Daycare Matching with Siblings: Social Implementation and Welfare Evaluation

In centralized assignment problems, agents may have preferences over joint rather than individual assignments, such as couples in residency matching or siblings in school choice and daycare. Standard preference estimation methods typically ignore such complementarities. This paper develops an empirical framework that explicitly incorporates them. Using data from daycare assignment in a municipality in Japan, we estimate a model in which families incur both additional commuting distance and a fixed non-distance disutility when siblings are assigned to different facilities. We find that split assignment generates a large disutility, equivalent to more than twice the average commuting distance. We then simulate counterfactual assignment policies that vary the strength of sibling priority and evaluate welfare. The sibling priority reform that we designed and that was implemented in 2024 increases welfare by 6.4% while reducing inequality in assignment rates across sibling groups; models that ignore sibling complementarities substantially understate these gains. At the same time, we uncover a clear efficiency-equity tradeoff: along the frontier, increasing mean welfare by 100 meters is associated with an increase in inequality of about 1.7 percentage points, and the welfare-maximizing policy reverses much of the reform's reduction in inequality, largely through the displacement of households without siblings.