Linking network centrality to performance revisit: understanding and dealing with simultaneity in the creation of soft innovation

Anotace: Empirical studies of social network analysis have provided sufficient evidence to support the general claim that high-centrality individuals in social networks outperform those with low centrality. We derive arguments from theories to question this claim and identify the needs for considering reversed causality when examining the centrality-performance relationship. It is puzzling that previous studies have assumed causality only in one direction from centrality to performance, but not the reverse. The possible origins of the endogeneity are discussed. Furthermore, the study re-examines the centrality-performance relationship in the context of movie directors’ networks and box-office performance. Our findings show that all measures of centrality significantly affect the outcome variable when the method of instrumental variables is applied to address the proposed reversed causality.