Please use this identifier to cite or link to this item: https://hdl.handle.net/10419/234984 
Year of Publication: 
2021
Series/Report no.: 
ZEW Discussion Papers No. 21-049
Publisher: 
ZEW - Leibniz-Zentrum für Europäische Wirtschaftsforschung, Mannheim
Abstract: 
The proximity framework has attracted considerable attention in a scholarly discourse on the driving forces of knowledge exchange tie formation. It has been discussed that too much proximity is negatively associated with the effectiveness of a knowledge exchange relation. However, little is known about the key factors that trigger the formation of the boundaryspanning knowledge ties. Going beyond the "dyadic" perspective on proximity dimensions, this paper argues that the key factor in bridging distances may reside at the "triadic" level. We build on the notion of "the strength of weak ties" and its recent development by investigating the innovative performance and relations of more than 600,000 German firms. We explored and extracted information from the textual and relational content of firms' websites by using machine learning techniques and hyperlink analysis. We thereby proxied the innovative performance of firms using a deep learning text analysis approach and showed that the triadic property of bridging dyadic relations is a reliable predictor of firms' innovativeness. Relations embedded in cliques (i.e., strong ties) that connect cognitively distant firms are more strongly associated with firms' innovation, whereas inter-regional relations connecting different parts of a network (i.e., weak ties) are positively associated with firms' innovative performance. Also, the results suggest that a combination of strong inter-community and weak inter-regional relations are more positively related with firms' innovativeness compared to the combination of other relation types.
Subjects: 
weak and strong ties
proximity
knowledge exchange
innovation
web mining
natural language processing
JEL: 
C81
D83
L14
O31
Document Type: 
Working Paper

Files in This Item:
File
Size





Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.