The Intergenerational Effects of Health and Health Policy

Project Outcomes Statement

This project sought to estimate how Medicaid, which began providing public health insurance to low-income families in 1966, affected the infant health of babies born to mothers who were themselves children when Medicaid was introduced. A large body of evidence ranging from studies of natural disasters in human populations to lab experiments in animal populations, supports the idea that improving the health of a child can also improve the health of her children. Recent evidence even shows that changes to Medicaid in the 1980s triggered this kind of effect. A key challenge to understanding intergenerational policy effects is that they only unfold after a substantial amount of time. Older policy changes thus offer the important advantage of being able to observe longer-run outcomes across generations.

This project planned to use differences in the share of children eligible for Medicaid at its inception, which came from a statutory requirement that the program cover all children receiving cash welfare, to measure which cohorts were “more” or “less” exposed to public health insurance in childhood. Early research has shown that this measure is correlated with immediate and sharp reductions in infant and child mortality as well as longer-run improvements in health and labor market outcomes. But at the outset of the project there were basic methodological questions about how to use a continuous variable to create valid “treatment/control” type comparisons in a setting like this. Therefore, much of the project’s effort was devoted to developing statistical techniques appropriate to estimate causal effects in a research design with a continuous treatment variable. This resulted in a working paper (Callaway, Goodman-Bacon, and Sant’Anna 2024) whose methods will be applied to the original question.