The recent election has caused many to question whether significant changes are about to happen to Medicaid. Repeal of the Affordable Care Act would, of course, lead to the elimination of the Medicaid expansion, which could result in significant numbers of adults losing their coverage overnight. But even without repeal, many Republican replacement proposals also result in significant changes to Medicaid, whether it be through funding, eligibility, or benefits.

When discussing such changes, it's worth examining the evidence on the effect of Medicaid on society. With respect to children, a new NBER working paper does just that. Andrew Goodman-Bacon used the original implementation of Medicaid to conduct a controlled analysis of how health coverage for children through Medicaid affected health, employment, and income in adulthood. Specifically, he exploited the introduction of Medicaid between 1966 and 1970, when it mandated that all cash welfare recipients receive health insurance coverage through the program.

This meant that in areas of greater welfare eligibility, more people were immediately eligible for Medicaid. Further, those children who were born closer to the implementation date spent more time in their life on Medicaid. Cumulative childhood Medicaid eligibility therefore phased in across cohorts gradually, and more speedily in higher-welfare states. By looking at cohorts born in different years and in different states, it is possible to estimate how Medicaid affected outcomes through a difference-in difference model. Moreover, since there are so many years since the implementation of Medicaid, it was possible to look at long-term impacts of the program.

One of the main results was that those who were eligible for Medicaid at an earlier age had a lower mortality rate as adults. They also had lower disability rates. 

People who were on Medicaid earlier tended to earn more as adults, to the tune of about $1800 on average. They therefore received fewer benefits from the government (averaging about $600 less per year), which meant that their overall income didn't change significantly. Medicaid didn't seem to reduce poverty rates either, because it affected where people got their income from, not how much income they received. But for people who favor getting people to get more of their income from the private sector, and less from welfare, Medicaid seems to do a decent job. Government spending for these individuals, on the other hand, did reduce overall government spending. Goodman-Baker's analysis showed that they saved on benefit payments and took in increased revenue through income taxes, to the tune of more than $21 billion per year. Granted, this must be compared to the cost of childhood coverage of Medicaid. But since that was $132 billion total, this means that even with a standard 3% discount rate, there appeared to be a 7% return on investment per year on childhood Medicaid.

Further, he found that (with additional discounting for both costs and benefits), the government likely recouped more than a quarter of the additional cost for Medicaid for children between 2000 and 2014. More than half of that was from reduced welfare benefits for adults, more than a quarter from increased income tax revenue, and 13% from reduced use of public insurance.

Although it didn't make the top-line results, the analysis showed that the longevity effects of childhood Medicaid alone, even with a conservative estimate of lives saved, were more than $291 billion. That didn't include the value of improvements in physical and mental health to the cost-effectiveness of the program.

This isn't the first study to show that Medicaid provides a significant return on investment. It's also not the first to show that it seems to benefit children even into adulthood. It also doesn't answer the arguments of some that Medicaid may not be as good as private insurance might be. But compared to nothing, which is what many without Medicaid had before it was instituted, the program has provided significant benefits in many ways, even since its inception.