State officials who are serious about reforming their health insurance markets will be confronted sooner or later by the complex issue of risk adjust­ment. Today, most Americans have limited or no choice of health insurance and are covered either by a plan selected by their employer or by a government-designed and government-managed program.

In order to move to a consumer-centered system in which individuals can choose and own their coverage in a competitive market, state laws governing health insurance need to be modified in a number of areas.[1] One technical issue associated with providing more individual choice of coverage in health insurance markets is how best to "risk-adjust" payments to insurance plans.

The risks and costs of providing medical care can vary significantly from individual to individual, depending on the individual's health status. There will always be individuals whose risks and costs are lower or higher than the average for the general pop­ulation. Indeed, sometimes the divergence from the mean is substantial. At one end of the spectrum are those who are likely to incur little if any medical expense. At the other end are a relatively few individ­uals with significant medical costs.

A Functioning Market

If a health system is structured to pay insurers or providers an average amount for each subscriber or patient, insurers or providers will have incentives to seek those whose costs are likely to be below average and to avoid those whose costs are likely to be above average. The way to limit this problem is to adjust payments in some fashion so that insuring or treat­ing sicker individuals is also profitable, or at least not a significantly money-losing proposition.

For hospitals, doctors, and other medical profes­sionals, this can be accomplished by paying them either on a fee-for-service basis, so that they are paid more for patients needing more services, or on a per-capita basis, but with the payment amounts var­ied according to each patient's "severity of illness." Under both methods, providers receive more money when treating sicker patients.

In the same fashion, the solution for paying insurers is to risk-adjust the payments to align them better with the divergent medical costs incurred by healthier and sicker subscribers. Some form of risk adjustment is particularly important in a consumer-choice system in which each individual can choose his or her preferred health insurance plan from a menu of different plan types and coverage designs.

The Better Path

Risk adjustment for health insurance can be pro­spective or retrospective. However, for a variety of practical reasons, a retrospective risk-adjustment system -- through the creation of a statewide risk-transfer pool -- is likely to be the more promising solution for state policymakers.

A risk-transfer pool is a mechanism to spread the costs of the relatively few individuals with very expensive medical conditions evenly among all insurers and policyholders in a given market. Therefore, no single insurer is disadvantaged if it happens to attract more high-cost subscribers than its competitors. Because the funding adjustments in a risk-transfer pool system are based on actual claims as opposed to projections of future costs, they are less complicated to design and administer than systems for prospectively risk adjusting insurer payments.

State officials should follow three major princi­ples when designing a statewide risk-transfer pool:

1. The scope of the risk-transfer pool should be as wide as possible. Ideally, participation in the pool should apply to all major medical health insurance policies sold in the state, and all insurers selling major medical coverage in the state should be required to participate.
   
   
2. The governance of the risk-transfer pool should be confined to the participating insurers. Since a risk-transfer pool effectively functions as a reinsurer for insurance compa­nies, the pool should be self-governing and operate according to a set of rules and proce­dures agreed upon by the participating insur­ance carriers. The state insurance department would provide regulatory supervision for the pool, just as it does for all other insurers oper­ating in the state.
   
   
3. There should be little or no public financing of the risk-transfer pool. The purpose of the pool is to redistribute premium dollars among insurers so that no insurer bears a dispropor­tionate share of the costs of covering expensive patients. It should not be a mechanism for using tax dollars to subsidize insurers. Any public subsidies should instead be directed to helping individuals buy private coverage. If a state does use tax dollars to offset pool losses, any such payments should be well defined, minimal, and time-limited to avoid inadvert­ently creating incentives for insurers to shift costs onto taxpayers.

State officials can design innovative insurance market reforms that effectively create new patient-centered health insurance markets in which insur­ers compete to offer the best value for consumers' dollars. An important feature of such reforms will be to include risk-adjustment mechanisms that give health insurers the right incentives to manage risks and costs better, as opposed to simply trying to avoid risks and costs. Comprehensive, consumer-centered reform holds greater promise for deliver­ing better value to patients than the alternatives of trying to patch up the seriously flawed status quo or merely expanding poorly performing public pro­grams, such as Medicaid or the State Children's Health Insurance Program, to cover discrete chunks of the uninsured population.

The Importance of Risk Adjustment

From the patient perspective, a consumer-choice health insurance system is desirable because it forces insurers and providers to compete for their business by offering the best results or value for consumers' dollars. However, to work effectively, a consumer-choice health insurance market needs to ensure that the payments to health plans are com­mensurate with the varying costs of care among individual enrollees. This will give plans the neces­sary incentives to provide both the healthy and the sick with the best value relative to their individual needs and health status.

The most basic form of health insurance risk adjustment is to permit some variation in premiums according to one or more associated factors, such as age or geography. For example, charging age-rated premiums for health insurance coverage offers a kind of rough equity as it adjusts somewhat for dif­ferences in the likely consumption of medical care. In general, older individuals tend to consume more medical care than younger individuals consume. Older workers also tend to earn more than younger workers earn and thus can afford to pay somewhat higher health insurance premiums.

However, within any age group, income bracket, or geographic area, some individuals will have bet­ter or worse health than the norm. Thus, policy­makers should look to additional, more formal risk-adjustment mechanisms as a way to ensure that health plans have incentives to accept rather than avoid higher-risk individuals and to optimize the care provided to those individuals in terms of cost and results.

Two Approaches

There are two basic approaches to constructing health insurance risk-adjustment mechanisms: pro­spective and retrospective. In the prospective approach, the premiums paid to insurers are adjusted in advance to reflect the expected medical expenses of different enrollees. In the retrospective approach, the insurers in a given market pool many of the costs associated with higher-risk individuals and then redistribute those costs proportionately among all subscribers. This is sometimes referred to as a "reinsurance pool," although the more accurate term is "risk-transfer pool."

Prospective Adjustment. Both risk-adjustment approaches have advantages and disadvantages. In theory, the prospective approach, to the extent that it more accurately matches plan payments to risks, should give health plans stronger incentives to man­age use of services by both healthier and sicker enrollees to provide better value for money. How­ever, prospective risk adjustment also has some dis­advantages.

First, it is difficult to design a system that predicts likely costs for different enrollees with an acceptable degree of accuracy.

Second, the system needs an adequate methodol­ogy for updating the risk-adjustment formula over time to accommodate changes in treatment costs driven by biomedical innovation. Medical innova­tions can change treatment costs significantly and rapidly, either upward or downward. For example, a new drug might provide much better results but also cost much more than previous treatments. In contrast, a new, less-invasive surgical technique might cost significantly less than the current proce­dure and produce better results.

In this regard, a prospective risk-adjustment sys­tem for insurers faces essentially the same issues as a prospective payment system for providers. Any errors, omissions, or lags in the updating process can throw the system out of alignment over time, reducing its effectiveness and creating unintended consequences.

Third, because of its tendency to become mis­aligned, a prospective risk-adjustment system also carries a higher level of political risk. That is, it is more susceptible to special-interest lobbying and legislative tampering to override decisions made by the experts who administer the system. Often, the justification is a supposed error in the methodology. Of course, those who benefit from any overpayment that arises from an error will seek to preserve their advantage, while any self-perceived losers will seek legislative changes.

Once lawmakers inject themselves into the details of the payment process, the exercise can quickly degenerate into endless arguments over whose pro­jections are more correct and whether supposed overpayments are justified either as inducements for desirable behavior or as compensation for sup­posed underpayments elsewhere in the system. In this regard, Medicare's experience is instructive. Congress began micromanaging the prospective payment systems for hospitals in the 1980s, intro­duced the Medicare fee schedule for doctors in the 1990s, and is now arguing about the payment rates of Medicare Advantage plans and Medicare Pre­scription Drug plans.

Fourth, using prospective risk adjustment to manage a multipayer market will be significantly more complex than managing a single-payer mar­ket, such as private Medicare Advantage plans[2] or Medicaid managed-care plans.

In public programs, such as when Medicare or Medicaid beneficiaries are given a choice of private coverage options, a prospective risk-adjustment system starts from the question of how to divide a given sum among a given enrollee population. Fur­thermore, because the target population is relatively well defined, adjusting for future changes in the size and characteristics of the enrollee population is a simpler exercise.

In contrast, the equation is much more complex in a multipayer system, such as when employers offer their workers a choice of coverage through defined contributions in a health insurance exchange.[3] The size and risk distribution of the enrollee population are harder to predict, both ini­tially and over time. Furthermore, the payments from multiple sources must be aggregated into one pot and then redistributed according to an agreed formula. While such an approach is theoretically possible, it raises additional technical and political issues, such as the need for some type of public or private entity to collect and distribute the funds.

Retrospective Adjustment: A Better Way. In contrast, retrospective risk adjustment largely avoids the uncertainties inherent in any prospec­tive approach. By using incurred claim costs as its starting point, a retrospective system not only avoids these issues at the outset, but also effectively self-corrects over time for changes in the underly­ing cost structure driven by advances in medical technology and biomedical innovation. The remaining projection issues, such as estimating the next six to 12 months of pool losses so that carriers can build most of these costs into their premiums for the next plan year, are relatively straightforward and easily managed.

Properly constructed, a retrospective system also avoids or mitigates the political risk inherent in pro­spective systems. Because a retrospective system is based on actual experience instead of projections, it becomes more difficult to argue with the methodol­ogy. A retrospective risk-pooling arrangement will still require resolving some methodological issues, but they are fewer and less complicated.

The major issue would be dealing with the differ­ences in excess claims costs associated with acute versus chronic conditions. For example, the distri­bution of expensive acute incidents (e.g., major trauma cases) among competing plans is largely by chance. In contrast, any skewed distribution among competing plans of patients with chronic conditions (e.g., diabetes) is more likely the cumulative result of decisions made by individuals when selecting their coverage.

From the perspective of patients, it is better if individuals with chronic medical conditions choose plans that help them manage their care better, but the payment methodology for the risk-transfer pool would need to adjust for the extra expenses incurred by those plans. The best way to address these different types of risks is for the pool to apply somewhat different rules to the different types of risks, such as different rules limiting the size and share of claims that carriers can cede to the pool.

The main disadvantage of retrospective risk adjustment is the potential for an improperly designed plan to create a moral hazard. The most obvious potential moral hazard is that retrospective risk adjustment could reduce incentives for plans to manage the medical costs that they would pass on to the pool. In risk-transfer pools, this is typically addressed by establishing "risk retention" or "risk corridor" rules, which stipulate that a carrier can cede only part of a particular risk to the pool. Thus, the carrier continues to pay a portion of the claims as an inducement to manage the total risk.

The second potential moral hazard is that the pool may become an avenue for carriers to shift risks and costs to a funding source outside the insurance market, particularly the taxpayers. For example, during the 2004 presidential campaign, Senator John Kerry (D–MA) advanced a federal reinsurance proposal that would have created this kind of moral hazard. His proposal would have allowed insurers and employers to transfer 75 per­cent of health claims costs in excess of $50,000 to the federal government, which would serve as the equivalent of a reinsurance pool.[4] The first problem with such an approach is that it gives insurers an incentive to try to transfer ever more costs to tax­payers. The second problem is that the government would likely respond to rising pool costs by impos­ing more regulations on providers and restricting patient access to expensive treatments as ways to hold down costs.

Reducing Taxpayer Burdens. The best solution to this second type of moral hazard is to construct the risk-pooling arrangement as a closed system with no public financing. In other words, the pool would simply be a mechanism for evening out dis­parities among private plans by shifting a portion of the excess costs incurred by plans with above-aver­age shares of expensive claims or patients to plans with below-average shares.

If lawmakers decide to include some public financing for a risk-transfer pool, these dangers can be mitigated by explicitly limiting the size and tim­ing of any public subsidies used to offset pool losses. However, state lawmakers could appropri­ate public monies to offset the startup costs of designing and administering a risk-transfer pool -- but not pool losses -- without risking the creation of moral hazard.

While government programs and the private sec­tor continue to make progress in developing more effective prospective risk-adjustment systems, retro­spective risk-pooling arrangements are on balance easier to design, can be implemented in a shorter time, and are less likely to develop significant oper­ational problems over time.

Designing an Effective Risk-Transfer Mechanism

Because insurance carriers naturally tend to seek their own advantage when designing a risk-transfer pool and because there is no single best way to design such a pool, state lawmakers should adhere to the following basic guidelines when crafting leg­islation for a retrospective risk-transfer pool:

• The scope of the pool should be as wide as possible. It should encompass all commercial major medical coverage sold in the state, both group and individual, in and out of a health insurance exchange. Participation in the pool should also be a requirement for insurance carri­ers to sell major medical insurance in the state.
  
  States cannot require self-funded employer plans, which are exempt from state regulation under the federal Employee Retirement Income Secu­rity Act, to participate in a risk-transfer pool. However, they might consider allowing such plans to participate voluntarily under "Lloyds rule" terms, meaning that they can join the pool and transfer their excess claims to the pool as commercial insurers do, but only if they agree to pay pool funding assessments for three years fol­lowing any decision to withdraw from the pool.
  
  
• All insurers participating in the pool should have equal rights to cede risks to the pool in accordance with rules established by the pool.
  
  
• All funding assessments should be applied on a proportional, covered-life basis, not accord­ing to the percent of premium. In this way, the expenses associated with high-cost individuals are spread evenly among all policyholders in a given market instead of falling disproportion­ately on those consumers who opted for more expensive plans.
  
  
• Member insurers should self-govern the pool in accordance with the rules that they estab­lish themselves. No outside individuals or enti­ties should be involved in governing the pool. This would not preclude the pool from contract­ing with third parties for services such as admin­istration, accounting, and actuarial support, but those entities would not be involved in pool governance.
  
  
• The pool would operate under the regulatory supervision of the state insurance commis­sioner, just as with any other insurance company.
  
  
• The enabling legislation should set a deadline for the carriers to submit a plan for the design and operation of the pool to the state insur­ance commissioner. The deadline should be determined based on discussions among law­makers, carriers, and the state insurance depart­ment about a reasonable time frame. The plan would take effect upon approval by the commis­sioner. In the event that the carriers cannot agree on a design or the commissioner finds one or more design features unacceptable, the commis­sioner should be given authority to modify the design or to draft another plan and then imple­ment the result.

This approach forces the affected carriers -- which likely have divergent interests -- to agree on a set of operational details, such as attachment points for ceding risks to the pool, premiums paid to the pool by ceding carriers, and the share of claims retained by the primary insurers. The additional political effect is to shift these issues off center stage in the overall reform discussion and into a separate room where the insurers sort out the details among themselves, with the insurance commissioner standing outside the door as the backstop.

Reasonable Expectations

Finally, state policymakers should clearly under­stand the limits of risk adjustment as a policy tool and not invest it with any unrealistic expectations.

First, neither a prospective nor a retrospective risk-adjustment mechanism will directly reduce total medical costs, nor is that the intention of such mechanisms. Rather, the purpose of risk adjustment is to redistribute costs more fairly so that health insurers do not have incentives to avoid covering sicker individuals, but rather have incentives to manage better the risks and costs of medical care for all enrollees, regardless of their health status. How­ever, to the extent that risk adjustment comple­ments and reinforces other structural reforms, such as consumer choice and ownership of coverage, it will drive health plans to focus on improving the value proposition for their enrollees and contribute to the creation of a value-maximizing health system.

Second, both prospective and retrospective risk-adjustment designs can address only "second-order" selection effects in health insurance markets. These selection effects can occur when individuals are free to choose among competing health insur­ance plans offering different benefit designs. In a consumer-choice market, the resulting distribution of high, low, and average risks will not necessarily be proportional among the plans in the market. Risk adjustment can adjust for any skewed distribution of risks so that plans are appropriately compen­sated, but in a way that does not otherwise restrict consumers' choices of coverage.

Risk-adjustment mechanisms cannot address the "first-order" selection effects of individuals choosing between coverage and no coverage. For that, other mechanisms are needed. In the case of employer-sponsored coverage or public programs, establish­ing "all or none" rules, together with some kind of auto-enrollment in a default plan -- with individuals then given the option of choosing another plan off the menu -- is a way to limit first-order selection problems.

For instance, if a company wants to offer its workers coverage through a health insurance exchange, it must agree to offer insurance to the whole group, not only to select individuals, and select one of the competing plans as the auto-enroll­ment plan for its workers. The burden is then on the worker to select different coverage or to explain why he or she should be allowed to decline cover­age altogether, as would happen if the worker is already covered under his spouse's plan. If the lim­ited guaranteed-issue provisions applied to employer-sponsored coverage and public program coverage (e.g., Medicaid managed care) are also applied to the non-group market, they would need to be accompanied by some form of personal responsibility requirement on the affected individu­als, either to obtain coverage or to take responsibil­ity for their medical expenses so as to avoid or minimize first-order selection effects.

Conclusion

Risk-adjusting payments to insurers can ensure that health plans have incentives to accept higher-risk individuals and to optimize their care in terms of cost and results. In particular, risk adjustment can improve the functioning of consumer-centered health insurance markets. It addresses the concern that health plans that do a better job of managing care for sicker patients would be disadvantaged because they would attract a larger share of high-cost subscribers when individuals are free to choose their health plans. Instead, well-designed risk-adjustment mechanisms give health plans the right incentives to provide better value to both healthier and sicker enrollees.

The purpose of risk adjustment is to correct for any natural tendencies toward risk segmentation that produce a skewed distribution of risks among competing insurers in a market while still preserving the benefits of insurer innovation in coverage design and consumer choice of coverage. Rather than attempting to limit risk selection by imposing stan­dardized benefit packages or restricting consumer coverage choices, risk-adjustment mechanisms accept the reality of selection effects in the insurance market and manage them in ways that accommodate the needs of both insurers and consumers.

Addressing the reality of risk segmentation when designing competitive health insurance markets is akin to addressing the reality of gravity when designing airplanes. Like gravity, risk segmentation is a natural phenomenon that the designer must respect and accommodate. However, in neither instance does the natural phenomenon pose an inherently insurmountable barrier to designers' achievement of their intended objectives.

While policymakers can design and implement risk-adjustment mechanisms that work either prospectively or retrospectively, retrospective approaches, such as risk-transfer pools for health insurers, are somewhat easier to design, can be implemented in a shorter time, and are less likely to develop significant operational problems over time.

Edmund F. Haislmaier is Senior Research Fellow in the Center for Health Policy Studies at The Heritage Foundation.