By Molly Frean and Benjamin D. Sommers, Harvard T.H. Chan School of Public Health
& Jonathan Gruber, Massachusetts Institute of Technology
One of the primary goals of the Affordable Care Act (ACA) was to reduce the number of individuals without health insurance in the United States. Prior to the passage of the legislation, this group was very heterogeneous in nature. The uninsured population included low-income families eligible for Medicaid but not yet enrolled; parents with incomes near or below the poverty level but in states with very restrictive eligibility criteria; and middle-income families for whom coverage was simply too expensive and/or considered unnecessary. Through its unique combination of Medicaid expansion, tax credits for health insurance purchased through the marketplaces, and the individual mandate, the ACA was projected to help millions of families and children obtain affordable coverage suited to their healthcare needs. Estimates from early 2016 indicate that as many as 20 million Americans have gained coverage under the law thus far. But beyond this overall population estimate remain several important policy questions: Which of the ACA’s policies were most responsible for the coverage gains? Which demographic groups and types of households were most likely to gain insurance under the ACA?
The objective of our research was to measure the experience of different demographic and geographic groups under the ACA and to understand how the law’s various provisions – Medicaid expansion, marketplace subsidies, mandate – worked together to produce changes in coverage. We focused on three family types: single adults, adult couples, and families with children. To accomplish our objective, we utilized the largest household survey in the United States, the American Community Survey. The only respondents excluded from our analysis were (1) the elderly, as the ACA’s coverage expansions did not apply to them; and (2) Massachusetts residents, due to their unique experience with health reform similar to the ACA in 2006. For each family in the dataset, we estimated their Medicaid eligibility, tax credits for marketplace coverage, and tax penalty for the individual mandate. We then evaluated how these characteristics corresponded to changes in the family’s insurance status in the survey.
Overall, between the pre-ACA period (2012-2013) and the expansion’s first full year (2014), we found that the non-elderly population experienced a 3.4 percentage-point reduction in the percent uninsured after the ACA, a 1.6 percentage-point increase in Medicaid, and a 1.4 percentage-point increase in private coverage. Which policies mattered? We found that the biggest drivers of the coverage gains were Medicaid and marketplace tax credits. For Medicaid, we found that the law not only increased take up among those newly eligible but also among those who were already eligible but had not previously enrolled, a phenomenon known as the “woodwork effect” or “welcome-mat effect.” The ACA simplified the Medicaid application process and also led to much greater levels of awareness about insurance coverage, which likely fueled this change.
When decomposing our results by family type, we found that premium subsidies were most important in helping adult couples and single adults obtain insurance. We found that new Medicaid eligibility was important for all three family types, while the welcome mat effect was substantially lower in families with children (consistent with already high enrollment rates among children eligible for Medicaid/CHIP even before the ACA).
We also completed analyses based on different ACA-related state policies. We compared marketplace type – state-based or federal – and found that take-up rates of tax credits for coverage were almost twice as high among states running their own marketplaces. We also compared states based on their Medicaid expansion status – early expanders (2011-2013), 2014 expanders, and non-expanders. This shed light on our welcome mat effect, showing that approximately half of the national effect occurred in the six early expansion states (California, Connecticut, Minnesota, New Jersey, Washington, and the District of Columbia). In other words, the early expansions laid the groundwork and essentially “primed the pump” for increased Medicaid participation in the future.
Our work has several important takeaways. First, state policies matter. The combination of expanding Medicaid and running a state-based marketplace was the most effective method of reducing the uninsured rate. Second, our analysis found no reduction in private insurance due to the expansion of Medicaid, though some of the law’s critics worried this “crowd-out” of private coverage would drive up program costs. Finally, we found little effect of the individual mandate’s tax penalty in 2014, but this finding should be interpreted with caution. This may reflect low awareness in the general public about the details of the mandate penalty and exemption rules, but it also may be attributable to the relatively low levels of the penalty itself in the law’s first year. Future research that evaluates longer-term impacts of the ACA will be invaluable in enhancing our understanding of the post-health reform landscape.
Note that this research is currently a working paper, and results are subject to change. For more details on this study, please see www.nber.org/papers/w22213.