For Whom Neighborhoods Matter: Heterogeneous Effects of Socioeconomic Neighborhood Composition on Child Achievement

• In Event: Heterogeneity in Neighborhood Effects on Child Development: An International Perspective

Abstract

Background
There is convincing evidence that the neighborhood children grow up in has a causal effect on school achievement and other life chances. However, much less is known about heterogeneity in neighborhood effects. This paper takes a novel approach to studying heterogeneity in neighborhood effects by integrating population-wide data from national registries with genetic data from a large health study.

Data
We use Norwegian register data covering individuals born in Norway between 1997 and 2008 (approx. 600,000 children), which are matched with genetic data on parents and children from the Norwegian Mother, Father, and Child Cohort Study (MoBa) (approx. 30,000 children). The treatment is average neighborhood socioeconomic composition (e.g., education, earnings, welfare) during the first ten years of the child’s life (mean=0, sd=1), where higher values mean a more disadvantaged area. The outcome is the average of standardized national 5th-grade tests in reading (Norwegian) and mathematics. Children’s genetic propensity towards education is measured using the polygenic scores for educational attainment (EA3 PGS).

Methods
Unobserved confounding has been a concern in the neighborhood literature for decades. We use the recently developed two-step Neighborhood Choice Model to identify the effects of neighborhood characteristics, as well as conditioning children’s genetic effects on parents’ genetic propensity. This modeling approach first creates correction terms that measure families’ propensity to sort into particular neighborhoods and then controls for these correction terms in a linear regression model.

Results
We find that neighborhood disadvantage during a child’s ten first years impacts test scores (column 1 in Table 1): compared to growing up in the most affluent areas, the children in the most disadvantaged neighborhoods get about 23% lower test scores in fifth grade (i.e., comparing -2SD with +2SD). Furthermore, our results show that neighborhood effects differ by individual characteristics and family background (columns 2-5). First, boys are considerably more affected by neighborhood exposure than girls; neighborhood SES affects boys’ test scores by 7% of a standard deviation, compared to 4.3% among girls. Second, children of socioeconomically disadvantaged parents are more affected by neighborhood exposure than those of advantaged parents. For example, among children of parents who have completed compulsory school, a one standard deviation increase in neighborhood disadvantage reduces test scores by 6.3%, compared to 2.8% among parents with a master’s degree. Unlike socioeconomic backgrounds and gender, children’s genetic propensity to take long education does not seem to moderate the neighborhood effect. Although the coefficient indicates that children with a high genetic propensity are less affected by the neighborhood SES, it is not statistically significant at conventional levels.

Conclusions
There is limited research on neighborhood effects on early school outcomes and even less research on for whom neighborhoods matter. Adapting the Neighborhood Choice Model to the entire country of Norway, we find that growing up in a disadvantaged neighborhood harms children’s test scores, especially for boys and children of parents with low socioeconomic backgrounds. However, children with a genetic propensity to do well in school are seemingly affected by their neighborhood surroundings as other children.

Authors

• Nicolai Topstad Borgen, Universitetet i Oslo
  Presenting Author

• Henrik D. Zachrisson, University of Oslo

• Torkild Hovde Lyngstad, University of Oslo

• Rosa Cheesman, King's College London

• Eivind Ystrøm, University of Oslo