Markets and Medical Care: The United States, 1993-2005
By Joseph White
Case Western Reserve University
The Milbank Quarterly
Volume 85, Number 3, 2007
Many studies arguing for or against markets to finance medical care investigate “market-oriented” measures such as cost sharing. This article looks at the experience in the American medical marketplace over more than a decade, showing how markets function as institutions in which participants who are self-seeking, but not perfectly rational, exercise power over other participants in the market. Cost experience here was driven more by market power over prices than by management of utilization. Instead of following any logic of efficiency or equity, system transformations were driven by beliefs about investment strategies. At least in the United States’ labor and capital markets, competition has shown little ability to rationalize health care systems because its goals do not resemble those of the health care system most people want.
Keywords: Competition, markets, capital financing, managed care, health care reform.
an markets give us the kind of health care system we want? This question, posed by Tom Rice (1998a) and many others, may provoke both visceral reactions from some of this journal’s readers and methodological objections from others. I am among those who have argued that to classify policy choices as “market” versus “government” or “competitive” versus “regulatory” is likely to confuse an analysis of alternatives. In the case of cost control, these and other common labels, such as “managed care,” deflect attention from how and why policies actually work (Hacker and Marmor 1999; White 1999).
Yet the question is unavoidable because the broad ideological battle over the role of markets remains a basic dividing line and dominant theme in American health policy (for examples, see Bodenheimer 2005; Cogan, Hubbard, and Kessler 2005). In nearly all its efforts, the Bush administration appears to be guided by a belief in markets and private business, whether through “modernizing” Medicare into a system of competing private plans or replacing private insurance with Health Savings Accounts. In his third debate with Senator John Kerry (D-MA) before the 2004 presidential election, President George W. Bush revealed his views when he declared that “costs are on the rise because the consumers are not involved in the decision-making process. Most health care costs are covered by third parties. And therefore the actual user of health care is not the purchaser of health care” (Commission on Presidential Debates 2004).
Broad ideological judgments are powerful because people generally can more easily judge whether a policy is “the kind of thing we do” or “the kind of thing that I think works” than they can assess its substantive details. Hence, even though contrasts between “competition” and “regulation” or “market” and “government” frequently say little about substance, they can be expected to greatly influence policy debate in the future. Precisely because these beliefs are deeply entrenched, they are not easily changed by evidence (Sabatier and Jenkins-Smith 1999). But the importance of these beliefs means that the long-term processes by which they change have a great influence on policy in any field (Mayhew 2001; Pierson 2001). This is why we should review the evidence about how market mechanisms work in medical care, even though classifying policy instruments as being more or less market oriented can be misleading and opinions can be changed only slowly by means of evidence.
Since the early 1990s, the U.S. markets for both medical services and health insurance have undergone a period of unprecedented dynamism as the traditional subservience of insurance companies to provider interests has finally been eliminated. Paul Starr’s (1982) “coming of the corporation” was real, if perhaps a bit delayed, and with the corporation came the pursuit of market logic above all. These events provide an opportunity to view market forces in action, and the health policy community has responded with extensive documentation and analysis, especially but not solely through the Center for Studying Health System Change. This work and especially the center’s Community Tracking Studies in twelve metropolitan areas document changes in organization, results, and thinking of participants in the medical care world.1 My article, therefore, is largely (though not exclusively) an interpretation of other scholars’ evidence. Because beliefs about markets carry so much ideological baggage, it is important to show that the judgments about detail and patterns from which I build my argument are not exclusively my own, but instead are statements by analysts who may have very different political views. Therefore I will quote rather more than the editors prefer.
What will happen to health care finance and delivery if the participants in these processes more closely resemble the players in a market than they did in the past for American medical care and at present for health care systems in other countries?
The evidence suggests that some of markets’ basic attributes— particularly how investors allocate capital—have been incompatible with the pursuit of a more equitable and efficient health care system. It also illustrates dynamics of market behavior, particularly herd behavior in response to compelling stories, which are not part of the standard economic explanations of how competition should work. For both these reasons, an interpretation of patterns in the United States since 1993 can shed new light on the prospect that markets could give us the kind of health care system we seem to want.
Methods and Plan of the Argument
“How markets work” and “how well markets work” are such general questions that they cannot be reduced to testable hypotheses. But they can be analyzed both logically and rigorously. In this article, I use the available mainstream evidence to make comparisons over time and between two forms of health care finance. The time series is mainly from 1993 through 2005, though I also look further back when that would be useful. National health expenditure and insurance access data provide macro trends, and the Health System Change studies and some others offer an extensive account of dynamics within the private insurance market. The two forms of insurance are the private, largely employment-based insurance that is available to most Americans under the age of sixty-five, and the Medicare program that is the main source of coverage for Americans who are aged sixty-five and older or are disabled. Medicare offers another perspective from both the outcome of attempts to include private plans within Medicare itself and a comparison of the costs and access for privately insured Americans to costs and access for Medicare beneficiaries. Accordingly, we can compare the performance of the market with the performance of an approach that weakly approximates the international standard (White 1995) for national health care/insurance systems. Medicare does not have the cost control potential of many national systems because it does not control system capacity and has not implemented hospital budgets. Yet it shares with other countries such attributes as compulsory contributions and membership, contributions related to ability to pay, cost controls applied across the universe of providers, a very wide risk pool for beneficiaries (one of the largest single pools in the world), and the resulting ability to be a price maker rather than a price taker, with only limited (but not insignificant) concerns that providers might exit the system.
Other scholars, from Kenneth Arrow to Mark Pauly to Tom Rice, have offered sophisticated discussions of how economic theory can be applied to medical care production and delivery (Arrow 1963, 2001; Evans 1998a, 1998b; Gaynor and Vogt 1998; Glied 2001; Pauly 1998a, 1998b; Reinhardt 2001; Rice 1998a, 1998b, 2002; Rosenau 2003; Sloan 2001). By contrast, this article focuses on “the market” in its actual, not theoretical, form, as it existed in the United States during this period. The outcomes of interest are the cost of medical care, access to health insurance, and what might be called the rationalization of medical care. Rationalization here stands in for quality, which is very difficult to measure in even small batches, much less across the whole system.
By rationalization, I mean a reorganization of medical care so as to resemble more closely a standard of what Henry Aaron and William Schwartz call “medical efficiency:” “that every medical service offered produce larger expected benefits per dollar of total cost than any medical service not provided” (Aaron and Schwartz 2005, p. 96). The premise of what has often been called “managed care” is that in too many cases the wrong services are provided and that the right set of incentives or form of organization would lead to the right services provided to the right people at the right time. The theory of managed competition thus promised that a more market-oriented approach would increase value for money spent, by leading to better-informed or more prudent purchasing. As we will discuss later, advocates expected that the dynamics of market competition would yield a reorganization of health services in which more efficient and integrated systems, epitomized by the group-/staff-model HMO, would come to dominate the system (Gitterman et al. 2003). Because this was a prominent goal for market-oriented reformers during the 1990s, whether the market can be expected to deliver such a result (or, alternatively, why it cannot) should be a subject for evaluation.
Part 1 of this article sets the stage by outlining the basic attributes of the market in American medical care and ends with a discussion of the relationship between markets and the rationalization desired by many advocates. Part 2 reviews the basic data on trends in costs, access, and the organization of medical care from 1993 through 2005. I begin with 1993 both because this was the time of the most concentrated political action to change the system and because it is the most obvious breaking point in the data, seeming to show the beginning of a period of successful market-based change. I chose 2005 as the end point both because it is recent and because the implementation of the Medicare drug benefit (and its attendant subsidies to private insurers) must change the meaning of comparisons between Medicare and the private sector.
Part 3 uses the Health System Change and other analyses to explore the market dynamics that explain the trends regarding cost, access, and organization.
Part 4 addresses the idea that the markets’ failure to solve health care problems was a political failure rather than a market failure, in other words, that a political “managed care backlash” undid a promising start to market-led reform. Neither the Health System Change data on plan and provider behavior nor other careful analyses support that theory.
In this article’s conclusion, I argue that in order for any kind of “market-oriented” reform of American health care to have significantly positive effects on cost, access, and quality, it would have to include such substantial restrictions on the normal ways of doing business in U.S. markets that it would be barely recognizable as “market oriented” in the American context. For example, it would have to greatly restrict the flow of capital and ban many forms of insurance contracts.
Part 1: Attributes of the Market
Arguments in favor of improving health care through more reliance on markets and competition tend to be slippery. The failure of any particular approach can be written off as not getting competition “right,” that is, a failure of execution rather than principle. “Despite widespread acceptance of the competitive market model in the U.S. health care system,” Bryan Dowd writes, “debate continues regarding the optimal form of competition and the patient-professional relationship” (Dowd 2005, p. 1501). In other words, no remotely optimal form has been identified in practice, but we can keep looking. In a fine review of recent developments, Paul Ginsburg (2005) explains many of the reasons why they did not lead to reliable cost control or improved access and quality. Yet he still suggests that perhaps some other kind of competitive approach might have better effects.
Although few hypotheticals can be dismissed entirely, some are quite improbable. The question for reform of American health care is not whether one can imagine virtuous competition in some theoretical world (although Rice 2002 and Rosenau 2003 give reason for doubt). Instead, the question is whether institutions of market competition in the current American political economy are likely to have virtuous results. “Markets” are not a form of organization that can be separated from the context of norms and other institutions. A “stock market,” for example, may be constructed in very different ways and for very different purposes in accordance with how politics, economic resources, and economic organization vary across countries (Lavelle 2004). The literature of political economy shows that the way capital is organized and the resulting corporate incentives and norms can differ across countries (as referenced in Arrow 2001; for more details, see Doremus et al. 1999; Shonfield 1965). The basic rules of national economic organization, such as wage and hours legislation, determine how both for-profit and not-for-profit firms can be managed.
At the most general level, a “market” can be defined as a network of buyers and sellers. According to this definition, a market day in classical Greece, the cloth industry of medieval northern Europe, the souk in Casablanca, and the American home-building industry all are the same phenomenon. Market exchange existed long before modern capitalism, but the ideology of markets and modern market institutions and the modern understanding of the concept are based on an economic system and its accompanying values that arose over the past two centuries (Barber 1995). Three aspects of modern markets are relevant to how behavior driven by “market incentives” and in a context of market institutions has affected and can affect the cost of, quality of, and distribution of access to American health care:
- Suppliers’ individual pursuit of profit (or income maximization). Note that this behavior exists in other systems as well. However, advocates of market incentives rely on this pursuit to create efficiency, so a system that gives greater scope to markets should impose fewer institutional constraints on income maximization and also create fewer social obstacles (such as norms of restraint) to that behavior. For example, the modern market differs from the medieval cloth trade because it lacks the norms and constraints associated with a guild system. Nonprofit organizations in a market system may pursue somewhat different values than do for-profit organizations, but their leaders still will be influenced by society’s approval of income maximization and by the market’s economic forces (Hall and Conover 2003; Rosenau 2003).
- Extensive shopping for care, whether by individuals or agents for individuals. In the United States, such agents could be employers, who are the main purchasers of insurance. Shopping in this case means choosing under some conditions of price constraint, not simply choosing which physician or hospital to go to without any price constraints (as can be done in Canada or Germany).
In theory, markets should maximize value because as suppliers pursue profit and invest in creating capacity in pursuit of profit, they will be disciplined by customers’ shopping. The need to offer lower prices than the competition’s should encourage efficiency, and the drive to satisfy customers should encourage the creation of a diversity of products to match different individual utilities.
These first two factors are the “facets of competitive theory” that Rice identified (1998a, p. 29). But a third is equally important:
- Medical care providers and insurers that have extensive access to capital when the suppliers of capital (either equity or debt) are motivated mainly by the pursuit of profit: in other words, capitalism. It is possible to have “competitive” reforms that extract capital from the equation. For instance, the British National Health Service’s “internal market reforms” under Prime Ministers Margaret Thatcher and John Major sought to create a form of competition among hospitals but kept a tight rein on capital. The result was an extremely limited form of competition (White 1995), and none of the theories of “competitive” reform for the United States that I have seen contemplates anything like this.
Instead, as J.B. Silvers writes, the development of the U.S. health care system since Kenneth Arrow’s classic article on the welfare economics of health care in 1963 has included a massive growth in assets, “fueled by an unprecedented use of tax-exempt debt, retained earnings, and new stock” (Silvers 2001, p. 1019). Silvers then argues that relying on private capital has much the same effect whether the source is debt or stock and whether the recipient is a for-profit or a nonprofit organization. Nonprofit organizations that go into debt can go bankrupt or be forced into mergers, reorganizations, or changes in management. “With the use of other peoples’ money came the responsibility to meet more stringent financial requirements . . . the threat of bankruptcy is the ultimate lever of control and may be even stronger than the votes of shareholders” (Silvers 2001, pp. 1025-26). The marketization of the American health care system includes the subjection of even nonprofit providers to the “discipline” of the financial markets, in ways that are dramatically different from the world that Arrow described. For example, a world in which local institutions were managed mainly by local nonprofit boards was largely superseded because “competitive or financial threats have compelled a very large portion of all providers to merge with larger entities with resulting loss of local managerial control” (Silvers 2001, p. 1026). That is what a capitalist market looks like.
Markets and “Managed Care”
In the 1990s, the mainline theory of how markets would improve the American health care system relied on two factors: competition in some form and “managed care.” The course of events over the following decade reflected the realities of both markets and “managed care.” Before reviewing the events, therefore, we must make some distinctions about the latter.
As used in the health policy literature, the term managed care has two distinct meanings (for references to the use of the meanings and more extensive discussion, see White 1999). The first, closer to the commonplace understanding of “managed care,” assumes that somebody other than medical providers will in some way manage treatment decisions. We can call this managing treatments. Measures for influencing treatment decisions include third-party enforcement of standards (utilization review), providers’ responsibility for the costs of referrals or prescriptions (risk-bearing gatekeeping), and the creation of large delivery systems that may develop a practice culture of and internal management routines to enforce more conservative treatment (the traditional group/staff HMO). The second meaning of “managed care” in the literature, especially when applied to “managed care organizations,” is insurers’ contracts with some providers and not others. Hence what is being “managed” most directly is the network of providers that is offered to the insurer’s customers. This form of “managed care” can more accurately be termed selective contracting.
Selective contracting as such requires no management of treatments. Instead, it may be employed mainly by purchasers who hope to obtain lower prices by threatening to take their business elsewhere. Conversely, in principle, treatments could be managed without any selective contracting. For example, some national health care systems require or encourage access to specialists only through primary care gatekeepers. If selective contracting is essentially a way to lower prices, it will not involve the kind of system rationalization—the guarantee that patients receive the right care from the right provider at the right time—that was at the heart of the ambitions for competition-led reform.
Much of the rhetoric about the insurance transformations of the 1990s seemed to suggest a big shift to managing treatments. For instance, the “managed care backlash” was largely framed as objections to utilization reviews (“1-800-Mother May I”) or to gatekeeping.2 Analysts would speak of “Jekyll and Hyde” forms of managed care (Bodenheimer and Grumbach 2005, p. 45). But we will see that the managed care revolution consisted more of selective contracting than of managing treatments, which leads to two empirical questions. First, was the marketization associated with savings related more to discounting or to rationalization? Second, if the period of relatively good cost control was associated more closely with discounting, why wasn’t it stable, and why was the management of treatments less important than the common rhetoric (including both sides of the managed care backlash) suggested?
Part 2: Health Care and Insurance, 1993-2005
To the health policy world, the most obvious initiative in 1993 was President Bill Clinton’s ill-fated attempt to enact a national health insurance plan. At the time of this battle, costs were expected to quickly hit 14 percent of GDP and rise to 18 percent by the end of the decade (White 1995, pp. 239-40). Yet even when those projections were made in 1993, the cost trend in the private sector was dramatically slowing. This suggested to corporate managers that they could control their health insurance expenses without giving government the job.
Cost Control
A longer view, however, can show both why the cost restraint after 1992 seemed impressive at the time and why it was quite temporary. We first need baselines for how well costs can or should be controlled, for which both American history and the performance of other countries provide perspectives.
Figure 1 shows the basic trend in total health care spending as a share of GDP in the United States and nine other rich democracies. After growing much more quickly than costs in any of the other countries from 1980 to 1992, costs in the United States suddenly stabilized, so that cost control was equal to or better than that in other countries (except, interestingly, Canada) through the end of the decade.
Table 1 provides a further comparison, between private insurance and Medicare. Private insurance’s benefit packages tend to be different from Medicare’s, in that they often are more generous, particularly regarding pharmaceutical benefits, but may offer less coverage of or need for benefits such as home health care. Therefore it is safest to compare all benefits and also those that the two systems share (such as physician and hospital services).
Table 1 shows that between 1993 and 1997 the spending trend for private health insurance slowed dramatically from the period before it and that spending increased much less quickly than for Medicare during the same years. This improvement was associated with an accelerating shift by employers toward more “managed” health coverage, as well as by a political stalemate that prevented significant Medicare cost control legislation after 1990. As the economy also improved, both costs and premiums grew more slowly than the per capita GDP through 1997, so that national health expenditures declined as a share of the economy. Instead of rising toward 18 percent, they fell from 13.4 percent of GDP in 1993 to 13.2 percent in 1998 (Heffler et al. 2005, pp. W5-W75).
Around 1997, however, trends for both private insurance and Medicare reversed. The particularly rapid increase in pharmaceutical costs since the late 1990s perhaps might explain some of the worse performance of private insurance, although the table shows that different benefits explain only 0.3 percentage points of the 3.1 percentage point annual difference in cost trends between Medicare and private insurance from 1999 to 2004.
The reasons for changes in the private-sector trend are discussed at greater length later. Here we need to understand only what happened in Medicare. In 1997 the warring factions in the U.S. government finally compromised, in the Balanced Budget Act (BBA), on a set of savings measures for Medicare. Payment controls were strengthened where they had previously been applied (to inpatient care and physician services) and extended to areas that had been relatively uncontrolled (nursing homes, physical therapy, and home health care). In the language of political dispute, savings were derived from regulation, not competition (Moon, Gage, and Evans 1997; O’Sullivan et al. 1997).
In addition, the federal government initiated a crackdown on Medicare “fraud and abuse.” Legislation (such as BBA-97 and the Kassebaum-Kennedy insurance reform law in 1996), increased financing for investigation and prosecution (in both the FBI and the U.S. Attorneys’ Offices), use of laws with harsh civil penalties (the Federal False Claims Act), and particularly visible prosecutions of large providers (the University of Pennsylvania system and the giant Columbia/HCA for-profit hospital chain) appear to have scared the wits out of health care managers. Putting people in jail and levying multimillion-dollar fines on institutions are not cost control methods easily adopted by private payers. In response to the antifraud initiative, “DRG creep”—the phenomenon in which more and more admissions were classified as more complex and costly diagnoses under the prospective payment system for inpatient care—suddenly stopped in 1997 (U.S. CBO 1999). Home health care providers offering questionable services vanished from the world of Medicare contracting, and Medicare cost increases suddenly slowed. In fact, total Medicare costs even shrank slightly between federal fiscal years 1998 and 1999 (U.S. CBO 1999, 2001).
The new Medicare trend was almost as big a surprise as the earlier savings in the private sector and not much more sustainable. Providers screamed with the pain of cost constraint, and there was great political pressure for “givebacks.” Meanwhile, the federal budget had shifted into surplus so the government had money to give. Medicare’s costs thus returned to a pattern of rapid increase between 2001 and 2003, but the private sector’s costs rose even more quickly.
At the time, the fluctuating trends in the private sector appeared even more extreme than these data reveal. What employers (except for those who self-insure) and commentators see most directly are insurance premiums. Insurance premiums both fell more quickly and then rose more quickly than the underlying cost trends because insurance tends to follow an “underwriting cycle” in which periods of higher profits are followed by greater price competition among insurers (reducing the spread between premiums and medical costs). The cycle then turns, as higher “medical losses” lead to less price competition and a higher spread between premiums and costs.3 The data on premiums paid by large insurers show that premiums grew more quickly than costs from 1990 through 1995, more slowly from 1995 through 2000, and then much more quickly from 2001 to 2003. Through 2005, they continued to increase a bit faster than costs.4 This underwriting pattern matters, as we will see, because it includes herd behavior by insurers in both restraining premiums and then aggressively raising them.
By 2003, the political retreat from Medicare cost control, the collapse of cost controls in the private sector, and health insurers’ pricing strategies combined to return the American health care system to the cost crisis in which it had been in 1993, even though Republican control of the federal government kept national health insurance far away from the political agenda. As a share of the economy, national health expenditures rose from 13.2 percent in 1998 to 16 percent in 2004. In early 2006 they were projected to rise to 20 percent of GDP by 2015 (Borger et al. 2006).
As table 1 shows, from 1970 to 2004, costs rose a bit more rapidly for private insurance than for Medicare. From 1993 to 2004, the gap for “all benefits” was almost as large as before, although the gap for “common benefits” narrowed substantially. Except for the brief period from 1993 to 1997, “the market” has not been able to control costs as well as Medicare has. In addition, Medicare beneficiaries appear to be, on average, “generally more satisfied with their health care than are privately insured people under age sixty-five” (Boccuti and Moon 2003a, p. 235).5
Coverage
The cost slowdown of the mid-1990s had a positive effect on access to private health coverage. The combination of rapid economic growth and significant cost control made health benefits more affordable for employers and therefore restrained the growth in the share of costs that employees would have to pay to cover their families. As a result, employment-based coverage reversed its decline and, between 1994 and 2000, rose from covering 64.4 percent of the population to covering 66.8 percent. Meanwhile, coverage for the needy through the Medicaid program declined from 12.7 percent of the population in 1994 to 10.5 percent in 1999, because of both the good economy (which reduced need) and the “welfare reform” that reduced participation in Medicaid. These trends, however, then reversed. By 2003 only 63 percent of Americans had health benefits through employment—a smaller percentage than in 1994. Meanwhile, governments in the late 1990s responded to favorable budget conditions by expanding Medicaid eligibility. When the economy then turned sour, Medicaid enrollments grew to 12.8 percent of the population by 2003—higher than in 1994 (Fronstin 2004, p. 5).
By March 2003, nearly 45 million Americans—about 17.7 percent of the population below the age of sixty-five (and thus ineligible for Medicare) had no health insurance. Holahan and Wang summarized the pattern: “The extent to which the loss of employer coverage resulted in people becoming uninsured depended on their access to public programs” (2004, p. W4-31).
Cost matters to access: when both governments and employers were doing well financially, they tended to maintain or expand coverage, and when they were not doing well, they reduced or, at best, maintained the same coverage. But the overall pattern suggests that government was a bit more likely to intentionally expand coverage in good times and to resist contracting coverage in bad times. The decrease in Medicaid enrollment in the mid-1990s did result in part from welfare reform. This reform was not driven by budget concerns, however, but by conservative ideology and general public disgust with the previous system. In addition, as written, the 1996 law was supposed to maintain the entitlement to Medicaid benefits. Much of the decline was viewed as a failure of state administration, and in the late 1990s, the states actually made efforts to rectify those failures of outreach. Medicaid turned out to have enough “support among coalitions of public officials, health care providers, and local advocates” to “protect the program in hard times and enlarge it when the clouds lift” (Hoadley, Cunningham, and McHugh 2004, pp. 143-44). It should be no surprise that private market dynamics are not as reliable a way to subsidize poor people as government is, even in the United States. Indeed, equity is not what markets are supposed to provide (Pauly 1998a).
Cost controls are good for payers and bad for providers. For example, the period of strong cost controls had particularly negative effects on major teaching hospitals. Not only were they (for a while) at some disadvantage in contracting with private insurers, but they were hit by the antifraud campaign in Medicare, and many of them made unsuccessful investments (such as purchasing physician practices) during the 1990s. MedPAC estimates that in 1999 at the peak of the effects of the Medicare restraint, the average operating margin for teaching hospitals had fallen to 0.2 percent, which means that many of them were in the red (MedPAC 2001, pp. 69-71).
Public policies did, however, ameliorate the effects on access. Safety-net hospitals benefited from the Medicare and especially Medicaid “Disproportionate Share Hospital” (DSH) programs (Zuckerman et al. 2001). Academic medical centers also received payments for medical education and benefited from the boom in National Institutes of Health funding at the turn of the century. Reduced income from the spread of “managed care” was associated with physicians providing less charity care (HSChange 1999). Shrinking inpatient capacity (in almost all markets) and facility closures (in many) did cause many problems with access to emergency departments; ambulances shunted from one emergency department to another became common by 2001. Yet these pressures were somewhat ameliorated by a mix of measures, including the expansion of community health centers, the reorganization of dispatching systems for ambulance services, and hospital managers’ choosing to expand emergency departments in hopes of catching more inpatients (Brewster and Felland 2004; Felland, Felt-Lisk, and McHugh 2004; Kellerman 2004; Melnick et al. 2004).
The basic pattern, then, was that the market did threaten the “safety net” but that the safety net was protected—mostly—by political decisions. By 2003 a larger proportion of Americans was uninsured than in 1993, and more Americans were dependent on government “safety-net” programs such as Medicaid and subsidies to community health centers and academic medical centers.