Neighborhood E¤ects on Performances at School¤

To repeat a grade in elementary school or junior-high school is the most direct indicator of early performances at school in France. At the end of junior high-school, about 40% of French adolescents have been held back a grade and our paper provides estimates of neighborhood e¤ects on this indicator of performances at school. We develop three di¤erent strategies in order to address the endogeneity issues. The ...rst approach uses the available information on the socioeconomic status of families who have just moved into a neighborhood. French families are bound to send their children to schools located in the administrative district where they live. Information on the families who move into a neighborhood are information on the quality of the neighborhood and may be used to control for the e¤ect of omitted neighborhood attributes. Our second strategy builds on the fact that in Francethe distribution of families across public-sector housing is under the control of administrative authorities and can be shown exogenous. The true neighborhood e¤ects can plausibly be evaluated by focusing on the sample of public-sector neighborhoods. Our last approach uses instrumental variables. In France, one of the determinant of having repeated a grade is the date-of-birth within the year. Children born at the beginning of the year are more mature and less likely to repeat a grade than children born near the end of the year. Within this context, the distribution of children’s date-of-birth in the neigborhood is a plausible instrument for identifying the e¤ect of the proportion of children who have repeated a grade in the neighborhood. Generally speaking, our three strategies provide consistent ...ndings. The estimated neighborhood e¤ects are signi...cant and almost as large as the direct e¤ect of parental education. ¤Previous versions of this paper were presented at the Conference in memory of Zvi Griliches (Paris, 2003) and at the CEPR Conference on Changing Condition in Education (Uppsala, 2003) and at the CREST (Paris, 2003). This reasearch has bene...ted from a grant of the French Commissariat General au Plan (n± 7-2002). We thank Per Johansson and Gerard Van den Berg for their comments on earlier versions of this paper.