We use a rich longitudinal dataset on department-level productivity in a contemporary field of science to identify and decompose the causal impact of hiring a star on local knowledge production. Specifically, we estimate the relative roles of knowledge spillovers versus recruiting externalities as they affect co-located researchers who are related or unrelated to the star in idea space. Hiring a star does not increase overall incumbent productivity, but this aggregate effect hides off setting effects on colleagues who are related (positive) versus unrelated (negative). Star hires improve subsequent joiner quality for both related and unrelated scientists, although the effect is significantly larger for related scientists. The overall positive impact of the star on department-level productivity is mainly due to joiner-quality effects. Furthermore, the productivity impact is more pronounced at mid- and lower-ranked institutions, suggesting implications for the optimal spatial organization of science and university strategies aimed at ascending departmental rankings.