Technological advancements now allow companies to report additional information to supervisors. In this dissertation, I examine how middle-level supervisors integrate additional information in their discretionary evaluation decisions, and how employees react to these evaluations. The findings presented in chapter 2 suggest that supervisors are more likely to reward unsuccessful exploration when they receive more frequent performance reports from their employees. In turn, employees do not appear to anticipate this and do not explore more when reporting frequency increases. The findings presented in chapter 3 suggest that supervisors with a wider span of control increase the rewards allocated to top performers and decrease the rewards allocated to the weakest performers. In turn, employees do not anticipate this and do not exert more effort when the span of control widens. The results of chapters 2 and 3 suggest that only changing how supervisors evaluate employees is not enough to change employee behavior because employees do not always anticipate how supervisors will evaluate them. The results of chapter 4 suggest that supervisors reward observable luck because they find it fair to do so. In turn, employees decrease their contribution when supervisors reward observable luck but only after employees learn how supervisors evaluate them through repeated interactions. These results suggest fairness concerns can diminish one of the intended benefits of allowing discretionary evaluations. Specifically, fairness concerns can prevent supervisors from using all available non-contractible information to decrease the weight of luck in employees' compensation.