Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
We study how performance metrics affect the allocation of talent by exploiting the introduction of the first citation database in science. For technical reasons, it only covered citations from certain journals and years, creating quasi-random variation: some citations became visible, while others remained invisible. We identify the effects of citation metrics by comparing the predictiveness of visible to invisible citations. Citation metrics increased assortative matching between scientists and departments by reducing information frictions over geographic and intellectual distance. Highly cited scientists from lower-ranked departments ("hidden stars") and from minorities benefited more. Citation metrics also affected promotions and NSF grants, suggesting Matthew effects.