Theoretical Framework

THE EFFECTS OF SOCIAL CAPITAL ON FIRM SUBSTANTIVE AND SYMBOLIC PERFORMANCE IN THE CONTEXT OF E-BUSINESS

Liu, Hefu*, Ke, Weiling, K. K., & Hua, & Lu, Yaobin 2016. The effects of social capital on firm substantive and symbolic performance: In the context of E-business. Journal of Global Information Management, 24(1): 61-85.

Abstract While it is well documented that social capital plays a critical role in affecting firm performance, the diffusion of Internet-based applications challenges our understanding of social capital. In this study, we examine the effects of social capital in the context of e-business. Specifically, we investigate how each of the three dimensions of social capital (i.e., structural, relational and cognitive) differentially influences a firm’s substantive and symbolic performance. We also explore how structural capital and cognitive capital indirectly affect firm performance through relational capital. The research model is generally supported by data collected from a survey administered to 205 firms in China. The results suggest that structural and relational capital positively influence substantive and symbolic performance, respectively. However, cognitive capital does not have significant effects on substantive performance, though it positively affects symbolic performance. Also, we found that structural capital and relational capital have stronger effects on substantive performance than symbolic performance. In contrast, cognitive capital has stronger effects on symbolic performance than substantive performance. Further, both structural capital and cognitive capital positively affect relational capital. We conclude with implications and suggestions for future research. Keywords: Social capital, substantive performance, symbolic performance, Internet

1. INTRODUCTION Social capital plays a critical role in affecting firm performance (Nahapiet and Ghoshal 1998; Baker 1990). As a set of resources rooted in relationships with stakeholders in a firm’s organizational field, social capital allows the firm to exchange information and knowledge, access resources required for development and achieve legitimacy status (Nahapiet and Ghoshal 1998). As such, social capital is regarded as a crucial predictor of performance and an enduring source of competitive advantage (Arregle et al. 2007; Carey et al. 2011; Moran 2005; Stam et al. 2014). For example, Batjargal and Liu (2004) posit that social capital significantly affects venture capitalists’ investment selection decisions. Also, Acquaah (2007) found that social capital developed from social networks with other firms, government officials and community leadership has positive effects on organizational performance in emerging economies. Similarly, Krause and Handfield (2007) found that social capital is directly related to firm performance improvement in the US automotive and electronics industries. All these studies are instrumental in providing us with knowledge about social capital. However, the findings and propositions of prior research may not be directly applicable to the Internet era. The Internet, with its open connections and low switching cost, is reconstructing managerial social networks and ties (Li et al. 2010; Porter 2001). The diffusion of the Internet is pushing organizations to reconstruct the fragmented, silo-oriented network with an open, real time, broad partner basis, low switching cost, and globally connected format (Boyer & Olson 2002; Liu et al. 2010). Yet, it also breeds more potential threats by allowing partners to switch from the existing relationship and to cooperate with others more easily within global markets (Porter 2001). Thus, the role played by social capital in the Internet era is important and it warrants scrutiny (Chiu et al. 2006; Uslaner 2000).

A careful review of the literature shows that the existing empirical research on firms leveraging Internet-based applications has mainly focused on only the structural dimension of social capital. According to Nahapiet and Ghoshal (1998), social capital has three dimensions, namely, structural, relational and cognitive. Structural capital refers to the overall pattern of network structures (Krause et al., 2007; Nahapiet and Ghoshal 1998). Relational capital reflects the nature of relationship, such as trust and norm of reciprocity (Xiao et al. 2007; Maurer et al. 2006; Nahapiet and Ghoshal 1998). Cognitive capital represents shared meaning and understanding between the network members (Nahapiet and Ghoshal 1998; Inkpen and Tsang 2005). In the view that all three dimensions are critical components of social capital and each plays a role in affecting the amount of resources a firm can acquire, integrate and transfer within its networks, an investigation considering only one dimension of social capital the Internet era is incomplete and can be dubious. Also, prior research has been primarily focused on assessing social capital’s effect on the firm’s substantive performance and ignored its effect on symbolic performance (e.g., legitimacy). Meanwhile, the positive impacts of social capital on substantive performance have not gained consistent empirical support although such effects are well touted theoretically. For example, the findings on the role of supply chain integration (SCI), a form of structural capital, are mixed and even controversial (Flynn et al., 2009; Frohlich et al., 2001; Rai et al., 2006). While a few studies indicate that SCI improves substantive performance (e.g., Cagliano et al. 2006; M. T. Frohlich & Westbrook 2001), others found that reaping benefits from SCI is more rhetoric than reality (e.g., Fawcett & Magnan 2002; Power 2005). Given such limitations with the extant literature, it is important to differentiate types of firm performance consequences and examine how social capital may affect substantive and symbolic performance, respectively. Indeed, institutional theorists indicate that cognitive capital, such as conformity to social norms, not only helps an organization to improve

substantive performance, but also symbolic performance (Heugens & Lander 2009). Also, the open and interconnect nature of Internet may make firms easier to achieve symbolic benefit than substantial benefit from social capital (Greenwood et al., 2002). Thus, a research differentiating substantial and symbolic performance would help us resolve the inconsistency in the extant literature and crystallize how social capital affects firm performance. Further, the extant literature on social capital has mainly treated structural, relational and cognitive dimensions of social capital as independent of each other. However, these three dimensions may operate interdependently to some degree (Chiu et al., 2006; Krause et al., 2007). Lee et al. (2004), for example, indicate that high level of relational capital could facilitate the organization to benefit from structural capital in the firm’s adoption of Internet-enabled SCI. As such, exploring the possible interrelationships among social capital’s three dimensions would shed new light on our understanding. In this research, we intend to address the above mentioned shortfalls in the literature. Drawing upon social capital theory, institutional theory and information systems literature, we develop our research model to investigate how the three dimensions of social capital would differentially affect a firm’s substantive and symbolic performance, respectively. Specially, in addition to arguing that structural, relational and cognitive capital has positive effect on substantive and symbolic performance, we contend that structural capital’s effects on substantive performance would be stronger than on symbolic performance. In contrast, we posit that relational and cognitive capital’s effects on substantive performance are weaker than those on symbolic performance. Also, we hypothesize that both structural and cognitive capital have positive effects on relational capital. The research model is tested with data collected in China as the context of e-business in China is of particular interest to researchers and practitioners due to two reasons (Batjargal 2007). First, research on Chinese institutional, regulatory, and human capital issues remains insufficient. Second, China’s relational, intense

cultural milieu makes social networks even more dynamic and more important for performance (Bat Batjargal 2007a; Xiao & Tsui 2007). We expect that the research will make the following contributions. First, it will improve our conceptual and operational understanding of social capital in the context of e-business. Be examining the appropriate components of each dimension of social capital in e-business, we intend to make the related constructs richer, though we do not have the intention to provide an exhausted list of all their components. Second, this study advances theory on the consequences of social capital. By differentiating substantive and symbolic performance and examining how social capital’s dimensions have different effects on them, this research enriches the literature by highlighting the importance of legitimacy for the firm in its organizational field. Third, this study provides theory-based managerial guidelines for developing social capital to maximize organizational benefits in the context of e-business. Given that numerous global firms are operating in China, this research will benefit them by offering insights into better managerial social networks management.

2. THEORETICAL BACKGROUND In the context of e-business, Internet-based social network is becoming the new era of supply

chain management. It challenges our current understanding of the impacts of social capital on firm performance. Specifically, the Internet makes managers face a new, dynamic, and information-rich business environment (Granados et al. 2010; Porter 2001; Zhu et al. 2006). Such an environment challenges firms’ cognitive capacity, leading to information overload for decision makers (Jones et al. 2004; Malhotra et al. 2005). While the Internet enables the firm to set up connections with supply chain partners across the world economically, the barriers imposed by culture and values differences may not be overcome easily (Dong et al., 2009). As such, the social network in which the firm is embedded becomes more complex

and diverse than in the pre-Internet age, which may make the benefits that could be easily materialized in the old days become far reaching. Indeed, scholars attribute the mixed findings on the effects of SCI to the complexity and challenges imposed by social factors such as mistrust and lack of commitment (Chiu et al. 2006; Patnayakuni et al. 2006). Also, while the Internet-enabled collaboration relationships allow the firm to market and distribute their products and services worldwide, the Internet also makes the firm’s competitors and customers better informed. It simply makes it difficult, if not impossible for the firm to conduct rent seeking by taking advantage of information asymmetry. In other words, in the context of e-business, social capital embedded in the network could become a double-edged sword. In particular, information transparency enabled by the Internet may prevent the firm from generating superior profits (Granados et al. 2010). Therefore, it is imperative to investigate how social capital affects firm performance in the e-business context (Dong et al. 2009; M. T. Frohlich & Westbrook 2001; Power & Singh 2007; Rosenzweig 2009). Since Coleman (1988) introduced the concept of social capital, it has evoked considerable interest. According to Nahapiet and Ghoshal (1998), social capital has three dimensions, namely the dimensions of structural, relational, and cognitive capital. The three dimensions reflect valuable assets for an organization, since they enable an organization to exchange a variety of information, knowledge, and other forms of capital through managerial social networks and ties (Krause et al. 2007; Villena et al. 2011). Specifically, structural capital reflects the overall pattern of network structures, such as a supply chain that connects customers and suppliers (Jansen et al. 2011; Krause et al. 2007; Nahapiet & Ghoshal 1998). In the existing literature, scholars increasingly propose the managerial social networks and ties, in general, and SCI in particular, as the specific structural capital. Meanwhile, due to the diffusion of the Internet in reconstructing managerial social networks, scholars and practitioners tend to focus on Internet-enabled SCI (IeSCI). It is well documented that IeSCI

involves the dimensions of online supplier integration, online customer integration, and online internal integration. Online supplier and customer integration reflect the extent to which a firm with its suppliers or customers to structure inter-organizational information, practices, and processes into interconnected, synchronized processes via the Internet (Rai et al. 2006); Online internal integration refers to the degree to which a firm structure its own organizational information, practices, and processes into interconnected, synchronized processes via the Internet. Relational capital reflects the nature of relationship, which is behavioral, as opposed to structural (Krause et al. 2007; Lawson et al. 2008b; Moran 2005; Nahapiet & Ghoshal 1998). Other than deriving value from its network configuration, an organization can benefit from the nature of relationships in the social network (Krause et al. 2007). In particular, relational capital allows organizations to govern relations based on trust, rather than strict formal contracts (Patnayakuni et al. 2006), which can help to reduce transaction cost (Williamson 1980). Based on its attributes, relational capital reflects organizational trust and commitment (Krause et al. 2007; Maurer & Ebers 2006; Xiao & Tsui 2007). Moreover, scholars argue that guanxi is a special and critical dimension of relational capital in Chinese culture (B. Batjargal 2007b; Gu et al. 2008; Xiao & Tsui 2007). Specifically, trust refers to one party’s willingness to depend on another party based on the anticipated beneficial behavior of that party (Ke et al. 2009), commitment is defined as a firm’s duty to undertake some activity for the relationship in the future (Nahapiet & Ghoshal 1998), and guanxi refers to “the durable social connections and networks a firm uses to exchange favors for organizational purposes” (Gu et al. 2008 p.12). It is well established that relational capital helps the firm improve performance as it plays a critical role in deciding the extent to which the firm can access resources embedded in the social network. Specifically, when the relational capital is high, the nature of the firm’s relationships with its supply chain partners and other related stakeholders such as

governmental agencies is more favorable, thereby making it more likely for the firm to access the resources at a lower cost (Gu et al. 2008; Acquaah 2007). As a third dimension of social capital, cognitive capital would be presented when members of a network develop shared meaning and understanding (Krause et al. 2007; Sherif et al. 2006). With the shared meaning and understanding, members can develop a self-reinforcing sense making process across the network (Krause et al. 2007). The concept of shared values has been widely treated as the critical dimension of cognitive capital (Inkpen & Tsang 2005; Krause et al. 2007). Shared values refer to the shared understanding and beliefs about what behaviours, goals and policies are important or unimportant, appropriate or inappropriate, and right or wrong among members of a network (Krause et al. 2007). It serves as norms of behaviour governing the relationships. As another component of cognitive capital, data consistency is proposed to be critical for the context of e-business (Chiu et al. 2006; Patnayakuni et al. 2006). With information technology (IT) playing an enabling role in extended social networks of supplier and customers, the parties involved need to reach mutual agreement in term of the syntactical, semantic and pragmatic meaning of data transmitted across the network. Data consistency refers to the degree of the establishment of common data definitions and consistency in shared data across a relational network (Rai et al. 2006). With data consistence, the firm is equipped with the capability to process shared information in an efficient and effective manner (Im and Rai 2008). While the positive effects of social capital have been well recognized more than two decades (e.g., Burt 1997; Leana and van Buren 1999; Nahapiet and Ghoshal 1998), it has been applied to the context of e-business only in recent years. For example, Cai and Zhang (2011) found that the collaborative linkages with supply chain partners help the firm achieve good performance. Flynn et al. (2010) investigate how configuration of intra- and inter-organizational processes allows the firm to enhance flows of product and services and

provide value to the customers. Yet, the research on the effects of IeSCI has primarily focused on the structural dimension of social capital. As discussed above, relational and cognitive dimension of social capital would also affect firm performance and this notion can be extended to the context of e-business. Further, social capital in the context of e-business may affect the focal firm’s symbolic performance as well as its substantive performance. Substantive performance refers to a firm’s performance in economic terms that can be measured by tangible value. In contrast, symbolic performance is defined as the extent to which a firm or its actions are regarded as legitimate in its organizational field (Greenwood 2002). Due to the network effects of e-business platforms, it is well recognized that firms should set up electronic linkages with supply chain partners and establish routines that take advantage of the new business opportunities that are provide by the Internet. According to the institutional theory, an organization’s expectations about the performance implication of a business practice are socially constructed and thus influenced by the stakeholders in its organizational field (Martinez et al. 1997; Roberts et al. 1997). In other words, social capital in the context of e-business would help the firm achieve symbolic performance in e-business. In the view that prior research has focused on investigating social capital’s effects on substantive performance, we extend the literature by considering symbolic performance.

3. RESEARCH FRAMEWORK AND HYPOTHESES Figure 1 shows the theoretical model that we develop based on the existing social capital, institutional, and managerial literature. Our theoretical model specifies how the three dimensions of social capital, namely structural, relational and cognitive capital affects firm performance. Based on a critical review of the literature against our investigative context and objectives, we posit that social capital have differential effects on substantive and symbolic

performance in the context of e-business. Also, we contend that structural, relational and cognitive social capital are interrelated and they can affect each other. In particular, we conjecture that structural capital and cognitive capital have positive influence on relational capital.

Figure 1. Conceptual framework

The impacts of structural capital in general, and IeSCI in particular, on substantive performance has gained support from previous studies (e.g., Devaraj et al. 2007; Flynn et al. 2010; M.T. Frohlich 2002). With the IeSCI, the focal firm can access, exchange and transfer information and knowledge with supply chain partners, which would allow the firm to develop new business process capabilities. Also, IeSCI enables the firm to work closely with stakeholders in product and services development. For instance, through the Internet-based platform, the firm can leverage the innovative ideas of its supply chain partners across the world. Indeed, there is consistent empirical support for the positive effects of structural social capital on the firm’s substantive performance. For example, Lee and Whang (2004) contend that the IeSCI indicates tight structural coordination and collaboration within the related network. Cagliano et al. (2006) and Fabbe-Costes et al. (2007) thus suggest that IeSCI has positive influence on operational performance while Vickery et al. (2003) posit that IeSCI helps the organization to improve customer service. Thus, we expect that online information

integration, online planning synchronization and online operational coordination have positive relationships with substantive performance. In addition, we expect structural capital can improve symbolic performance. Scholars indicate that the prevalence of the Internet is leading a new era of supply chain management, in particular it enables IeSCI (Devaraj et al. 2007; Esper et al. 2010; M.T. Frohlich 2002; M. T. Frohlich & Westbrook 2002). In this view, conducting IeSCI would allow the focal firm to convey a message that it is making a rational and effective decision within the network. As mentioned above, the stakeholders in the firm’s organizational field would identify the focal firm’s behaviors and have positive attitude to the firm, which would enhance the firm’s legitimacy and thus positively affects its symbolic performance (Greenwood et al. 2002). Further, institutional theorists suggest that stakeholders in the focal firm’s organizational fields are usually indifferent to certain amounts of differentiation and thus allow organizations a “range of acceptability” around the ordained template (Deephouse 1999). In other words, the enhancement of symbolic performance due to structural capital is attenuated compared to the actual economic benefits the focal firm can derive. As such, the impact of structural capital on symbolic performance is weaker than that on substantive performance in the context of e-business. Accordingly, we hypothesize the following.

H1a: An organization’s structural capital has positive effects on its substantive performance H1b: An organization’s structural capital has positive effects on its symbolic performance H1c: The effects of structural capital on substantive performance are stronger than on symbolic performance.

According to social capital theory, relational social capital can lead to superior substantive performance. Specifically, with the relationships characterized by trust, the firm can expect reliable and benevolent behaviors from other parties in the social network. The firm is willing to be vulnerable and enter into collaborative relationships rather than getting stuck in drafting a complete and detailed contingent contract to protect its own interest (Nahapiet and Ghoshal 1998; Florin and Lubakin 2003; Acquaah 2007). Also, commitment to the social network enables the firm to focus more on the development of long term relationship and avoid opportunistic behaviors. Further, guanxi allows the firm to get the access to information and resources that are typically unavailable to its counterparts. Therefore, relational social capital would have positive effects on the firm’s substantive performance. This thesis has gained consistent support from a plethora of empirical studies (Krause et al. 2007; Lawson et al. 2008a; Min et al. 2008; Villena et al. 2011). In the context of e-business, the open and lean nature of e-business, compared to the traditional business relying on face-to-face interactions, would lead to even higher uncertainties about the trading parties’ identities, partners’ opportunistic behaviors, and network stability (Ba & Pavlou 2002; Zhu et al. 2006). High relational capital could allow organizations to avoid the opportunistic behaviors within the network, to decrease business risks, and to improve operational performance and customer service (Ba & Pavlou 2002; Jøsang et al. 2007; Villena et al. 2011). Thus, we expect that relational capital will be critical drivers for substantive performance improvement in the context of e-business. Relational capital is also an important source of an organization’s legitimacy (Hitt et al. 2002; Tang 2009), which is the essence of symbolic performance (Heugens & Lander 2009). Specifically, high relational capital could enhance an organization’s image of trustworthiness and acceptance, the critical nature of legitimacy, in its organizational field (Lu & Xu 2006).Also, the open and global connection nature of the Internet helps to diffuse such image

within the organization’s network. Therefore, we expect that symbolic performance can be significantly affected by relational capital in the context of e-business. Further, Park and Luo (2001) note that relational capital helps an organization to position itself in the market, such effects would be amplified by the wide connections enabled by the Internet. In particular, relational capital enables the firm to collaborate with partners through a variety of mechanisms that involve online and offline resource reconciliation and orchestration. As such, relational capital would have an even larger influence on the firm’s market shares and substantive performance in the context of e-business. Accordingly, we propose that the impact of relational capital on substantive performance would be stronger than on symbolic performance.

H2a: An organization’s relational capital has positive effects on its substantive performance. H2b: An organization’s relational capital has positive effects on its symbolic performance H2c: The effects of relational capital on substantive performance are stronger than on symbolic performance.

Cognitive social capital, including shared values and language within a network, has been regarded as a desirable weapon to improve substantive performance (Inkpen & Tsang 2005; Krause et al. 2007; Lee 2007; Maurer & Ebers 2006; Patnayakuni et al. 2006; Rai et al. 2006). Specifically, shared values and language could allow an organization to decrease the potential conflicts and misunderstanding within its network, which allows the firm to obtain requested resources and knowledge to improve operational performance and customer service (Jap &

Anderson 2003; Rossetti & Choi 2005; Zaheer et al. 2000). Indeed, empirical studies have lent support to this argument (Krause et al. 2007). In the context of e-business, cognitive capital would play an even more important rule in affecting firm performance because of the open standards and global connection feature of Internet (Zhu et al. 2006). With the Internet connecting firms with diverse cultural backgrounds located in different regions and nations, the parties involved could easily have their own schemas to interpret the world in general and what happens within the network in particular (Zhu et al. 2006). With shared values, the network would have norms governing appropriate behaviors and thus can avoid certain partners rigidly pushing their own agenda in their own ways (Inkpen and Tsang 2005). Also, data consistency, with syntax, semantic and pragmatic standards, enables mutual understanding about the definition, meaning and operations of data and information exchanged within the network (Rai and Tang 2010). It helps the firm to develop efficient and effective business processes within and across the organizational boundary (Wang et al. 2013). Thus, we expect that high cognitive capital would lead to superior substantive performance in the context of e-business. According to the institutional theory, conformity to social norms can improve an organization’s symbolic performance (Heugens & Lander 2009). With cognitive capital, the firm can better understand and accept the collective expectations to conduct work professionally, and to establish legitimization successfully (DiMaggio & Powell 1983). Also, coupled with the open nature of the Internet, cognitive capital facilitates the diffusion of the trustworthiness image of the firm and makes the firm more widely accepted by other parties in the network. Thus, we expect that cognitive capital has positive influence on symbolic performance. Further, we follow the institutional theory and argue that cognitive social capital’s impacts on symbolic performance would be stronger than that on substantive

performance (Heugens & Lander 2009; Meyer & Rowan 1977). As such, we have the following hypotheses.

H3a: An organization’s cognitive capital has positive effects on its substantive performance. H3b: An organization’s cognitive capital has positive effects on its symbolic performance H3c: The effects of cognitive capital on substantive performance are weaker than on symbolic performance.

Recent research has revealed that dimensions of social capital are not independent of each other (Krause et al. 2007; Chiu et al. 2006). Accordingly, we propose that structural capital, in the form of IeSCI, could directly affect relational capital. Specifically, IeSCI connects the parties in the network, allows them to communicate, coordinate and coordinate in a more efficient way (Zhu et al. 2006). The increased amount of interaction could help develop relational capital by building trust, commitment, and guanxi (Wasko and Faraj 2005). Also, IeSCI involves a lot of interconnected and synchronized processes, which would help firms to avoid misunderstanding, and have better knowledge about each other’s purposes, tactics and strategies. In addition, we contend that cognitive capital positively influences relational capital. With shared values and data consistency within the network, related parties’ behaviors become more predictable (Nahapiet and Ghoshal 1998). Also, with the shared values and data consistency, firms are more likely focus on the long term development of the network and thus would have higher level of commitment toward the collaborative relationships (Xiao and

Tsui 2007). Further, with the mutual understandings about what is and should be happening within the network, firms would be more likely to back up each other and have better guanxi (Gu et al., 2008; Tsai and Ghoshal 1998). Accordingly, we hypothesize the following.

H4a: An organization’s structural capital has positive effects on its relational capital. H4b: An organization’s cognitive capital has positive effects on its relational capital.

4. RESEARCH METHODS 4.1 Sample and data collection

We collected data using the designed questionnaire survey to test our proposed model. This survey is conducted in China. We randomly chose 1000 firms from lists provided by local institutions and government officials, such as industrial park administrators. The lists included contact information on key informants, such as CEOs, senior vice presidents for operations management, and chief technology officers. We sent invitations to key informants to participate in the survey. These senior executives are at the top positions in firms, so they tend to have a more comprehensive view of their firms to answer all questions. We explained the process of obtaining their contact information and the objectives of the research before inviting them to participate in our study. We sent questionnaires to the potential respondents after receiving their consent to participate in the survey. Follow-up phone calls and reminder e-mails were sent two weeks after the questionnaires were delivered. A total of 270 questionnaires were returned, including 65 incomplete questionnaires, which were discarded. A total of 205 useful questionnaires were obtained, a response rate of approximately 20.5%. Table 1 shows the demographic information of the

sample. The use of a single respondent may not be ideal for firm-level studies, but this approach is common among empirical studies, such as those investigating management issues and firm performance (Cao & Zhang, 2011; Liu et al., 2013). The senior executives were knowledgeable about the issues examined in this research, such as social capital, and can access information about the market and the performance and competitor situations of their firms. We tested for potential non-response bias according to Armstrong and Overton (1977). We compared demographic information about the responses from the first 25% of the respondents and that of the final 25% through a chi-square test. The chi-square value had no significant difference between the two groups on these demographic items, indicating that non-response bias is not an issue in this study. Measure Ownership types

Industry types

Firm size

Firm history

Items State-owned Privately Owned Foreign-controlled Joint Venture Others Mechanical Equipment Manufacturing Finance Industry Wholesale and Retail Trade IT Industry Real Estate Industry Construction Industry Others Less than 100 100-299 300-499 500-999 1000-1999 More than 2000 1-5 Years 6-10 Years 11-25 Years 26-50 Years More than 51 Years

Frequency 68 67 11 6 5 20 35 9 11 13 12 57 37 46 16 17 13 28 25 43 61 13 15

Percentage 43.31% 42.68% 7.01% 3.82% 3.18% 12.74% 22.29% 5.73% 7.01% 8.28% 7.64% 36.31% 23.57% 29.30% 10.19% 10.83% 8.28% 17.83% 15.92% 27.39% 38.85% 8.28% 9.55%

Less than 2 2-5 IT department size 6-10 11-15 More than 16 Table 1. Sample demographic

36 48 28 8 37

22.93% 30.57% 17.83% 5.10% 23.57%

4.2 Measures The measures used for the constructs in our research model are shown in Appendix. Instrumentation was developed based on previously validated measures. The domain of each construct was clearly defined in terms of the items that should be included or excluded. Related literature was also reviewed to find relevant scales. The measures were assessed with a five-point Likert scale that ranged from “strongly disagree” to “strongly agree.” We invited three academicians from fields related to social capital and IT to review the initial questionnaire and offer feedback. According to their comments, we removed items that were irrelevant to the domains of the designated constructs, and modified the wording of some items. A translation committee of bilinguals translated the English instruments into Chinese because the survey was conducted in China. Three native Chinese speakers who are fluent in English were invited to translate the questionnaire into Chinese. An expert with no knowledge of the research was also invited to back-translate the questionnaire into English to ensure that the Chinese and original English versions had no semantic discrepancies.

5. DATA ANALYSIS AND RESULTS 5.1 Common Method Bias Since all data were perceptual and collected from a single source simultaneously, we checked the possible common method bias using Harman's one-factor test to get rid of the

threat to the research validity. The results revealed that all the items can be categorized into six distinct factors with eigenvalues above 1.0, and account for 71.307% of the variance. The first factor didn’t account for the majority of the variance (only 14.955%), which indicated that the common method bias was unlikely to be a major issue in our study.

5.2 Reliability and Validity We employed confirmatory factor analysis (CFA) to assess the construct reliability and validity of the measurement. Specifically, we assessed the reliability of each construct following the guidelines outlined by (Fornell & Larcker 1981). The results showed that Cronbach’s Alpha ranged from 0.75 to 0.92, and composite reliability ranged from 0.86 to 0.94. As shown in Table 2, these values of all constructs were higher than the benchmark of 0.700, which indicated the good reliability of the measurements. Further, we tested construct validity via convergent and discriminant validity. The convergent validity was tested by the items’ loadings and average variance extracted (AVE). As Table 2 reports, the loadings of all items varied from 0.71 to 0.93 at a significance level of 0.001, and the AVE values ranged from 0.63 to 0.80, which were above the 0.500 recommended level. The results showed that the measures had satisfactory convergent validity. To assess the discriminant validity, we compared the relationship between the correlations among the constructs and their square root of AVEs. The data in Table 3 indicated that the square roots of AVEs for all constructs were higher than the correlations between constructs, which verified the discriminant validity of the measurement model. As one inter-construct correlation in Table 3 had value over the criteria of 0.60, we

conducted a test to verify the potential threat of multicollinearity. Generally, if variance inflation factors (VIFs) are greater than 10 or tolerance values are less than 0.10, the existence of multicollinearity is proved (Mason & Perreault Jr 1991). The results of our study showed that the highest VIF was 2.077 and the lowest tolerance value was 0.482. Thus, multicollinearity did not appear to be a significant problem in our study. Loadings Cronbach’s Composite AVE range Alpha Reliability Cognitive capital Data consistency 0.78-0.88 0.84 0.90 0.68 Shared values 0.83-0.88 0.82 0.89 0.73 Structural capital Customer integration 0.78-0.83 0.82 0.88 0.65 Internal integration 0.71-0.84 0.80 0.87 0.63 Supplier integration 0.78-0.86 0.75 0.86 0.67 Relational capital Affective trust 0.87-0.91 0.86 0.92 0.78 Cognitive trust 0.85-0.90 0.86 0.91 0.78 Business ties 0.78-0.84 0.82 0.88 0.65 Political ties 0.87-0.91 0.87 0.92 0.79 Commitment 0.87-0.90 0.86 0.91 0.78 Substantial performance Business performance 0.80-0.89 0.92 0.94 0.72 Operational performance 0.78-0.84 0.89 0.91 0.64 Symbolic performance Regulatory endorsement 0.85-0.93 0.88 0.93 0.81 Media endorsement 0.87-0.92 0.87 0.92 0.80 Agency ratings 0.80-0.87 0.87 0.91 0.72 Table 2. Results of Confirmatory Factor Analysis (CFA). Second-order

Table 3.

First-order

Mean, standard deviation, and correlation Mean SD 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1. Data consistency 3.15 0.91 0.83 2. Shared values 3.57 0.78 0.48 0.86 3. Customer integration 3.43 0.78 0.52 0.67 0.81 4. Internal integration 3.38 0.78 0.38 0.46 0.51 0.80 5. Supplier integration 3.50 0.75 0.45 0.50 0.64 0.52 0.82 6. Affective trust 3.66 0.77 0.35 0.59 0.59 0.37 0.50 0.88 7. Cognitive trust 3.66 0.81 0.37 0.58 0.59 0.35 0.50 0.79 0.88 8. Business ties 3.64 0.77 0.42 0.54 0.59 0.48 0.53 0.58 0.55 0.81 9. Political ties 3.65 1.00 0.25 0.30 0.40 0.30 0.33 0.34 0.26 0.55 0.89 10. Commitment 3.85 0.80 0.34 0.55 0.55 0.39 0.42 0.67 0.63 0.60 0.37 0.88 11. Business performance 3.64 0.80 0.43 0.49 0.53 0.43 0.42 0.39 0.34 0.50 0.45 0.47 0.85 12. Operational performance 3.66 0.76 0.38 0.51 0.56 0.54 0.48 0.58 0.56 0.54 0.32 0.62 0.51 0.80 13. Regulatory endorsement 3.55 1.00 0.34 0.41 0.41 0.43 0.38 0.44 0.35 0.51 0.49 0.42 0.49 0.44 0.90 14. Media endorsement 3.67 0.97 0.29 0.37 0.37 0.35 0.36 0.42 0.36 0.50 0.53 0.41 0.44 0.37 0.73 0.89 15. Agency ratings 3.64 0.88 0.37 0.50 0.51 0.43 0.43 0.46 0.35 0.55 0.49 0.47 0.54 0.50 0.78 0.74 0.85 *. Correlation is significant at the 0.05 level (2-tailed). Table 3. Assessment of discriminant validity

5.3 Hypothesis testing Given that all the constructs in the model are formative constructs, we used the partial least squares (SMARTPLS) analysis technique to test the research model. Fig. 2 provides the results of the weights of the sub-constructs of formative constructs. The existing literature has presented that the weights of the sub-constructs can act as the beta coefficients in a regression model. Thus, their values would normally be lower than loadings. Under this condition, to retain the content validity, using insignificant indicators in the model should be acceptable. In this view, although the weight of regulatory endorsement and media endorsement of symbolic performance were insignificant, we kept them to retain the content validity of symbolic performance. Fig 2 further presents the results of the structural model, which includes the R2 of endogenous variables and the structural path coefficients and their significance. The model explained 39 to 56 percent of the variances. The results showed that H1a and H1b which demonstrated the relationship between structural capital and substantive performance (β=0.32, p