The Long-Run Effects of Disruptive Peers

Administrative elementary-school records linked to later test scores, college attendance and completion, and earnings at ages 24-28

US; elementary school cohorts followed into the labour market.

Extends the Carrell & Hoekstra design to adult outcomes, using population variation in the proportion of classmates from families linked to domestic violence. The outcome data are administrative earnings and enrolment records rather than surveys, and the identifying variation is the same plausibly-exogenous source used in the 2010 paper.

Exposure to one disruptive peer in a class of 25 during elementary school reduces earnings at ages 24-28 by 3 percent. Differential exposure to children linked to domestic violence explains 5 percent of the rich-poor earnings gap in the data. Each year of exposure to a disruptive peer reduces the present discounted value of classmates' future earnings by about USD 80,000.

Low. Peer composition variation is not chosen by the affected families.

Replicates and extends Carrell & Hoekstra 2010 on the same identification strategy with adult outcomes; converges with Figlio 2007, which uses an entirely different instrument.