brown bag Travail et Économie Publique

Du 19/09/2024 de 12:30 à 13:30

PSE- 48 boulevard Jourdan, 74014 Paris, salle R1-09

HEIM Arthur (PSE)

Rage against the matching: fairness and inequalities in a market design experiment of daycare assignments in France

The choice of childcare alternatives is a central decision which affects several key societal dimensions such as child development, mothers' labour supply, and economic and gender inequalities. While families from lower socioeconomic backgrounds tend to gain the most from formal childcare, these services are overwhelmingly used by more affluent families - a phenomenon often called the Matthew effect. In this paper, we consider access to daycare as a matching problem controlled by local authorities. Based on policymakers' definitions of the procedure, we use market design to define assignment mechanisms and we analyse the consequences of important design choices in a field experiment. The daycare assignment problem is similar to school choice, but includes two additional features: multidimensional constraints that cover weekdays and diversity constraints, typically age groups. Policymakers' design choices affect the definition and range of stable matchings. Our algorithms deliver student optimal fair assignments (SOFA) in the different versions of the problem. From 2020 to 2023, we assigned daycare slots to families in nine urban districts in France. Our objectives were twofold: i) to provide automated assignment mechanisms with desirable properties and ii) to introduce random variation in assignment probabilities for future work on causal effects of accessing daycare. We use two case studies to demonstrate that our assignments meet their intended objectives and compare different assignments. Using a change in priorities in one case study and counterfactual simulations of alternative priority scores, we show that i) giving larger weights to some group (e.g. dual earners) increases their assignment probabilities and share in daycares, ii) increasing priorities with time since registration strongly penalises single parents, in part because iii) dual-earner couples compete for early entry and strategically register as soon as possible, as incentives. iv) Being the largest group and also receiving high social weights, their strategies crowd-out parents who cannot register that early and create inequalities of opportunities even among strategic parents, correlating assignment probabilities with birth month. Other analyses show that diversity constraints may create sharp discontinuities in assignment probabilities that are unrelated to priority scores. Our results provide clear evidence of the mechanisms that contribute to the Matthew effects in childcare, and they are mostly political choices. However, our tools can be used to satisfy other distributional objectives and achieve higher transparency in the assignment processes.