Vejleder: Anders Holm 

Abstract:

Exploiting an institutional feature of the Danish school system we investigate class-size effects in an optional 10th grade of lower secondary school in Denmark. Given the setting we argue that pupils have low chances of predicting class size prior to enrolment, and that behavioural responses to realized class size are limited. This implies that the traditional endogeneity problem is limited in our case. Estimating value-added models using registry data from 2002-2009 we find small but significant negative effects of class size on achievement.