Panel Paper: On Peer Effects and Pollution: Does Exposure to Lead Affect Everyone in the Classroom?

Saturday, November 9, 2019
Plaza Building: Lobby Level, Director's Row I (Sheraton Denver Downtown)

*Names in bold indicate Presenter

Ludovica Gazze, University of Chicago, Claudia Persico, American University and Sandra Spirovska, University of Wisconsin, Madison


A growing body of evidence indicates that lead exposure has deleterious effects on children’s educational and behavioral outcomes (Amato et al., 2013; Aizer et al., 2018; Aizer & Currie, 2017; Feigenbaum & Muller, 2016; Ferrie, Rolf, & Troesken, 2015; Grönqvist, Nilsson, & Robling, 2017; Reyes, 2007, 2015a, 2015b). However, there also might be potential spillover effects of lead exposure on children who, despite not being directly poisoned by lead, attend school with lead-exposed peers. Previous research suggests that children who are disruptive can negatively affect the performance of other children in the same class (Carrell & Hoekstra, 2010; Figlio 2007). Because children exposed to lead are more disruptive and have lower achievement, we show that the effects of lead exposure spill over to affect everyone in the classroom. If lead has spillover effects, quantifying these externalities would change the estimates of the social cost of lead exposure to be much larger than previously supposed. The Department of Housing and Urban Development estimates that nationwide, 5.5 million houses inhabited by small children have lead paint (HUD, 2011). In addition, the American Water Works Association estimates that in the US, there are at least 6.1 million lead service lines—the pipes that connect each home to the water main—and replacing all of these service lines could cost up to $30 billion. Moreover, because lead hazards are more concentrated in disadvantaged communities, educational spillovers from lead exposure might have important distributional consequences.

This project investigates spillovers in lead exposure using unique student-level education data linked to student-level blood lead tests by age six. Our education data cover every child born between 1990 and 2008 and attending public schools in North Carolina. First, we show that the negative effects of lead exposure on educational and behavioral outcomes are robust to controlling for siblings fixed effects, thus assuaging concerns of omitted variable bias. We also investigate the persistence of these effects across grades, and their heterogeneity across socioeconomic statuses. We find that lead-exposed peers have worse outcomes in the classroom.

Second, we estimate the effect of having more lead-exposed peers on educational achievement, suspension from school and dropping out of school in different grades and school environments controlling for sibling, school, and grade fixed effects. We exploit variation in the proportion of lead-exposed children across years in the cohort composition at a given school and compare siblings who attend the same school a few years apart. Our preliminary estimates suggest that having more lead-poisoned peers decreases test scores and increases the likelihood of being suspended from school after controlling for one’s own lead exposure.