2004

Abstract
This study exploits two features of the Dutch schooling system to estimate the effect of expanding enrollment possibilities on test scores of young children. One feature is that children are allowed to enroll in school on their fourth birthday. The other that children having their birthday before, during and after the summer holiday are placed in the same class. Together these features generate sufficient exogenous variation in children’s potential time in school to identify its effects on test scores. Allowing disadvantaged pupils to start school one month earlier increases their test scores on average by 0.06 of a standard deviation. This effect is of the same magnitude for pupils with lower educated parents and for minority pupils. For nondisadvantaged pupils we find no effect. Results are similar for language and math scores.

Key words: Early childhood intervention, early test scores
JEL Codes: I21, I28, J24