Expanding Health Insurance Coverage by Creating New Options: Some Thoughts on the Relative Attractiveness of Association Health Plans, Health Marts, and High-Risk Pools


Committee on Commerce Subcommittee on Health and Environment U.S. House of Representatives

( Testimony )

Author(s): Len Nichols

Published: June 16, 1999

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, its funders.

As a health economist and as an American citizen I am very pleased to be here before you today. My remarks, both written and oral, reflect my opinions alone and do not represent those of my employer, the Urban Institute, nor of its trustees or sponsors. * Most of the papers emanating from the research work that is reported upon were co-authored with my esteemed colleague, Linda J. Blumberg, who is here today and can answer any questions should I falter.

Four facts have brought us to this policy crossroads: (1) most Americans get their health insurance through some employment relation; (2) most of the uninsured are either workers or dependents of workers; (3) despite all the policy attention in the last few years, small employers are still much less likely than large firms to offer health insurance to their workers; and (4) workers in small firms are the most likely to be uninsured. These facts suggest that focusing on ways to get coverage to workers in small firms is the most important incremental reform strategy we could undertake in this country at this time. I applaud the committee for recognizing these facts and addressing this issue in a thoughtful and serious way.

At the same time I want to impart just a few words of caution, because I've learned the hard way that we all need to maintain modesty and think through the complicated interactions among the many parts of our health care system before proposing specific policies that will most likely help some people but hurt others. Real-world health policy is almost always about choosing among alternative positions along some kind of tradeoff function. "First Do No Harm," was the title of one of my first published papers on health insurance reform, and I think the admonition is a wise one for all policy discussions.

Most of my research effort is devoted to studying the theory and actual consequences of health insurance reform legislation as introduced by state, federal, and, increasingly, by foreign governments around the world. Pursuant to that research interest, with the help of my colleagues at Urban and elsewhere, I have built an elaborate model that can simulate the choices different kinds of employers make about offering health insurance to their workers. If our firms decide to offer coverage, the model then simulates their preferences among possible insurance vehicles, which of course in our country are many: commercial insurance, self-insurance, as a participant in a MEWA [multiple-employer welfare arrangement], or perhaps in a new AHP [association health plan] or a Health Mart, where these new options were modeled as described in the legislation that was proposed last year in the House.

Our results at this point are best characterized as preliminary, but since our ongoing work is so relevant to the topics you are discussing today, it seemed useful to share with you our as yet unpublished research findings.

Our simulated employers' choices among their many alternatives depend upon the factors we and others believe are most important to them in the real world: the relative premiums of the options, which in turn depend upon the presence or absence of benefit mandates, premium taxes, solvency requirements, different administrative loads, and state insurance market regulations.

All microsimulation models depend upon a long list of technical assumptions, which we try very hard to make consistent with what we can observe about the real world. These unavoidable assumptions do make all our conclusions contingent upon them. Different assumptions would produce different results, and Linda and I would be glad to discuss the details of our model with you or your staffs at your leisure. Still, the overarching lesson that can be distilled from our modeling exercise is, I think, quite intuitive: By far the most important element of health insurance choices is the risk pool to which groups or individuals have access.

By stressing the importance of the risk pool, I hope to remind you all of the principle that all insurance is about pooling risk. Indeed, the fundamental health insurance policy debate can be boiled down to a question of what kind of risk pools we really want: homogeneous and purely voluntary, versus heterogeneous and partially coerced. Homogeneous and voluntary pools have the virtue of forcing no one to pay for someone else's expected costs, and they serve the healthy with low premiums, but they also often leave the unhealthy uninsured. Heterogeneous and community pools have the virtue of subsidizing access to care for the unhealthy, but they can discourage coverage of the healthy, who may prefer the risk of being uninsured to the burden of this implicit tax. Policymaking is about choosing among these desirable yet imperfect alternatives, and wise policymaking is about trying to balance our competing objectives along the feasible paths which good analysts try to describe for you.

All health insurance market reforms-guaranteed issue, guaranteed renewal, portability, limits on pre-existing condition restrictions, restrictions on the variance of premiums-all of these reforms force more risk pooling than the market alone would achieve. What our research shows is that the nature of the resulting risk pools to which different firms have access is more important to employer choices about health insurance than the presence or absence of benefit mandates, premium taxes, and solvency requirements.

Some firms clearly prefer to go it alone: They have low-risk workers and dependents and can do quite well through self-insurance, especially if they're large enough to enjoy administrative economies of scale.

But most small firms appear to prefer more risk pooling than self-insurance allows, hence the relative popularity of commercial insurance and the potential popularity of AHPs and Health Marts, if carefully structured.

One reason these options are appealing is because they exempt participants from benefit mandates. And while benefit mandates, as the research literature suggests, may add little to costs on average, they can surely add considerably to the cost of some benefit packages, especially those preferred by some small firms. At the same time, more mandates make the packages more attractive to workers and probably increase worker take-up, so once again we have a tradeoff.

Again, our research simulations suggest that by far the most important factor determining the attractiveness of various health insurance options is the pool with whom the firm's workers will be joined for premium rating purposes. AHPs and Health Marts, to the extent they are exempted from state premium rating rules, will be more attractive to the good risks and less attractive to high risks in search of more heterogeneous pools.

MEWAs are attractive to firms of all sizes in some industries, but they are not a very large part of the overall private health insurance market today, and our simulations do not suggest that they're likely to grow a lot in the future. The interesting policy question is, which types of firms would want to join AHPs or Health Marts, and what would happen to the commercial, self-insured, and MEWA markets if these options came into existence next year?

Our simulations, based on our detailed assumptions, predict that Health Marts are not likely to be very popular, for the simple reason that their only real advantage over commercial products is exemption from benefit mandates, and that is simply not enough of a price advantage on average to entice many firms to choose to purchase health insurance through one, at least not when AHPs are also an option.

AHPs then appear to be the most popular new option that federal legislation might create, and our model suggests that they will be more popular after four years than MEWAs are now. Still, our results suggest that AHPs are not likely to capture huge shares of the market, with a little more than six percent of all workers. Somewhat surprisingly to us, almost four-fifths of the workers in our model who work for firms that will choose AHPs are currently in large firms (with more than 100 workers) as opposed to small firms (with fewer than 100 workers).

Perhaps even more interestingly, most new AHP enrollment appears to come from the currently self-insured, not from the commercial insurance market. This finding suggests to us that at least medium-sized firms also want a bit more pooling than self-insurance affords, but not as much as commercial insurance would impose, even without extensive regulation.

This result also suggests that much of the opposition to AHPs may have been a bit shrill, since the commercial risk pool does not appear to be destroyed by this new option. This opposition was based on the likelihood that AHPs are most likely to appeal to the healthiest risks. AHPs will appeal to good risks since they can practice more segmented premium rating practices than the commercial insurance industry is expected to, whether that commercial sector is regulated or not (we simulated both scenarios).

This segmentation increases the chances that firms will be pooled only with firms with similar cost structures. Thus AHPs represent a step toward homogeneous pools from a moderately heterogeneous status quo. And as such, they do represent a threat to established heterogeneous risk pools. But the real-world-based price responsiveness that we built into our model suggests that this threat is not likely to destroy the commercial market, though some firms may stop offering as low-risk groups leave and commercial premiums rise, as I discuss below.

At the same time, in our simulations, extremely few new firms are enticed to offer health insurance which they did not offer before the reform options were made available. Some firms do drop coverage, and the average firm size of those which do offer declines. These findings translate into the result that introducing the new options-AHPs and Health Marts-may actually reduce overall offer rates on net, though by such a small amount-less than one percent of all workers-that it probably should be considered as no net effect on employer offerings. Net coverage is reduced because the commercial and MEWA pools lose some of their best risks to the AHPs, and thus their pools deteriorate. Because of this risk-pool deterioration, some firms drop coverage rather than pay the new higher prices that go with this deteriorating risk pool.

These firms do not join AHPs, however, because that risk pool is too segmented for their taste and risk profiles. Most of the firms that drop coverage after AHPs are made available, by the way, were initially insured through MEWAs, not through commercial insurance. Our preliminary results also suggest that about half of all employment-based insurance policyholders experience a premium change of more than five percent, with winners slightly outnumbering losers. So we predict rather a lot of premium churning for relatively little coverage impact.

We intend to continue this line of research, to test the results against alternative assumptions, and to refine this model and to adapt it to address the implications of switching to an individual tax credit system rather than the current exemption for employer contributions to employee premiums. We expect to have results to report on this new model by September of this year.

High-Risk Pools

I'd like to devote the remainder of my testimony to reporting on one empirical result that was found in a couple of different studies, including one of my own with other Urban Institute colleagues. The studies tested for the effects of a number of state policies, including high-risk pools, on private insurance coverage. While the results are not definitive, they are strongly suggestive that the existence of certain kinds of high-risk pools leads to more nongroup coverage than would otherwise be the case.

Now you all know that most high-risk pools are very, very small, and I don't think the household surveys that underlie the best empirical work in this area are picking up many people who are actually enrolled in them. But the results do suggest that when high-risk pools provide reasonably comprehensive coverage, are reasonably subsidized, and are not limited by enrollment caps, the individual insurance market seems to work better. I think this is most likely to be because when insurers are confident the truly hard-to-insure are safely cordoned off in reasonable high-risk pools, then they are less fearful of adverse selection and thus offer lower prices, which in turn entice more, and especially more low-risk individuals to purchase nongroup health insurance. This suggests that adequately funded high-risk pools, without enrollment caps and with statutory definitions of high risk, can be useful components of a coverage-enhancing policy mix-not to mention the fact that they provide immense financial relief to the unfortunate families involved, and thus could serve a major equity role in our free society.

It is worthwhile to remember that in the absence of high-risk pools, these individuals still get care, but it is often uncompensated. We all share in that risk, in that we collectively subsidize their care, either with tax dollars going directly to the public facilities that provide the care, or through implicit surcharges that are added by providers to privately financed services and thus to private health insurance premiums. An appropriate high-risk pool strategy, which preserves family dignity and enables needed comprehensive care to be delivered at a time in the patient's illness episode when it is most likely to be effective care, can be a much more efficient risk-sharing mechanism than those "backdoor" channels we otherwise use.