www.sciedu.ca/ijba International Journal of Business Administration Vol. 3, No. 6; 2012 Published by Sciedu Press 82 ISSN 1923-4007 E-ISSN 1923-4015 Business Network Simulation: Combining Research Cases and Agent-Based Models in a Robust Methodology Frans Prenkert BI Norwegian Business School, Center for Cooperative Studies Department of Innovation and Economic Organisation, NO-0442 Oslo, Norway Tel: 47-46-410-559 E-mail: frans.prenkert@bi.no Received: October 5, 2012 Accepted: November 1, 2012 Online Published: November 9, 2012 doi:10.5430/ijba.v3n6p82 URL: http://dx.doi.org/10.5430/ijba.v3n6p82 Abstract This paper describes the use of an agent-based model (ABM) with the purpose of simulating interaction processes in business networks. Such an ABM must be able to capture time and process in networks while retaining a level of comprehension and overview. The ABM rests on the assumption that business networks can be viewed as complex adaptive systems (CAS) and draws on the opportunities to build and use ABMs to simulate processes in such systems. This paper outlines the conceptual foundations and the methodological challenges and opportunities associated with such an endeavor with special reference to issues concerning the modeling of the complexities of business networks. Specifically it discusses the use of ABM and research case in combination in an effort to produce a robust methodology. It contributes with suggestions on how one concretely can handle these issues and challenges when using ABMs to model non-linear dynamic interaction processes in networks. Keywords: Social simulation, Agent-based models, Multi-agent simulation, Business networks, Interaction, Complexity, Process 1. Introduction How entities such as business relationships and networks develop over time is an area that is generally under-researched. Despite being a recurring theme in the field of industrial marketing (Holmen, Pedersen, & Torvatn, 1999; Håkansson & Snehota, 1995; Medlin, 2004; Aastrup, 2000), there is still a gap in our understanding of how business networks develop and evolve over time and the roles managers and governments and other actors play in these processes (Wilkinson, 2008). One reason for the relatively little research on time and process in business networks can be due to the inherent challenges associated with both the associated conceptualizations and methodologies (Halinen, Medlin & Törnroos , 2012; Halinen & Törnroos, 2005; Medlin, 2004). In this circumstance, the argument in this paper is that there may be a remedy to this methodological challenge. I argue that we can model time and process in business networks in a fruitful way by considering these networks to be complex adaptive systems (CASs) and that we can investigate them by the use of agent-based simulation models (ABMs) run on ordinary desktop computers. This gives us the opportunity to view time and process through the lens of complexity science–a view that I argue to be very helpful when it comes to dealing with both the conceptualization problem and the methodological challenges associated with time and process in business networks, indeed with the complexities of management and organizational research in general as suggested by Harrison, Lin, Carroll, & Carley (2007). In relation to business networks, this is recognized by Ian Wilkinson: “A major gap in understanding business relations and networks is the way they develop and evolve over time [...] this gap exists because most research and theory to date is dominated by comparative-static, variance-based, survey-type approaches to describing and explaining relationship and network behavior and performance, which ignore temporal processes including development and evolution, interaction and order effects, and feedback effects. The overall aim must be, therefore, to build agent-based models (ABMs) of such business relationship and network development and evolution as complex adaptive systems.” (Wilkinson, 2008: 265) Wilkinson develops his line of argument as follows: “The traditional focus of research on static, variance-based models of business markets, tested often with one-shot cross-sectional surveys, has reached the limits of its usefulness. The models developed and tested typically assume