Peer Effects across Firms: Evidence from Security Analysts

We study peer effects among workers in the same occupation across firms in the context of security analysts (i.e., we study whether the productivity of a worker is a function of her peers in other firms). We use microdata concerning analysts as well as an identification strategy that employs one feature of social networks: the existence of partially overlapping peer group. The results show strong evidence of spillovers in terms of peers’ outcomes and characteristics. We find a significant positive causal effect of peers’ accuracy on an analysts’ accuracy. For example, a one standard deviation in peers’ earnings forecast accuracy increases an analyst’s accuracy by 25.7%. We also find that peers’ characteristics have a significant effect on earnings forecast accuracy, and the estimate of the social multiplier is around 1.46. ∗First Draft: August 2014. The University of Texas at Austin McCombs School of Business, 2110 Speedway, Stop B6600, Austin, TX 78712 ([email protected]). I would like to thank Clemens Sialm, Jonathan Cohn, Robert Parrino, Karsten Muller and seminar participants at the University of Texas at Austin, and 18th IZA Summer School in Labor Economics and LBS TransAtlantic Doctoral Conference.