People

After-school tutoring has risen globally despite limited evidence of effectiveness. We implement a randomized after-school tutoring program in rural China where many children are left-behind by parents in care of grandparents. Compared to tutees cared for by parents, those in care of grandparents reported much smaller home-tutoring reductions but larger test-score gains. We interpret our data analysis with a model with tutoring efficacy and substitution between private and public inputs both differing by family background: Increased public tutoring generates larger test-score gains for children who experience greater tutoring efficacy and lesser substitution with household inputs, consistent with our estimates.