There has been a significant amount of literature on the relationship between household income and children’s health, and in particular, whether this relationship is constant as a child ages. That is, are there breaks in the income-health profile according to the age of the child? In this paper we re-examine this relationship but hypothesize that the use of ‘standard’ age ranges (as in the existing literature) may mask ‘true’ changing gradients and breakpoints: there is no reason to suppose that the ‘standard’ age bands used to date reflect the actual ages at which breakpoints occur. Indeed, we allow the data to endogenously determine if, and where, breaks occur. We find positive gradients in both US and English data sets, and breakpoints which coincide with school changeover ages. Our results have significant policy implications in terms of allocation of resources.
This papersis written by Mark N. Harrisa, Bruce Hollingsworth, Brett Indera and Pushkar Maitrac.