The search-transfer problem: the role of weak ties in sharing knowledge across organization subunits

Administrative Science Quarterly, March, 1999 by Morten T. Hansen

Why are some subunits in an organization able to share knowledge among themselves whereas others are not? Addressing this question, organization scholars have analyzed factors that inhibit knowledge sharing among subunits, in particular, the lack of direct relationships and extensive communication between people from different subunits (e.g., Lawrence and Lorsch, 1967; Galbraith, 1973; Allen, 1977). More recently, two other lines of research have addressed the topic of knowledge sharing among people in an organization. In the product innovation literature, the argument is often made that close and frequent interactions between research and development (R&D) and other functions, teams, and operational subunits lead to project effectiveness because of the timely integration of knowledge across organizational boundaries (e.g., Clark and Fujimoto, 1991; Leonard-Barton and Sinha, 1993; Henderson and Cockburn, 1994; Eisenhardt and Tabrizi, 1995; Szulanski, 1996). In this literature, efficient knowledge sharing is typically characterized by tight coupling between people from different organization subunits. Some social network scholars, however, provide a different argument. According to the weak-tie theory originally advanced by Granovetter (1973), distant and infrequent relationships (i.e., weak ties) are efficient for knowledge sharing because they provide access to novel information by bridging otherwise disconnected groups and individuals in an organization. Strong ties, in contrast, are likely to lead to redundant information because they tend to occur among a small group of actors in which everyone knows what the others know.

The question thus arises whether it is strong or weak relationships between people in different organizational subunits that lead to efficient knowledge sharing among them. The discrepancy between the different arguments about the effects of relationship strength on knowledge sharing that are proposed in the product innovation literature and the weak tie perspective may be partly due to different foci. Social network research tends to concentrate on the problem of finding relevant information and other resources, a search activity in which weak ties may provide access to new information, while product innovation research tends to focus on the movement of knowledge from various areas in the organization to an R&D team in situations in which people know each other beforehand (a transfer activity). I draw on each line of research to consider knowledge sharing among people from different subunits as a dual problem of searching for (looking for and identifying) and transferring (moving and incorporating) knowledge across organization subunits, taking into account the complexity of the knowledge that flows through interunit relationships.

Although several social network scholars have argued that weak ties only provide information benefits under certain conditions and are less beneficial than strong ties in providing socio-emotional support and solving conflict, social network research has largely remained agnostic with respect to the content of what flows through instrumental relations between actors (Nelson, 1989; Wegener, 1991; Krackhardt, 1992; Podolny and Baron, 1997). Whether it is simple information or richer forms of knowledge (e.g., a complex technology) that flows through the ties has not been studied. In contrast, researchers studying product innovation have analyzed the difficulties in transferring complex knowledge, including noncodified or tacit knowledge (Teece, 1977; Zander and Kogut, 1995) and components that are dependent on larger systems (Winter, 1987). When such complex forms of knowledge are considered, the instrumental benefits of weak ties are called into question. Weak ties may lead to search benefits in a social network but they may also cause problems in transferring complex forms of knowledge.

I limit my discussion to one task undertaken by subunits in many multiunit firms - new product development. A product development team situated in an operating unit can use established interunit relations - which exist prior to the start of the project - to search for and transfer to the project various types of knowledge residing in other operating units. For simplicity, I define interunit relations as regularly occurring informal contacts between groups of people from different operating units in an organization. I confine the discussion to relations between operating units, such as divisions, and not functional departments. These informal relations may have an effect on the time it takes to develop a new product from concept development to market introduction by affecting the ease with which project team members search for and transfer knowledge across subunits. Although project completion time is only one dimension of the effectiveness of new product development (Wheelwright and Clark, 1993), it has become an important outcome measure in many industries, notably in the electronics and computing industries (Eisenhardt and Tabrizi, 1995) and is a relevant measure for the empirical study of the electronics and computer company reported in this paper.(1) Product development time captures both the benefits and costs of sharing knowledge across other organization subunits. On one hand, a project team stands to benefit to the extent that it obtains useful knowledge from other subunits, shortening completion time. Knowledge from other subunits can help projects avoid duplication of efforts (e.g., using an existing software module) or provide them with complementary expertise, as when an expert helps solve a technical problem (Teece, 1986). Such knowledge, as defined here, includes product-specific technical know-how, knowledge about technologies and markets, as well as knowledge embodied in product components (e.g., in a software module). On the other hand, the project team may expend considerable search time and transfer efforts to be able to use fully the knowledge from other subunits. If search and transfer take a long time, then knowledge sharing is likely to hamper the performance of the tasks, prolonging completion time.