Critical links in knowledge networks – What about proximities and gatekeeper organisations?

Abstract

The paper analyses links in knowledge networks that are essential for the integration and knowledge diffusion properties of the entire network. By focusing on critical links, as defined in network science, we evaluate these links’ properties from the perspective of the proximity and regional gatekeeper literature. We thereby gain insights into likely conditions of their emergence and functions. Moreover, we extend the dyadic perspective on regional gatekeeper organisations and link it more strongly to the network science and proximity framework literature. An empirical study applies these arguments and investigates the proximity characteristics of critical links in 132 technology-specific subsidised knowledge networks in Germany. The results show that critical links tend be formed between regional gatekeepers that offer related knowledge resources. The links bridge institutional distances by utilising the benefits of geographic and social proximity.

Keywords:

• Proximities
• knowledge networks
• gatekeeper
• R&D subsidies
• critical links

JEL Classifications:

• D85
• L14
• O33
• R10

Notes

¹ They are also sometimes referred to as ‘bridging’ or ‘gatekeeper’ links.

² Cassi and Plunket (2015) also argue for critical links being characterized by large social distances. However, in the context of inter-organizational knowledge networks, we think that the relationship between social proximity and critical links is much less clear.

³ Newman (2001) suggests to account for the number of nodes related to an event when projecting from two-mode to one-mode networks with the following weighting: with N _a being the number N _a − 1 of nodes related to an event in the two-mode network.

⁴ The geodesic distances in the weighted network are estimated according to the method proposed by Dijkstra (1959).

⁵ The measure is moreover significantly different from link betweenness. In the later empirical analyses, the correlation between bridging and link betweenness turns out to be just 0.281***.

⁶ Note that it is our aim to identify general characteristics of critical links that apply on average in most technology-specific networks.

⁷ E.g. associations, ministries, and societies.

⁸ As cooperative projects might be formed on the basis of resource complementarity and not similarity, these are excluded.

⁹ Note that this approach is prone to identify large organizations as regional gatekeepers. In the paper’s context, this has no further implications.

¹⁰ The variance inflation factor (VIF) test indicates the absence of multicollinearity in each model. Maximum VIF scores are depicted in Table 3. The use of robust standard errors also does not yield different results.