Optimal Grading

In the framework of static mechanism design games with non- pecuniary rewards, we solve for optimal student grading schemes and attempt to explain the observed mismatch between students? grades and their abilities. The model predicts that the more pes- simistic the teacher is about her students, the more generous she should be in grading them. Generally, the "no distortion at the top" property ceases to hold for optimal contracts with cost- less non-pecuniary rewards, and we argue that the compression of ratings as witnessed in job performance appraisals could be an equilibrium outcome. The presented theoretical ?ndings are strongly supported by empirical evidence from the related litera- ture in psychological and educational measurement.