Description:
We explore whether teachers have different predictions for the examination performance of ethnic minority students relative to White British students. We exploit an exogenous change in assessment methods to compare grades based on teacher predictions to grades received through actual blindly marked examinations. Relative to White British students, teachers appear to have higher predictions for ethnic minority students' examination performance in maths and lower predictions for ethnic minority students' examination performance in English. These effects do not disappear when observable differences between groups and cohorts are accounted for, with differential teacher predictions of examination performance across ethnic groups remaining a convincing explanation of the results.