The University of Michigan
555 South Forest Street
Third Floor
Ann Arbor, MI 48104-2531

T 734-936-9842
F 734-998-6341



Mention the topic of risk segmentation - when individuals with differing health care risk are divided into different products or different risk pools for setting insurance premiums - and eyes glaze. Yet, the issue of segmentation is a key reason, albeit a subtle one, why insurance is becoming unaffordable to millions of Americans and why a sudden illness or accident may also financially wipe out you, your family or your next door neighbor at any moment. As a result, policymakers need to address risk segmentation in order to close the widening uninsurance hole that threatens each individual whether or not they are healthy or insured at this moment. That may mean limiting choice or getting everyone covered and truly spreading the risk.

Interview with Linda J. Blumberg, Ph.D., Economist and Principal Research Associate in the Health Policy Center of the Urban Institute

  Urban Institute's Linda J. Blumberg, Ph.D., is a nationally renowned expert on private health insurance, health care financing and health system reform. She writes on and advises Congress on issues related to small businesses and insurance, adverse selection in private insurance markets, mechanisms for expanding coverage and tax incentives for health insurance purchases. Blumberg, who also advised the Clinton administration during its health care reform effort, spoke to ERIU about the role of risk segmentation, its effect on health insurance markets and how it hinders efforts to expand coverage and reform the health care system.


What does risk segmentation lead to?

BLUMBERG: It leads to healthy individuals ending up with lower premiums and people with higher actual and expected medical costs facing much higher premiums and/or out-of-pocket costs, either because they are in more expensive products or because they are in a pool with other unhealthy individuals or because benefit levels have been cut back.

If most Americans are healthy—10 percent of Americans account for 70 percent of health care costs—why is this bad?

BLUMBERG: While most people are healthy at a particular point in time, as we age, our health care needs have a tendency to increase. We also face substantial uncertainty about what our health situation will be from random bad events (getting cancer or another serious disease, getting in a car accident and sustaining serious injuries that need to be tended to). A system that provides accessible, adequate and affordable insurance coverage to all people protects all of us in the case of bad luck and in the face of aging, and eliminates the worry that a single bad event or job loss will cause us to lose our insurance, our access to necessary care, or lead to financial ruin. And be clear, the U.S. does not have a health care safety net designed to catch all who fall—there is no guarantee of a source for purchasing health insurance coverage in this country, regardless of how high your income is. 
Because most people are healthy at a given time, spreading the costs associated with high-need populations broadly would lead to increased costs per person that are relatively small. Conversely, leave the financial burden to be carried by the person who is sick or injured, and that burden will likely be crushing.

Can you give me some examples of risk segmentation and how it occurs?

BLUMBERG: The larger a group, the more similar the distribution of risk is to the population overall. With employment-based coverage, people are coming together to get insurance but they primarily are together for their job. In the large employer group insurance market, it is unlikely that any particular employer will experience average health care costs in a year that are well in excess of the population average, and the costs paid by the employer and workers generally reflects the firm’s own health care experience.  In fact most of the large employers self-insure their workers.   This is different than in the small employer group market, where the year to year variability in average health care costs for the small numbers of workers in a given firm is much greater, and it is more likely that some firms will be high risk and others low risk in a particular year.  Most small employers who insure their workers do so through commercial health insurance. In the small group market, every state has its own regulations governing small groups, including defining what a small group is. States have various rules about how much pooling of health care risks they force on those markets. So, if I am a commercial insurer and I am rating insurance in a highly regulated state, I may be forced to spread the costs of the high-cost firms broadly across all of the firms that I’m insuring. But if I am in a state that has looser insurance regulations, then I may have more leeway to rate up the premiums for the higher-cost small firms.  The more premiums are set to reflect the expected costs of a particular group being insured, the more segmentation occurs, leaving the higher cost groups to shoulder a greater share of their own health care financing burden and providing financial relief for those that are healthier in a given year. 

What is risk selection and what does it lead to?

BLUMBERG: Risk selection is really the outcome of enrollment decisions.  Certain insurance plans will end up with enrollees that are more or less healthy than average, i.e. with positive or negative risk selection. When purchasers are deciding whether or not they want health insurance coverage and, if they decide they want it, where they are going to try to get it and what kind they are trying to get, they make those decisions as a function of what their health care needs are expected to be. For example, an individual who is relatively healthy may select not to have health coverage or decide that they want to have coverage lower in actuarial value than someone who wants and expects to use a lot of services. So, a healthy person may choose a plan with a much higher deductible and higher cost sharing or more limited benefits. Whereas someone who expects to use medical services is more likely to look for comprehensive coverage and make sure that they get the providers they need, while minimizing their cost sharing.

As people or groups are making decisions based on their own expectations, you end up having risk selection in the market that is driven by that behavior. Insurers also drive some of the selection by the way that they set premiums and by the way that they market and limit access to particular types of plans.  For example, if an insurer denies coverage to enrollees with histories of health problems, its plans will generally enroll a healthier than average population (positive selection).  If a plan is the only one in an area that provides guaranteed issue of insurance to all applicants, it will tend to enroll a higher than average cost population (adverse selection).

Is the individual market where risk segmentation is most pronounced?

BLUMBERG: Yes. This is the market for which the classic adverse selection concerns are the greatest. Commonly, in the non-group markets there isn’t guaranteed issue—so people can be excluded completely, another segmentation tool. Or people are offered only certain types of coverage as a function of their health status. For example, benefit riders, allowed in many states, are written to exclude body systems or body parts or pre-existing conditions permanently from the policies that are being sold. Many states allow many more tools for non-group insurers to use to segment risks.

Some states like New York have tried to spread out the risk in the individual market. How has that fared?

BLUMBERG: In states like New York, which has pure community rating in the non-group market and guaranteed issue policies, you get the potential for a market death spiral because it’s a voluntary market. All individuals who buy into that market are paying the same rate for the same coverage regardless of their health status and everybody has to be issued a policy. The individual can come and go as they please. This approach requires any healthy people deciding to enroll to share in the costs of the high-cost enrollees in a very significant way. So the healthy people exit or never enroll, and the premiums go up for the people remaining in the pool, making the pool unsustainable over time without outside subsidization or support.

You’ve highlighted how segmentation works in the individual market, what about group markets?

BLUMBERG: One example may be the firm that offers two kinds of plans: a high deductible plan with or without a Health Savings Account (HSA), and a comprehensive plan. The people who are older or who expect to use more health care tend to gravitate towards the comprehensive plan, whereas younger, healthier people or the higher-income people may take advantage of the HSA.  The HSA has tax advantages that make them attractive to high income people, and the high-deductible plans that are associated with them do not dissuade higher income purchasers who can cover those costs if need be.  Lower-income people don’t have the funds to pay the high deductibles of HSA-compatible plans,  so a high deductible is unattractive to them. For the unhealthy, having very high deductibles to pay up front before reaping the benefits associated with these plans makes them unattractive.

So, if you’ve got a group situation where you have a HSA option and a comprehensive option, you are going to see people peeling off into these different products as a function of these preferences.  The lower cost workers will tend to gravitate toward the HSA-compatible plans and the higher cost workers will tend to move toward the comprehensive plan.   And, it’s really dependent on the employer in that context to decide how they are going to have the competing choices priced. Premiums charged to workers could reflect the difference in the risk of enrollees choosing the different options, or an employer could decide to price the options in a way that pools risk more broadly.  For example, premiums could be set so that those who go into the comprehensive policy pay a little extra for having more generous benefits, but they would not have to pay extra to reflect that those enrolling in that option tend to be more expensive than those enrolling in the high deductible plans. HSA-plan enrollees would experience some savings from enrolling in a lower actuarial value plan, but would still share in the health care costs associated with those enrolling in more traditional coverage.  The firm has a lot of leeway in terms of how to price the products their workers are facing, with their choices leading to more or less segmentation or risk within the firm.

How does segmentation hinder efforts to expand coverage in the reform of the health care system?

BLUMBERG: The biggest difficulty comes when you’ve got these variations in plans and how plan premiums are rated based upon the pool of individuals or groups that have voluntarily enrolled, while at the same time you have individuals who can opt not to get coverage at all. The result is the kinds of complexities that you see in non-group insurance markets occur on a broad and national scale in new programs.
Can you give me an example?

BLUMBERG: Let’s say I set up a new purchasing pool with easily compared pricing for standardized packages open to individuals and small groups, so you have a new location for purchasing insurance that hopefully lowers administrative costs and creates some efficiencies while guaranteeing access to coverage for all. The problem comes when individuals also have the option of buying coverage outside the purchasing pool, with differing rules in the non-purchasing pool market. There may not be guaranteed issue outside the pool like there is inside the pool. The people who can get better premiums outside the pool are going to stay out because they can get a better deal, or maybe they won’t buy coverage at all. The premiums inside the pool then increase to the point that they are just not affordable to anybody. There are real segmentation problems, especially if you do not create a mechanism for subsidizing the risk inside that pool.

It seems that choices only serve to segment insurance pools?

BLUMBERG: Yes, people love choice and choice always sounds like an unconditional positive characteristic of a plan. The truth is with insurance that is not the case. Choice lends itself to segmentation of risk. So, if you are going to maintain choice, which seems to be a political necessity in the U.S. at the moment, it actually makes it more difficult and more complicated to do reform effectively. You can do it, but it’s going to take more effort, and in my opinion, it is very likely to require either some explicit or implicit subsidization of high-cost individuals. You’ve got to either figure out a way to set individual prices so that they are not based upon the risk of individuals making a particular insurance choice or you’ve got to explicitly subsidize high-cost individuals in their purchase of coverage.

Going back to pools, what rules would help prevent healthy people from bailing out?

BLUMBERG: No. 1 would be an individual mandate—require everybody to be in coverage of some type. Once everybody is in the system, it’s a lot easier to redistribute the costs across people of different risk. If healthy people can get out of having to share in the risk with other people, you’re creating a gaping hole to achieving effective coverage. Without a mandate, those who choose not to purchase coverage should have to contribute to the costs of caring for the unhealthy through some other means, such as  a broad-based tax of some type.  This revenue can then be used to subsidize and help fund medical care for the high-cost population. So even if the healthy person doesn’t choose to participate in health insurance, he’ll still end up having to help pay for these people who are unlucky and have high medical needs. I think you either get everybody in and set prices in a way that makes it fair and affordable for everybody regardless of their health status, or you say I’m not going to force everybody to be in but I am going to find a way to collect money to help the people who are in. 
Regarding the problem of risk segmentation and the requirement for subsidies for high-cost individuals. How would you go about doing that?

BLUMBERG: There are a number of things you can do. John Holahan, Len Nichols and I developed a purchasing pool plan that would have individuals come in and buy comprehensive coverage. First, income-related subsidies would be available in the pool to make coverage affordable for those with modest resources. Second, if the pool attracted a group of enrollees whose average health care costs exceeded that of the population as a whole, those excess costs would be subsidized by the government.  In essence, premiums would be set as if everybody in the state with insurance coverage had enrolled in that pool. The difference between the actual costs in the pool for those who voluntarily enrolled and the population-based average health care costs would be paid for through a broad-based tax. That’s one way to do it. There are other ways too.
What about proposals to enlarge FEHBP or Medicare for purchasing pools? Do they make sense?

BLUMBERG: There is no guaranteed source for purchasing health insurance in this country. So, an individual without access to employer coverage, who may be unhealthy and in a state where he or she is unable to get a non-group policy and who is not eligible for a public program, may have zero sources for buying health insurance coverage. Reform needs to start with some guaranteed source for buying coverage and you can set that up as a new purchasing pool, SCHIP, Medicare, FEHBP, or state employee health plans as your base.  All are viable options to consider. Then you need to subsidize the low-income to make it affordable. Additionally, you need to think about the relative risk of those individuals likely to enroll, and how premium pricing is going to be set, and whether you are likely to have to subsidize health care risk in your pool, as we’ve discussed. But, you can use any of those sources for your core pool.

You suggest that catastrophic-like plans are one way that insurers are able to avoid the risk of enrolling sicker or at-risk people. Yet many economists suggest an efficient way to provide affordable coverage and perhaps expand coverage is via barebones-like plans. What’s your view? 

BLUMBERG: High-deductible policies are, absent other reforms, risk segmentation devices.  Higher deductible/higher cost sharing plans may be a reasonable insurance vehicle for higher income individuals and those without serious health conditions. However, they provide inadequate coverage for those who are modest income and they may very well impinge their access to necessary care. The same may be true for somewhat higher income people who have large health service needs -- the high cost sharing becomes a tremendous financial burden. If I take more of health care spending out of insurance and place it on the users of care, I'm increasing the financial burden of the high users and reducing it for the low users. The higher the cost sharing, the better off are the healthy, but at the direct expense of the unhealthy. As there are many more healthy than unhealthy people, the financial burden per person goes up for the sick more than it goes down for the healthy. If we found a broad based way to raise revenue (income tax, value added tax, etc.) that could be used to subsidize the higher cost population and the low income, we could undo that increased segmentation. But without such redistribution, a shift to high deductible policies will hurt the most vulnerable.

So how do we address the issue of the uninsured, rising costs of coverage (particularly for those who are sick or at-risk) when we look at it from risk segmentation and risk selection. What are the fixes policymakers should be considering?

BLUMBERG: I think there are 5 key components to successful comprehensive health care reform.  First, we need to develop a guaranteed source for purchasing insurance coverage.  Second, a comprehensive set of insurance benefits must be made available to the low and modest income population on a subsidized basis.  The third component is a mechanism for broadly spreading the costs associated with the high medical need population.  Ideally, the fourth component would be an individual mandate that everyone participate in insurance coverage.  And finally, we will have to develop more effective mechanisms for containing medical cost growth.  Our understanding of how to design a system of bringing adequate and affordable insurance to all Americans is actually much more advanced than our understanding of effective cost containment.  But cost containment will definitely have to include strategies for more efficient provision of medical care to high cost populations, as that is where the bulk of health care spending is concentrated. Evaluating the cost-effectiveness of new and existing technologies and procedures should be part of that, as should identification of effective and efficient strategies for managing the care of the chronically ill.  Health education and preventive efforts may also be helpful. But we are in our infancy with regard to these types of strategies. They require a significant upfront investment in research and analysis, and we don't have a great deal of strong information yet on what will work and what will not.

Back to top

Funded by The Robert Wood Johnson Foundation, ERIU is a five-year program shedding new light on the causes and consequences of lack of coverage, and the crucial role that health insurance plays in shaping the U.S. labor market.