On the Other Side of the Creek: Historical segregation, neighborhood social capital, and human capital accumulation

Abstract

This paper examines how long-run neighborhood segregation shapes the formation of localized social capital and, in turn, affects human capital accumulation. I study Guadalajara, Mexico, where a colonial-era administrative decision in the 16th century divided the city into two settlements separated by a small creek. Spanish settlers and their allies established themselves on the west bank, while Indigenous communities were resettled in the east. Despite convergence in infrastructure and public goods (school supply and quality) over time, residents on the eastern side of this historical boundary continue to show significantly lower educational attainment. Using modern census microdata and a spatial regression discontinuity design, I show that schooling levels decline sharply when crossing into the historically Indigenous east side, even within a few meters. I trace this persistent gap to differences in neighborhood social capital. Residents in historically marginalized areas report significantly higher trust in neighbors but much lower confidence in public institutions and broader civic networks. A mediation analysis suggests that this inward-facing, bonding form of trust explains approximately 20 percent of the educational gap between sides. I develop a theoretical model in which localized social capital competes with investment in portable human capital, generating multiple self-reinforcing equilibria. Cross-national evidence from the World Values Survey reveals similar patterns. Individuals who express high trust in neighbors relative to other groups are more likely to drop out of upper secondary school. These findings suggest that spatially concentrated trust, when rooted in historical segregation, can constrain long-run human capital formation.