Leniency in rating behavior compromises the interrater comparability that is necessary in hiring and promotion decisions. Because accountability may be one way to improve the information content of different raters’ ratings, in this paper, we exploit a natural experiment by comparing rating outcomes in the same institutions before and after an increase in accountability. More specifically, we use a large administrative dataset of 295,065 high school exit exams for 59,013 students in the German state of Hesse to study the effect of introducing centralized written exit exams on grammar versus comprehensive schools, which tend to have higher and lower quality peer groups, respectively. We find that the rating differences between these schools increases by about 50 percent when standards are centrally enforced. These results are consistent with the lenient behavior that occurs when raters subjectively adjust their rating standards to those of their subordinate team members in the absence of centrally enforced accountability standards.