| Summary: | Socioeconomic segregation is often decried for denying poorer children the benefits of positive 'peer effects'. Yet standard, linear-in-means models of peer effects (a) implicitly assume that segregation is zero sum, with gains and losses to rich and poor perfectly offsetting, and (b) rule out theories of 'social distance' whereby peer effects are strongest among similar pairings. The paper exploits the random assignment of pupils between classes to identify more general peer effects in Argentine test-score data. Estimates violate both assumptions (a) and (b), and provide micro foundations for the correlations between school segregation, average test-scores, and test-score inequality in municipality-level data
|
|---|