Agenda > Tous les événements

Séminaire d’économie de Bordeaux

Laurent Bergé 
GREThA

Nicolas Jonard (Université du Luxembourg) et Ruth Samson (Université du Luxembourg)

“Should I stay or should I go? Unravelling the link between job mobility and productivity in the context of innovation: A case study of Nortel”

 Abstract : This paper investigates the link between job mobility and innovation. The central question is whether job mobility is a driver of productivity, in particular in the context of innovation in which the recombination of knowledge plays a critical role. The key problem in deriving causal estimates for job mobility is that most mobility events come from employees deciding to move, therefore confounding any effect of the mobility with the selection of the moving employees. We overcome this problem by focusing on a unique historical event: the fall of Nortel Networks. Nortel was the largest Canadian IT company at the turn of the century and unexpectedly collapsed following unforeseen strategic missteps combined to an accounting fraud. We use the sudden dismissal of thousands of R&D employees from Nortel to infer the consequences of mobility on innovation. Using a novel data set of USPTO patents matched to online CVs, we are able to track the full career histories of thousands of inventors across several decades. In a difference-in-difference setup, we compare employees who moved for exogenous reasons (the dismissed) to (i) employees who moved voluntarily and (ii) employees who did not move (non-movers). These comparative statics allows us to infer causal evidence on job mobility, net of the selection effects linked to voluntary mobility. Using various patent-related measures to track innovative productivity, the preliminary results show that dismissed employees are not less productive than the non-movers in their new companies (A), despite the negative event they had to endure. Their performance is however significantly lower than the performance of voluntary movers (B), although this effect does not show up for quality-weighted measures. Overall these first results suggest that the level of job mobility may be lower than what would be economically optimal (A), but not too much so (B).