We take advantage of repeated cross-sectional household surveys and a sharp discontinuity created by the introduction of an unconditional cash transfer to elders in Bolivia, to evaluate its impact on educational expenditures on children within a household. We find positive and significant impacts of the program at the aggregate level. We also find that the program has stronger effects on indigenous populations as well as on female and rural populations. Our results are robust to a series of falsification tests, survey structure, model specification, and estimation methods.