Modifying Medicare's Benefit Design: What's the Impact on Beneficiaries and Spending?
The model’s primary data source is the 2009-2012 Medical Expenditure Panel Survey (MEPS), an annual survey of households and medical providers conducted by the Agency for Healthcare Research and Quality. This dataset includes individual-level information on demographics, income, health status and conditions, use of medical care, health expenses, and insurance coverage for nonelderly and elderly people. We supplemented the MEPS-based analysis with 2009-2011 data from the Medicare Current Beneficiary Survey (MCBS), a nationally representative survey of Medicare beneficiaries sponsored by the Centers for Medicare & Medicaid Services, to impute additional information where needed (for example, for SNF use and spending) and to validate some of our results (for example, the spending distribution by payer).
Although MEPS includes Medicare beneficiaries who are enrolled in Medicare Advantage, we excluded this group when evaluating the individual-level spending effects of the benefit design options because the options modify traditional Medicare. The model does incorporate indirect effects on aggregate Medicare Advantage spending and enrollment, based on the assumptions that changes in traditional Medicare reimbursement would be reflected in Medicare Advantage payments, and that aggregate Medicare Advantage payments will change to the extent that some beneficiaries switch between traditional Medicare and Medicare Advantage.
We aligned the poverty distribution to conform to the income and poverty distribution in the 2011 DYNASIM microsimulation model developed by researchers at The Urban Institute, with additional refinements using a CBO distribution of Part B enrollees by poverty status and Medicaid enrollment from 2004. We estimated the number of enrollees in the Medicare Savings Programs (with and without Medicaid) and the Part D Low Income Subsidy (LIS) program, using the Chronic Conditions Data Warehouse (2011 data) and CBO total enrollment counts, to determine the number of beneficiaries receiving cost-sharing assistance versus assistance with their Part B and/or Part D premiums only, projected to 2018, in order to model the effects of options to extend additional financial protections to low-income beneficiaries in traditional Medicare.
We used MEPS to assign beneficiaries to one of five supplemental coverage groups: 1) traditional Medicare only; 2) Medicaid; 3) employer-sponsored insurance (ESI), including TRICARE; 4) Medigap; and 5) other insurers, including Veterans’ Administration (VA), Indian Health Service, Worker’s Compensation, other federal, state, and local sources, and unclassified/unknown sources. Beneficiaries who listed more than one supplemental insurer were assigned to the coverage source they had for the most months during the calendar year. Enrollment estimates under the current-law baseline and the four options are provided in Table A1.
Calculating changes in cost sharing
The analysis looks exclusively at spending associated with Medicare-covered services. We excluded spending on non-Medicare covered services—such as dental care and other specific procedures, providers, or visit types that Medicare does not cover—from the MEPS data prior to conducting the analysis. As a survey of the non-institutionalized population, MEPS also does not include data for long-term services and supports. We began by estimating, at the individual level, what share of Medicare-covered services would be covered by the beneficiary and what share would be covered by each payer under current law in 2018. We controlled the MEPS Medicare channel of payment (i.e., Medicare reimbursement) by service to data from the August 2015 update1 of the CBO March 2015 Medicare baseline2 for Medicare benefits spending, with refinements at the
|Table A1: Enrollment in Traditional Medicare, Medicare Advantage, and Other Types of Coverage under Current Law and Four Medicare Benefit Design Options if Fully Implemented in 2018|
|Enrollment (in millions)||Current law||Option 1
Basic benefit redesign option
Lower deductible/cost-sharing limit option
Low-income subsidies option
|Employer (including TRICARE)||15.1||15.1||15.1||14.4||15.1|
|None (traditional Medicare only)||5.7||6.5||6.5||5.3||6.5|
|New low-income subsidies2||N/A||N/A||N/A||2.3||N/A3|
|NOTE: N/A is not applicable. 1The number of traditional Medicare beneficiaries under Options 1-4 includes 0.5 million beneficiaries who switch from Medicare Advantage to traditional Medicare, but who are not included in the analysis of individual-level cost-sharing effects due to differences in the cost-sharing structure of each coverage type. 2Applies to beneficiaries enrolled in SLMB, QI, and Part D LIS who are not already receiving assistance with Medicare cost-sharing requirements from Medicaid or Medicare Savings Programs. 3Beneficiaries with incomes below 150 percent of poverty would have a lower deductible and cost-sharing limit under Option 4 than those with higher incomes.
SOURCE: Kaiser Family Foundation/Actuarial Research Corporation, June 2016.
service and aged/disabled level (where available) based on the 2015 Medicare Trustees report3 and the 2013 Medicare & Medicaid Statistical Supplement.4 Total covered charges for each service in the record are then calculated based on the controlled Medicare reimbursement amount and the cost sharing for the service corresponding to the utilization data in the record. In this way, spending for each record is limited to Medicare-covered services only. The cost sharing (total covered charges minus Medicare spending) is distributed across the beneficiary and third-party payers according to their respective shares of total cost sharing for that service. To determine an individual’s out-of-pocket spending liability, we applied Medicare cost-sharing rules under current law to each beneficiary’s spending to divide the amount between Medicare and the beneficiary. Next, cost-sharing obligations were distributed among beneficiaries and supplemental insurers based on the share covered by each in the baseline.
We then applied alternative cost-sharing requirements under the Medicare benefit redesign options to determine shifts in the amount and distribution of spending for Medicare, beneficiaries, and other payers. (See the last section in this appendix for an example of how we model changes in the distribution of cost sharing under current law versus benefit redesign.) We first calculated the impact of the new cost-sharing rules on each individual’s exposure to cost sharing holding utilization constant. We also subject any increased cost sharing to the supplemental insurance the individual has, assuming insurers pay the same percentage of cost sharing under the benefit redesign options that they pay under current law. We then adjusted these calculations to account for changes in utilization and spending that would be expected to occur in response to cost-sharing changes, based on research showing that individuals tend to reduce their use of services, and thereby their spending, when their exposure to cost sharing increases (and vice-versa). The magnitude of these adjustments are described below. After making these adjustments and determining the new spending levels for each beneficiary, we recalculated the share of spending covered by each payer.
Because Medicaid is jointly-financed by the federal and state governments, we divided spending between the two payers based on current average spending patterns, with the federal government covering 57 percent of Medicaid spending on average and states covering the remaining 43 percent.5
Calculating changes in premiums
We estimated total spending by Medigap plans on Medicare-covered services and derived an expected average premium using current administrative expense loads and profit rates. We adopted a similar approach for ESI premiums, using a different administrative expense load and assuming employers would pick up half of the additional expenses, with the other half converted into a base premium. Using data from the Person Round Plan (PRPL) file in MEPS, we looked at deciles of Medigap and ESI premium amounts and used this information to make adjustments to the base premiums. We then randomly distributed these varying levels of premiums to enrollees’ records. It is important to note that for the purposes of this analysis, our emphasis is on the magnitude of changes in out-of-pocket spending (including premiums), rather than on absolute levels.
Medicare Part B premiums. Part B premiums are adjusted annually to cover 25 percent of the total predicted Part B costs for that year. However, the actual amount of premiums paid by enrollees is complicated by issues such as the income-related premium. We thus determined the Part B premium under modified benefit design options by adjusting the total premiums paid—taken from the CBO March 2015 Medicare Baseline, adjusted for the August 2015 update—by the percent change in Part B spending calculated by the model. The Part B premium was then adjusted to reflect beneficiaries’ incomes. We used a similar adjustment to CBO’s estimate for the federal share of premiums paid by Medicaid on behalf of beneficiaries dually eligible for Medicare and Medicaid.6
Supplemental insurance premiums. For supplemental insurance, we calculated premiums based on the insurers’ total spending on Medicare-covered services and related administrative expenses.
- Medigap: Medigap plans were assumed to set premiums in order to cover Medigap costs (including administrative expenses) as well as a constant rate of profit. Thus, the Medigap premium was simply total Medigap expenses divided by the number of enrollees.
- Employer-sponsored coverage: Employers were assumed to cover half of their employees’ expenses. The other half was converted into a uniform premium.
- TRICARE, Medicaid, and other insurers: We assumed no premiums for these groups. TRICARE does not charge its enrollees a premium, and premiums are prohibited for most Medicaid enrollees. We also assumed that other supplemental insurers do not charge a premium because this category is primarily VA, Worker’s Compensation, and other federal and state programs.
- Administrative expenses. For every dollar of spending, we assumed that payers would spend the following amounts on administrative costs: $0.017 and $0.015 for Medicare Part A and Part B, respectively; $0.052 for Medicaid; $0.25 for Medigap; $0.15 for ESI; and $0.10 for TRICARE and other insurers. These figures are based on the 2015 Trustees Report, CBO’s March 2015 Medicare Baseline, and data from the National Association of Insurance Commissioners (NAIC). Because administrative expenses vary by payer, benefit redesign options that change the distribution of spending also change total administrative costs. For instance, an option that shifts spending from Medicare to Medigap plans would also increase administrative expenses because the latter spends more on administration than the former.
Modeling changes in the distribution of cost sharing under current law versus a modified Medicare benefit design
Suppose we have a record for a Medicare beneficiary enrolled in an employer-sponsored retiree health plan who has one inpatient stay under 60 days, two primary care physician (PCP) visits, and three specialist visits. Under current law, total costs for these services are distributed among payers as follows (Table A2):
|Table A2: Example of Distribution of Total Costs Under Current Law in 2018|
|Total spending||Medicare spending||Beneficiary
|SOURCE: Kaiser Family Foundation/Actuarial Research Corporation, June 2016.|
Under current law, for inpatient services, this beneficiary is responsible for the Part A deductible ($1,376 in 2018, according to the 2015 Medicare Trustees Report), which the MEPS record shows is split evenly between beneficiary out-of-pocket and private spending ($688 each). For physician services, the beneficiary is responsible for the Part B deductible ($171 in 2018, according to the 2015 Medicare Trustees Report), along with 20 percent of costs above the deductible: (($3,500 – $171) x 0.2) + $171 = $836.80. The MEPS record shows that the beneficiary pays 10 percent ($83.68) and the employer plan pays the rest ($753.12).
To model the Medicare benefit redesign with a single A/B deductible of $650, a $6,700 cost-sharing limit, and varying copayments by service, we first apply the $650 deductible amount to inpatient services only, since physician visits are exempt from the deductible.
The cost-sharing amount for inpatient services is $750 per inpatient stay. For physician services, cost sharing is $25 for each of the two PCP visits plus and $50 for each of the three specialist visits for a total of
(2 x $25) + (3 x $50) = $200. Adding the deductible to the inpatient copayment produces a total cost sharing for inpatient of $650 + $750 = $1,400. Note that if other services had been used, the combined deductible would have been split pro-rata across the applicable services. This method enables us to retain the underlying cost-sharing distributions for each service as well as apply induction effects by service.
For inpatient services, the total cost sharing is split evenly, as under current law, between beneficiary out-of-pocket and private spending ($700 each), and for physician services, 10 percent ($20) of the cost sharing is paid out of pocket (as per the MEPS record under current-law cost-sharing rules). Thus, under the new benefit design, the beneficiary’s total costs would be distributed among payers as follows (Table A3):
|Table A3: Example of Distribution of Total Costs Under Medicare Benefit Redesign in 2018|
|Total spending||Medicare spending||Beneficiary
|SOURCE: Kaiser Family Foundation/Actuarial Research Corporation, June 2016.|