Research Collaboration in Co-inventor Networks: Combining Closure, Bridging and Proximities

Abstract

Cassi L. and Plunket A. Research collaboration in co-inventor networks: combining closure, bridging and proximities, Regional Studies. This paper investigates the determinants of co-inventor tie formation using micro-data on genomic patents from 1990 to 2006 in France. In a single analysis, it considers the relational and proximity perspectives that are usually treated separately. In order to do so, it analyses various forms of proximity as alternative driving forces behind network ties that occur within existing components (that is, closure ties) as well as those between two distinct components (that is, bridging ties). Thus, the paper investigates not only the respective impacts of network and proximity determinants, but also how they overlap, interact and possibly act as substitutes or complements.

Cassi L. and Plunket A. 共同发明者网络中的研究合作：结合闭合、桥连带与邻近性，区域研究。本文运用法国自 1990 年至 2006 年间的染色体专利微资料，探讨形成共同发明者连带的决定因素。本研究在单一分析中，考量经常被分开处理的相对性与邻近性视角。为了进行上述研究，本文将分析各种形式的邻近性，做为存在于既存元素（意即闭合连带）之中、以及两个相异元素（意即桥连带）之间的网络连带背后的另类驱动力。因此，本文不仅探讨网络与邻近性决定因素的各别影响，亦探讨这些影响如何相互交叠、互动，并可能相互替代或补充。

Cassi L. et Plunket A. La coopération en matière de recherche dans les réseaux de coinventeurs: combiner la fermeture, le pontage et la proximité, Regional Studies. À partir des données microéconomiques sur le brevetage du génôme en France de 1990 à 2006, ce présent article cherche à examiner les déterminants de la création de liens entre les coinventeurs. En une seule analyse, on considère les points de vue relationnel et de proximité que l'on traite séparément d'habitude. Pour pouvoir le faire, on analyse les différentes formes de proximité comme forces motrices alternatives derrière les liens de réseau qui se produisent au sein des éléments en vigueur (à savoir, les liens de fermeture) ainsi que ceux qui relient deux éléments distincts (c'est-à-dire les liens de pontage). Donc, l'article examine non seulement les retombées relatives aux déterminants des réseaux et de la proximité, mais aussi comment ils se chevauchent, interagissent et servent éventuellement de substituts ou de compléments.

Cassi L. und Plunket A. Kooperative Forschung in Erfindernetzwerken: die Kombination von geschlossenen und überbrückenden Verbindungen mit Nähe, Regional Studies. In diesem Beitrag untersuchen wir die Determinanten der Bildung von Verknüpfungen zwischen gemeinsamen Erfindern mit Hilfe von Mikrodaten über genomische Patente im Zeitraum von 1990 bis 2006 in Frankreich. Hierbei betrachten wir die Perspektiven der Beziehungen und Nähe, die normalerweise getrennt behandelt werden, in einer gemeinsamen Analyse. Zu diesem Zweck analysieren wir verschiedene Formen der Nähe als alternative treibende Kräfte hinter Netzwerkverbindungen, die innerhalb von bestehenden Komponenten auftreten (d. h. geschlossene Verbindungen), und Netzwerkverbindungen, die zwischen zwei verschiedenen Komponenten auftreten (d. h. überbrückende Verbindungen). Auf diese Weise untersuchen wir nicht nur die jeweiligen Auswirkungen der Netzwerk- und Nähedeterminanten, sondern analysieren auch, wie diese überlappen, sich gegenseitig beeinflussen und einander eventuell ersetzen oder ergänzen.

Cassi L. y Plunket A. Investigación cooperativa en las redes de coinventores: la combinación de vínculos cerrados y vínculos de conexiones con la proximidad, Regional Studies. En este artículo investigamos los determinantes de la formación de vínculos entre coinventores mediante datos micro sobre las patentes genómicas de 1990 a 2006 en Francia. En un análisis único, consideramos las perspectivas de relación y proximidad que generalmente se tratan por separado. Para ello, analizamos varias formas de proximidad como factores alternativos que impulsan los vínculos de redes que ocurren entre componentes existentes (es decir, vínculos cerrados) así como aquellos entre dos componentes distintos (es decir, vínculos de conexiones). Por tanto, en este artículo analizamos no solamente los impactos respectivos de los determinantes de redes y proximidades, sino también cómo se solapan, interactúan y posiblemente actúan como sustitutos o complementos.

Keywords:

• Social networks
• Relational perspective
• Proximity
• Co-patenting
• Network formation

• 社会网络
• 相对性视角
• 邻近性
• 共同专利申请
• 网络形成

• Réseaux sociaux
• Point de vue relationnel
• Proximité
• Brevetage
• Création de réseaux

• Soziale Netzwerke
• Relationale Perspektive
• Nähe
• Gemeinsame Patente
• Netzwerkbildung

• Redes sociales
• Perspectiva relacional
• Proximidad
• Copatentes
• Formación de redes

JEL classifications:

• D85
• O31
• R12
• Z13

Acknowledgement

This research was funded by a grant from the Agence Nationale pour la Recherche (ANR) Corpus Genomique project. The authors are grateful to the seminar and conference participants of: Association de Science Régionale de Langue Française (ARSDLF) Clermont 2009; Colloque proximité, Poitier 2009; Analyse des Dynamiques industrielles et Sociales (ADIS) internal seminar, Fudan-Paris 1 joint seminar, June 2010; Toulouse Eurolio, 2010; Dime workshop Utrecht, September 2010; and especially to Koen Frenken for comments. The authors wish to thank Endri Nocka, Sounia Chanfi and especially Olivier Antelo for research assistance on this project.

Notes

1. Social network analysis computation was programmed by the authors themselves with SAS. The SPAM modules developed by James Moody (Moody, 2000) were extremely helpful.

2. Even for industry–university collaborations, usually there is only one affiliation for a given patent. For this reason, inventors of a given patent have the same affiliation even if the applicant designated in the patent does not employ them.

3. The logit and probit models were used, as is usual in this literature, which assumes that each tie formation is independent of the ties created at the same period. However, it is recognized that this is a strong assumption since the decision to form a link by an inventor is affected by the choice of other collaborations and the simultaneous behaviour of other individuals. The authors thank one of the referees who highlighted this point.

4. The Euclidean distance was also calculated and similar results were obtained.

5. The estimations are limited to bridging and closure ties since the authors cannot estimate geographical, or organizational and institutional distances for the pendant and new component ties, because these ties are formed by new inventors for which there is no information about their characteristics in t – 1.

6. The multinomial probit model is preferred since it enables the independence of irrelevant alternatives (IIA) assumption to be relaxed. The Hausman–McFadden and Small–Hsiao tests (available from the authors upon request) are not entirely conclusive and thus provide little guidance to the violation of the IIA assumption (Long and Freese, 2003). However, similar results are obtained as for the multinomial logit.

7. See note 4.

8. The latitude and longitude coordinates for the Earth's curvature were adjusted; thus, the distance (km) between two points A and B is computed as:

9. In order to ensure that dropping all non-European inventors does not affect the regressions, all models were estimated with a proxy of the geographical distance to non-Europeans by introducing a geographical distance of 6000 km for all North American inventors. Observations were substantially increased with 1,999 observations for closure ties and 2671 for bridging ties. The results remain similar overall for signs, magnitude and significance.

10. A logit model with cluster robust errors and year fixed effects was also estimated, and similar results to the conditional logit were obtained. These tables can be provided by the authors upon request.

11. If n is the number of active inventors, n*(n – 1) is the number of potential ties between these inventors. This number is estimated to be approximately 3000.