In this paper we measure the effect of entering preelementary school at age 2 rather than 3 in France. Our identification strategy relies on ratios between the number of young children and the capacity of preelementary schools observed at the very local level. This information allows us to solve...