|Home | About | Journals | Submit | Contact Us | Français|
Medicare spending accounts for a substantial fraction of Federal spending, and significant program changes may be necessary for long-run fiscal balance. We used a microsimulation approach to estimate how benefit changes to Medicare–including Part A, for hospital care, premiums, premium support credits, and changing the eligibility age–affect long-term Medicare spending and enrollment. All policies considered reduce spending, with reductions ranging from 2.4 to 24 percent between 2012 and 2036. However, the policies also reduce coverage among the elderly. To achieve significant costs savings without causing substantial uninsurance among seniors, benefits changes would likely need to occur in combination with other options.
The Congressional Budget Office projects that–without policy intervention–Medicare will account for 21 percent of all federal spending and 6 percent of gross domestic product by 2035.1 The increase in Medicare spending, which currently accounts for 14 percent of federal outlays, is a major factor in projected growth of the national debt. The Patient Protection and Affordable Care Act of 2010 (ACA) introduced a number of policy changes to reduce Medicare costs, with the largest expected savings stemming from reductions in payments to Medicare Advantage plans (private plans that have historically received more funding per enrollee than traditional fee-for-service Medicare) and reductions in annual payment increases for certain types of providers, including hospitals. However, some studies suggest that lower provider fees increase the volume of services provided, reducing the efficacy of these policies as a cost containment option.2 Others have argued that, without policies that fundamentally change Medicare’s cost structure, Congress may face irresistible pressure to reverse the features of the Affordable Care Act intended to limit Medicare expenditure growth.3
The newly created Independent Payment Advisory Board (IPAB)–which will advise Congress on additional ways to reduce Medicare spending–may not consider most changes that would increase costs to Medicare beneficiaries, including altering Part B premiums, which cover physician visits, lab services, and equipment like wheelchairs, adding Part A premiums, or increasing deductibles or copayments for care. The Independent Payment Advisory Board is also prohibited from considering policies that would modify eligibility requirements, such as increasing the age of eligibility. Despite these restrictions, from 2015 through 2019 the Independent Payment Advisory Board is charged with making recommendations that keeps annual Medicare cost growth at a level between the average inflation rate for all goods (measured using the Consumer Price Index) and the average inflation rate for medical services. After 2019, the Independent Payment Advisory Board’s recommendations must constrain Medicare growth to be less than gross domestic product growth plus 1 percentage point. Because the Independent Payment Advisory Board is limited in the policies that it may consider, many fear that the Board’s primary tool to address cost increases will be further reductions to provider payments. The CMS Office of the Actuary has cautioned that the Board’s requirement to reduce costs given its constrained set of options represents an “exceedingly difficult challenge.”4
Hesitancy to consider policies that directly affect beneficiaries demand for care is not a new phenomenon for the Medicare program. The Medicare eligibility age, for example, has remained constant at 65 since the program began nearly 50 years ago, while life expectancy at birth in the US increased by 7 years. Further, Medicare Part A has never required a premium. One policy lever that has been used frequently over the years is the Part B premium. Currently, single individuals with incomes under $85,000 pay a monthly premium of 25 percent of Part B costs ($104.90 in 2013); this amount increases on a sliding scale, reaching 80 percent of costs for individuals with incomes above $214,000. Thresholds for couples are twice the individual amounts (e.g. $85,000 becomes $170,000). In this analysis, we evaluate the long-term spending and enrollment effects of three large changes to the Medicare program that would go beyond the policies allowed under the Affordable Care Act. Specifically, we consider imposing means-tested premiums for Medicare Part A, converting the program to a premium support plan, and increasing the eligibility age to 67. The policies considered reflect a range of strategies that address the fundamental trade-off of whether to provide a rich benefit for few individuals, or a spare benefit for many. At one extreme, increasing the eligibility age preserves current structure of the Medicare benefit but provides it to fewer people, while–at the other extreme–premium support credits provide a defined contribution amount to all individuals and require them to shoulder the costs of health care cost inflation above the credit amount. Because the Part A premiums we considered are means-tested, they fall in the middle of the spectrum, providing a richer benefit for those with lower incomes. The Independent Payment Advisory Board is prohibited from considering these policies, although such policies may be necessary for long-run cost containment.
Funded primarily through hospital insurance payroll taxes, Medicare hospital insurance, or “Part A,” has never required a premium contribution from enrollees. The possibility of adding a means-tested premium for Part A was originally suggested in the mid-1990s by the bipartisan Kerrey-Danforth commission on entitlement reform.5 Although the Kerrey-Danforth proposals were never adopted, means-testing has recently been suggested in the context of averting the fiscal cliff.6 In our main scenario, we impose a Part A premium that scales with income, requiring premium contributions of 5 percent of expected total spending for individuals with incomes below $85,000, 10 percent for incomes between $85,001 and $107,000, 15 percent for incomes between $107,001 and $160,000, 20 percent for incomes between $160,001 and $214,000, and 25 percent for incomes above $214,000. Income limits are double for couples, and no contribution is required for individuals who are dually eligible for Medicare and Medicaid due to their low income. We assume that individuals with incomes below $85,000 must pay 5 percent of Part A costs, rising on sliding scale to 25 percent for individuals with incomes above $214,000. We assume that thresholds for couples are twice the individual amounts. In scenario testing, we consider a means-tested Part A premium that mirrors the Part B schedule, with premiums starting at 25 percent of expected spending for individuals with incomes under $85,000, rising to 80 percent of expected spending for individuals with incomes over $214,000. We also consider an alternative scenario where the Part A premium equals 10 percent of expected Part A spending.
In all three scenarios, we assume that hospital insurance tax payments, currently a 1.45% tax on wages paid to employees with earnings under $200,000 ($250,000 for couples) and a 2.35% tax on higher earners, are required regardless of Part A take-up. In effect, the tax payments are providing guaranteed access to hospital insurance at a subsidized price, where the amount of the subsidy may vary depending on the individual’s income. There are at least two arguments for requiring a premium as opposed to increasing payroll taxes. First, because current payroll taxes are used to fund current Medicare spending, the payroll tax transfers resources from younger to older generations. Imposing a Part A premium would ensure that those currently receiving the benefit share at least part of the burden of cost containment. Additionally, while all workers are required to contribute payroll taxes, not all workers will receive the same benefit because some–particularly minorities and lower income individuals–will die earlier than others. Imposing a modest premium, as opposed to increasing payroll taxes, would provide workers with a hedge against the possibility that they may not live long enough to recoup their payroll tax investment.
In our second scenario, we assume that Medicare enrollees receive a premium support credit that compensates them for current Medicare Parts A and B spending, less the Part B premium. For example, in 2014, an individual with income under $85,000 would receive a credit of $8900, which would cover approximately 85 percent of estimated expenditure for hospital and medical services. The credit could be used to purchase health insurance, but could not be refunded if an individual opts not to enroll. Although we do not model the details of new enrollment options, we assume that the credit could be used to purchase private health insurance or to buy into a Medicare-like plan. Over time, the credit is indexed to the growth rate of gross domestic product.
In our third scenario, we consider increasing the Medicare eligibility age from 65 to 67. Such a change would mirror current Social Security eligibility, and has been discussed in numerous proposals.7
We assume that all of the hypothetical options would take effect in 2014. Further, we assume that the policies are rolled-out all at once, rather than being phased-in over time.
We project spending using the Future Elderly Model (FEM),8 a simulation model that tracks cohorts over age 50 to project health status and economic outcomes, using data from the Health and Retirement Study (HRS), the Medical Expenditure Panel Survey (MEPS), and the Medicare Current Beneficiaries Survey (MCBS). Status quo Medicare enrollment in the model reflects currently observed patterns. Since Part A has no premium, virtually all adults over the age of 65 are currently enrolled. Enrollment is slightly lower for Part B. Those not enrolled in Part B tend to have low incomes but are not Medicaid-eligible.
Medicare enrollment and total health spending are projected using regressions which account for age, health conditions, disability, and other characteristics. All spending estimates are reported in net present value, using a discount rate of 3 percent. In our baseline scenario we constrain the rate of Medicare cost growth to reflect targets set forth in the Affordable Care Act, which are to be achieved through the policy levers above, notably reductions in provider payment rates and the recommendations of the Independent Payment Advisory Board. The Affordable Care Act-related Medicare spending policies play a role complementary to the strategies we consider, since they focus largely on providers rather than on consumer demand for services.
Modeled Part A premiums have two effects on enrollment. First, there is a one-time drop in enrollment caused by the initial imposition of the premium, which we set at 3 percent. Second, there is a gradual change in enrollment as premiums rise with medical inflation. We set this elasticity at −0.1 based on the literature,9 and because this number accurately predicts Medicare Part B take-up under pre-Affordable Care Act policies.
We assume that premium support credits can be used only for health insurance, and that enrollees pay any difference between health insurance costs and the credit amount. We assume that health insurance costs continue on the path specified in the Affordable Care Act, even after imposing the premium support program. After 2018, because this growth rate is pegged at gross domestic product plus 1 percent, the credit becomes less valuable over time. We treat the difference between predicted Medicare cost growth and the credit amount like a health insurance premium. In sensitivity analyses, we consider an alternative policy where premium credits are indexed to the consumer price index.
Exhibit Exhibit11 shows the predicted savings for the three policies considered, relative to the Affordable Care Act. Under the law, total Medicare spending between 2012 and 2036 cumulates to a net present value of 16.7 trillion. Imposing a Part A premium reduces spending by the smallest amount of the three options considered, achieving a 2.4 percent reduction between 2012 and 2036. At the other end of the spectrum, increasing the Medicare eligibility age reduces cumulative spending by 7.3 percent. The premium support plan provides relatively small savings through 2019, the time period during which the Independent Payment Advisory Board is required to maintain strict limits on spending growth. However, after 2019, the premium support plan achieves savings that outpace increasing the Medicare eligibility age.
Exhibit Exhibit22 considers the Part A enrollment effects of each of the three policies, relative to the Affordable Care Act baseline. Because this analysis is intended to show who is at risk for being uninsured, we consider enrollment among people ages 65 and over even in the case where we are evaluating a change in the Medicare eligibility age. Imposing a Part A premium using the main premium scenario described above reduces projected enrollment in 2020 from 57.7 to 56.6 million, a decline of just over 2 percent. However, the other two policies lead to a roughly 13 percent decline in enrollment, leaving between 7 and 8 million seniors without hospital coverage. Disenrollment under the premium support program stems from the fact that enrollee contributions will be required to cover the difference between the credit amount and the actual enrollment premium.
Part B enrollment is unaffected by Part A premiums, so–in Exhibit Exhibit3–there3–there is no distinction between the Affordable Care Act and the Part A premium scenarios. Reducing the Medicare eligibility age reduces Part B enrollment from 52 to 45 million by 2020, and from 72 to 65 million by 2036. The premium support program has little effect on Part B enrollment before 2020, due mostly to Independent Payment Advisory Board’s strict spending targets. However, by 2036, the premium support program reduces Part B enrollment by approximately 6 percent.
The Congressional Budget Office has estimated that increasing the Medicare eligibility age to 67 will lead to $148 billion in savings between 2012 and 2021, compared to our estimate of $405 billion in savings.10 The difference is explained primarily by the fact that we assume policy takes full effect in 2014, while the budget office assumes it begins to take effect in 2014, and is not fully phased in until 2027. The budget office’s projected cumulative savings from 2012–2035 is 5 percent reduction, which is more similar to our 7.3 percent cumulative reduction between 2012 and 2036. While the budget office has analyzed the effects of a Medicare premium support plan, these estimates are not directly comparable to ours because they were estimated only in the context of larger policy changes (such as Rep. Paul Ryan’s Path to Prosperity proposal).11 Moreover, the premium support program analyzed by the budget office would take effect in 2021 or later, while we assume implementation in 2014. The Congressional Budget Office has not considered the effects of Part A premiums.
Exhibit 4 summarizes our main results, along with results from sensitivity analyses. The sensitivity analysis in line 2 shows how spending projections change if we assume the savings provisions set forth in the Affordable Care Act fail to constrain costs. Eliminating these savings increases cumulative Medicare spending by 14 percent between 2012 and 2036. This large difference points to the importance of the assumption that Medicare payment level adjustments are maintained, and that the Independent Payment Advisory Board successfully achieves its target spending reductions.
Line 3 of Exhibit 4 reports sensitivity analyses related to Part A premiums. The first row summarizes results from the main Part A premiums scenario, which leads to a reduction in spending of 2.4 percent between 2012 and 2036, and a similar decline in Part A enrollment. We next consider a steeper premium schedule, based on the current schedule for Medicare Part B. Imposing a larger premium contribution leads to much greater savings, with cumulative spending falling from $16.7 to $13.8 billion between 2012 and 2036, a decline of 17 percent. However, steeper premiums also lead to a much greater decline in coverage, resulting in only 46 million seniors with coverage in 2020, a 20 percent reduction relative to the Affordable Care Act baseline scenario. The large decline in coverage may make such large premium contribution requirements untenable, especially given that most beneficiaries have paid taxes with the expectation of receiving hospital insurance later in life.
We also consider a flat premium contribution equal to 10 percent of per capita Part A spending for all enrollees, regardless of income. Relative to the means-tested scenarios, this approach is simpler and does not create cliffs where a small increase in income can lead to a large change in contribution requirements. This scenario yields an additional $100 billion in savings relative to the main means-tested scenario, but also leads to an additional 500,000 people without coverage.
In line 4 we compare the baseline premium support scenario to consumer price indexed premium support credits. Relative to the original premium support scenario, indexing to the consumer price index saves considerably more money, since consumer price index growth tends to be lower than gross domestic product growth. Under consumer price-indexed credits, cumulative spending declines by 24 percent relative to the baseline Affordable Care Act scenario between 2012 and 2036, compared to only 5.4 percent with gross domestic product indexing. However, consumer price indexing causes an additional 5 million people to drop Part A coverage relative to the GPD-indexed scenario, since the value of the credit relative to the expected cost of health services is diminished.
Finally, in line 5, we show the effects of increasing the Medicare eligibility age to 70, an option that reduces cumulative spending between 2012 and 2036 by approximately 20 percent. However, this option leads to a more substantial decline in enrollment than any of the other options, reducing enrollment in both Parts A and B by approximately 30 percent. The reduction in spending is smaller than the reduction in enrollment because younger Medicare enrollees (those in the age range of 65 to 70) tend to have lower average spending than older enrollees.
Despite their likely effectiveness at reducing costs, all of the policy changes considered have advantages and disadvantages. Importantly, the policies impose a burden on enrollees, by increasing the out-of-pocket spending amounts required to obtain health insurance. Increasing the eligibility age leads to a loss of coverage among people just over 65 years of age, but it preserves the current generosity of the Medicare benefit for the oldest individuals. Due to the Affordable Care Act, individuals losing Medicare coverage may have access to affordable insurance either on the health insurance exchange or through the Medicaid program. However, particularly in states that do not expand their Medicaid programs, some low income seniors may fall through the cracks. There will also be secondary effects on federal spending (not measured in this report) if seniors are shifted from Medicare to the exchanges or to Medicaid. For example, the Kaiser Family Foundation estimates that exchange premiums would increase by approximately 3 percent if 65 and 66 year olds became ineligible for Medicare.12
Instituting Part A premiums or a premium support plan would preserve at least the offer of Medicare coverage to everyone who is eligible today. However, due to the new cost-sharing requirements associated with these options, some individuals will not enroll. In our main Part A scenario, which institutes a premium contribution that is generally smaller than what would be required for employer-sponsored coverage, spending decreases by about 2.4 percent between 2012 and 2036, and Part A enrollment declines by about the same amount. A back-of-envelope calculation suggests that, to achieve the same saving with a payroll tax increase, taxes would need to be increased by approximately $60 per worker per year (in net present value). Instituting a Part A premium schedule based on the current schedule for Part B leads to more significant savings, but–because Part B requires premium contributions that range from 25 to 80 percent of expected health spending ($1796 to $5749 in 2010 dollars)–Part A enrollment declines by 17 percent, leaving 12 million seniors without hospital insurance. Such a massive reduction in enrollment could lead to considerable challenges, including a large number of seniors foregoing necessary treatment or receiving uncompensated care. Requiring a flat 10 percent contribution for all Medicare Part A enrollees leads to spending declines of 4 percent between 2012 and 2036, while Part A enrollment falls by just under 3 percent.
A premium support program indexed to gross domestic product reduces spending by approximately 5.8 percent between 2012 and 2038, but this effect varies over time, with most of the savings coming after 2019. A consumer price-indexed program would lead to greater savings, but would also increase the share of the population without coverage. Both of these policies will have greater effects if Medicare expenditures grow more rapidly than expected, for example if the Independent Payment Advisory Board did not achieve its goals.
Anticipated growth in Medicare spending is a major challenge for the federal budget. We project that cumulative Medicare spending between 2012 and 2036 will amount to $16.7 trillion in net present value. Without the cost-saving provisions included in the Affordable Care Act, which require reductions in Medicare payment levels and assume the Independent Payment Advisory Board can successfully contain health care costs, cumulative Medicare spending could be as high as $19 trillion. The policies considered in this analysis lead to significant savings, ranging from a 2.4 to 24 percent spending reduction between 2012 and 2036. Yet, each policy requires sacrifices in terms of eligibility and burden imposed on enrollees, and policy makers must weigh the costs and benefits of alternatives. The most effective policies in terms of reducing spending–raising the Medicare eligibility age to 70 years or imposing a consumer price-indexed premium support program–lead to enrollment reductions of 13 and 30 percent respectively. If such loss of enrollment were to occur, there could be dire consequences for population health. Many of those who opt to disenroll would be low income individuals without good alternative health care options. These uninsured elderly would put a strain on hospital emergency departments, and some individuals would go without needed care. The Affordable Care Act’s required reductions in disproportionate share hospital payments, which compensate hospitals for serving low-income patients, would add to the problem by making it even more difficult for hospitals to shoulder this unfunded burden.
Consequences for enrollment are less extreme in scenarios with gross domestic product-indexed premium support credits or more modest Part A premiums. However, because these policies are also less effective at constraining costs, these options represent at best a partial solution to the challenge for rising Medicare costs, and would need to be simultaneous to, or in combination with, other cost containment approaches to substantially change the Medicare cost trajectory.
Despite these challenges, growing budget concerns make it unlikely that we can sustain the Medicare program as it exists currently. Optimal policy decisions in the face of resource constraints depend on broad societal goals–such as whether it is preferable to provide a generous benefit for few or a small benefit for many. Although no analysis can make these difficult choices for us, projections like these are essential to making informed policy decisions.
Christine Eibner, RAND Corporation, Arlington, VA.
Dana P. Goldman, University of Southern California, Los Angeles, CA.
Jeffrey Sullivan, Precision Health Economics, Boston, MA.
Alan M. Garber, Harvard University, Boston, MA.