Résumé : Fertility has declined almost everywhere, yet the pattern of decline differs sharplyacross countries, education groups, and family forms. We argue that this heterogeneity is best understood through the value of marriage, that is, the surplus generated bypartnership relative to outside options. This surplus depends on preferences, the organization of work and childcare, bargaining positions, marriage market opportunities,social norms, and public policy. Children therefore do not simply enter utility; theyreshape both the gains from partnership and the way these gains are shared over time.Collective models enriched with marriage markets and limited commitment providea coherent framework for analyzing these mechanisms jointly. This perspective showshow similar fertility outcomes can emerge from very different underlying forces, such ashigh maternal career costs in one context and fragile unions with limited commitmentin another. It also highlights why structural models are essential for counterfactualpolicy analysis: they help isolate whether observed fertility patterns reflect changes inchildcare burdens, bargaining positions, or marriage-market conditions, and how thesemechanisms interact.