Policy Feedbacks and Medicaid on Its 60th Anniversary

Abstract

Context: Despite early skepticism about Medicaid's ability to withstand retrenchment as a program of “welfare medicine,” it has proved remarkably durable. Existing analyses explain durability from a policy feedbacks perspective—how program provisions affect the subsequent political environment and policy-making options. This article updates earlier feedback accounts to the Affordable Care Act (ACA) era.

Methods: The article examines extant findings on policy feedbacks in Medicaid at the elite and mass levels since the 2010 passage of the ACA.

Findings: Mass feedbacks have been modest. Medicaid expansion under the ACA only slightly increased beneficiary political participation, if at all. Medicaid attitudes among beneficiaries and the larger public have become somewhat more supportive. Elite-level feedbacks are the most powerful, with the federal contribution—increased for expansion populations under the ACA—inexorably shaping state incentives. However, continued rejection of Medicaid expansion and attempts to add conditions to Medicaid eligibility in Republican-led states with large shares of Black residents demonstrate that federalism, race, and the program's welfare medicine image continue to threaten the program.

Conclusion: Medicaid survives as the nation's largest health insurance program by enrollment, and it is deeply woven into the health care system. However, it remains chronically vulnerable and variable across states despite robust aggregate enrollment and spending.

Even before the Affordable Care Act (ACA) expanded Medicaid's reach, analysts noted the protective political dynamics around the program. Characterized early on as a “poor people's program,” Medicaid proved remarkably durable (Brown and Sparer 2003; Rose 2013; Sparer 2015; Thompson 2012). Whether durability was measured as enrollment, spending, or ability to withstand retrenchment attempts such as block granting, the program defied expectations and avoided the pattern seen among many other means-tested programs that are capped or wait-listed. Pre-ACA authors attributed the program's durability to the lure of federal matching dollars for states, robust program defense among stakeholder groups, and the inclusion of more sympathetic and highly resourced constituencies such as the working poor and the middle class as beneficiaries beyond the low-income core. Some analysts explicitly used a policy feedback approach to explain Medicaid's trajectory, asserting that policy provisions affected the interests and resources of stakeholders and constituencies, reshaping the political environment in ways that strengthened the program politically.

And yet, policy threats have always loomed. Some are economic, such as the ever-increasing cost of American health care, state budget pressures, and federal deficit concerns. Some are political, such as federalism and the race-to-the-bottom dynamics among states that it enables; political polarization; the rightward movement of the Republican party; the enthusiasm among both fiscal conservatives and social conservatives for trimming spending; the increase in Republican party control of state governments; the enduring racial politics around means-tested programs; and so on.

This article assesses the state of feedback processes around Medicaid in the post-ACA period, examining the status of Medicaid durability on the program's 60th anniversary. Updating Karch and Rose's initial (2019) examination, the article is a synthesis of extant research that considers what has happened with each of the feedback factors cited by earlier authors since the ACA was passed in 2010 and since most of its provisions came into effect in 2014. I argue that at the aggregate level, Medicaid's remarkable record of durability continues. Policy feedbacks at the elite level seem particularly important, as mass feedbacks have been more modest. Medicaid has long had two images: an “entitlement” image and a “welfare medicine” image (Grogan and Patashnik 2003). The program's continuing resilience squares with the former, that is, with the notion that Medicaid is an essential entitlement that extends into the middle class and that is “popular, . . . durable, and has considerable potential for expansion over time” (Lanford 2020: 141).

However, the alternative image of Medicaid as welfare medicine continues to lurk and often influences policy outcomes, despite the many positive, protective feedback effects that have accumulated. In this perspective, Medicaid is primarily seen as a program for poor people, with the public viewing beneficiaries as “undeserving loafers and minorities” (Lanford 2020: 141, following Quadagno [1994] among others). Aggregate measures such as total spending or total enrollment mask variation across states and continued threats to Medicaid. Examples of the power of the welfare-medicine policy image include the refusal of 10 Republican-led states (many with large uninsured Black and Latino populations) to expand Medicaid; the attempts to impose work requirements and other forms of conditionality (approved by the first Trump administration and thwarted only by Biden's narrow victory); the periodic attempts to block grant the program; the aggressive disenrollment of recipients during the “unwinding” of pandemic-era protections; and the enormous gulf between Democratic-identified and Republican-identified people in their support for Medicaid and the ACA (even if Republicans with personal experience are more supportive of Medicaid than of other means-tested programs). Medicaid has proven more politically robust than other social assistance programs, but it is chronically susceptible to attempts to diminish its reach. The ACA was intended to make Medicaid more uniform nationwide, with eligibility based on income only. Instead, cross-state variation is more pronounced than ever, enabled by federalism and fanned by race.

Policy Feedbacks at the Elite and Mass Levels

Policy feedback is a theoretical perspective that argues that public policies are not merely the outcomes of political processes but also are crucial inputs (for an overview see Béland, Campbell, and Weaver 2022). Existing policies can alter the political environment among both policy-making elites and members of the mass public, affecting the course of future policy making.

At the elite level, policies can alter resource levels for relevant stakeholder groups, shape policy makers’ incentives and views about what good policy is, impose budget constraints, and provide models for subsequent policies. As existing policies change the resources, interests, and incentives of policy-relevant actors, they either facilitate subsequent policy initiatives and reforms (positive feedback) or undercut them (negative feedback).

Extant policies can also shape the factors driving political participation and attitudes among beneficiaries and the larger public. Social welfare policies such as Medicaid can enhance program support and political participation rates among recipients (positive feedbacks) by providing politically relevant resources (income, financial security, health); enhancing political interest by tying well-being to government activity; increasing political efficacy by “hearing” and addressing citizens’ needs; fostering feelings of “reciprocity” (gratitude to government that gets channeled into support and participation); and providing opportunities for mobilization by political parties and advocacy groups by defining groups by virtue of program receipt (Campbell 2012; Mettler 2005; Mettler and SoRelle 2018). By fostering program support and facilitating political participation among recipients who come to value and defend their benefits, a protective constituency dynamic emerges (Pierson 1993). Alternatively, negative program experiences—such as being treated poorly by gatekeeping caseworkers or encountering significant administrative burdens—can block access to resources and undermine political efficacy, decreasing political participation rates (Michener 2017; Soss 1999). Positive and negative feedbacks have been demonstrated for a variety of US social policies, including cash welfare (Soss 1999), Social Security (Campbell 2003), and the GI Bill (Mettler 2005).

Policy Feedback Perspective on Medicaid before the ACA

Medicaid's status as a means-tested program suggests political vulnerability. Originally, eligibility was tied to cash welfare, and Medicaid seemed likely to face political challenges similar to those threatening that unpopular program. Many members of the larger public, who saw program recipients as only marginally deserving, especially after the media-fueled racialization of welfare and attendant programs in the late 1960s, were skeptical about health insurance for the poor (Gilens 1999). Opposition among small-government conservatives grew with the rightward shift in the Republican party beginning in the 1980s; stakeholder groups were lukewarm, with providers concerned about the program's low reimbursement rates. The program's vulnerabilities could be summed up as “poor policy and poorer politics” (Brown and Sparer 2003: 31).

Yet Medicaid proved remarkably durable. Over the decades, eligibility expanded and enrollment and spending grew. Retrenchment efforts, such as block-grant proposals in 1995 and 2003, were defeated. Frank Thompson (2012) and Shanna Rose (2013) used policy feedback arguments to explain the program's persistence and growth despite its status as a means-tested program.

In the Thompson and Rose accounts, the policy feature most central to the program's durability is the open-ended federal match. Both scholars argue that the federal commitment to match each dollar of state Medicaid spending with a federal dollar (or more, for lower-income states) created an irresistible fiscal carrot. The combination of entitlement to the states and shared federal-state control fueled growth, fed by an asymmetric logic: states could cut Medicaid, but every state dollar saved meant at least a federal dollar lost. The entitlement structure's incentives fueled “catalytic federalism,” in which state and federal leaders prodded each other to expand the program.

A second feedback factor Thompson and Rose specify is interest-group support. Providers, especially hospitals, perpetually lamented Medicaid's low reimbursement rates; but low reimbursement is preferable to no reimbursement. Hospitals are politically powerful interest groups in most states, and they fought program cuts, especially the safety net and rural hospitals for which Medicaid is an especially important insurer.

Third, Thompson and Rose cite a mass feedback arising from Medicaid's middle-class (or at least nonpoor) constituencies. Although low-income pregnant women and children have always constituted a large share of the program's clientele, also covered are a multitude of middle-class constituencies for whom other public or private coverage is difficult or impossible to acquire. These include formerly middle-class elders in need of nursing home and home health care (and their adult children) and intellectually disabled children (and their parents). Even before the ACA, eligibility also expanded to some lower-income workers at the federal level, thanks to the advocacy of lawmakers such as Henry Waxman (D-CA), and to those in some richer and more liberal states. These nonpoor constituencies were thought to exert a politically protective effect on the program (even if direct evidence for this dynamic is scarce).

This perspective on Medicaid's pre-ACA dynamics attributes the program's durability to policy feedbacks arising from specific design features. The open-ended federal match, dependence among hospitals for safety net insurance coverage, and inclusion of higher-resource nonpoor constituencies altered the political environment around the program. Unlike many means-tested programs, Medicaid enjoyed political protection from powerful allies well beyond the impoverished beneficiaries at its core. The program may have begun life as “welfare medicine,” but its entitlement status provided both substantial stakes to key actors and an alternative policy image as a partly middle-class program that other social assistance programs do not enjoy, helping explain its nearly unique upward trajectory, at least according to aggregated, national-level measures such as spending and enrollment.

Mass Policy Feedbacks since the ACA

As Karch and Rose (2019) noted in their earlier post-ACA assessment of Medicaid's status, many of the positive feedbacks that politically protected Medicaid in previous decades have continued and even blossomed. On the eve of its 60th anniversary, the program covers one fourth of Americans, more than any other private or public insurer, and constitutes the third-largest domestic item in the federal budget, just behind Social Security and Medicare. Threats such as block granting and ACA repeal-and-replace have come and gone (at least for now). Assessed at the 30,000-foot level, the program continues to display remarkable resilience.

Medicaid remains politically vulnerable nonetheless. Pre-ACA elite-level feedbacks either remain in place or are stronger (the federal match is even greater for expansion populations, and the substance abuse and rural hospital financial crises have only heightened the importance of Medicaid funding to state budgets). At the mass level, however, additional policy feedbacks are limited or compromised. The public seems mildly more supportive of Medicaid and the ACA but still deeply divided by party, and political participation has not increased much, if at all, among new beneficiaries. Moreover, the politically weaker of Medicaid's two images, that of welfare medicine, remains latent, always lurking in the background and waiting to be triggered in the public mind, a continual political threat to Medicaid's durability.

Behavioral Feedbacks among Medicaid Recipients

As many policy feedback researchers have hypothesized, receiving Medicaid could enhance the political participation of its beneficiaries, and turnout and other activity could increase among lower-income individuals as the reach of Medicaid spreads with ACA implementation. Political participation is a function of politically relevant resources, political engagement, and mobilization (Verba, Schlozman, and Brady 1995), all of which could be enhanced by gaining government health insurance. Having health insurance provides some measure of financial stability, which may facilitate political participation. Health insurance can also lead to improved self-reported physical and especially mental health (Finkelstein et al. 2012), which are associated with higher levels of political participation (Ojeda 2015; Pacheco and Fletcher 2015). Gaining government health insurance could enhance one's interest in public affairs (Clinton and Sances 2018) or increase one's sense of external political efficacy (the sense that government “listens to people like me”), a positive “interpretive” effect (Pierson 1993) that can occur when government confers a benefit and recognizes the recipient as a worthy citizen of the polity. Medicaid's spread could enhance political participation through a third mechanism: mobilization. The 1993 National Voter Registration Act requires social assistance agencies to provide voter registration services, which may have increased participation in Medicaid expansion states (Clinton and Sances 2018). Many health navigators who assisted individuals in signing up for health insurance under the ACA provided voter registration services as well (Families USA 2016).

Despite these plausible hypotheses connecting Medicaid recipiency and political behavior, the evidence does not suggest that the program enhances voter turnout rates in a linear way, at least not for traditional low-income Medicaid populations. Both observational and causal studies indicate that in the short term, newly enrolling in Medicaid boosts turnout. In an observational study of congressional district–level turnout, Haselswerdt (2017) found that increases in Medicaid enrollment were associated with higher aggregate turnout, which he argues was the result of higher voter participation among new beneficiaries and among nonbeneficiaries opposed to the ACA, a backlash effect (see also Haselswerdt and Michener 2019). Causal models using individual-level Oregon Health Plan lottery data (Baicker and Finkelstein 2019) and comparing Medicaid expansion states and nonexpansion states under the ACA at the aggregate level (Clinton and Sances 2018) both found a small increase in turnout arising from Medicaid enrollment.

However, in both of these studies, the small increase in turnout disappeared by the next election. Looking at the longer term, Jamila Michener's work (2017, 2019) suggests that the continued racialization and stigmatization of the program may undercut any initial resource boost provided by gaining insurance. That is, the “experiential” aspects of program receipt, which are marked by administrative burden and, in some states, conditionality such as monthly premiums and lock-out periods arising from noncompliance with such requirements, both undercut any resource boost and send negative messages to recipients about their worth as citizens. Given the transient nature of any positive feedback on Medicaid recipients’ political participation, and the tremendous variation in the recipient experience across states enabled by the program's hybrid federal-state design, it is difficult to attribute much of Medicaid's continuing political durability to increased political participation among its low-income clientele.

Medicaid may have produced policy feedback effects among program constituencies beyond the low-income core, and perhaps the political participation of these groups has helped protect the program from retrenchment. These include the disability movement, which has long been an organized and vigorous defender of Medicaid, and middle-class parents of children with intellectual or developmental disabilities, who are a high-resource group with a large stake in the program. Both of these groups fought Republicans’ ACA repeal-and-replace plans and block-granting proposals vociferously (Abrams 2018). The evidence that these groups constitute a protective constituency remains anecdotal rather than systematic, however, as they are typically not identified in the survey or other data researchers use to assess feedback effects on citizens.

Finally, another form of behavioral policy feedback could be vote choice. This could take the form of electoral punishment. For example, in the statewide election held after Tennessee disenrolled a large number of Medicaid recipients in 2004, the incumbent Democratic governor experienced greater vote share loss in counties where more people lost coverage (Kogan 2022). Or it could take the form of electoral reward: between 2012 and 2016 under the ACA, Democratic vote share in counties in Medicaid expansion states increased relative to neighboring counties in nonexpansion states, particularly in counties where the opioid epidemic was less severe (where the epidemic was worse, Democratic vote share gains disappeared; Shepherd 2022).

Attitudinal Feedbacks among Recipients and the Larger Public

Although the evidence that Medicaid has increased recipients’ political participation rates is modest to nonexistent, a variety of studies suggest that with its extension under the ACA, Medicaid has enjoyed increased support among both recipients and other members of the public, a positive attitudinal feedback effect. Policies could affect attitudes in several ways (Béland, Campbell, and Weaver 2022). Recipients and their family members have a personal stake in programs, with this sense of self-interest leading to greater program support. Such a sense may be heightened when such benefits are threatened. Receiving a government benefit that enhances one's well-being also sends positive citizenship messages to recipients, increasing their efficacy and in turn their program support. Among nonrecipients, a program's very existence may constitute a “new normal” that fosters approval. Cues from supportive political elites may also encourage assent among other members of the public.

A central limitation in assessing whether Medicaid has had attitudinal feedback effects on recipients is that few studies both directly identify Medicaid recipients and measure Medicaid sentiment. One exception is Grogan and Park's (2017a) analysis of cross-sectional survey data from 2015. They find that support for Medicaid is higher among the two thirds of Americans who reported some connection to Medicaid. Those who had an indirect connection (close friend or family member enrolled) or a current or previous direct connection (the respondent or their child was enrolled) were more likely than those with no connection to say that Medicaid is important, that spending should be increased, and that the program should be expanded (see also Grogan and Park 2018).

A number of other studies examine whether ACA support has grown among those who newly gained insurance through Medicaid expansion (or among those who would be eligible in expansion states). Note that in these studies, the feedback is support for the ACA generally rather than for Medicaid specifically (although perhaps individuals were thinking of Medicaid when they responded to the ACA survey item). Some of these studies use aggregate data. For example, a difference-in-differences analysis of 2010–2017 survey data shows that possible Medicaid enrollees—low-income respondents—became more favorable toward the ACA in Medicaid expansion states relative to nonexpansion states. Because there was no change in ACA attitudes among high-income respondents, the change in attitudes among low-income respondents is attributed to their greater Medicaid eligibility (Hopkins and Parish 2019). Similarly, Sances and Clinton (2021) collected 200 surveys on ACA support fielded from 2009 to 2017. The varied timing of states’ Medicaid expansions allows them to estimate a causal model that indicates that favorability toward the ACA increased 1.5 percent in expansion states compared to nonexpansion states. The increase is concentrated among low-income respondents younger than age 65, the target population, with no attitudinal change among higher-income groups.

A smaller number of studies employ individual-level data and are able to identify respondents with Medicaid insurance (not merely in the Medicaid target group). In her causal analysis of individual-level panel data from the early days of ACA implementation (September 2013–November 2014), Adrienne Hosek (2019) was able to assess attitudes toward the ACA among newly insured Medicaid individuals. Support increased, although the result is not statistically significant because of small numbers of respondents (those who enrolled in health plans on an insurance exchange also had more positive opinions of the ACA, an effect that was statistically significant).

Other studies assess attitudinal feedbacks on the public writ large, not just on possible or actual Medicaid recipients. Support for the ACA increased in the face of congressional Republicans’ ACA repeal attempts, especially after the 2016 election ushered in unified Republican control of the federal government, heightening the sense of threat (Mettler, Jacobs, and Zhu 2023; Sances and Clinton 2021). But these studies assess feedback effects of the ACA on ACA opinion rather than feedback effects of Medicaid on Medicaid opinion. They also put the question to the entire public, not just the Medicaid target population. Thus, we cannot know which component of the ACA respondents had in mind when answering the question (i.e., Medicaid expansion, insurance exchanges, the ban on preexisting condition exclusions, the extension of parents’ insurance to children younger than age 26, or other regulations).

Another way to assess whether Medicaid has had mass feedback effects is to consider the fate of ballot initiatives in which the larger public was asked to determine whether to expand Medicaid in states where elected officials had declined to do so. Early polling suggested such efforts might be successful: two-thirds of respondents in nonexpansion states favored expansion in a 2015 Kaiser survey (Grogan and Park 2017a). Since then, seven of the nine nonexpansion states with a direct democracy option have approved Medicaid expansion by ballot measure (Idaho, Maine, Missouri, Nebraska, Oklahoma, South Dakota, and Utah, leaving only Florida and Wyoming as potential expansion-by-direct-democracy possibilities).¹ These election results constitute a behavioral indicator of pro-Medicaid attitudes arising from the ACA's expansion framework.

Limits to Positive Mass Feedback Effects among the Public

Although numerous studies support the view that Medicaid has generated positive feedback effects among members of the public who have connections to Medicaid, and even among voters without direct ties, it is important not to overstate the protective role that public opinion plays in Medicaid's durability. The program's welfare medicine image is a constant undercurrent, and opinion about the program continues to be divided by party identification and racial sentiment.

The partisan divide in Medicaid attitudes remains vast. Although Republicans with personal experience with the program are more supportive, most Republicans remain much less favorable toward the program than Democrats are (Campbell 2015). ACA support is also deeply divided by party, and perceptions of the effectiveness of the ACA in improving access to health insurance differ by partisanship and level of government distrust (Jacobs and Mettler 2016). The ACA also seems to have elicited a backlash effect among Republican voters: Haselswerdt (2017) found that voter turnout in House races declined less in the 2014 midterm election compared to the 2012 presidential election in Medicaid expansion states, but this positive turnout effect was apparent not just among Democrats but also among Republicans, suggesting that expansion mobilized opponents as well as supporters.

Racial resentment also continues to influence Medicaid attitudes. For example, work requirements are a policy tool that appeals to racialized impressions among some members of the public that social assistance recipients are lazy. Survey evidence shows that support for work requirements is higher among conservatives and the racially resentful (Haeder, Sylvester, and Callaghan 2021). Across states, white people's opinions about eligibility for parents of Medicaid-eligible children vary with attitudes toward Hispanic Americans and toward undocumented immigrants, and opinions about eligibility for nonparent adults vary with attitudes toward Hispanics and the undocumented as well as racial resentment (Lanford and Quadagno 2022).

Republican leaders in the 10 states that have still not expanded Medicaid continue to invoke racialized welfare imagery when discussing the program (on racial disparities in coverage, see Sommers, Smith, and Figueroa in this issue). In the second Republican presidential debate in September 2023, Florida Governor Ron DeSantis was asked about the large number of Floridians without health insurance. DeSantis said there had been a population boom in the state and that there are not “a lot of welfare benefits in Florida. . . . We basically say this is a field of dreams, you can do well in this state, but we're not going to be like California and have massive numbers of people on government programs without work requirements. We believe you work, and you gotta do that, and so that goes for all of the welfare benefits” (Perry 2023).

There is some anecdotal evidence of negative attitudinal feedbacks arising from the ACA's design, with some individuals with Marketplace plans angry about the high monthly premiums and deductibles they face in comparison to the “free” insurance Medicaid recipients enjoy. Such effects may contain an element of racial resentment. They are also an example of a “self-undermining” design feature in the ACA that undercuts its popularity (Jacobs and Weaver 2015).

In sum, the evidence of Medicaid feedbacks on mass publics is positive but mostly mild (or we lack direct evidence). The extension of Medicaid to new beneficiaries may have increased their voter turnout, but not in the long term. The vociferous efforts of disability activists to thwart Republicans’ repeal-and-replace efforts seem to have worked, although we have only anecdotal evidence; and another factor—the inability of the ACA's opponents to devise a plausible replacement plan—also undermined the repeal movement. Personal experience with Medicaid fosters support for the program, and the inclusion of Medicaid in the ACA's suite of coverage expansions seems to have bolstered support among the larger public too, as the ACA has become an established part of the American health care landscape. But partisan differences remain large: overall favorability toward the ACA peaked at 62 percent in the March 2023 KFF Health Tracking Poll, but the gap between Democrats and Republicans was an enormous 61 percentage points, at 90 percent to 29 percent (Rudowitz et al. 2023). And although Medicaid is the most popular means-tested program (Campbell 2015), racial resentment remains a strong attitudinal driver among white Americans, which the program's expansion under the ACA has done little to mitigate.

Elite Policy Feedbacks

Although it is difficult to know how much of Medicaid's durability is the result of mass feedbacks, the elite-level feedbacks described in pre-ACA accounts remain intact, although not without limits, as the example of the remaining nonexpansion states shows.

Positive Elite Feedbacks

The fiscal logic of the federal match is as inexorable as ever and has even been strengthened because of the higher match for expansion populations, with the federal government providing 100 percent of the cost of covering the newly eligible for two years and 90 percent thereafter. The power of the match is seen in the gradual adoption of expansion in many red states. Although the majority of states with Republican governors filed lawsuits against the ACA immediately after its passage, some Republican-led states chose to expand. In Nevada, Arizona, and Ohio, Republican governors pursued expansion; in Arkansas, Kentucky, and Massachusetts, Republicans who followed Democratic governors kept expansion in place. And these elite actions have mass consequences: when Republican governors announce support for Medicaid expansion, attitudinal polarization in the public diminishes as Republican identifiers become modestly more supportive of the ACA (Pacheco, Haselswerdt, and Michener 2020).

Beyond the budgetary logic of the federal match, pressure from hospital groups and other stakeholders is also a key elite-level feedback surrounding Medicaid. Located in every state legislative and congressional district, providers such as hospitals, home health care agencies, nursing homes, community health centers, and physicians are organized and exert pressure on lawmakers to resist cuts and bolster reimbursements (Brown and Sparer 2003). As Frank Thompson (2012: 26) puts it, “To a greater degree than other redistributive programs, Medicaid can count on a coalition of politically potent providers to defend it.” Since threats crop up regularly—from the ACA's proposed cuts to Disproportionate Share Hospitals to the rural hospital crisis of recent years—these economically crucial stakeholders make regular trips to state capitols and the halls of Congress, repeatedly reminding lawmakers of the interests at stake.

Another elite-level feedback arises from Medicaid's middle-class constituencies. Elite invocations of middle-class constituencies have proven politically important. In 2023, North Carolina became the 40th state to expand, and the program's racialized image was nowhere to be found in elite rhetoric. Appealing to their fellow state legislators, proexpansion Republicans cited two factors: first, the state was losing out on more than $500 million per month by not expanding, invoking the federal match logic; second, expansion would help a variety of deserving groups, including farmers and other rural residents, veterans, the middle class, and children in need of mental health services as well as rural hospitals. Some examples from an analysis of the language Republican legislators used to justify the expansion (Little and Searing 2023): “It will go a long way toward helping our middle class.” Expansion will “help our rural hospitals, and North Carolina is still 80 percent rural.” “Expansion is about that person, whether they be a veteran, or a farmer that risks losing their farm because they have some catastrophic health care bill that they can't pay. It's really about helping the people of North Carolina.”

The ability to invoke middle-class constituencies helps the backers of expansion convince their colleagues that expansion is “good policy” and will have positive electoral payoffs.

Limits to Positive Elite Feedbacks

Despite the elite forces that undergird Medicaid's durability, the program remains vulnerable. The entitlement image provided by the federal match and the appeal of covering middle-class constituencies are always in tension with the program's welfare medicine image. Program opponents who try to combat program growth continue to invoke the program's cost and the “undeserving” portions of its clientele. For example, although the North Carolina expansion decision was a triumph for program proponents, the Republican-led legislature gave themselves one more out before committing to expansion by making it contingent on passing the state budget.

A chief source of Medicaid's continuing vulnerability is federalism and the significant authority that subnational units have over the program's implementation. As Béland, Rocco, and Waddan (2016, 2020) have written, federalism fractures authority over Medicaid and “frustrated” the implementation of the ACA, allowing partisanship and race to undermine health insurance expansion. State decisions to expand are a function of Black share of the population and hinge more on the preferences of whites than nonwhites (Grogan and Park 2017b; Olvera, Smith, and von Lockette 2023). And the toll of federalism is reflected in local action as well. A study of county-level Medicaid and Children's Health Insurance Program transfers found a “patchwork of inequality” at the substate level, with transfers received or requested by counties varying with the Black share of the population (Olvera, Smith, and von Lockette 2023: 535). The authors find a curvilinear pattern in which Medicaid transfers are modest at low Black population levels, increase with share of the Black population (consistent with a Black empowerment argument), but then level out when Black composition reaches 50 percent (consistent with a racial threat or conservative backlash thesis). The results show “how federalism can be wielded to increase racial disparities through the politics of policy implementation” (Olvera, Smith, and von Lockette 2023: 535).

Political conservatives’ repeated attempts to impose conditions on Medicaid eligibility such as work requirements, monthly premiums, and lockout periods also demonstrate Medicaid's continuing vulnerability. No states that have implemented Medicaid expansion have repealed it, Béland, Rocco, and Waddan (2020) note, as revoking eligibility for thousands outright is a bridge too far for even the most conservative governor or state legislature. Instead, states have tried to reinstate moral probity and “personal responsibility” as criteria beyond the ACA's income-only guidelines with these forms of conditionality. Many states submitted waivers to the Centers for Medicare and Medicaid Services (CMS) seeking approval for these provisions, which the Trump administration granted for the first time in the program's history. Work requirements in particular evoke Medicaid's racialized policy image of welfare medicine, rather than its entitlement image, “reinsert[ing] the distinction between the deserving and undeserving poor in Medicaid, which the ACA discarded” (Huberfeld 2019: 209). Work requirements “are hallmarks of benefits for the poor and are heavily racially coded. They assume that the poor, unlike the wealthy, are in need of behavioral control. They assume that poor people are in need of our teaching” (Bach 2019: 39). Those seeking waivers included Republican governors in expansion states who faced backlash from their voters. Proposing work requirements and other forms of conditionality helped them shore up their political support. The practice then diffused to nonexpansion states, which imposed them in their existing Medicaid programs, making them even more restrictive than they had been before the ACA (Fording and Patton 2020). Only some court decisions and then Biden's 2020 presidential win prevented these work requirements from being adopted, and no doubt such waivers will be entertained by future Republican administrations.

Another example of Medicaid's continued vulnerability is state variation in implementation of the postpandemic “unwinding” process. During the COVID-19 public health emergency, the federal government required states to keep recipients continuously enrolled. Coverage grew to the highest in program history, reaching more than 94 million in March 2023, up from 71 million before the pandemic (CMS 2023). In April 2023, the continuous coverage requirement ended, and “unwinding” began, as states reinstated eligibility redeterminations. Millions lost coverage, some because they were no longer eligible; others (including eligible individuals) because of administrative barriers (for more on administrative barriers as a tool for rationing care, see Herd and Johnson in this issue). Even states committed to supporting Medicaid experienced backlogs or vetted statuses incorrectly, while others, especially nonexpansion or Republican-led states, used the opportunity to trim their Medicaid rolls aggressively. As of mid-September 2023, more than 6.4 million people had lost coverage (CBPP 2023), with some estimating that as many as 17 million would lose coverage during the unwinding process (Burns et al. 2023).

State variation in unwinding reflects the welfare medicine image of Medicaid. Arkansas moved particularly quickly, seeking to complete the process in six months rather than the year allowed by CMS. In a Wall Street Journal opinion essay, Arkansas governor Sarah Huckabee Sanders wrote, “We're simply removing ineligible participants from the program to reserve resources for those who need them and follow the law,” despite efforts among “some Democrats and activist reporters” who “want to keep people dependent on the government” (Sanders 2023). An examination of KFF data from October 2023, six months into the unwinding, found that nonexpansion states and states with Republican legislative control disenrolled a greater share of those reviewed than expansion states and states with Democratic legislative control.² Thus even elite-level policy feedbacks provide uneven protection to the program across states.

Theoretical Lessons and Research Avenues

Medicaid's trajectory sheds light on policy feedbacks theory itself. One question concerns the sources of political influence and whether constituencies can protect programs by themselves or need the influence of elite allies as well. Consider the case of Social Security, which has had a multitude of positive feedback effects on its clientele, transforming them from the group that participated the least in American politics to the one that participates the most. The pension program granted senior citizens stable resources generous enough to pull nearly all of them out of poverty and made retirement and its large dose of free time a reality for most. It fostered political interest by tying seniors’ financial status so visibly to a government program, and it encouraged mobilization by interest groups and politicians by creating a group identity as program beneficiaries (Campbell 2003). As a result of these policy feedbacks, older Americans have become the quintessential protective constituency, so much so that R. Douglas Arnold, in his analysis of what Congress is likely to do on the eve of the program's trust fund insolvency in 2034, predicts that seniors will prevail over the preferences of the donor class. Outcomes are somewhat contingent on party control of government at that moment, but he concludes that the program will not be privatized, as the financial industry desires, and will not be greatly cut, as donors hostile to entitlement spending want. Instead, the preferences of 84 million highly attentive, highly participatory recipients who are key to lawmakers’ reelection chances will prevail, and the program will be shored up through some combination of tax increases and perhaps some cuts to less vulnerable subpopulations such as affluent retirees (Arnold 2022).

What does it take for a program to gain that protective shell? In Medicaid's case it seems that mass feedbacks are not enough. The program's low-income beneficiaries are not very politically active to begin with, and the ACA's rollout shows that participatory boosts from newly achieving insurance coverage are small. The program's more politically active middle-class beneficiaries, whom lawmakers need to court, may augment the core population's influence, as the rhetorical strategies of Medicaid boosters in North Carolina suggest. But the program's durability to date is due primarily to elite feedbacks and only modestly to mass feedbacks. One lesson for policy feedbacks theory and research is that mass and elite feedbacks may need to be considered together rather than separately, as is often the case. Their interaction or net effect may be what determines program trajectories.

A second question concerns the conditions under which programs have positive and negative feedbacks. Early research tended to paint means-tested programs with a broad brush, suggesting that all of them have negative effects on recipient participation rates and attitudes toward the state because of the demeaning processes for accessing benefits and the surveillance regime that comes with work requirements and other forms of conditionality. But as Carolyn Barnes's (2020) research has shown, means-tested programs can be designed with an incorporating customer orientation just like Social Security and can have positive effects on citizen engagement. For example, recipients of assistance from the Women, Infants, and Children nutrition program report supportive encounters with caseworkers, in contrast to “stigmatizing” interactions with caseworkers for the Supplemental Nutrition Assistance Program and Medicaid (Barnes, Michener, and Rains 2023: 3). Research into feedback effects arising from varied levels of administrative burden, differing orientations of and incentives for caseworkers, and other practices across states would shed light on this question.

A third theoretical question concerns the social construction of target populations. In an old debate, Lieberman (1995) argues that Schneider and Ingram's (1993) conceptualization of the relationship between policy designs and the social images of recipient populations lacks causal clarity. Do program designs create popular images of beneficiary groups? Or are program designs chosen because of existing popular images (e.g., the elderly have always been more sympathetic than the poor)? With the expansion of Medicaid more visibly to working poor and middle-class constituencies, we have the opportunity to examine which way causality runs: will these middle-class constituencies be seen in a less favorable light because of their association with a means-tested program, or will the program be seen in a more favorable light because of its association with these more sympathetic and deserving constituencies? Researchers could also examine which of Medicaid's many constituencies are foremost in individuals’ minds when they think of the program and under what conditions. This would address theoretical questions about the policy images of multiheaded policies like Medicaid.

Finally, policy feedbacks researchers need to continue to examine how federalism and race affect the ways in which existing policies affect future politics. Thanks to the work of Jamila Michener, Daniel Béland, and their coauthors we know more than we used to, but clearly both subnational control over policy parameters and implementation and persistent stereotypes about the deservingness and behavior of Black people and other nonwhite recipient groups continue to shape program trajectories.

Conclusion

Medicaid has had a different trajectory than many other social assistance programs with limited funding and constrained enrollments. Its central place in the American health insurance landscape only grew with the ACA, and even Republican-led states now number among those embracing expansion to new populations. The protective factors cited in pre-ACA research, especially the generous federal match and stakeholder pressure, continue to influence policy making around the program. Causal analysis made possible by ACA expansion suggests that the program has had only a modest short-term effect on the political participation of its low-income clientele. But the relevance of the program for some “deserving” and higher-turnout groups such as the middle class and rural populations helps protect the program, if nothing else providing rhetorical cover for conservative politicians voting for expansion. Growing support for the ACA over time may in part indicate increased acceptance of Medicaid expansion among the larger public.

Medicaid's welfare medicine image persists, however. Ten states have still not expanded the program, and opposition forces continue to rely on racialized imagery. No doubt work requirements will return as policy proposals, and block-grant attempts may return too as concerns about the federal debt increase. The program's hybrid federal-state design means that similar individuals living in different states and localities have differing access to Medicaid and face differing administrative burdens and sanctioning regimes. The ACA aimed to make Medicaid more uniform nationwide, but subnational control and the concerted efforts of program opponents to block or complicate access has heightened variation in access, worsening health care inequities.

On its 60th anniversary, Medicaid is in many ways as robust as ever. In 40 states it now covers all low-income citizens, not just the categorically eligible, as was the case before the ACA. It covers a huge swath of Americans and remains a true safety net for those left out of other public and private insurance. But old tropes about recipients being lazy and undeserving endure, and large proportions of Republican lawmakers and identifiers among the public remain opposed to the program and to the ACA, in which it is a central plank. The program remains simultaneously robust and fragile, durable yet vulnerable.³ Black Americans and other nonwhites remain the chief victims of opponents’ access-limiting strategies.

Despite advances, our ability to assess policy feedbacks in Medicaid, especially at the mass level, is thwarted by lack of data availability. Public opinion evidence on Medicaid opinion is thin, in contrast to the large amount of polling on Medicare. Many surveys ask about ACA approval but shed limited light on Medicaid sentiment, as it is difficult to know which of the ACA's many components respondents have in mind when responding to such survey items. Moreover, most examinations of behavioral feedbacks around Medicaid, such as the analyses comparing expansion and nonexpansion states, have focused on one group (Medicaid's low-income beneficiaries) and one political act (voter turnout, with vote choice examined in a few instances). We need more evidence to examine feedback effects among the multiplicity of Medicaid subpopulations and for political acts beyond voting (see Michener in this issue for a qualitative examination of other types of political activity around Medicaid). Finer-grained analyses examining variation in feedbacks arising from state variation in generosity or level of administrative burden would also be welcome. Given that such a large share of Americans is enrolled in Medicaid, future data collection efforts and analyses should give Medicaid the attention it deserves.

Acknowledgments

I thank Morgan Gillespie for her superb research assistance. I am grateful to participants in the “Medicaid at 60: From Welfare Medicine to America's Most Important Healthcare Program” conference (Brown University, October 20–21, 2023), Eric Patashnik, and the anonymous reviewers for their very helpful comments.

Notes

References