Health Insurance Market Reforms

Linda J. Blumberg, Len Nichols

---

The nation rejected a comprehensive restructuring of the U.S. health care system in 1994, but reform of the health insurance market is still very much on the table. Insurance reforms can lower costs for some and enable an imperfect market to work better for many. Still, the danger with the current discussion of this issue is that it often neglects two important points. First, insurance reform can make limited improvements in increasing the proportion of Americans with health insurance in a voluntary market but cannot be expected to significantly reduce the rate of growth in system-wide health spending. Second, and even more important, piecemeal reform without the proper safeguards could actually make things worse.

The fundamental problem with a voluntary insurance market is the possibility of risk segmentation—disproportionate numbers of persons with higher than average risk of health problems being congregated in particular risk pools. The principle behind insurance is to spread individual risks across a group. In a large and diverse risk pool, the premiums paid by those with better than average experience go to cover the costs of those with worse than average experience. Low risk individuals might accept this cross-subsidy because they want to be able to buy coverage at the pooled community rate when they become worse than average health risks in the future. Risk segmentation concentrates the high risks in one pool and the low risks in another, leading some to face premiums systematically higher than the actual risk they face or to be unable to buy insurance at any price.

Our intent here is to clarify the meaning of the terms used in the insurance reform debate, and briefly review the current state of the health insurance market. We then proceed, on the fundamental principle that change should do no harm, to lay out what tools are available for reforming the health insurance market. We also define packages of reforms that can improve the market without aggravating selection problems. We describe these packages in ascending order of aggressiveness in terms of the changes that would be required.

• The Health Insurance Market Today
• Insurance Market Reforms Defined
• Selection Bias Effects in Reformed Voluntary Markets
• Goals of Reform and the Tools Available for Achieving Them
• Four Insurance Reform Packages
• Conclusions
• About the Authors

The Health Insurance Market Today

Issues of insurance reform are almost entirely issues related to risk segmentation. The basic fact underlying risk segmentation in health insurance markets is the highly skewed distribution of health expenditures (Figure 1). The most expensive 1 percent of our population accounts for 30 percent of all health spending. The most expensive 50 percent of the population engenders 97 percent of total national health spending, meaning that the least expensive 50 percent account for only 3 percent of spending. This disparity in the financial consequences of differential health risks absolutely dwarfs any feasible savings from managing the care of the sick. When coupled with the voluntary character of insurance markets in the United States today, it is clear why health insurers have extremely strong incentives to identify and insure below-average risk populations. In the absence of specific rules governing market rules and conduct, competition for good risks will be intense. This paper will explore the consequences of risk selection and explain the rules that could improve insurance market performance.

The lower the health risk of an insured group, the lower the expenses to the insurer, and the lower the premiums the insurer must charge in order to be profitable. Lower premiums, in turn, make the group-specific insurance product more attractive to purchasers. But while risk segmentation of the insurance market provides short-run benefits for low-risk populations, higher risk populations may find it prohibitively expensive or impossible to obtain insurance for themselves and/or their families. In addition, individual members of today's lower risk groups may become part of tomorrow's higher risk population due to their own or a family member's future illnesses or injuries. Aging will inevitably increase the risk associated with insuring any population, making long-term insurability an uncertainty for all individuals in a market organized like our current one.

The ability to segment the insurance market has created a multi-tiered system of coverage (Figure 2). Coverage is found in both the public and private sector, though public sector coverage is limited to particular categories of individuals. The elderly (those age 65 and over) as well as disabled persons meeting specific eligibility requirements receive coverage through the Medicare program. Some groups of low-income persons are eligible for benefits through the Medicaid program, with eligibility and included services varying significantly by state. Other public programs, such as the military health system,[1] provide insurance for some narrowly defined populations. Another example of public coverage is state-sponsored high-risk pools. Twenty-five states had operational risk pools as of 1993,[2] providing insurance to individuals who, as a function of pre-existing illnesses or conditions, have been denied private insurance coverage or who have difficulty obtaining comprehensive coverage at a rate below that offered in the high-risk pool.

Private sector coverage includes those people who are ineligible for public sector coverage (either at any time during the year or for part of the year), people who choose to supplement their public coverage with additional private insurance, such as Medicare beneficiaries who purchase Medigap policies to cover Medicare's cost-sharing requirements, and people who choose to purchase private coverage despite being eligible for public programs (e.g., certain military health and Medicaid eligibles). Individuals in the private sector are sorted into four basic tiers of coverage. First tier private sector coverage is through an Employee Retirement Income Security Act (ERISA) protected plan. Under ERISA, firms, multi-employer welfare arrangements (MEWAs), or associations of individuals choosing to self-insure—i.e., to bear the risk of health insurance for the enrollees, usually purchasing a financial stop-loss arrangement—are exempt from state insurance laws. For example, self-insured plans are not required to abide by state mandated benefits laws—they may include any benefits that they choose. In addition, self-insured plans are currently exempt from state premium taxes that are intended to subsidize high-risk or uninsured populations. Large groups and groups with a better-than-average risk profile are attractive candidates for self-insurance.

The second tier of private coverage comprises groups that purchase commercial experience-rated insurance through actuarially balanced risk pools. These are firms of 500 to 1,000 or more workers. Given the size of these groups, a single high cost case in a given year would not be able to destroy the financial viability of the group. With large numbers of people, the vast majority of whom are not high cost, the random risk of an individual having a high-cost year is averaged over the whole group, implying low variation in expenditures per person in the group.[3]

The third tier of private coverage includes those groups, often small or moderately sized firms, which purchase experience-rated insurance, and those purchasing individual policies. This tier we refer to as the "persistence of risk rating" group. These groups are not large enough or diverse enough to be considered an actuarially balanced risk pool. At any point in time the group may be more or less expensive than average, and as a consequence of their size have little bargaining power. Premiums for these groups are subject to abrupt upward shocks, the effects of which can persist for a long time, while groups with extended periods of low utilization may see premiums stable at significantly lower levels.

Inefficiencies arise in tier 3 because insurers tend to rate small groups with a high-cost year as being high risk in the following years, regardless of whether or not the high-cost health care episode is persistent. While in a perfectly competitive market one would expect the group to find an alternative insurer willing to issue a plan at a more actuarially fair rate, this does not seem to be a frequent result in the current market. A recent survey found that two-thirds of small firms that dropped coverage did so because the premiums they could obtain increased substantially.[4] Economic theory suggests that it does make sense for the high premiums to persist, because the gains to an insurer of selecting groups with relatively healthy experience are much greater than the net gain resulting from analyzing a group with recent high-cost experience in order to determine if the high-cost episode is completed or unlikely to recur. Consequently, small groups or individuals with even a single "bad" year can find themselves continuously penalized with high insurance premiums and some have difficulty finding vendors.[5] These problems are particularly severe for individual purchasers who have the least opportunity to pool risk with others. Insurers perceive individuals seeking health insurance on their own as very high risk.

The fourth tier of "coverage" is the complete absence of insurance. This group comprises those unable to purchase insurance in the private market (including many who were once in the third tier of coverage and were later dropped by their insurer) and those choosing not to purchase insurance.

Insurance Market Reforms Defined

Insurance market reforms can be thought of as mechanisms intended to "level the playing field" for purchasers of insurance, as well as for imposing long-run stability on the character of insurance. Although the objective of such policies is to reduce the risk selection in the current marketplace, if done inappropriately reforms have the potential to increase the selection problems relative to what is seen today and threaten the existence of a private insurance market.

The Rules of Issue

Guaranteed issue, renewability, portability, and limits on pre-existing condition exclusions often tend to be grouped together in discussions of insurance reforms. The appropriateness of implementing one (or some) without the others will be addressed in a later section.

Guaranteed issue is a requirement that insurers sell a health benefit plan to any eligible party that agrees to pay the applicable premiums and to fulfill the other plan requirements. It can be thought of as a "take all comers" rule. The term "eligible party," as used here, means that a guaranteed-issue rule can be written to apply only to firms of certain sizes, individuals, or other market subgroups. In general, guaranteed-issue rules pertain to specifically defined open-enrollment periods each year, in order to discourage individuals/groups from waiting until an expensive illness or injury occurs before enrolling. Guaranteed issue alone does not regulate the premium charged to a given individual or group. In other words, absent other regulations, it could allow insurers to charge enrolling individuals/groups substantially different premiums for the same plan, based upon the insurer's expectation of future incurred expenses and their confidence in that expectation.

Complete guaranteed-issue rules require that any insurance product offered by the insurer in the relevant market be open to all applicants. Some states apply guaranteed issue only to a few specified insurance products. Without concomitant price regulations, this kind of guaranteed issue can be circumvented by charging a high price for the guaranteed product and indicating the existence of other alternatives only to good risks. Guaranteed-issue policies are theoretically designed to eliminate the problem of uninsurables—those groups whose expected risks are sufficiently high that most insurers refuse to offer coverage to them, regardless of the price. Such insurance "red-lining" practices are seen today for workers in industries as diverse as cab companies, hair salons, and mining operations.[6] Such policies alone, however, would not do away with the problem of "economic uninsurables"—those high-risk groups that are in practical effect uninsurable because the premium rates demanded for their coverage are so high.

Guaranteed renewability ensures that those currently covered by a particular firm cannot have that coverage discontinued by their current insurer in a subsequent continuous year as long as that insurer continues to do business in that particular market.[7] Such rules are intended to eliminate practices where insurers refuse to renew coverage of groups or individuals once a substantial expense is incurred. As with the guaranteed-issue rules, however, guaranteed renewability alone does not constrain the premium that can be charged a covered group or individual. Following a serious illness, a small group or individual could, therefore, face increases substantial enough to make continuing coverage unrealistic. Thus, many states restrict the range of annual premium increases in the small group market—though in practice, most ranges are quite large.

Pre-existing condition exclusions occur in today's market in a number of forms. Some insurers, most notably those issuing policies to individuals, permanently disallow coverage of treatments related to any previous conditions.[8] Many group policies have pre-existing condition waiting periods, whereby new enrollees are excluded from coverage pertaining to such conditions for a specified period of time. Limits on pre-existing condition exclusions would serve as maximum time periods for which conditions could be excluded. Most often, reform proposals set these limits at 6 to 12 months. Eliminating pre-existing condition exclusions completely in a voluntary market would undoubtedly lead to horrific adverse selection problems. Without any restrictions there would be no incentive to purchase insurance until the onset of an illness or injury; consequently, premium prices would escalate dramatically.

Portability is, on its own, a form of a limit on pre-existing condition exclusions. Under such rules, those individuals maintaining continuous coverage would be exempt from all pre-existing condition exclusions applying to new policies. The objective of such a rule is to decrease the problem of "job-lock". Empirical studies[9] indicate that job mobility could increase substantially in the absence of pre-existing condition exclusions. Portability would allow individuals to move from one job with employer sponsored insurance to another job which offers employer sponsored insurance without being subject to pre-existing condition exclusions. What portability alone does not provide is a guarantee that job changers could continue coverage in their particular health plan, nor does it guarantee that the job changer will be offered coverage at his or her new job. Though portability could be a valuable policy for workers who would like to move from one insuring job to another but fear losing coverage for an illness that they or one of their dependents has, portability does not mean that all the currently insured have the right to retain their current plans after a job change. Some continuation coverage is available, however, as a result of the Consolidated Omnibus Budget Reconciliation Act (COBRA) of 1987. Among other reforms, COBRA requires that employers of 20 or more workers offer continuation of the firm's insurance for up to 18 months to workers separating from the firm. The worker choosing this coverage pays an amount set by the employer not to exceed 102 percent of the premium previously paid by the employer and the employee.[10]

Mandated Benefits

There are two basic motivations for specifying benefit package provisions. One is to facilitate comparison shopping by consumers, by ensuring that standard services and providers are covered by all plans. The other is to simply guarantee that certain benefits (e.g., mental health) are available to all the insured. Both motivations preempt absolute market freedom. The conditions under which competitive market performance is enhanced or not are outlined in the next section (page 17). This sub-section will simply describe the three major policy options: standard benefit packages; mandated benefits; and minimum suggested benefit packages. In addition, we will describe a medical savings account and its potential use in conjunction with a high-deductible or catastrophic insurance policy.

Standard benefit packages specify exactly which services and which providers are covered, and what cost-sharing obligations are imposed on the insured. The details may be left to commissions or professional bodies acting pursuant to legislative guidance. Most analysts agree that standard benefit packages would facilitate comparison shopping and encourage competition among health plans.[11] In addition, standard benefit packages are probably a prerequisite for effective risk adjustment, a necessary adjunct to community rating, both of which are discussed below. Most of the Federal reform bills in the 103rd Congress included standard benefit packages, as do some state laws creating purchasing cooperatives.[12] In addition, some purely private cooperatives have specified the details of the benefit package which plans must offer in order to sell to the group members.[13]

Mandated benefits laws predate the recent comprehensive reform debate, and represent attempts to guarantee that particular health services are covered in every insurance product sold in that state, regardless of the other provisions of the insurance policy. These benefits range from inpatient alcohol and drug abuse programs to the services of chiropractors.

Minimum suggested benefits laws represent attempts to partially accomplish both goals. If the minimal package became the real floor in a state, then a step would have been taken toward standard benefit packages without a mandate. Similarly, minimum benefit packages represent a partial consensus on what benefits should be included in every policy. At the same time, minimum suggested packages permit a variety of benefits to be offered, so that specific packages may be tailored to attract low-risk individuals and maintain the risk segmentation already in the market. Finally, if the packages are merely suggestive, there is no guarantee that any particular service will be universally available even to the insured population.

A medical savings account (MSA) is a fund, set up and owned by an individual, that can be drawn upon to finance uncovered medical expenses as the need arises. If they choose to set up an MSA, most individuals would also purchase a high-deductible or catastrophic health insurance policy (perhaps as a condition of a tax-preferred status) to cover all expenses in a given year above the threshold deductible amount (e.g., $3,000). Recent legislative proposals, at both the federal and the state levels, would make income used to create the MSA exempt from income and payroll taxes, as are employer contributions for employees' health insurance premiums currently.[14] Unused year-end balances in the MSA fund might be available for other uses, with a penalty or not, or allowed to accumulate and earn interest. Requiring that a particular MSA-catastrophic combination be made available to workers, or dictating the conditions if it is made available (as most Congressional bills do), is equivalent to mandating the cost-sharing requirements of a benefit package, as opposed to the services or providers covered.

Community Rating

Two basic options currently exist for setting the premiums faced by purchasers of private health insurance. The first, and most widely seen in today's market, is experience rating. Under experience rating, insurers use characteristics of the insured group, including past patterns of health service utilization for that group and other groups similar to it in composition, to determine the applicable premium level. In this way, insurers are able to charge groups that are older and that have had above average spending in the recent past more than they charge other groups.

Community rating, on the other hand, is more akin to charging a premium that is averaged over an insurer's entire book of business. In its pure form, community rating would allow price differences based only upon geographic location, benefit package, and family size. In this way, the higher costs of less healthy groups are spread over all of the groups insured by the same company; the result being that in any one year healthier groups pay more than they would under experience rating, and less healthy groups pay less than they would otherwise. This rating option essentially eliminates price variation based upon health status and risk within the block of a company's insureds.

In a world of voluntary insurance coverage, however, pure community rating can be problematic. These issues will be discussed fully in the section on adverse selection. Modified community rating, however, is another option. This alternative entails adjusting pure community-rated premiums to take into account a limited set of differences across insured groups or individuals. Age rating, for example, sets broad age bands, allowing premiums to vary across those bands; the differences can be confined to certain boundaries. For example, the highest age band can be required to be no more than three times the lowest age band. While age rating is the most widely considered, rating bands can be determined along other parameters as well (e.g., gender, higher risk versus lower risk industries).

Risk Adjustment

Given the skewed distribution of health expenditures illustrated in Figure 1, there are already enormous incentives for insurers to select the healthiest enrollees. If community rating were implemented, the health plan which ended up with the healthiest enrollees would do considerably better than the health plan which managed to attract the least healthy enrollees. Without any other change, the incentive to select risks could actually increase, for community rating is typically accompanied by guaranteed issue and other reforms which would make it harder for insurers to protect themselves than in the current environment.

Risk adjustment is a mechanism for spreading among all plans the above-average costs of the bad risks, so that the consequences of poor risk selection are ameliorated and the incentives to engage in aggressive risk selection are reduced. It is unlikely, however, that risk adjustment could ever perform perfectly, so some incentive to select risks will remain. There are two polar extremes of risk adjustment—prospective and retrospective—as well as "blended" combinations of both. In this sub-section we describe each type of risk adjustment generally, and offer a detailed example of how risk adjustment might work in practice in the following section.[15]

A prospective risk-adjustment mechanism takes into account the objective risk factors a person or group brings to the combined risk pool (e.g., age, gender, past utilization) and adjusts payments to the health plan according to the differential risk profile their enrollees reflect. People and their employers would pay their community-rated premiums, but part of the payments would be transferred among insurers to reflect differential expected claims costs, given their actual enrollees. The key concept is adjusting for differential ex ante or expected claims costs, as opposed to actual experience.

A purely retrospective risk-adjustment mechanism is one that adjusts payments to health plans based on ex post claims experience. Reinsurance mechanisms are an example.[16] An assessment is made on each insurer, usually as a fixed percentage of premium revenue. Then, for all patients whose claims' exceed a predetermined amount, the insurer can draw from the pool. This dampens the financial penalty from enrolling individuals who turned out to be sicker than average. This mechanism can also be tailored for sharing the costs of specific high-cost medical conditions.[17]

A blended risk-adjustment mechanism is a combination of both the prospective and retrospective approaches. It would adjust premiums received by plans based on predetermined factors (e.g., age or prior utilization) that are believed to predict expected claims costs. In addition, acknowledging our limited technical ability to predict expenditures in any given year, the blended risk adjustor would enable some of the cost of extremely high-cost cases to be recouped from a reinsurance pool, with an appropriate incentive (e.g., a threshold claim size) to manage high-cost cases efficiently. In this sense the blended risk adjustor combines the best we can do at the present time on prospective risk adjustment with a promise to largely but not completely recompense insurers who end up with financially disastrous cases.

Regulation of Marketing Practices

All states have prohibitions on unfair trading which apply to the business of insurance, prohibiting false and misleading claims and the like. Reforms in this area tend to be specifically targeted at marketing and advertising practices that may serve to enhance risk segmentation. For example, insurers may be prohibited from offering brokers or agents a bonus for directing low-risk individuals to them. Insurers may also be prevented from avoiding certain neighborhoods where a higher percentage of higher risk individuals are believed to live.[18] Advertisements could be made subject to administrative approval to maximize the factual content and to minimize targeting to healthier groups.

Purchasing Cooperatives

As many have noted,[19] small groups and individuals are disadvantaged purchasers of health insurance. Purchasing cooperatives (sometimes called Health Insurance Purchasing Cooperatives, or HIPCs) can redress these disadvantages when they are designed to accomplish two major goals: (1) achieving economies of size, i.e., enabling small groups and individuals to purchase insurance with the same administrative efficiencies and bargaining power that only larger groups can achieve on their own; and (2) reducing risk segmentation (adverse selection) problems in the small group and individual insurance markets. The second motivation is more ambitious than the first, and requires more extensive reforms, including information gathering and dissemination.

Purchasing cooperatives would, in practice, contract with health plans and eligible employers,[20] collect and provide plan-specific information, market plans to eligible employers and individuals, handle enrollment, and manage the flow of funds from employers and individuals to health plans. The purchasing cooperative could also perform risk adjustment for health plans selling to cooperative members. In short, purchasing cooperatives could act in much the same way that many large firms' benefit management departments currently act as purchasing agents for their workers.[21] Under broader reform proposals, HIPCs would have been administrators of subsidies for low-income families and the enforcers of cost-containment regulations.[22]

HIPCs are not risk bearing entities, and they would not contribute to covering individuals' cost of insurance coverage, though a government subsidy program could be administered through them. In the same way, employers do not, in essence, contribute to the costs of their workers' coverage; workers pay for their employers' "share" of health benefits through lower wages. Within a HIPC-based system, employers would make contributions on behalf of their covered workers directly to the HIPC. The workers would then have the choice of enrolling in any of the plans under contract with the HIPC. Individual purchasers would have the same plans available to them and would also make their premium payments directly to the HIPC.

HIPCs created under state law would be bound by the rules imposed by the law. Typically, for the class of eligibles defined by state law, HIPCs would not be allowed to exclude employer groups or individuals based upon health status or risk. In their role as monitors of the quality of care delivered in its plans, they may be allowed to exclude plans that deliver poor quality but otherwise comply with the rules set out in state law. If HIPCs are granted this authority, then their potential price and quality bargaining power would be significantly enhanced.[23]

While the 1993/94 debate and recent state actions indicate the HIPC concept has bipartisan support, there is some dispute over whether these entities should be mandatory or voluntary for eligible firms and individuals choosing insurance coverage. The issues relevant to this debate are largely related to adverse selection, and are explored in a later section of this paper.

Repeal of Anti-Managed Care Laws

The term "anti-managed care laws" refers to a set of laws and regulations that share one common goal: the protection of traditional fee-for-service medicine and indemnity insurance. The most important of them today are "any willing provider" (AWP) and "freedom of choice" (FOC) laws. AWP refers to statutes which require that any managed care organization accept any provider into its "network" of preferred providers who is willing to comply with the plan's publicly explicit criteria. This necessarily limits a health plan's ability to select cost-effective providers for its network, and retards managed care plans' ability to offer the low premiums they otherwise might.

FOC laws require a plan to permit enrollees to choose to use any provider of their choosing and to get substantial (sometimes equal to the maximum) reimbursement from the health plan. This is like requiring a point-of-service option on every plan, and could, depending on the legislated restrictions, retard the ability of the managed care plan to offer lower premiums based on expected health expenditures. Both AWP and FOC laws typically list the specific provider types and health plan types for which they apply.[24]

Stop-Loss Policies for the Self-Insured and the Definition of Health Insurance

The exemptions from state laws governing the business of insurance that are afforded employers under ERISA have proven to be very attractive to many firms—63 percent of insuring firms with more than 500 employees and 17 percent of insuring firms smaller than 500 self-fund their health insurance.[25] Thus, approximately 45 percent of workers and their dependents are in self-insured plans.[26] Some employers do indeed bear all the risk of health insurance themselves, but a large majority purchase a stop-loss policy that covers medical expenses above certain threshold levels both for individuals and for the insured group in the aggregate.[27] For example, stop-loss coverage may be triggered if an individual's covered expenses exceed $20,000, or if the group's covered expenses exceed 120 percent of expected aggregate costs.[28] The lower are the individual and aggregate thresholds, the more risk is actually borne by the stop-loss insurer rather than by the self-insured employer or group. If most risk is actually borne by the stop-loss insurer, then the employer is actually buying health insurance and not merely protection from financial catastrophe. In this case, the employer is not really "self-insured" and the ERISA exemption from state laws may be inappropriate. A state may choose to define health insurance to include stop-loss policies with individual or aggregate thresholds below certain amounts, and thereby apply the same solvency and issue laws and regulations to these policies as are applied to other health insurance policies.

Selection Bias Effects in Reformed Voluntary Markets

There is an unfortunate tradeoff—some might say paradox—in the current health insurance market. The more choice individuals and groups have over whether or not to purchase insurance coverage, the types of insurance policies, and the locus for purchasing those policies, the greater the risk of systematic problems related to selection bias. In some types of mandatory coverage environments selection problems arise as well; however, by and large, selection issues are considerably easier to handle in mandatory environments than in voluntary settings.

Adverse selection occurs when individual insurers (or insurers as a group) enroll a disproportionately high-risk clientele. This can occur because healthier individuals, in general, are not risk averse enough to prompt them to purchase insurance when they do not need medical care, or because certain insurance plans are relatively more attractive to higher cost groups. As the average enrollee cost in a particular plan or plans in general rises, the premium price per enrollee rises as well. Each time the premium increases, enrollees must weigh the costs and benefits to them of retaining that insurance. As premiums increase, enrollees at the lower end of the risk spectrum tend to opt out of their plan, forcing the average premium up even further. Such a dynamic, in theory, could result in such a substantial upward spiral of premiums and loss of coverage that the private health insurance system might no longer be viable over time.[29]

Does this mean that voluntary systems are unworkable? Not entirely. Our current system is a voluntary one, and although it has many flaws, it has shown itself to be viable in the long run for certain groups. As we strive to reform the system in a voluntary environment, however, we must be cognizant that many of the serious problems in today's structure are the results of mechanisms that were put in place by insurers in order to protect themselves from adverse selection in a voluntary purchasing world.

The Rules of Issue and Adverse Selection

The general absence of uniform rules of issue, discussed previously, enables insurers to enroll and maintain a lower than average cost population. In most states today, insurers can refuse to issue coverage to groups or individuals based upon their specific past health care use or general indicators of their health status (such as industry or demographic profile). Most states also allow insurers to refuse to renew policies based upon previous years' claims experiences, and to deny coverage for pre-existing illnesses or conditions.[30] In addition, insurers issuing new policies or renewing old policies for high-cost groups/individuals can, in general, charge premiums reflective of their high expectations of future costs.

Simply disallowing these practices in a voluntary environment, however, would generate tremendous adverse selection problems for insurers, and would in all probability, eventually lead to the eradication of private insurance. The reason is that there would be no incentives for individuals to insure themselves until they became seriously ill or expected high levels of future medical care consumption. Given that, the average cost of insured persons would be very high, pricing most people out completely. Those that remained would not be pooling risk of uncertain future medical needs—the medical needs would be virtually certain in this context.

Changing the rules of issue without destroying the market is possible. Limiting guaranteed issue to a pre-determined open enrollment period is one example of approaches for restricting the extent of adverse selection. Setting maximum allowable time spans (e.g., 12 months or 6 months) for pre-existing condition exclusions, without completely eliminating them, is another. Guaranteed issue and renewability will be most effective at reducing selection if they apply to all products sold by insurers in a given market.

Mandated Benefits and Adverse Selection

Standard benefit packages (SBPs) could reduce selection, for competition is more likely to be over price than the package itself. When packages can vary infinitely, insurers can use particular benefits to attract good risks (e.g., by offering well-baby care with zero cost-sharing) and to discourage bad risks (e.g., by not offering mental health coverage). However, the net effect of standard benefit packages on adverse selection will depend upon the package of reforms (if any) that is simultaneously implemented. Without community rating and effective risk adjustment, standard benefit packages are likely to have a negligible impact on the degree of risk segmentation in the insurance market.

Mandated benefits provisions, by forcing coverage of particular services, have the potential for reducing selection in a manner similar to SBPs. However, in the absence of community rating and risk adjustment, mandated benefits laws may end up significantly increasing the incentives for insurers to select risks (e.g., if generous mental health benefits are mandated), and thereby worsening the actual degree of risk segmentation, for seeking out the least utilization prone among those with mental health coverage could be highly profitable. There is some evidence that mandated benefit laws, per se, increase premiums.[31] Generous mandated benefits and standard benefit packages thus may encourage firms to self-insure, thereby exempting themselves from regulation. By removing themselves from the general private insurance market, risk is segmented further.

Medical savings accounts and the catastrophic health insurance policies that go with them would increase risk segmentation, at the least, and probably adverse selection as well. Simulations by the American Academy of Actuaries[32] confirm economic theory's prediction that the MSA/catastrophic combination will be most attractive to relatively young and healthy workers. Firms that maintained traditional comprehensive options would find them selected against and would be required to increase their premiums—classic adverse selection.[33] Firms that abandoned traditional options and just offered the MSA/catastrophic plan would engender significantly higher expected total payments (premium plus out-of-pocket payments minus net tax savings) from older and sicker workers or workers with sicker family members, thus increasing risk segmentation within the firm. This may not be a big problem among firms with relatively young workforces which also have relatively high labor turnover, but for many firms increasing risk segmentation and the attendant intrafirm tensions will be the likely result.

Community Rating and Adverse Selection

Pure community rating in a voluntary insurance world may not be advisable either. Without requiring coverage, lower risk individuals, such as the young and healthy, may opt out of the insurance market completely rather than subsidize the insurance of higher cost populations through higher premiums. The implications of low-risk "drop-out" are:

1. Increases in the community rated premium for comprehensive coverage: this occurs due to an increase in the average risk of those maintaining insurance and due to an influx of higher cost individuals who had previously been priced out of the insurance market by experience rating. In the extreme case, the dynamic of insurance dropping by those at the low range of risk could lead to price increases beyond some high-risk individuals' ability to pay.

2. A potential increase in the number of uninsured: this occurs if the number of lower risk individuals leaving coverage is greater than the number of higher risk individuals newly taking up coverage.

One option for moderating the potential negative results of community rating is to use age-rating bands. Modified community rating might allow those in different broad age bands to pay different amounts for premiums. For example, those aged 50-65 might pay at most three times the premium payment for an 18- to 25-year-old. In this way, younger and healthier individuals would not face as large a premium increase as under pure community rating, while the premium charged the older individuals is bounded at a level below what they are likely to face in the current market.

Other options besides age for rating bands include industry categories (individuals/employers in industries such as mining, for example, would face higher premiums than those in industries that were perceived as lower risk), and work status. All the options for developing rating bands would result in more redistribution of premium prices than is seen today but would result in less redistribution, and consequently lower the risk of dropping, relative to a pure form of community rating. Modifications such as industry rating could mean, however, that some high-risk groups would still be priced out of the market.

While age rating bands attempt to lessen the probability of individuals and firms dropping insurance completely, another related issue is opting out of the community-rated pool but not out of insurance all together. To the extent that firms are allowed to purchase insurance either in the community-rated (or modified community-rated) pool or through other insurers or through self-insurance, further opportunities for selection bias occur.

For example, say a community-rated pool was established for firms of 100 workers or fewer. Along the spectrum of voluntary system options, the most stringent option would be to require all employers of that size choosing to insure their workers to purchase community-rated insurance. The least stringent option would be to allow all employers of that size to purchase community-rated insurance or to purchase experience-rated insurance through another insurer or to self-insure. Under the least stringent option, those firms with lower-than-average cost employee groups would likely opt not to enroll in community-rated plans. This would leave the higher cost groups to the community-rated plans, but the community-rated premiums would be high given that there would not be any lower cost groups with which to share the risks. As under the discussion of pure community rating, this situation would be likely to lead to inaccessibility of insurance for the higher cost groups.

The handling of Multiple Employer Welfare Arrangement (MEWA) plans is one example of exemptions that could lead to dramatic selection problems. MEWAs are arrangements whereby, in the current market, small employers voluntarily band together for` purposes of purchasing insurance. During the 1993/94 health care reform debate, many of the later bills allowed community-rating pool exemptions to firms insuring their workers through MEWA plans. Some proposals (S. 2374, the "Dole Bill" of the 103rd Congress, is one example), allowed exemptions for not only MEWAs providing health insurance prior to enactment, but for those choosing to provide such coverage in the future as well. The practical effect would have been to allow virtually any employer with a lower-than-average cost group to band together with other firms similarly situated, and to purchase experience-rated insurance or to self-insure. The result, again, would have been high and potentially unsustainable community-rated premiums.

In order to minimize selection problems resulting from insuring groups opting out of the community-rated pool, more stringent restrictions are necessary. The firms and individuals eligible for the community-rated pool and choosing to insure should be restricted to purchasing their insurance through that pool to the extent possible. If some type of MEWA exemption was necessary for political reasons, it would probably not threaten the integrity of the community rate to allow the MEWAs currently providing health insurance coverage through self-insurance to continue to do so; however, expansion of the size of those plans should be severely restricted, perhaps by the rate of growth of employment in the applicable industry.[34] Another option might be to develop a risk-adjustment mechanism that would redistribute some costs from the community-rated pool to those opting out of that pool.[35]

In addition, it is advisable to keep the number of firms eligible for participation in the community rate as large as possible given a rule that, with few exceptions, requires that those eligible can only purchase insurance through the community-rated pool. This is particularly true in situations where individual purchasers are in the same community-rated pool with the small firms. Although the health care costs of workers and their dependents do not appear to differ appreciably along the continuum of firm sizes, the health care costs associated with non-working households tend to be higher than those of working households.[36]. Consequently, the larger the community rating pool, the greater the number of relatively less costly households over which to spread the higher costs of non-working households.

Risk Adjustment and Adverse Selection

A risk-adjustment mechanism is designed to reduce the incentives for risk selection. A purely prospective mechanism, if its forecasting ability is not much improved over currently available candidates,[37] might not reduce the incentives for risk selection, for insurers would still expect to be undercompensated for bad risks and overcompensated for good ones. A purely retrospective mechanism, like a reinsurance pool, would reduce the incentives to select by ameliorating the consequences of having bad risks. However, there is a tradeoff between complete recompense for bad risks and maintaining the incentives to manage the care of the very sick in a cost-effective manner. Thus, a fairly high threshold and/or some sharing of the costs above this threshold may be required of the "unlucky" insurer with a retrospective risk-adjustment mechanism. This liability attenuates the protection from bad risks and thereby reinstates the incentive to select in the first place. A blended combination of prospective and retrospective mechanisms is probably the most practical solution at the present time.[38]

A blended mechanism would work like this. The prospective factors might be age and sex, and the retrospective threshold might be $20,000 in claims. After the enrollment period, each health plan's age/sex profile would be calculated, converted to an index, and compared to the market average age/sex profile. The indexes would reflect expected age/sex differences in utilization and expenditures, on average. Plans with lower-than-average risk profiles would be assessed for a fraction of their premium revenue and plans with higher-than-average risk profiles would receive these assessments. These prospective payments and receipts would be proportional to each plan's variance from the market average risk profile.

All plans would contribute either some fixed amount per enrollee or some fraction of risk adjusted premium revenues into a retrospective claims fund. Plans with individual patients whose utilization, evaluated at pre-established fair market prices, exceeded $20,000, would submit those claims and draw some fraction of the excess claims from the fund. As experience accumulates, the proportional adjustment factor for prospective payment and receipts, the high-cost case threshold, the size and type of assessment for the retrospective fund, and the fraction of excess claims that can be drawn from the fund would all be adjusted to balance the incentives against both risk selection and managing high-cost cases efficiently.

Marketing Practices and Adverse Selection

This is fairly obvious: insurers and their agents have strong incentives to use the marketing techniques mentioned above (targeting neighborhoods, withholding information about alternatives, etc.) to select favorable risks, and some states have tried to restrict this behavior. Effective enforcement is significantly aided by standard benefit packages and guaranteed issue requirements. The absence of guaranteed issue means that permitting medical underwriting is state policy. Without guaranteed issue, market conduct examiners (most states require an examination every 3-5 years) are limited to trying to prevent false and misleading statements, a considerably lower standard than inhibiting selection strategies.

Health Insurance Purchasing Cooperatives and Adverse Selection

The establishment of purchasing cooperatives raise more adverse selection issues related to community rating. As mentioned in a previous section, a reform system which includes HIPCs can structure them in two basic ways. Under a system of voluntary HIPCs, if an individual or group chooses to insure, they can insure through a HIPC and under its community rate or they can insure outside of a HIPC through self-insurance or experience rating. Under a system of mandatory HIPCs, any individual or group choosing to insure must insure through a HIPC.

The voluntary HIPC structure invites the same types of selection issues as mentioned above, when firms or individuals are allowed to insure themselves outside of the community rate. Absent further regulation[39], those insuring through the HIPC would tend to be those having difficulty attaining coverage and those with relatively higher costs. Some potential offsetting differences exist, however.

First, the administrative savings for small firms that could result from larger group purchasing might be sufficient to attract a broader group of employers. The administrative costs involved in insuring an individual or an individual small employer can be prohibitively high.[40] Second, the wider array of insurers that would likely be available to workers of enrolling firms could be another lure for a range of employers. In the current market, small employers are significantly less likely than large employers to provide their workers with a choice among insurance plans.[41] And third, if the HIPCs are permitted to be tough negotiators, the greater purchasing power of a HIPC could be a pull for employers. If HIPCs are not required to flatly accept any insurer bid that complies with community rating standards, but instead are given powers to exclude insurers and/or to negotiate premiums around an expenditure target, small employers might find HIPCs to have a premium price advantage over their individual capabilities.

Even so, Enthoven, among others, suggests the implementation of compelling incentives or legal requirements for all small employers to participate through the HIPC to avoid a "spiral" of adverse selection.[42] One compelling incentive is contained in a Jackson Hole proposal which would make HIPC participation by small employers a condition for exclusion of employer contributions from employees' taxable income.

The second "layer" of the HIPC structure, after the voluntary versus mandatory decision is made, is the choice of exclusive versus competing HIPCs. Will an individual or firm located in a specific area have a single HIPC through which they may enroll, or will there be a number of different HIPCs in the area from which they may choose? The exclusive HIPC option minimizes adverse selection problems relative to the competing HIPCs. Allowing competing HIPCs opens up the possibilities of "cream-skimming" of good risks among HIPCs of the same sort that is seen today among insurers. HIPCs could potentially compete for the lowest cost clientele in order to keep premiums lower than average. The result would again be a segregated market, with little risk sharing across individuals of different levels of utilization, and potentially prohibitively high premiums among HIPCs with a disproportionate share of high-cost individuals.

In order to contain the level of selective enrollment across geographically overlapping HIPCs, a fairly elaborate system of regulation would be necessary. In addition to requiring all HIPCs to offer guaranteed issue, renewability, portability, and limits on pre-existing condition restrictions (which would also be necessary in a mandatory HIPC world), HIPC marketing practices would have to be carefully monitored. In addition, a risk-adjustment mechanism that would work across HIPCs (as opposed to across plans within a single HIPC) would likely be necessary. Such measures might provide a network of mechanisms for reducing HIPC incentives to cream skim enrollees.

Anti-Managed Care Laws and Adverse Selection

In principle, both any willing provider and freedom of choice laws increase the range of providers that managed care enrollees have access to. A major reason for remaining in indemnity plans is to have access to specific providers to whom patients have developed attachments but who have not joined managed care plans. Since patients with strong provider attachments are likely to be sicker, on average, indemnity plans competing with managed care plans suffer adverse selection. So, increasing the range of providers should moderate this adverse selection against indemnity plans. Given the fairly limited nature of anti-managed care laws in most states at present, however, and some providers' continued reluctance to abide by the conditions of managed care plans, the practical importance of this effect has not been conclusively demonstrated. Furthermore, the tradeoff here is that stringent anti-managed care laws could stifle efficient reorganizations in the health delivery system. This risk is arguably greater than the improvement in selection effects they may entail.

Stop-Loss Policies, the Definition of Health Insurance, and Adverse Selection

The larger the number of employers who self-insure, the smaller the potential pool for commercial insurers who might be inclined to at least partially pool risks within a given class of enrollees (firm size, industry, etc.). The availability of very low threshold stop-loss policies, i.e., de facto health insurance, makes it easier to "self-insure," and thus increases the degree of risk segmentation in the smaller group market.

Goals of Reform and the Tools Available for Achieving Them

Four goals seem to dominate the intent of insurance market reform:

• to extend economies of size to small groups and individuals;
• to increase security of insurance coverage;
• to promote competition in the private market; and
• to expand insurance coverage.

For each of these major goals we will delineate the policy tools available to address them, and the contingent policies necessary to ensure that these strategies are effective and do not worsen the risk segmentation in the current marketplace.

Economies of Size. As has been discussed, small groups and individuals are disadvantaged relative to large groups in two ways: the administrative costs associated with insuring them are substantially higher, and they have very limited or no opportunities for spreading their health care risks with other individuals or groups. The administrative cost handicap leads to higher premium levels for virtually all small firms and individuals. The risk spreading issue threatens access to insurance for high cost individuals and some small groups while endangering long-run insurability for all individuals and small groups.

Community rating pools are the predominant tools for increasing the spreading of insurance risk. We have already mentioned that community rating in voluntary markets should be modified using limited age bands; decreasing or eliminating the age bands slowly over time may be feasible. Community rating must be done in association with other policies as well. At a minimum, community rating in voluntary markets must include guaranteed issue (during specified open enrollment periods), renewal, portability, and limits on pre-existing condition exclusions (6 to 12 months). Without these reforms of the rules of issue, high-cost individuals and groups would continue to be excluded from risk sharing. And without the conditions (limited enrollment periods, etc.) placed on these rules of issue, low-cost individuals might opt out of the insurance system completely until the time at which they became ill. In addition, there must be limits on the exemptions that are allowed from the community rating pool; for example, only MEWAs and association plans that were in existence prior to reform could opt out of the community rate. With easy exit available, the community rating pool would become the insurer of last resort for the high cost, and would not be sustainable over time.

For the longer run effectiveness of community rating it may be necessary to introduce standard benefit packages and risk adjustment across plans. If benefit packages are allowed to vary considerably, insurers may continue to use the design of covered services to select the less costly individuals and groups in the community rating pool. And if keeping the number of plans available in the pool high in the long run is valued, some risk adjustment mechanism may be required. Without one, plans which tend to attract higher cost individuals may see their prices escalate over time, making their plans unaffordable due to that adverse selection—not due to differences in quality of care or efficiency of service delivery.

Administrative and risk pooling economies of scale can be addressed effectively through HIPCs. A pared down HIPC could be designed to address administrative costs without fully confronting issues of adequately spreading risk. For example, a state could develop a HIPC for all firms of a particular size (say, 2 to 100 employees). All qualifying firms could participate, taking advantage of the larger group purchasing and consequent lower administrative costs. Such a pared down HIPC would not, however, include standard benefit packages or risk adjustment. Plans could vary considerably in their premiums and the benefits offered, although insurers could not price-discriminate by health status across the employers joining the HIPC. The HIPC might list a limited number of factors (such as age) that insurers could use to develop rate bands, but the state would not place boundaries on the factors (for example, a state would not legislate that the oldest age group not pay more than four times the youngest age group). This type of HIPC might not make any significant strides toward broader spreading of risk across groups, but it should not increase segmentation of the market either. Individuals might also be permitted to purchase insurance inside the HIPC, but at an individual rate which would be allowed to vary from the employee rate. The primary interest of this more modest HIPC, therefore, is not to spread risk but to lower administrative costs. A more traditional HIPC, as described in earlier sections, would address both the administrative and risk sharing aspects of economies of size.

HIPCs seeking to address both administrative costs and risk spreading would require community rating and all of the policies necessarily associated with community rating. In addition, it is critical that the insurance rules inside the HIPC are in congruence with those rules outside the HIPC. For example, if the rules of issue were reformed for the market inside the HIPC but not for the outside market, the HIPC market would attract the higher cost groups, with the effect of segmenting the insurance market as a whole. Policy must be consistent inside and outside of the HIPC otherwise the voluntary nature of these reforms will provide leeway for self- selection by individuals and groups.

Security of Coverage. A number of policy tools are available for increasing the security of insurance coverage. These include: the rules of issue, community rating, proscribed limits on annual premium increases, and high-risk pools for medically uninsurables. The rules of issue prevent insurers from excluding those with high-cost experiences from coverage. Community rating insures that coverage at average rates will be available regardless of past or future medical needs. Annual limits on premium increases help to limit insurers' ability to effectively price groups out of the market because of a high-cost year. And high-risk pools can serve to provide coverage of last resort to individuals who have been denied coverage (or affordable coverage) due to pre-existing conditions.

All of the rules of issue do not necessarily have to be implemented together; however, doing so increases the security value of the reforms. Portability, being a type of pre-existing condition limitation, must be implemented with general pre-existing condition limitations in order to be meaningful. If guaranteed issue is not included in reforms, guaranteed renewability would not be effective in increasing coverage security unless fairly tight limits on annual premium increases were also included. Without such limits, the price at which renewal comes could be set sufficiently high enough that it is effectively a termination of coverage. If guaranteed issue were to be included in the package, annual premium limits might not need to be as restrictive, because groups and individuals would have the ability to purchase coverage from other firms. Premiums at first issue might, of course, be set sufficiently high that access would still be severely limited for some groups. As a consequence, long-run effective coverage security for all groups would require community rating and its associated policies: reformed rules of issue, standard benefit packages, and risk adjustment. Guaranteed issue will, however, increase the average premium level somewhat. In order to keep such increases as moderate as possible, guaranteed issue requires broad risk pooling—the more average and below-average risks included in the pool, the less of an effect the new entrants will have on the average premium level.

High-risk pools can theoretically be effective in removing very high-cost individuals from the private insurance market, reducing the incentive for insurers to select lower risk groups and individuals. In practice, however, these pools often are not well funded.[43] As a result, only those able to afford the high cost of the premium (usually set at 150 percent of the premium for a comparable benefit package) are able to participate, and total enrollment may be limited. In order to have a significant impact in lowering the overall risk of the privately insured, high-risk pools must be associated with sources of funding sufficient to provide a broad group of high-cost individuals with insurance on an ability to pay basis. The pools must also be available to individuals whose employers provide insurance coverage but where the employer group is sufficiently small that the group's access to coverage might be hampered if the high-cost individual were included in the insurance group.

Promote Competition. Some analysts believe that a more competitive insurance market will lead to more efficient delivery of health services, higher service quality, and decreases in average premium levels. The tools available for moving the insurance market in this direction include data collection and dissemination, standard benefit packages, regulation of marketing practices, and repeal of anti-managed care laws. Collection and dissemination of data on diagnoses, treatments, outcomes, and costs can allow individuals and insurers to make more informed choices about the providers that they choose and with which they contract. The more informed are the purchasers of services, the more providers of services will compete on quality and cost. Uniform benefit packages allow individuals to easily compare the prices of each plan available to them, while repeal of anti-managed care laws remove state imposed impediments to the ability of network plans to price themselves competitively. To the extent that insurers' ability to selectively market themselves to lower cost groups could be limited through regulation, insurers would be forced to compete on service quality and efficiency as opposed to their ability to enroll low-risk individuals.

While anti-managed care laws can be repealed in the absence of other reforms, both data-based approaches and marketing regulations require a central organizing body for information flows. HIPCs are one alternative, but other structures could conceivably be designed to serve the same purposes. Such an organizational structure is necessary in order to assure uniformity in data collection and to closely oversee compliance with marketing guidelines. If standard benefit packages are implemented, risk adjustment may also be necessary in order to maintain long-run diversity in available insurance plans.

Expand Coverage. The insurance reform tools available for expanding the number of covered individuals and groups are chiefly those that make insurance policies more affordable. Given the low estimated price elasticities of demand for health insurance found in the literature,[44] such increases in coverage, in general, should be expected to be fairly modest.[45] The competition enhancing strategies are approaches that should be considered in this context. While community rating and reform of the rules of issue would lower premiums for higher than average cost individuals and groups, such reforms would also increase premiums for others. The net coverage effect will be a function of the success with which insurance dropping by the currently insured is contained by strategies such as age rating.

Four Insurance Reform Packages

Depending upon the political climate in the state considering insurance reform options, some legislative approaches may not be feasible. Regardless of the level at which reforms are pursued, the utmost in caution should be used to ensure that reforms do not lead to further segmentation of risk in the private insurance market. We present four alternative insurance reform packages, representing increasing levels of aggressiveness. We believe that these packages adhere to the basic tenant of "do no harm"; while they may help only limited groups or individuals, they should not lead to further deterioration in the overall market. We assume that the cautionary steps advised in previous sections are taken in conjunction with each specific package component. For example, guaranteed issue assumes that the policy applies only during a pre-specified limited time period each year.

In each package below, we recommend that the state adopt the NAIC's Model Stop Loss Insurance Act, to standardize the definitions of health insurance and self-insurance in reasonable ways. Such rules can significantly limit current abuses, and can be implemented without concern about potential interactive effects with other insurance reforms.

Package 1—Minimalist Reform: regulation of definition of stop loss, guaranteed renewal, portability, 6- to 12-month limits on pre-existing condition exclusions, limits on annual premium increases. This minimalist package would provide some increased security to the currently insured. Such a package should not have a substantial effect on average overall premiums. While the potential negative effects of Package 1 are minimized, so too are the potential positive effects. Those groups and individuals that are currently uninsured would not have their access to coverage enhanced, and there is some chance that insurers would become more selective of new insureds since renewal is guaranteed. In addition, even with what might be considered fairly "tight" limits on annual premium increases, say 50 percent, premiums could be made unaffordable for some within a fairly short time horizon—one to three years. The Package 1 approach should not, however, increase the segmentation of risk seen in the current market; it is also unlikely to lead to any significant redistribution.

Package 2—Modest Reforms: regulation of definition of stop loss, guaranteed renewal, portability, 6- to 12-month limits on pre-existing condition exclusions, limits on annual premium increases, well-funded high risk pool, HIPC focused on reducing administrative costs. By adding a well-funded high risk pool to Package 1, we should be able to pull a significant number of very high-cost individuals out of the private insurance market and out of the ranks of the uninsured. This could help to counterbalance the potential effect of insurers being more aggressive in screening new applicants because of the guaranteed renewal rule. Such a pool also would provide additional coverage for the most vulnerable state residents. The pared down HIPC, reminiscent of the federal government employees health benefits plan (FEHBP), should provide some savings to the groups (and individuals, if included) hardest hit by the current insurance market's administrative costs structure. This package would also provide access to HIPC eligible employer groups and individuals who have been unable to attain coverage at any price, as guaranteed issue would be a necessary component of the HIPC.

It is not feasible to have guaranteed issue inside the HIPC and not have guaranteed issue outside the HIPC if the HIPC is voluntary. The options are: (1) make the HIPC mandatory for individuals and groups below the cutoff size; or (2) apply the same insurance rules, including guaranteed issue, outside the voluntary HIPC as well as inside the HIPC.

This package does not sufficiently address the issue of risk spreading, leaving ample opportunities for insurers inside the HIPC to continue to select good risks through a variety of mechanisms. The HIPCs could choose to offer policies to individual purchasers at a different price than employment-related enrollees, and the absence of standardized benefit packages would allow insurers to continue to attract lower risk enrollees through benefit design. The lack of risk adjusting across plans also keeps the insurers' incentives to select high.

Under this package, modest redistribution would result. Depending upon the size of the high-risk pool, groups with high-cost individuals should see their premiums fall as those individuals are removed from the employer risk pool. There would also be some redistribution of income from state residents to low-income individuals enrolling in the high-risk pool. The character of this redistribution would be a function of the mechanism used to raise revenue to fund the pool. The administrative cost savings are also likely to result in a redistribution of dollars away from insurance companies back to the purchasers of insurance.

Package 3—Moderate Reforms: regulation of definition of stop loss, repeal of anti-managed care laws, full reform of rules of issue (including guaranteed issue), HIPC for administrative efficiencies and risk pooling, regulation of marketing practices, modified community rating, and retrospective risk adjustment. This package would increase security for all those individuals and groups that are included in the community rating pool and that are subject to the reformed rules of issue. Administrative costs should be reduced for the large majority of those enrolling in the HIPC. In addition, those previously unable to obtain insurance coverage due to high costs should have significantly increased access to coverage. In general, the insurance market environment should be more competitive. In such an environment, managed care would be allowed to operate on truly equal footing with more traditional types of insurance plans. Retrospective risk adjustment should help mollify insurers' fears of potential financial disasters resulting from adverse selection.

Retrospective risk adjustment is unlikely to be as successful as a blended approach in striking a balance between incentives to manage care and financial protection for insurers with bad risks. With broad age bands, the community rating may still make insurance unaffordable for some older low-income individuals and for small groups with low wages and high proportions of older workers. And without centralized collection and dissemination of data, purchasers will not be as informed as they could be, perhaps limiting to some extent their tendency to choose plans within the HIPC efficiently. Loosely defined or numerous standard benefit packages may leave significant room for selection, potentially threatening the long-run viability of the most comprehensive plans.

Depending upon the breadth of the age bands, significant redistribution could occur, with younger individuals paying more than they would under experience rating and older people paying less than they would have otherwise. The actual magnitude of these changes will be a function of the number of previously uninsured higher cost persons that enter the market and the number of lower cost individuals that leave the market. At least some of the increased costs to younger individuals will be counterbalanced by the administrative savings resulting from larger group purchasing.

Package 4—Comprehensive Reforms: regulation of definition of stop loss, modified community rating phasing down to tighter age bands (2:1), narrowly defined and fewer standard benefit packages, blended approach to risk adjustment, full service HIPC including data collection and dissemination. This package provides groups and individuals with the most security of any of the suggested packages, with the most stable long-run premiums. By constraining the standard benefit packages and using a blended approach to risk adjustment, the incentive and capacity of insurers to selectively enroll low-cost persons is minimized. By including a data strategy, consumers can make better informed and efficient plan choices.

This package also implies the greatest degree of redistribution of any of the four options presented. Younger individuals and healthier groups would likely see their premiums increase the most relative to those in the current system. The costs associated with the seriously ill would be redistributed to healthier individuals to a greater extent than under any of the alternative packages. In return, the young would get a promise of an available community rate in the event they get older or sicker or both. Insurance companies which have selected good risks most aggressively would find this strategy substantially less rewarding under this reform package, for insurance market profits would be distributed to those more efficient at managing care than selecting risks.

Conclusions

As we have seen, insurance reforms are not simply structured policies that can be implemented without care or risk. Voluntary markets make these types of reforms difficult to achieve without increasing risk segmentation beyond what is seen in the current market. For this reason, packages of insurance reforms must be carefully designed. Additionally, the best insurance reform tools available for addressing the major goals of reform can also be quite politically sensitive—using community rating to pursue the goal of long-term security in access to insurance coverage is one example.

Whether or not proposed packages of insurance reform policies increase or decrease risk segmentation in the market is an appropriate way to assess such policies. Some economists argue that risk segmentation makes insurance markets more efficient in the neoclassical sense, for it brings expected health costs and premiums into closer alignment for smaller and smaller sub-groups of the population. There are two major arguments against this position. First, this is a narrow definition of efficiency. An unfettered insurance market fails to deliver competitive outcomes, because many systematically pay premiums higher than their expected costs and some cannot find vendors willing to sell to them at any price.

In addition, the neoclassical or libertarian construct of efficiency is fixed at a point in time. Today's healthy person or group could become tomorrow's uninsurable. The existence of a functional community-rated market, then, is like an insurance policy against the risk of becoming less healthy. A purely segmented market wherein all pay their expected costs, even if it did exist, would underinvest in this security value of a community rate. Finally, a highly segmented health insurance market raises significant and inequitable barriers to the sick who are not poor enough to qualify for Medicaid.

While many states have made progress in implementing some insurance reforms for small groups, all but two of those states have passed reforms that apply only to the employer-sponsored insurance market of 50 or fewer workers. A significant number of those states limit their legislation to firms of 25 or fewer workers. Most exclude the individual market and leave other small groups unprotected—firms of size 26 or 51 are clearly not actuarially sound for purposes of spreading risk.

Although some improvements in access and security of coverage can be made for some people through carefully planned insurance market reforms, substantial gains in either insurance coverage or cost containment should not be expected. While making the market more competitive may have some modest effects on the level of premiums (less likely is that the rate of increase in premiums would be significantly affected), such price changes are not likely to prompt substantial changes in the number of individuals purchasing insurance or in the number of firms sponsoring insurance for their workers. Low-income individuals, particularly the working poor without employer-sponsored insurance and those not Medicaid- eligible, are unlikely to be significantly helped by reforms that do not include sizable premium subsidies.

About the Authors

LINDA J. BLUMBERG is a research associate at the Urban Institute. Dr. Blumberg has focused her career and research interests on issues of health care policy and economics. From August 1993 through October 1994 she served as health policy advisor to the Clinton Administration during its health care reform effort. While at the Urban Institute, she has done research on the distributive impacts of national health reform proposals and on potential improvements in the inter-state equity of federal Medicaid funding. She is currently evaluating components of state level health care reforms as well as analyzing the effect of Medicaid eligibility on private insurance coverage.

LEN M. NICHOLS joined the Urban Institute's Health Policy Center in November of 1994. He was the Senior Advisor for Health Policy at the Office of Management and Budget (OMB) from 1993–1994, where he managed and coordinated cost and revenue estimation for the President's Health Security Act (HSA) and its Congressional successors. He was the lead administration analyst for coordinating estimation methodology with the Congressional Budget Office. Since coming to the Urban Institute, Dr. Nichols has worked on insurance market reform, medical savings accounts, evolving integrated health service networks, Medicare reform, and health workforce issues.

---

Notes

---

---

---

Usage and reprints: Most publications may be downloaded free of charge from the web site and may be used and copies made for research, academic, policy or other non-commercial purposes. Proper attribution is required. Posting UI research papers on other websites is permitted subject to prior approval from the Urban Institute—contact [email protected].

If you are unable to access or print the PDF document please contact us or call the Publications Office at (202) 261-5687.

Disclaimer: The nonpartisan Urban Institute publishes studies, reports, and books on timely topics worthy of public consideration. The views expressed are those of the authors and should not be attributed to the Urban Institute, its trustees, or its funders. Copyright of the written materials contained within the Urban Institute website is owned or controlled by the Urban Institute.