The aim of this paper is to provide empirical analysis of the geographical mobility of PhDs during the beginning of their career, by using long term data (1970-2000) about France. First of all, we highlight the low mobility of PhDs: more than 60% of them supervise their first PhD in the region where they defended their own one. Then, thanks to a second database, we show that this mobility level is not specific to the Higher Education and Research system. We test the impact of structural determinants with a gravity model (sample selection model) and we observe in particular the influence of the scientific size of regions and spatial distance on the PhDs' mobility.
