Filling the need for trusted information on national health issues…

Modifying Medicare's Benefit Design: What's the Impact on Beneficiaries and Spending?

Limitations and Assumptions


An important limitation of this single-year analysis is that it does not consider or evaluate the potential long-term effects on health outcomes or the long-term spending effects of the Medicare benefit design options. Our model incorporates the assumption that as beneficiary cost sharing increases, beneficiaries would use fewer services, which could produce short-term savings to Medicare but which could also result in poor (or worse) health outcomes—thereby increasing costs to Medicare over the longer term. It is outside the scope of our analysis to determine whether beneficiaries would forgo necessary or unnecessary services, and the extent to which this would affect their health or expenses over the longer term, but previous research has suggested the existence of secondary (and unintended) effects of increasing cost sharing on both patients and insurers.1 For example, Trivedi and coauthors found that Medicare Advantage plans that nearly doubled copayments for ambulatory care experienced increases in hospitalizations, especially for enrollees with certain chronic conditions.2 Chandra and coauthors also found increases in hospitalizations after a large retiree health plan introduced copayments for physician services and increased copayments for pharmaceuticals.3 We also do not evaluate or incorporate any potential substitution effects that might offset certain reductions in utilization. For instance, beneficiaries who use fewer home health services because of new cost sharing might use more physical therapy or physician visits.

Based on the available evidence, it was not possible to model these longer-term effects with any degree of confidence. As a result, our model may overestimate the amount of savings to Medicare and underestimate the cost to beneficiaries that could result from increased cost sharing in the long run—for instance, if beneficiaries simply substitute some types of care for other services, or if their health deteriorates, requiring additional care in the future. At the same time, the single-year analysis could underestimate the share of beneficiaries with high out-of-pocket costs who could experience spending reductions associated with the new annual cost-sharing limit.4 These are important areas to explore in future research.

MEPS does not include beneficiaries in long-term care facilities, a group that tends to use a relatively large amount of services. Therefore, our model may understate the number of beneficiaries who would benefit from a cost-sharing limit. Another caveat is that many beneficiaries in long-term care facilities are dually eligible for Medicaid and Medicare. As a result, our model may understate the effects on Medicaid spending associated with the benefit redesign options. Altogether, only 5 percent of Medicare beneficiaries are in long-term care facilities, so we do not expect that including this population would have a substantial effect on our outcomes.


Modeling programmatic and policy reforms involves some degree of uncertainty and invariably requires a number of assumptions that may oversimplify individual decisions and responses, while averaging out variations in circumstances. We nonetheless took this approach to develop a greater understanding of the possible effects of reforming Medicare’s benefit design on beneficiaries and spending.

Full implementation in 2018. We modeled full implementation as of January 1, 2018 to assess the effects of Medicare benefit redesign options if implemented in one year, rather than phased in over time. We recognize the administrative challenges of implementing such changes by then. We also acknowledge possible legal issues associated with prohibiting first-dollar coverage for current Medigap policyholders rather than an approach that only applies to new policyholders (as in the new Medigap restrictions included in MACRA).

Changes in utilization. We assumed that an increase in cost sharing would cause beneficiaries to use fewer services (and vice-versa). To model this assumption, we used “induction factors” based on those used by the Health Care Financing Administration (HCFA, now the Centers for Medicare & Medicaid Services, or CMS), which in turn were based on the RAND Health Insurance Experiment. Specifically, for every $1 increase in cost sharing, we assumed that total spending would decline by $0.70 for physician and outpatient services, by $0.50 for home health care, and by $0.20 for inpatient hospital and SNF services.5

No changes in Medicaid eligibility. We assumed that Medicaid eligibility would not change under any of the benefit redesign options. State Medicaid programs currently pay all or a portion of Medicare’s cost-sharing requirements on behalf of beneficiaries who are enrolled in both programs. Because benefit redesign would alter beneficiaries’ cost-sharing obligations, it would subsequently affect Medicaid expenditures. For example, introducing a cost-sharing limit would shift spending from Medicaid to the Medicare program. If Medicaid programs chose to scale back coverage for optional populations in response to these changes, then our analysis would understate the spending effects of these options for low-income beneficiaries.

No erosion of employer-sponsored or other supplemental coverage. We assumed that the benefit redesign options would not affect the share of enrollees’ cost-sharing liabilities (including premiums, if applicable) covered by ESI and other supplemental coverage. We also assumed that the options would not affect employer offer rates or take-up rates by retirees. If some employers decided to drop retiree health benefits altogether, retirees would likely incur higher out-of-pocket costs and subsequently use fewer services, resulting in a decrease in Medicare spending. The magnitude of the effect on employer spending would depend on how much employers reduced coverage and how many employers did so. If employers or other supplemental payers were to make changes in coverage by shifting some, if not all, of the additional cost-sharing requirements onto beneficiaries—and therefore no longer covered the same share of beneficiary spending as under current law—we would expect savings to supplemental insurers and higher costs for beneficiaries with this coverage, relative to the effects we observe.

No effect on retirement decisions. We assumed the benefit redesign options would not affect retirement decisions in the short term. If changes in Medicare’s benefit design prompted workers to delay retirement and remain covered by employer-sponsored insurance as their primary source of coverage, out-of-pocket spending would decline for younger workers if their employer policies’ cost-sharing rules are more generous than the redesigned Medicare benefit, and Medicare spending would decline somewhat if Medicare then became the secondary payer for these workers rather than the primary payer. In this case, employer costs could increase.

Changes in Medicare Advantage and Medigap enrollment. We assumed that changing traditional Medicare’s benefit design—most importantly, with the addition of an annual cost-sharing limit—would lead to changes in Medicare Advantage and Medigap enrollment. Our switching assumptions are shown in Table 6. We used the estimates in the third column (“benefit redesign plus Medigap coverage restrictions”) for all four options in this report.

Table 6: Share of Medicare Advantage and Medigap Enrollees Switching Coverage Under Medicare Benefit Redesign Options in 2018
  Benefit redesign only Medigap coverage restrictions only Benefit redesign plus Medigap coverage restrictions
% of Medicare Advantage enrollees switching to traditional Medicare 2.4% No effect 2.4%
% of Medigap enrollees switching to traditional Medicare 2.4% 0.8% 3.2%
% of Medigap enrollees switching to Medicare Advantage No effect 3.1% 3.2%
SOURCE: Kaiser Family Foundation/Actuarial Research Corporation, June 2016.

The switching assumptions are expressed as a percent of current law enrollment in that supplemental type. The pool of possible switchers consists of an estimated 36.4 million beneficiaries in 2018: 21.3 million Medicare Advantage enrollees (58 percent of the pool), 9.5 million Medigap enrollees (26 percent), and 5.7 million traditional Medicare beneficiaries with no supplemental coverage (16 percent).

The literature does not provide clear guidance on how to establish these “switching” parameters. Our estimates are based on the reasoning described below.

  • Enrollment switches under Medicare benefit redesign. One reason why beneficiaries might choose to purchase Medigap coverage or enroll in Medicare Advantage is that traditional Medicare currently lacks protection from catastrophic medical expenses through a cost-sharing limit. Thus, we assume that adding a cost-sharing limit to traditional Medicare would induce a small share of Medigap enrollees to disenroll (2.4 percent) and a small share Medicare Advantage enrollees to switch to traditional Medicare (2.4 percent).
  • Enrollment switches under Medigap coverage restrictions. Because the restrictions would only apply to Medigap, we assume that some Medigap enrollees (3.1 percent) would switch to Medicare Advantage where they would still have the option of paying a premium (as applicable) for more generous coverage. We also assume that because some beneficiaries purchase Medigap policies in order to avoid the hassle of paying medical bills directly, a small share of Medigap enrollees (0.8 percent) would drop this coverage and revert to traditional Medicare only if coverage restrictions were imposed on Medigap plans.
  • Enrollment switches under the combined options. For all four options in this report that combine benefit redesign plus Medigap coverage restrictions, we added the estimates for each separate option and increased the total slightly to incorporate the effects of compounding.

In general, we do not expect Medicare benefit redesign options to have a dramatic effect on switching between coverage types. Because the switching effects we modeled are small relative to overall Medicare enrollment, even moderate changes in these assumptions are not likely to affect the direction of or conclusions drawn from our main outcomes of interest. However, if a larger share of Medigap enrollees drop their Medigap policies and switched to traditional Medicare only (or to Medicare Advantage) in response to the modified benefit design, then federal savings would be of larger magnitude, while more beneficiaries would see spending reductions since the former Medigap policyholders would no longer be paying Medigap premiums. If a larger share of Medicare Advantage enrollees switched to traditional Medicare, then we would expect to see greater federal savings, to the extent that payments made to Medicare Advantage plans are slightly higher than what traditional Medicare would pay per beneficiary. It is not possible to quantify more specifically what spending effects we would observe under different switching assumptions. The magnitude of the effect on spending would depend on the specific benefit design option under consideration, since the overall effects are highly sensitive to design details, such as the deductible amount and the cost-sharing limit, for example.

Take-up of new subsidies for low-income beneficiaries. We assume that all Medicare beneficiaries who are SLMBs, QIs, and Part D LIS enrollees who are not already receiving full cost-sharing assistance from Medicaid or Medicare Savings Programs would automatically receive the new 100 percent cost-sharing subsidies. However, Option 3, as modeled, does not subsidize all low-income Medicare beneficiaries. We did not include other low-income beneficiaries who might be eligible for but are not enrolled in SLMB, QI, or LIS, and beneficiaries with incomes below 150 percent of poverty but assets above program eligibility thresholds. Including these low-income beneficiaries would increase federal spending on this option relative to the current effects estimated by the model, but would also increase the share of low-income beneficiaries who would experience spending reductions under the modified benefit design that includes low-income subsidies.

Findings Discussion

The Henry J. Kaiser Family Foundation Headquarters: 2400 Sand Hill Road, Menlo Park, CA 94025 | Phone 650-854-9400
Washington Offices and Barbara Jordan Conference Center: 1330 G Street, NW, Washington, DC 20005 | Phone 202-347-5270 | Email Alerts: | |

Filling the need for trusted information on national health issues, the Kaiser Family Foundation is a nonprofit organization based in Menlo Park, California.