Authors:
Muge Ozman | Telecom Ecole de Management, Institut Mines Télécom | France
Andrew Parker | University of Exeter Business School | United Kingdom
Previous research on knowledge transfer within organizations suggests that having closed networks facilitates the flow of sticky knowledge (Hansen, 1999). In contrast, Burt (2004) shows the benefits of having structural holes, i.e., open networks, for access to good ideas. The research on boundary spanners is mixed with evidence that they can both inhibit and promote knowledge flow. Overall the research indicates that effective knowledge transfer depends on a mixture of strong ties/cohesion and range (Burt, 2005; Tortoriello et al., 2012). In this paper, we investigate knowledge transfer processes taking into account the embeddedness of individuals in higher-level structures such as teams, departments and units. According to previous research such higher level structures tend to form thought worlds (Dougherty, 1992) or foci (Feld, 1981) where individuals develop common frames of thought, which may inhibit the extent of knowledge transfer between them. In addition, the way in which these higher-level foci are related with each other potentially influences the learning at the individual level as well. For these purposes, we carry out an agent based simulation and distinguish between two levels: individual advice networks and the higher-level foci they belong to. In the agent-based simulation model, agents are members of different foci, and they learn from their alters through advice networks. We take into account different parameter spaces, according to (1) the structure of the advice network and (2) the similarity (relatedness) of foci. As for (1), the structure of the advice network is taken as a parameter in the simulations, which ranges between a perfectly closed (cohesive) structure where agents are connected only to others in the same focus, and a complete random structure where any agent is equally likely to be connected to any other. As for (2), we model the similarity between foci by drawing upon the knowledge-based theories of organisations. Specifically, we use a learning function in which, neither too much similarity nor too much dissimilarity is good for effective knowledge transfer, and that there exists an optimal cognitive distance (Nooteboom et al, 2007) for learning. In a parameter space defined by (1) and (2) above, we examine how the aggregate and foci knowledge evolves. In addition, we seek to understand how the cohesiveness (and randomness) of advice networks influence learning under different foci network structures as well as in more complex structures where agents can be members of various foci at the same time. Some preliminary results reveal that, the effect of advice networks on learning depend on the similarity between foci. When similarity between foci is low, a cohesive structure benefits overall learning. When similarity between foci is high, a random network structure benefits overall learning.