Cost control choices for the Affordable Care Act are a logical topic for this special issue because of both the policy and political importance of the choices and what they illustrate about the use of policy analysis in policy making. This article argues that the common trope that research-based evidence is ignored because of “politics” misunderstands the patterns of policy analysis. Politicians did listen to interest groups, voters, and their own attitudes, but the visibility and credentials of different policy analyses were also important. The advice structure and validation of ideas for budgetary decisions also differed distinctly from that for choices about how to control overall health care spending. This article illustrates patterns of influence with regard to three types of cost controls: price regulation, limits on insurance, and an aspirational agenda of system reengineering measures such as accountable care organizations, pay for performance, and cost-effectiveness analysis. Even the meaning of evidence for some measures (e.g., reengineering) was different from that for others (e.g., price regulation). Being heavily promoted by the most prestigious policy experts did not mean ideas were either well supported by evidence or politically safe. The question may be more why politicians listened to some analysts than why they do not listen to analysis.
What explains US government policies to control health care spending? More specifically, what explains the spending control policies included in or excluded from the 2010 health care legislation commonly known as the Affordable Care Act (ACA)? And what may that tell us about how policy analysis can and should be used in making policy?
Decisions about spending control require information about effects on both spending and other valued outcomes, such as quality of and access to care. Yet cost controls also have huge political stakes, because of both ideological disagreements about the roles of government and markets and efforts by provider interests to maintain and increase their incomes. Therefore, some might expect this to be a story of how “research evidence,” as Mark Peterson calls it, is ignored because of politics. That is a common theme in discussions of policy analysis and policy making. Eric Patashnik and Justin Peck (2017: 115, 129), for instance, cite David Mayhew about the “battle between the popular and the high-minded” and the need to “reconcile the pressures of democratic politics with the dictates of policy analysis.” As Jonathan Oberlander describes in his article on the Independent Payment Advisory Board in this special issue, politicians and journalists are as likely as disappointed academics to take this view.
Yet whether the evidence generated by credentialed experts (in Peterson's words) is influential is not the right question. Politicians not only “power” but “puzzle”—figure out what to do to accomplish their policy goals (Heclo 1974). Policy makers have to get their ideas from somewhere, and policy communities (Kingdon 2002) in which analysts play a prominent part generate both problem definitions and policy alternatives. The right questions are which analysis is influential, and why.
In this article I emphasize the contentious and contended nature of policy analysis among “experts” as well as among politicians.1 That involves the following observations:
The enterprise of policy analysis is marked by disagreements over both core assumptions and evidence. A major reason policy makers do not listen to some policy analysts is that they are listening to other analysts. Analysts compete for attention, craft proposals to attract support or defuse opposition, and seek to build and buffer analytical institutions (Bimber 1996; Oliver 1993).
Traditional political factors—such as formal decision-making processes, partisan conflict, ideology, and interest group power—shape which analyses become incorporated into policy. Yet they alone do not explain why policies are adopted. As policy makers puzzle about how to pursue their interests, analyses affect how they define those interests (Wolbrecht and Hartney 2014).
In any established policy arena, experts, politicians, journalists, people whose incomes are earned through that activity, and other advocates tend to group into different advocacy coalitions (Sabatier and Weible 2007). For health care cost control there were three main approaches, associated with different analytic approaches, organized interests, and sections of the two main political parties.
Two different kinds of research evidence or credible knowledge base (Peterson's terms) pass for policy analysis. One consists of direct evidence: examples or plausible analogies from experience, in which something like the policy had somewhat measurable effects. For instance, there is significant evidence about the effects of cost sharing from which to base arguments about a cost-sharing proposal. The other kind of analysis might be called theoretical fit. It begins with arguments about the causes of conditions. Since conditions (especially spending) may have multiple causes, these arguments tend to be based on a mix of research evidence and disciplinary leanings (e.g., economists look for deviations from market logic, and public health professionals look for underlying health inequities). Then proposals are judged by whether they seem to address the definition of the cause. If so, that makes them credible—regardless of whether there is evidence that they will work. A clear example of this dynamic, as shown by Sandra Tanenbaum (2009) and elaborated below, is “paying for performance” (P4P).
Budget analysts such as the staff of the Congressional Budget Office (CBO), if following their own professional norms, insist on direct evidence. They are not willing to quantify savings from theory alone. So, for example, CBO would project savings from cost sharing far more readily than from P4P. Many eminent academics, however, follow the theoretical fit approach.
Policy makers use policy-analytic knowledge (or beliefs) to assess how to accomplish their policy goals and maximize the chance they will get credit for policies rather than blame.2 But they also worry about which constituencies will be directly hurt or helped by a policy. These distributive effects, such as who will get paid to build an aircraft carrier or will receive a tax cut, may be much clearer than the policy effects, such as whether the ship will improve national security or the tax cut will increase economic growth. And policy makers have to worry about ordinary knowledge—how citizens tend to think about issues—because policies have to be explained to voters (Peterson 1995; Schick 1991). So in the competition among policy analyses, those that align best with distributive and ordinary knowledge factors will be strongly favored.
The fact that experts' perceptions may differ from those of the average voter (the popular view) does not mean they are right or “high-minded.” In some cases experts and voters care about different outcomes. In this case, experts cared about total national health spending but voters cared about whether they could afford the care they believed they needed. Neither democratic theory nor political wisdom suggests policy makers should choose experts' values.
All seven of these observations help explain the cost control choices made in the ACA and their consequences. Analysis must begin, however, with the motives or goals for cost control. Different goals led to different politics and different roles for policy analysis.
The Outcomes to Be Explained
Drafting the ACA involved two cost control challenges. The first was to limit the federal budget effect of the new coverage: the law could not increase the federal deficit over any of the periods used in budget process scorekeeping. So the costs of new subsidies for coverage had to be offset by savings from preexisting government programs (mainly Medicare) and by new revenues.
The second challenge was to write legislation that would be perceived as reducing future costs of health care (most directly through premiums for health insurance) for voters who did not benefit from the new coverage: personal costs for the already insured. During his campaign, President Obama promised that health care reform would reduce premiums for a family, when fully implemented, by $2,500 per year from what they otherwise would have been. Policy makers knew that political benefits from reform depended on reassuring voters worried that rising costs would make their existing coverage unaffordable (Nather 2009; Wayne and Armstrong 2009). The uninsured were on average less likely to vote, and polls showed that controlling costs of existing insurance was more popular than expanding coverage.3
There is some argument, of course, about what the law accomplished. Not surprisingly, conservatives argued the expansion would increase budget deficits (Holtz-Eakin and Ramlet 2010). Equally unsurprising, representatives of the Obama administration insisted it reduced premiums (relative to previous trends) for the already insured (Cutler 2010; Furman 2015; Obama 2016).
Both projections at the time and the balance of evidence since, however, indicate the ACA hit its budget goals but did little to improve affordability for the already insured. CBO (2011: 2) estimated the law paid for just under $800 billion in net costs of the insurance expansion with just under $500 billion in savings in other programs and just over $400 billion in new revenues. These estimates were mostly accepted by more neutral observers, such as the fact checkers at Politifact (Farley 2011; Holan 2012), and continued to seem reliable (for a good discussion, see Elmendorf 2014).
In contrast, CBO projected very little reduction in premiums for most private insurance.4 Some distinguished health economists claimed the legislation would “bend the cost curve,” but they could not base this on much direct research evidence. A Commonwealth Fund report even said as much, arguing the savings estimates by CBO and the Centers for Medicare and Medicaid Services Office of the Actuary relied too much on large and systematic studies. It relied on a “less formal, but no less important” anecdotal literature—declaring it was “imperative to cast a wider net than traditional evidence standards” (Cutler, Davis, and Stremikis 2009: 10).
Presented with a call to rely on anecdotes from one side, and even less credible arguments from conservatives that the ACA would make most private insurance a worse value, a typical journalistic attempt to be fair at the end of 2011 concluded there was “little to back up the president's $2500—or even $2000—savings claim” and that “what happens to health care spending and premiums in the future is an educated guessing game” (FactCheck.org2011).
Although insurance premium increases slowed substantially, this was weakly related, if at all, to the ACA. The early stages were strongly related to lower incomes and employment due to the recession (McMorrow and Holahan 2016; Goldsmith 2015; Kellogg Insight 2016). All health care payers benefited from a dramatic slowdown in the number of new drugs hitting the market, combined with a “patent cliff” in which a number of expensive drugs went off patent, allowing substitution by much cheaper generics (McMorrow and Holahan 2016; CBO 2014). Over time, increases in deductibles reduced enrollees' consumption of services (Claxton, Levitt, and Long 2016; McMorrow and Holahan 2016) but, as I discuss below, that did not look like cost control to the average voter.
In short, the ACA did not meet the political goal of reassuring most voters about the affordability of their coverage. Even the law's supporters among the public did not say making care more “affordable” was a significant reason (Kaiser Family Foundation 2010a). The ACA was not popular, and its effects in the 2010 election favored the Republicans (Kaiser Family Foundation 2010b; Konisky and Richardson 2012; Nyhan et al. 2012). Although this surely was related to partisan lies about the law, the fact that credible neutral sources concluded the law would not accomplish what voters most wanted could not have helped the Democrats.
Why did the Democrats pass legislation that would meet such skepticism about accomplishing what the voters wanted? Liberals who wanted stronger controls clearly chose to expand coverage in spite of their disappointment. Yet it appears that many drafters thought they had done more. For example, Title III, “Improving the Quality and Efficiency of Health Care,” and other sections of the law (such as creating the Patient Centered Outcomes Research Institute and some workforce policies) followed advice that was strongly promoted by expert policy analysts.5
Advocacy Coalitions and Competing Policy Analyses
The concept of a policy community presumes ongoing relationships among holders of formal authority, their staffs, and experts. Problem definitions (especially documentation of conditions) and policy alternatives (usually called “solutions”) are mainly generated within these communities. The Advocacy Coalition Framework (ACF) provides a useful way to think about divisions and relationships within these communities.6
Policy communities normally are divided into stable coalitions representing coherent combinations of values, analytic perspectives, and interests. A coalition's members largely share “deep core beliefs” such as “the relative priority” of values such as “liberty and equality” or “the proper role of government vs. markets in general.” These beliefs are formed early in life and people rarely change them. Members within coalitions also share “policy core beliefs” such as “the relative seriousness and causes of policy problems in the subsystem” (Sabatier and Weible 2007: 194, 195). These change a bit more easily, partly because they might be changed by perturbations within the system itself (such as large increases in the number of people without insurance, or a lot of white people dying of opioid overdoses). People tend, however, to defend both their deep core and policy core beliefs against contrary evidence.
Opposing coalitions often include members divided by material interests, such as labor versus business, or environmentalists versus land developers. But their attitudes still will be shaped by broad values or, in the case of experts, disciplinary training. Associations between disciplinary training and views will not be absolute; there surely are public health professionals who do not favor universal health insurance, and many microeconomists are willing to interfere with markets to create it. Nevertheless, in health policy as in all others, policy core beliefs tend to be related to professional training and experience.
Analysis then can influence policy makers' decisions in various ways, but three paths seem especially important. First, in rare cases gathering specific information or making specific judgments may be delegated directly to policy analytic institutions, with their premises then largely accepted by policy makers. CBO's role in assessing budgetary impacts is a nearly uniquely strong example.7
More frequently, individual policy makers choose to take advice from some sources and not from others. The question then is what validates the influential advisers. One answer is familiarity that builds credibility. The Medicare Payment Advisory Commission (MedPAC) appears to have had this kind of credibility with the ACA's drafters, and its judgments about many details were incorporated into the law.
Third, some ideas are promoted widely in a policy community, become familiar to policy makers and journalists who spend time in that community, and so become part of what “everybody” (or, sometimes, “our side”) believes. In Carol Weiss's (1977: 531) interpretation, analysis generates “ideas, information, and orientations to the world”—what she called the “enlightenment function.” Ideas may be associated with specific scholars but become a core policy belief for many.
These processes may reinforce one another; for example, a congressional agency may be influenced by a common trope within the policy community. Yet formal advisory bodies also may disagree, siding with different advocacy coalitions. In the case of the ACA, MedPAC strongly supported the coalition arguing for a set of delivery and payment reform ideas that I call the “aspirational agenda,” while CBO was more skeptical.
One advocacy coalition, including many liberal Democratic politicians, allied pronational health insurance interest groups, and some health policy scholars, supported regulating prices to control total spending. Their evidence included comparisons of other countries to the United States and Medicare to private insurance (Anderson et al. 2002; Angrisano et al. 2007; Ginsburg 2008; White 2007, 2009).8 Some members of this coalition believed price constraint required a single payer, but others emphasized coordinating multiple payers in an all-payer price-setting mechanism.
The regulatory coalition was weakened both by opposition from the medical industry and by having only modest support from the experts whose credentials would seem most relevant to policy makers: economists. Although a few have argued for all-payer regulation, that approach violates core beliefs in markets and against price controls. Some leading scholars oppose it so strongly that they insist it is not real efficiency (Aaron and Schwartz 2005; Garber and Skinner 2008). Discomfort with the idea was nicely illustrated by Joseph Newhouse in an article on potential effects of the ACA. Newhouse (2010: 1723) concluded the law's system-wide cost controls were so weak that, “despite all the substantive and political problems of price-setting, some sort of all-payer regulatory regime may be the only feasible alternative.” But he does not appear to have done more to advocate that approach.9
In spite of this skepticism, price restraint had been the major way that public programs limited spending in the past. For legislation to meet fiscal targets under congressional rules, CBO must validate savings. CBO had experience estimating and would “score” (give credit for) savings from price policies. Medicare involves many highly technical issues on which expert advice is needed. MedPAC was created to provide such analyses.
Therefore, the Medicare savings in the ACA were mostly recommendations by MedPAC that CBO would score. MedPAC's authority is cited dozens of times in reports on the bills (e.g., Committee on Energy and Commerce 2009; Committee on Ways and Means 2009). The Committee on Ways and Means, for instance, cited MedPAC in explaining payment update and therapy payment changes for skilled nursing facilities (sections 1101 and 1111), removing Part B drugs from calculation of the Sustainable Growth Rate formula (section 1121), changing assumptions about utilization rates of equipment so as to cut payments for imaging services (section 1147), penalties for preventable readmissions (section 1151), and reducing overpayments to Medicare Advantage plans (section 1158).
But the ACA excluded proposals to use price regulation to reduce costs either in the system as a whole or for the new individual insurance market. Liberal advocates pushed for a “public option”—that health insurance exchanges offer one plan sponsored by the government and linked to Medicare. Its supporters said the public option would be a better value than other insurance because it could pay fees based on, though perhaps modestly higher than, those paid by Medicare (Hacker 2009; Holahan 2009). It usually received positive responses in opinion polling (e.g., Blendon and Benson 2009), yet it received little support from key Senate leaders and the Obama administration.
The weight of expert advocacy overwhelmingly supported alternative approaches. Disinterest in rate regulation has persisted even as prices have received more attention in recent years.10 These attitudes made it easier to rationalize conceding to the primary cause of the public option's defeat: interest group power. A senior congressional aide stated flatly that, “you just can't do it politically. Every provider group would be 4-square against the bill. That's not an assumption, that's a known. That's Armageddon for them.” Similarly, asked why his “liberal” think tank did not push the public option, a source explained that, “our view was that it wouldn't do much good,” because an option paying close to Medicare rates “didn't seem likely to be in the cards.”11 The Obama administration chose instead to cut modest deals with pharmaceutical and medical interests (Oberlander 2010; Cohn 2010). As I describe below, this strategy also reflected the perspectives of the swing group of legislators: more conservative Democrats.
Jacob Hacker (2010: 865) summarized the results: the administration “consistently acted as if the crucial swing votes in Congress depended not on wavering citizens, but on organized interests with the greatest ability to shape the positions of congressional moderates within the Democratic Party.” He added that these “up-front concessions . . . limited the law's ability to deliver tangible benefits to the middle class and largely took off the table tools of cost control used in other nations, such as provider rate-setting and government negotiations for lower drug prices.”
A conservative advocacy coalition believes the best way to reduce the costs of health care is to reduce insurance coverage, such as by reducing government subsidies or encouraging health savings accounts instead (Republican National Committee 2008; Buchmueller et al. 2008).12 People with more “skin in the game” will be careful to consume only necessary care. This view had become deeply entrenched among Republican politicians and think tanks (Jost 2007), but it was represented in the ACA due to the influence of economic analyses on more centrist Democrats.
Some, but not all, health economists emphasize that moral hazard created by excess insurance causes people to purchase more than they would in a perfect market (Stone 2011). This analysis is rejected by traditional advocates of national health insurance who think in terms of rights and needs (Melhado 1998). But the health exchanges were set up to encourage less generous insurance. The basic subsidies were based on the cost of a “silver” plan, which would cover about 70 percent of “essential health benefits.” So it intentionally encouraged less coverage than in typical large-employer plans, which on average covered about 84 percent of an essential package (Davis 2010).13
The high cost sharing was in part a trade-off based on budget constraints: better coverage would have spread the available funds over fewer new enrollees. Sherry Glied and Erin Miller (2015) argue that policy research about how cost sharing can make insurance less useful for poorer people was acknowledged by a further provision: the ACA included extra cost-sharing subsidies for those with incomes below 250 percent of the federal poverty level. Yet the fact that policy makers would consider such a low standard remains striking. It should have been obvious that high cost sharing would cause future dissatisfaction with the coverage (Altman 2017; Hamel et al. 2016).
Worries about excessive insurance were incorporated even more directly through the “Cadillac tax.” This is an excise tax of 40 percent on premiums for insurance that exceeds a standard for cost, with cost viewed as a marker of excessive benefits. Here the division between economists and other experts was clear. The former enthusiastically supported the tax (Gruber 2009; Rampell 2009). Others, especially insurance market experts (Gabel et al. 2010), criticized it. They argued that more expensive plans were normally more expensive because they were in places with higher than average costs or had sicker than average members. Therefore, the tax would require lesser benefits for groups with greater need—a reverse risk adjustment (Jost and White 2010; Lemieux and Moutray 2016).
Larry Levitt (2015) observed that, aside from economists and budget hawks, “pretty much everyone else . . . seems to detest” the tax. Business interests were no exception, as shown in congressional testimony (American Benefits Council 2009; ERISA Industry Committee 2009; National Business Group on Health 2009). In Glied and Miller's (2015: 389) words, it had “no political constituency whatsoever—not unions, not business, not conservative taxpayers, not liberal taxpayers. It was a victory only for health economists.”
So why did it happen at all? One reason was that it offered savings to help pay for the coverage expansion. CBO scored budgetary savings because money that employers did not spend on insurance would be taxed either as individual or business income. Jonathan Cohn (2014) argued that Obama administration officials who hoped that the aspirational agenda described below would generate sufficient savings accepted the Cadillac tax in part because CBO would not score those other savings. I describe below how the tax also fitted the distributive interests of the key swing legislators: conservative Democrats.
Yet at least some Democratic policy makers echoed the economic arguments that the excise tax was both fair and a good way to reduce total spending. Asked why it happened, a senior legislative aide responded:
Because it has been the wet dream of health economists for thirty years. Every economist across the board agrees. And then when you look at it as the single largest tax exclusion, and that it is growing like crazy, and at its regressive nature, and you see that if you cut it in half it would pay for everything that's in the bill. . . . And then a lot of Republicans seemed to like it, so if you were going to try to get Republicans, it seemed that it would likely be needed. And [Senator Max] Baucus was for it, and he is chair of Finance.
Baucus's own remarks at one hearing could have been made by many economists: “We should also look at the current tax treatment of health care . . . the current tax exclusion is not perfect. It is regressive. It often leads people to buy more health coverage than they need. We should look at ways to modify the current tax exclusion so that it provides the right incentives” (Committee on Finance 2009a).
The regressive nature of existing tax policy was the kind of thing “everybody knows”—an example of Weiss's “enlightenment”—although it may well not be true (EBRI 1992; Field and Shapiro 1993; Schoen et al. 2009; White 2017; for further explanation, see White 2017). After enactment, the tax continued to be strongly endorsed by members of the Obama administration and economists, who argued it was a crucial tool to reduce total health care spending (Aaron et al. 2015; Furman 2015). But it was the prime example in the law of experts defining an issue in a way that made little sense to ordinary voters. If it reduced total spending, that was very likely to be by making care less “affordable” to the average voter, through higher cost sharing (Altman 2015; Lemieux and Moutray 2016; Levitt 2015).
In the Shadow of “Enlightenment”: The Aspirational Agenda
Discussions of system-wide cost control were dominated by a third approach. I call it the “aspirational agenda” because it is based more on hope than on experience. Its major theme was that costs were high because of unnecessary care, which could be reduced by a mix of spending on prevention (including getting people to take better care of themselves), avoiding medical errors and improving communication among providers through electronic health records and health information technology; creating new organizations to coordinate care, such as accountable care organizations (ACOs) and “medical homes”; improving knowledge about cost-effective treatments through more cost-effectiveness analysis and evidence-based medicine; and most especially “paying for performance” (P4P) rather than simply rewarding physicians and other providers with extra fees for each extra service.
These views required dramatic reengineering of health care's financing and organization. They became conventional wisdom among central policy makers because of broad promotion from credentialed experts. They also were based far more on arguments of theoretical fit than on direct evidence.
We can trace some of the process both of communicating and of receiving this analysis. The premise about excess and unnecessary care was based especially on the research of John Wennberg, Elliot Fisher, and colleagues through the Dartmouth Atlas of Medicine (e.g., Dartmouth Institute for Health Policy and Clinical Practice 2008). They showed large geographic variations in spending within Medicare, driven by variation in volume of services, with no better outcomes in the areas with higher volume. Health Affairs (2007) demonstrated the prominence of this work by selecting Wennberg as the “Most Important Health Policy Researcher of the Past Twenty-Five Years.” The Dartmouth scholars summarized their argument with a powerful slogan: 30 percent of spending was for unnecessary care. In politics this kind of seemingly precise number (such as “1 in 9” women will have breast cancer) is powerful, even if the basis for the number is obscure.14
If a lot of care was unnecessary, the logical response was P4P, not pay for volume. MedPAC's chair illustrated theoretical fit logic when he said that evidence showed costs were too high and quality too poor; there was evidence that providers responded to economic incentives, and so “it does not seem like much of a leap to conclude that P4P is a step in the right direction” (Tanenbaum 2009: 729).
The aspirational agenda measures in the ACA were the next stage in a long pursuit of that “right direction.” Tanenbaum (2009) documented the hope and hype behind P4P in this journal. In 2003 fifteen leading experts published an open letter calling for P4P to become a “top national priority.” Congress in the 2003 Medicare Modernization Act directed the Institute of Medicine (IOM 2007: 2) to “identify and prioritize options for aligning performance with payment in the Medicare program.” The IOM's response strongly endorsed P4P even as it acknowledged that “literature evaluating the effectiveness of pay for performance consist[ed] of fewer than 20 studies, yielding mixed conclusions on overall impact” (3) and that “pay for performance will not necessarily reduce the cost of care” (3). Such uncertainty only meant, to IOM, that the secretary of the US Department of Health and Human Services “should implement pay for performance in Medicare using a phased approach as a stimulus to foster comprehensive and systemwide improvements in the quality of health care” (6). MedPAC continually promoted P4P. “In previous reports,” as the commission summarized in 2010, it had “described the need for Medicare to move away from payment policies that encourage service volume and are indifferent to quality and toward policies that promote better value for Medicare and its beneficiaries” (MedPAC 2010a: xi; one such report is MedPAC 2008).
This broad support did not make the analysis correct. Strong critiques of the 30 percent number were available at the time (Zuckerman et al. 2010; see also sources cited in White 2011b). It has become less credible since.15 As CBO (2008) reported, direct evidence that any of the aspirational proposals would work was quite scarce.
Yet policy makers continually repeated the claims of excess volume caused by flawed incentives, citing the experts. The House Committee on Ways and Means (2009: 343) declared that, “according to MedPAC, the fee-for-service payment system encourages volume growth and fails to encourage care coordination delivered across an episode of care.” The Senate Committee on Finance (2009b: 3) reported that “rewarding providers that furnish better quality care, coordinate care, and use resources more judiciously could reduce costs and, most importantly, better meet the health care needs of millions more American patients.” Key Obama appointees, especially Office of Management and Budget director Peter Orszag and Orszag's adviser Ezekiel Emanuel, promoted these views (e.g., Orszag and Ellis 2007; Emanuel 2008).
When Orszag testified about the potential 30 percent savings before the Senate Finance Committee on March 10, 2009, “around the dais, lawmakers from both parties nodded” (Armstrong and Wayne 2009). The Senate Committee on the Budget (2009: 23), reporting on its draft budget resolution, proclaimed “widespread agreement that Americans are not getting good value for the money we are spending on health care. According to work by the Dartmouth Atlas Project, nearly 30 percent of total spending in our health care system is wasteful and does nothing to improve health outcomes.” Business groups cited the figure in congressional testimony (American Benefits Council 2009; ERISA Industry Committee 2009; National Business Group on Health 2009).
The ACA therefore included a wide range of reengineering measures (Zuckerman 2010), especially promotion of ACOs, that supporters (e.g., Cutler 2010) asserted would bend the cost curve. Obama endorsed the approach strongly at the time and continued to claim it would increase the system's value through the end of his administration (Nather 2009; Obama 2016). As this article was being written, direct supporting evidence remained weak but aspiration based on theory remained strong. Len Nichols and colleagues (2017) illustrated the pattern neatly. They reported that, in spite of the “bipartisan consensus” on payment reform, recent “evaluations, reviews, and published perspectives have cast doubt on the promise and spending reduction potential of care-coordination initiatives, shared savings accountable care organizations (ACOs), patient-centered medical homes, and bundled payments in particular.” But, they hopefully added, “it is important to avoid being overly discouraged in the face of the mixed results we have seen so far.”
We should not be surprised that direct evidence that proposals would work was not so important—except for budget scorekeeping. Kent Weaver (2000: 152–53) has described how a lack of evidence about policies to discourage teen pregnancy made them a preferred option for welfare reform because, unlike other measures, they had not failed yet. Bruce Vladeck (1999) argued that much of health policy debate essentially claims unicorns are prettier than horses, as experts promote their clever but rather imaginary solutions. At any point, believers in an approach can blame failures on the need for further development, as in the MedPAC and IOM reports on P4P (or, for that matter, communism). The ACF tells us that experts, like everyone else, protect their core beliefs against contrary evidence.
Party Politics and Pivotal Decision Makers
Yet evidence can matter in ways analysts do not intend. Arguably the most puzzling cost control choices were adoption of the Cadillac tax and the combination of price regulation inside and none outside Medicare. They are easier to understand if we remember that conservative Democrats were the key swing legislators who had to be satisfied and consider how they interpreted research in distributive terms.
The lower costs in rural areas, as shown by the variations literature and other data, meant that premiums were especially likely to be below the thresholds for the Cadillac tax. So rural legislators like Senator Baucus did not have to worry much about backlash from their supporters.
The reports that Medicare spending per capita was much lower, without worse health outcomes, in many rural areas led their representatives to conclude, in the words of Representative Earl Pomeroy (D-ND), that their providers were being “underpaid by Medicare” (Pear and Herszenhorn 2009). They therefore opposed extending Medicare rates to other insurance, fearing, in Pomeroy's words, that “if we have a public plan option at Medicare rates, we will bankrupt the North Dakota health care delivery system” (C-SPAN 2009, at 8:32).16
Yet if conservative Democrats have a dominant policy value, it is “fiscal responsibility” (Blue Dog Coalition 2009, n.d.; White 2011a). They could support stronger price regulation in Medicare, while opposing it outside Medicare, because of different consequences for the budget. Because the federal government would pay only part of the costs of coverage for the newly insured, the pain to providers from extending Medicare prices to the new insurance would be greater than the savings for the government. All the savings from greater restraint within Medicare, however, would improve the government's fiscal position.
As these examples show, there is no hard line between political influences such as constituency interests and policy-analytic influence. Analysis led to conclusions about political interests.
Analyzing for an Audience
As Mark Peterson writes in this issue, researchers and policy makers tend to be engaged in an ongoing exchange, in which researchers learn what policy makers think they need and draw conclusions both about what to propose and about how to frame it. Unless they truly believed in those options, both researchers and policy makers tended to fear the Scylla of price regulation and the Charybdis of coverage reduction. The aspirational agenda was promoted as a way to avoid those perils. Cadillac tax supporters, for example, insisted insurers could meet targets not by raising cost sharing but by following that agenda and making care more efficient (Furman 2015). Francis Crosson (2011: 1250), a MedPAC member from 2004 to 2010 and now its chair, wrote that the concept of ACOs was “too vitally important to fail” because the alternative was “indiscriminate cuts to health care payments, with resulting reductions in access, service, and quality.”
The aspirational agenda fits a history of “third ways” devised by experts to bypass the known objections to existing alternatives. Managed competition was a similar health policy initiative; the classic case may be Keynesian macroeconomics as described by Ikenberry (1992).17 The strongest political recommendations for such proposals is what they are not. In the absence of direct evidence, other support from experts helps validate such proposals for policy makers who want to “do something” but “not that!”
MedPAC provides a striking example of how analysts can craft arguments to resolve political dilemmas. While promoting reengineering, MedPAC also had to recommend savings that CBO would validate and would still maintain affordable care for beneficiaries—thus, payment regulation. MedPAC's leadership and staff dealt with this tension by arguing that reengineering and regulation did not contradict but instead complemented each other. Thus, its chair during the debate, Glenn Hackbarth (2009: 2), testified that among the “tools to surmount barriers to increasing efficiency and improving quality” within the program was “creating pressure for efficiency through payment updates.” “By limiting and altering Medicare's unit prices,” MedPAC (2010b: 2) reported as the ACA was passing, “Medicare provides an impetus for providers to volunteer for experiments with new payment methods.” In short, price regulation would encourage system transformation by changing incentives for providers!
Experts, Voters, and Ordinary Knowledge
Throughout the drafting of the ACA, there was a curious disconnect between how experts discussed cost control, which was largely in terms of total spending, and how voters thought about it, which was whether they would be able to afford care. That disconnect has continued, as Drew Altman (2014) summarized in an essay titled “Health Cost Growth Is Down, or Not. It Depends Who You Ask.” After presenting statistics of the overall slowdown in health care spending and premium growth, he proclaimed that “no one disputes that the slowdown is real.” “No one,” he then added, “except the American people, who see health costs from a different perspective.” As he explained, voters saw both their own share of the premiums for employer-sponsored coverage and the deductibles within that coverage rising much more quickly than their incomes. Meanwhile, medical bills remained confusing, and there were many stories about “surprise” bills for services provided out of network (Altman 2014; Brill 2015; Rosenthal 2013–14). In short, policy-analytic knowledge did not match ordinary knowledge. Yet the voters had a point: reducing total spending is not the same as making care “more affordable” for voters if the former is mainly accomplished through higher cost sharing.
Policy analysts' tendency to think in terms of statistics and population totals, rather than individual experience, may be even more problematic for policies about reducing “unnecessary” care. When he addressed cost control in public meetings, President Obama proclaimed, “That's not rationing, it's being sensible” (quoted in Nather 2009). But what looks sensible in retrospect—that if some services within a population were unnecessary, then they could have been avoided—leads to the question of how choices can be made for specific individuals, prospectively. Whom can patients trust to decide what care is “necessary” in advance?
Trust issues were exaggerated through false claims about “death panels” (Gitterman and Scott 2011).18 But as one public opinion analysis reported, “Many experts believe that reforming health care in the United States does not just imply, but requires constraining consumer and provider choices” (Bernstein 2009: 13). This view was scary because people need to believe in their own doctors. They have to trust physicians to do things like probe them and cut them and feed them mysterious chemicals, so ordinary knowledge about medical care prioritizes the idea that your doctor knows what to do.
Therefore, even the “idea of coverage decisions being based on treatment effectiveness” received mixed responses in surveys. For example, if a policy to create an independent board to assess whether treatments should be covered was worded as saying it could reject “treatments for drugs recommended by a person's own doctor,” it was opposed by 63 percent of respondents (Bernstein 2009: 7). According to a Gallup poll in September 2009, opposition to restrictions on treatment, even if a plan expanded coverage to nearly all Americans, was 84 percent versus 15 percent, as opposed to a mere 73 percent versus 26 percent against higher taxes on the middle class (Jones 2009).
Ordinary knowledge was flawed here, as it often is. Some care is unnecessary, and some doctors overprescribe and overtreat. But the policy analysis was also flawed, because it dismissed the trust problem as a matter of how to persuade the public, failing to recognize its deep roots in the patient-physician relationship.
I do not want to suggest that ideas with better analytic evidence than the aspirational agenda are necessarily better politics. Yet policy analysis that largely ignores how both costs and medical care are experienced by individuals will not serve either the public or the politicians well.
Conclusion: Cost Control Choices and Policy Analysis
As MedPAC's reframing of regulation shows, policy choice was not simply a matter of one advocacy coalition winning. The ACA's cost controls represent all three views, in different ways and for different purposes. The answers to this article's core questions, therefore, are complex. Policy analyses influenced the choices in fundamental ways. One cannot imagine the Cadillac tax being adopted based on political interests alone. I have described how it responded to expert criticism of the tax preference for employer-sponsored insurance, a critique that was widely shared among economists of all political stripes and became what “enlightened” policy makers believed. Yet the political conjuncture in which Blue Dog support was needed to pass the bill—and they wanted measures that would help the budget but not hurt their districts much—was also important. The aspirational agenda was fundamentally a product of health policy analysts. It was pursued in spite of a lack of support from the public, in part because providers did not vocally object. Payment regulation was anathema to providers but necessary for budget savings, according to the analysts who mattered, at CBO, who wanted direct evidence.
Cost control policy choices reflected the first six observations at the beginning of this article. But that does not quite answer why some analyses gain prominence within policy communities. The following factors seem important, but further research is required. One factor is financial and other institutional support. That could be through government or foundation grants, hiring by think tanks, or positions in university departments. The development of the field of health services research is an example of perspectives being funded and institutionalized.
Another factor is whether ideas “speak the language” or fit the premises of key policy makers. The tax preference argument should have fitted well with the analytic presumptions of key Obama administration figures such as Larry Summers and Timothy Geithner.19 Yet the influence of conventional management talk or thinking may be an even better example. Measuring performance simply fits the management zeitgeist of our time.
The rise of P4P also shows how ideas can become more prominent through elite commissions or study groups. Appointments and charges to those groups favor some ideas over others. They may be helped or suppressed by dynamics of consultation within such groups, as with MedPAC's synthesis of price regulation with the aspirational agenda.
The authority of expertise vests only indirectly in the research, which policy makers often cannot judge. It resides largely in the prestige of the expert. At any time some fields, such as economics, will receive more credence than others, such as sociology (Mechanic 1990). Attention from policy makers can feed on itself, creating a bandwagon effect. Conversely, if policy makers seem uninterested, then strategic scholars may not invest in promoting ideas. Scholars have traced some of these processes both for health policy (e.g., Melhado 1998; Tanenbaum 2009) and for other policies (e.g., Weaver 2000; Ikenberry 1992).
The evidence in this article does not suggest, however, that the prominence of an analysis normally derives from strong direct evidence. That is not simply because “politics” intervenes but also because much of the research evidence used by credentialed experts focuses on theoretical fit. The policies were hypotheses about what might work based on perceptions of the nature of the problem—such as insufficient incentives for effective care, or moral hazard. Belief in the diagnosis was the key argument for the cure. This tends to lead to a cycle of hype, hope, and disappointment (Vladeck 1999). Texts on policy analysis may explain why it is dangerous to conflate a diagnosis with problem definitions, thereby defining the solution into the problem (Bardach and Patashnik 2016: 11–12). Yet that is precisely how much health policy analysis has proceeded.
The story of cost control choices for the ACA therefore suggests two cautions about how policy analysis should be used. First, policy makers need to be aware that analysts often define issues in ways that do not match voters' concerns—and that are not necessarily more appropriate. Second, they need to recognize that analysis is frequently based on disciplinary bias and hope rather than solid evidence that proposals should succeed.
Joseph White is Luxenberg Family Professor of Public Policy in the Department of Political Science at Case Western Reserve University. His research focuses especially on health care policy and politics in the United States and other advanced industrial countries, budget politics and policy in the United States, and the intersection of the health and budget realms in cost control policies, health care reform, and budgeting for health care systems.
An early version of this study was prepared for a conference “The Politics of Ideas and the Politics of Representation: The Case of Health Policy in Rich Democracies,” hosted by the Rothermere American Institute of Oxford University, November 10–11, 2010. This article benefits from participants' comments and from a few off-the-record telephone interviews with policy experts from interest groups, think tanks, or on Capitol Hill, which are quoted without attribution.
By expert I mean someone whom other people view as a source of useful information and opinions, based on that person's (perceived) knowledge. This authority of expertise derives from such factors as credentials (e.g., medical training or academic appointments) or other sources of prestige.
When I use the term knowledge that is because others use it; in practice what people think they know may not be true, so beliefs is more accurate.
Asked in September an open-ended question about the major problems facing the health care system, 38 percent of Gallup's respondents mentioned the cost or affordability of health care; only 15 percent mentioned the number of uninsured Americans (Saad 2009). In November, Democrats ranked “making sure affordable health insurance plans are available” slightly above “providing enough government financial help so as many uninsured people as possible can get health insurance.” But Independents favored affordability by a twenty-point margin, and Republicans by twenty-four (Kaiser 2009: 5).
CBO (2011) said only that premiums in the large group market would be “slightly lower,” referring to an earlier report (CBO 2009) that estimated large group premiums could eventually be 0–3 percent lower.
Another factor is that legislators who shared the cost control beliefs of the regulatory advocacy coalition described below voted for the bill because they would not sacrifice the chance to expand coverage.
For a further overview of the ACF, see Weible et al. 2011.
Although Republicans in 2017 appear to accept this norm less than Democrats in 2009–10.
Even research by strong advocates for alternative views showed the importance of prices (see Cutler, McClellan, and Newhouse 2000).
Perhaps Newhouse was seeking to live up to his own distinction between academic economists and policy entrepreneurs (see Newhouse 1995). I am arguing this is more of a continuum than a dichotomy; scholars move along the continuum in response to opportunities and their own preferences; and the norms of publication do not make it easy, in any case, for policy makers to tell the difference.
A good example of attention is Brill 2015; a particularly telling response is by the National Academy of Social Insurance (NASI 2015). One would not expect NASI to be biased against regulation, but it received little attention in the report.
Off-the-record telephone interviews with policy experts from interest groups, think tanks, or on Capitol Hill are quoted without attribution. Business interests sided with the providers, viewing it as likely to create a cost shift as providers, forced to charge less to the public option plan, would raise charges to employers (American Benefits Council 2009; ERISA Industry Committee 2009; Johnson 2009). This argument itself is a policy analysis and a highly suspect one (Reinhardt 2011).
A second approach, limiting liability for alleged medical errors, is also prominent in GOP health policy thinking. But this view has less support within the health policy community and strong opposition from trial lawyers and thus received only modest attention from the Obama administration and congressional Democrats.
In 2015 average benefits in large employer plans remained substantially higher than the silver level (Gabel et al. 2015).
One problem, noted at the time, was that Medicare variations depend more on volume than on prices because Medicare prices are set administratively. Price variation is much greater in the private market. Cooper et al. (2015) showed this quite clearly. Leading experts have also redefined the 30 percent figure in a way that emphasizes factors such as administrative complexity and pricing failures as much as or more than overtreatment (Berwick and Hackbarth 2012).
Pomeroy was especially important because, as a former state insurance commissioner and as a member of the Ways and Means Committee, he had serious expertise and credibility about health insurance reform.
Occasionally they work!
However, arguments about trade-offs between care for the very old and sick and the good of society are prominent among bioethicists and were visibly represented in the administration by Peter Orszag's adviser Ezekiel Emanuel.
My thanks to Sherry Glied for this point, in personal communication as a comment on the first draft of this article, via e-mail on May 5, 2017.