Contrary to standard microeconomic principles, it is by now well understood that income is not fungible. For example, the label of a government transfer can induce individuals to make expenditure decisions that are skewed towards the label. In this paper, we show that child benefits are disproportionately used for savings assignable to children. We exploit a policy reform in a difference-in-differences approach to estimatethe effect of child benefits on savings while holding total household income constant.Our results suggest a significant positive labeling effect on long-term savings, but noeffect on assignable consumption. We conclude that labeling effects should be consideredcarefully by policy makers, if not for nudging individuals, then to avoidaffecting decisions unintentionally.