This paper establishes a new fact about educational production: ordinal academic rank during primary school has long-run impacts on later achievement that are independent from underlying ability. Using data on the universe of English school students, we examine a setting in which the same baseline score on a national tandardized test can correspond to different ranks among students situated in different primary school classes, where we calculate ranks using this baseline score. Institutional factors cause students to be re-assigned to a new set of secondary school peers and teachers that are unaware of the student's prior ranking. We find large and significant positive effects on test scores and subject choices during secondary school from experiencing a high primary school rank, conditional on the underlying primary baseline score. The effects are especially large for boys, contributing to an observed gender gap in end-of-high school STEM subject choices. Merged survey data suggest that the development of confidence is a likely mechanism.
