Early childhood interventions and their impact on poor children and their mothers: Evidence from Ecuador
This paper evaluates two early childhood interventions in Ecuador, a home-visit program and a child care center program, funded through the same proposal-scoring mechanism. For each program we estimate effects on poor children and their mothers in a regression discontinuity design around the funding cutoff. The forcing variable is a proposal-level score that takes only 34 distinct values, so we assess every estimate with inference methods designed for discrete running variables and few clusters. Our most robust finding is that home visits raise children's overall development: the bias-aware honest confidence interval excludes zero, also when we double the smoothness bound, apply the fuzzy correction, or restrict the sample to an age range that is balanced across the cutoff. The point estimates further suggest that home visits reduce mothers' labor supply and stress, and that child care centers raise maternal working hours and income with mixed effects on children, but these estimates do not survive the most demanding inference procedures. The pattern of point estimates is consistent with a trade-off between child development and maternal labor supply; we present the cross-program comparison as descriptive because the two programs serve different populations.