National efforts to greatly reduce the number of uninsured Americans have made little progress for decades because achieving majority support for any one approach has proved to be impossible. While as authors we remain unreconciled on the best solution, we share the belief that federally supported state experimentation is a promising way to make progress. States should be allowed to try widely differing solutions with federal financial support under legislated guidelines, including specific protections and measurable goals. Congress would enact a "policy toolbox" of federal initiatives that states could include, and funding to states would be linked to success in reaching the goals.

Nearly everyone thinks that something should be done to reduce the number of Americans lacking health insurance. Unfortunately, while numerous plans exist on how to reach that goal, few agree on any one. Indeed, as authors we disagree on how best to extend and assure health insurance coverage. Nonetheless, we believe that using the pluralism and creative power of federalism is the best way to break the political logjam and to discover the best way to expand coverage.

Accordingly, we believe that states should be strongly encouraged to try any of a wide range of approaches to increasing health insurance coverage and rewarded for their success. This approach offers both a way to improve knowledge about how to reform health care and a practical way to initiate a process of reform. Such a pluralist approach respects the real, abiding differences in politics, preferences, traditions, and institutions across the nation. It also implies a willingness to accept differences over an extended period in order to make progress. And it recognizes that permitting wide diversity can foster consensus by revealing the strengths and exposing the weaknesses of rival approaches.

Despite our abiding disagreements on which substantive approach to extending coverage is best, we believe that people of goodwill must be prepared to countenance the testing of ideas they oppose if progress is to be made. Moreover, we believe that there is no hope for legislation to begin to transform the largest U.S. industry-health care-unless such legislation enjoys strong support from both major political parties.

Using Federalism To Spur Action

Proposals to reduce the number of uninsured Americans abound. Some favor expanding government programs, such as Medicaid. Others favor refundable tax credits to help families buy private health insurance. Still others favor regulatory approaches, such as changes in insurance rules. But working together in health care to achieve a goal shared by virtually everyone has proved to be impossible. One reason for this is that the capacity to reach substantive compromise in Washington has seriously eroded.¹ Among the causes is the widespread view that reforming the complex health care system requires very carefully designed and internally consistent actions. Some say that it is like building a new airplane: Unless all the key parts are there and fit together perfectly, the airplane will not fly. Thus, many proponents of particular approaches fear that abandoning key components of their proposals to achieve a compromise will prevent a fair test of their favored approach and lead to failure. Another obstacle is that many lawmakers believe that approaches that might conceivably work in one part of the country, given the cultural, philosophical, or health industry conditions prevailing there, will not work in their state or district because of different local conditions. This view leads many in Congress to resist proposals that might work in some areas because they believe that those proposals could make things worse for their constituents.

These and other factors have stalled efforts to extend health insurance and achieve other reforms for decades. The enactment of Medicare and Medicaid stands as one notable-and instructive-exception to that pattern. Medicare sprang from comprehensive social insurance initiatives of congressional Democrats, Medicaid from limited needs-based approaches of congressional Republicans. The passage of each program was possible only because the two initiatives were linked in the form of a trade-off, not so much by blending some elements of each approach but by moving forward with two programs in parallel: Medicare for the elderly and disabled, and Medicaid for the poor of all ages. That experience illustrates a principle of politics: that progress often requires combining elements of competing proposals into a hybrid legislative initiative, in which internally consistent approaches operate in parallel.

In our view, federalism offers a promising approach to the challenge of building support to tackle the problem of uninsurance. While proponents of nationwide measures to introduce health insurance tax credits, or to extend Medicare or the State Children's Health Insurance Program (SCHIP) to other groups, should of course continue to make their case for national policies, we emphasize an initiative designed to support states in launching a variety of localized initiatives. Under this process, the federal government would reward states that agreed to test comprehensive and internally consistent strategies that succeeded in extending coverage within their borders. In contrast to block grants, federal-state covenants would operate within congressionally specified policy constraints designed to achieve national goals for extending health insurance. These covenants would include plans ranging from heavy government regulation to almost none, as long as the plans were consistent with the broad goals and included specified protections. States could also select items from a federally designed "policy toolbox" to include in their proposals. Allowable state plans would include forms of single-payer plans, employer mandates, mandatory individual purchase of privately offered insurance, tax credits, and creative new approaches. States would be free not to undertake such experiments and continue with the current array of programs, but sizable financial incentives would be offered to those that chose to experiment and financial rewards given to those that achieve agreed-upon goals.

The model we propose builds upon proposals we have outlined elsewhere.² It is also compatible with some other federalism approaches, such as the plan advanced by the Institute of Medicine.³ We favor a wide diversity of federal-state initiatives for three reasons. First, fostering a bold program in a state will produce much information that will aid the policy discovery process. Successes will encourage others to follow, while unanticipated problems will force redesign or abandonment and will be geographically contained. Second, encouraging bold state action will quickly and directly extend coverage to many of the uninsured. Instead of facing continued national inaction or the potential for disruption of state initiatives by future federal action, states would have the incentive and freedom to act decisively. Third, we see no evidence of an emerging consensus on how to deal with these problems at the national level. But our proposal is based on the observation that advocates of rival plans trust their preferred approaches enough to believe that a real-life version would persuade opponents and create a consensus. Not all can be right, of course, but all advocates of health insurance reform, like residents of Lake Wobegon, seem to believe that their plans are above average. Thus, they should be open to the idea of testing diverse proposals. Our proposal is a process to enable policymakers to discover which is right, either for the whole country or for a region.

Core Elements

We propose that Congress provide financial assistance and a legal framework to trigger a diverse set of federal-state initiatives. To help break the impasse in Congress over most national approaches, we propose steps designed to enable "first choice" political ideas to be tried in limited areas, with the support of states and through the enactment of a federal "policy toolbox" of legislated approaches that would be available to states but not imposed on them. Our view is that elected officials would be prepared to authorize some approaches now bottled up in Congress if they knew that the approach would not be imposed on their states. Our proposed strategy would contain six key elements.

Goals and protections. First, Congress would set certain goals and general protections. Goals would be established for extending coverage, and perhaps improving the coverage of some of those with inadequate coverage today. One such goal could be a percentage reduction in the number of uninsured people in a state. The more precise the goals, the more contentious they are likely to be. But clear and measurable goals under the proposed covenants are necessary if the system of financial rewards described below is to work effectively.

What is "insurance"? For a coverage goal to mean anything, it would have to define what constitutes "insurance." Specifying adequate coverage in health care is no easier than quantifying an adequate high school education, and when money follows success, drafting such definitions becomes even more difficult.

In defining what is meant by adequate insurance, agreement on two characteristics is vital: the services to be covered and the maximum residual costs (deductibles and copayments) that the insured must bear. States could be more generous than these standards. Instead of speciying precisely what states must do in each of these dimensions, we suggest that Congress establish a required actuarial minimum-such as the cost of providing the benefit package of the Federal Employees Health Benefits Program (FEHBP) for the state's population-as the standard, with states retaining considerable latitude on which services to include and how much cost sharing to require. Whether to set this actuarial standard high or low will be controversial and will determine the overall cost to the federal government of eliciting state participation.

Both high and low benefit standards suffer from well-known problems. High standards would raise program costs and weaken individuals' incentives to be prudent purchasers of health care. Low standards expose patients to sizable financial risk and raise questions about whether to restrict patients' right to buy supplemental coverage. Thus, federal legislation would not specify the content of insurance plans beyond some such actuarial amount. States would then be free to design plans as they wish, although certain types of plans might be presumptively acceptable (see below), and others could be negotiated as part of a covenant. The exact mix of benefits could vary within reason, but no further limits would be imposed. One goal of this approach, after all, is to encourage experimentation to generate information on whether particular configurations of benefits work better than others. It might turn out, for example, that states would adopt quite different plans with similar actuarial values. One group might opt for high-deductible plans covering a wide range of services with no cost sharing above the deductible and generous relief from the deductible for the poor, while others might adopt a system with low deductibles and modest cost sharing but covering a much narrower range of benefits. Discovering how individuals' and providers' attitudes and behavior differ under such plans and how health outcomes vary would provide valuable information for private health insurance planners and government officials.

Protections for individuals. In addition to the definitional question, the question also arises, What limitations and protections should be applied to state experiments? If a simple net reduction in uninsurance guaranteed a financial reward to a state, for example, the state would have the incentive to drop coverage of costly high-risk adults and extend coverage to less costly (healthier and younger) workers. Some such concerns could be addressed in negotiating covenants, but some broad protections and policy "corridors" would be established under our proposal and would be necessary to achieve political support.
One of the most politically sensitive would be a primum non nocere limitation. That is, states could not introduce a plan that reduced coverage for currently insured populations, most notably the Medicaid population, beyond some minimum amount. We believe that no reform proposal is likely to be achievable without that restriction. Most Medicaid outlays in many states are not strictly mandated by federal law, in the sense that some beneficiaries and some services for all beneficiaries are optional. States provide optional coverage because federal law permits it, and the federal match makes its provision attractive to states. If incentives were introduced to cover the non-Medicaid population, states might find it financially and politically attractive to increase the total number of insured people by curtailing Medicaid eligibility and benefits and using the money saved, together with federal support, to cover a larger number of people who are uninsured but less poor.

Designing and enforcing rules to prohibit or limit such "insurance swapping" would be extremely challenging but politically-and, one could argue, morally-essential. On the other hand, we believe that states should have some opportunity to propose different ways of delivering the Medicaid commitment to the currently insured population, as long as the degree and quality of coverage were not diminished. That form of Medicaid protection could stimulate creativity and improvement in coverage for the poorest citizens while avoiding any threat to their existing coverage. To be sure, there are disagreements, including between us, on the degree of freedom states should have in deciding how to deliver the Medicaid commitment. Positions range from only minor tweaking to sweeping changes in the delivery system, such as allowing states to use Medicaid money to subsidize individual enrollment in an equivalent private plan. The degree of flexibility states should have, while maintaining eligibility and level of coverage, is a difficult political issue for Congress to decide.

Acceptable state proposals would also have to limit cost sharing and features analogous to pension nondiscrimination rules. We believe that requirements, consistent with the general goals and protections we propose, are needed to ensure that lower-income households do not face unaffordable coverage. Without such limits, states could reduce the number of uninsured people and secure attendant federal financial support, for example, by instituting an individual mandate with a high premium that would effectively make insurance universal among the financially secure and do little for the poor. States would need to propose a fair, plausible way of meeting the requirement, such as by mandating some form of community rating or through a cross-subsidy to more vulnerable populations.

The federal government should establish broad guidelines, but no more. A key principle of our proposal is that state officials are more likely than federal officials to design successful solutions to those problems that members of the policy or congressional staff community have failed to solve. Congress can and should set the parameters, but it should avoid micromanagement.

"Policy toolbox" of federal policies and programs. A feature of the congressional impasse noted earlier is that many plausible health initiatives that might merit testing, and have support in some states, are blocked by other lawmakers who oppose the introduction of the approach in their own state or across the country. Thus, we propose that Congress enact presumptively legitimate approaches to the expansion of health insurance coverage as a "policy toolbox" that would be available to states à la carte to apply within their borders. Lawmakers could safely vote to permit an initiative, confident that it would not be imposed on their states. In this way, potentially useful policies and programs could be "unlocked" from Congress and become available for states to use in their own initiatives.

A policy toolbox likely would include expansions of existing policies, such as raising income limits under Medicaid or lowering the age of Medicare eligibility. It could include arrangements to subsidize individual buy-ins to the FEHBP, refundable tax credits or their equivalent (perhaps with some steps to modify the federal income tax exclusion for employer-sponsored health insurance costs), mandating employer or individual coverage, or creating a single state insurance plan though which everyone may buy subsidized coverage.

Other possible examples might include the following: (1) Remove regulatory and tax obstacles to churches, unions, and other organizations providing group health insurance plans. This could open up new forms of group coverage offered though organizations with an established membership and common values. (2) Allow Medicaid and SCHIP to cover additional populations, with greatly enhanced federal matching payments, and perhaps to operate in very different ways-with appropriate safeguards to protect those who are covered under current law. Both federal welfare legislation and SCHIP, for example, included safeguards to preserve existing Medicaid coverage. (3) Extend limited federal Employee Retirement Income Security Act (ERISA) protection to large corporate health plans willing to enroll nonemployees, and extend the tax exclusion to those enrollees. This could lead in a state to expanded access to comprehensive coverage. (4) Provide a voucher to individuals designed to mimic a comprehensive refundable tax credit for health insurance. This could allow the practical issues of a major tax credit approach to be examined. (5) Enact legislation to make forms of FEHBP-style coverage available to broader populations within states. This would enable states and federal government to explore the issues associated with extending the program to nonfederal employees and retirees. (6) Enable states to establish association plans and other innovative health organizations.

We emphasize that any menu of tools would be optional for states. None would be required. Members of Congress would be more likely to agree to the inclusion of elements they would deplore in their own states if they knew that no state, including their own, would be forced to adopt them than they would be in a nationally uniform system. Some lawmakers, for instance, oppose association plans because they believe that such plans would disrupt successful state insurance arrangements. Under the menu approach, association plans would be introduced only in states wishing to use them as part of their overall strategy.

State proposals, federal approval. Under our proposed strategy, states interested in a bold, creative initiative would design a proposal consistent with the goals and restrictions established by Congress. Typically this proposal would include some elements from the federal policy toolbox in conjunction with state initiatives.

Needless to say, a critical congressional decision would concern mechanisms for approving state plans and monitoring state performance. States would no doubt seek to take advantage of every financial opportunity to game the system and to stretch agreements to the limit, as the almost zany history of the Medicaid upper payment level (UPL) controversy makes painfully clear. Yet monitoring state behavior, determining state violations, and enforcing penalties on states is enormously difficult. Moreover, the entity could (and we think should) have the power to negotiate parts of a proposal, not merely approve or reject it, so that refinements could be made consistent with Congress's objectives.

But what entity should this be? It might seem natural to designate an executive agency that reports to the president, such as the Department of Health and Human Services (HHS). We suspect, however, that many members of Congress would refuse to cede so much selection authority to another branch of government and that roughly half would fear partisan decisions by an administration of the "other" party. Congress would likely insist on adding suffocating selection criteria and other restrictions to executive-department decisions, jeopardizing the very creativity we intend. Thus, we favor instead an existing or newly created body that has independence but ultimately answers to Congress. A new bipartisan body might perform this function with members selected by Congress and the administration or with members also representing the states, with technical advice from the U.S. General Accounting Office (GAO). This body would evaluate and negotiate draft state proposals according to the general requirements specified by Congress and then present a recommended "slate" of proposals to Congress for an up-or-down vote without amendment. Once the state proposals had been selected, HHS would be responsible for implementing the program.

Bipartisan willingness to authorize state programs and to appropriate sufficient funds to elicit state participation also requires that members of Congress believe that approaches they find congenial will receive a fair trial and agree that approaches they reject will also receive a fair trial. Unfortunately, current federal legislation makes two key approaches difficult to implement in individual states or even groups of states: a single-payer plan and an individual mandate combined with refundable tax credits. A federalist approach should include mechanisms that would enable states to give such proposals as fair and complete a test as possible, both because that would provide valuable information and because the political support of their advocates is important in Congress.

Crafting a single-payer experiment. ERISA, which exempts self-insured plans from state regulation, is the primary technical obstacle to testing single-payer plans. The political sensitivity to modifications in ERISA is difficult to exaggerate. Any attempt to carve out an exception from ERISA for state programs to extend coverage would probably doom federal legislation. But states could create "wraparound" plans to cover all who are not currently insured, or even to cover all who are not insured under plans exempted by ERISA from state regulation. While such an arrangement would not be a single-payer plan, it could achieve universal coverage, which is one defining characteristic of single-payer plans, and arguably be sufficient for a valid test. After all, the U.S. health care system is characterized by different subsystems for certain populations and has a form of single-payer coverage for military veterans. But of course the real test is whether advocates of single-payer plans regard such a limited arrangement as a fair trial.

An individual tax credit approach. The obstacles to a state-level individual mandate with a refundable credit are also serious and complicated. We presume that an individual mandate would require some contribution from people with incomes above defined levels. Such a mandate raises both political and practical questions. Testing federal tax reform in selected geographic areas also raises constitutional and practical issues, although advocates of the approach maintain that other site-specific programs involving federal tax changes, such as enterprise zones, have passed muster. In addition, for a limited experiment it might be possible to design subsidy programs that would mimic tax relief.

Administering a refundable tax credit would pose formidable difficulties for some states, particularly those that do not have a personal income tax. In all states, the logistics of providing a credit with reasonable accuracy on a timely basis would be challenging. So, too, would deciding how to address such administrative problems as households that live in one state yet work in another. Advocates for tax credits say they have solutions to these and similar challenges, just as supporters of single-payer approaches or employer mandates claim to have answers to challenges facing those approaches. For instance, some maintain that the employment-based tax withholding system could serve as a vehicle for refundable credits or equivalent subsidies and would make individual enrollment practical.⁴ Whether or not they are right is of course disputed by their critics. The beauty of a "put up or shut up" federalism initiative is that it offers a chance for advocates to offer such solutions in practice instead of in theory.

Using "managed federalism" to build support? Deciding how many states could qualify for experiments is an open political and technical question. One approach would be to limit it to a few states. This would limit costs but has little else to be said for it. Accordingly, we would favor opening the program to all states wishing to accept a federal offer. Nevertheless, we recognize that some lawmakers would be reluctant to vote for a process of federal-state innovation unless they were sure that certain "generic" or "standard" approaches were included-especially if the number of states in the program were to be limited. In particular, we believe that our proposal can win congressional support only if liberals and conservatives alike are fully convinced that the approaches each holds dear will receive a fair and full trial in practice.

While we believe that any state initiative that meets approval should be welcomed, political considerations thus might require that no state's proposal would be approved unless a sufficient range of acceptable variants was proposed. For example, strong advocates of market-based or single-payer approaches might find the federalism option acceptable only if each was confident that favored approaches would be tested

Adequate data collection. To determine whether a state was actually making progress toward a goal, accurate and timely data would be needed. These data would include surveys of insurance coverage, with sufficient detail to provide state-level estimates. Such surveys would be essential to show whether the states were making progress in extending health insurance coverage. They are vital to the success of the whole approach because payments to states (apart from modest planning assistance) should be based on actual progress in extending coverage, not on compliance with procedural milestones.

Congress should also assure that states report on use of health services, costs, health status, and any other information deemed necessary to judge the relative success of various approaches to extending coverage. Only a national effort could ensure that data are comparable across states. States' cooperation with data collection would be one element of the determination of whether a state was in compliance with its covenant and was therefore eligible for full incentive payments. The experience with state waivers under welfare before enactment of the 1996 welfare reform clearly illustrates the power and importance of such data collection. The cumulative effect of the reports showing the effectiveness of welfare-to-work requirements in reducing rolls, increasing earnings, and raising recipients' satisfaction transformed the political environment and made welfare reform inescapable.

Rewarding progress. Congress would design a formula under which states would be rewarded for their progress in meeting the agreed federal-state goals of extending insurance coverage. As experience with countless grant programs attests, haggling over such formulas can become politics at its grubbiest, with elected officials voting solely on the basis of what a particular formula does for their districts. Even without political parochialism, designing a formula that rewards progress fairly is no easy task. For one thing, states will be starting from quite different places. The proportion of states' uninsured populations under age sixty-five during 1997-1999 ranged from 27.7 percent in New Mexico and 26.8 percent in Texas to 9.6 percent in Rhode Island and 10.5 percent in Minnesota and Hawaii.⁵ Designing an incentive formula to reward progress amid such diverse conditions is both an analytical and a political challenge. Moreover, the per capita cost of health care varies across the nation, which further complicates the assessment of progress. The cost of extending coverage depends on the geographic location, income, and health status of the uninsured population. Having financial access may be hollow in communities where services are physically unavailable or highly limited. Extending coverage may require supply-side measures to supplement financial access.

We believe that the only way to design such a formula is to remove the detailed design decisions from congressional micromanagement. We suggest that Congress be asked to adopt the domestic equivalent of "fast-track" trade negotiation rules or base-closing legislation. Under this arrangement, Congress would designate a body appointed in equal numbers by the two parties, to design an incentive formula that Congress would agree to vote up or down, without amendments. Such a formula would have to recognize the different positions from which various states would start. Any acceptable formula would have to reward both absolute and relative reductions in the proportions of uninsured people. Whether financial incentives would be offered for other dimensions of performance and how performance would be measured constitute additional important challenges.

Sources of funding. Bleak budget prospects could cause one to give up on this or any other attempt to extend health insurance coverage broadly. But as recent history amply illustrates, the political and budgetary weather can change dramatically and with little notice. What funding approach would be desirable if funds were available? Under our proposal, the federal funding would be intended for several broad purposes: (1) A large portion of the money would be used to help states actually fund approaches to be tested. (2) Some funding (perhaps with assistance from private foundations) would provide national support and technical assistance to states. A model to consider for such support is the Health Resources and Services Administration (HRSA) State Planning Grants program, which both funds state planning activities and provides federal support and technical assistance. (3) Some funds would cover the cost of independent performance monitoring. (4) Some funds would be set aside to reward states for meeting the goals in their agreed-upon plan. Congress might consider an automatic "performance bonus" system similar to the mechanism used in welfare reform. Congress could also consider withholding the periodic release of part of a state's grant pending a periodic assessment by the independent monitor of the degree to which the state is accomplishing the objectives specified in its covenant. Only those states willing to offer proposals designed to achieve the national goals would be eligible for a share of the funding or for the menu of federal policy tools. A state could decline to offer a proposal and remain under current programs.

Federalism enables the states to undertake innovative approaches to challenges facing the United States. Federal legislation often grants states broad discretion in designing even those programs for which the federal government bears much or most of the cost. In health care as well as education or welfare, states have been the primary innovators. But the federal government limits, shapes, and facilitates such innovation through regulation, taxation, and grants. Such a partnership is bound to be marked by conflict and tension as state and federal interests diverge.

A creative federalism approach of the kind we propose would change the dynamics of discovering better ways to expand insurance coverage, just as a version of this approach triggered a radical change in the way states addressed welfare dependency. By actually testing competing approaches to reach common goals, rather than endlessly debating them, the United States is far more likely to find the solution to the perplexing and seemingly intractable problem of uninsurance.

An earlier version of this paper was presented at a conference convened by the Council on Health Care Economics and Policy, 19 September 2003, in Washington, D.C.

NOTES

1. P.S. Nivola, "Can the Government Be Serious?" in Agenda for the Nation, ed. H.J. Aaron, J.M. Lindsay, and P.S. Nivola (Washington: Brookings Institution, 2003), 485-452; and J.Q. Wilson, "Reflections on the Political Context," ibid., 527-549.
2. S.M. Butler, "Using Federalism to Spur Action on the Uninsured," Heritage Foundation Working Paper (Washington: Heritage Foundation, May 2003); H.J. Aaron and S.M. Butler, "How Federalism Could Spur Bipartisan Action on the Uninsured" (Paper presented to the Council on Health Care Economics and Policy conference, "Health Insurance Expansions 2004: Examining the Options," in Washington, D.C., September 2003); H.J. Aaron, "Template for Health Care Coverage," Washington Post, 25 November 2002; and H.J. Aaron and S.M. Butler, "Four Steps to Better Health Care," Washington Post, 6 July 2003.
3. J.M. Corrigan, A. Greiner, and S.M. Erickson, eds., Fostering Rapid Advances in Health Care: Learning from System Demonstrations (Washington: National Academies Press, 2002).
4. L. Etheredge, "How to Administer Health Insurance," Backgrounder no. 1516 (Washington: Heritage Foundation, 31 January 2002).
5. S. Raetzman, L. Craig, and C. McDougall, Reforming the Health Care System: State Profiles 2001 (Washington: AARP Public Policy Institute, 2001).

Henry Aaron ([email protected]) is a senior fellow in health economics at the Brookings Institution in Washington, D.C. Stuart Butler ([email protected]) is vice president for domestic and economic policy studies at the Heritage Foundation, also in Washington.