Context: A political science literature has emerged on the policy feedback effects of alternative health care coverage expansions, focusing on whether programs like Medicare, Medicaid, and the Affordable Care Act can generate robust public constituencies. Yet, scholars have mainly examined direct governmental benefits and rarely studied the feedback effects from the underlying regulatory frameworks that allow health care markets to function, such as regulations that determine eligibility for nongroup and continuation coverage.
Methods: Drawing on literature from multiple disciplines, this article analyzes political and policy factors that promote and undermine such regulations' capacity to stimulate positive feedback and the emergence of citizen constituencies.
Findings: While some factors may help these regulations to cultivate constituencies, there are barriers to this positive feedback generation. The most notable is the failure of these regulatory policies to confer politically useful resources on citizens unless other, mutually supportive policies are also in place, such as policies to control premiums.
Conclusions: Steps in the policy feedback model that are typically left unexplored in the literature deserve greater scrutiny when scholars examine the weak self-reinforcing effects of health insurance regulations. A take-away lesson, relevant to the Affordable Care Act, is that even when policies broaden nongroup coverage and ostensibly assist many people, a robust citizen constituency is unlikely to emerge, leaving key protections of access to health insurance vulnerable to erosion or reversal over time.
Public policy is an unfolding process in which citizens' reactions to preexisting public policies may influence subsequent rounds of politics and policy making. Lawmakers may repeal policies that have sparked opposition and citizen protest (Street 1993). In contrast, lawmakers will face pressure to preserve policies that citizens have embraced (Campbell 2003). Citizens' reactions to specific policies may also transmit lessons to lawmakers about which general policy approaches are most likely to be politically viable over time (Barabas 2009; Lerman and McCabe 2017). The feedback effects that policies stimulate—or fail to stimulate—can also shape material inequalities since policies (re)distribute resources across groups, advantaging some and disadvantaging others (Mettler and Soss 2004).
For all these reasons, policy feedback scholars study the conditions under which policies generate robust citizen constituencies. Such constituencies can actively push for benefit expansions, or they can be a “latent” force poised to mobilize should repeal threats emerge (Arnold 1990: 68). Positive group feedback effects are likelier to occur when a policy creates “interpretive effects,” signaling that a constituency is morally deserving and held in high esteem (Campbell 2003; Mettler 2002; Pierson 1993). Constituencies can also emerge when a policy creates “resource effects” by conferring economic or political resources, such as money or time, that are of sufficient value to stimulate political engagement (Campbell 2003; Mettler 2002; Pierson 1993).
Health policies may be especially likely to generate citizen reactions (positive or negative) due to the “intensely personal nature of health and health care and citizens' keen attention to this issue area” (Campbell 2011: 971). Policies that govern health insurance would seem to be particularly good candidates for positive feedback generation since insurance mediates access to such resources as health care, health status, financial security, and job flexibility.1 Such resources substantially impact citizens' day-to-day lives (Campbell 2014; Pierson 1993: 606) and relate positively to political participation.2
However, scholars are still assessing which of the several policy approaches used in the United States to link people to insurance are most prone to generate a high level of citizen enthusiasm. Seniors clearly defend Medicare against direct benefit cutbacks (Campbell 2003; Kelly 2015; Oberlander 2003), yet they do not consistently mobilize against subtle declines in the value of the coverage they receive under the program (Campbell 2011). The Medicaid program—which serves diverse constituencies, including the poor, the disabled, and formerly middle-class persons who have spent down their assets—has generated such varied mass and beneficiary reactions as enthusiasm, ire, and passivity (Haselswerdt 2017; Michener 2018). The public has been divided over the Affordable Care Act (ACA) as a whole (Jacobs 2014a) and has shown muted or mixed political reactions to specific ACA provisions that ease access to private coverage (Chattopadhyay 2017; Lerman, Sadin, and Trachtman 2017; Sances and Clinton 2018).
Ideally, the literature on constituency formation in health policy will develop a framework to explain and reconcile these diverse empirical findings. Greater precision is required to isolate the feedback effects from specific policy instruments. One set of instruments that has received too little attention in efforts to explain these patterns is health insurance regulations (Hackey 2000; Meier 1991; Oliver 2000). In particular, there is a paucity of research on Americans' political reactions to eligibility-focused regulations (EFRs). In short, an EFR is a regulation that ensures that a person is eligible to receive a private health insurance offer, of some type at some price, even if that person's health status or employment circumstances may make him or her unattractive to insurance companies. Two types of EFRs appear to have received sufficient attention from scholars and lawmakers to make them good candidates for studying this larger class of policies. One is “guaranteed issue” in the nongroup market, which requires insurers to offer coverage regardless of applicant health history. A second type is “continuation coverage” mandates, such as the 1985 Consolidated Omnibus Budget Reconciliation Act (COBRA) and similar state-level laws. To the degree that scholars study mass politics around such regulations, they mainly predict citizen indifference due to these policies' complexity and poor visibility (Chen and Weir 2009; Hackey 2000; Jacobs 2014b; Meier 1991; Mettler 2010).3
Yet, the factors that favor or discourage citizen constituencies around EFRs deserve greater study for four reasons. First, many citizens possess the capacity to reason about the health insurance arrangements they use despite their complexity (Gusmano, Schlesinger, and Thomas 2002: 762). Second, several circumstances can leave Americans with no opportunities to seek insurance other than through continuation coverage or in the nongroup market (see table 1). Yet, as discussed below, the ability of people to procure an insurance offer from either of these two settings hinges on the terms of EFRs that historically have varied across time and state lines and remain subject to change today. Third, public policy structures the nongroup market's operation and role in insurance provision (Buntin, Marquis, and Yegian 2004; Marquis et al. 2005; Swartz and Garnick 1999). The ACA contains tools to expand this market, and its importance tends to grow when job-based insurance contracts (CMS, n.d.-a) and when expansions of public programs are not politically feasible (Swartz and Garnick 2000). Fourth, if complexity or invisibility is indeed the main barrier to citizen enthusiasm for EFRs, then EFRs' potential to generate constituencies may not be fixed but may instead be responsive to “political communication” (Mettler and Soss 2004: 58) that frames and explains the content of EFRs to the public (e.g., Mettler and Soss 2004; Mettler 2010). Understanding whether even complex regulatory instruments like EFRs can generate feedback matters since citizens' assessments of specific health policy tools inform their preferences on the direction of health policy more broadly (Barabas 2009; Lerman and McCabe 2017).
This article synthesizes literature from several disciplines to examine factors that may generate supportive public constituencies for EFRs (discussed in the second section), as well as factors that may impede the cultivation of such constituencies (discussed in the third section). My bottom-line conclusion is that, despite their importance, EFRs will generally fail to generate citizen constituencies for numerous reasons, including not only their complexity and invisibility but also, and more stubbornly, their inability to generate substantial resource effects unless they are linked to other policies that control medical prices and make insurance affordable. In other words, EFRs are a necessary but insufficient policy instrument for constituency generation.
This dynamic is an example of a more general phenomenon that I call reliance on policy intersection (discussed in the third and fourth sections). Reliance on policy intersection, which relates to Arnold's notion of “multi-stage policy” (1990: 20), has not been integrated into policy feedback research, but it clearly can disrupt the flow of political resources and thus lead to weak feedback generation. Literature suggests that resource conveyance failures will avert citizen constituencies around a policy even if politicians communicate about the importance of that policy to citizens. The take-away lesson for health policy is that, even if lawmakers enact and “sell” policies that broaden eligibility for nongroup coverage—as they have in the ACA—constituency formation may be disappointing, creating relatively little pressure from citizens to strengthen and deepen those provisions over time.
EFRs' Constituency Potential
Three factors suggest that citizen constituencies—at least of a latent sort (Arnold 1990)—could be stimulated to protect and defend EFRs. First, some people lack ways to obtain insurance other than through continuation coverage or the nongroup market, and EFRs shape eligibility for these insurance pathways. Second, EFRs can have a nearly universal impact since they establish basic insurance rules of the system. Third, while EFRs may indeed be complex and less visible, elites can explain the importance of EFRs to people's access to health care and thereby render their operation more salient to citizens. This section discusses these considerations.
EFRs Can Ensure or Foreclose an Insurance Offer
First, the categories of people identified in table 1 may have no ways to obtain insurance other than through continuation or nongroup coverage. Only subsets of Americans are eligible for Medicare, Medicaid, and military coverage. Private group coverage is typically job based and most consistently available to full-time workers in large organizations (Gruber and Madrian 2002: 15; Long et al. 2016).
While EFRs are thus critically important to Americans' health care, their terms have historically varied across time and state lines. These terms can determine whether people who lack or lose other coverage options are eligible for either continuation or nongroup coverage. COBRA requires that employers with 20 or more workers who provide health plans allow qualifying employees and their dependents to buy their existing coverage for 18–36 months after job exit (DOL 2016). Only 58% of workers aged 19–64 would have been COBRA eligible in 2010 had they lost job-based coverage (Doty et al. 2011: 4, 7). State continuation coverage laws varied before COBRA (Gruber and Madrian 2002: 11) and vary today (Moriya and Simon 2016: 320).
For people in table 1 ineligible for continuation coverage, nongroup coverage may be the only possibility.4 By one estimate, it was the only option for over 20% of Americans, pre-ACA (Buntin, Marquis, and Yegian 2004). Yet, guaranteed issue determines who can procure an offer. The ACA has required guaranteed issue nationwide for nongrandfathered nongroup plans since 2014, but only six states used it in their individual markets in 2012 (KFF 2012). Before the ACA, insurers may have denied coverage offers to 8–18% of nongroup applicants (Buntin, Marquis, and Yegian 2004) and 29% of nongroup applicants aged 60–64 (Claxton et al. 2016).
EFRs Impact Many People
A second reason EFRs could generate citizen constituencies is that they are conceivably “proximate” (e.g., Soss and Schram 2007) to all Americans at some point in their lives. Proximity means that people “believe that they might benefit” from a given policy “someday” (Campbell 2012: 340). The shares of people using continuation and nongroup coverage at a given time are modest. Roughly 6 million Americans used COBRA in 2006 (Bovbjerg et al. 2010: 1). Seven percent of Americans had nongroup coverage as their primary insurance in 2016 (KFF 2016). Yet, the “potential market” for nongroup insurance in particular is larger (e.g., Buntin, Marquis, and Yegian 2004; Claxton, Levitt, and Pollitz 2017; Pauly and Percy 2000: 14). Of the 91.2% of Americans insured at some point in 2016, 16.2% used direct-purchase insurance for at least part of the year, similar to the 16.7% of insured Americans who used Medicare during part of 2016 (all percentages from Barnett and Berchick 2017: 3). Second, everyone is at risk of poor health (Campbell 2014: 9; Claxton, Levitt, and Pollitz 2017). In 2018, 57% of American adults reported that they or a member of their household had a preexisting condition (KFF 2018). Further, an estimated “15 to 30 percent of people in perfectly good health today are likely to develop a pre-existing condition over the next eight years” (CMS, n.d.-a: n.p.). While not all illnesses generate a coverage denial absent guaranteed issue, at least 27% of working-age Americans have conditions that would likely render them uninsurable in the nongroup market under pre-ACA policies in most states (Claxton et al. 2016).
An open research question is whether policies with such potentially broad proximity can acquire the type of widespread support that surrounds universal policies like Medicare. One reason for skepticism is that people may underestimate their chances of incurring some types of misfortune and thus being in a position to benefit from a regulatory protection (e.g., Campbell 2011: 970; Soss and Schram 2007: 125). However, many Americans do appear to see EFRs as being relevant to their lives. In 2018, a candidate's position on “continuing protections for people with pre-existing health conditions” was the “single most important” factor to 14% of registered voters and a “very important” factor to another 52% (KFF 2018). Additionally, 75% of Americans said it was “very important” that the ACA's guaranteed issue rules remain in law (Kirzinger et al. 2018).
Poor Visibility Is Changeable
Third, “political communication” (e.g., Mettler and Soss 2004: 58) by elites may be able to help low-information citizens grasp the importance of EFRs despite their policy complexity and invisibility. To first take a step back, complexity and invisibility do indeed make it difficult for EFRs to generate citizen constituencies. In fact, the existing literature that makes this point (e.g., Chen and Weir 2009; Hackey 2000; Jacobs 2014b; Meier 1991; Mettler 2010) may understate the challenges that complexity and invisibility pose in two respects. First, as discussed in detail later, EFRs are not just what Arnold (1990: 20) calls multistage policy—meaning a policy whose value grows if and when its “first-order effects” spur “later-order” or nth-order effects. Rather, they are a version of multistage policy that Arnold implicitly suggests, which this article calls reliant multistage policy. A reliant multistage policy must intersect with other, separate policies to link key segments of its target population to any usable resources.5 This structure may hinder citizens from giving an EFR credit for resources they derive from insurance, since “traceability is nearly impossible for any [policy] effects that stem from a multitude of complex governmental actions” (Arnold: 50). Second and related, lawmakers typically lack incentives to communicate about such complex policies (97). “Building a constituency in support of . . . guaranteed issue . . . is not easy,” be it composed of citizens or other actors (Hackey 2000: 220). Thus, there are indeed obstacles to making EFRs more visible and legible to Americans—a sine qua non for a citizen constituency to develop.
That said, three considerations suggest that, with adequate political communication, an active citizen constituency for EFRs could be generated. First, citizens can in fact see EFRs at both the federal and state levels, at least in some instances. For example, between April 2010 and November 2011, majorities (58–70%) of Americans knew that the ACA “prohibit[s] insurance companies from denying coverage because of a person's medical history or health condition” (KFF 2011: “Topline,” 13). Citizens also “respond[ed] enthusiastically to the initial stages of guaranteed issue” in state reforms predating the ACA (Kirk 2000: 168), which were possibly less salient (but see Hackey 2000: 217, 221).
Second, some elite communication about EFRs' content and value has emerged following policy threats to the ACA (Lucas 2018; Pelosi 2017), and such communication appears capable of encouraging people to credit government for risk protection stemming from EFRs. For instance, in cross-national data, there is evidence that people credit government for policies that “delink benefits from the labor market” (Gingrich and Ansell 2012: 1624), which EFRs partly do. And in the United States, there is experimental evidence that “simple, clear, policy-relevant information” can improve citizens' knowledge of policies whose positive effects are otherwise difficult to trace to government (Mettler 2010: 809).
Third, Americans appear supportive of EFRs when prompted to consider their content and existence. For instance, survey questions that make EFRs visible find support for these policies to be moderately high. One survey found 47% of Americans holding a “very favorable” and 20% a “somewhat favorable” view of the ACA's guaranteed issue mandate (KFF 2011: “Topline,” 15). Protests against ACA repeal in 2017—while doubtless inspired by many considerations—also suggest that when a policy containing EFRs faces reversal, people who seemed indifferent mobilize in its defense, as Arnold (1990: 68) might predict (also KFF 2011: “Findings”). These considerations all suggest that, with improved communication, a citizen constituency for EFRs could form.
Impediments to EFR Citizen Constituencies
While elites' political communication could thus render EFRs less complex and more visible to mass publics, other barriers to the stimulation of constituency formation around EFRs are more challenging to overcome. Policies are more likely to generate such constituencies when their benefits (a) are of meaningful magnitude, (b) are visible and traceable, (c) reach concentrated beneficiaries who concur on their value, (d) last over time, and (e) lack stigma (Campbell 2012). While EFRs may at times satisfy the visibility and traceability requirements, the benefits they provide are often diffuse, temporary, and small (see Hackey 2000).
Diffuse and Divisible Beneficiaries
EFRs may fail to provide concentrated benefits for two reasons. First, nongroup users are dispersed geographically, as are continuation coverage users save for regionally concentrated layoffs. This dispersion is one example of how EFRs' benefits are diffuse (e.g., Hackey 2000: 219). Spatially dispersed beneficiaries are relatively unlikely to see or be mobilized around a policy's personal proximity (Arnold 1990: 28; Campbell 2012: 340).
Second, as Hackey (2000) emphasizes, EFRs' target population members have unequal interest in the first-order resource that EFRs provide—an assured coverage offer. Nongroup customers' needs and financial means are especially diverse (Buntin, Marquis, and Yegian 2004; Chollet 2000; Hackey 2000: 218). Hackey (2000: 218) posits that “the difficulty of obtaining coverage [offers] rivals cost as a policy concern” mainly for people in poor health; consequently, “guaranteed issue . . . will win little support from young, healthy individuals.” Nongroup insurers' risk-segmentation efforts may amplify age- and health-based differences in shoppers' views of EFRs (Chen and Weir 2009: 687–88; Claxton, Levitt, and Pollitz 2017; Kirk 2000: 169–70; Marquis et al. 2005: 394; Swartz and Garnick 1999: 181). Constituencies in support of a policy will tend to be fragile and hard to mobilize when some members of the group see a stronger stake in the policy than others (Campbell 2003: 145; Hackey 2000: 218; Oberlander 2003: 50).
Another issue is that the benefits from EFRs are necessarily temporary in many cases—continuation coverage lasts for limited periods of time by statute, and nongroup coverage is often temporary in practice. Regarding the latter, some people do rely on nongroup insurance coverage for long time spans (Claxton et al. 2013; Marquis et al. 2005; Swartz and Garnick 2000). For example, California data from 1997–2001 suggest that 63% of nongroup enrollees retain their coverage for at least 1 year and 31% for at least 3 years (Buntin, Marquis, and Yegian 2004). That said, many perceive it as mere bridge coverage until they can gain access to other insurance plans (Buntin, Marquis, and Yegian 2004; Marquis et al. 2005), and people rarely scrutinize programs they expect to use briefly (Campbell 2012: 340). Further, the nongroup market sees high turnover (Chollet 2000; Claxton et al. 2016). Churn of this type can prevent a policy from generating a robust coalition of citizen supporters (Arnold 1990: 137).
Reliant Multistage Policy
Finally, EFRs may fail to offer large enough benefits to be meaningful—for a nontrivial share of their target populations—because they rely on other, auxiliary policies that jointly determine the insurance coverage options that people actually face. EFRs are what Arnold calls “multistage policies” (1990: 20) in that their first-order effect—assurance of a coverage offer—might or might not catalyze later-order policy effects, namely, the ability of citizens to buy insurance and obtain the security and economic value that the plans confer on recipients. EFRs can be of crucial importance to citizens, but many citizens can realize their benefits only when other mechanisms are in place to make access to insurance both a meaningful and affordable opportunity. In this sense, EFRs are not just multistage policies but reliant multistage policies. If those other auxiliary health insurance policies are absent, deficient, or politically unstable, then the only resource that these citizens can derive from the EFR is the first-order effect of eligibility for an unusable coverage offer, keeping robust feedback effects from ever occurring.
To be sure, scholars have observed that EFRs are reliant multistage policies, in not so many words. They have noted that COBRA and nongroup coverage offers can be unaffordable, insufficient, or both absent tools that address insurance price and content (Collins, Gunja, and Doty 2017; Gruber 2001; Oliver 2000). Scholars have compared health insurance regulation to a “three-legged stool” (Gaba 2017; Gruber 2010) since different policies often address eligibility, cost, and quality problems (Carroll 2012). In such instances, removing one provision may destabilize the entire structure (Gaba 2017).
Yet, scholars have not examined these observations through a policy feedback lens and so have not spelled out how reliance on auxiliary policies can prevent EFRs from generating resource effects and thus a citizen constituency. Figure 1 does so by modifying the policy feedback logic model sketched by Mettler (2002: 353) that is widely used in feedback studies. Whereas in Mettler's figure an arrow directly connects “Payments, Goods, and Services” to resource effects, figure 1 distinguishes resource conveyance from resource effects, and weaves in Arnold's (1990: 47) notion of a “policy effects chain” for multistage policy, by drawing first-, second-, and third-order resources emanating from EFRs. This revision reflects ideas that already exist in the feedback literature (see Campbell 2003: 4; Mettler 2005: 32) but have not been emphasized in the logic model to date. Figure 1 incorporates the idea of reliance on policy intersection by adding price- and content-focused “support policies” to the left side of the model, alongside the EFR “root policy,” and diagramming their role in enabling second-order resources to accrue. It adds political communication (Mettler and Soss 2004) since mobilization (Campbell 2012) and information (Mettler 2010) can help facilitate the chain of events in the logic model. The take-away lesson of figure 1 is that, for some share of an EFR's target population, resource conveyance can come to a halt if vital supporting policy mechanisms that are needed to make the insurance system work perform poorly, are vulnerable to elimination, or are absent at the outset.
These situations are not theoretical but are indeed live possibilities in the contentious world of contemporary US governance. Health policies that “rely upon voluntary efforts in the private market” face an inherent risk of poor performance (Barrilleaux and Brace 2007: 664). For example, many insurers left Kentucky's nongroup market in the 1990s when that state joined rate restrictions to guaranteed issue (Kirk 2000; see also Chollet 2000: 42; Hackey 2000: 213; Pauly and Nichols 2002; but see Swartz and Garnick 1999). Price-focused policies not susceptible to such pushback—in other words, approaches not plagued by “inadequate state capacity to make policy operational” (Patashnik and Zelizer 2013: 1076)—can also fall short of expectations. For instance, the 2009 American Recovery and Reinvestment Act (ARRA) temporarily subsidized some workers' COBRA premiums at a rate of 65% but ended up helping fewer people obtain COBRA than projected (Graetz et al. 2012: 1992). This shortfall may have happened because ARRA subsidies were insufficient to mitigate COBRA's price (Bovbjerg et al. 2010: 10; Graetz et al. 2012: 1993).
The political controversy surrounding certain policies contained in the ACA that were intended to help control health care costs illustrates the inherent vulnerability of these market interventions to repeal. The ACA's individual mandate was designed to keep nongroup premiums under control for older, sicker individuals by ensuring that younger, healthier people enter risk pools (Gruber 2010). Yet, some elites and parts of the public on both the ideological right and left opposed the mandate, and lawmakers ultimately repealed it in 2017 (New York Times2009).6 The repeal may cause insurers to raise premiums 10% (Congressional Budget Office 2017; Jost 2017). To be sure, ACA subsidies may protect most marketplace shoppers from the burden of such increases (Sheils and Haught 2011). Yet, insurers view subsidies as insufficient to curb adverse selection absent an individual mandate (Kail, Quadagno, and Dixon 2009: 1357; Japsen 2012), raising questions of whether the market will remain stable over time (Hall 2012; Jost 2017). Insurance subsidies to the needy can also be “watered down” (Patashnik and Zelizer 2013: 1080) or phased out (Cantor and Monheit 2016: 793). Since policies that “lack needed institutional auxiliaries and supports” likely “fail to generate strong effects” (Patashnik and Zelizer 2009: 35), policies with “multiple ‘separable’ parts” are at risk of being undercut (Patashnik 2008: 167). Partisan discord (e.g., Jacobs 2014a) could even invite lawmakers to remove necessary institutional supports.
Many people will find offers to buy insurance provided by EFRs unusable for all practical purposes in the absence of other price-control policies. For instance, COBRA's usual take-up rate is 19–20% (Bovbjerg et al. 2010: 2; Graetz et al. 2012: 1991) and perhaps as low as 5–10% (Graetz et al. 2012: 1981, 1991). With ARRA, take-up reached 38–39% (Bovbjerg et al. 2010; Graetz et al. 2012). The difference may be an indication of the share of people who want COBRA for whom an offer is unusable absent a subsidy equaling or exceeding ARRA's (e.g., Bovbjerg et al. 2010). That gap is at least 18 points (38% vs. 20%) and may be larger; over half of COBRA takers in one study “would not have enrolled without the subsidy” (Graetz et al. 2012: 1991). Similarly, nongroup guaranteed issue may be sufficient to link at least the “40 percent of nongroup candidates . . . who are not income constrained” to insurance without support policies (Pauly and Nichols 2002: W334). Yet, “about half” of potential nongroup buyers in 2003 “had incomes below 200 percent of the federal poverty level” (Buntin, Marquis, and Yegian 2004: 83). For the very income constrained, even substantial subsidies may not make an offer affordable (Hackey 2000: 215; Oliver 2000). In 2009—when annual nongroup premiums “averaged $2,985 for single coverage and $6,328 for family plans” (America's Health Insurance Plans 2009: 2)—4% of uninsured working-age adults said they could not pay a premium of $25/month ($300/year) (KFF 2009). To use Andersen's (1995: 4) language, EFRs give income-constrained people “potential access” to coverage, but without price-mitigating policies, they do not convey “realized access,” meaning “the actual use of services.”
These considerations suggest that the toughest barrier to generating a constituency around EFRs is not policy complexity or invisibility per se but, rather, the failure of EFRs to provide citizens economic value and other directly usable resources. While material and other resources alone cannot build a constituency around a policy (Sances and Clinton 2018), public attitudes do reflect objective policy experiences (Barabas 2009; Gusmano, Schlesinger, and Thomas 2002; Lerman and McCabe 2017). People can become frustrated with policy approaches that yield disappointing outcomes (Barabas 2009; Hacker 2004: 248), though this does not always happen (Campbell 2011: 970).
Notably, this frustration is probably most likely to occur among low-income Americans, who are in most need of price-control policies to make the benefits of EFRs meaningful in their daily lives. While they may be less likely than the affluent to protest poor institutional performance (Holbein 2016), lower-income earners notice when policies implicitly promise but fail to deliver later-order resources (Mettler 2005). And while people may still think well of such a policy when it provides valuable early-order benefits (see Mettler 2005 on the G.I. Bill), they are unlikely to favor policies whose resource flows are unusable.
I have argued that a full explanation of why EFRs may lack a citizen constituency must reach beyond the complexity and poor visibility of these policies. The toughest barrier to positive feedback at the mass level is EFRs' reliance on other, auxiliary policies to connect citizens to usable resources. These other policies may be absent, inadequate, or vulnerable to removal.
My argument does have some limitations. First, my argument assumes that, when EFRs fail to link citizens to usable resources, policy disappointment outweighs any positive political communication from elites to help citizens grasp the value of these policies. Existing research offers support for this assumption. Specifically, Soss and Schram (2007: 122) indicate that people who experience tangible policy effects likely test elite assertions about that policy against those personal experiences. And, Gusmano, Schlesinger, and Thomas (2002: 761–62) offer evidence that “specific lessons about the performance of institutions in a particular domain represent the most important effect of existing policy on public attitudes,” outweighing elite messages. Such findings suggest that people who are disappointed with the resources they can draw from EFRs are likely to be unmoved by statements about EFRs' value. Nonetheless, this article does not test this hypothesis. Future research could do so; for example, survey experiments could be performed to test public opinion about EFRs in the presence and absence of policies to mitigate premium hikes and other complementary insurance market reforms. Outcome measures could include respondent support for the EFR or respondent assessments of whether a person whose insurance is regulated by an EFR would receive sufficient resources to notice them. To be sure, the external validity of survey experiment findings requires attention (Barabas and Jerit 2010). In-depth interviews can also provide insight into how people understand and react to their policy experiences (Campbell 2012: 347; Michener 2018).
Second, my argument assumes that citizens' objective policy experiences with EFRs inform their policy evaluations. This premise rests on a wealth of evidence (Barabas 2009; Gusmano, Schlesinger, and Thomas 2002; Lerman and McCabe 2017; Mettler 2005). Yet, symbolic considerations like partisanship can also lead people to self-select into different health policy experiences (Lerman, Sadin, and Trachtman 2017), to interpret policy performance differently (James and Van Ryzin 2017; McCabe 2016), and to interpret self-interest in a policy differently (see Campbell 2012: 346). Future research should thus study whether citizens who incur similar policy disappointments when using EFRs respond to those experiences in diverse ways. It is worth noting, however, that even if this is the case, such patterns would likely bolster this article's central conclusion that EFRs face steeper barriers to building citizen constituencies than has been recognized to date, since political fragmentation undermines policy constituencies (e.g., Campbell 2003).
More broadly, my analysis offers lessons for research on the politics of policy feedback and for a substantive understanding of the ACA. By building on Arnold's (1990) concept of multistage policy, and the related concept of reliance on intersecting policy mechanisms, I have outlined steps in Mettler's (2002) classic feedback model that merit scrutiny when the policies under examination cannot provide usable resources to key population groups without well-functioning policy supports. This insight can help researchers unpack variation in the emergence and mobilization of citizen constituencies across different health policies. Scholars note that feedback research should study not only policies that have generated robust clienteles but also policies that have not (Campbell 2012: 347; Patashnik and Zelizer 2013: 1075). A testable hypothesis of my argument is that a policy's potential to generate a mass constituency is inversely related to the share of its target population that can only draw usable resources from the policy when it intersects with support policies. In addition, my argument should help us predict and understand changes in the strength of a policy's public constituency over time. This is so because a policy's degree of reliance on support policies can rise or fall—not only through exogenous events but also through policy revisions that introduce policy intersections and thus new mechanisms of either reinforcement or tension.7
Finally, this article suggests that the ACA's ability to build and sustain a citizen constituency may change over time together with shifts in the operation of marketplace subsidies and other market supports and auxiliaries. Specifically, my analyses imply that elite efforts to build mass support for the ACA by highlighting the value of regulatory provisions like guaranteed issue may mobilize citizens in the short run but are unlikely to engender a robust constituency in the long run if ACA subsidies prove unstable or fail to keep pace with medical inflation.
An earlier version of this article was presented at the 2018 “JHPPL Special Issue on Health and Political Participation Conference.” Some ideas here draw from papers I prepared for earlier conferences, particularly the 2013 Midwest Political Science Association and 2015 Association for Politics and the Life Sciences meetings. For helpful suggestions, I thank an anonymous reviewer, Jennifer Costanza, Cody Drolc, Sarah Gollust, Jake Haselswerdt, Eric Patashnik, and fellow participants in the conference.
Wolfe and Serakos (2015) and Schoen et al. (2005) particularly discuss these resources. While insurance is not always sufficient (Andersen 1995; Mechanic 1999) or necessary (Herring 2005) for health care, experimental (Finkelstein et al. 2012: 1057) and observational studies suggest the insured receive care more ably (IOM 2002a), see better health (IOM 2002a), and incur less financial loss in illness (IOM 2002b) than the uninsured. Insurance not tied to one's job relates positively to early retirement and other job mobility measures (Gruber and Madrian 2002).
On health, see Burden et al. 2017, Ojeda 2015, and Pacheco and Fletcher 2015; but for qualifications, see Gollust and Rahn 2015 and Pacheco 2018. On finances, see Brady, Verba, and Schlozman 1995. On job flexibility, see Newman, Johnson, and Lown 2014; but for caveats, see Schur and Kruse 2000.
Meier (1991: 710) posits that insurance regulation is “complex but not salient” and notes that such policies see “only modest input from . . . consumers.” Hackey (2000: 219) notes that governance of the nongroup market is “low visibility” compared to other aspects of health policy. Chen and Weir (2009: 688) write, “the process of preserving broader risk pools does not feature self-reinforcing dynamics.” This observation suggests that citizen political behavior in support of EFRs is unlikely. Jacobs (2014b: 637) hypothesizes that “the ACA's ‘submerged’ elements (such as insurance regulations) may fail to register with Americans as tangible benefits.” And while EFRs may not precisely fit Mettler's (2010: 803) definition of submerged policy since they do not provide incentives or subsidies to individuals, Mettler identifies regulations on job-based private insurance as submerged (804). Her assessment that citizens are unaware of “what is at stake” in submerged policies and lack “facts about how such policies function” (809) likely applies to EFRs insofar as citizens have incomplete insurance knowledge (Parragh, Okrent, and Mehryar 2015).
COBRA-eligible persons can seek nongroup coverage (DOL 2016), but not all nongroup shoppers are eligible for continuation coverage. Those who are not may systematically differ from those who are (Moriya and Simon 2016: 339).
Arnold (1990: 20) appears to leave open the possibility that not every multistage policy is reliant, in that the example of multistage policy he provides is federal aid to school districts. This policy's “first-order effect” is usable: higher district education budgets (20). That said, Arnold appears to suggest that some multistage policies must intersect with other policies to deliver usable benefits in noting that some policy effects “are the result of multiple governmental decisions” (48). Conceivably, these effects could include first-order effects.
For a discussion of how policy design can predispose regulations like the individual mandate to face backlash, see Vogel 1990: 462–63.
Regarding exogenous events, for instance, EFRs' reliance on price-mitigating policies would fall if premiums fall.