Childhood health shocks and parental labor supply

Poor health in childhood has been identified to be detrimental for schooling outcomes and labor market outcomes in adulthood. However, it is less well understood if shocks to children’s health have negative spillovers to parents as well. Using Danish registry data on health care utilization and socio-economic outcomes for the entire population, we leverage childhood onset diabetes (Type 1 diabetes) as an unforeseeable, exogenous shock to childhood health. We show that having a child diagnosed with diabetes reduces maternal labor income by 5%, and that this drop in income is persistent at least 5 years after the year of diagnosis. The effect is mainly driven by moving from full time to part time work. We do not find similar reductions in paternal labor income, nor does the diagnoses impact the probability of family dissolution.

The fundamental problem in estimating the causal impact of poor health on outcomes is that poor health is not randomly assigned. Those who are in poor health are likely to suffer from other disadvantages which would cause their outcomes to be worse for reasons other than health. By utilizing very detailed registry data, we demonstrate that children (and their parents) who get diabetes are similar to their healthy peers along an array of observable characteristics

Type 1 diabetes has – from an identification viewpoint – a nice set of features: Heritability is low; the vast majority of individuals affected do not have any immediate relatives with the condition. The condition cannot go undiagnosed nor can it be prevented. Symptoms develop within a very short timeframe of weeks to a couple of months, and treatment with insulin injections is fundamental for survival.

Type 1 diabetes is a complex disease to manage because it is cognitively demanding to constantly monitor glucose and adjust insulin levels. Thus, parents will likely need to shift some of their resources to helping manage the disease. Applying detailed information on disease management from blood samples drawn from affected individuals, we find the level of disease management varies by the education of the mother (academic degree versus less educated). At the time of diagnosis, children have the same high level of glucose concentration as measured by the HbA1c. One year after diagnosis and beyond we see gaps in disease management (HbA1c levels) by maternal education. This has important implications for clinicians who are trying to help patients better manage the disease.