Grades and exams are an important part of schooling systems that have marked effects on students. They ration scarce educational resources, provide a range of information to schools, and may promote learning. At the same time, they provide information to students about their academic progress and ability, and potentially, their suitability for future study. In practice, there is a large variety in how grades are assigned where a key variation is whether this is teacher assessment or exam-based. This paper uses a setting, Norway, where both are used, but exam grading is randomly assigned to students at the end of middle school. Using register data, we demonstrate systematic differences in grade assignment by teachers across different student groups. Most notably, teachers are in general generous relative to exams, and we document large systematic across school variation. In some schools, assignment to a math exam results in an average one grade reduction (i.e. B to C).
