Effect of Contextual Factors on Organizational Learning and Knowledge Transfer
As is shown in Figure 29.2, the context interacts with experience to create knowledge. That is, the context moderates the relationship between experience and knowledge creation and transfer. In the sections that follow, we discuss how characteristics of the context can facilitate or impede organizational learning and knowledge transfer in geographically distributed organizations. Thus, we identify contextual conditions that enable organizations to overcome the barriers posed by geographic dispersion. We organize our analyses according to whether the contextual dimension pertains to national, technical, or social factors.
National context
While globalization may in some situations lead to greater commonality across nations (Hirst and Thompson, 1996), national diversity persists on a wide number of dimensions. Some of these dimensions are immediately obvious to the casual observer: different languages, different physical resources, and different legal rules and policy regulations. Other dimensions of national context can be subtle. Nations have different institutions—the formal and informal rules of the game that constrain human interaction (North, 1990). As a consequence of these different institutions, nations also have different national organizations. These organizations can be of many forms including national firms, universities, government laboratories, or other organizations that are part of the government. In addition, in different nations, individuals and organizations also have different local knowledge (Patel and Pavitt, 1994). Combined, the different physical resources, organizational resources, knowledge resources, and rules of the game in each nation lead to very different contexts within which learning and knowledge transfer occur.
There are many examples of the different organizational structures, production environments, and team environments that emerge as a consequence of national contexts. In the US, industries tend to be vertically disintegrated with large firms often outsourcing technology development to small and medium-sized firms (Cohen and Levinthal, 1990; Lamb and Spekman, 1997; Chesbrough, 2003). In contrast, in Japan, firms tend to be vertically integrated, keeping many different functions in-house (Fransman, 1995; 1999). Thus, while knowledge and learning must often be coordinated across many different organizations in the US, this learning is more often centralized in a single firm in Japan.
The most efficient or economically viable production processes and products can also vary widely with national context. In the case of production environments, the Ford system of mass production emerged in the US (Hounshell, 1985), the Toyota production system and lean manufacturing emerged in Japan (Womack, Jones, and Roos, 1990), and the Mexican Maquiliadora system emerged in Mexico (Morris and Pavett, 1992; Vargas and Johnson, 1993; Prasad, Tat, and Thorn, 1995). In many ways, each of these production systems emerged by learning from and adapting the Ford mass production system to better fit a particular national context (Zeitlin and Herrigel, 2000).
In the case of products, research has shown in both the automotive and the photonic semiconductor industries, the most competitive design switches with the choice of manufacturing location, due to national differences in the organization of production (Fuchs, Field, Roth, and Kirchain, 2011; Fuchs and Kirchain, 2010). Significant work remains to understand the extent to which these differences are understood by managers, and the speed at which organizations may be able to learn and integrate these differences into their production decisions. Along these lines, Leonard-Barton (1988) described the implementation of new production technologies as mutual adaptation of technology and organization (Leonard-Barton, 1988). Also, Zeitlin and Herrigel (2000) found that countries that adapted mass production techniques to match the communication and work norms in their national contexts were most successful (Zeitlin and Herrigel, 2000).
Finally, accepted and successful methods of communication and working within and across teams can vary widely across national contexts (Cramton and Hinds, 2005). Working across different national contexts can lead to unexpected costs due to context mismatch (Zeitlin and Herrigel, 2000) and slowed cross-context learning or knowledge transfer (Gibson and Gibbs, 2006), but also can lead to many opportunities for increased learning through the adaptation of existing knowledge and the creation of new knowledge (Leonard-Barton, 1988; Zeitlin and Herrigel, 2000; Cramton and Hinds, 2005). Cramton and Hinds (2005) found that teams with mutual positive distinctiveness were likely to learn from sub-group differences, and developed sophistication in their understandings of cross-national relationships and competence in managing them. We return to the issue of managing differences across geographically distributed groups in our discussion of the social context.
Technical context
Learning and knowledge transfer often occur in a technical context. In the case of this chapter, we use technical context to encompass the extent of uncertainty around a problem, the amount and complexity of information, the architecture (or modularity) of a technology or design, the equipment and tools used in production, and the type of technology used to store or transfer the requisite knowledge. These dimensions of technical context have been shown to be important moderators of the relationship between experience and knowledge when geographic distance is involved.
Several technical contexts are particularly challenging for learning and the transfer of knowledge across geographic distance. Research has shown that technical contexts involving unfamiliar, unstructured problems often require experts to be physically present to recognize embedded clues, exploit specialized tools, use tacit knowledge, and interpret relevant information (Leonard-Barton, 1988; Tyre and von Hippel, 1997; Nadler, Thompson, and van Boven, 2003). The opportunity for in-person problem solving can be critical in production environments (Leonard-Barton, 1988), in particular in the early stages of technology development in chemical and process-based industries such as semiconductors, pharmaceuticals, and standard chemical production (Fuchs and Kirchain, 2010).
Another context involving unfamiliar, unstructured problems is new product development. Managing product development can involve a greater degree of process, marketing, and technical uncertainty than found in many other settings (Anderson, Davis-Blake, Erzurumlu, Joglekar, and Parker, 2008). In this context, ensuring information flows, cooperation, and collaboration can be critical (Anderson, Davis-Blake, Erzurumlu, Joglekar, and Parker, 2008). Thus, new product development can be particularly challenging across geographic distance. At the same time, new product development is a particularly interesting setting for understanding what systems, tools, and frameworks alleviate challenges in learning and knowledge flows created by distance. Previous research has suggested that distributed product development faces greater challenges than non-distributed product development (Sosa, Eppinger, Pich, and McKendrick, 2002; Anderson, Davis-Blake, Erzurumlu, Joglekar, and Parker, 2008). However, contrary to conventional wisdom, recent research has also suggested that project organization types that span country boundaries can outperform co-located insourcing projects, particularly in projects with higher uncertainty (Mishra, 2009). Mishra suggested that this finding may be a result of client firms increasingly leveraging offshore locations for strategic reasons that go beyond cost considerations—a discovery that echoes recent findings that companies are increasingly locating activities offshore to leverage local talent (Lewin and Couto, 2007; Mishra, 2009). Further research will be necessary to disentangle in what contexts distributed teams outperform local ones by leveraging location-specific capabilities versus in what contexts challenges in learning and knowledge transfer across geographic distance mean that co-location is most advantageous.
In addition to affecting the nature of problems, the technical context can also affect the quantity and complexity of the information to be transferred. For example, Von Hippel (1994) argued that successful anticipation and avoidance of all field problems that might affect, for example, a new airplane or a new process machine, require a very large amount of information transfer and, thus, are extremely costly, if not impossible. Likewise, a number of authors have argued that technical contexts with high complexity can lead to knowledge being difficult to communicate or transfer across distance (Patel and Pavitt, 1991; Pavitt, 1999; Novak and Eppinger, 2001). In their study of the automotive industry, Novak and Eppinger suggested that product complexity has three main elements: (1) the number of product components to specify and produce, (2) the extent of interactions to manage between these components (parts coupling), and (3) the degree of product novelty. Variations in product complexity are, in turn, driven by choices in performance, technology, and product architecture. Novak and Eppinger found a significant positive relationship between product complexity and vertical integration (Novak and Eppinger, 2001). Relatedly, Pavitt and Patel (1999) defined complexity as the levels of multi-field knowledge and differentiation required for a given technological competency, and found that multinational firms tend to concentrate activities with high complexity in their home country (Pavitt and Patel, 1999). Notably, however, these relationships may be changing. In particular, recent research suggests that more and more firms are choosing to source their research and development activities overseas, and that with firm experience access to local skills becomes an important driver of overseas location (Lewin and Couto, 2007).
While certain technical contexts can prove particularly challenging or even impossible to manage across geographic distance, other technical contexts can help enable or even support the separation of productive activities across geographic distance. One such technical context is design modularity. A module is a unit whose elements are powerfully connected to one another but weakly connected to elements in other units. Modules have thin crossing points, or points with low transaction costs, at their boundaries; and thick crossing points, or points with high transaction costs, in their interiors (Baldwin and Clark, 2000). Traditionally, product and process modularity have been considered critical to distributed production and product development (Sturgeon, 2002; Sosa, Eppinger, and Rowles, 2004; Gereffi, Humphrey, and Sturgeon, 2005; Eppinger and Chitkara, 2006; MacCormack, Rusnak, and Baldwin, 2008). More recent literature, however, suggests that this relationship may be more nuanced than originally thought (Hoetker, 2006). Colfer and Baldwin (2010) found that almost one third of the cases in their study did not support the hypothesis that organizational patterns, including employment ties, hierarchical groupings, and geographic location and communication links would correspond to the technical patterns of dependency in the system under development (Baldwin and Clark, 2000; MacCormack et al., 2008; Colfer and Baldwin, 2010). In all of the cases where independent and dispersed contributors made highly interdependent contributions to the design of a single technical system, ‘actionable transparency’ was used as a means of achieving coordination (Colfer and Baldwin, 2010). Colfer and Baldwin (2010) defined actionable transparency as the extent to which everyone with an interest in improving a given design has the right and the means to act on it.
In addition, Tripathy and Eppinger (2008) found that most components or processes can neither be termed completely modular or integral at the task level, and suggested an alternative structuring framework for distributed product development. Further, work transfer across distance did not behave monotonically with respect to modularity. Others, however, continue to find that design–interface misalignment has a significant negative impact on project performance, and that this impact is particularly severe when projects are distributed across nations and across firm boundaries (Mishra, 2009). Future work will need to explore the sources of these different findings.
The use of IT to coordinate and collaborate around problems and projects can also ease the difficulties of working across spatial and temporal distances. IT can include synchronous communication methods (i.e. instant messaging, group note-taking or document-editing systems, tele- and video-conferencing), asynchronous communication methods (i.e. email, electronic bulletin boards, web-based software, and data sharing software), information holding databases, and collaborative tools or tools that are designed to automate a manual task. IT tools can provide effective means for communicating and collaborating across distances. This ‘virtual co-location’ can allow collaborators to share information (Hameri and Nihtilä, 1997), provide a ‘technical grammar’ which can create social conventions around collaborating and coordinating (Argyres, 1999), and improve coordination (Sproull and Kiesler, 1991; Yates and Orlikowski, 1992; Sosa, Eppinger, Pich, and McKendrick, 2002).
The extent of integration of technology into processes and routines also matters. Boone and Ganeshan (2001) found productivity benefits from technology that was integrated into the production process but no benefits from technology that just held documents or served as a repository. Ashworth, Mukhopadhyay, and Argote (2004) found that the introduction of IT facilitated organizational learning and knowledge transfer across six geographically distributed units of a financial services firm.
Kane and Alavi (2007) noted that information systems support organizational learning. Hansen, Nohria, and Tierney (1999) argued that the effectiveness of knowledge management systems was contingent on the extent to which work was standardized. In particular, developing detailed databases or knowledge repositories was effective when work was standardized while providing directories that identified member expertise was effective when work was not standardized. These directories facilitated communication, which enabled the solution of non-standard problems.
A couple of studies examined empirically the effectiveness of knowledge management systems. Kim (2008) found that using a knowledge management system contributed positively to the performance of stores in a retail grocery chain. Further, the magnitude of the effect was greater for managers who were remotely located, for those with fewer alternative sources of knowledge and for managers dealing with products that did not become obsolete quickly.
By contrast, Haas and Hansen (2005) found that the number of documents used from a knowledge management system in a consulting firm was negatively associated with consulting team performance. Further, the researchers found interactions between the number of documents used and two contextual variables, team experience and the number of competitors. Using documents from a knowledge management system was particularly harmful for experienced teams and for teams with many competitors. It seems likely that experienced teams already possessed relevant knowledge. Because standard knowledge rather than knowledge that is a source of competitive advantage is likely to populate knowledge management systems, those systems are not very useful in competitive environments. Thus, although knowledge management systems can facilitate organizational learning and have positive effects on organizational performance (Kim, 2008), the positive effect is not guaranteed and depends on important contextual conditions, such as the experience of team members, their alternative sources of information, and their task environment. These factors will be discussed in our section on the social context.
Knowledge management systems are evolving and therefore their capability to contribute to organizational learning and knowledge transfer is also evolving. While early generations of knowledge management systems provided document repositories and directories of declared expertise of organization members, recent systems include communication capabilities and the identification of expertise based on who is consulted and who answers questions. Research is needed on the effects of knowledge management systems with these new capabilities on organizational learning and knowledge transfer.
Social context
By the social context we mean characteristics of the organization’s members and relationships among members. Dimensions of the social context that have been shown to be important moderators of the relationship between experience and knowledge include: the similarity of the contexts; characteristics of members such as their social identity and experience working together; characteristics of relationships among members such as transactive memory systems; leadership; and the organization’s structure, culture, and practices.
The extent to which a context is shared moderates the effect of geographic distance on knowledge transfer. Hinds and Mortensen (2005) found that geographically distributed teams had more conflict than co-located teams. Further, a shared context in which members possessed the same tools, information, and priorities weakened the effect of geographic distance on increased conflict. Thus, providing a shared context to members can be helpful in overcoming the negative effects of geographic distribution on member relations and knowledge transfer.
The similarity of social contexts affects knowledge transfer across them (Bhagat, Kedia, Harveston, and Triandis, 2002). Based on an analysis of three cases of multinational firms, Makela, Kalla, and Piekkari (2007) concluded that similarity in national-cultural backgrounds, language, and organizational status led to more interaction among the members of multinational firms, which increased knowledge transfer. Similarly, Tsai (2002) found that strategic relatedness facilitated knowledge transfer. Darr and Kurtzberg (2000) also found that strategic similarity increased knowledge transfer while they did not find evidence that geographic distance affected knowledge transfer. The units in the Darr and Kurtzberg study were located throughout the UK. Thus, geographic dispersion and language differences may not have been large enough to pose challenges to knowledge transfer. Alternatively, all of the units in the Darr and Kurtzberg (2000) study were affiliated with the same parent corporation, a superordinate structure that may have facilitated knowledge transfer.
When members identify with a superordinate group or organization, they are more likely to transfer knowledge across its units (Kane, Argote, and Levine, 2005). Members who feel that they belong to the same superordinate group are more likely to share information and thoughtfully consider the ideas of others than members who do not feel that they share an identity. Similarly, Sosa, Eppinger, Pich, and McKendrick (2002) found that team interdependence and strong organizational bonds helped overcome the negative effect of distance.
Investigating how a shared identity interacts with geographic distribution, Hinds and Mortensen (2005) found that a shared identity weakened the effect of geographic dispersion on interpersonal conflict. Although geographically distributed teams experienced more task and interpersonal conflict than co-located teams, the effect of geographic dispersion on interpersonal conflict was weaker for teams that shared an identity. In addition, communication weakened the effect of geographic distribution on conflict. Thus, a shared identity and communication hold promise for overcoming the negative effects of geographic distribution.
A transactive memory system can also be useful in mitigating the negative effects of geographic dispersion. A transactive memory system is a collective system for storing, encoding, and distributing information (Brandon and Hollingshead, 2004; Wegner, 1986). In organizations with well-developed transactive memory systems members know ‘who knows what.’ Transactive memory systems have been found to improve the performance of virtual (Kanawattanachai and Yoo, 2007) as well as co-located groups (Austin, 2003; Lewis, 2004; Hollingshead, 1998; Liang, Moreland, and Argote, 1995) and organizations. Further, Borgatti and Cross (2003) found that the effect of geographic distance on knowledge transfer was mediated by knowing what others know and being able to access that knowledge. Thus, a well-developed transactive memory system can overcome the effect of geographic dispersion on knowledge transfer.
Leadership affects the success of most organizational units, including geographically distributed ones. Joshi, Lazarova and Liao (2009) found that inspirational leaders fostered attitudes and relationships critical to the success of geographically distributed teams. Inspirational leadership enhanced members’ trust in each other and commitment to the team, which in turn increased team performance. Further, the beneficial effects of inspirational leadership were stronger in geographically distributed teams than co-located teams.
Experience working together can also help groups overcome the challenges of geographic dispersion. Based on study of geographically distributed software teams, Espinosa, Slaughter, Kraut, and Herbsleb (2007) found that: (1) geographic dispersion had a negative effect on team performance; (2) team familiarity, or experience working together, had a positive effect on team performance; and (3) the positive effect of team familiarity or team performance was stronger for geographically distributed than co-located teams. Thus, team familiarity helped the teams overcome the negative effect of geographic dispersion on team performance.
Moving personnel across ‘donor’ and ‘recipient’ sites can facilitate knowledge transfer (Galbraith, 1990; Davenport and Prusak, 1998; Sole and Edmondson, 2002). Moving personnel is an especially effective mechanism for knowledge transfer because both tacit and explicit knowledge move with people when they move to a new task (Berry and Broadbent, 1984; 1987). Almeida and Kogut (1999) demonstrated that knowledge flowed across organizations through personnel movement. Similarly, Kane, Argote, and Levine (2005) showed that personnel rotation was an effective mechanism for transferring knowledge across groups, especially when members shared an identity.
How members are configured across geographically dispersed groups has important implications for knowledge transfer. Polzer, Crisp, Jarvenpaa, and Kim (2006) studied teams of graduate students from universities in different countries. Teams were assigned to three conditions: fully dispersed (six members in different, unique locations), three sub-groups (two members in each of three unique locations), and two sub-groups (three members in each of two unique locations). Results indicated that trust was higher in the fully dispersed than in the other two conditions, which did not differ significantly from each other. Team conflict was significantly higher in the two sub-group condition than in either the three sub-group condition or the fully dispersed condition, which did not differ significantly from each other.
O’Leary and Mortensen (2010) also analyzed the effect of the geographic configuration, or the number of team members at each location, on group processes and performance. Similar to the Polzer et al. results, O’Leary and Mortensen (2010) found that totally dispersed teams, where members had no teammates at their sites, scored the best on outcomes such as identification with the team, effective transactive memory, coordination, and low conflict. Teams with two or more members per site score weaker on these outcomes than totally dispersed teams. Further, an imbalance in the size of sub-groups (i.e. the uneven distribution of members across sites) invoked a coalitional approach that led to even more negative effects than those observed in the balanced sub-groups.
The organizational structure also affects knowledge transfer. Tsai (2002) found that centralization impaired knowledge transfer in a large multi-unit firm. By contrast, social interactions facilitated knowledge transfer. In a study of multinational firms, Gupta and Govindarajan (2000) found that both formal integrative mechanisms and lateral socialization mechanisms facilitated knowledge transfer.
Organizational practices also affect the success of geographically distributed work. Leonardi and Bailey (2008) analyzed two cases of offshoring, including one case of offshoring from Mexico to India and another of offshoring from the US to India. The Mexican site interacted more directly with India than the US site, which interacted via third-party on-site coordinators or gatekeepers. Two work practices appeared to contribute to successful offshoring at both sites: defining task requirements by monitoring progress of task, and fixing work that was returned because the offshored site did not understand task requirements. From the perspective of the home site, satisfaction with the offshored arrangement was higher under the US gatekeeping model than under the Mexican direct contact model. By contrast, the recipients in India were more satisfied with the Mexican model of direct contact and felt they learned more from it than from the US model. This learning resulted in larger and more complex tasks being offshored from Mexico to India than from the US.
Interactions between national, technical, and social contexts
A given geographically distributed project will invariably have a particular instantiation of national, technical, and social contexts. Due to interaction effects between the different contexts, the specific combination of instantiations of each of these three contexts in a single project has significant implications for how the context moderates the relationship between experience and organizational learning and knowledge transfer. We discuss the possible interaction effects between different national, technical, and social contexts below.
The national and technical contexts found in a project can interact with one another. As discussed earlier, the national context can change the economic viability of particular production processes or products (Zeitlin and Herrigel, 2000; Fuchs, Field, Roth, and Kirchain, 2010; Fuchs and Kirchain, 2010). National context, in the form of culture, can also interact with the technical context, in the form of IT, and affect the extent to which IT can ease the negative impact of geographic distance on organizational learning. Particularly relevant to our discussion is the interaction of IT and national culture. The technology itself and cultural norms around its use can contribute to the challenges and successes of distance collaboration. Although some work has found no difference in use of technology by users from different national cultures (Setlock, Quinines, and Fussell, 2007), more recent work from Setlock and Fussell (2010) found that Asian users valued the support of social processes in addition to task processes in their online work more than American users. Other related work has found varying degrees of difference in technology usage as a function of cultural differences such as the extent of use of video and audio chat features in instant messaging (Kayan, Fussell, and Setlock, 2006), and the extent of participation in video chat based brainstorming sessions (Wang, Fussell, and Setlock, 2009). These differences are commonly related to contextual richness of communication expected in different cultures.
There are other national cultural aspects that affect collaboration when using IT to coordinate across distance. Echoing the findings detailed above, the more task-oriented Americans who are used to short-term teams can find it challenging to work with Europeans or Asians who may value long-term personal relationships (Hall and Hall, 1990). This difference can seem particularly pronounced when video or audio chats are used and norms about the extent of social conversation before and after business discussions are unclear (Olson and Olson, 2000). Speech norms that vary across cultures can create difficult environments for productive conversations over audio and video chats (Olson and Olson, 2000). These challenges in interacting and communicating over new technologies can extend to differences in management style and cultural norms around relationships with management (Hofstede, 1991). The subtleties around these issues can easily be lost over distance, but in some cases overcome with adaptation of the user. For example, if American team members are more conscious to build in pauses after their sentences, particularly in video conferences where timing delays are longer, it allows foreign collaborators more opportunities to comment (Olson and Olson, 2000). Finally, even inconsequential issues, such as different conventions about attire, are apparent on video conferences and can cause misunderstandings and misconceptions (Olson and Olson, 2000).
The social context also interacts with the technical context to affect organizational learning and knowledge transfer. For example, the frequency and timing of meetings and interaction can moderate the effectiveness of technology usage. In a study of three virtual teams within one organization, Maznevski and Chudoba (2000) found that effective global teams had a regular rhythm of face-to-face meetings with virtual communication in the interim. These face-to-face meetings helped build relationships and provided long-term stability and continuity to the teams. In a concurrent engineering context, Loch and Terwiesch (1998) concluded, based on an analytic model, that project characteristics (such as the speed of the evolution of a product design and extent of concurrency between design activities), pre-communication, and uncertainty all contribute to determining communication frequency.
Olson and Olson (2000) provides a rich overview of research on IT and argued that its impact depends on social factors, including common ground, coupling (dependencies) of group work, collaboration readiness, and collaboration technology readiness. They conclude that geographically distributed collaborative work will be challenging for a long time, but as technology improves and work systems are adapted to utilize these evolving technologies, some of the hurdles that existed then (and still do now, a decade later) can be overcome. Common ground, context, trust, different time zones, culture, and interactions of these factors with technology are elements that make distance collaboration qualitatively different than physical collocation.
Social networks are another arena in which the social context interacts with the technical context. Social networks interact with characteristics of the task to affect organizational learning and knowledge transfer. Hansen et al. (1999) found that tacit knowledge was best transferred through strong ties while explicit knowledge was best transferred through weak ties.
The national context can also interact with the social context to affect organizational learning and knowledge transfer. For example, in their study of geographically distributed teams, Polzer, Crisp, Jarvenpaa, and Kim (2006) found that geographic distance between sub-groups led to more conflict and less trust when sub-group members were homogenous with respect to nationality. Thus, their results supported extending Lau and Murnighan’s (1998) concept of ‘fault lines’ to geographically distributed teams. The overlapping of geographic dispersion and national differences strengthened fault lines and exacerbated conflict between groups. Despite such potential challenges, as discussed earlier, teams can also be constructed so as to be more likely to learn from sub-group differences, becoming more sophisticated in their understanding of cross-national relationships and competent in managing them (Cramton and Hinds, 2005).