A striking feature of social policy in the United States is the heavy reliance of public agencies on private institutions. At the state and local level, social services are delivered through a welter of nonprofit organizations, community institutions, and profit-oriented providers. At the national level, longstanding federal policies leave a wide range of social welfare responsibilities in the hands of private corporations, whose provision of health insurance, pensions, and other fringe benefits is subsidized through the tax code and governed by an expanding complex of regulations. Even America's most quintessential social programs are not exempt from this pattern. Federal housing, public assistance, and education policies all have long mixed direct government spending with indirect support for private provision through loans, insurance, tax breaks, and vouchers. Yet, in the roster of federal social interventions, no program better illustrates the private sector impulse-in its best and worst manifestations-than Medicare.

When Medicare was created in 1965, it was meant to emulate the Blue Cross Blue Shield plans that were then the gold standard for private coverage. And not just emulate them: under the law, hospitals and doctors could elect to have their Medicare payments funneled through the Blues or other private plans, and not surprisingly, most did. Not only that, the legislation promised to reimburse medical providers at rates that were "reasonable and customary," meaning essentially whatever they chose to charge. Ironically, a program that critics have savaged for decades as socialism run amok initially foreswore any intervention in the practice or structure of private medical care. Social insurance it was, but "socialized medicine" it surely was not.

In the years since, of course, Medicare has become a more interventionist program, its reimbursement system increasingly restrictive, its largesse entailing obligations on the part of medical providers that may well have killed the legislation in 1965. Yet, in many ways, Medicare today is more dependent on the private medical sector-and the private sector on it-than ever. This is because of a critical but, until recently, little-noticed development in the program: the establishment and expansion of Medicare risk contracts. Under such contracts, Medicare pays HMOs and other private health plans a preset amount to care for Medicare beneficiaries who enroll in private plans. Initiated as a demonstration project in the early 1980s, risk contracting has grown dramatically in the past half decade, expanding from 114 plans and fewer than two million enrollees in 1993 to more than three hundred plans and over 5.5 million enrollees in 1998. More than a seventh of Medicare beneficiaries now receive their care from private plans, and the Congressional Budget Office (CBO) projects that under current policy about half will do so by 2030 (CBO 1998). Critics and proponents alike have described this development as "privatization," but this is misleading, inasmuch as it implies that Medicare does not incorporate substantial elements of private administration and provision already. Although the growth of contracting does indeed represent a fundamental change within Medicare, that change has more to do with the distribution of risk rather than with the delivery of services-which is why it cuts much closer to the heart of Medicare qua social insurance than either side seems to have recognized.

Medicare contracting became a leading political issued in 1997, when President Clinton and Congress agreed to restructure and expand the policy as part of the so-called Medicare+Choice provisions contained in the Balanced Budget Act (BBA). The issue became prominent again in early 1999, when a bipartisan commission established by the BBA failed to reach the necessary supermajority to recommend a comprehensive Medicare reform proposal crafted by the commission's chair, Democratic Senator John Breaux. That proposal went well beyond the BBA-which, among other things, increased the range of plans participating in Medicare and changed the way they were paid-to outline an entirely new program based on regulated competition among private health plans, with traditional Medicare becoming simply another contender in the competitive fray. The demise of the commission and the subsequent release of a Medicare Trustees' report indicating a rosier-than-expected fiscal outlook have pushed this ambitious set of changes off the immediate political agenda. Yet reforms of this sort will remain a leading alternative in the debate over Medicare's future and, regardless of their fate, the place of private health plans within the program is certain to remain fiercely contested.

Into these unsettled waters wades Richard Kronick and Joy de Beyer's recent edited volume, which presents as informed and timely an analysis of the dilemmas raised by contracting-and of the mechanisms proposed to address them-as health policy experts are likely to find. The chief selling point of this focused book-a collection of essays commissioned under a grant from the Commonwealth Fund-is its careful review of the problem of risk-selection and how it might be ameliorated through payment mechanisms, the provision of information, and the regulation of plan behavior. The book is clearly meant to offer an optimistic assessment of such measures. In the concluding chapter, Kronick and de Beyer express their hope that "the material presented in this book will further understanding of the tasks ahead and perhaps facilitate their accomplishment" (p. 226). Yet the problems that their volume identifies are so endemic and the proposed solutions so institutionally and politically delicate and of such uncertain efficacy that the overall effect is less than reassuring.

Medicare HMOs is billed as a manual for "making HMOs work for the chronically ill." This is false advertising, or at least something of a bait and switch. The volume has surprisingly little to say about how vulnerable patients fare in HMOs or other private plans. Kronick and de Beyer do claim in the preface that "HMOs can reduce the costs of providing healthcare to a defined population, increase quality of care, and lead to improved health outcomes" (p.1), and they offer a one-paragraph literature review of the research on comparative health quality in the first chapter. Yet none of the essays assesses in any depth the relative performance of Medicare and private plans, nor do any evaluate the degree to which HMOs live up to their promise of preventive medicine, coordinated delivery of services, and active clinical oversight. As a result, the book is less an analysis of the suitability of private health plans for vulnerable patients than a detailed discussion of risk selection and how to discourage it. And while risk selection clearly affects the care of individual patients, its broader implications for the structure of Medicare are in many ways more profound. Unfortunately, because the book is framed around the incentives that risk selection creates for the delivery of care to individual patients, it does not give enough attention to these larger structural concerns.

Nonetheless, on the realities of risk selection and the potential means to fight it, the essays in this collection are knowledgeable and well crafted. Kronick and de Beyer set the tone with a brisk tour of the evidence that Medicare HMOs attract a disproportionate mix of healthy patients. Although the story will be familiar to most health policy observers, they tell it well, and it is hard not to be impressed anew by the success that HMOs have had in avoiding high-cost patients. Compared with beneficiaries who remain in traditional Medicare, enrollees in Medicare HMOs "have lower prevalence of chronic conditions, better self-reported health, and better functional health status" (p.10). They are less likely to be disabled, have significantly lower age-adjusted mortality rates, and cost Medicare less before enrollment than those who remain in the traditional program. As de Beyer shows in chapter 8, this is not just because healthier patients are more likely to enroll in HMOs, but also because sicker, older, poorer, and disabled patients are more likely to leave them. Yet, despite extensive biased selection of this sort, Medicare pays private plans a fixed amount that is essentially equal to what it costs the program to treat an average Medicare beneficiary in any given region, with minimal adjustments for demographic and other differences. Private plans can thus reap considerable profits and cover their higher administrative costs even while offering broader benefits than Medicare does. This is a good deal for private health plans and for those who gain enhanced coverage through them. But it is a bad deal for Medicare, which has consistently lost money on the program (at least $2.2 billion in the past three years (p. 217)). And it poses risks to high-cost beneficiaries, whom private health plans have incentives to avoid enrolling, retaining, or adequately treating.

The remaining essays all consider ways to deal with the risk-selection problem. Inevitably, many of the solutions are familiar from the 1993-1994 debate over the Clinton plan for "managed competition." Now as then the Holy Grail of competitive health care reform, "risk adjustment," emerges center stage. This time, however, it surfaces with more real-world experience and short-term significance attached to it, since some states and large employers have already tried to implement risk adjustment and the BBA requires that Medicare develop a provisional scheme by the turn of the century. In their well-researched chapter on the subject, Tony Dreyfus and Kronick make a convincing case that risk adjustment (or, as they term it, "health-based payment") could bring the amount that Medicare pays plans and the amount that beneficiaries cost plans more closely into line. Their case is weaker, however, when they turn to the feasibility and effectiveness of the approach. Even the best available techniques for adjusting payments son the basis of a patient's past health experience still have relatively poor predictive ability (accounting for between 8 and 11 percent of an elderly patient's expenditures in the coming year (p. 5)). And these measures require extensive diagnostic information that, for many beneficiaries, would have to be collected and reported by private health plans themselves. Putting aside the privacy concerns that seem to be a fixture of medical care debates in the United States, it is worth noting that private health plans have to date been acutely reluctant to take on this task, and many would need to make substantial investments to do so (pp. 49-50).

Recognizing these difficulties, Dreyfus and Kronick consider various complements to risk adjustment. They discuss risk sharing and partial capitation formulae that mix prospective and retrospective reimbursement, and they argue that Medicare should provide special payments for costly end-of-life care (a policy that has so far been considered taboo because of the specter of "paying for death" (p. 383). They also rightly argue that risk adjustment neither can nor must be perfect to weaken the incentives for risk selections. Still, Dreyfus and Kronick do not specify how good risk adjustment has to be to pose a serious barrier to risk selection-in part because there is not yet sufficient evidence to know. Within any given risk category, however sophisticated its construction, patients will always turn out to differ greatly in their actual medical need. And, once enrollment occurs, health plans will always have more information about the health of patients than Medicare does. So, too, for that matter, will patients, who may move in and out of plans based on their current need for specialty services or ready access to care. The combination of plan "cherry picking" and patient self-selection, along with the subtle effects of marketing, the structure of benefits, and the way in which care is delivered, all could lead to biased selection or undertreatment even in the presence of risk adjustment.

Accordingly, the chapters that follow Dreyfus and Kronick's discuss a number of auxiliary measures for discouraging health plans from engaging in risk selections or providing substandard care to costly patients: Mark Merlis examines the provision of comparative information on health plans to Medicare beneficiaries, Mark McClellan and Sontine Kalba weight the possible effects of benefits standardization, Gerard Anderson looks at managed care regulations at the state level to assess the prospects for federal quality-assurance efforts, and de Beyer takes up the particular regulatory challenge of monitoring and policing disenrollment from private plans. Perhaps the most inviting chapter is the one by Thomas Rice, who explores a reform option that has only a tenuous connection to selection bias in Medicare HMOs. He proposes the creation of an annual open enrollment period for Medigap supplementary policies that would be coordinated with the new enrollment period for private plans mandated by the BBA. The benefits of reforming the Medigap market and coordinating enrollment are clear. Less unambiguous, however, is Rice's assertion that coordinated open enrollment would reduce the extent of selection bias. His premise-that unhealthy beneficiaries would find they could receive broad benefits through private insurance and hence choose it over traditional Medicare plus supplementary coverage-assumes, rather implausibly, that private plans would continue to be able to offer such comparatively generous benefits if risk selection were less prevalent or payments to private plans better reflected the health status of enrollees.

As in any edited volume, the quality of the analyses vary, and the authors evince different degrees of enthusiasm for the broader project of the book. Merlis's fine chapter on comparative information, for example, ends on a decidedly skeptical note, citing the difficulties of comparing quality across private plans and, even more so, between such plans and Medicare. He also points out the irony that better comparative information might actually increase, not decrease, risk selection by encouraging patients with special conditions to enroll in the plans that excel in treating their needs. As Merlis sensibly concludes, "To advocate improved consumer information as a means of reducing selection bias assumes that higher-risk beneficiaries making fully informed and rational choices will be more likely than at present to select managed care options. This is by no means certain" (p. 102).

The collective impression conveyed by these chapters, besides the obvious conclusion that Medicare contracting will keep health policy specialists in high demand well into the next century, is that improving the operation of Medicare's internal market will require a daunting progression of delicately calibrated reforms, nearly all reliant upon one another to achieve their desired ends. The closing chapter by Kronick and de Beyer does little to dispel this impression. It lists at least twelve separate reforms discussed in the previous pages, rating each with regard to its feasibility and potential effect on risk selection. To their credit, Kronick and de Beyer include a crude measure of political feasibility in their ratings-the degree to which proposed reforms threaten "vested interests" in Medicare (presumable private health plans and other industry interests, not beneficiaries). But, in addition to being ungrounded in any broader discussion of Medicare politics, this gesture toward political realism unrealistically considers each of the reforms in isolation from one another. It also appears much too sanguine: during the debate over the BBA, Medicare HMOs managed to turn a proposed reimbursement cut into a guaranteed payment floor and a promise of minimum annual future increases, and even so many have since left or threatened to leave the program. As with the complex Clinton health plan, the inspiration for which was similarly rooted in the managed competition design, the clever ideas catalogued in Medicare HMOs fail as a whole to give sufficient weight to the difficulties of achieving coordinated and interdependent policy aspirations through the fragmented and conflictual American political process.

The major fault of the book, however, is more serious. For a volume about risk selection in Medicare, Medicare HMOs has almost nothing to say about the broader dynamics of risk shifting and risk segmentation that might be unleashed by movement toward a fully implemented system of competing health plans, such as that proposed by Senator Breaus and apparently endorsed by Kronick and de Beer (pp. 225-226). The correspondence between what Medicare pays private plans and what patients end up costing obviously has implications for how well high-risk patients fare in the private sector. Yet, even more directly, it concerns Medicare's continued ability to pool the varied individual medical risks of beneficiaries within a common program, which is, after all, the essence of social insurance. None of the authors has much to say about this aspect of the distribution of risk within Medicare, and none addresses the salient worry that within a competitive system, the traditional Medicare program could become a last refuge for the sickest and poorest beneficiaries, who in any given year account for virtually all of the program's costs. These concerns are nagging today, but they will become pressing if Medicare+Choice and reforms like those advocated in Medicare HMOs have the desired effect of increasing enrollment in private health plans.

It may be too much to ask of American policy analysis that it treats politics not as an afterthought that complicates the best-laid plans, but as an integral aspect of constructing and advocating policy solutions that can stand the test of democratic deliberation and debate. Yet surely it is not too much to ask of a book on Medicare that it consider the relationship between programmatic change and the continued vibrancy of Medicare as social insurance. Ultimately, this is the main failing of Kronick and de Beyer's otherwise worthy volume: a restricted vision that focuses on the incentives for treating patients but not on the larger social insurance vision that animated Medicare's initial creation and that still represents its most substantial achievement.

Medicare is part of American medical policy, but it was not intended to change American medical care. It pays for health care services, but that hardly distinguishes it form many other actors in the health care sector. What marks Medicare as distinctive is the conviction that, in this realm of life and for this group of citizens, a common umbrella of protection should be available to all regardless of individual health, income, or status. The question for those assessing Medicare reform is not merely how it will affect vulnerable groups-important though this question may be-but how it will affect the larger social insurance vision that Medicare aims to realize. And the debate should not be over whether health care under Medicare should be private-it always has been-but over whether risks can still be shared in common when they become the province of private institutions.