Worldwide children’s access to after-school learning activities is highly dependent on family backgrounds. Concern over the implications of such activities for child development and educational inequality has led to a global rise of public provision of after-school learning support. However little is known about inter-actions of public after-school activities and household investments in children’s learning. This paper contributes to the literature on the effects of public inputs on household inputs and student achievement in after-school settings. We build a model that integrates public and private inputs to produce student achievement through two competing mechanisms – diminishing returns to total inputs and complementarity between public and private inputs. When diminishing returns dominate complementarity, the model predicts the substitution away of private inputs due to increases in public inputs for all households, although the extent of crowding-out is smaller and therefore the test score gains are larger for children from disadvantaged family backgrounds facing higher costs of private inputs. We implement a randomized controlled after-school tutoring experiment in rural China where many children are left-behind by both parents and cared for by grandparents. During the program, tutees living with parents reported large and significant reductions in the amount of tutoring received at home, whereas tutees living apart from both parents reported much smaller, and often insignificant, reductions. We find that tutees’ math scores improved significantly, and more for children living without parents, although there is no evidence for improvement in tutees’ endline reading scores.