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 © January 1999. The Urban Institute. All rights reserved. Except for short
quotes, no part of this paper may be reproduced in any form or utilized in any form by
any means, electronic or mechanical, including photocopying, recording, or by
information storage or retrieval system, without written permission from the Urban
Sang Chi and Misha Segal contributed substantially to this project, helping to develop
the basic model and producing the tables and charts contained herein.
TABLE OF CONTENTS
Early next century, the Medicare program must undergo a number of changes to respond to the aging of the baby boomers and the continued increases in health care costs. Searching for ways to reduce growth in future taxpayers' contributions to keep Medicare operating, many policymakers have argued for major structural reforms that combine attempts to influence the growth rate of health care spending with attempts to change the financing obligations of the federal government. This is the second in a series of papers focusing on options for reforming Medicare. Two major restructuring options have received considerable attention: defined contribution and premium support approaches.
Defined contribution options would simplify the government's obligation by tying the beneficiaries' per capita contribution to a set rate of growthoften keyed to a major indicator such as the Consumer Price Index (CPI) or gross domestic product (GDP)then turning that amount over to beneficiaries in the form of a voucher to help pay for a private plan. Premium support options would also stress private plans, but would link the government's obligation to a portion of the premium charged by one or more of the private plans serving Medicare beneficiaries. Premium support options also would replace the current program, although traditional Medicare might remain one of the "plans" vying to serve the beneficiary population.
These two restructuring approaches to Medicare could begin to look alike or take on very different characteristics depending upon the specific details of the options. In theory, at least, one of the most important distinctions between the two is whether beneficiaries bear the full risks of rising health care costs (in a defined contribution framework) or whether some of that risk is to be shared by government (under premium support). But in practice, even that distinction can be blurred. For example, if premium support options are set against the lowest priced plans or the percentage of health care costs declines over time, that option, like defined contribution, will shift the risks of higher-costs onto the beneficiaries.
These restructuring approaches also share a common view that competition among private plans will hold down costs above what the traditional Medicare program covers. But these claims are theoretical rather than demonstrated in practice. Nonetheless, for either option to work well and achieve savings without shifting significant cost burdens to beneficiaries, the option must successfully moderate growth. Even then, the variations in premiums likely to be associated with new plans can create special problems for beneficiaries, potentially placing the more vulnerable beneficiaries at risk.
The defined contribution approach modeled in this paper is, of necessity, limited to a set of basic assumptions. In the simulations presented, per capita amounts for the vouchers are tied to the overall growth rate of GDP and to the per capita GDP growth rate (where per capita refers to the entire U.S. population). The first limit allows Medicare to expand as a share of GDP in response to growth in the number of beneficiaries participating in Medicare. The second limit is the more stringent of the two. The impact on beneficiaries is projected as the amount of new costs that a beneficiary would have to absorb in higher premiums if the voucher does not grow fast enough to cover the costs of covered services over time. This "defined contribution gap" may be reduced if health care costs grow more slowly because of the competition engendered by this restructuring. If so, the gap would be the difference between the voucher and the new, lower rate of per capita spending necessary to provide basic
Medicare services. We model estimates of the defined contribution gap faced by beneficiaries assuming both no change in health spending and reduced expenditures (with per capita spending 0.5 percent lower each year).
This restructuring approach places Medicare beneficiaries at risk. When the constraint on what the federal government will pay is the stringent per capita GDP growth limit, the taxpayer share of Medicare as a share of GDP could be reduced by nearly 20 percent as compared to the baseline (from 4.62 percent of GDP to 3.71 percent). But even if health care spending slowed, beneficiary liability would rise by $889 (see summary table 1). And if health care spending did not moderate by 2025, beneficiaries would be hit with a 65 percent increase in liability and as much as a 41 percent increase in the amount they would have to spend on acute care services over the projected baseline amount. At that point, beneficiaries would be liable for more than one-third of the costs of Medicare spending.
The premium support approach would likely achieve smaller savings and put beneficiaries less at risk than would a defined contribution approach. The basic idea behind premium support is for the government to pay a share of the costs of the premiums charged by plans participating under Medicare. As with the Federal Employees Health Benefits Program, offering a range of plans with beneficiaries paying higher premiums for more expensive plans would give individuals a range of options to choose from. But the details matter significantly because they help determine how much protection is actually available to beneficiaries and how the benefits are distributed across the population. Setting the premium based on the lowest bidder, for example, would not ensure that most beneficiaries would be able to find a plan at that premium level. If the guarantee were set assuming a premium that few would be able to obtain, the premium support approach would begin to look more like a defined contribution option. Tying the guarantee instead to a median priced plan would provide greater protections for beneficiaries. But even in this case, there would likely be substantial variation in the costs of plans and in what benefits beneficiaries would receive.
Two levels of premium support are estimated: requiring either a 15 or a 20 percent contribution from beneficiaries as a condition of receiving Medicare. This would replace the existing Part B premium, which under current law is expected to reach a maximum just below 14 percent of the total costs of Medicare. Since this analysis focuses on just one typical beneficiary, the simulation of premium support looks essentially like a straight premium increase.
For this option, potential savings to the federal government depend both upon the level of the required premium from beneficiaries and upon whether the rate of increase in health care spending falls over time. Since the risks of health care spending would be shared between beneficiaries and the government, lower health care spending would be felt at the federal level as well as by beneficiaries. For example, when beneficiaries are required to pay 20 percent of the median priced plan, by 2025 the taxpayer share of GDP would decline to 4.45 percent from its baseline projection of 4.62 (summary table 2). But if we assume that the growth rates in health care spending decline by the same percentages assumed for the defined contribution approach, the taxpayer share of GDP would decline further, to 3.98 percent by 2025.
The impact on beneficiaries from premium support options would depend upon how much variation would arise in premiums and which plan beneficiaries chose. For example, if the rate is set at the median priced plan, but not all plans (including traditional Medicare) are able to hold the line on growth in spending, many beneficiaries might face premiums over time that are substantially larger than the increases implied by looking at the median priced plan. In the example developed for this paper, the premium could rise by as much as $1,657 in real terms in 2025 in a high-cost plan as compared to a $433 rise for the average priced plan.
Restructuring options could reduce federal spending on Medicare over time, but they would likely also result in substantial increases in beneficiary liabilities as well. These results hold even if restructuring slows the rate of growth in health care spending. Further, the impact of higher burdens would vary across beneficiaries, depending upon which plans are available to and/or chosen by them. The key questionwhether this type of restructuring is necessary or desirablecan be answered only after a broader range of reform options is explored.
|Summary Table 1
Effects of Defined Contribution Options
on Medicare and Its beneficiaries, 2025
||Growth Rate Limits
|Medicare Taxpayer Share
as a Percentage of GDP
|Per Capita Defined
Assuming no change
Projected Out-of-Pocket Costs
as a Share of Income
Assuming no change
|Source: Urban Institute analysis.
Summary Table 2
Effects of 80 Percent Premium Support Option
on Medicare and Its Beneficiaries, 2025
Health Care Costs
|Medicare Taxpayer Share
as a Percentage of GDP
|Increase in Beneficiary Liability
for Persons in Median Priced Plan
|Increase in Beneficiary Liability for
Persons in Traditional Medicare
|Projected Out-of-Pocket Costs
for Beneficiary in Traditional
Medicare as a Share of Income
|Source: Urban Institute analysis.
As the next century nears and baby boomers begin to approach retirement, the Medicare program, which serves the elderly and disabled, will need to undergo a number of changes. A variety of proposals to help resolve the financing problems facing Medicare have been suggested, ranging from major restructuring efforts to minor adjustments in the current system. The debate about Medicare's future needs to move beyond rhetoric to analyses of the implications of proposals; understanding the tradeoffs across various approaches can contribute to the debate about potential changes. Each option will have different consequences in terms of savings generated and impacts on beneficiaries. And since much of the concern about Medicare extends to the period after the baby boom generation becomes eligible for the program, it is important to examine these consequences over time.
In the first in a series of papers on Medicare's future, we presented projections of Medicare spending and beneficiary out-of-pocket costs assuming continuation of Medicare under current law.(1) The focus of both these papers is on the impact of various options on
beneficiariesa consequence that sometimes gets lost in discussions about potential savings to the federal government. This second paper examines the effects of two major approaches to restructuring Medicare that have been discussed as possible options for the future: defined contribution and premium support options. While the lines sometimes blur between these two and some analysts use the terms interchangeably, there are some clear distinctions that can be made that, at a minimum, define choices along a continuum.
Describing the Basic Approaches
A defined contribution approach would move Medicare away from offering a guaranteed insurance benefit for those eligible for the program and instead establish a fixed contribution toward the cost of health insurance for those beneficiaries. This would replace both the traditional Medicare program and the current Medicare+Choice option. In its purest form, the contribution made bears no necessary relation to the cost of the insurance, but rather provides a voucher tied to some target level of spending on Medicare. The generosity of that guarantee can vary considerably across proposals, but essentially the goal is to shift the risks of higher health care spending costs from the federal government to the beneficiaries themselves.
A premium support approach, as initially defined by Aaron and Reischauer, would shift some but not all of these financing risks to beneficiaries.(2) Here the idea is that both the government and the beneficiary would share the costs of insurance in ways designed to make individuals more responsible for the costs of their own care. As with a defined contribution approach, beneficiaries would be made more aware of the costs of various types of insurance plans in which they could choose to enroll, but under this approach the government contribution is more closely linked to the actual costs of care, implicitly allowing for more controls over the system than a pure, defined contribution approach would allow. In that sense, the premium support option is a more intermediate step between a defined contribution option and the current program. How close it is to either approach depends on design details.
A central feature of both these options is variation in the premiums that beneficiaries pay. Under the current system, beneficiaries are guaranteed the basic insurance package for the Part B premium and any required cost sharing. If they enroll in one of the private plan options available under Medicare, most plans are required to meet or exceed the same standard. That is, managed care plans and most of the new private plans allowed under Medicare+Choice can only charge a premium for services offered in excess of the basic Medicare package.(3) This is the nature of a defined benefit.
Under both restructuring approaches being analyzed here, variation in premiums is not only allowed, it is an intended goal of each option. That is, if beneficiaries are liable for at least part of any additional costs of insurance above what is assumed to be the level for the average plan, beneficiaries will face incentives to seek lower-cost plans. Plans can differ in the extent to which they place bounds on these variations, but the reforms explicitly penalize those who choose higher-cost plans, thus encouraging beneficiaries to shift into less costly options (often assumed to be managed care). Indeed, this is one rationale for claiming savings under these plans.
It is generally assumed that individual premium payments would be larger for defined contribution proposals than for a premium support approach, because under defined contribution (voucher) plans, the federal contribution usually is stated as a fixed dollar amount and not related to the costs of care. It is possible, therefore, that under voucher schemes all plans would charge premiums high enough to exceed the level of the Part B premium (even after adjusting for the projected growth in the Part B premium under current law). A very tight limit on growth in the dollar amount of the premium over time likely would lead to such a result if, for example, health care spending grew substantially faster than the voucher amount. Further, if this represented a full-fledged voucher approach with few controls on plan behavior, there would likely be substantial variation in the costs of insurance policies offered to Medicare beneficiaries.
Premium support options are usually described in terms of a contribution tied to a fixed percentage of the overall insurance premium, either set by the federal government or derived through a process of negotiated rate setting or competitive bids. In a premium support option that allows for multiple premiums, the percentage guaranteed by the federal government could be tied, for example, to the median plan's offering or to that of the lowest-cost plan. If set at the median level, those choosing lower-cost plans would face lower premiums both because of the premium itself and because the federal contribution would be a dollar amount representing a higher-cost plan; hence, the actual premium share would represent a higher percentage for less expensive plans. Alternatively, premium percentages could be set to vary across high- and low-cost plans. For example, one early version of such a model from the Medicare commission would gradually reduce the percentage contributed by the federal government as the premium levels rose.(4) Over time, if the premium percentage that the government absorbs remains constant, the beneficiary would receive more protection against higher health care costs than under the defined contribution approach. That is, the share would remain constant under
premium support, but could decline over time with the defined contribution option. But as Aaron and Reischauer note, policymakers could reduce the percentage over time as a way to further save costs to the federal government.(5)
Modeling the Options
Both of these options seek to reduce the per capita costs of Medicare, which are projected to grow over time at rates shown in chart 1. This illustrates growth both from projected increases in general prices of goods and services as measured by the Consumer Price Index (CPI) and additional amounts reflecting a combination of changes in the age distribution of the covered population, new technology, and medical care price growth above the general CPI rate. This baseline projection from the Trustees' Report of 1998 shows a decline in growth net of the CPI toward the end of the analysis period.(6)
These growth rates result in projected real (inflation-adjusted) per capita Medicare costs that rise from $7,425 in 1998 to $13,309 in 2025. As shown in chart 2, this can be divided into the federal government's taxpayer share and the liability that beneficiaries face. Our measure of beneficiary liability captures the average spending that individuals are responsible for paying as part of participating in the program. Medicare requires both payment of a Part B premium and cost-sharing contributions from its beneficiaries. In 2025, the taxpayers' burden is $8,836 per capita, while the beneficiaries' projected liabilities would be $3,074.
Potential savings for Medicare from both these restructuring approaches could come from three sources: shifts in costs to beneficiaries as a result of the level set for the government's contribution, changes in beneficiary behavior that reduce costs when a greater share of beneficiaries switch to lower-cost plans, and price reductions in premiums when private plans compete more aggressively. The shifting of costs to beneficiaries is the easiest source of savings to model and predict. For each of the approaches, we consider a number of different levels of public commitment.
The claim for savings from behavioral changes by beneficiaries is based on theoretical arguments about competition that have not been tested for this group of the population and for which only limited evidence exists. Thus, this is a more difficult estimate of savings to model. There is no strong evidence to point to lower-costs over time for private insurance performance as compared to Medicare; indeed, most of the evidence cited in favor of the private insurance market refers only to comparisons based on experience in the 1990s. But in fact, across a longer period of time, per capita Medicare spending has had a lower cumulative rate of growth than has the private sector.(7)
Somewhat stronger evidence can be brought to bear for comparisons between Medicare and the Federal Employees Health Benefits Program (FEHBP). Over little more than a decade, FEHBP has had lower average rates of growth than Medicare.(8) But because we are comparing different population groups and benefit packages, a more in-depth analysis would be needed before concluding that the choice arrangements established by FEHBP would generate savings for Medicare. For example, one of the rationales for arguing that FEHBP is more efficient than Medicare is that it encourages individuals to shift to less expensive managed care arrangements. But ironically, the share of FEHBP retirees in managed care is nearly the same as the share enrolled in Medicare's managed care option at present. Moreover, only 2 percent of such federal retirees shift their FEHBP coverage in any year, and many of those retirees are in plans specifically criticized by experts as undesirable.(9) That is, the costs of the high option plans are extraordinarily high, due not to differences in benefits offered but to the poor health status of enrollees. This "adverse risk selection" is another concern arising from relying on private plans. At best, modest savings might be achieved from making Medicare more like FEHBP.
Proponents of greater reliance on private plans also argue that insurers who would offer coverage to Medicare beneficiaries would respond to these lower contribution rates by competing more aggressively on price, thus lowering the amount of additional costs that would be shifted to beneficiaries. In areas of the country where multiple insurers offer coverage under Medicare, fear of losing business might be a stimulus to hold the line on health care costs. On the other hand, if these limits were imposed on Medicare and not on other parts of the health care system, insurers might focus their business on the under-65 population and offer premiums for Medicare beneficiaries that would pass on most of the difference between the projected costs of health care and Medicare's fixed contribution. Indeed, the negative response by the HMO industry to even modest limits on Medicare payments over the past two years in response to Balanced Budget Act (BBA) changes suggests less pricing flexibility than advocates
of managed care have claimed.
In practice, the specific details of any option will greatly influence what, if any, savings actually occur. For example, will traditional Medicare be retained as an option and if so, what would be the contribution arrangements by the federal government for traditional Medicare? (This is more likely to be an issue in premium support, since many defined contribution plans would do away with traditional Medicare.) Unless traditional Medicare is priced beyond the reach of many beneficiaries, it is likely to remain the choice of a majority of beneficiaries for some time to come; in that case, savings attributed to lower overall spending would depend to a considerable degree upon whether traditional Medicare is modified to achieve savings (such as relying upon extensions of BBA changes). If Medicare is not retained as the "default" option, will some other plan serve that role? And if so, will there be special protections offered for those with health problems, since this plan is likely to receive a greater than average share of high risks?(10)
Thus, it is difficult to argue conclusively for a substantial set of savings to be attributed to these options over and above the costs that would be shifted to beneficiaries. For both the defined contribution and premium support options, we assume a set of savings estimates derived from cost shifting only and another set that adds modest reductions in the rate of growth of per capita Medicare spending.
For both types of options, we assume that any new system would be put in place in 2002, since time would be needed to implement changes and that is the date when most of the current BBA changes expire. For simplicity, we assume immediate full implementation of a constant set of rules affecting how the voucher rates and the premium support levels are set. In addition, we project one alternative that generates savings (also assumed to begin immediately) that would stem from improved efficiencies by relying upon a market approach. That is, we assume a 0.5 percentage point reduction in each year's projected per capita rate of growth in Medicare expenditures.
Several important issues related to these proposals are beyond the scope of the current analysis but will be modeled at a later time. It is important to recognize that they are also key to interpreting the effects of these options. If some beneficiaries do shift into lower-cost plans, one of the key issues for modeling would be who would remain in higher-cost plans, effectively paying a penalty for such a choice. If the trend occurs as it has for Medicare's current managed care option, the older and sicker beneficiaries would be more reluctant to shift arrangements and thus might be the ones left in traditional Medicare or in plans whose costs rise rapidly. If so, then the market mechanism will work to the disadvantage of the most vulnerable.(11) Recognizing the impact of variations in the amounts of cost shifting onto beneficiaries is of crucial importance, particularly with regard to determining the likely winners and losers under these approaches.(12)
Another factor that will be modeled later is the generosity of the Medicare benefit package. Many approaches for restructuring Medicare, such as the premium support option, have assumed an expansion of the basic benefit package. That may be critical for issues of risk selection and whether managed care plans can remain competitive if fee-for-service options offer many of the same benefits. Further, since premium costs to beneficiaries for such an expanded package might be much higher, it is essential to consider what further protections for low income beneficiaries would be relevant.
Defined Contribution Options
There are many ways to design a defined contribution approach. The practical details, such as whether traditional Medicare would remain in place, what benefits would be offered and/or required, and what geographic variations would be allowed are crucial design details that would have to be addressed in evaluating any serious proposal. Nonetheless, it is possible to consider some of the implications of a defined contribution approach by focusing on the basic principle of restricting the government's contribution to the costs of a health insurance package. The model used here is intended to illustrate the magnitude of impacts of various options, and of necessity is limited to a set of basic assumptions in order to streamline the projections.
A key to savings from this option is the mechanism used to establish the annual rates of growth allowed for the vouchers; in turn, this will also affect beneficiaries to the extent that it results in a shift in costs to them in the form of higher premiums or greater out-of-pocket costs. To bound the estimates on beneficiary burdens, we assume a worst case scenario in which the full costs of the difference between the federal contribution established by formula and the projected costs of providing Medicare services are borne by beneficiaries. A lower bound estimate of beneficiary burdens assumes that only part of that "gap" is passed on, assuming that greater efficiency in the provision of health care over time could be achieved by lowering the rate of growth of per capita Medicare expenditures.(13)
In a defined contribution world, the impact of these lower growth rates would be to reduce the "premium gap" that individuals would be assumed to facedefined as the difference between the fixed federal contribution and what beneficiaries would have to pay. This means that if the federal contribution grows more slowly than costs of insurance, the premium would increase over time, shifting burdens from the government to beneficiaries. It could be argued that if , in practice, this gap were lower than implied by a full shift of costs, the federal government could more successfully maintain the growth rates established for the voucher payments. But for purposes of our analysis, the impact of lower expenditures affects only costs to beneficiaries.
Assigning Growth Rates for the Voucher
A range of growth rate targets was considered for this simulation, which would be applied to the per capita voucher amount. Some proposals have suggested using the CPI as the basis for the prescribed per capita voucher growth rate. Establishing a rate above the CPI would ensure positive real growth in the level of contribution from the federal government. A growth rate of CPI+1, for example, would allow an annual rate of growth of one percentage point over the CPI for the voucher. The justification is that health care spending prices rise faster than prices for other goods and services, and new technologies dictate that spending will likely increase over time at a positive real rate. But CPI+1 would still achieve substantial savings over time, since the baseline projections for per capita spending average a little over 2 percent per year above the CPI over the period between 2002 and 2025 (as illustrated in chart 1).
An alternative and more frequently proposed approach is to tie spending to GDP or a related measure. This is viewed by many proponents of a defined contribution approach as a way to link Medicare spending to society's ability to absorb new health care spending over time. At its most stringent, growth in the government's contribution could be set so as to remain a constant share of GDP. This is equivalent to setting the per capita growth rate of the voucher to growth in GDP divided by the projected Medicare enrollee population. That is, the growth in GDP would have to absorb the increase in the number of enrollees as the baby boomers become eligible for Medicare before allowing any increase in spending on a per capita basis. The difficulty is that this limit would result in a real decline in the government's
contribution for premiums over time. That is, the inflation-adjusted contribution would fall from $5,649 per capita in 2002 to $4,629 in 2025an 18 percent drop in the federal contribution. This approach is simulated but treated here as illustrative rather than as a serious option.
More likely would be less restrictive limits still linked to GDP growth and designed to achieve savings, but that are more realistic in the stringency of the constraint. For this analysis, we examine two alternatives: tying the defined contribution to aggregate GDP growth and tying the contribution to per capita GDP growth. The first approach effectively constrains the federal contribution to rise as a share of GDP only in response to growth in the Medicare beneficiary population. Per capita contributions for spending would not be allowed to grow faster than the economy. This is not a very stringent limitation relative to projected rates of growth in Medicare spending, but it does make logical sense that Medicare be allowed to grow as a share of the economy when it is serving a larger and larger share of the U.S. population. The second alternative, based on per capita GDP growth, would keep Medicare spending per enrollee from rising any faster than the growth in our national product after controlling for general population growth. Per capita GDP growth is essentially a measure of what we are
producing on average for each U.S. citizen. In that sense, it is a measure of average individual worth at any point in time. This is a more stringent limit on Medicare spending than the overall rate of GDP growth but less stringent than the option that would freeze the share of GDP that Medicare is allowed to consume.
The impacts on how Medicare would grow as a share of GDP as a result of these defined contribution options are shown in chart 3. The top line is the projected trend for Medicare assuming no change in policy. It is referred to as the baseline "taxpayer share" because it nets out the premiums that individuals now pay on Part B of the program. The remaining amount is the federal contribution required each year to cover expenditures under both Parts A and B of the program.(14) Over time, with no change in policy, the taxpayer share would rise to 4.62 percent of GDP by 2025 from its starting point of 2.52 in 1998. The bottom line in chart 3 keeps Medicare's share of GDP constant after 2002 by allowing per capita spending to grow only by what is available after netting out the effects of growth in the beneficiary population. The impact of the other two options, GDP growth and per capita GDP growth, would limit taxpayer share to 4.35 and 3.71 percent, respectively, in 2025.(15) The first of these effectively indicates what Medicare's share of GDP would have to be if Medicare were allowed to rise as a share of GDP as its number of enrollees grows.
Impacts on Beneficiaries from Defined Contribution Approaches
The greater the savings to the federal government from a defined contribution approach, the potentially greater the costs that are passed onto beneficiaries. This can be clearly seen in a comparison between charts 4 and 5, which illustrate the full gap that would exist if Medicare expenditures remained at their baseline projections but the federal contribution were limited by one of the two rate of growth assumptions shown in chart 3. The less restrictive approach, limiting the voucher to the rate of aggregate GDP growth, would result in an $835 increase in beneficiary liability by 2025 in 1998 dollars (chart 4). If a lower per capita GDP limit were placed on the voucher, the additional amount that beneficiaries would face in greater liabilities could total $2,011a 65 percent increase in the liabilities they would face as compared to the baseline level (chart 5).
These numbers are shown in the first line of table 1. The table also indicates the potential increase in liabilities from these defined contribution options if this approach were able to slow the rate of growth of spending on Medicare. In that case, not only would the gap faced by beneficiaries be lower, but the liability from the existing premium and the cost sharing that beneficiaries face as compared to baseline liabilities would also go down as overall spending moderated. These amounts are summarized for 2025 in the table. The assumption is for a 0.5 percentage point decline in baseline spending. For the voucher option based on overall GDP growth, this assumptions about lower spending would lead to reduced beneficiary liabilities because the voucher constraint would actually grow more rapidly than spending on
health care over the period. (It seems unlikely that policymakers would keep in place a "constraint" that rises faster than what it seeks to control, however.) The tighter per capita GDP constraint would still result in some shifting of costs onto beneficiaries, even with optimistic assumptions about moderating health care spending growth.
Another way of looking at this impact is to focus on the beneficiary liability as a share of total Medicare spending in 2025. The baseline projection is for beneficiary liability to reach 25.8 percent of total Medicare. As chart 6 shows, the liability would range from 23.4 percent to 42.7 percent of total Medicare spending depending upon the growth target used for the voucher and the assumption about how much Medicare spending would moderate over time.
Finally, chart 7 indicates projected out-of-pocket costs for a beneficiary remaining in traditional Medicare in 2025 for several of the scenarios outlined above. If costs were fully shifted to the beneficiary, the more restrictive voucher option would leave the typical beneficiary paying over 40 percent of her income for acute care expenses. Under that option, even the most optimistic projections of health care spending moderation would still result in a rising share of income devoted to out-of-pocket spending.
In sum, beneficiaries are placed at risk under this approach to Medicare restructuring. If the constraint on what the federal government will pay is fairly stringent, the rate of growth of Medicare as a share of GDP could be reduced by nearly 20 percent as compared to the baseline (from 4.62 percent of GDP to 3.71 percent), but even if health care spending slowed substantially, beneficiary liability would rise. And if health care spending did not moderate as its supporters claim, beneficiaries would be liable for a 65 percent increase in liability and as much as a 41 percent increase in the share of income they would have to spend on acute care services. Finally, as will be discussed in detail in the premium support options below, there could also be substantial variation in the burdens that individuals would face depending upon the range of choices offered the cost of those choices.
Premium Support Options
The basic idea behind premium support is for the government to pay a share of the costs of the premiums that plans participating under the Medicare program charge. The share would likely be set as a percentage of the costs of plans that compete for Medicare's business. As with FEHBP, this would have the advantage of giving individuals a range of options to choose from, varying in price and likely in delivery arrangements and benefit packages. This approach would establish incentives for beneficiaries to shop each year to find plans with lower-costs or more desirable characteristics. In doing so, it moves away from Medicare's standard Part B premium, which does not vary across individuals, and toward a system with highly variable premiums. Further, the details matter significantly because they help determine how much protection is actually available to beneficiaries and how the benefits are distributed across the population.
Setting the premium against the lowest bidder, for example, would not assure most beneficiaries that they would be able to obtain a plan at that premium. Assume the premium is set at 80 percent of the low-cost plan, for which the bid is $5,000. This price might be available only to a small number of beneficiaries. The resulting $4,000 government contribution would be at most a 67 percent premium support for those paying $6,000 or more (which could be the majority of Medicare beneficiaries). If the guarantee were set assuming a premium that few would be able to obtain, this approach would begin to look more like a defined contribution that does not protect beneficiaries well against the risk of rising health care costs. Tying the guarantee instead to an average or median priced plan would provide
greater protections for beneficiaries. But even in this case, there would likely be substantial variation in the costs of plans and in what beneficiaries would receive.(16)
Under yet another alternative approach, the premium share could be set at a fixed percentage across a range of plans, but then the federal contribution would rise with the cost of plans, perhaps creating a regressive structure that reduces incentives to shift to low-cost plans.(17) Such options also tend to place an upper bound on the government contribution, limiting what it will pay for those in the highest cost plans. But this approach could be
considered unfair to many lower income beneficiaries who might be able to afford only the lowest cost plans.
A further complication is whether these rates would be based on a national figure or set regionally. Either approach has several substantial disadvantages. Using a national figure recognizes the universal social insurance nature of Medicare but ignores the enormous geographic differences in the current distribution of Medicare costsdifferences that go well beyond costs of living or costs of doing business.(18) A national rate would mean that beneficiaries in rural Midwestern states would be heavily subsidizing persons in New York and Florida, for example. But setting the premiums locally would mean that few private plans would wish to set up business in the low-cost rural areas of the country, thus continuing the very uneven balance we now have in the availability of private plans participating in Medicare. This geographic conundrum would likely pose a major stumbling block in implementation since it would inevitably generate highly visible winners and losers across the country.
Another key issue is what would happen to traditional Medicare under this scenario. If traditional Medicare remains an option, will its premium, for purposes of being a competing plan, be set at the national average or at the local rates? And if it grows faster over time than other plansperhaps because it remains the default plan that many of the sickest beneficiaries choosewill it increasingly become unaffordable to the most vulnerable of Medicare's enrollees?
Defining the Premium Support Options
For this analysis, since we are focusing on just one typical beneficiary, the simulation of premium support looks essentially like a straight premium increase. This allows us to compare basic levels of savings achieved by such an option, but makes it more difficult to present results of interest on beneficiary burdens. Consequently, we provide here a few additional examples of how the variation in premiums could affect beneficiaries.
Although a premium support option that does not generate any savings solely from passing on higher costs to beneficiaries could be devised, it seems unlikely that a whole restructuring of the program would take place without using this mechanism to achieve some assured savings. Under current law, the Part B premium will rise gradually over time, reaching a maximum of about 14 percent of the costs of total Medicare. To achieve modest savings, we estimate an 85 percent premium support option that means a 15 percent contribution from beneficiariesan amount only modestly above the 12 percent that is now
expected under current law in 2002 and close to the maximum the premium is scheduled to reach.(19) We treat this as a set percentage that would be applied to the median plan under Medicare. As a consequence, the federal contribution would not vary depending upon who signed up for various plans; thus, it can be projected as a set percentage of projected per capita spending on Medicare. We also consider a premium support level of 80 percent, resulting in a higher 20 percent required beneficiary contribution.
In examining potential savings from the premium support option, we are faced with the same question facing defined contribution options of whether to assign savings from inducing a more efficient health care system. For simplicity and ease of comparison, we take the same approach as with the defined contribution options. This allows us to make comparisons on the basis of orders of magnitude between the two approaches, using similar assumptions about these savings. The mechanism for how these savings would flow to beneficiaries and the federal government differ, however, and it could be argued that they might have differential effects on reducing spending. For example, if traditional Medicare is retained in premium support, will that lead to smaller reductions in the trendline of health care spending than a
strict voucher approach?
Because premium support shares the risks between the government and beneficiaries, there are some key differences in this simulation as compared to the defined contribution simulations. The federal government would share in the savings that would be generated if premiums charged by private plans grew more slowly than Medicare's baseline. Not only would the government be paying a lower share of Medicare's costs, those costs would be assessed against a lower base.
These potential savings to the federal government are illustrated in chart 8, which assumes an 80 percent premium support (and that beneficiaries pay a 20 percent share). The taxpayer share of GDP would decline to 4.45 percent by 2025 as compared to its baseline projection of 4.62. If we assume that the growth rates decline by 0.5 percentage points each year, as assumed earlier for the defined contribution approach, this would lower the share of GDP to 3.98 percent by 2025.
Impacts on Beneficiaries from Premium Support Options
As noted above, the interesting part about the impact on beneficiaries for the premium support option is the variation in premiums that individuals might face. The impact of premium support options for the average individual in this model will simply show up as a cost shift equal to the difference between the current premium and whatever is established for the beneficiary requirement under premium support. This makes charts on the share of beneficiary liability less interesting, as these shares would remain constant across all the expenditure assumptions and vary only depending upon how much the premium increased. For example, moving to a 20 percent premium would raise beneficiary liability, on average, by about 6 percentage points.
It is possible to estimate how the beneficiary liability could vary depending upon choice of plans, however, by comparing liability for a beneficiary enrolled in the median cost plan with a beneficiary remaining in traditional Medicare. For purposes of simplicity, we assume initially that Medicare and the median price private plan would start from the same spending levels, and over time that savings from a lower growth in expenditures would be achieved by the median priced plan but not by traditional Medicare, which would remain at the baseline projection level. In practice, it would not be just traditional Medicare, but any plan that failed to grow at a rate as low as the median priced plan, where beneficiary liabilities would rise. While this is a somewhat oversimplified example, it illustrates what could happen under premium support.
As table 2 indicates, if there is no change in baseline spending levels, traditional Medicare would remain a competitive option that would grow at the same rate as private plans, and hence the premiums would be similar. Under assumptions of slowing rates of growth in private plans, those enrolled in the median priced plan would actually experience reductions in their premium contributions at a 15 percent level as compared to baseline projections. Even if the rate the beneficiaries were required to pay were increased to 20 percent, their premium would rise only modestly. But for the person remaining in traditional Medicare, a slower rate of growth for the median priced plan relative to what we expect under policy changes for Medicare would mean substantially larger premium increasesperhaps as much as $1,657 higher in 2025. This would occur because the premium contribution would be established on a plan experiencing lower growth rates. Consequently, the effective percentage contribution from the federal government for its 80 percent support level would fall to 71.6 percent for traditional Medicare if growth rates in private plans slowed by 0.5 percent per year.
The baseline projections for total out-of-pocket costs for beneficiaries in traditional Medicare in 2025 (in real dollars) would be $4,855. Under the premium support approach this could rise to a minimum of $4,992 when the beneficiary requirement is 15 percent and no change in baseline spending occurs, and up to $6,512 assuming a 20 percent beneficiary contribution coupled with a reduction in the rate of growth of private plans (see chart 9).
And if these numbers for traditional Medicare beneficiaries were translated into shares of income spent on health care, the numbers would range up to 39.4 percent of income for those in traditional Medicare or other high priced plans in 2025 (chart 10). Thus, while beneficiaries able and willing to choose median priced plans would likely face only moderate increase in out-of-pocket costs, the upside risk for beneficiaries from this approach are substantial. Would the beneficiaries in high cost plans mainly be those with higher incomes, or would they be the very old or those in ill health who are afraid to switch? The characteristics of those faced with substantial premiums make a great deal of difference in how this option is viewed.
In the search for ways to reduce what taxpayers will be expected to pay to keep the Medicare program operating, many policymakers have argued for major structural reforms, often to combine attempts to influence the rate of growth of health care spending and to change the financing obligations of the federal government. Two such options that have received considerable attention are defined contribution and premium support approaches. Defined contribution options would simplify the government's contribution by tying it to a set rate of growthoften keyed to some major indicator such as the CPI or GDPand turning over that amount to beneficiaries in the form of a voucher to apply to the purchase of a private plan. Premium support options would also stress private plans, but would link the government's obligation to a share of the actual premium charged by one or more of the private plans serving Medicare beneficiaries.
These two approaches to restructuring Medicare could begin to look alike or take on very different characteristics, depending upon the details defined around the options. In theory, at least, one of the most important distinctions between the two is whether beneficiaries bear the full risks of rising health care costs (in a defined contribution framework) or whether some risk is shared by government (under premium support). But in practice, even that distinction can be blurred. For example, if premium support options are set against the lowest priced plans or if the percentage of the costs of care guaranteed declines over time, those options, like defined contribution approaches, will shift the risks of higher-costs of care onto the beneficiaries.
These restructuring approaches also share a common view that competition among private plans will hold down the costs of care over and above what the traditional Medicare program can do. But these claims are theoretical in nature rather than demonstrated in practice. Nonetheless, for either of these options to work well and achieve savings without placing beneficiaries at significant disadvantages in terms of higher burdens, they have to be successful in moderating growth. Even then, however, the variation in premiums that will likely be associated with plans offered to beneficiaries can create special problems, potentially placing many vulnerable beneficiaries at risk.
These simulations indicate that restructuring of Medicare could create a number of problems for beneficiaries. For example, if a defined contribution plan were set to grow at the same rate as per capita GDP, even if modest savings were achieved in the rate of growth of health care spending, a 20 percent drop in taxpayer share of Medicare would be achieved while increasing beneficiaries' liabilities by 29 percent by 2025. At that point, beneficiaries would be liable for more than a third of the costs of Medicare spending.
Further, both restructuring options would result in differential burdens on beneficiaries, depending upon the costs of the plans that were available to them and what they chose. Under a premium support proposal that would raise the beneficiaries' average contribution to just 20 percent of the costs of the premium, savings would come disproportionately at the expense of those who remained in traditional Medicare or chose a plan with higher than average costs.
One of the key questions that can be answered only when a broader range of reform options is explored is whether this type of restructuring is necessary or desirable. The next paper in this series will examine incremental options that could be compared with these restructuring options. A key question that remains to be answered about balancing downside risks for beneficiaries from these restructuring options with potential savings is whether this is the best that we can expect for the future of Medicare.
1. See Marilyn Moon, "Growth in Medicare Spending: What Will Beneficiaries Pay?"
for a description of the issues surrounding these projections and the basic sets of assumptions that had to be made.
2. Aaron, Henry, and Robert Reischauer. "The Medicare Reform Debate: What is the Next Step?" Health Affairs vol. 14, no. 4 (Winter 1995), pp. 830.
3. The exception is the private fee-for-service option that was explicitly exempted from this requirement and touted in the Conference Report on the Balanced Budget Act as the first step toward a defined contribution model. See Federal Register, July 29, 1997, p. H6177.
4. This model would, however, allow the dollar contributions from the federal government to rise with the premium price, potentially resulting in greater federal subsidies for those who could afford higher priced plans. This could then reduce the progressivity of the Medicare benefit. Details of this option can be found in Bipartisan Commission on the Reform of Medicare, "Preliminary Staff Estimate: FEHB-Style System in Medicare," November 11, 1998, mimeo.
5. Aaron and Reischauer 1995.
6. Board of Trustees of the Federal Hospital Insurance Trust Fund. 1998 Annual Report of the Board of Trustees of the Federal Hospital Insurance Trust Fund (Washington: USGPO, 1998). The Medicare Trustees' intermediate projections assume that over time there is a gradual decline in the rate of growth of per capita Medicare expenditures. Ultimately it is assumed to approach the rate of growth of GDP. A second baseline adopted by the Commission assumes instead that Medicare per capita spending will not slow over time, but will continue at growth rates experienced in recent years. All other factors are held constant.
7. Moon, Marilyn, and Misha Segal. "Beneath the Averages: An Analysis of Medicare and Private Expenditures." The Henry J. Kaiser Family Foundation, forthcoming.
8. Bipartisan Commission on the Future of Medicare. "Key Issues in Creating a Premium Support Program for Medicare," staff paper. November 1998.
9. William Smith, presentation at Alliance for Health Reform. November 30, 1998.
10. One answer often posed for such a situation argues that we will improve the risk adjustment mechanisms used to balance Medicare's payments to private plans. But there are currently no ideal risk adjusters that could fully offset a substantial concentration of high-risk patients in one plan.
11. This is analogous to the situation of the high option Blue Cross plan in FEHBP, where, despite very high premiums, many retirees are reluctant to risk moving to another plan. For more discussion of related issues, see Marilyn Moon, "Will the Care Be There? Vulnerable Beneficiaries and Medicare Reform," Health Affairs, forthcoming.
12. Later analysis will allow us to look at a range of cases and hence to focus on this variability. To keep the model as simple as possible, we include two simple potential cases: one where the individual chooses to remain in traditional Medicare and one where the individual participates in an average priced plan.
13. This is a relatively modest figure in part because the Medicare baseline being used for this analysis assumes a slowdown in the rate of growth of health care spending after 2010. If we were using the higher "no slowdown" Commission baseline, a larger assumption about the ability to obtain savings might be in order.
14. This implicitly assumes that a defined contribution would be tied to the taxpayer share and that the individual's new contribution could be thought of as an addition to the existing level of the Part B premium.
15. If we put CPI+1 on this chart as an option it would be nearly identical to per capita GDP until about 2020, when it would be a somewhat less constraining limit.
16. In fact, for this option to achieve its goals, such variation is important. If plans clustered around the median, there would be less pressure for beneficiaries to shop and less pressure on plans to hold down their premiums.
17. This is closer to the approach taken by the Medicare Commission in its example plan presented in December 1998.
18. Barbara Gage and Marilyn Moon, "Analysis of State Level Variations in Medicare Expenditures," The Commonwealth Fund, forthcoming.
19. Aaron and Reischauer, for example, proposed an initial formulation that would set the federal contribution to start at about 5 percent below its current level. They also suggested the government share could be lowered over time to achieve further savings.
Tables & Charts