indicators, are mostly used to reflect the market orientation construct. However, more recent literature has disputed this approach, claiming that the three components should be modeled to form market orientation (i.e., as a second order formative construct), because they are not interchangeable (Cadogan et al., 2008). While being aware of these discussions, in line with the evidence provided by Siguaw and Diamantopoulos (1995), we treat customer-oriented behaviors and competitor-oriented behaviors as two separate constructs that are measured reflectively by their respective three indicators.
Relationship-coordinating behaviors is measured with four items adapted from Walter et al. (2006). These indicators tap into the extent to which a focal firm coordinates resources and activities according to the match of resources and activities with each partner, which can be its customer or supplier.
We developed a new scale for one of the moderation variables, closeness to end user, based on two items measuring the extent to which a focal firm feels close to the end users of their offerings, with which it has only indirect relationships. In addition, we adapted constructs for technological turbulence, competitive intensity and market turbulence from Jaworski and Kohli (1993), each of which has three items. Note that only closeness to end user and technological turbulence were included in the hypothesized nomological model. Competitive intensity and market turbulence, together with technological turbulence were aggregated to create a composite construct, i.e., 'environmental volatility', which was used for a multi-group analysis.
There are two performance variables in the measurement model. The relational performance is measured by relationship portfolio effectiveness, which is adapted from Johnson et al. (2004). Three items are used to measure the effectiveness of a focal firm's overall relationship portfolio. The financial performance is measured by firm profitability with three items adapted from Venkatraman (1989) that indicate a focal firm's assessment of its profitability position in relation to competition.
We also employed a range of control variables. Industry growth is measured by a single item indicating the overall industry growth in the UK ranging from poor to excellent (7 point scale). Market presence is measured by a single item indicating the number of years that a firm has been established in the UK. Firm size is measured by number of employees (based on 10 categories).

Assessing the measurement model
The measurement model as a whole was assessed by a confirmatory factor analysis using Mplus (ver. 7.1) in order to establish the level of model fit. All 11 constructs are modeled as reflective measurements (including four first-order constructs that form the second-order formative construct of network-oriented behaviors) based on their respective theorized factors. The measurement model specification allows each construct to covariate with all others. The model fit indices are as follows: χ 2 (574) = 881.169 (p b 0.000), comparative fit index (CFI) = 0.96, Tucker-Lewis index (TLI) = 0.95, root mean square error of approximation (RMSEA) = 0.039 and standardized root mean square residual (SRMR) = 0.036. A significant χ 2 can be expected and still indicates a good model fit, when the sample size is more than 250 (sample size = 354 in our study) and the observed variables are more than 12 (number of items in the measurement model = 37) (Hair, Black, Babin, & Anderson, 2008). In addition, the ratio of χ 2 / degree of freedom at 1.54 (b2), indicates a very good fit (Hair et al., 2008). We therefore conclude that these indices are in support of a good measurement model fit.
For assessing the convergent validity of all the constructs in the measurement model we closely follow a comprehensive procedure proposed by Hair et al. (2008). First, all the items have factor loadings above 0.7 (0.71-0.92) (see Appendix A), which is well above the cut-off point of 0.5. According to the results presented in Table 2, average variance extracted (AVE) by each factor (0.63-0.80) is well above the cut-off point of 0.5. All factors show very good levels of internal consistency, as their composite reliability (CR) is in the range of 0.86-0.92, which is well above the suggested threshold of 0.6-0.7 as a minimum. Based on the above evaluation, we conclude that the measurement model has satisfied the criteria of convergent validity. In addition, correlations between any given two factors are substantially smaller than 1, and the AVE for any given two factors is greater than the squared correlation between these two factors (Fornell & Larcker, 1981). Therefore, all factors in the measurement model display adequate discriminant validity.
Common method bias is assessed next. The general consensus suggests that wherever possible a procedural prevention should be taken in the first place to mitigate the threat of such bias derived from common methods, such as same source data (e.g., self-report survey). We have carefully designed certain aspects of the measurement instrument based on our assessment of the possible sources of method variances (Spector, 2006). For example, we intentionally randomized the question order so as to break up the causal relationships of the substantive constructs under study (Podsakoff, MacKenzie, Jeong-Yeon, & Podsakoff, 2003). We also used Likert as well as semantic scales interchangeably and appropriately without overloading respondents' cognitive tasks by using 7-point rating scales throughout when applicable (Podsakoff, MacKenzie, & Podsakoff, 2012). We employed a knowledgeability question at the beginning of the on-line questionnaire to ensure that only those respondents who are capable of answering the following questions will continue filling out the questionnaire (Spector, 2006).
The statistical assessment of common method bias involves two steps as part of the most widely used Harman's single factor test using both exploratory factor analysis and confirmatory factor analysis. First, in the exploratory factor analysis all the items in the measurement model were entered and the result reveals that the first factor explains less than half of the total variance. Of the 11 constructs in the measurement model, 4 are first-order reflective constructs of network-oriented behaviors, which are all based on a firm's interactions with others. It can be expected that these items would share a considerable amount of variances because of the fact that they are all interactions between a firm and its counterparts. Secondly in the confirmatory factor analysis we compared the theorized multi-factor measurement model against a single-factor model with all indicators loaded on it. The single factorsolution produces a significantly inadequate fit (χ 2 (665) = 3318.49 (p b 0.000), CFI = 0.66, TLI = 0.64, RMSEA = 0.110, SRMR = 0.083) compared to the multi-factor solution. The χ 2 -difference test shows that the hypothesized measurement model fits the data significantly better than the single-factor model (p b 0.001). Given the procedural remedies we have taken and the results of the above analyses, we reasonably conclude that common method bias is not cause for concern in the assessment of the hypothesized structural model.

Assessing hypothesized structural model
We modeled the four components of network-oriented behaviors as formative indicators based on a priori theory (Thornton et al., 2014). One fundamental issue of any formative measurement is the extent of multicollinearity among the formative indicators, the presence of which will make it difficult to assess the unique contribution from each of them (Diamantopoulos & Winklhofer, 2001). To assess multicollinearity, four multiple regressions were performed. In each of them a formative indicator was regressed on the remaining three in order to obtain the variance inflation factors (VIF). The VIF ranges from 1.820 to 2.192, which is well below the suggested threshold of 10 (e.g., Hair et al., 2008), and within the more stringent cutoff point of 3 (Petter, Straub, & Rai, 2007). Multicollinearity therefore does not pose a threat for modeling organizational networking behavior as a formative measurement, the disturbance term of which has been set to 0, in line with the specification of a formative measurement embedded in a structural model.

Main effects
We test the structural equation model containing all the hypothesized direct effects, using Mplus with adjusted maximum likelihood estimation (Satorra & Bentler, 2001). The model provides a good fit: χ 2 (df = 185) = 377.63 (p b 0.000), CFI = 0.95, TLI = 0.94, RMSEA = 0.054, SRMR = 0.047). Almost all hypothesized paths are statistically significant and in the expected direction except one path related to the relationship of customer-oriented behaviors on firm profitability (see Table 3). Note that given the ongoing debate surrounding the best way to model a formative construct (e.g., Diamantopoulos, 2013;Lee, Cadogan, & Chamberlain, 2014), we compared the results to that of an alternative solution in which the formative construct is modeled as a composite variable, while other constructs remain reflective. The comparison suggests that there is no substantive difference in interpretation of results in both the hypothesized relationships and model fit, except a higher SRMR is found in the alternative solution. We therefore present the original solution here. First, as can be seen from Table 3 network-oriented behaviors positively affect customeroriented (β1 = 0.68, p b 0.001), competitor-oriented (β2 = 0.44, p b 0.001) and relationship coordinating behaviors (β3 = 0.41, p b 0.001), in support of H1, H2 and H3 that network-oriented behaviors are indeed a driver of other important firm behaviors.
Finally, we examine the role of network-oriented behaviors in relation to relationship portfolio effectiveness (β9 = 0.30, p b 0.001), as well as the effect of the latter on firm profitability (β10 = 0.47, p b 0.001). The results support both H9 and H10.
In addition, we also test whether portfolio effectiveness mediates (1) the effect of network-oriented behaviors on firm profitability, and  z-value is the standardized coefficient divided by its standard error (Byrne, 2012). A value of greater than 1.96 indicates the value is significantly different from zero based on a 95% confidence level.
(2) the effect of relationship coordinating behaviors on firm profitability. We test the mediation effects within the hypothesized structural model by using the delta method (MacKinnon, 2008), with two further previously non-hypothesized direct relationships of (1) and (2). This approach allows for the test of a mediation effect of interest in the context of multiple mediation effects, taking into consideration total indirect effects (Preacher & Hayes, 2008). With regard to (1), the result shows that while network-oriented behaviors do not affect firm profitability directly, the full mediation effect of the relationship through portfolio effectiveness is positive and significant (p b 0.05). Similarly, given the direct relationship of (2) is non-significant, the test of the mediation effect corroborates that portfolio effectiveness fully mediates the impact of relationship coordinating behaviors on firm profitability (p b 0.001).

Moderation effects
Although we do not expect the causal relationships in the overall structural model to differ under the influence of the contextual factors, we treat them as control variables and see whether the model holds for large as well as small firms, and for low and high levels of environmental volatility. To test changes in causal relationships across different groups, multi-group analyses were performed. We utilize a composite variable, formed by market turbulence, technological turbulence and competitive intensity (Jaworski & Kohli, 1993) to denote the level of volatility in a firm's business environment. We take a median split to form two groups: lower (n = 177) and higher volatility (n = 177). After partial metric invariance was established (Hair et al., 2008), we specify a structural model allowing all path parameters to be freely estimated across two groups, against which a model is also specified with all path constrained to be equal across two groups. A χ 2 -difference test between these two models (χ 2 (df) = 13.26 (10), p = 0.21) reveals that there is no significant difference in model fit, which means that the causal relationships do hold across lower and higher volatility groups.
The procedure was repeated to assess whether the model holds across smaller and larger firms. Again we use a median split for 10 categories of firm size (measured by the number of employees) to form two groups: smaller firms with employees under 249 (n = 201) and larger firms with employees more than 250 (n = 153). A χ 2 -difference test between the freely estimated model and the constrained model (χ 2 (df) = 34.04 (10), p b 0.001) indicates that the causal relationships in the structural model do not hold across smaller and larger firms. The most notable differences based on a loose multigroup comparison, using the same model estimated separately in the two groups, are the paths for customer-oriented and competitororiented behaviors on firm profitability. Strikingly, for smaller firms, competitor-oriented behaviors do not contribute to firm profitability (β = 0.20, p = 0.071), whereas for the larger firms, the same set of behaviors significantly and strongly affect firm profitability (β = 0.52, p b 0.001). Furthermore, customer-oriented behaviors have no impact on smaller firms' profitability (β = − 0.15, p = 0.21); whereas the same behaviors have a significant and negative impact on larger firms' profitability (β = −0.48, p b 0.001).
In order to test the two hypothesized moderation effects, we employ a Latent Moderated Structural Equations approach (LMS) with adjusted maximum likelihood estimation specifically developed for dealing with "the distributional characteristics of the nonnormally distributed joint vector in a latent interaction model" (Klein & Moosbrugger, 2000, p. 473). LMS is a relatively robust method for assessing interaction effects embedded in a structural model (Little, Bovaird, & Widaman, 2009). We have also taken the decision to adopt this approach over a multi-group analysis on the basis that the latter does not allow us to assess multiple interaction effects; neither does it take into account the measurement errors in the structural model. Since Mplus is used for testing the latent interaction effects, we closely follow Muthén (2012) and Muthén and Asparouhov (2003) for model specification and result interpretation.
Within the direct effect model we added the two moderation constructs, i.e., technological turbulence and closeness to end users, on which the dependent constructs, relationship portfolio effectiveness is regressed, according to H11 and H12. We specified two interaction terms, network-oriented behaviors and closeness to end users, and relationship portfolio effectiveness and technological turbulence, on which relationship portfolio effectiveness and firm profitability are regressed. Note that although Mplus is capable of handling a structural equation model with multiple latent interactions, using the LMS approach, it only provides unstandardized coefficients and very limited model fit output. Although it does not allow assessing the effect size (R 2 ), it does allow for testing the hypothesized multiple interaction effects in the direct effect model. The results show that the relationship of network-oriented behaviors on relationship portfolio effectiveness is strengthened by the degree of end user closeness (p b 0.01), but not by the level of technological turbulence. Therefore, H11 is not supported, whereas H12 is. The unstandardized coefficients of networkoriented behaviors (b = 0.107, p b 0.001), closeness to end users (b = − 0.223, p = 0.117) and their interaction term (b = 0.030, p b 0.01) were used to calculate the predicted values of relationship portfolio effectiveness based on high and low values (+1 and −1 standard deviation) of the predictor and the moderator (Aiken & West, 1991;Dawson, 2014). The predicted values are depicted in Fig. 2, which shows that the relationship of network-oriented behaviors on relationship portfolio effectiveness is strengthened by a high degree of end user closeness (the dotted line) compared to a low degree of end user closeness (the continuous line).

Discussion and implications
We set out to understand, from a resource dependence perspective, whether a firm's network-oriented behaviors in response to its business network help it to perform better (Pfeffer & Salancik, 1978;Salancik, 1995). The consensus in the network literature suggests that firms that are better positioned in the network perform better than their counterparts that occupy worse positions. However, it is also evident that firms with a similar network position have different levels of performance (Zaheer & Bell, 2005). One explanation for this relates to the fact that the network position of a firm can be differently exploited, or strategically changed through network-oriented behaviors by adapting either the structure of the relationship portfolio or the patterns of interactions with interaction partners (Gulati, 1999;Robert, 1992). On the other hand, the actions by others in the network can also change a focal firm's network position (Ford & Håkansson, 2006). Networkoriented behaviors are therefore strategic acts that have the purpose of responding to the dynamics of the network and proactively creating ways of leveraging resources (Thornton et al., 2013).
Against this backdrop of a theoretical framework, we offer a conceptual model that outlines how a firm can utilize different organizational behaviors, specifically its network-oriented behaviors, to understand its customers and competitors, and coordinate with its important business partners within the network. This framework provides an explanation as to how firms' strategic behaviors contribute to firm performance, either directly or indirectly. The following discussion of our empirical analysis is structured around the three themes underlying our nomological model, followed by managerial implications, limitations and future research directions.

Network-oriented behaviors as a driver of market-and relationship-oriented behaviors
Network-oriented behaviors are hypothesized in this study as the driving force of a firm's behaviors toward its direct customers and its competitors, and the relational interactions toward its important business partners. This is partly built on Day's (1994) theory of a market driven firm, including the concept of market orientation. He argues that from a strategic management perspective, a market driven firm should be equipped with both 'customer linking' and 'market sensing' behaviors. He also suggests that a firm's market orientation needs to be enhanced by its constant learning that brings about the development of necessary capabilities and competencies for sustaining its organizational success.
It is evident from our study that a firm's network-oriented behaviors positively affect its customer-, competitor-and relationship-oriented behaviors, and explain a large proportion of the variances of these constructs (46%, 67% and 77%, respectively). Our findings therefore provide empirical evidence for Day's (1994) argument that network-oriented behaviors act as sensing and seizing activities, which are the configuration of four broad sets of goal-driven network-oriented behaviors, namely information acquisition, opportunity enabling, strong-tie resource mobilization and weak-tie resource mobilization. Note that all of these four formative indicators significantly contribute to the overall network-oriented behaviors (standardized coefficients are 0. 22, 0.44, 0.29, 0.21), which largely corroborates the result of Thornton et al. (2014).

The effects of market-and relationship-oriented behaviors
Although assessing the effects of market-and relationship-oriented behaviors is not our main contribution, some relevant aspects regarding firm performance need to be mentioned. We deviate from the predominant approach of aggregating different components of market-oriented behaviors, and treat customer-and competitor-oriented behaviors as two separate constructs. The results show that customer-oriented behaviors have no direct influence on a firm's profitability, whereas competitor-oriented behaviors have a significantly positive impact. This is somewhat surprising against the backdrop of the extant literature suggesting that the core of market-oriented behaviors is a 'customer focus' with some even arguing that a customer orientation is analogous to a market orientation (Deshpandé & Farley, 1998;Deshpandé et al., 1993). It is evident from more recent studies that market orientation does not necessarily lead to a superior firm performance. Cadogan et al. (2009) show that the impact of market orientation on performance is an inverted U shape, which implies that being overly market-oriented can be detrimental to a firm's success. Grewal and Tansuhaj (2001) conclude that a firm's market orientation has a negative effect on firm performance after an economic crisis. In addition, the effect is weakened by demand and technological uncertainty and is strengthened by competitive intensity.
The result that a customer focus does not impact on a firm's profitability can be due to the fact that being customer oriented has become a necessity, rather than a competitive advantage, which is in part echoing the argument provided by Kumar et al. (2011). This implies that being overly customer-oriented can be detrimental to a firm's profitability since it might require unnecessary capital investments in order to fulfill customer needs. This type of investment arguably would contribute to sales growth, but it is questionable as to whether it is profitable to do whatever customers require. However, based on the proposed and validated nomological model, customer-oriented behaviors are a significant driver of competitor-oriented behaviors, which subsequently positively impact on firm profitability. In this context, a customer focus is still important since it helps a firm to become aware of its competitors. Therefore, our result does not imply that a firm should not be customer-oriented. Rather, being customer-oriented is necessary for developing a firm's competitororiented behaviors, and it is the latter that facilitates firms' ability to increase their profitability.
A multi-group analysis based on firm size reveals that for smaller firms customer-oriented behaviors have no impact on firm profitability, whereas the same behaviors have a significant negative impact on larger firms' profitability. For smaller firms being market-oriented or not is independent of how they perform, since they rely on the effective use of their relationship portfolio to sustain their profitability. This may imply that due to the resource constraints of smaller firms, they need to leverage their counterparts' resources in order to compete in the market place, in line with the resource dependence theory (Pfeffer & Salancik, 1978). Larger firms, on the other hand, face a negative impact on their profitability from being overly customer-oriented, whereas other organizational behaviors, such as understanding their major competition in the market and utilizing their relationship portfolio effectively, are both important in driving firm profitability.
A firm's relationship-oriented behaviors, such as its ability to coordinate with its important partners, positively affects the effectiveness of its overall relationship portfolio, which in turn serves as an influential factor that contributes to a firm's profitability. However, the multigroup analysis, again based on firm size, reveals that in the case of smaller firms competitor-oriented behaviors do not contribute to their profitability (β = 0.20, p = 0.071), whereas for larger firms, the same set of behaviors significantly and positively affect firm profitability (β = 0.52, p b 0.001).
In this context, our findings echo Kumar et al. (2011) that market oriented-behaviors have gradually become a prerequisite for any firm to compete in the market place, rather than a competitive advantage. Furthermore, being market-oriented could be a mere 'self-portrait' rather than a true representation of a market-centric approach (Deshpandé et al., 1993). Our finding is certainly not to suggest that firms should not be customer-focused, but rather that a customer focus is essential according to our research results particularly for firms to develop their competition awareness and foster coordination with important business partners in order to better utilize resources to fulfill customer needs.

The role of network-oriented behaviors on firm performance
The network-oriented behaviors are outward facing and can be seen as a constant evolving learning mechanism that represent a firm's orientation toward its network context (Day, 2000;Thornton et al., 2013). This learning mechanism is not directly a 'profit spinner', i.e., a driver of firm profitability, but rather it serves as an enabling force of other organizational behaviors (customer, competitor and relationship specific ones) (Day, 2000). However, besides these indirect effects, networkoriented behaviors also directly affect relationship portfolio effectiveness, which is in turn an important driver of firm profitability. Networkoriented behaviors serve to effectively influence the way in which a firm coordinates with its important partners. Through networkoriented sensing and seizing efforts firms are more likely to recognize the scarce resources in the network and how they can be mobilized by effective coordination with business partners (Cui & O'Connor, 2012;Mouzas & Naudé, 2007). These network-oriented behaviors have a bearing on the effectiveness of a firm's relationship portfolio. Strong-tie resource mobilization fosters resource synergies within the existing relationship portfolio (Roseira et al., 2010), and prompts an understanding of redundant resources within the existing portfolio (Hagedoorn et al., 2006). The effectiveness of the overall relationship portfolio also benefits from introducing new relationships through network-oriented behaviors. This approach fosters the use of novel resource combinations through weak-tie resource mobilization, which provides new opportunities for a firm (Burt, 2000;McEvily & Zaheer, 1999).
Our research results suggest that network-oriented behaviors do not affect a firm's profitability directly. However, their indirect impact on firm profitability through relationship portfolio effectiveness (as evident from the result of the mediation test) on the one hand, and other organizational behaviors on the other, provides important theoretical contribution toward explaining firm performance in business networks. The effect of network-oriented behaviors on portfolio effectiveness is strengthened by a firm's degree of end user closeness, but not by the technological turbulence, according to the latent interaction model results. First, this implies that regardless of the levels of technological turbulence, network-oriented efforts are influential for a firm to increase the effective use of the existing relationships within its relationship portfolio. Secondly, when a firm is able to understand the end users of its offerings by being close to them, the sensing and seizing as part of network-oriented behaviors can be utilized to form a superior understanding of demand chain integration and thereby allows for an optimization of the effectiveness of the firm's relationship portfolio (Jüttner et al., 2010).

Managerial implications
Our study suggests that firms operating in business-to-business markets need to strategize in networks (Holmen & Pedersen, 2003), which means that they need to employ behaviors beyond customer, competitor and relationship orientation. We offer three implications for the practitioners based on the research findings. First, firms need to take a configurational approach to the planning of their networkoriented behaviors as the four dimensions are complimentary to each other (Meyer, Tsui, & Hinings, 1993;Vorhies & Morgan, 2003). As such, our study supports Thornton et al.'s (2013) suggestion that "firms need to carefully plan these different types of networking activities/routines/ practices, using a portfolio approach, to maximize the utility their network context can afford" (p. 1163).
Secondly, the presence of a firm's customer-oriented behaviors is not necessarily fostering a superior firm profitability, but rather, firms need to beware of competition within the context of the network. Competitor-oriented behaviors can be better gauged and adjusted through assessing how competitor actions are likely to affect the existing use of resources for fulfilling current and future customer needs. Thirdly, rather than focusing on business partners in a dyadic sense, i.e., via individual relationship management activities, firms need to plan their interactions with each business partner, which include activity adjustment, knowledge exchange and resource configuration, based on the overall picture of other direct as well as indirect relationships. This signifies that a relational portfolio approach within the context of the network is necessary. This will allow firms to plan their relationships with each important business partner accordingly, including the interactions between these relationships, which will in turn fosters a more effective use of relationship portfolio (Johnson et al., 2004;Roseira et al., 2010).
Lastly, firms may try to get to understand the end users of their offerings, although some firms might find it difficult even to identify them since the demand chain might be lengthy in some instances. Despite empirical evidence regarding firms' considerations of their indirect customers is limited, the understanding of possible value creation in the context of the wider demand network is an important managerial issue (Henneberg & Mouzas, 2008). Our findings suggest that when firms feel close to their end users, their network-oriented behaviors are more effective in impacting on overall relationship portfolio effectiveness. This can be achieved through three means. First, a firm can gather relevant information about the end users of its offerings from its direct customers. This could be relatively effective if these relationships are well established, which allows effective transfer of valuable information (Uzzi, 1996). Secondly, a firm can gather insights directly from the end users of its offerings with the help of its direct customers, who can act as a go-between to provide a bridge for the interactions (Smith & Laage-Hellman, 1992). Thirdly, if the existing relationships do not allow such interactions, a two step-leverage can be employed to form new relationships with relevant parties in the network in order to be closer to the end users (Burt, 2000;Gargiulo, 1993).

Limitations and future research directions
Any research study exhibits certain limitations, based on one's research design choices. We would like to focus on two such limitations. First, industry specifics may have affected our findings in some way, based on our choice of the research setting. We chose both manufacturing and services sectors for our study based on the evidence of a measurement invariance test by Thornton et al. (2014). The results of their study show that there exist no significant differences in the way managers in these two sectors use network-oriented behaviors. We are mindful that differences could be significant even among different industries in the manufacturing or the services sector. However, the non-significant Levene statistics of the key variables suggest that there are equal levels of variance across these variables, which means that we can be confident in the assessment of the causal relationships among the constructs across these two industry sectors. Secondly, the use of Mplus as the main tool for assessing the latent interaction effects does not allow us to further evaluate the standardized coefficients and the effect size of the interactions. Because of this limitation, we can only state that the significant interaction effect suggests that the hypothesis is supported, but we cannot ascertain whether or not the interaction has a substantial effect on the outcome. Although we did provide the interaction plot to show the effect, this is only for interpretational purpose, rather than precisely assessing the strength of the interaction and how much variance of the outcome variable has been explained by the interaction term (Aiken & West, 1991;Dawson, 2014). The advancement of Mplus computational capabilities in relation to producing output for latent interaction will enhance future research's ability to provide more precise estimation of such effect.
In relation to the first limitation, we propose that future research could duplicate this study in different research settings, such as in a specific industry. Although there is no evidence to suggest that technological turbulence affects the effectiveness of a firm's network-oriented behaviors, a comparison of the specific configurations of the four different sub-dimensions of network-oriented behaviors between firms in a high-technology industry (high environmental turbulence) and those firms in traditional industries (low environmental turbulence) would provide an interesting avenue for further research. Based on configuration theory, this would also mean that identifying different 'recipes for success', such as equifinal configurations of the four types of networkoriented behaviors within and across industries (Doty, Glick, & Huber, 1993;Meyer et al., 1993;Vorhies & Morgan, 2003), could be a potentially fruitful research direction. A latent class analysis (McCutcheon, 1987) or a qualitative comparative analysis (Rihoux & Ragin, 2009) might serve as a tool to identify the underlying different configurational types of firms, which can be subsequently characterized based on company and industry characteristics. While firm size does not seem to affect the effectiveness of a firm's network-oriented behaviors in our study according to the result of the multi-group analysis, this may be due to the fact that we did not provide analyses based on specific industries. Therefore, future research should include the effect of firm size on the effectiveness of a firm's network-oriented efforts within an industry since networking is often associated with small and medium size firms (e.g., Chetty & Campbell-Hunt, 2003;Ferris et al., 2007;Semrau & Sigmund, 2010).

Constructs
Factor loadings Network-oriented behaviors Information Acquisition (α = 0.89) IA1. We ask our business partners when we need information regarding any of the following: new business opportunities, competition or technology developments in the market. 0.82

IA2.
Information provided by our business partners is helpful for us to make an informed decision. 0.88

IA3.
By speaking to our business contacts, we are able to obtain the information that is crucial to us. 0.80

IA4.
Information from our business contacts who work in a similar market can be useful for us.
We make every effort to go out and network in order to increase our reputation in the market. 0.81

OE2.
We recognize that the value of working well with our business partners adds to the reputation of our products or services. 0.71

OE3.
We invest in building up our reputation in the market by networking with our business partners. 0.88

OE4.
We work toward becoming an effective business partner for other companies in the market (e.g., potential customers or suppliers).

0.88
Strong-tie Resource Mobilization (α = 0.89) SRM1. Matching our suppliers' capacity to the demands of our customers has been an important practice in our organization. 0.81

SRM2.
Our suppliers' ability is critical for us to satisfy our customers. 0.80

SRM3.
Having good relationships with both suppliers and customers has enabled us to adapt to changes in the market place. 0.83

SRM4.
Our customer-focused approach is communicated to suppliers, so that they are aware of how we serve our customers and can contribute to the success of delivering the offerings.

0.85
Weak-tie Resource Mobilization (α = 0.87) WRM1. We initiate relationships with new business partners to gain local knowledge in a new market. 0.79

WRM2.
We interact with the customers of our customers. 0.75 WRM3.
We work closely with influential parties who have relationships with our direct customers to stimulate demand.
0.83 WRM4. Identifying our competitors' major customers helps us to getting to know the needs and requirements of potential customers.

Market-oriented Behaviors
Customer-oriented Behaviors (α = 0.90) CUS1. We closely monitor our level of commitment in serving customers' needs.