Although insurance reform in a voluntary system will not necessarily expand overall coverage or contain health care costs, strong reforms can set the stage for the broader, system-wide change that is necessary for the achievement of universal coverage and cost containment. State governments have been enacting and implementing varying degrees of insurance reforms for several years, and some limited reform may be enacted at the federal level in the 104th U.S. Congress. In the wake of a failed comprehensive reform effort, the strength of incremental insurance reforms will serve as a barometer of the national commitment to the eventual achievement of universal coverage. National reforms that call on the broader community to reasonably share the risks and costs of both healthy and sick persons can provide a cornerstone for the building blocks of universal coverage. Although important, reforms that solve "job lock" (the inhibition of employment mobility due to preexisting conditions) stop short of defining the broader community's responsibility for the sick cannot.

Insurance reform as a bridge to universal coverage requires some resolution of the dichotomous roles of private health insurance in the U.S. political economy. In this paper, the roles of private health insurance as both an economic commodity and a social welfare tool are considered. Next, the roots of the market's problems are examined, and the reforms and rules that have been proposed or undertaken to resolve the conflicting roles of private insurance are analyzed. Finally, the need for federal involvement in setting industry standards is considered.

Health Insurance as an Economic Commodity

As an economic good, all insurance is a vehicle for reducing risk and increasing peace of mind. As such, it carries a price. The development of an insurance product for any particular risk is subject to the same rules of supply and demand as any other commodity. On the demand side, the size and predictability of the risk affect whether consumers will pay to insure against it. The risk must be large enough to exceed the ability of most persons to finance it through personal savings, and it must be common enough to worry about. Supply-side factors that affect the availability of insurance include the insurer's ability to control adverse risk selection and "moral hazard" [2].

Adverse risk selection occurs when sick subscribers congregate in certain plans, making them more costly. Insurers rely on waiting periods, exclusions made on the basis of preexisting conditions (uncovered through medical underwriting), and the formation of random groups (for purposes other than insurance, such as business and professional associations) to control adverse risk selection. "Moral hazard" is the industry's term for the tendency of insurance to increase the risk for loss. For example, homeowner's insurance increases the risk for fire, and health insurance increases both utilization of health care and medical costs. Health insurers respond to moral hazard by providing disincentives to use health coverage, including co-payments, deductibles, and premiums adjusted for health status (as determined by medical underwriting or past claims experience).

The significance of health status goes beyond its role as a supply-side factor in an economic equation. Health status is integral to the principle of actuarial fairness [3]—according to which policy holders should pay according to their risks—that is deeply ingrained in the insurance industry. Stone [3] ascribes a Calvinist flavor to the principle of actuarial fairness. She suggests that the founders of the insurance industry, reflecting social norms, drew inferences about an individual person's insurability from that person's station in life and perceived moral character. Over time, these inferences influenced statistical measures of risk in determining insurability. Something specific about individual persons, usually their trade or profession, determined the insurability of their lives.

From an economic perspective, actuarial fairness is said to contribute to the efficient allocation of insurance resources. If persons pay according to their risks, they are paying the "right" price and can make accurate choices about the risk they seek to bear and the risk they seek to share—the level of coverage they want. From the perspective of the classical economist, these kinds of individual preferences are essential to the efficient allocation of resources. If low-risk persons are forced to subsidize high-risk persons through artificially set higher prices, then they may purchase less insurance or no insurance, resulting in inefficiently low levels of coverage. Conversely, artificially low prices for high-risk persons will encourage them to purchase too much insurance, resulting in inefficiently high levels of coverage. This, in turn, contributes to higher health care costs [4].

Health Insurance as Social Welfare

On the other hand, health care coverage is considered so vital that much of U.S. society believes that everyone must have at least partial access to this coverage. Unlike other forms of insurance, health coverage is directly related to survival, to life and death. The social belief that health insurance is different from other types of insurance and should be accessible to everyone is manifested in society's intervention—through public hospitals, Medicaid, and other mechanisms—into the deterioration of health of persons who cannot or will not purchase health insurance. However late, inefficient, or ineffective, some degree of mutual aid is triggered by sickness [3].

The social welfare view of health insurance also derives from the fact that individual control over health status is much more limited than individual control over other risk factors, such as those for fire or automobile loss. Although lifestyle choices certainly contribute to health status, many other factors—including genetic, biological, and environmental influences—are beyond the control of the individual. Researchers are finding that even smoking and dietary habits are attributable in some measure to biological influences. From this perspective, actuarial fairness through risk rating is unfair.

For other types of risks, society clearly supports actuarial fairness and risk rating; it is considered unfair to make everyone responsible for the practices—such as bad driving, the purchase of expensive cars and homes, and activities such as skydiving and race car driving—of the few. Moreover, segmenting these risks invests individual persons with responsibility that can lead to safer driving, safer buildings, and safer hobbies. When risks can be substantially controlled, risk segmentation helps insurance to function well as an economic commodity and even contributes to public welfare by reducing risky behavior. Risk rating enables individual persons to choose any combination of risk shifting, risk reduction, or risk bearing for a corresponding price, and the segmented markets that are formed as a result reach stable equilibriums.

The social welfare dimension of health insurance is perhaps most evident in the historical transformation of this insurance from a traditional risk protection product to a payment mechanism for routine health care. This is partly the result of tax incentives and historical influences that have encouraged employers to buy increasing amounts of health care, but the fact that health insurance was readily sold as a group product and that employees were eager to trade wages for health coverage reflects strong support for health insurance as a social welfare tool. As a financing mechanism, health insurance defies the logic of actuarial fairness. Unlike life, fire, and automobile insurance, health insurance is used over and over to pay for preventive and other minor services as well as critical care. To apply actuarial fairness through risk rating to this type of insurance is inherently contradictory and renders the insurance product ineffective. Taken to its logical extreme, actuarial fairness would separate risk by individual person, in which case the rational economic response would be to pay one's own health care bills. This, of course, would eliminate the market for health insurance altogether.

To the rest of the industrialized world, the essential nature of health care and the inherent difference between health risks and other risks are self-evident. Except for the United States and South Africa, all industrialized nations finance health care on a community-wide basis, and almost all members of the community essentially have equal access to the health care system. The concept of private health insurance—driven perforce by economic principles—as the major financier of a nation's health care is comparatively odd.

The American College of Physicians reaffirms its commitment to universal health care coverage. To that end, the College recommends reforms of the private insurance market that 1) harness the benefits of economic principles, including competition based on price and quality but not risk selection and 2) spread risk, financing, and access broadly across communities. The College's insurance reform recommendations are put forth not as independent goals but as stepping stones along the path toward universal coverage.

Because the objectives of the economic and social roles of private health insurance are deeply contradictory, implementing the reforms necessary to reconcile these roles is difficult both substantively and politically. An understanding of the historical underpinnings of the private health insurance market and its disintegration will illuminate reform efforts that seek to blend the two seemingly exclusive functions of health insurance.

The Roots of Modern Market Failure

Health insurance began unraveling almost as soon as it appeared, in the 1930s [5, 6]. Although the first health insurers, the nonprofit Blue Cross hospital plans, were conceived by providers as a way of paying them, "the Blues" approached health coverage as social insurance that encompassed the needs and resources of everyone: The Blues charged everyone the same premiums regardless of risk.

The success of these plans—various states had licensed 35 of them by 1945—showed the viability of health insurance to commercial life insurers, who had always considered persons with health risks to be uninsurable. Before the 1930s, the cost of medical care (as opposed to wages lost because of illness) was insufficient to generate demand for insurance, and the prospect of determining limits to care and to the insurer's liability (controlling for moral hazard) was perceived as unwieldy if not impossible. As the effectiveness and therefore the costs of health care increased in the 1930s, so did demand. Commercial life insurers entered the market, immediately undercutting the community rates of the Blues by picking off the best risks. Early on, the long-standing principle of actuarial fairness, implemented through medical underwriting and risk rating, justified a competitive strategy of carving out healthy and profitable markets. Paradoxically, the very tool designed to prevent adverse selection made it spread like cancer.

The phenomenal growth of employer-based insurance—more an artifact of history than a grand design—forestalled the problems caused by adverse selection. Employers have been encouraged to provide health benefits in several ways. For example, a wage and price freeze during World War II exempted fringe benefits, allowing employers to increase benefits as a recruitment tool; unions have viewed bargaining over employee benefits as a way of solidifying their power base; and tax law excludes benefits from personal income taxes. Employer subsidies kept healthy workers from bailing out of coverage, the size of the employee groups was large enough to dilute risk, and the random membership of employee groups kept risks and prices stable.

Although the individual market always included a dynamic of adverse selection, leaving some sick persons without access to insurance, problems with access became more noticeable as underwriting and risk rating spread to the small employer market. Originally, insurers rated all of their small employers together as one block of business, adjusting premiums only for age. Again, risk rating became a marketing tool as "upstart" small insurers found new market niches by picking off the best small-group risks through extensive medical underwriting. This forced larger insurers and many Blue Cross organizations to do the same; eventually, adverse selection overwhelmed the small-group market. Small businesses dropped coverage altogether, excluded unhealthy employees from coverage, or shopped aggressively for lower premiums if their group was healthy. The result was an ever-deepening spiral of adverse selection and access problems.

The Fix: Insurance Reforms

It is precisely the potential for the adverse selection spiral that makes meaningful insurance reform, in the absence of a mandate for coverage, so difficult. That is, inclusion of the sick in risk pools will increase premiums for the healthy, causing some of them to drop coverage and prompting the repetition of the destructive dynamics. The key question is, Will the spiral stop at some reasonable price and coverage level, at which an equilibrium can be established? Early evidence from some state reform efforts suggests an affirmative answer.

Reforms at the state level have been limited to the small-group and sometimes the individual markets, and the states vary in their definitional thresholds; most use 25 to 50 employees as the cutoff. This is partly by design: The access problems caused by risk avoidance are more severe in the individual and small-group markets, whereas the size of large employers offers an inherent protection against adverse selection. Focus on these markets is also the inadvertent result of the Employment Retirement Income Security Act's exemption of self-insuring, large employers from state regulation.

A discussion of insurance reforms can be divided into four parts: general market rules, rating rules, reinsurance and risk adjustment provisions, and the role of purchasing groups. This discussion focuses on the first two categories. Reinsurance and risk adjustment are vital to the success of insurance reforms but are beyond the scope of this paper. Purchasing groups are the topic of another position paper from the College [7].

General Market Rules

All but four states—Michigan, Pennsylvania, Alabama, and Hawaii—have adopted some sort of insurance market rules during the past 4 years. (Hawaii has a long-standing employer mandate that predates ERISA [Employee Retirement Income Security Act].) About half of the states have adopted general market rules, which address portability of coverage through job changes with rules on guaranteed issue, guaranteed availability, guaranteed renewal, and preexisting condition clauses. The others have combined market rules with rating restrictions [8, 9].

Guaranteed Issue or Open Enrollment

These laws require insurers who participate in the small-group market to accept all applicants, even those who are expensive to cover. Alternatively, some states have required only large insurers to guarantee issue; this is referred to as guaranteed availability.

Elements of guaranteed issue can include requirements that insurance be purchased for all or most employees of a business and that employers who choose to offer coverage make a certain contribution. These requirements guard against selection problems by retaining healthy employees in the group and preventing employers from excluding unhealthy employees from coverage.

The College supports guarantee-issue requirements on all insurers in all markets.

Guaranteed Renewal

These measures prevent insurers from dropping subscribers who become sick. Refusal of insurers to renew policies is disturbing because the motivation for the refusal is so obviously the search for market advantage in a process known as "churning." To attract subscribers, insurers may offer unrealistically low initial rates to small groups. As claims come in and losses accumulate—as they inevitably will because of statistical regression to the mean—the insurer drops the group, all the while seeking new small groups to "churn" [2, 3]. A variation on churning is durational rating, in which premiums are rated according to the duration of coverage—a kind of preemptive strike against losses [3, 10]. If a fair premium is set initially and the market reasonably controls for other causes of adverse selection, then eliminating sick subscribers is unnecessary because the random occurrence of losses will not make the group costlier [2].

The College supports guaranteed renewal requirements on all insurers in all markets.

Preexisting Condition Clauses and Exclusion Waivers

In a voluntary insurance market, some sort of inhibition is necessary to encourage persons to purchase coverage before they get sick. Preexisting condition clauses stabilize coverage for existing subscribers who have contributed to the pool while healthy. A risk pool that does not require members to insure themselves over time cannot remain stable; if purchasers can jump in at any time, rates will increase and healthy persons will drop out, setting off what policy analysts term a "death spiral" of adverse selection.

However, preexisting condition clauses, along with exclusion waivers, also keep sick persons away from physicians in the name of actuarial fairness. These devices are yet another way of deflating the expectation that private health insurance will serve the public interest.

Exclusion waivers are explicit. For a specified period, they exempt coverage for a specified body part or disease uncovered by medical underwriting. Exclusion waivers are more commonly applied to individual than to group policies. Unlike exclusion waivers, preexisting condition clauses require little specificity and are therefore more troublesome. These clauses are common even among large firms. A Foster Higgins survey of 2000 employers found that preexisting condition clauses were used in the indemnity plans offered by 61% of firms with fewer than 500 employees, 57% of firms with 10 000 to 20 000 employees, and 61% of firms with more than 20 000 employees. The percentages were even higher in preferred provider organization plans. About half of employers used preexisting condition clauses in their point-of-service plans [11].

Preexisting clauses have two dimensions: 1) the period of time before a policy takes effect, during which a subscriber can be deemed to have a preexisting condition and 2) the period after the policy takes effect, during which a subscriber must be treatment free. The length of these periods typically ranges from 3 months to 2 years, depending on the insurer and the policy. Length is one contestable issue that arises from preexisting condition clauses; another is vagueness, which allows insurers to engage in post-claims underwriting (which is far cheaper than advance underwriting). Preexisting condition clauses can be invoked for undiagnosed conditions that are unknown even to the subscriber, and courts have upheld this right [12]. Stone [3] aptly describes the preexisting condition clause as a wild card.

To provide portability of coverage and to solve the problem of job lock, current and recent reform proposals at both the state and federal levels limit preexisting condition clauses for recently insured persons. In the 104th Congress, these provisions are contained in at least three bills: One is sponsored by Senate Labor and Human Resources Committee chair Nancy Kassebaum (R-KN) and Ranking Minority Leader Ted Kennedy (D-MA); one is sponsored by Senate Minority Leader Tom Daschle (D-SD); and one is sponsored by House Ways and Means Health Subcommittee Chair William Thomas (R-CA) [13-15]. The Kassebaum bill, which was unanimously approved by the Senate on 23 April 1996, gives credit against exclusion periods for previous coverage. Other bills allow a brief insurance gap before exclusion clauses are applied. The Physician Payment Review Commission, an advisory body to Congress, has recommended allowing a gap of 1 to 3 months [16].

The College supports limits on preexisting condition clauses and exclusion waivers in both the individual and small-group markets and for those persons moving between these markets. Exclusion of coverage for a maximum of 1 year for conditions existing as long as 6 months before coverage would serve as a reasonable disincentive to remaining uninsured. As a protection against retrospective underwriting, an exclusion should not be enforceable for a condition unless the condition was actually treated during the applicable period before coverage.

Preexisting condition clauses and exclusion waivers should be prohibited altogether for previously insured persons. The term "previously insured" should include insurance applicants covered within the previous 6 to 12 months or within the previous 18 months if the coverage gap is due to a job layoff. (This relief period is based on statutory precedent in the Consolidated Omnibus Budget Reconciliation Act of 1986, which allows laid-off workers to purchase health coverage for as long as 18 months from their ex-employers at group rates.)

Standardized Benefits

Insurance reforms commonly require insurers in the small-group and individual markets to offer one or two standardized insurance plans, one basic and one more comprehensive. The sale of supplemental policies is usually permitted. Some state reforms also require insurers to make available to the individual and small-group markets the same plans they sell to larger groups. The intent of these requirements is to prevent risk selection through plan design, to make insurers and employers focus on delivery system efficiencies rather than on risk selection, and to foster similarity and price competition among plans. Standardized benefit packages are also important to the functioning of voluntary purchasing groups [7].

The College supports requirements for standardized benefit plans, including one comprehensive plan. Insurers should make all of the plans they sell available in all markets, including voluntary purchasing pools.

Rating Rules

Market rules are far more meaningful if combined with effective rating rules. Guaranteed issue and renewal mean little in a market that offers stratospheric premium prices.

The College recommends limiting premium variation to the following factors: geography, family composition, plan design, age, and group size, which should be phased out over time. Variation in age ratings should follow the 1995 model law developed by the National Association of Insurance Commissioners. This model allows rating by 5-year age increments, beginning at age 30 years and ending at age 65 years, and the age factor for the oldest group may be no more than twice that of the youngest group. The College opposes the use of health status, claims experience, sex, duration of coverage, industry, and occupation as rating factors. These additional factors segment the market into ever-smaller communities that exclude those who most need health care, allocating health resources according to risk and payment ability rather than health care needs.

Definitions of community rating vary, but the College essentially supports modified community rating. Pure community rating is commonly defined as that which allows adjustments only for family status, geography, and plan design; modified community rating is commonly defined as that which allows age and sex to be used as rating factors or as that which allows limited variation by age; sex; and variables other than health status, claims experience, and duration [17, 18].

At least 19 states have adopted community rating laws. Some (New York, New Jersey, Massachusetts, Vermont, and Maine) truly compress the spread of rates. Others permit so many rating classes (or rating bands) that "community" seems inappropriate as a descriptor. Moreover, the extent of variation permitted within and among the classes can be so wide that the classes are ineffective, if not meaningless.

A model rating scheme proposed by the Health Insurance Association of America (HIAA) provides an example of this. The model, which some states have adopted in whole or in part, permits any combination of rating factors forming bands limited to a 70% spread, with an additional 30% spread allowed for industry. There is no limit on the variation allowed between premium classes, which are multitudinous because of the use of age as a rating factor. The result is a more than 10-fold difference in the rates between the extreme bands. For a plan with an average annual rate of $1500 per single enrollee, a group of three healthy 28-year-old male computer software engineers might pay $1865 a year, whereas a group of three sickly 58-year-old physicians could pay $30 555 [19]. Eliminating health status as a rating factor would compress these rates somewhat. Rates could also be compressed by allowing health status to be used as a rating factor while sharply limiting its variation. California and Oregon allow health status as a variable but limit variation; this compresses rates substantially.

In addition to permitting variation within and among bands, the HIAA model allows unlimited variations among insurers' so-called books (or blocks) of business. Insurers have traditionally separated inherently different insurance products (such as health maintenance organizations and indemnity) or products sold by different sales forces into separate books of business. Some state laws limit price variation among books of business just as they limit it among rating bands, but the HIAA model does not. The absence of such limits creates an opportunity to circumvent rating rules by gerrymandering books of business, for example, by putting all older persons in one book. Books of business can also be used for the durational rating of groups: When the risk of a healthy group increases only because time passes, the group is dropped, and the reason given is that the book of business has been closed. This alleged practice gives new meaning to the phrase "to close the book on."

Modified community rating is also supported by the Physician Payment Review Commission and the National Association of Insurance Commissioners (NAIC), which is the only national influence on state insurance regulations. For several years, NAIC has played an influential role in encouraging states to adopt market and rating reforms through the development of model laws. In 1995, NAIC tightened rating rules, eliminating industry as a permissible rating factor in its model regulation and tightening recommended limits on the use of age and sex as variables [20, 21]. The NAIC model limits the allowable variation within a premium class to a 50% spread and, more significantly, limits the rate difference between the lowest and highest premium rates to 20%. Annual increases can be no more than 15% higher than an insurer's trend, as measured by the rate increase for new business; this is considered the most competitive.

Effectiveness of State Insurance Reform

It is too early to draw definitive conclusions about the effects of insurance reform, especially rating reforms, but two studies [4, 18] have given some early indications that strict rating rules, combined with other market rules, have been successful. That is, these rules have expanded coverage to include less healthy persons and may have expanded overall coverage in some small-group and individual markets without hemorrhaging healthy persons from the insurance pool and further destabilizing markets. Indeed, the reforms seemed to have increased market stability.

A study supported by the Commonwealth Fund and conducted by the Intergovernmental Health Policy Project at George Washington University in 1994 [18] examined the effect of community rating (both pure and modified) in the five states that have used it longest—New York, New Jersey, Massachusetts, Vermont, and Maine. These states had each had about 1 year of experience with the rating reforms. Another study [4], funded by the Robert Wood Johnson Foundation and conducted by the Alpha Center, studied New York and Vermont. The findings were similar.

1. The rating reforms worked as expected to compress rates: High premiums decreased significantly, and low premiums increased less significantly (in dollar amounts). More policy holders experienced price increases than price decreases, a statistical expectation. A mitigating factor is that those who had the greatest rates of increase were those who had had the lowest rates to begin with; thus, the dollar amounts of the increases were tempered. (By definition, community rating will increase rates for persons below the mean because the distribution of rates in an unregulated market is such that the number of persons who pay less than the mean is greater than the number of persons who pay more than the mean. Also, by definition, the rate of increase for the cheaper policies will exceed the rate of decrease for the higher-priced policies because the base on which rates are figured—the denominator—is lower for the cheaper policies. Thus, comparing rate increases and rate decreases may create a distorted perception of the results of rating reforms.)

2. Except in New York, premium increases were smaller than anticipated, and the rating reforms may have generated increased price competition in the small-group market as a result of diminished ability to risk select. Assuming that an overarching goal of health care and insurance reform is to encourage insurers to compete on the basis of price and efficiency of delivery, it is telling, if not humorous, that one carrier asked a state insurance department for permission to offer a nonstandard plan because "otherwise the only way we can compete is on price."

3. Carrier flight was minimal and was confined largely to small insurers who covered relatively few persons.

4. The effect on overall coverage could not be ascertained because of the difficulties of accounting for movement into and out of different coverage markets. For example, some small companies may have responded by self-insuring; some very small companies that had been forced to rely on individual policies may have been able to obtain small-group policies; and small companies that had excluded sick employees may have included these employees after the enactment of reforms. Rating reforms may have resulted in coverage of fewer lives but more expenditures.

New York, which was the first state to implement pure community rating, may provide the best example of the dynamics of insurance reforms. In the first 9 months after the enactment of this reform, coverage in the individual and small-group markets declined by 1.2%. However, overall state coverage had declined by the same amount in the year before enactment. The decline in coverage after enactment included an increase of 4286 small-group policies (0.5%) and a decrease of 43 666 individual policies (12.4%). Coverage declined even more in the Blues plans, which had always been community rated.

Whether New York's reforms directly caused substantial numbers of healthy persons to drop their coverage is unclear because the movement of persons between markets was not tracked. Younger men may have made up the bulk of those who dropped coverage. Mutual of Omaha reported that 80% of its individual "droppers" were younger than 45 years of age and that 60% were male.

In the first year after reforms were enacted, the average premium increase in the small-group market was 4.6%. In the year before reforms, commercial insurers had requested an average price increase of 19%, and Empire Blue Cross and Blue Shield had requested a 28% increase. Enactment of reforms resulted in decreased premium rates for 34% of small groups and increased premium rates for the rest; 18% of small groups had increases of more than 40%. According to the New York State Insurance Department, carriers initially set prices high and then decreased them substantially during the first year of reform because of competition.

Also of interest were findings in New Jersey and Vermont. In New Jersey, where reforms included the individual market, 20 new carriers and 11 000 previously uninsured persons entered the fully community-rated individual market. In Vermont, small-group coverage increased by 15%. An outside evaluation by William M. Mercer, Inc., reported that the reforms had greatly expanded coverage, caused a small overall cost increase, and stabilized premiums.

The Federal Role in Insurance Reform

The federal government has the sole constitutional right to regulate insurance as interstate commerce, but the U.S. Congress deferred regulatory authority to the states by statute 50 years ago [22]. Currently, insurance reforms lie within the prerogative of state regulators, who have adequately established general market rules, such as guaranteed issue and renewal. However, insufficient rating restrictions in many states make it unlikely that substantial numbers of higher-risk persons and small employers will purchase coverage as a result of reforms.

Federal participation in insurance regulation would do more to improve access to care for higher-risk groups and persons than would states acting individually. The College supports a federal minimum standard for both market and rating reforms, based on the recommendations contained in this paper. States wishing to establish stronger standards could do so.

Federal involvement would ease the ability of interstate corporations to comply with insurance regulations. It would also make coverage portable across state lines, allowing, for example, a person from Maryland to take a job in Virginia and continue to receive coverage without regard to preexisting condition clauses. Finally, federal involvement would facilitate the collection of useful data with which to study the effects of policy change on health coverage.

The need for federal intervention stems from some inherent limitations on state insurance regulators. Insurance is a very local business, and state legislatures are very locally represented. Three percent of legislators are themselves insurance agents, and many more are business persons who have long-standing relationships with the agents in their communities [23]. Small businesses and agents perceive their self-interests to be similar, even though agents' incentives encourage them to sell high-cost policies and congregate higher risks into separate high-cost pools. Agents' commissions are sometimes tied to their loss ratios so that low claims expenditures increase commissions. This is an incentive for agents to segregate apparently high-risk applicants even before submitting applications to a carrier for underwriting. Moreover, the health insurance industry—both carriers and agents—are experienced and skilled at molding state reform efforts to their own interests. The expectation that the federal government may be more formidable partly explains the support of health care providers for federal rather than state certification of Medicare risk-bearing provider health plans in the federal budget debate.

Lastly, federal assumption of a role in insurance regulation is a requisite step in advancement toward the goal of universal coverage. A federal role, even if initially confined to portability provisions, by definition expands the protected community of insured persons beyond state borders, enforcing the charge that private health insurance be socially responsible. Ultimately, however, a national commitment to both portability and strong rating reforms is needed to establish insurance reform as a foundation for universal coverage.