Neighborhoods or Schools


Children growing up in high-poverty neighborhoods appear to be particularly disadvantaged when it comes to school performance, a fact which is easily observed in this graph showing the relationship between a neighborhood’s income levels and the local schools’ test scores.

The lack of progress in closing the racial achievement gap has fueled the concern that schools alone do not have the capacity to significantly increase achievement among the poor. Some have argued that the challenges impoverished children bring into the schools are simply too much for educators to overcome. (See, for example, Brooks-Gunn and Duncan, 1997 and Gonzalez et al, 1996.) In this view, real change will not occur unless significant social and economic reforms are put in place to help ailing neighborhoods. (See, for example, William Julius Wilson’s work on concentrated poverty and Richard Rothstein’s Class and Schools.)

The ideal way to understand whether schools alone can break the intergenerational cycle of poverty or whether community reform is necessary would be to design experiments that either alter school quality, neighborhood quality, or both simultaneously. However, because of the inherent complexity in researching a neighborhood’s effect, this approach is largely untenable.

As a result, no such experiments exist, and the social and political implications of such work are numerous and complex. Fortunately, there is a growing body of evidence using credible experimental and quasi-experimental sources of variation in existing neighborhoods and schools that allow researchers to draw conclusions about the impact of each on children’s academic and social outcomes. In this way, we can deepen our understanding of the reforms most likely to close the achievement gap.


The primary source of data on the impact of neighborhoods on life outcomes is theMoving to Opportunity experiment, which distributed housing vouchers to eligible families from Baltimore, Boston, Chicago, Los Angeles, and New York City. Ten to fifteen years of data were gathered on participating families. While MTO successfully placed families in higher quality neighborhoods, it had only a modest effect on the quality of schools children attended. The study showed that there was no impact on math and reading achievement, educational attainment, risky behaviors, or labor market outcomes for either male or female youth.


There are also many sources of variation within schools that allow for an analysis of the impact of school quality changes in the absence of any neighborhood interventions. Chetty et al (2011) examines class size reduction, using data fromProject STAR, and finds positive effects on both college attendance and long-term socioeconomic success. Chetty, Friedman, and Rockhoff (2011) quantifies the value of a highly effective teacher, showing that a student with even a single top teacher is more likely to attend college, earn a better living, and save for retirement. That same student is also less likely to become a teenage parent.


Evidence from the Harlem Children’s Zone, a unique anti-poverty program that covers a 97-block area in Harlem, allows an analysis of school changes coupled with neighborhood improvements. This environment offers the compelling opportunity to investigate the interaction of school and neighborhood quality on student experiences.

Comparing lottery winners outside the Zone to lottery losers outside the Zone reveals the impact of school quality alone, and shows positive effects from attending a high-quality school. However, comparing lottery winners within the Zone to lottery losers in the Zone shows that in only one out of thirteen possible outcomes is the effect of being in the Zone larger than for those out of the Zone. In other words, there is no important interaction of a neighborhood and school quality.

Based on this collection of evidence, it appears that strategic interventions towards improving school quality will have a greater impact on reducing persistent inequalities across a range of important life outcomes.