Essays on the economics of education

This thesis consists of three standalone papers which explore the extent to which education policies can change children’s outcomes, whether they tend to do this for better or for worse, and what factors get in the way. Chapter 1 explores changes in the attainment gaps between pupils from different ethnic groups when grades are assigned by teacher predictions rather than through blindly marked examinations. When grades are assigned by teachers, ethnic minority pupils in England receive higher grades in maths and lower grades in English relative to White British pupils and compared to when grades are assigned through exams. Using Gelbach decompositions, we show that observed characteristics partly explain the maths gap changes but amplify those in English. We conclude that group-specific stereotyping is a convincing explanation of the results. Chapter 2 evaluates the medium-term effects of an extended play-based learning policy for early childhood in Wales – the Foundation Phase – on a range of school-related outcomes. I use a staggered difference-in-differences research design to compare pupils who received the Foundation Phase to pupils who instead received formal education between ages five and seven. I find no evidence of effects at ages 11 or 16. Finally, Chapter 3 studies sibling spillover effects in temporary exclusions and unauthorised absences in England. We estimate sibling spillover effects when siblings are in the same grade and use a novel instrumental variable strategy exploiting variation in the behaviour of older siblings' peer groups within and across school cohorts. We find evidence of modest spillover effects from the older to the younger sibling in both exclusions and absences. We also find that sibling pairs are more likely to be excluded for the same stated reason, suggesting that role modelling is a plausible mechanism for our results.