Network dynamics and field evolution: The growth of interorganizational collaboration in the life sciences
Citation: Walter W. Powell, Douglas R. White, Kenneth W. Koput, Jason Owen‐Smith (2005) Network dynamics and field evolution: The growth of interorganizational collaboration in the life sciences. American Journal of Sociology (RSS)
DOI (original publisher): 10.1086/421508
Semantic Scholar (metadata): 10.1086/421508
Sci-Hub (fulltext): 10.1086/421508
Internet Archive Scholar (search for fulltext): Network dynamics and field evolution: The growth of interorganizational collaboration in the life sciences
Tagged: Sociology (RSS) Complexity (RSS), Social Networks (RSS), SNA (RSS), Biotechnology (RSS), Entrepreneurship (RSS), Alliance (RSS), Strategy (RSS)
The goal of the paper is to use network-based methods to help account for and describe the development of the commercial field of biotechnology by showing how the formation, dissolution, and "rewiring" of alliances and network connections between 1988 and 1999 shaped the opportunity structure for biotechnology firms. In other words, the papers to treat networks as dynamic and to map the evolution of the entire network over time.
In particular, they aim to test between four hypotheses about what drives network connections over time:
- Accumulative advantage (a rich-get richer story)
- Similarity, (i.e., homophily-based mechanisms)
- A follow-the-trend story of firms emulating other firms and dominant trends.
- Multiconnectivity which is a new model suggested by the authors and which describes the linking of multiple partners both directly and through chains and a preference for diversity in ties.
Throughout the paper, the authors return to a dance-floor metaphor of choosing partners for a dance floor. Since their model is dynamic, they also can describe both the way in which an individual may desire different connections at different times and way in which dynamics can drive the evolution of the dance floor and the choices made by the individuals on it in general.
The authors assemble an incredible dataset on tie formation for the first 12 years of the biotechnology industry to test their results. In terms of their network measures, the authors use a broad definition of collaborative tie that describes any alliance or contractual relationship to exchange or pool resources between a dedicated biotech firm (DBF) and a partner organization. For ties that are not bewteen DBFs, the authors code the ties by the type of organizations (e.g., universities, pharmaceutical companies, VCs). Tie data is available for all DBFs and for links from DBFs to outside organizations, but not for links between the outside organizations.
The results show little support for accumulative advantage, mixed to bad results for homophily, weak support for follow-the-trend, and strong support for the authors' multiconnectivity hypothesis. Indeed, they show that as DBFs consider new partnerships, a patterns diversity of ties acts as a valuable marker of resources and information.
The results also show that diversity has become more important with time and that the industry has changed hugely. The industry was initially dominated by alliances to large pharmaceutical companies but that, over time, this has changed to R&D partnership and venture-financing and the large multinational companies being pushed toward the periphery of the network.
While the methods are incredible, there remain important questions about the generalizability of the results to fields other than biotechnology. One important takeaway is about the degree to which a preference for partners that are multiply or differently linked can drive the evolution of a network or field and how it drove the biotechnology in particular.
Theoretical and Practical Relevance
Powell and White did this work as part of the Santa Fe Insitute and the fingerprints of that organization are visible all of this piece. The methodological innovation and the complexity of the analysis are made clear in the "hairball" pictures sprinkled throughout the article give a good view of the complexity that the authors are struggling with in their analysis. It also puts important limitations on the generalizability and, indeed, the basic comprehensibility, of the papers results.
The paper has been cited more than 500 times its publication 5 years ago. It is an important paper in the study of biotechnology in particular but has important applications for entrepreneurship, innovation, networks, and geographical localization.