When and Why Do Supervisors’ Evaluations Overweight Subordinates’ Performance Outcomes? Evidence from a Team Setting in the Field

ABSTRACT To better understand supervisors’ outcome bias, I use a regression discontinuity design that compares coaches’ performance assessments of professional football players involved in narrow wins and losses. I document that supervisors over-react to negative outcomes, sharply lowering performance ratings and tripling subordinate turnover. I find that supervisors’ evaluations that subjectively incorporate information from more incomplete objective performance measures are more prone to outcome bias than those that draw on less incomplete objective measures. I also document that supervisors’ evaluations of high-performing team members are more prone to outcome bias than supervisors’ evaluations of low performers. Finally, I find that outcomes affect supervisors’ ex post information collection. My findings, consistent with predictions that I derive from the theory of cognitive reconstruction, shed light on supervisors’ outcome bias in team settings and how effectively firms’ use of objective performance measures, direct monitoring, and information gathering can mitigate this bias. JEL Classifications: M50; D91; Z20.