The Patient Protection and Affordable Care Act, commonly referred to as the ACA and signed into law on March 23, 2010, was the most significant reform of the American health-care system since the passage of Medicare and Medicaid a half century earlier. As former President Barack Obama noted in his personal assessment, the law was intended to “improve the accessibility, affordability, and quality of health care” (Obama 2016). In service of these goals, the “affordable care” portion of the measure sought to expand coverage to the uninsured through Medicaid expansion and the creation of insurance marketplaces with sliding-scale premium subsidies, cost-sharing subsidies, and rate restrictions, as well as the requirement that dependents be permitted to remain on parental insurance plans up to age twenty-six. The “patient protection” portion included new regulations aimed at increasing access and improving insurance coverage, such as guaranteed issue, a prohibition on preexisting condition exclusions, no annual or lifetime caps on expenditures for covered services, coverage of essential health benefits, and free preventive care, among others. This portion also included provisions implementing pilot and demonstration projects aimed at exploring new payment and care models such as accountable care organizations or bundled payments, and new care coordination models for dual Medicare-Medicaid eligibles and other populations. Last were a number of additional provisions—such as increased funding for community health centers and incentives for states to continue rebalancing their Medicaid long-term care spending toward home- and community-based services—also intended to improve the availability of health care and its alignment with need (for summaries of ACA provisions, see American Public Health Association 2012; Kaiser Family Foundation 2013).
Viewed narrowly, a primary focus of the law was to extend health insurance to the approximately forty-nine million non-elderly individuals who were uninsured in 2010 (DeNavas-Walt, Proctor, and Smith 2012). Although determining exactly how much of the ensuing increase in insurance coverage can be attributed directly to various elements of the law is challenging for reasons we discuss, simple estimates derived from extending the preexisting trend in uninsured rates suggest that more than eighteen million non-elderly adults gained insurance, amounting to a 46 percent reduction in the number of non-elderly adults without insurance (Blumberg, Garrett, and Holahan 2016). In promising to extend health insurance to most of the uninsured, the ACA had the potential to offset the toll that low incomes and financial uncertainty take on the large share of poor and uninsured citizens. In a neoliberal era, it also promised to achieve some measure of redistribution, with benefits funded largely by the affluent through new taxes on high earners and new fees on health-care stakeholders.
Although the number of non-elderly adults gaining insurance potentially due to the ACA is a relatively small fraction of the total population, virtually the entire population was affected to some extent by its provisions, including substantial regulatory components that changed important rules on the ground for private insurance and the existing public insurance programs. However, although these regulatory effects were broad and important, they were in many ways more difficult for individuals to recognize. The ACA used multiple levers of change, many of which were quite hidden to the ordinary observer, which may help explain why such a significant reform had in many cases modest behavioral effects, especially on individuals, as discussed in this introduction and in many of the articles in this issue.
Although nominally focused on changing various components of the health-care system, the ACA has touched on a broad variety of social institutions and societal relationships. Connections between states and the federal government, between governments and health-care providers, between governments and individuals, and between individuals and firms all were altered by the ACA. Taken together, the elements of the ACA had the potential to spur major societal changes beyond extension of health insurance coverage. Indeed, the law’s passage was followed by continuous challenges in Congress, in the courts, and in the states, due in part to the far-reaching nature of the law. In addition to spurring considerable political discourse and action, these challenges affected the ACA’s implementation and may have changed its impacts. Six years after the law passed, elections ushering in unified Republican control of government at the national level and Republican control of government in many states potentially shifted the environment surrounding the law and its implementation as well.
The health reform has prompted a great deal of research among social scientists exploring its origins and effects. However, much of what researchers, policymakers, and the general public might want to know about the Affordable Care Act is difficult to learn. The ACA was sweeping in its reach, touching many aspects of the health-care system specifically and American society more broadly, and its passage coincided with the early years of recovery from a profound economic shock, the Great Recession. In addition, important but not fully understood long-term changes in health care and labor markets were occurring before the ACA and may have been affected by it in ways that are difficult to disentangle. Consequently, identifying specific effects of the ACA from those of other factors likely to affect outcomes of interest is challenging. Research designs that involve comparing outcomes before and after some aspect of the ACA took effect are fraught with the difficulty inherent in pulling apart competing causal factors, any or all of which may be operating. The coincidence of the ACA’s passage with the start of an economic recovery makes focusing on deviations from existing trends less convincing than such an analysis might be in calmer economic circumstances. In many cases, both opponents and supporters of the ACA can point to results from such analyses that support their views.
These difficulties imply that more credible research assessing the ACA has largely taken one of two forms: descriptive assessments coupled with an explicit recognition of their descriptive nature and research focusing on aspects of the ACA that offer the possibility of usable exogenous variation. As a result, research on some aspects of the ACA has been considerable but on other aspects minimal. The extension of parental coverage to young adults, which affected those age twenty-six and younger but not those older than that; competition in markets for individual insurance across the country, which had differential levels of preexisting market depth; and especially the expansion of Medicaid coverage, which was made optional to states by the Supreme Court’s decision in the 2012 case National Federation of Independent Business v. Sebelius have received more attention. 1 In addition, substantial literatures on the effects of the ACA on health-care delivery reform (for a review, see Blumenthal, Abrams, and Nuzum 2015) and on health outcomes have emerged. However, other components—often the less visible ones—have received less consideration because of both the difficulties in finding credible research designs and the data limitations; in addition, work on the economic, political, and sociological effects of many components of the ACA is scanty. The outbreak of the COVID-19 pandemic in 2020 heightens the importance of many of these questions.
This issue of RSF begins to fill these gaps with a series of articles from social scientists assessing these broader effects of the health reform. In this introduction, we situate these articles by reviewing the relevant literature on the ACA’s economic, political, and social effects. We examine extant discussion of the implications of the ACA’s design for private insurance markets and the major public insurance programs as well as the effects of the law on noninsurance components (such as the health-care workforce, providers, and so on). We examine the responses of states to the many decisions the ACA required of them regarding insurance exchanges, Medicaid expansion, and navigator support. We then turn to effects of the ACA on individuals, both nonpolitical effects (health insurance coverage and access, financial security, labor-market effects, and effects on family structure), and political effects (changing patterns of political behavior and attitudes). We confine our review to topics relevant for the examination of the broad social, economic, and political effects that the articles included in this issue examine; we do not review the voluminous literatures on health-care delivery reform and health outcomes, including literature on changes in the treatment of various health conditions (such as substance use disorder, cancer, obesity, or many others) affected by expanded access to coverage for such treatment, nor do we review health reform case studies of individual states. Our review reveals that despite the great volume of research the ACA has inspired thus far, many additional areas are in need of examination. In the hope of fostering a continued research agenda among social scientists, we conclude by highlighting areas where more work is needed.
We begin our review by discussing various ways in which the ACA’s provisions have changed the health-care landscape in the United States, how implementation has proceeded, how the ACA’s design elements affect private insurance, what the implications are of the ACA’s noninsurance and public insurance components, and how the states have responded to the ACA’s provisions. After providing this context on the health and policy impacts, we turn to the extant work on the economic, social, and political effects of the ACA.
The Affordable Care Act is an extraordinarily complex law with thousands of provisions, not to mention a politically contested one subject to unrelenting attacks by partisan opponents. The status of high-profile provisions garners much media attention and scholarly consideration (for example, Blumenthal, Abrams, and Nuzum 2015), whereas far from the public glare a great deal of quieter activity is under way, many provisions going into effect and others falling by the wayside. Helen Levy, Andrew Ying, and Nicholas Bagley (this issue, 2020) go beyond existing overviews of high-profile provisions to analyze the implementation status of approximately two hundred “key” provisions as identified in prior research. They discuss each of the ACA’s titles in turn, providing a helpful overview of the law, and delineate five categories of reasons some provisions were invalidated, repealed, or abandoned, including “legal challenges,” “born to fail,” “interest group pressure,” “failure to thrive,” and “executive branch sabotage.” The authors discuss examples of key provisions that fall into each category and provide a brief narrative about salient events surrounding each one. Overall, however, their analysis indicates that a majority of the law has been implemented. Subsequent articles in the issue—and the rest of our literature review—in turn examine the social, economic, and political effects of many of the provisions that have been at least partly implemented.
Before the ACA was enacted, private insurance could be obtained either through a group, such as an employer, or in the nongroup market. By far the largest share of private coverage was employment based: according to data from the 2010 Current Population Survey, in 2009 only 9 percent of the individuals younger than sixty-five who had private insurance at some point during the year had directly purchased insurance only (see U.S. Census Bureau 2010, table HI01). The private nongroup market suffered from a number of problems, including lack of access to insurance for individuals with preexisting health conditions, high administrative costs, limited choices, and continued exposure to health expenditure risk, with caps on coverage or exclusions of coverage for certain conditions being common features of privately purchased plans. Employer-sponsored insurance (ESI) markets by many measures functioned better, although some plans also had annual or lifetime limits on coverage so that enrollees were exposed to the risk of catastrophic health expenditures, some imposed waiting periods on coverage for preexisting conditions for new employees, and small-group plans in some states had higher premiums attributable to required medical underwriting. Nevertheless, ESI represented the largest source of insurance for the under sixty-five population, nearly 73 percent in that age group who had insurance at some point in 2009 having employment-based coverage (see U.S. Census Bureau 2010, table HI01). The ACA was therefore intended to address the problems of private insurance markets but not reduce the extent of ESI coverage.
The most visible aspect of the ACA’s impact on insurance markets was the establishment of health insurance exchanges (also called Marketplaces) in which individuals could shop for individual or family policies. Importantly, these policies must be offered to anyone, with pricing variation permitted only on the basis of geography (market rating area, typically metropolitan statistical areas [MSAs] plus the remainder of the state not included in an MSA), family composition, age (the ratio of premiums for the oldest to the youngest enrollees not to exceed three to one), and tobacco use. Plans offered must fit into one of five tiers within which all plans must be actuarially equivalent: one catastrophic high-deductible tier generally available only to younger individuals, and four “metal levels”—bronze, silver, gold, and platinum—that correspond to increasingly generous coverage. Within tier, plans compete primarily on price (premium plus cost-sharing requirements) and the network of providers included in the plan. Regulations also apply to both employer-sponsored and Marketplace plans, including a minimum ratio of benefits to premiums (different levels for large-group and small-group or individual plans), standards for “essential health benefits” that must be covered, an annual out-of-pocket expenditure limit, and a ban on annual or lifetime coverage limits. Tax credits to reduce the cost of the premium and reduced cost sharing on a sliding-scale basis are available in the individual market to families with incomes below 400 percent of the federal poverty level (FPL). In addition, if offered, dependent coverage must be made available to unmarried adult children younger than twenty-six.
To ensure that employer-sponsored coverage was not reduced, firms with more than fifty workers were required to offer “affordable” coverage meeting minimum value standards to full-time employees or pay a penalty, although the implementation of the penalty was delayed until 2015; small firms (fewer than fifty workers) were given the opportunity to purchase a health plan to offer through the Small Business Health Options Program (SHOP). Small firms could also drop coverage and allow their workers to enter the individual Marketplaces. Group or individual plans in existence when the law was passed were grandfathered—that is, did not have to meet all the requirements of the law until the insurer or employer made a significant change in coverage or pricing.
These regulations, and the rulemaking that accompanied them, were fundamental changes for private insurance markets, beginning with the clear increase in information available to potential consumers of individual plans. The “metal” tiers and standards for coverage made comparing plans more straightforward, though the extent of the networks of providers available in each plan was a key remaining variable, and one that has generated questions about the trade-offs inherent in “narrow network” plans between lower premiums and ease of access to health care. Both Leemore Dafny and colleagues (2017) and Daniel Polsky, Zuleyha Cidav, and Ashley Swanson (2016) show a clear association between the narrowness of the network and the premium, estimating a 6 to 9 percent reduction in premium with a narrowing of the network, and a larger reduction if both physician and hospital networks are reduced. Aditi Sen and colleagues (2017) find that individuals who are Hispanic or low income constitute a disproportionate share of enrollees in plans with networks that include fewer than a quarter of the physicians in a local area. However, further research is needed on whether narrow network plans have resulted in any health-care access or health implications or function primarily as an effective check on health-care prices. Another concern was that consumers would be unlikely to shop around in subsequent open enrollment periods, particularly given low rates of plan switching in Medicare Part D (Sanger-Katz 2014). Although some evidence indicates that a fair number do change plans on the ACA exchanges (Pearson, Carpenter, and Sloan 2016), to date little research has been undertaken on such plan switching and its effects.
Perhaps because so many of our day-to-day activities are conducted online, it may not be immediately apparent what a structural change Section 3021 of Subtitle C (“Health Information Technology Enrollment Standards and Protocols”) represented in its call for “Electronic matching against existing Federal and State data, including vital records, employment history, enrollment systems, tax records, and other data determined appropriate by the Secretary to serve as evidence of eligibility and in lieu of paper-based documentation.” The requirement for online enrollment capability meant that various electronic records systems, both private and public (state and federal) needed to be able to exchange information that could then be used in an eligibility determination for Medicaid or premium subsidies. This was such an immense undertaking that roll-out of the online enrollment system for the exchanges was, as is well known, less than smooth. Moreover, evidence suggests that problems have persisted in states experiencing particularly difficult roll-outs (Scheuer and Smetters 2018). The establishment of this capability, however, represents a significant change in ease of access to Medicaid—as of January 2019, according to a Kaiser Family Foundation survey of states, for the first time individuals can apply for Medicaid online in all states and can receive eligibility determinations within twenty-four hours in forty-six states (Brooks, Roygardner, and Artiga 2019). In addition, it represents an opportunity for states to streamline eligibility determination for a wider variety of programs (Dorn, Minton, and Huber 2014).
The ACA envisioned certain roles for states in insurance regulation, the federal government taking on some of the regulatory roles that states had formerly held, such as establishing the benefits that qualifying insurance plans would have to provide or mandating employer offering of insurance, but the states being given the opportunity to establish their own state-level marketplaces. However, the roles as envisioned were not necessarily the same as the roles that occurred, many states choosing not to design their exchanges and instead adopting the federal one. Nevertheless, individual insurance markets continued to have considerable variation at the state level. One decision left to states was market definition, in particular, the geographic area that would be considered a single market. Michael Dickstein and colleagues (2015) find that counties that are smaller or more rural have more insurers and lower premiums when they are “bundled” with larger counties, although more heterogeneous regions (in terms of proportion urban versus rural) have fewer insurers and higher premiums, suggesting a trade-off for states between bundling smaller counties with larger ones and keeping more dissimilar counties separate.
Once the markets are defined, the decision about whether to enter the market is up to insurers. Researchers have noted two countervailing effects of additional entry into insurance markets. First is the typical effect of price competition arising through additional entry, which would tend to push down prices to consumers. However, because of bargaining between insurers and providers, the entry of additional insurers into a market is not guaranteed to lower prices to consumers because individual insurers have less bargaining power vis-à-vis providers when insurers are more numerous (see, for example, Moriya, Vogt, and Gaynor 2010; Ho and Lee 2017). Researchers have studied the impact of competition on consumer prices in the ACA’s individual insurance market. Focusing on arguably exogenous sources of variation in the number of insurers, they have found that the entry of an additional insurer has generally resulted in a reduction in prices to consumers of between 4 and 5 percent (Dafny, Gruber, and Ody 2015; Abraham et al. 2017; Lissenden 2017).
Particularly given that markets with more insurers have been shown to offer lower prices to consumers, a concern persistently expressed by observers of the ACA has been exit of insurers from the individual markets. In a series of issue briefs, analysts at the Kaiser Family Foundation have tracked the performance of insurers in the individual market since the passage of the ACA (see, for example, Cox, Levitt, and Claxton 2017; Fehr, Cox, and Levitt 2018a, 2018b). Examining medical loss ratios (the share of premiums paid out in claims) in the individual insurance market, they find that medical loss ratios rose to unprofitable levels in the first two years of the ACA marketplaces but began to decline thereafter, suggesting that after an initial period of inadequate information about the individuals purchasing insurance that led insurers to set their prices in the market too low for the level of risk, insurers in the market have begun to gain information that allows them to set prices more accurately. Nevertheless, many of the markets have been characterized by instability and uncertainty about the level and nature of enrollment. Mark Hall (this issue, 2020) assesses the sources of instability in individual insurance markets using documentary research and case studies from ten states. He focuses particularly on the roles of actuarial uncertainty (which arises because insurers must account for unknown responses to known changes in market rules) and political uncertainty (which arises because insurers do not know whether and how regulations might change) in insurer pricing and entry decisions. Based on interview evidence, he concludes that actuarial uncertainty is not inherently destabilizing, although political uncertainty is; he points to regulatory flexibility on the part of the states and the subsidy structure as ensuring the resilience of the individual insurance markets in the face of political uncertainty. Jean Abraham (this issue, 2020) also studies instability in the individual markets, classifying local markets as more or less volatile based on changes over time in insurer participation and premiums and more or less vulnerable based on insurer participation and premiums in 2019. She finds that by her measure nearly a third of local markets experienced high volatility in the segment of the market offering subsidized plans, a slightly smaller share experiencing high volatility in the unsubsidized portion. She classifies markets as vulnerable if they have below-median insurer participation and a premium level above the median premium, and finds that vulnerable local markets are more likely to be rural, to have less healthy populations, and to be in states that have not expanded Medicaid.
It is not surprising that the individual insurance markets are affected by the states’ decisions on whether to expand Medicaid, given that Medicaid expansion removes more risky low-income individuals from the private risk pools. As a result, individual market premiums are expected to be lower, on average, in states that expanded Medicaid. Aditi Sen and Thomas DeLeire (2018) find this to be the case, comparing premiums for plans on both sides of a state border where one state expanded Medicaid and the other did not. Similarly, Lizhong Peng (2017) finds that premiums fell in Pennsylvania and Indiana when Medicaid expansion occurred.
Researchers have also assessed whether either the hope of the ACA’s designers that employer-sponsored insurance coverage would continue as a mainstay of the insurance coverage structure, or the concern of the ACA’s opponents that ESI coverage would fall significantly, have occurred. Overall, employer-sponsored insurance seems to have largely remained stable: Abraham, Anne Royalty, and Coleman Drake (2016), Frederic Blavin and colleagues (2016), and Adele Shartzer, Blavin, and John Holahan (2017) all find little change in ESI offerings or take-up post-ACA. One area that has elicited some concern is the small employer market, innovations such as the SHOP marketplace never becoming operational (Curran 2017). Focusing on small employers, Sabrina Corlette and colleagues (2017) use qualitative data from six states to describe changes in the small-group ESI market, noting that though rates improved for some small groups, employers with younger, healthier employees saw rising premiums, which has led some small employers to pursue lower priced but less complete coverage or other options. However, in comparing the small-group market with the individual market, Abraham, Royalty, and Drake (2019) note that on average, small-group markets appear to be functioning better than individual markets, offering more plan types and lower premiums. Overall, how people have fared in the small-group and individual insurance markets as a result of the ACA would seem to be important areas for further research.
Although much of the scholarly work on the Affordable Care Act has thus far centered on the private insurance components of the law, such as the exchanges and new regulations, the law contained multiple provisions for Medicaid and Medicare as well as noninsurance provisions affecting many aspects of health care. A number of these provisions affect health-care delivery and, ultimately, health outcomes; and a robust and growing health services research literature examines those outcomes. However, we have found relatively little scholarly analysis thus far of the social, political, and economic effects that these many provisions might exert on individuals and institutions. Hence we highlight hypotheses that observers have put forth about their likely effects and describe analytical findings where they exist.
By increasing health insurance coverage, the ACA was projected to increase the demand for health care, putting pressure on providers and provider participation. For example, Stephen Parente and colleagues (2017) estimate that demand for physicians, licensed practical nurses, and medical aides will increase more than 10 percent between 2014 and 2021 relative to a no-ACA baseline, and demand for other occupations—such as technician, registered nurse, and home health aide—will grow at somewhat lower rates. Their model predicts differential wage increases across provider types, with greater wage increases among health-care occupations requiring more education and training, such as physicians and registered nurses. Upward pressure on wages could also counteract other ACA provisions meant to control rising health-care costs.
The ACA includes additional measures intended to meet the increased demand for providers created by expanded health insurance coverage. The law permanently authorized and increased funding for the National Health Service Corps (NHSC) program, which provides scholarships and loan repayment to primary care providers who work in underserved areas. However, despite some increases in field strength—the total number of clinicians in the program—the number of open NHSC positions is higher than the number of NHSC providers (Heisler 2018). The ACA also increased Medicaid payments to primary care providers to Medicare levels for two years, 2013 and 2014. However, studies have found no apparent effect on physician participation in Medicaid, perhaps because the bump was temporary (Decker 2018; Neprash et al. 2018).
According to at least one survey, efforts to increase the size of the health-care workforce or to encourage physician participation in Medicaid could improve enrollee satisfaction. A 2014–2015 survey of Medicaid recipients found more patient satisfaction and greater access in states with higher physician participation per capita in Medicaid (Barnett, Clark, and Sommers 2018). Inequalities persist, however: racial and ethnic minorities report less satisfaction and access than white recipients. Ongoing economic, political, and sociological questions remain about whether the size of the health-care workforce and distribution across health occupations and geographic locales will meet patient needs in the future and how to reduce disparities in access across income, race-ethnicity, and other demographic categories.
Although the impacts on the health-care workforce have not yet been fully assessed, several studies have examined the impact of the ACA on hospitals and community health centers. By expanding Medicaid, the ACA has reduced the amount of uncompensated care that hospitals provide (Blavin 2016; Dranove, Garthwaite, and Ody 2016), particularly among hospitals serving a disproportionate share of low-income patients (Camilleri 2018). Although some substitution of Medicaid expansion funding for existing state or local safety net hospital funding was likely (see Duggan, Gupta, and Jackson 2019 for evidence on this substitution for California), the evidence to date indicates that hospitals, particularly those serving poorer populations, have benefited financially from the ACA. Richard Lindrooth and colleagues (2018) find that Medicaid expansion under the ACA is associated with better financial health for hospitals and lower likelihood of closure, particularly in rural areas. This could be good financial news for safety net hospitals if as a result they have greater capacity to treat more private insurance patients (at higher reimbursement rates), but alternatively could represent a financial threat if non–safety net hospitals attract healthier Medicaid patients and sicker ones remain at safety net hospitals.
Another ACA provision doubled federal funding for community health centers (CHCs), which provide care to twenty-six million Americans (Rosenbaum 2017). Researchers report that this increase in funding, together with Medicaid expansion, has significantly increased patient volume and reduced the shares of uninsured patients at CHCs (Han, Luo, and Ku 2017). Congress extended the original five-year grant for CHCs several times, but future funding remains uncertain (Lewis et al. 2019). Cuts in CHC funding would disproportionately affect access for low-income individuals and rural residents (on the latter, see Cole et al. 2018).
The ACA includes demonstration projects aimed at better coordination of care for those dually eligible for Medicaid and Medicare. However, enrollment in these demonstration projects remains below projections (Grabowski et al. 2017), and advocates have voiced concern that the dual eligible demonstration projects do not do enough to address racial and ethnic health disparities (Sharma 2014). The ACA also provides new options for creating medical home models of coordinated care. Questions remain as to whether the medical home concept reduces racial-ethnic disparities in access for health services (National Academies of Sciences, Engineering, and Medicine 2015) or brings parity to mental health services (Sahasranaman 2017).
One difficulty the patient-centered medical home model faces is that its principles of comprehensive care that includes disease prevention and management of chronic conditions may be challenging for small physician practices to implement due to the financial and other resources required. Radhika Gore and colleagues (this issue, 2020) study the implementation of two particular population-health strategies espoused in the ACA—electronic health records and community health workers—in the context of efforts to implement hypertension control strategies among small practices serving South Asian immigrant communities in New York City. Using a method of semi-structured interviews and on-site observation of clinic workflow before and after implementation, they find that although some aspects of the strategies strengthened care provision and patient engagement, others proved challenging to implement or were not perceived as helpful by providers; they outline some of the challenges faced in making these ACA population-health strategies successful in the context of small providers and culturally distinct communities.
Beyond demonstration projects for dual eligibles and other populations, a notable feature of the ACA was an effort to make institutional investments in demonstration projects more generally to better ensure their translation into policy, such as creating a new Center for Medicare and Medicaid Innovation (CMMI), increasing the budget for demonstration projects and allowing them to be non-budget-neutral, exempting some elements from judicial and administrative review, and increasing the authority of the secretary of the Department of Health and Human Services to expand Medicare and Medicaid demonstration projects without congressional approval. Philip Rocco and Andrew Kelly (this issue, 2020) examine fourteen demonstration models CMMI pursued between 2012 and 2018. Despite the increased budgets and authority, they find that only two new payment and delivery models have been certified for expansion. The actuarial certification process, they conclude, requires measurements of quality and attributions of savings that are difficult to meet, particularly for complex demonstration projects involving many types of stakeholders. Barriers to innovation therefore remain, even with increased discretion and resources.
The ACA contained many provisions affecting Medicare. A substantial health services research literature examines the effects of delivery system and payment reforms such as programs to reduce unnecessary hospital readmissions, develop accountable care organizations, introduce bundled payments and other value-based rather than volume-based reimbursement, and so on, which we do not examine here. Other provisions were intended to improve Medicare benefits by expanding preventive services, providing an annual wellness visit, and closing the prescription drug coverage gap known as the “donut hole” (reducing cost sharing in the coverage gap from 100 percent before the ACA to 25 percent in 2019 for brand name medications and in 2020 for generic drugs). Research indicates unequal patterns in utilizing these new benefits. For example, only a small share of Medicare beneficiaries received an annual wellness visit in the provision’s first four years, increasing from 7.5 percent in 2011 to 15.6 percent in 2014, whites, women, urban residents, non-dual eligibles, and those from higher-income areas being more likely to do so (Ganguli, Souza, and McWilliams 2017). The ACA also introduced changes to private Medicare Advantage (MA) plans, which enroll a large proportion of Medicare recipients. MA plans are subject to the medical loss ratio provision, limiting the amount they can spend on administrative costs, profits, and other non-health-care aspects to 15 percent of their Medicare payments. The ACA also sought to reduce payments to MA plans, which were higher than traditional Medicare payments, 14 percent higher per capita in 2009 (ASPE 2014), in part by reducing the per-enrollee “rebate” an MA plan received when its bid was below the benchmark rate in the county, based on traditional Medicare spending in the county. Senior citizens exerted pressure on Congress to avoid such cuts (Kelly 2015), a kind of protective constituency policy feedback, although MA plan payment rates did fall, on average, to be equivalent to traditional Medicare, and MA enrollment grew significantly (Guterman, Skopec, and Zuckerman 2018).
The ACA also contained a large number of provisions addressing public health, including the creation of a public health council and a $15 billion public health fund, “the first time that a comprehensive public health strategy, with dedicated funding, was articulated in federal law” (Chait and Glied 2018, 508). Other provisions aimed at prevention, oral health, immunizations, laboratory capacity, minority health, diabetes, childhood obesity, women’s health, tobacco cessation, and so on (for a summary, see Chait and Glied 2018). Although some work on the effects of these initiatives on health outcomes has started to emerge, to the best of our knowledge none on the political, social, or economic effects has, perhaps because the impact of the ACA on such effects would be difficult to disentangle from myriad other social factors.
The ACA included multiple streams of financing that were intended not only to support new spending obligations such as Medicaid expansion and health insurance subsidies, but also to change both health-care and health insurance incentive structures and to magnify the law’s redistributive effects. Some financing sources were imposed on health-care stakeholders, such as new annual fees on pharmaceutical manufacturers and health insurers as well as taxes on medical devices and indoor tanning services. More relevant for possible social, political, or economic effects were tax changes for individuals, ranging from lower limits on flexible spending accounts for medical expenses and an increased threshold for itemized deduction of unreimbursed medical expenses, to changes with clearly redistributive implications, such as increased capital gains and Medicare payroll taxes for high earners and the so-called Cadillac tax.
Among the financing streams, the Cadillac tax was the subject of the most hypothesizing and analysis. A 40 percent excise tax on employer-sponsored health benefits that exceed certain thresholds, the Cadillac tax was intended not only to raise revenue but also to partly offset the tax exclusion for employer-sponsored insurance and to discourage employers from offering health plans that are so comprehensive that they encourage overuse. The thresholds, $10,800 for individuals and $29,100 for families, were indexed to the Consumer Price Index, which tends to increase at a lower rate than health-care costs, meaning that the tax would apply to more plans each year. Because of this and other controversies around the tax, its implementation was delayed several times. Thus scholarly analysis could not assess the effects of the Cadillac tax, but instead focused on estimating which workers would be affected by the tax when it came into effect (Claxton and Levitt 2015; Herring and Lentz 2011–2012; Lowry 2015). Mark Warshawsky and Michael Leahy (2018) estimate, for example, that 12 percent of workers would be affected at the outset, the highest concentrations among those who tend to have more generous benefit packages: union members, workers in education occupations, and workers in the top quartile and top decile of earnings. Workers in the Northeast and West would be more likely to be subject to the tax than those in the South or Midwest because of regional variations in health-care costs. Sherry Glied and Adam Striar (2016) speculate that the tax would ultimately be more progressive than first thought because it would most affect workers with health savings accounts. The tax may have induced employers and workers to move toward plans with greater cost-sharing or narrower provider networks, and it could have increased financial risk for the demographic categories most likely to be affected. That the Cadillac tax would fall on the health plans of more highly resourced and organized individuals resulted in political pressure to prevent or blunt its implementation. Congress finally repealed the tax altogether (along with the ACA’s taxes on the health insurance and medical device industries) in a budget bill passed in December 2019 and signed by President Donald J. Trump (Keith 2019).
Finally, by extending health insurance to the previously uninsured with funds extracted from higher-income households, the ACA will have effects on American patterns of inequality. Scholarly work on the social, political, and economic effects of the law’s distributional consequences has just begun. Before most ACA provisions were implemented, Henry Aaron and Gary Burtless (2014) predicted that money incomes would increase slightly for the bottom quintile and fall slightly for other income groups, though none of the changes were large and most changes were concentrated in the bottom two deciles. Kevin Griffith, Leigh Evans, and Jacob Bor (2017) find that the ACA decreased socioeconomic disparities in insurance coverage, the gap in the insurance rate between those with incomes above $75,000 and below $25,000 shrinking over time. The decreases were larger in Medicaid expansion states, where the gap fell from 31 percentage points to 17 percentage points, than in non-expansion states, where the gap decreased from 36 percentage points to 28 percentage points. As the ACA extended health insurance access to adults in ways unrelated to their relationships with employers, spouses, and children, disparities were also decreased across sociodemographic groups, men, black and Latino adults, and adults with less education gaining insurance at greater rates and narrowing preexisting gaps by gender, race and ethnicity, and education (Gutierrez 2018). Naomi Zewde and Christopher Wimer (2019) use the ACA’s Medicaid expansion to estimate that Medicaid coverage reduced the nation’s poverty rate by about 1 percent, the effect concentrated in the non-elderly adult population that was the focus of the expansion.
One of the defining characteristics of the Affordable Care Act was the large role assigned to states. Although the law created a national framework for extending health insurance to more Americans and for addressing practices in the private insurance market that reduced access, states were charged with important implementation and policy responsibilities. These included deciding whether to create an insurance exchange, implementing new insurance regulations, and determining whether to participate in various initiatives and demonstration projects aimed at health education, healthy living, health-care delivery, payment structures, and so on (Weil and Scheppach 2010). The Supreme Court’s decision rendering Medicaid expansion optional further heightened the importance of state decision making. Proponents of state control have traditionally lauded the possibilities for innovation and for tailoring policy to local conditions that such subnational policy responsibility affords. But states vary in their levels of expertise, previous policy experience, administrative capacity, and fiscal strength, not to mention political climates and partisan control of government. Delegating important aspects of the ACA to the states ensured substantial uncertainty and variation in the law’s operation on the ground.
A number of scholars have examined the drivers of state choices in implementing the Affordable Care Act, including their decisions regarding health insurance exchanges (Jones, Bradley and Oberlander 2014; Rigby and Haselswerdt 2013; Shor 2018) and expanding Medicaid (Barrilleaux and Rainey 2014; Jacobs and Callaghan 2013; Shor 2018). Most find that party control of state legislatures and governorships is a dominant factor in explaining state policy choice, Democratic-led states implementing the ACA more enthusiastically. Some also find that state administrative capacity and previous policy experience, such as previous expansions of Medicaid, were also related to ACA implementation (Rigby and Haselswerdt 2013; Jacobs and Callaghan 2013; Haeder and Weimer 2015). Evidence on the influence of interest groups is mixed. On the one hand, the greater presence of business and professional lobbying groups is associated with less state progress on Medicaid expansion (Callaghan and Jacobs 2016). On the other, the greater presence of pro-expansion interests was influential in one study—public interest groups (Callaghan and Jacobs 2016)—but not another—safety net providers (Grogan and Park 2017). Public opinion has an independent effect on policy choices in some analyses (Rigby and Haselswerdt 2013; Grogan and Park 2017), but only an indirect effect, through elected officials, in others (Shor 2018). Also, race matters. In the degree to which public opinion is associated with Medicaid expansion decisions, the opinions of white state residents matter, not those of nonwhites (Grogan and Park 2017). That ACA implementation occurs at a time of heightened political polarization complicates the usual pathways by which policies diffuse from one state to another (Volden 2017). Finally, and in some ways most normatively concerning, actual health insurance need in the state does not seem to matter for politicians’ decision making (Jacobs and Callaghan 2013; Barrilleaux and Rainey 2014).
Most of these studies use state-year as the unit of analysis, although Boris Shor (2018) is able to use the state legislative district due to methodological advances that permit the estimation of public opinion and legislator ideology at that level of disaggregation. Unlike some of the individual-level studies, which use causal designs, most state studies are based on observational data.
One early decision states faced was whether to implement their own health insurance exchange or use the federal one. A state-level exchange could be attractive to conservatives as a market-model solution meant to foster competition and drive down health insurance prices, as was extolled during the Massa-chusetts reform under Republican Governor Mitt Romney and in various Republican health reform plans over the years (Jacobs and Skoc-pol 2010). Also, a state-level exchange would heighten state control and ward off the “partial preemption” of the federal exchange (Rigby and Haselswerdt 2013). On the other hand, running a state exchange could be viewed as embracing the progressive ACA reform championed by President Barack Obama, a Democrat. Also, the federal rules on exchanges and minimum standards meant that the exchanges were a form of “one-tailed devolution” (Conlan and Posner 2011), in that it was easier for states to proceed in a liberal direction than in a conservative one (Rigby and Haselswerdt 2013). Republican state lawmakers were caught in a dilemma: state-level exchanges appealed to partisans who sought to minimize federal intervention, but also represented entrenchment of a law they opposed and might complicate legal challenges to the ACA (Jones, Bradley, and Oberlander 2014).
In examining early steps that states could take in setting up exchanges, Elizabeth Rigby and Jake Haselswerdt (2013) find more activity in states with more supportive public opinion and with either Democratic governors or a high share of Democrats in the state legislature. Simon Haeder and David Weimer (2015) find greater state cooperation in setting up exchanges in states with unified Democratic legislatures and less cooperation in those with a Republican governor or a Republican elected insurance commissioner. Shor (2018) examines state legislator roll-call votes on several ACA outcomes. He too finds that state exchanges were opposed more by conservative and Republican state legislators, legislator ideology having a stronger effect than legislator partisanship; district public opinion did not have an effect independent of the legislator characteristics. Thus analyses of both state-level decisions and individual legislator roll-call voting indicate that conservatives and Republicans were less likely to support state-level exchanges than liberals and Democrats were, despite the possible attractiveness of such exchanges to conservatives.
Another decision states had to make was whether to expand Medicaid after the Supreme Court effectively made expansion optional. An examination of early decisions regarding Medicaid expansion in 2012 and 2013 (such as issuing gubernatorial or legislative statements supporting expansion, applying for federal planning grants, or streamlining Medicaid application processes) found that party control is an important though not perfect determinant (some Republican-led states moved toward expansion). States were also more likely to have taken steps toward expansion if they had previously expanded Medicaid or if they had more administrative capacity (as measured by insurance oversight, policies against Medicaid fraud, and existing high-risk pools for the medically needy). However, states with greater need—those with lower average per capita income—were less likely to have done so (Jacobs and Callaghan 2013).
Lawrence Jacobs and Timothy Callaghan (2013) examined bivariate patterns only. Rigby (2012) analyzed state resistance to the ACA in 2010 and 2011, as measured by a three-item index adding whether the state had filed a lawsuit challenging the ACA, had passed legislation in opposition, or had forgone federal planning grants, finding that GOP control of government (governor, attorney general, or insurance commissioner) was the most important factor, accounting for half of the total variation in outcomes. Also important was state public opinion. State capacity (for example having a less professionalized legislature) was more modestly associated with resistance, as was the degree of change the ACA represented from current policy, such as the magnitude of Medicaid enrollment or the net costs to state budgets the ACA would bring (Rigby 2012).
Similarly, an analysis of governors’ decisions to support Medicaid expansion found that gubernatorial partisanship and legislative party control were the most important factors; public opinion did not exert an independent effect, nor did need—support for Medicaid expansion declined with the share of state uninsured population (Barrilleaux and Rainey 2014). In yet another analysis of Medicaid expansion decisions in 2012 and 2013, supportive public opinion is associated with expansion, but only whites’ opinions are statistically significant, not nonwhites’ opinions (Grogan and Park 2017). As in his analysis of insurance exchanges, Shor (2018) finds that state legislators’ roll-call votes on Medicaid expansion are associated more with legislator ideology than legislator party, public opinion again working through those pathways rather than exerting an independent effect on legislators’ voting.
Several researchers have assessed the role of organized groups in state choices on Medicaid expansion. Colleen Grogan and Sunggeu Park (2017) find no statistically significant role for safety net interest group influence (measured by the number of community health centers and the number of patients served per capita) on state Medicaid expansion decisions. However, Callaghan and Jacobs (2016) do find a significant effect: states with Democratic control of government and more public interest and nonprofit lobbyists per capita had taken more steps toward Medicaid expansion, but those with a stronger professional and business lobbyist presence had taken fewer. State affluence, past policy choices, and administrative capacity did not exert independent effects.
Another policy choice was funding navigators, assisters, and certified application counselors to assist consumers in comparing plans on the health insurance exchanges, applying for subsidies, and enrolling (Goodell 2013). Such navigators are typically members of advocacy groups or social service organizations. Variation in state funding of navigator programs is wide; those states that established marketplaces had more funding available given the nature of federal funding sources. States could elect to use their own money as well. During the first open enrollment period, California and Maryland spent as much on navigators as all states with federally run marketplaces combined (Goodell 2013). State variation in advertising and navigator budgets per capita of the uninsured population remained pronounced in the ACA’s sixth open enrollment period in late 2018 (Corlette and Schwab 2018).
Some states erected barriers to navigators, such as stringent licensing and training requirements. In many cases, insurance agents and brokers lobbied for these regulations, viewing navigators as “government-funded competition” (Kusnetz 2013). In some states, local health departments have engaged in the outreach that navigators might otherwise play, as in Houston, where the city’s Department of Health and Human Services headed a collaborative effort aimed at increasing health insurance enrollment even in the absence of Medicaid expansion (Runnels et al. 2016; Williams et al. 2016). Nonetheless, significant state variation in outreach remains, and evidence of navigator effectiveness suggests that this variation—and sharp reductions in navigator funding for federal marketplaces under the Trump administration—could be particularly harmful to vulnerable populations. In the early years of ACA implementation, African Americans and Latinos were more likely than white consumers to seek navigator assistance (Enroll America 2014; Mosqueira, Hua, and Sommers 2015); navigators also proved particularly helpful for those seeking insurance who had low incomes, low levels of health literacy, complex family situations, or limited English proficiency (Pollitz, Tolbert, and Diaz 2018). Overall, those receiving in-person navigator assistance were about twice as likely to enroll as those who tried to access insurance without help (Enroll America 2014).
The ACA includes many initiatives and demonstration projects in which states could participate. State choices to participate have not been analyzed in many instances. One exception is the article by Lisa Beauregard and Edward Miller (this issue, 2020), which examines state adoption of the ACA’s home and community-based services (HCBS) initiatives from 2011 to 2015, using both cross-sectional and longitudinal models. Many individuals in need of long-term supports and services prefer to receive care at home rather than in a nursing home. Although the number of individuals receiving Medicaid HCBS has increased over time, the initiatives in the ACA were meant to accelerate the shift to noninstitutional care and to address cross-state variation in long-term services and supports rebalancing. The authors find that HCBS initiative adoption was more common among more liberal states, those that had previously adopted HCBS policies, and those with neighboring states that had adopted; in the cross-sectional model, states that had expanded Medicaid under the ACA were also more likely to adopt the ACA’s HCBS policies. Thus several of the factors associated with Medicaid expansion, as mentioned, were important in explaining HCBS expansion as well: ideology, existing policy, and policy diffusion. Going against expectations, HCBS initiative adoption was also more common in states with less bureaucratic capacity (fewer state employees per capita) and more nursing home beds per elder. It could be that states with less capacity saw the HCBS initiatives as a way to bolster hiring in an understaffed area, and that states with more nursing facility beds had more incentive to increase HCBS options.
Given the importance and structure of the ACA, legal scholars and political scientists have examined the implications of the ACA for the operation of American federalism. For example, Abbe Gluck and Nicole Huberfeld (2018) argue that the ACA is valuable for illustrating how federalism operates contemporarily in the United States. They conclude that the ACA illustrates the “conceptual confusion” inherent in American federalism: is “healthcare federalism”—including the structural elements by which the federal and state governments divide responsibility—meant to produce “particular policy outcomes” regarding cost, access, or quality, or is it meant to “service structural aims regardless of policy ends,” for example “reserving power to states.” Frank Thompson, Michael Gusmano, and Shugo Shinohara (2018) use the ACA as a case of “executive federalism,” showing how some Republican governors abetted Trump administration efforts to undercut the insurance exchanges and Medicaid expansions through waivers, funding decisions, executive orders, and administrative roles, and how a few resisted ACA retrenchment. Some researchers express concern that enhanced state control and flexibility under the Trump administration is used less for innovation than for retrenchment and “intergovernmental blame shifting” (Jones 2017).
Many studies examine state choices to engage in various components of the ACA, but scholars are just beginning to examine the next phase in the ACA’s political effects: feedback effects arising from earlier state choices. In this issue, Richard Fording and Dana Patton (2020) examine one feedback that has emerged in the negative direction: how the expansion of Medicaid to new populations incentivized the adoption of work requirements in some states. Such requirements emerged first in expansion states led by Republican governors, who sought to assuage Republican voters’ objections to expansion by imposing additional terms and conditionality on Medicaid eligibility. The policy then diffused to newly expanding states, which incorporated work requirements from the outset of expansion, and even to non-expansion states, which imposed them in their existing Medicaid programs, rendering those programs even more restrictive than they had been before the ACA. In this way, the ACA provides an important example not just of positive policy feedbacks (policy entrenchment in many liberal states), but also of negative feedbacks of various forms in more conservative states (policy modification, policy reinvention, and policy regression).
Unsurprisingly, dozens of studies estimate the impact of the Affordable Care Act on health insurance coverage, the vast majority focusing on the Medicaid expansion due to its importance in targeting the uninsured (just over half of the uninsured had incomes less than the new Medicaid eligibility income cutoff in 2011) and the opportunity for causal inference afforded by the Supreme Court’s decision to make the Medicaid expansion optional to the states. Research has also been substantial on the extension of parental coverage to young adult dependents that began in 2010, researchers comparing young adults eligible for the dependent coverage with those who were somewhat older and therefore ineligible. The most empirically convincing estimates indicate that insurance rates for young adults increased by about 3.5 percentage points (Slusky 2017). Given the general increase in insurance coverage following the implementation of the main components of the ACA, the literature has focused on determining the contribution of the ACA to the rise in coverage and estimating the contribution of the Medicaid expansion and the ACA’s other elements.
All studies focusing solely on the Medicaid expansion and examining insurance coverage find that Medicaid expansion resulted in sizable and statistically significant reductions in uninsured rates (for a review and comprehensive list, see Antonisse et al. 2018), virtually all of these studies relying on comparing outcomes between states that did and did not take the Medicaid expansion. Two take the analysis further, examining the Medicaid expansion in the broader context of the ACA as a whole. Charles Courtemanche and colleagues (2017) distinguish between substate areas with high and low rates of insurance prior to the ACA, noting that the ACA’s provisions will have more impact in areas where more individuals lack insurance coverage. Using this third dimension of variation in impact, along with the variation over time and across states, they find that in areas with average levels of uninsurance prior to the ACA, the uninsured rate fell by 5.9 percentage points in states with a Medicaid expansion and by 2.8 percentage points in states that did not expand, suggesting that the Medicaid expansion explains just over half of the overall fall in uninsurance on average, the contribution for certain subsets of the population being considerably larger. Molly Frean, Jonathan Gruber, and Benjamin Sommers (2017) estimate the premium subsidy for which a family would be eligible (which varies depending on the area of residence and family structure) and the Medicaid eligibility of the family under both pre-ACA and post-ACA Medicaid rules. Decomposing the change in coverage, they find that approximately 60 percent of the decline in uninsurance explained by their model can be attributed to expansion in Medicaid eligibility and 40 percent to the premium subsidies. A study that focuses on the impact of the premium subsidy policies (Hinde 2017) finds a statistically significant 5.4 percentage point increase in private non–group insurance coverage for individuals with incomes just above the 138 percent of the federal poverty limit cutoff in Medicaid expansion states and a statistically insignificant 2.3 percentage point increase in private non–group coverage for individuals with incomes just above 100 percent of FPL in non-expansion states.
Given the increase in health insurance coverage, demand for health care is likely to rise as the price falls. However, to the extent that supply of health care may respond more slowly, health-care use may not increase as quickly. Moreover, the causal linkages between health insurance, health-care use, and health are not obvious: individuals with health insurance tend to use more care and are healthier, but disentangling the causal effect of the insurance itself from other characteristics of insured individuals is difficult. Consequently, research examining the effects of the expansions of health insurance availability that occurred with the ACA on health care and health has been considerable. Researchers have examined a variety of health-care access and use measures, some of which may plausibly be affected quickly by an increase in insurance coverage; others may take more time for any effect to be seen. One of the goals of the ACA was to improve the appropriateness of care used, thus researchers have been particularly interested in examining whether appropriate preventive care such as blood pressure screening increased and use of the emergency department (where the uninsured are more likely to go to obtain care) fell.
Concomitant with the expansion in insurance coverage, measures such as whether someone needed care but could not afford it have been found to be lower in many studies (for a survey of Medicaid expansion studies, see Antonisse et al. 2018; for a survey of studies of the dependent coverage expansion, see Breslau et al. 2018). Other use measures have a less obvious relationship with insurance coverage given the possibility of supply side constraints, but studies examining use of primary and preventive care have generally found evidence of increases. For example, using the common state difference-in-differences approach, Laura Wherry and Sarah Miller (2016), Miller and Wherry (2017), Kosali Simon, Aparna Soni, and John Cawley (2017), and Ausmita Ghosh, Simon, and Benjamin Sommers (2019) find that at least some measures of preventive care use increased in a statistically significant way due to the Medicaid expansions. Courtemanche and colleagues (2018) find that both the Medicaid expansions and the impact of the ACA in previously high uninsured areas increased use of preventive care. Thomas Selden, Brady Lipton, and Sandra Decker (2017) strike a cautionary note, however, finding that for adults with incomes between 100 and 138 percent of the federal poverty level increases were similar in having a usual source of care and primary care visits, but adults in expansion states reported facing greater difficulty accessing physician care than those in non-expansion states, although those in expansion states saw larger reductions in out-of-pocket spending.
Some evidence indicates that use of acute care increased as a result of the Medicaid expansion, although estimates of such effects are more variable. For example, Wherry and Miller (2016) find that Medicaid expansions were associated with increased overnight hospital stays but no change in emergency department use in the first year of the expansion; results are somewhat different when they add additional data (Miller and Wherry 2017). Estimates of impacts on emergency department use vary, some researchers finding no change in overall emergency department use (Pines et al. 2016) and others finding an increase (Nikpay et al. 2016). Research on the dependent coverage mandate suggests that it led to a slight decrease in emergency department visits (see Akosa Antwi et al. 2015). Overall, hospital admissions seem to have remained largely unchanged, though the payer mix shifted toward Medicaid and away from uninsured admissions (Pickens et al. 2018). Focusing on substance abuse-related admissions, Angélica Meinhofer and Allison Witman (2018) find that opioid admissions to specialty treatment facilities increased in expansion states, especially those with comprehensive medication-assisted treatment coverage under Medicaid. Joanna Maclean and Brendan Saloner (2019) find some evidence of increases in prescriptions and specialty admissions for substance use disorder, but stronger evidence of a shift in payer away from uncompensated care and state and local government payments and toward Medicaid and private insurance.
Although having insurance may lead to moral hazard effects, research thus far has not found the ACA Medicaid expansion to have increased risky behavior such as smoking (Simon, Soni, and Cawley 2017; Courtemanche et al. 2018; Cotti, Nesson, and Tefft 2019). Instead, Chad Cotti, Erik Nesson, and Nathan Tefft (2019) find that Medicaid expansions were associated with reduced cigarette consumption and increased smoking cessation product use among the Medicaid-eligible population. However, Silvia Barbaresco, Courtemanche, and Yanling Qi (2015) find evidence of an increase in risky drinking following the dependent coverage provision.
Although insurance coverage and many measures of access to care have clearly improved under the insurance provisions of the ACA, the health impacts are as yet not clear. Researchers have found results ranging from an improvement in self-assessed health due to the Medicaid expansion (Simon, Soni, and Cawley 2017), to no effect on self-assessed health (Courtemanche et al. 2018), to reduced reporting of being in excellent or very good health in the first year after the expansion (Miller and Wherry 2017). Because changes in self-assessed health are subjective, they may reflect reductions in stress from greater financial security due to insurance or new information learned from new contacts with health-care professionals as well as changes in physical health, so it is perhaps not surprising that the results for self-assessed health vary. In addition, health impacts of insurance coverage may arise over the longer term. For example, Wherry and Miller (2016) find increased rates of diagnosing chronic conditions, and Benjamin Sommers and colleagues (2017) find evidence of increased treatment for chronic conditions, both of which might be expected to have longer-term impacts on health.
Although improving the health of the population is an underlying goal of the Affordable Care Act’s provisions to move toward universal health insurance coverage, another goal of insurance coverage is to protect against the financial consequences of poor health. Health insurance is in many ways unique among types of insurance and, from the perspective of protecting against financial consequences, society’s interest in ensuring access to health insurance may be greater than for other insurance types, because though the risks of a bad health shock can be reduced by behavioral changes, they cannot be eliminated. Moreover, publicly subsidized health insurance plays a key role in the safety net supporting low-income Americans. As a result, household financial security is an important area where the ACA may have had economic impacts, and research in this area is considerable.
Much of the research examining changes in access to care has also examined whether provisions of the ACA affected reported difficulty in paying medical bills, inability to afford care, or magnitude of out-of-pocket payments. Researchers have found evidence for reductions in such measures with Medicaid expansion (see, for example, Miller and Wherry 2017; for a review, see Antonisse et al. 2018). Researchers have also found a reduction in such measures correlated with the ACA insurance expansion more broadly (see, for example, McKenna et al. 2018). Most of the extant studies examining the effects of the ACA on financial security have examined the Medicaid expansion; other ACA impacts, such as changes in rating rules, cost-sharing subsidies, and an increase in patient cost-sharing, may have had an effect as well, but are less studied.
More generally, research has shown that the ACA, and in particular the Medicaid expansion, has improved the financial circumstances of low-income families. Exploiting the fact that some California counties expanded Medicaid earlier than 2014, Heidi Allen and colleagues (2017) examine the use of payday loans—short-term, unsecured loans characterized by high annual interest rates and more commonly used by low-income families. The authors find an 11 percent reduction in the number of loans taken out each month in the early expanding counties relative to others. They also find reductions in the expansion counties in the number of unique borrowers each month and the amount of payday loan debt. Kyle Caswell and Timothy Waidmann (2019) use credit bureau data to compare individuals in expansion and non-expansion counties and across counties with more previously uninsured individuals relative to fewer, and find that the expansion improved consumer financial health on a number of dimensions, including credit scores, balances past due as a percent of total debt, probability of new medical collections, and probability of experiencing a new derogatory balance of any type. Luojia Hu and colleagues (2018) use a panel of consumer credit data and a synthetic control approach to deal with the issue of inconsistent pre-trends across expansion and non-expansion states. They also find evidence of improved financial well-being: the Medicaid expansion reduced the number of unpaid bills and the amount of debt sent to collection among individuals living in zip codes with a high share of previously uninsured low-income individuals. Kenneth Brevoort, Daniel Grodzicki, and Martin Hackmann (2017) examine medical debt more specifically as well as measure the indirect benefits to households of improved credit profiles. They find that the Medicaid expansion reduced the incidence of new medical debt, reduced the probability of becoming newly delinquent on a debt, and improved credit scores. These improvements, they show, translate into better credit outcomes, using novel data on credit offers to show that after the expansion individuals in adopting states received more offers of credit and at substantially better terms than individuals in non-adopting states. Their results indicate that the effects are larger for individuals with subprime credit scores. Because low-income individuals are more likely to have subprime credit scores, these results point to an improvement in financial security among low-income families as a result of the Medicaid expansion. Similarly, Dahlia Remler, Sanders Korenman, and Rosemary Hyson (2017) show the importance of Medicaid and insurance subsidies in reducing health-inclusive poverty, which they define as the poverty rate when accounting for health needs. They find that the ACA’s insurance provisions had a particularly strong impact on health-inclusive poverty among groups such as two-parent families and nondisabled childless adults.
Given the long-standing connection between employment and health insurance, a law such as the ACA may have consequences for the labor market, different provisions having possibly different and even contrasting effects, making overall impacts difficult to disentangle. In addition, the context of an economic recovery and improving labor market adds an additional layer of complexity, limiting the use of variation over time and making provisions that applied equally more difficult to study. The provisions that are most likely to result in labor-market effects, and thus which have received the most attention from researchers, include the dependent coverage mandate, the Medicaid expansion and exchange subsidies, and the employer mandate.
By offering insurance to young adults through their parents, the dependent coverage mandate would be predicted to reduce the incentive to work in jobs offering health insurance and to increase the incentive to work in jobs that do not offer health insurance or in self-employment. In addition, the income effect arising from newly available health insurance at low or no additional cost may reduce labor supply; it may also have differential effects on school enrollment, increasing the probability of enrollment in school for dependents who might have been reluctant to enroll in school if it meant not working or taking a part-time job without health insurance, or reducing the probability of enrollment in school for the group of dependents who previously would have been covered only while a student. Finally, wages among the newly eligible may rise if they take jobs that offer higher wages and fewer benefits such as health insurance, or if employers with large numbers of workers newly eligible for dependent coverage shift their wage-benefit packages accordingly. As Bradley Heim, Ithai Lurie, and Kosali Simon (2018b) note, however, young adults may be less responsive to incentives arising from health insurance perhaps because of general good health or myopia, indicating that any effects of the dependent coverage mandate may be small.
Indeed, the results from the literature studying the dependent insurance mandate indicate small effects or no effects. Using similar approaches to those discussed earlier, Yaa Akosa Antwi, Asako Moriya, and Simon (2013) find some evidence of reduced work hours. David Slusky (2017), however, points out that these estimates are likely an overestimate due to differential trends for the treatment and control age groups used; he finds no evidence of changes in labor supply. Following Slusky’s suggested refinements but using new data, Gregory Colman and Dhaval Dave (2018) find some evidence that newly eligible dependents spent less time working and more time searching for work. However, James Bailey and Anna Chorniy (2016) find little evidence of increased job mobility among young adults. Finally, Heim, Lurie, and Simon (2018b) use tax data to examine employment, self-employment, wages, and enrollment in education. They find effects that are in the theoretically predicted directions but quite small and only for a subsample of young adults whose parents have an employment-based retirement plan (a proxy for having employer-sponsored insurance). Overall, it appears that the dependent insurance mandate has had no substantial effect on labor-market behavior.
The Medicaid expansion and subsidies for insurance purchased through the exchanges have a variety of theoretical impacts on the supply side of the labor market. Most straightforward is the incentive for individuals to reduce labor supply, either to get below the subsidy eligibility level (or to qualify for a more generous subsidy) or because of an income effect of the additional resources provided by Medicaid or insurance subsidies (Congressional Budget Office 2014). This incentive implies reductions in employment and hours and thus increases in (voluntary) part-time work. However, individuals such as parents who previously were eligible for Medicaid if their income was very low have an incentive to increase their labor supply because they can now earn more and still have insurance. In addition, individuals in states that do not accept the Medicaid expansion also have an incentive to increase their labor supply to qualify for exchange subsidies that only apply to individuals earning more than the poverty line. In addition, the availability of health insurance through a source not tied to employment makes it easier for workers to change jobs or become self-employed.
Although theoretical predictions indicate a variety of possible labor supply effects, empirical work to date has found little evidence of statistically or economically significant labor supply effects of the Medicaid expansion. Using standard state difference-in-differences methods, multiple studies find no evidence of changes in labor-force participation, employment, usual hours worked, propensity for full-time versus part-time work, or wages as a result of the Medicaid expansion (Gooptu et al. 2016; Kaestner et al. 2017; Frisvold and Jung 2018; Leung and Mas 2018). This lack of a result is also found when using somewhat higher-income individuals as a control group (Gooptu et al. 2016), when examining outcomes for individuals observed for two years (Gooptu et al. 2016; Leung and Mas 2018), when focusing just on childless adults in states with no previous coverage for such adults (Leung and Mas 2018), and when using a synthetic control method to better account for the possibility of differential trends across expansion and non-expansion states (Kaestner et al. 2017). Similarly, researchers have found no change in the probability of retirement or part-time work among workers ages fifty through sixty-four beginning in 2014 in Medicaid expansion states relative to non-expansion states (Levy, Buchmueller, and Nikpay 2018). In addition, the Medicaid expansion does not appear to have affected exits of workers from unemployment, suggesting no detectable impact on job search behavior among the unemployed (Buchmueller, Levy, and Valetta 2019).
The only indication of labor supply effects comes from two working papers that focus on substate geographies. Mark Duggan, Gopi Goda, and Emilie Jackson (2019) find no aggregate effect on labor-force participation but an increase in participation and employment in Public Use Microdata Areas (PUMAs) with a higher pre-ACA uninsured rate among the Medicaid-eligible population and a reduction in labor-force participation in areas with a higher pre-ACA uninsured rate among the subsidy-eligible population. However, they find no effect on part-time employment, self-employment, or hours worked conditional on employment. Lizhong Peng, Xiaohui Guo, and Chad Meyerhoefer (2018) find evidence of a transitory decline in employment in border counties in expansion states relative to neighboring counties in non-expansion states, although they find no impact on wages. However, the data they are using correspond to the location of the job rather than the location of the potentially eligible individual, so it is not clear to what extent they are measuring a labor supply effect. In another study focusing on substate geographies, Lucie Schmidt, Lara Shore-Sheppard, and Tara Watson (2019b) find no evidence of labor supply effects for nonparents or married parents for any outcome (labor-force participation, employment, hours, or earnings) and at most a small increase in labor supply among single parents, comparing individuals in expansion PUMAs with those in PUMAs in bordering non-expansion states. The vast majority of the evidence thus suggests no economically or statistically significant effect of the Medicaid expansion on labor supply.
Finally, although the Medicaid expansion and the availability of subsidized insurance through the exchanges raises the possibility of increased job flexibility, research to date has not found evidence of increases in job changing or self-employment. Kavan Kucko, Kevin Rinz, and Benjamin Solow (2018) examine the universe of individual tax returns and find an increase in reported income just above the poverty level among taxpayers with self-employment income in states that did not take the Medicaid expansion. However, by matching the tax returns to survey data, they show that there were no differences in actual labor market outcomes, indicating an increase in reported income in response to the tax incentive but no real change in labor-market behavior.
The employer mandate, which requires firms with more than fifty workers to offer affordable coverage or pay a penalty, could have several possible impacts on labor demand, including incentives for firms to reduce their size below the cutoff if they are close to that level, to use fewer full-time workers but increase their hours, to increase use of temporary or contract workers, and to increase use of part-time (less than thirty hours per week) rather than full-time full workers. Firms forced to begin offering coverage have an incentive to reduce wages to compensate, although the minimum wage limits changes on this margin. However, 95.7 percent of firms with fifty or more employees offered health insurance in 2013 according to data from the Agency for Healthcare Research and Quality, indicating that any impacts of the employer mandate on labor demand are likely to be small.
The evidence to date on the employer mandate, much of which is primarily descriptive rather than demonstrating causal effects, largely bears this prediction out. Moriya, Selden, and Simon (2016) compare trends adjusted for economic conditions for different firm sizes and find little evidence of differential change in part-time employment by firm size. Using similar methods, Bowen Garrett, Robert Kaestner, and Anuj Gangopadhyaya (2017) find no difference between trends in actual labor-market outcomes and those predicted based on economic conditions and demographics for employment or usual hours per week, but do find an increase in voluntary part-time employment, particularly among women, and a decline in involuntary part-time employment. To obtain estimates that are more plausibly causal, William Even and David Macpherson (2019) try to incorporate differences between occupations more and less likely to be affected by the employer mandate. They find that among less-educated workers, involuntary part-time employment fell more slowly after 2014 in occupations with higher shares of workers likely to be affected by the employer mandate (working in firms with more than one hundred workers, not offered health insurance, and working thirty or more hours per week), which suggests that the employer mandate contributed to higher levels of involuntary part-time work, although it is not definitive given that trends across occupations may have differed in the absence of the ACA. They find no evidence of a change in voluntary part-time work.
Finally, the requirement that insurance policies must cover preexisting conditions and cannot charge higher premiums for them has important theoretical implications for job mobility because it allows an individual to change jobs or to take a job not offering health insurance even if a family member has such a condition. Little research thus far has focused on job mobility changes as a result of the preexisting condition limitation, but Pinka Chatterji, Peter Brandon, and Sara Markowitz (2016) find that the elimination of preexisting condition exclusions led to an increase in voluntary job changes among parents of a child with a chronic condition.
Overall, the evidence to date on labor-market impacts of the ACA suggests that they have been limited, even in areas such as labor supply and job flexibility where theoretical arguments suggest effects. Further work, particularly that which investigates possible sources of heterogeneity in outcomes across individuals and places, is needed before full conclusions can be drawn, however. In addition, although short-term effects of the ACA on labor-market outcomes may be smaller than long-term ones, as the time since the ACA’s passage lengthens, researchers will face additional challenges in attempting to disentangle ACA effects from the effects of other changes in the economy.
The extensiveness of the ACA and the importance of health insurance to individual and family well-being suggests that in addition to the direct impacts on coverage, health, and the labor market discussed earlier, its passage may have affected more indirect social outcomes, including marital and fertility decisions. Prior to the ACA, marriage was an important way for individuals to gain access to insurance if their employer did not sponsor it or they did not have a job but their prospective spouse’s employer did. The incentive offered by the possibility of insurance coverage thus may have induced couples in a relationship to marry or to marry earlier than they would have otherwise. Various provisions of the ACA changed this incentive, however. In particular, the young adult dependent coverage mandate provides young adults an additional source for health insurance coverage outside of marriage, reducing the incentive to marry. The provisions of the premium tax subsidy, like other tax subsidy programs, may have more complex effects on marriage incentives, penalizing marriage in some cases and rewarding it in others, depending on the income and employment circumstances of the potential partners. The requirement that preexisting conditions be covered also has implications for marriage (and divorce) since an individual no longer faces constraints on moving between insurance plans. In addition, as Joelle Abramowitz (2016) points out, indirect impacts on marriage may operate through other channels (for example, if young adults increased their school enrollment in response to the new coverage, their marriage propensity could be affected indirectly). Finally, the Medicaid expansion included a provision eliminating asset tests. In that case, when a spouse is diagnosed with an expensive medical condition couples do not need to divorce to preserve assets for the healthy spouse while obtaining Medicaid coverage for the unhealthy spouse (“medical divorce”).
In addition to marriage, the ACA’s provisions may have impacts on other aspects of family structure, such as fertility. These impacts are theoretically ambiguous given that having health insurance reduces the cost of any medical care, including the cost of childbearing, which would be predicted to increase fertility, but having insurance also lowers the price of contraception, thus reducing fertility. In addition, the ACA included a provision requiring insurers to cover contraception, which would be expected to reduce fertility further.
Despite the variety of possible impacts on family structure, relatively little research on family structure outcomes has been undertaken. In the area of marriage and divorce, the research thus far suggests that marital status decisions are affected by the policy. Abramowitz (2016) examines the impact of the dependent coverage mandate using a typical age-based difference-in-difference strategy, finding reductions in the likelihood of marriage and increases in the probability of divorce following the implementation of the mandate. Matthew Hampton and Otto Lenhart (2019) use longitudinal data to estimate the impact of coverage of preexisting conditions, finding that the probability of being married declines for men with preexisting conditions after 2014 relative to men without such conditions, a finding that is robust to a variety of specification checks, including a placebo test using alternate time periods not including a policy change. Finally, Slusky and Donna Ginther (2017) use state difference-in-differences to examine the impact of the Medicaid expansion on “medical divorce” and find that the Medicaid expansion decreased the prevalence of divorce among those ages fifty through sixty-four with a college degree.
As is true of research on marital status, research on fertility effects of the ACA has focused on the impact of the young adult dependent coverage mandate. Both Abramowitz (2018) and Heim, Lurie, and Simon (2018a) take advantage of the age variation in the mandate, Abramowitz using data from the American Community Survey and the National Survey of Family Growth and Heim, Lurie, and Simon using tax records. Both studies find that the dependent coverage mandate modestly reduced childbearing, Abramowitz also showing evidence of a reduction in abortion rates and an increase in the use of long-term contraceptives. Because such little research has been published to date on fertility impacts of the ACA, this would seem to be an important area for future research, particularly given that previous work on Medicaid expansions and fertility has found equivocal effects of income-based expansions on birth rates but more consistent impacts of expansions to contraceptive access (for a review, see Buchmueller, Ham, and Shore-Sheppard 2016).
Because rates of uninsurance were higher in vulnerable populations before the ACA, many but not all such populations were intended to be helped by the policy. Evidence from Medicaid expansions in particular but also the ACA overall indicates that when subpopulations of low-income, low-education, or racial or ethnic minorities are studied, coverage gains are substantial (for a review, see Antonisse et al. 2018, 3). In addition, health outcomes tend to have improved for those groups (see, for example, Sommers et al. 2015; Antonisse et al. 2018). One key group excluded from the intended effects of the policy, however, was immigrants who were either undocumented or had been in the United States fewer than five years. Researchers have pointed out the likely importance of additional funding for community health centers that was included in the ACA in providing care for this group (see Ortega, Rodriguez, and Vargas Bustamante 2015). However, relatively little work evaluating the impact of this part of the ACA has been done.
Similarly, the ACA, and particularly the Medicaid expansion, represents a potentially important new source of health-care funding for a group that has historically had low rates of insurance—the criminal justice–involved population (Boutwell and Freedman 2014). This population faces health challenges including rates of infectious disease, chronic illness, and trauma that are higher than in the general population (Rich, Wakeman, and Dickman 2011). A lack of health insurance among the formerly incarcerated therefore suggests that many of these issues have gone unaddressed. However, little is known about the impact of the ACA on this population, an important omission that Carrie Fry, Thomas McGuire, and Richard Frank (this issue, 2020) address in their article. Comparing arrest data from six urban county jails, they show that Medicaid expansion is associated with small decreases in rates of recidivism in two of the three county pairs examined. The declines are of similar magnitude across gender and racial-ethnic subgroups.
Another way the ACA may affect vulnerable populations is its interaction with other parts of the safety net. This area has received more attention, particularly in regard to programs for the disabled—Supplemental Security Income (SSI), a means-tested program requiring that family income and resources be below a cutoff in addition to the individual being determined to have a disability, and Social Security Disability Insurance (SSDI), a social insurance program for disabled individuals with significant work history that pays benefits based on an individual’s past earnings. Both SSI and SSDI provide beneficiaries with health insurance as well as cash benefits, SSI recipients being eligible for Medicaid and SSDI recipients receiving Medicare. The health insurance benefit is likely to be particularly important for the disabled, who are likely to have high levels of health-care needs and low levels of access to employment-based health insurance. Jae Kennedy and Elizabeth Blodgett (2012) note that several provisions of the ACA have the potential to affect disability program participation both by allowing disabled workers to remain privately insured (the elimination of preexisting condition exclusions, the elimination of lifetime caps on insurance payments, and the parental coverage mandate) and by granting public coverage through Medicaid even without a formal disability assessment. The possibility of health insurance not tied to disability program participation thus provides an incentive to reduce disability program participation. However, incentives are also in place to increase disability program participation. Because the disability determination process is time consuming, and SSDI recipients face an additional waiting period before they can receive Medicare, disabled workers may be reluctant to leave their jobs and health insurance to claim disability benefits because it could mean a long period without health coverage. The Medicaid expansion of the ACA might therefore encourage disability program participation among such individuals. In addition, the Medicaid expansion might encourage disability program participation if potentially eligible individuals become aware of their disability program eligibility in the process of applying for Medicaid.
Several groups of researchers have examined the question of the net effect of the Medicaid expansion on disability program participation and applications. Chatterji and Yue Li (2017) examine SSI participation in three states and Washington, D.C., that expanded Medicaid before 2014 under an optional provision of the ACA or a federal waiver. Notably, all of the expansions they study were built on previous state-run programs that had limits on benefits or the number of enrollees rather than being entirely new opportunities. Using synthetic control methods, they find a marginally statistically significant reduction in SSI receipt in only one state, Connecticut. Aparna Soni and colleagues (2017) also find a reduction in SSI recipients when they examine expansions that occurred in 2014 and 2015 using a simple difference-in-differences approach, although no effect of the expansion is evident when SSI participation is measured as a fraction of the population (Schmidt, Shore-Sheppard, and Watson 2019a).
Because exit from disability programs is relatively low, disability program participation is affected by previous policies as well as new policies, suggesting that the stock of program participants may change more slowly than changes in policies would suggest. Moreover, the lag between application and enrollment may be substantial, raising the question of when the level of disability program participation might realistically be expected to reflect changes in public insurance policy. By contrast, applications to disability programs are likely to reflect policy changes more immediately. Two studies have examined research on SSI and SSDI applications using administrative data from the Social Security Administration. Priyanka Anand and colleagues (2019) run a difference-in-difference model using only PUMAs that match well on preexpansion characteristics with at least one other PUMA, finding that SSI applications were slightly higher in PUMAs in states that expanded in the first quarter of 2014 than in non-expansion PUMAs between one and five quarters after the expansion. However, as Schmidt, Shore-Sheppard, and Watson (2019a) note, because Anand and colleagues pool all expansion and non-expansion PUMAs in their matched sample rather than comparing specific matched PUMAs, there is some evidence of dissimilar preexpansion trends in the two groups, raising the possibility that their results partially reflect differential trends across groups. Their estimates for SSDI applications are similarly inconclusive.
Schmidt, Shore-Sheppard, and Watson (2019a) use annual applications for SSI and SSDI, but observed at the county level. To control for the likelihood of differential trends in disability program applications across expansion and non-expansion states, they compare application rates in state border counties that expanded Medicaid with those in counties in non-expansion states just across the border, showing that border counties are more similar to the county just across the border than to those with differential expansion status located elsewhere. They find no significant effects of the Medicaid expansion on applications or awards to either SSI or SSDI. Overall, the research in this area indicates that any impact of the new availability of public insurance on disability program applications or caseloads is negligible.
One concern about a research design examining Medicaid expansion that compares geographically proximate areas across state lines would be the possibility that individuals who would be eligible for Medicaid if they lived in an expansion state would move across state lines in order to obtain public insurance. Lucas Goodman (2017) investigates migration in response to the ACA Medicaid expansion and finds no evidence that migration out of non-expansion states to expansion states increased relative to migration in the reverse direction. His results are precise enough that he can rule out effects on migration rates that would produce Medicaid enrollment changes detectable in the data.
Finally, using a similar comparison of outcomes in expansion to non-expansion border counties, Schmidt, Shore-Sheppard, and Watson (2019b) examine participation in two additional safety net programs, the Supplemental Nutrition Assistance Program (SNAP) and the Earned Income Tax Credit (EITC). As means-tested programs, EITC and SNAP may be affected by the Medicaid expansion if individuals change their labor supply in response—either reducing it to qualify for Medicaid, which might increase participation in other programs, or increasing it because for individuals such as parents the Medicaid expansion offers more generous means testing than previous Medicaid eligibility limits. Even if labor supply does not change, eligibility for Medicaid or the process of enrolling in it may provide information about eligibility for other means-tested programs or make the process of enrolling in other programs relatively easier. Schmidt, Shore-Sheppard, and Watson (2019b) use county-level administrative data and PUMA-level data from the American Community Survey and find evidence of small increases in SNAP and EITC receipt. They find no evidence of labor supply changes, however, suggesting that information about program eligibility or enrollment is the primary mechanism behind the increases.
The largest expansion of social policy in the United States in a generation, the Affordable Care Act could potentially have profound effects on the political behavior and attitudes of ordinary Americans—those who may benefit from the law’s provisions, those who pay for the new benefits, those who embrace the law’s expansion of health insurance access, and those who resent or oppose it. As noted, several scholars have examined whether state decisions on the ACA were associated with public opinion, with mixed results. A much larger literature has explored whether the ACA and its implementation has had feedback effects shaping subsequent attitudes and behaviors. That is, scholars have examined political activity and preferences among the public not as an input into policymakers’ decisions but rather as an outcome of the law. Indeed, although much of the existing policy feedbacks literature examines policy initiatives of the past, such as Social Security, the GI Bill, and Aid to Families with Dependent Children (for an overview, see Campbell 2012), the ACA enables both the study of feedback effects as they emerge in real time and enhanced causal inference, due to the quasi-experimental roll-out of various provisions. That the ACA was the subject of a well-publicized and enduring political debate during its creation, enactment, and implementation would seem to heighten the possibilities for feedback effects as well (Sances and Clinton 2019b).
Thus far, much scholarship has used the optional Medicaid expansion to estimate causal models of political behavior. Most analyses have found positive effects: increased voter turnout in states that expanded Medicaid. For example, Haselswerdt (2017) finds that aggregate voter turnout in House races declined less in 2014 (a midterm year) than in 2012 (a presidential election year) in states that had expanded Medicaid. An accompanying analysis using individual-level survey data shows that the positive turnout effect is evident among both Democrats and Republicans, the latter suggesting a backlash effect, given surveys indicating that most Republicans opposed the ACA. Joshua Clinton and Michael Sances (2018) compare counties sharing a border between expansion and non-expansion states, focusing on citizens between eighteen and sixty-four and below 138 percent of the federal poverty limit. Examining midterm and presidential elections before and after Medicaid expansion, they find that both voter registration and turnout increased in expansion counties, particularly those with a high share of citizens who are newly eligible for Medicaid. Unlike Haselswerdt, they find the participation effect concentrated in Democratic-leaning rather than Republican-leaning counties, evidence of a positive effect among recipients but not an anti-ACA backlash effect.
In this issue, Charles Courtemanche, James Marton, and Aaron Yelowitz (2020) estimate the impact of the ACA on voter registration and turnout, focusing not just on the Medicaid expansion but also on provisions intended to increase insurance coverage overall. Using the Current Population Survey’s November Supplement between 2006 and 2016 and capitalizing on variation across time, state Medicaid expansion status, and within-state pre-ACA uninsurance rates, they find that the ACA had small and statistically insignificant effects on registration and turnout. These results are in contrast to the positive Medicaid effects reported in analyses of aggregate data at the county level (Clinton and Sances 2018) and the congressional district level (Haselswerdt 2017), but are consistent with estimates of Medicaid effects on turnout outside of the ACA (Michener 2017).
In contrast to some of the findings for Medicaid expansion, the dependent care provision, by which individuals under age twenty-six can stay on their parents’ health insurance, does not have an effect on the political participation of these youth (Chattopadhyay 2017). The lack of an effect could be due to a variety of reasons: the dependent care benefit may not be visible as government activity because it depends on parents’ having private health insurance; it may be difficult to identify others benefiting from the ACA in this way, complicating potential mobilization by advocacy groups; the benefit targets youth, who are a low-participation group to begin with, and confers a short-term rather than lifetime benefit. The failure to find a positive participatory effect of the dependent care provision supports the view of other scholars who argue that the ability of policies to produce positive feedbacks may be contingent and fragile, particularly policies with designs as complicated, contested, and submerged as those of the ACA (Patashnik and Zelizer 2013; Galvin and Thurston 2017; Jacobs and Weaver 2015). We also do not know whether the dependent care provision changed the political behavior of parents, a possible question for future research.
Scholars have also begun to assess the mechanisms that may link ACA benefits to increased political participation, though much of this work remains speculative. A leading possibility is that gaining health insurance has a positive effect on individuals’ politically relevant resources. As discussed, the Medicaid expansion improved the financial stability of low-income families. Such stability may enhance recipients’ ability to engage in the “luxury good” of political participation (Rosenstone and Hansen 1993). Another resource effect of ACA-provided health insurance coverage could be improved mental or physical health, as research outside the ACA suggests. Better physical health is associated with greater political participation (Burden et al. 2017; Pacheco and Fletcher 2015; Gollust and Rahn 2015). Poor health could have an attention-interest effect—drawing focus from political matters to personal ones—or have a cognitive effect that inhibits political participation (Pacheco and Fletcher 2015; Blais 2000). As noted, effects of the ACA on health may not yet have emerged, although diagnosis and treatment of chronic conditions have risen under the ACA (Wherry and Miller 2016; Sommers et al. 2017).
A second mechanism linking health insurance and political participation is political engagement, including political interest, knowledge, and efficacy (Verba, Schlozman, and Brady 1995). Being newly insured because of the ACA could enhance recipients’ awareness of the stakes of public policy, linking their self-interest to government policy and enhancing their interest in government action, a form of “conscious mobilization” (Clinton and Sances 2018). Gaining insurance could have a positive “interpretive effect” (Pierson 1993), the government conferring a benefit and recognizing the recipient as a worthy citizen. Newly insured citizens might feel gratitude toward the government (De La O 2013) or become civically and politically engaged because of a “reciprocity” effect (Mettler 2005). Endorsement by politicians could dampen stigmatizing effects (Clinton and Sances 2018).
Third, the ACA could enhance political participation through a mobilization effect—either a positive effect on recipients or a participation-enhancing backlash among program opponents. One mobilization effect for recipients may simply have been mechanical: the 1993 National Voter Registration Act requires social assistance agencies, including the health exchanges, to provide voter registration services, which may explain increased turnout in Medicaid expansion states (Clinton and Sances 2018). Many navigator organizations assisting citizens in signing up for health insurance facilitated voter registration as well (Hagan 2016). Another possibility is group mobilization—that advocacy groups are actively organizing ACA recipients as pressure groups to defend the legislation—although we did not identify any scholarly accounts of such activity.
Fourth, policy threat could be a mechanism linking the ACA and political participation. Journalistic accounts describe protests defending the ACA from Republican repeal efforts after the 2016 election resulted in unified Republican control of the presidency and Congress (see, for example, Stein 2017). Scholarly accounts of anti-Trump and anti-Republican grassroots organizing are emerging (Gose and Skocpol 2019; Meyer and Tarrow 2018), protecting the ACA from repeal playing a significant role in this mobilization (see also Nadash et al. 2018). The ACA is an important case for analyzing the effect of policy threat on participation: previous analyses of policy threat and political participation find differing effects for means-tested versus universal programs: threats to cut Social Security and Medicare in the 1980s and 1990s elicited surges of senior citizen letter-writing to Congress (Campbell 2003); in contrast, cuts to Tennessee’s Medicaid program in 2005 resulted in greater turnout declines in counties with the largest disenrollment (Haselswerdt and Michener 2019). Thus more work is needed to assess the effects of policy threat for social policies of various designs and target populations, including the ACA and its many components.
One additional contribution the scholarship on policy feedbacks and the ACA has made is showing how the effects of public policy on behavior may vary by partisanship. Take-up of health insurance under the ACA varies by party identification: Republicans have been shown to oppose the ACA even when they might personally benefit (Kliff 2016); indeed, Amy Lerman, Meredith Sadin, and Samuel Trachtman (2017) find in their recent field experiment that Republicans in need of insurance are more likely to sign up if shown a private interface (healthsherpa.com) rather than the government interface (HealthCare.gov). The effect of party extends to political behavior as well. The ACA appears to elicit a backlash or “thermostatic” effect in which those who are opposed to the reform for ideological reasons or who perceive that they will not benefit or may have to pay for the reform react with increased political participation (Haselswerdt 2017; McCabe 2016).
The advent of the ACA also triggered a great deal of work examining effects on the political attitudes and preferences of recipients and other members of the public. Speculation circulated that support for the ACA, which hovered below 50 percent as the law was being debated, would rise once implementation began and people gained insurance through its provisions (Jacobs and Mettler 2011). However, previous scholarship looking for attitudinal changes after reforms of welfare and Medicare failed to detect them (Soss and Schram 2007; Morgan and Campbell 2011). Further, both political and design reasons to believe that the ACA also would not generate policy feedback effects have merit: the law was debated and implemented in a highly partisan environment, suggesting that partisanship might dominate personal experience as a driver of attitudes (Patashnik and Zelizer 2013). The law’s complicated, often hidden design elements might also undercut possibilities for attitudinal change (Chattopadhyay 2018, 2019).
Early in the ACA’s trajectory—as the legislation was being debated and before anyone actually benefited from its provisions—partisan and racial considerations dominated the public’s attitudes toward the reform. Pooled cross-sectional surveys in 2009 and 2010 showed that party identification was more important in shaping support or opposition toward health reform than were demographic factors such as being older, higher income, or African American (Kriner and Reeves 2014). An analysis of individual-level change from a 2008–2010 panel survey found that during the debate over the law, opinions were more likely to change toward opposition than toward support for reform, and that Republicans were more likely to switch to opposition than Democrats were, the effect being more muted among Republicans concerned about health-care costs, presaging later findings that material stakes affected ACA attitudes (Henderson and Hillygus 2011). Scholars also detected motivated reasoning among partisans in the absence of actual experience as well as an effect of elite rhetoric: as elite rhetoric about the law stabilized, variance in attitudes diminished (Kriner and Reeves 2014). The power of partisanship and symbolic attitudes such as little trust in government remained powerful even after implementation began. For example, panel data from 2010 through 2014 showed that Republicans and those with less trust in government were more likely to say the ACA was increasing their tax burden, regardless of their experience with the law (Jacobs and Mettler 2016, 2018).
Racial attitudes were also found to be highly correlated with support for the ACA (Henderson and Hillygus 2011). That the reform was debated and passed under the first black president made a difference; the relationship between racial attitudes and health reform attitudes was not apparent during the Clinton reform effort of 1993 and 1994, but materialized after it was clear Obama would be the Democratic nominee for the 2008 presidential election (Tesler 2012).
After implementation began, the question became whether actual experience—such as gaining insurance—might shape attitudes beyond party identification, racial attitudes, and elite rhetoric. Although in the past scholars have found that political attitudes often do not correspond to individuals’ apparent material interests, health insurance potentially has the characteristics that can trigger self-interested considerations: gaining insurance is a tangible, large, and visible policy event (Citrin and Green 1991). Evidence of increased support among those benefiting from the legislation has begun to emerge. As the ACA was implemented, fewer survey respondents said the law had no effect on health-care access (Jacobs and Mettler 2016). Both pooled cross-sectional data and panel data from the early years of ACA implementation also show that the gap between Republicans and Democrats in favorability toward the law was smaller among those who gained insurance through an ACA marketplace than those with employer-based insurance (McCabe 2016). These findings echo those of Daniel Hopkins and Kalind Parish (2019), who pool Kaiser Family Foundation surveys from 2010 through 2017 and find that Medicaid expansion made lower income Americans more favorable toward the ACA, the effects being stronger among nonwhite and Democratic respondents (and absent among higher-income respondents, who are more likely to have insurance from other sources, suggesting a self-interest mechanism at work). Similarly, those with personal or family experience with the ACA (such as using subsidies, gaining insurance, or getting prescription drug help as a senior citizen) are more likely to say the law has had a favorable impact on health access (Jacobs and Mettler 2018). Yet more evidence of personal experience affecting attitudes comes from an analysis of those using the exchanges to buy insurance, which found that they were much more positive toward the ACA after implementation commenced than those who remained uninsured (Hobbs and Hopkins 2019; see also Hosek 2016). The same study found that those in their early sixties whose insurance premiums were newly capped by the ACA became more favorable toward the law after implementation (Hobbs and Hopkins 2019; see also Nadash et al. 2018). However, the effect of personal experience varies across individuals in quite specific ways. William Hobbs and Daniel Hopkins (2019) also find that those purchasing insurance on the exchanges who experienced local premium spikes became less favorable toward the ACA.
Those with other forms of government health insurance are more supportive of the ACA as well (Jacobs and Mettler 2018; Lerman and McCabe 2017). Lerman and Katherine McCabe use a causal design—comparing those immediately above and below the Medicare eligibility threshold at age sixty-five—and find that getting Medicare increases both support for Medicare spending and support for the ACA. The effects of personal experience were stronger for Republicans, showing that personal experience and self-interest can offset some of the effects of partisanship.
That said, the effect of partisanship is so strong that Republicans in need of health insurance are less likely than Democrats to sign up under the ACA (Lerman, Sadin, and Trachtenberg 2017; Sances and Clinton 2019a). The gap is larger in non-expansion states; in Medicaid expansion states there is no partisan difference in Medicaid uptake, suggesting that elite cues and partisanship are not just filters but actual bars to addressing a tangible need (Sances and Clinton 2019a).
Another factor that might influence attitudes toward the ACA is news coverage and television advertising, although the existing work examining these relationships is correlational rather than causal. During the first two weeks of open enrollment in October 2013, news coverage of the ACA was more negative in states using federal marketplaces than those using state marketplaces (Gollust et al. 2014). Individuals in locales with a higher volume of insurance advertising and more local news coverage of the ACA in Fall 2013 were more likely to say that they were well informed about the law and more favorable toward it, although the effect was stronger for Democrats than Republicans (Fowler et al. 2017). After Marketplaces opened for business, a greater volume of television ads opposing the ACA was associated with lower Marketplace participation (Gollust et al. 2018). Looking at the effects of different types of health insurance and health-related political advertising on Marketplace participation, Paul Shafer and colleagues (this issue, 2020) use a county fixed-effects model to control for underlying differences across counties. They find that when more health-insurance-related ads sponsored by states are aired, enrollment increases, as it does, somewhat surprisingly, when more pro-Republican political ads that mention health care are aired (although this result is driven by a few locales where off-year gubernatorial contests or early federal primary elections were held). Other types of ads, including federal ads not dealing with Medicare, private ads, and pro-Democrat health-care-related political ads, do not show statistically significant correlations with enrollment, suggesting that both the volume and source of advertising matters for Marketplace participation, and that private-source advertising lacks the positive effect of state-sponsored advertising.
Yet other scholars examine the effects of elite rhetoric on public opinion. As noted, as elite rhetoric stabilized during the 2009 and 2010 health reform debate, public opinion became less volatile (Kriner and Reeves 2014). A comparison of word choice in senators’ press releases and in open-ended survey questions of the public during the debate reveals that elite framing did not appear to change public opinion but did influence the words and arguments members of the public used in explaining their ACA attitudes, the public adopting both Republican and Democratic rhetoric as debate ensued (Hopkins 2018). Last is the question of whether elite action, as opposed to rhetoric, can influence public opinion. One study of policy diffusion found that pro-ACA gubernatorial announcements in one state increased public support for the ACA in nearby states, a pattern the authors term “policy spillover” (Pacheco and Maltby 2017).
Julianna Pacheco, Haselswerdt, and Jamila Michener (this issue, 2020) provide another example of state policy choice and partisanship influencing public opinion (see also Pacheco and Maltby 2019). The authors estimate support for the ACA at the state level by quarter between 2009 and 2016, and find that differences between Republicans and Democrats in ACA attitudes are greatest in states in which Democratic governors established state-run health insurance exchanges. In the few states where Republican governors did the same, polarization in support by party identification is less, though the evidence is less clear. They conclude that attitudinal polarization is greater where state ACA policy choices are “aligned”—that is, where Democratic governors implemented the law—and more modest in the presence of greater “misalignment”—elite decision making that cuts across partisanship.
Finally, attitudes toward existing policies may become more positive in the face of political threat. Sances and Clinton (2019b) pool a large number of surveys from 2009 through 2017 and find that approval of the ACA is 1.3 percentage points higher in Medicaid expansion states than in non-expansion states, though this effect is apparent only after the 2016 election, when unified Republican control of the federal government made the threat of repeal more credible. The largest increase in approval was among lower-education non-senior adults, the expansion’s target population. They also find that support for repealing the ACA is 2 points lower in Medicaid expansion states than in non-expansion states, beginning immediately after ACA implementation commenced and corresponding to the well-established asymmetry of gains and losses (Kahnemann and Tversky 1979).
The largest expansion of social policy in more than a generation, the Affordable Care Act has garnered a great deal of attention from scholars, as both this review and the articles in this issue attest. However, much work remains to be done in exploring its economic, political, and social effects. Some of this work will become more feasible as more data become available; more will emerge as longer-term effects come to fruition. We also hope to see more work based on qualitative methodologies such as ethnographic fieldwork and interviews added to the results reviewed here, which have largely been based on administrative data, survey data, and field experiments.
As we searched for extant social science research on the ACA, we identified a number of areas where interesting questions appear to be unaddressed thus far. Having noted some apparent lacunae at various points, we pose additional questions here in the hope of inspiring continuing research. Many of these questions arise because of uncertainty in the policy environment or differences between the ACA as written and as implemented, although others would have arisen even had the ACA been more broadly welcomed across the political and social spectrum.
We have noted throughout that much of the research about the ACA’s effects on individuals and families—whether on economic and financial outcomes or on political behaviors or attitudes—has examined the effects of the Medicaid expansion. Although a crucial part of the law and a component lending itself to quasi-experimental analyses, the Medicaid expansion was not the only provision affecting individuals’ financial security. Research is still needed on the effects of other components of the ACA that affect the cost and availability of health insurance, such as cost-sharing subsidies or the rating rules.
Some of the questions arising from the uncertain federal policy environment concern state responses: how are states responding to federal decisions such as approval of work requirements and other previously denied Medicaid waivers, the elimination of cost-sharing reduction payments to insurers, approval of association health plans and short-term insurance plans with narrower benefits, and how are these state decisions affecting individual well-being? Given the repeated unsuccessful attempts to repeal the law entirely and journalistic accounts of rising public support for it, more work is needed on the development of such support. To what extent are constituencies being mobilized to defend the ACA, and by whom? To what extent are new advocacy groups forming or existing groups redeploying resources or adopting new strategies to combat challenges to the law? How does the COVID-19 pandemic alter politics around ACA support or retrenchment? Other questions surrounding the uncertain policy environment concern individual outcomes: what are the effects on enrollment and other individual outcomes of changes such as reduced federal funding for advertising and for navigators or shortened enrollment periods? Last are questions about how the portions of the ACA focused on reducing health-care expenditures have been affected by policy uncertainty; for example, how has the policy environment affected what can be learned from accountable care organizations and other demonstration projects?
Some of the unaddressed questions are largely unrelated to issues of policy uncertainty but arise because of the far-reaching nature of the ACA or the structure of its provisions. One important question that has received less attention from researchers than we were expecting is how the ACA has affected measures of inequality, including income, earnings, or wealth. The ACA’s insurance provisions are implicitly redistributive, but little is known about the distributional consequences of various provisions of the ACA. We also know little about how much freedom of choice low-income people have among insurance plans and health-care providers and what the implications are for their health-care access, financial security, and feelings toward the reform and government, given that low premiums have generally been attached to narrow networks of providers. In addition, as time goes on, it will become possible to assess the downstream effects of change in one health insurance arena on another. For example, if private insurance premiums fall where Medicaid expands, will there be increased budgetary pressure on Medicaid given that the shift implies adverse risk selection into Medicaid? Another area with surprisingly little research is the effect of the ACA on employment, safety net participation, and other outcomes for various vulnerable populations, including immigrants and families with mixed immigration status, early retirees, and the unemployed. Other questions are largely unexamined: What are the economic, social, and political effects of the various public health interventions contained in the ACA, a question made particularly pertinent by the COVID-19 pandemic? To what extent and how has the implementation of the ACA affected nongovernmental providers of social services?
Although we describe the results from some research on employer behavior, many questions about how firms have responded to the ACA remain unaddressed: To what extent have firms and nonprofits altered their business models, patterns of lobbying, and enterprise investment decisions in light of the ACA? How have small employers been affected by the changes in the small-group insurance market? How have large employers changed their practices in response to regulations in the ACA? Also largely uninvestigated is the nature of the interaction between the ACA and another statute affecting employee benefits, the Employee Retirement Income Security Act of 1974 (ERISA). Because self-insured employer plans fall under ERISA’s jurisdiction, they are not subject to state insurance regulation, but the ACA exempted ERISA plans from some of its requirements, leaving regulatory boundaries unclear. As Gluck, Allison Hoffman, and Peter Jacobson (2017) point out, unclear boundaries between ERISA’s jurisdiction and the portions of the ACA devolving regulatory authority to the states raises difficulties for states in using their regulatory authority to implement state-specific health reforms. Gluck, Hoffman, and Jacobson give as an example Gobeille v. Liberty Mutual Insurance Co., in which the Supreme Court ruled that ERISA preempted Vermont’s ability to require ERISA plans to participate in an all-payer claims database. 2 Another issue raised by ERISA-ACA interactions is the ACA’s exemption of self-insured ERISA plans from needing to cover “essential health benefits” as required of fully insured small-group (fewer than fifty employees) plans, an exemption that Corlette and colleagues (2017) note in their description of changes in small-group markets in six states has led some small employers to move toward self-insured plans.
The excellent articles in this issue expand our social scientific knowledge of the economic, social, and political effects of the Affordable Care Act. Just as earlier important social policy reforms have done, we fully expect the ACA to prompt new research for years to come.
Open Access Policy: RSF: The Russell Sage Foundation Journal of the Social Sciences is an open access journal. This article is published under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.