Association for Public Policy Analysis & Management

Understanding the spillover effects of schooling from students to their family members is potentially important for educational policy, both in determining the optimal level of public investment in education and in creating the optimal mechanism for allocating students to schools. This paper focuses on the causal effects of older siblings attending a higher-quality school on the achievement and academic behaviors of younger siblings in Mexico City’s public high school system, contributing to a small but emerging literature on sibling spillovers in education. In particular, this paper is among the first to investigate sibling externalities from educational quality, and is unique in addressing such effects in a developing country.

Estimating sibling spillover effects from school quality is difficult because students are not randomly assigned to schools of differing quality levels. To overcome this issue, I use exogenous variation in older siblings' school assignments arising from Mexico City’s centralized allocation mechanism. Students rank several preferred schools and then take a standardized exam to determine assignment priority. Some students score barely high enough to attend a more-competitive school, while others score barely too low and are rejected, giving rise to a regression discontinuity design that compares outcomes for younger siblings whose older siblings were marginally admitted or rejected from a “better” school.

In addition to proxying for school quality by average peer ability, I estimate value-added measures in a model that uses data on students’ stated preferences to account for non-random selection into schools. These value-added measures are then incorporated into the regression discontinuity design, which is robust to misspecification in the value-added model. The data come from several complete cohorts of students participating in the entrance exam and allow for linking siblings within a family across cohorts.

Younger siblings of students attending schools with relatively high value-added in mathematics experience gains in their own math scores at the end of middle school. Spillovers in Spanish are weaker and there is no discernible effect on middle school grade point average. Using extensive survey data on self-reported behaviors and attitudes, I find limited effects on study habits, student self-image, or parental investment, suggesting that the academic gains are likely to stem from having a more-able sibling in the household to assist in learning.

In contrast, simply attending a school with higher-achieving peers has little average effect on the outcomes of students or their younger siblings, with the latter set of effects being bounded tightly around zero. The findings suggest that school quality, rather than simply peer composition, is the key driver of between-sibling spillovers, such that investments in provision of educational quality are likely to have externalities for other members of a student’s household.