1 Does Geographical Proximity Still Matter? An Empirical Analysis of Two Long- Term Relational Databases

Abstract: The purpose of this article is to provide possible answers, at an empirical level, to the question “Does geographical proximity still matter in collaborations for innovation?", since the lack of available data has made it impossible to provide real answers up to now. Relying on two real long-term relational databases relating to science-industry collaborations in France, and on two complementary indicators of geographical proximity, we will show that proximity continues to count (number of significant intradepartmental collaborations and low average distance between the partners). However, this dynamic analysis does allow us to state that the closest partnerships and the most distant ones increase the most over the study period. We shall also show that the role of geographical proximity and its evolution over the course of time differ according to the type of science-industry contract and the sectoral specialisation of the partners