CONGRESS OF THE UNITED STATES CONGRESSIONAL BUDGET OFFICE CBO A Premium Support System for Medicare: Updated Analysis of Illustrative Options Net Federal Spending for -15% Second-Lowest-Bid Option Medicare Parts A and B for -8% Average-Bid Option Affected Beneficiaries Premiums Paid by 35% Affected Beneficiaries -7% Total Payments by 18% Affected Beneficiaries -5% Combined Net Federal Spending for -8% and Total Payments by -7% Affected Beneficiaries Estimated Difference From Outcomes Under Current Law, Without Grandfathering, in 2024 OCTOBER 2017 Notes All years referred to in this report are calendar years. The estimates were generated using the Congressional Budget Office’s March 2016 baseline projections of Medicare spending. The amounts in the text and tables are in nominal (current year) dollars. Numbers in the text, tables, and figures may not equal totals because of rounding. Supplemental information accompanies this report on CBO’s website (www.cbo.gov/publication/53077). www.cbo.gov/publication/53077 Contents Summary 1 What Are CBO’s New Estimates? 1 How Much Did CBO’s Estimates Change and Why? 1 What Is the Current Role of Private Plans in Medicare? 2 What Policy Options Did CBO Analyze? 2 The Federal Contribution 3 Grandfathering3 Other Features 3 Key Design Decisions for Future Proposals 5 What Were CBO’s Analytical Methods? 5 What Are CBO’s New Estimates? 5 Budgetary Effects Without Grandfathering 6 Budgetary Effects With Grandfathering 8 Other Effects 8 How Much Did CBO’s Estimates of Effects Without Grandfathering Change and Why? 10 Changes in the Estimates 10 Reasons for the Changes in the Estimates 11 BOX 1. THE ROLE OF THE MEDICARE FEE-FOR-SERVICE PROGRAM AND ITS PROVIDER PAYMENT RATES 14 How Much Did CBO’s Estimates of Effects With Grandfathering Change and Why? 15 About This Document 17 Table 1. Estimated Change in Net Federal Spending for Medicare Under Illustrative Premium Support Options, Relative to Spending Under Current Law, 2022 to 2026 6 Figures 1. Estimated Difference From Current Law in Net Federal Spending for and Total Payments by Affected Medicare Beneficiaries Under Illustrative Premium Support Options, Without Grandfathering, 2024 7 2. Ratio of Actual and Projected Medicare Advantage Bids to Medicare FFS Spending per Capita Under Current Law, in Two CBO Studies 13 A Premium Support System for Medicare: Updated Analysis of Illustrative Options Summary effect instead of requiring all beneficiaries to enter the Over the past two decades, policymakers and analysts premium support system once it began. have advanced a variety of proposals for converting Medicare to a premium support system as a way to What Are CBO’s New Estimates? reduce federal spending. Under such a system, beneficia- CBO’s new estimates indicate the following: ries would choose health insurance from a list of compet- ing plans, and the federal government would share the ■■ Without grandfathering, the second-lowest-bid cost of their premiums. The proposals have differed in option would reduce net federal spending for many respects, notably in the way that the federal con- Medicare by $419 billion between 2022 and 2026; tribution would be set and how that contribution might the average-bid option would reduce such spending change over time. by $184 billion. The Congressional Budget Office has in the past ana- ■■ With grandfathering, the second-lowest-bid option lyzed the budgetary effects of some illustrative options would reduce net federal spending for Medicare by for a premium support system.1 This report updates $50 billion between 2022 and 2026; the average-bid the agency’s work on the topic, presents new estimates option would reduce such spending by $21 billion. of the budgetary effects of those options, and examines the reasons for the changes in the estimates, including Those savings would arise because private insurers’ bids changes in law that have affected the Medicare program. would generally be lower than FFS costs per capita and CBO constructed its estimates for this report under would substantially influence the federal contribution. the assumption that the system would be implemented Savings would be much smaller if the options included in 2022. Depending on their details, future cost esti- a grandfathering provision because only a small portion mates for legislative proposals that resemble the options of the Medicare population would be covered by the analyzed in this report could differ substantially from the new system initially, and that portion would increase estimates presented here. gradually. In the options CBO analyzed, the federal government’s On average, CBO estimates, beneficiaries’ total payments contribution would be determined from insurers’ bids, for Medicare premiums and cost sharing (enrollees’ and Medicare’s traditional fee-for-service (FFS) program out-of-pocket spending on copayments, coinsurance, would be included as a competing plan. CBO examined and deductibles for Medicare-covered benefits) would two approaches for determining the federal contribu- be higher under the second-lowest-bid option, but lower tion: One would set the contribution on the basis of under the average-bid option, than under current law. the second-lowest bid in each region; the other would Under either option, the total payments made by partic- use the region’s average bid. CBO also examined the ular beneficiaries could differ markedly from the national effects of grandfathering, which would keep beneficiaries average. For example, in many regions, total payments by in the current Medicare program if they were eligible beneficiaries who chose to enroll in Medicare’s FFS pro- for Medicare before the premium support system took gram would be substantially higher than under current law because of the increases in beneficiaries’ premiums. 1. See Congressional Budget Office, Options for Reducing the Deficit: How Much Did CBO’s Estimates Change and Why? 2014 to 2023 (November 2013), pp. 204–210, www.cbo.gov/ content/options-reducing-deficit-2014-2023, and A Premium CBO’s current estimates of the federal savings from the Support System for Medicare: Analysis of Illustrative Options premium support options without grandfathering are (September 2013), www.cbo.gov/publication/44581. much higher than its earlier estimates. In a November 2 A Premium Support System for Medicare: Updated Analysis of Illustrative Options October 2017 2013 report, CBO estimated that if a premium sup- indicating the per capita payment they will accept for port system was implemented without grandfathering, providing benefits to enrollees under Medicare Parts A the second-lowest-bid option would reduce net federal and B. The resulting federal payments depend in part on spending for Medicare by $275 billion between 2018 the insurers’ bids and on how those bids compare with and 2023 and the average-bid option would reduce net county-level benchmarks, which range from 95 percent federal spending over that period by $69 billion.2 to 115 percent of local spending per capita in Medicare’s FFS program. Federal payments to insurers are adjusted CBO’s savings estimates increased primarily because the to account for the health status of their enrollees, and agency’s current projections of the bids that Medicare plans receive bonus payments if they earn high ratings Advantage plans would submit under current law are for quality of care. (Private insurers also participate in a lower relative to FFS spending per capita than the pro- separate bidding process that determines payments under jections used in its earlier analysis. Medicare Advantage Medicare Part D, the prescription drug program.) plans submit bids to Medicare for the amount that it would cost to provide enrollees with Medicare bene- What Policy Options Did CBO Analyze? fits covered under the Hospital Insurance (Part A) and In the current analysis, CBO examined two sets of Medical Insurance (Part B) programs. Medicare pays illustrative options for converting Medicare to a pre- plans based on those bids, and then Medicare Advantage mium support system. For each, the federal government’s plans assume responsibility for paying providers for contribution would be determined from insurers’ bids, beneficiaries’ care. (In contrast, Medicare’s FFS program including the “bid” of the Medicare FFS program, pays providers directly for services covered under Parts A which would be a competing plan. The nation would and B.) CBO used its projections of the bids Medicare be divided into regions within which competing private Advantage plans submit under the current program insurers would submit bids indicating the amount they to estimate the bids of private insurers under the pre- would accept to provide Medicare benefits to a benefi- mium support options. The lower current projections of ciary in average health.3 Similarly, Medicare’s FFS bid Medicare Advantage bids suggest that those insurers’ bids in each region would be based on the projected cost of would be lower than CBO had previously anticipated. providing benefits in Medicare FFS to an enrollee in Other factors also affected CBO’s budgetary estimates, average health. but with smaller net effects. Insurers would submit bids for a benefit package that CBO lowered its projections of Medicare Advantage covered the same services as Parts A and B of Medicare bids relative to FFS spending per capita for two reasons. (with a few exceptions, noted below) at the same actuar- First, Medicare Advantage bids have declined relative ial value as Parts A and B combined. (That is, each policy to FFS spending in recent years. Second, legislation would cover the same benefits and percentage of total affecting updates to Medicare’s FFS physician payment expenses for a given population that would be covered rates caused CBO to revise its projections of how much under current law by Medicare’s FFS program.) As under Medicare Advantage bids will change relative to FFS current law, Medicare Part D would be administered spending. separately. What Is the Current Role of The options CBO examined differ from each other Private Plans in Medicare? along two dimensions: the approach used to deter- In 2016, about 30 percent of Medicare’s 57 million mine the federal contribution, and whether the option beneficiaries were enrolled in Medicare Advantage plans. included a grandfathering provision so that beneficiaries Almost all other beneficiaries were enrolled in Medicare’s who became eligible for Medicare before the premium FFS program. Insurers who wish to participate in support system took effect would remain in the current Medicare Advantage submit bids to the government 2. For that estimate, CBO assumed implementation in 2018, four years earlier than the current estimate. See Congressional 3. Throughout this report, the term bid refers to the standardized Budget Office, Options for Reducing the Deficit: 2014 to 2023 bid for a beneficiary in average health. As under current law, (November 2013), pp. 204–210, www.cbo.gov/content/ federal payments to plans would be adjusted to account for options-reducing-deficit-2014-2023. differences in their enrollees’ health. October 2017 A Premium Support System for Medicare: Updated Analysis of Illustrative Options 3 Medicare system rather than enter the new system. Other participate in the premium support system initially, but program features would be the same. that portion would increase gradually over the long term. The Federal Contribution Other Features CBO analyzed two approaches to determining the The other features of a premium support system were benchmarks for setting the federal contribution: common to all options. Some illustrate the potential for savings from a premium support framework; others were ■■ A second-lowest-bid approach would set the chosen for feasibility of implementation or to simplify regional benchmark at the lower of a pair of bids: the modeling approach. Many other variations are pos- either Medicare’s FFS bid or the second-lowest bid sible, and none of the options presented in this report submitted by a private insurer. should be considered a recommendation by CBO. ■■ An average-bid approach would set the regional Under each option, beneficiaries would choose a plan benchmark at the weighted average of all bids, when they first entered the premium support system. including the FFS bid, with weights equal to the Beneficiaries who did not select a plan at that time proportion of beneficiaries enrolled in that plan in would be assigned (with equal probability) to a plan that the preceding year. had submitted a bid at or below the regional bench- mark, including the FFS program if it met that criterion. For each enrollee, the federal government would pay Beneficiaries would remain in the plan they chose (or insurers an amount equal to the benchmark for the were assigned to) in subsequent years, unless they chose a region minus the standard premium paid by enroll- different plan during an annual enrollment period. ees (discussed below). Insurers would receive larger or smaller payments for beneficiaries whose health was To clarify the choices for beneficiaries (and thereby worse or better than average. Neither the amount nor the heighten competition based on differences in premiums), growth rate of the federal payment would be capped. private insurers would be allowed to submit bids for the basic Medicare package for just one or two plans in each Beneficiaries who enrolled in a plan with a bid that region. If they chose to submit bids for two plans, each equaled the benchmark would pay a standard premium could have different features—offering a larger or smaller directly to the insurer. That premium would be the same provider network, for example—but both would need to everywhere and would be set to cover approximately have the same actuarial value. Insurers also could offer a one-fourth of the total cost, excluding cost sharing, for package of enhanced benefits (with a single, fixed, higher services covered in Part B (physicians’ services, hospital actuarial value that would be the same for all insurers) to outpatient care, durable medical equipment, and other accompany each basic package offered. Enrollees would services, including some home health care)—a formula pay the full additional cost of the enhanced packages that is similar to that under current law for Part B premi- through higher premiums. ums. Beneficiaries who chose a plan with a bid above the benchmark would pay the insurer the standard premium CBO assumed that there would be no changes to the plus the difference between the bid and the benchmark. current FFS program, either in the mechanisms for Those who chose a plan with a bid below the benchmark setting the rates paid to providers or in the tools avail- would pay the standard premium minus the difference able to contain costs. As under current law, beneficia- between the benchmark and the bid. Income-related ries who remained in the FFS program could purchase Part B premiums for higher-income beneficiaries would supplemental coverage (known as medigap coverage) continue as under current law. from private insurers. Such policies cover some or all of Medicare’s cost sharing and may also cover certain Grandfathering services that are not covered by Medicare. For each approach to determining the benchmark, CBO analyzed options with and without grandfathering. To simplify the analysis, CBO assumed that the pre- (Grandfathering would keep current beneficiaries from mium support system would not affect certain types of having to adjust to a new system.) Under grandfathering, federal spending for Medicare. Specifically, the agency only a small portion of the Medicare population would assumed that dual-eligible beneficiaries—people who are 4 A Premium Support System for Medicare: Updated Analysis of Illustrative Options October 2017 simultaneously enrolled in Medicare and Medicaid— First, CBO assumed that beneficiaries who had Part A– would be excluded from the premium support system only coverage would be excluded from the premium and that federal spending for their health care would support system. That analytical choice resulted in mod- continue as it would under current law. CBO made estly smaller budgetary savings, relative to CBO’s prior that assumption because of the additional complexity estimate, because Medicare is the secondary payer for of structuring a premium support system to include most such beneficiaries and thus typically spends much dual-eligible beneficiaries, although a system could be less to cover them.6 devised to include them. Second, CBO assumed that the federal government In a change from past analyses, CBO assumed that ben- would apply a greater reduction in the risk scores of eficiaries with coverage only for Medicare Part A would private-plan enrollees under the premium support be excluded from the premium support system and options than it would under the current Medicare that federal spending for their benefits would continue Advantage program. Risk scores are computed for all as it would under current law. CBO chose that feature Medicare beneficiaries on the basis of their diagnoses because most such beneficiaries have primary coverage and other characteristics, and the government uses those through employment-based insurance and have second- scores to adjust payments to plans. (CBO assumed that ary coverage through Medicare.4 a comparable risk-adjustment system would be used for the premium support options.) Research pub- CBO also assumed that Medicare’s spending for Part D lished in the past few years has shown that, on average, would continue as projected under current law, as would Medicare Advantage enrollees have higher risk scores spending for items and services that are not included in than FFS beneficiaries in similar health and that the the calculation of the benchmarks or bids for current- difference has increased recently.7 The difference between law Medicare Advantage plans—such as Medicare’s risk scores for the two groups of enrollees appears to additional payments to hospitals for medical education, arise more from the intensive diagnostic coding used by hospice benefits, and certain benefits for patients with Medicare Advantage plans than from actual differences end-stage renal disease. in health among the two groups.8 In the current analysis, CBO assumed that the federal government would take The categories of spending that CBO assumed would steps to ensure that the risk scores of private-plan enroll- be unaffected by the premium support system—which ees would be no more than 5 percent higher, on average, include spending for dual-eligible beneficiaries and bene- than the risk scores of Medicare FFS beneficiaries with ficiaries enrolled in Part A only, all spending on Medicare Part D, and the other categories of spending discussed 6. In certain situations—such as when a Medicare-eligible above—made up about 40 percent of net federal spend- beneficiary has health insurance coverage through a current ing for Medicare in 2016. (Net spending consists of total employer or a spouse’s employer—Medicare acts as the secondary Medicare spending minus beneficiaries’ premiums and payer. That is, Medicare only pays for covered benefits after the other offsetting receipts.) primary payer has met its responsibility for the beneficiary’s costs of care. CBO made many other detailed assumptions concerning 7. For example, see Medicare Payment Advisory Commission, the options that have been described previously.5 With “MA Risk Adjustment and Coding Intensity Adjustment,” in Report to the Congress: Medicare Payment Policy (March 2016), the following three exceptions, the specifications used in pp. 344–346, www.medpac.gov/-documents-/reports; and this analysis were the same as those that applied in 2013. Richard Kronick and W. Pete Welch, “Measuring Coding Intensity in the Medicare Advantage Program,” Medicare & Medicaid Research Review, vol. 4, no. 2 (2014), pp. E1–E19, https://go.usa.gov/xN5DU. 8. Because they receive larger payments for covering enrollees with 4. Under current law, beneficiaries must be enrolled in both Part A higher risk scores, Medicare Advantage plans have an incentive and Part B of Medicare to be eligible to enroll in a Medicare to code all diagnoses that are included in the risk-adjustment Advantage plan. mechanism. Many providers (particularly physicians) have no 5. See Congressional Budget Office, A Premium Support System such incentive to code every diagnosis for their Medicare FFS for Medicare: Analysis of Illustrative Options (September 2013), patients; they are paid on the basis of the services furnished, not pp. 7–15, www.cbo.gov/publication/44581. the diagnoses reported. October 2017 A Premium Support System for Medicare: Updated Analysis of Illustrative Options 5 similar health status.9 That difference is smaller than the ■■ What rules would be established for beneficiaries who published estimates of the difference under current law. receive retiree coverage from a former employer or union? Third, for this analysis, CBO assumed that legislation to establish a premium support system would be enacted ■■ How would the federal government change risk late in 2017. To allow time for the federal government to adjustment to account for differences in the health develop the necessary administrative structures and for status of enrollees in various plans (including beneficiaries and insurers to prepare for the new system, Medicare FFS)? CBO assumed that the system would not be imple- mented until 2022. What Were CBO’s Analytical Methods? CBO’s estimates of the effects of the premium support Key Design Decisions for Future Proposals options on federal spending and beneficiaries’ total pay- Options considered by the Congress, and the result- ments were based on detailed modeling of the behavior ing costs or savings, could differ significantly from the of buyers and sellers of health insurance policies. That options analyzed in this report. Policymakers who wished modeling was similar for both sets of options.11 to develop such proposals would need to make many complex decisions about the design of a premium sup- First, the agency projected the amounts of the bids that port system, with important implications for Medicare would be submitted by plans in the Medicare Advantage spending. In its earlier report, CBO discussed several program under current law. Then, the agency adjusted such decisions that would be specific to a system with those projected bids, given the downward and upward grandfathering.10 pressures that would be a likely result of a premium sup- port system. CBO used that information (and data about Some more broadly applicable design questions include past enrollment for the average-bid option) to estimate the following: regional benchmarks and premiums for each plan. ■■ Would dual-eligible beneficiaries be included in the CBO then simulated the enrollment choices of a large premium support system, and if so, how would the sample of beneficiaries in different plans on the basis of system accommodate them? premiums and previous patterns of enrollment, calcu- lated federal spending as the sum of the risk-adjusted ■■ Would enrollment in Part B remain voluntary, and if federal contribution for each beneficiary, and compared so, how would beneficiaries who are enrolled only in that estimate of total federal spending with its baseline Part A be treated by the new system? projection of federal spending under current law. To project beneficiaries’ total payments, CBO used claims data to estimate cost-sharing payments by each benefi- ciary for the services covered by Medicare and combined those estimates with estimates of the plans’ premiums. 9. Recent trends informed CBO’s expectation that, under current law, the unadjusted difference between the risk scores of Medicare What Are CBO’s New Estimates? Advantage enrollees and FFS beneficiaries would be greater CBO estimates that the options considered in this than it anticipated in 2013 and substantially above 5 percent. analysis would reduce net federal spending for Medicare For the premium support options, CBO assumed that coding differences would be limited to 5 percent. That limit is illustrative but that the savings would be substantially greater for and arbitrary. Pressure to have a low limit would stem from the second-lowest-bid option than for the average-bid concerns that a greater divergence between risk scores under option. Beneficiaries’ total payments, on average, would premium support would allow private plans to reduce their be higher under the second-lowest-bid option but lower bids. Reductions in those bids would tend to lower the federal under the average-bid option than under current law. For contribution but would not affect the FFS bid. Thus, premiums would increase for beneficiaries who chose to remain in the FFS program. 11. For additional information, see Congressional Budget Office, 10. See Congressional Budget Office, A Premium Support System A Premium Support System for Medicare: Analysis of Illustrative for Medicare: Analysis of Illustrative Options (September 2013), Options (September 2013), Appendix A, www.cbo.gov/ pp. 32–33, www.cbo.gov/publication/44581. publication/44581. 6 A Premium Support System for Medicare: Updated Analysis of Illustrative Options October 2017 Table 1 . Estimated Change in Net Federal Spending for Medicare Under Illustrative Premium Support Options, Relative to Spending Under Current Law, 2022 to 2026 Billions of Dollars Total, 2022 2023 2024 2025 2026 2022–2026 Without Grandfathering Second-Lowest-Bid Option -61 -80 -84 -91 -102 -419 Average-Bid Option -14 -33 -41 -46 -50 -184 With Grandfathering Second-Lowest-Bid Option -2 -5 -9 -14 -20 -50 Average-Bid Option * -2 -4 -6 -8 -21 Source: Congressional Budget Office. Options without a grandfathering provision would apply to everyone who would be enrolled in Medicare under current law other than dual-eligible beneficiaries (those enrolled simultaneously in Medicare and Medicaid) and those with coverage under Medicare Part A only. With a grandfathering provision, all beneficiaries who became eligible for Medicare before implementation of a premium support system would remain in the current Medicare program, and all beneficiaries who became eligible after that would be included in the premium support system. Dual-eligible beneficiaries and beneficiaries with Part A–only coverage also would be excluded. Net federal spending for Medicare consists of total Medicare spending minus beneficiaries’ premiums and other offsetting receipts. CBO used data from 2016 for its current analysis. * = between zero and $500 million. this analysis, CBO considered total payments to consist of spending for Medicare by 9 percent, and the average-bid premiums plus cost sharing for Part A and Part B benefits. option would reduce that spending by 4 percent. Under either option, a particular beneficiary’s total pay- Another way to measure the options’ effects is to exam- ments could differ markedly from the national average. ine their impact on net federal spending just for affected For example, in many regions, premiums would be much beneficiaries for benefits that would be included in the higher for Medicare’s FFS program, which would result premium support system—rather than for the Medicare in substantially higher total payments by FFS beneficia- program as a whole. That group would include every- ries than would be the case under current law. Moreover, one (other than dual-eligible beneficiaries and those under either option, the savings over the next decade with Part A–only coverage) who would have enrolled in would be substantially lower if a grandfathering provi- Medicare under current law. (The measure of spending sion was included. included in that calculation consists of federal spending for those beneficiaries for Part A and Part B benefits, Budgetary Effects Without Grandfathering excluding spending for items and services not covered by If the premium support system covered currently eligible Medicare Advantage bids, minus beneficiaries’ premi- and future beneficiaries (but excluded dual-eligible ums and other offsetting receipts.) Without a grand­ beneficiaries and those with coverage under Part A only), fathering provision, the second-lowest-bid option would the second-lowest-bid option would reduce net federal reduce net federal spending for affected beneficiaries in spending for Medicare by $419 billion between 2022 2024 by 15 percent, and the average-bid option would and 2026, CBO estimates (see Table 1). The average­ reduce such spending by 8 percent, CBO estimates (see bid option would reduce net federal spending over Figure 1). Those percentages are larger than the percent- that period by $184 billion. Compared with projected age reductions in total Medicare spending because the spending under current law, by 2024 (an illustrative savings are measured relative to the portion of Medicare year shortly after implementation of the new system) spending that would be covered under the premium the second-lowest-bid option would reduce net federal October 2017 A Premium Support System for Medicare: Updated Analysis of Illustrative Options 7 Figure 1 . Estimated Difference From Current Law in Net Federal Spending for and Total Payments by Affected Medicare Beneficiaries Under Illustrative Premium Support Options, Without Grandfathering, 2024 Percent -20 0 40 Net Federal Spending for -15 Second-Lowest-Bid Option Parts A and B for -8 Average-Bid Option Affected Beneficiaries Premiums Paid by 35 Affected Beneficiaries -7 Total Payments by 18 Affected Beneficiaries -5 Combined Net Federal Spending for -8 and Total Payments by Affected Beneficiaries -7 Source: Congressional Budget Office. Net federal spending consists of Medicare spending for affected beneficiaries on services covered under the premium support system, minus beneficiaries’ premiums and other offsetting receipts. Medicare spending includes all spending for services under Part A and Part B except spending that was excluded because it is not covered by the bids that Medicare Advantage plans submit under current law: spending for medical education, hospice benefits, and certain benefits for patients with end-stage renal disease. Spending for prescription drug coverage under Part D also is excluded. Affected beneficiaries consist of everyone who would be enrolled in Medicare under current law other than dual-eligible beneficiaries (people who are enrolled simultaneously in Medicare and Medicaid) and those with coverage under Part A only. Total payments by beneficiaries include premiums and out-of-pocket spending for copayments, coinsurance, and deductibles for services and supplies covered by Part A and Part B. Premiums are for the basic package of Medicare benefits covered in the premium support system. They exclude any additional amounts paid for enhanced benefits or supplemental (medigap) coverage. support system, rather than relative to total Medicare Beginning in the second year, CBO anticipates, low- spending. er-bidding plans would capture a larger share of enroll- ment under premium support than under current law, Under either option, the savings to Medicare between which would reduce benchmarks under the average-bid 2022 and 2026 would be similar in percentage terms option. to the savings estimated for 2024, with one exception. Under the average-bid option, CBO estimates, federal In CBO’s estimation, the two premium support options savings in 2022 would be about half the savings in the would yield federal savings because the benchmarks in other years, mainly because CBO assumed that the most regions—and hence, the federal spending per weights applied to the bids in constructing benchmarks ben­ ficiary—would be lower than federal spending e in the first year of the new system would reflect the per beneficiary under current law. That would occur share of beneficiaries enrolled in 2021 in Medicare’s even if private plans’ bids were no lower than the bids FFS program and in private plans under current law. Medicare Advantage plans would submit under current 8 A Premium Support System for Medicare: Updated Analysis of Illustrative Options October 2017 law, which CBO projects would average about 10 percent labor and equipment), and the federal government has less than FFS spending per capita. Specifically: broad authority under current law to expand demon- stration projects that successfully reduce spending for ■■ Under the second-lowest-bid option, the benchmark Medicare. Those provisions are discussed in more detail would be either Medicare’s FFS bid or the second- in CBO’s earlier report.13 lowest bid submitted by a private insurer. In most cases, the latter would be lower. Budgetary Effects With Grandfathering Federal savings would be much smaller under a pre- ■■ In the average-bid option, the benchmark—the mium support system that excluded people who already weighted average of all bids—usually would be were eligible for Medicare. CBO estimates that if a reduced by the inclusion of lower-bidding plans in its system applied only to beneficiaries who qualified for calculation. Medicare in 2022 or later, spending on such a system for the 2022–2026 period would be only 10 percent of Additionally, under both options, CBO anticipates the spending on a system without grandfathering. With that private insurers would face greater price compe- grandfathering, CBO estimates, the second-lowest-bid tition than under the Medicare Advantage program, option would reduce net federal spending for Medicare which would lead them to reduce their bids to attract by $50 billion through 2026; the average-bid option more enrollees and thus increase federal savings even would reduce such spending by $21 billion. more. A third important source of federal savings in the average­­ -bid option is the projected shift of enrollees Thus, modifying the second-lowest-bid option to from FFS into private plans in many areas, and from include grandfathering would yield savings between higher-cost to lower-cost private plans, which would 2022 and 2026 that were 12 percent of the savings that reduce benchmarks.12 would be achieved without such a provision. Under the average-bid option, the estimated savings with grand- For roughly another decade after 2026, under either fathering would be 11 percent of the savings without it. option, CBO expects that annual federal savings would Those percentages are similar to the share of Medicare remain roughly stable in percentage terms, although the spending that would be covered by the premium support dollar amounts would increase. Over the longer term, system with grandfathering relative to the share without increased price competition from implementing either grandfathering. The savings differ slightly because some option would probably reduce the growth of Medicare factors affect the bids of private plans differently if a spending by decreasing demand for expensive new tech- grandfathering provision is included. nologies and treatments and by increasing demand for cost-reducing technologies, although the magnitude of In the longer term, grandfathering also would reduce the such changes is highly uncertain. incentives created by a premium support system to limit the development and use of new medical technologies. The potential for a premium support system to produce Thus, the constraints on the growth of Medicare spend- additional savings, however, would be limited because ing that would probably occur under a premium support certain provisions of current law are already designed to system would be substantially weaker for many years. restrain the growth of Medicare spending. For example, updates to Medicare’s payment rates for most providers Other Effects in the FFS program are scheduled to be smaller than the The premium support options would affect the pre- projected increases in the costs of their inputs (such as miums that Medicare beneficiaries paid for Part A and Part B benefits, their total payments for those benefits 12. Most beneficiaries live in counties where CBO expects enrollees (premiums plus cost sharing), the combined payments of would shift from FFS to private plans under the premium the federal government and beneficiaries, and enrollment support options. However, many beneficiaries live in areas where in private plans. CBO estimated those effects for 2024, FFS would be the least expensive option, and the opposite would occur. CBO anticipates that beneficiaries also would shift to lower-bidding plans in the second-lowest-bid option, but that 13. See Congressional Budget Office, A Premium Support System change would not affect benchmarks and thus would not be a for Medicare: Analysis of Illustrative Options (September 2013), source of additional federal savings. pp. 19–20, www.cbo.gov/publication/44581. October 2017 A Premium Support System for Medicare: Updated Analysis of Illustrative Options 9 focusing on beneficiaries affected by the two options Under all of the options, including those with grand- without grandfathering. Although the options also could fathering, beneficiaries in all regions would be offered affect beneficiaries’ access to care and the quality of that at least one plan with a premium that was at or below care, CBO does not have the tools to study such effects the standard premium. Although it is possible that some and does not anticipate having them in the near future. regions would have no participating private insurers, CBO estimates that, in 2024, only about 1 percent of Effects on Beneficiaries’ Premiums. CBO estimates that beneficiaries would live in such areas. In those cases, under the second-lowest-bid option, affected beneficia- Medicare’s FFS program would be the only plan avail- ries in 2024 would pay a total premium that was about able, and beneficiaries would enroll in that program and 35 percent higher, on average, than the current-law pay the standard premium. Part B premium projected for that year.14 (The total premium for a beneficiary enrolling in a given plan Although all beneficiaries could select a plan with a under a premium support system would be the stan- premium below the current-law Part B premium, most dard Medicare premium, plus or minus any difference who wished to remain in the FFS program would pay between the bid of that plan and the benchmark.) Under much higher premiums under either option than they the average-bid option, the total premium would be would under current law (because the benchmarks in about 7 percent lower, on average, than the current-law most regions would be lower than the FFS bid).16 CBO Part B premium. Under either option, those amounts estimates that the premium for enrolling in the FFS pro- would depend on the premiums charged by the available gram under the second-lowest-bid option in 2024 would plans (which would vary by region) and on beneficia- be about twice as much, on average, as the current-law ries’ choices of plans. CBO expects that those choices Part B premium projected for that year. Under the would depend partly on premiums but also would be average-bid option, the premium would be 57 percent affected by other plan features, such as the size and com- more, on average, than the projected current-law Part B position of the provider network, the reputation of the premium. insurer, and its customer service. The increase in the premium required to enroll in the The standard premium under either option would be FFS program would be larger in regions in which FFS lower than the current-law Part B premium, CBO spending per beneficiary was higher. For example, under estimates, because both options would reduce federal the second-lowest-bid option, CBO estimates, the FFS Medicare spending, and thus the standard premium premium in 2024 would be about three times higher (which would be equal to the same share of spending than the current-law Part B premium that year in coun- that the Part B premium equals under current law). ties in the top fourth of FFS spending per beneficiary. That reduction in the standard premium is the main It would be about 2.5 times more than the current-law reason that the average premium paid by beneficiaries Part B premium under the average-bid option in those under the average-bid option would be lower than the same counties. In counties where spending was in the projected current-law Part B premium. The additional nation’s bottom fourth of FFS spending per capita, amounts paid by beneficiaries who enrolled in plans the FFS premium would be 20 percent higher under with bids above the benchmark would roughly offset the the second-lowest-bid option and roughly equal to the reductions for beneficiaries who enrolled in plans with current-law Part B premium under the average-bid bids below the benchmark. Under the second-lowest-bid option. option, however, the regional benchmarks would generally be lower than they would be under the average-bid option, As a result of those increases in premiums for the FFS so CBO expects that many beneficiaries would enroll in program, CBO estimates, in 2024, about 20 percent plans with bids above the relevant benchmark, resulting in more beneficiaries would be enrolled in private plans much higher average premiums than under current law.15 16. Based on the findings of prior research on beneficiaries’ choice of health plans, CBO anticipates that some beneficiaries would 14. Under current law, most Medicare beneficiaries do not pay Part A enroll in the FFS program even though they would pay higher premiums. premiums than under current law because that program would 15. On the basis of past research, CBO anticipates that in choosing offer greater freedom of choice among providers and fewer plans, beneficiaries would consider additional characteristics restrictions on care than private plans. However, CBO estimates beyond premiums, including a plan’s quality, its reputation, or that fewer beneficiaries would enroll in the FFS program under the providers included in its network. the premium support options than under current law. 10 A Premium Support System for Medicare: Updated Analysis of Illustrative Options October 2017 under both the second-lowest-bid and the average-bid Under either option, the change in total payments for options than under current law. Changes in the percent- individual beneficiaries could differ markedly from the age of beneficiaries enrolled in private plans would vary national average. For example, people who chose to across the country, depending, in part, on FFS spending remain in Medicare’s FFS program would generally face per beneficiary and on changes in premiums for the much higher premiums and would not see any reduction FFS program. For example, enrollment in private plans in their cost sharing. would roughly double under both options in counties in the top fourth of FFS spending per beneficiary because Effects on Combined Spending by the Government the increase in premiums for the FFS program would and by Beneficiaries. The sum of net federal spending be greatest in those places. The share of beneficiaries for Medicare and beneficiaries’ total payments in 2024 enrolled in private plans would increase by smaller under the second-lowest-bid option would be about amounts in most other areas of the country, and that 8 percent lower than under current law, CBO estimates, share would decline in some areas where the FFS spend- and about 7 percent lower under the average-bid option ing per beneficiary is lower. than under current law. (Those effects are measured as a percentage of projected net federal spending and bene- Effects on Beneficiaries’ Total Payments. CBO estimates ficiaries’ total payments, measured for beneficiaries and that affected beneficiaries’ total payments for Part A and benefits affected by the premium support system.) Part B benefits in 2024 under the second-lowest-bid option without grandfathering would be about 18 per- The estimated reduction in total spending is slightly cent higher, on average, than under current law. In gen- larger under the second-lowest-bid option because the eral, the premiums paid by beneficiaries would increase federal contribution would be smaller under that option. under that option, but beneficiaries’ out-of-pocket costs The result would be increased competitive pressure, would decline slightly (because more beneficiaries would lower bids by private plans, and more enrollment in enroll in lower-bidding private plans, which would tend lower-bidding plans. The federal savings would be much to reduce the total costs of care while maintaining the larger under that option than under the average-bid required actuarial value).17 That reduction would offset option, but those larger savings would be partly offset by part, but not all, of the increase in premiums. larger payments by beneficiaries. Under the average-bid option without grandfathering, How Much Did CBO’s Estimates of Effects beneficiaries’ total payments for Part A and Part B bene- Without Grandfathering Change and Why? fits in 2024 would be about 5 percent lower, on average, CBO’s current estimates of the federal savings from the than under current law. That reduction would result both two options without grandfathering are much higher from lower average premiums, which are lower than in than its earlier estimates, primarily because the agency’s the second-lowest-bid option because the federal contri- current projections of the bids that Medicare Advantage butions are higher, and from lower out-of-pocket costs. plans would submit under current law are lower relative As in the previous option, the difference in beneficiaries’ to FFS spending per capita than were the projections out-of-pocket costs would be attributable primarily to used in its earlier analysis. larger enrollment in lower-bidding private plans. Changes in the Estimates In a November 2013 report about approaches to reduc- ing the federal deficit, CBO estimated that if a premium support system was implemented without a grand- 17. Under the options evaluated in this report, plans would be fathering provision, the second-lowest-bid option would required to maintain the same actuarial value as current-law FFS Medicare, or cover, on average, the same proportion of total reduce net federal spending for Medicare by $275 billion expenses covered by Medicare’s current-law benefit package. between 2018 and 2023, and the average-bid option CBO expects that beneficiaries enrolled in lower-bidding plans would reduce such spending over the same period by would use less health care and would therefore pay less in cost $69 billion.18 By comparison, CBO now estimates that sharing than would enrollees in higher-bidding plans. CBO’s analysis did not account for possible differences among plans 18. See Congressional Budget Office, Options for Reducing the Deficit: in the additional cost sharing enrollees might incur for services 2014 to 2023 (November 2013), pp. 204–210, www.cbo.gov/ received outside a plan’s network. content/options-reducing-deficit-2014-2023. October 2017 A Premium Support System for Medicare: Updated Analysis of Illustrative Options 11 the second-lowest-bid option would reduce net federal values for five key parameters and determining the effects spending by $419 billion between 2022 and 2026 and of varying those parameters in 2020 (an illustrative year the average-bid option would reduce spending over the shortly after that new system would be implemented).20 same period by $184 billion. The results from that exercise indicated that for the second-lowest-bid option, net federal spending for Those comparisons of multiyear totals are affected by affected beneficiaries in 2020 would probably be reduced two differences in the projection periods for the two by between 9 percent and 14 percent (CBO’s central analyses. First, CBO’s use of a later projection period estimate was 11 percent). For the average-bid option, for the current analysis increased estimated savings net federal spending would probably be reduced by because those savings are expressed in nominal dollars between 1 percent and 7 percent (the central estimate and Medicare spending has grown over the period. was 4 percent). Second, in the current analysis, the premium support policy would be in effect for fewer years of the projec- CBO’s current estimates of savings are slightly greater tion period, which would reduce cumulative savings.19 than the high end of those ranges: The second-lowest-bid Without those changes, however, the savings estimated option would reduce net federal spending in 2024 for would still be substantially higher in the current analysis affected beneficiaries by an estimated 15 percent; the than previously. average-bid option would reduce such spending by 8 percent, CBO now estimates.21 Some of the changes in For any given year, savings are larger in CBO’s cur- CBO’s projections result from legislative actions, which rent analysis than in the earlier analysis. For instance, were not included in CBO’s estimates of the likely range earlier, CBO estimated that the net federal savings in of savings. (For instance, changes to Medicare’s physician 2023 would be $56 billion for the second-lowest-bid payment system enacted after the 2013 analysis caused option and $17 billion for the average-bid option. CBO an increase in the savings estimates, as discussed below.) now estimates that net federal savings in 2023 would be $80 billion for the second-lowest-bid option and Reasons for the Changes in the Estimates $33 billion for the average-bid option. Savings would In both the current analysis and the earlier one, CBO be significantly larger in 2023 in the current analysis, estimated the budgetary effects of the premium sup- even though that year would occur sooner after the port options by using the most recent bids of Medicare implementation of the premium support options. (CBO Advantage plans to project what those bids would be expects that federal savings would be slightly smaller in under current law. The agency then estimated the bids percentage terms in the early years of the premium sup- that private plans would submit under the premium sup- port options than in later years.) port options by estimating certain downward or upward pressures on the projected Medicare Advantage bids. Estimates of the effects of the premium support options Such pressures would result from important differences on federal spending for the next decade are highly uncertain, given the substantial changes to the Medicare program that would be required, the government’s lack of experience with such a system, the rapid evolution of health care and health insurance, and the significant 20. See Congressional Budget Office, A Premium Support System for changes occurring in the Medicare program under Medicare: Analysis of Illustrative Options (September 2013), www. current law. Estimates for the period after 2026 are even cbo.gov/publication/44581. The ranges for the parameters’ values more uncertain. were chosen on the basis of CBO’s judgment that, accounting for many sources of uncertainty, there would be about a two-thirds’ In its September 2013 report, CBO characterized chance that the effect on federal spending would be within the uncertainty in its estimates by specifying ranges of range of values estimated (assuming that the premium support system was implemented as specified and other laws remained generally unchanged). CBO conducted that analysis for the two 19. Because of differing assumptions about when legislation premium support options without grandfathering. establishing the premium support system would be enacted, Options for Reducing the Deficit: 2014 to 2023 presented 21. The current estimates for 2024 are comparable to the earlier budgetary estimates for 2018 to 2023 (six years); the current estimates for 2020; in each case, they capture the third year under report presents such estimates for 2022 to 2026 (five years). the new system. 12 A Premium Support System for Medicare: Updated Analysis of Illustrative Options October 2017 between the premium support options and the Medicare FFS spending per capita in 2012 was 0.92; in its current Advantage program.22 analysis, CBO estimates that the average ratio in 2016 is 0.90 (see Figure 2).23 The increases in CBO’s estimates of federal savings from the illustrative premium support options without grand- The second reason for the decline in the projected ratio fathering are the net effect of several factors: of Medicare Advantage bids to FFS spending per capita is the repeal of the sustainable growth rate (SGR) for- ■■ CBO’s projections of Medicare Advantage bids mula for updating Medicare’s physician payment rates.24 relative to FFS spending per capita under current law In CBO’s earlier analysis, the agency projected that the are lower than in the agency’s earlier analysis. That SGR formula would result in a substantial reduction change is by far the largest contributor to the increase in payment rates in Medicare’s FFS program but that in estimated savings. Medicare Advantage plans would not achieve compara- ble reductions. As a result, CBO projected that the ratio ■■ CBO currently expects that some of the downward of Medicare Advantage bids to FFS spending per capita and upward pressures on bids would differ from its would rise from 0.92 in 2012 to 0.96 in 2020. earlier analysis. On net, changes in those pressures lowered CBO’s projections for the bids and thus CBO’s current estimates incorporate the effects of 2015 increased estimated savings. legislation that replaced the SGR formula with new systems for updating Medicare’s physician payment ■■ CBO modified the specification of the premium rates. CBO projects substantially higher payment rates support options based on the assumption that the in Medicare FFS than would have been the case if the federal government would ensure that the risk scores SGR formula had been retained. CBO also now projects of private-plan enrollees would be, on average, only that Medicare Advantage plans will bid lower relative to 5 percent higher than the risk scores of beneficiaries FFS Medicare spending per capita and that those bids in similar health in Medicare’s FFS program. The and FFS spending per capita will grow at roughly the reduction in the risk scores would tend to reduce same rate. As a result, in CBO’s projections, the ratio of federal payments to the plans (and thus increase Medicare Advantage bids to FFS spending per capita will federal savings). But CBO expects that private remain at 0.90 throughout the projection period. The insurers would raise their bids in response to that replacement of the SGR formula was a more important change in risk adjustment, partially offsetting the contributor to the lower ratio of projected Medicare other effects of reducing the risk scores. 23. Those figures reflect Medicare Advantage bids and FFS spending Reduction in Projected Medicare Advantage Bids for beneficiaries in average health. CBO excluded private fee- Relative to FFS Spending per Capita Under Current for-service plans, special needs plans, and employment-based group plans because the agency does not regard the bids of those Law. CBO currently projects that, under current law, the plans as providing a good basis for estimating the bids that bids of Medicare Advantage plans will be lower relative would be submitted by private insurers under the illustrative to FFS spending per capita than it estimated in 2013. premium support options. For each future year in its analyses, That reduction arises from two factors. CBO estimated the FFS spending per capita in the service areas of Medicare Advantage plans by using projections of county-level FFS spending per capita developed by the Centers for Medicare First, on average, the bids of Medicare Advantage plans & Medicaid Services and adjusting those values so that the have declined relative to FFS spending per capita in national estimate matched CBO’s estimate in its most recent recent years. In its earlier analysis, CBO used Medicare baseline budget projections. CBO included in its calculations the Advantage plans’ 2012 bids as the basis for its projec- government’s cost of administering the FFS program. tions; for this analysis, CBO used the bids submitted for 24. The SGR mechanism was designed to control spending for 2016. At the time of its earlier analysis, CBO estimated physicians’ services in Medicare FFS by setting an overall that the average ratio of Medicare Advantage bids to target amount for such spending (measured both annually and cumulatively). Payment rates were to be adjusted each year to reflect differences between actual spending and the spending 22. For details, see Congressional Budget Office, A Premium Support target. The Medicare Access and CHIP Reauthorization Act System for Medicare: Analysis of Illustrative Options (September 2013), of 2015 replaced the SGR formula with new systems for www.cbo.gov/publication/44581. establishing the annual updates to the payment rates. October 2017 A Premium Support System for Medicare: Updated Analysis of Illustrative Options 13 Figure 2 . Ratio of Actual and Projected Medicare Advantage Bids to Medicare FFS Spending per Capita Under Current Law, in Two CBO Studies 1.00 Projected 0.95 The much lower projection for the Actual 2024 ratio is attributable partly to a decline in Medicare Advantage bids 0.90 between 2012 and 2016 and partly to a change in law affecting FFS physician payment rates that led CBO to raise its projection of FFS 0.85 spending per capita. 2012 2020 2016 2024 0.80 2013 Report 2017 Report Source: Congressional Budget Office. Data on Medicare Advantage bids exclude private fee-for-service plans, special needs plans, and employment-based group plans. Those bids reflect spending for a beneficiary in average health. Data on Medicare’s FFS spending per beneficiary exclude spending on benefits not covered by the bids that Medicare Advantage plans submit under current law: spending for medical education, hospice benefits, and certain benefits for patients with end-stage renal disease. Medicare Advantage bids and Medicare FFS spending per beneficiary exclude spending for prescription drug benefits covered under Part D. CBO used data from 2012 for its 2013 analysis and data from 2016 for its current analysis. Projections are shown for the third year of an illustrative premium support option: 2020 for the 2013 report and 2024 for the 2017 report. FFS = fee for service. Advantage bids to FFS spending than the decline in that At that time, no data were available on the provider ratio between 2012 and 2016. payment rates of Medicare Advantage plans. However, discussions with industry sources suggested that the Changes in the Pressures on Bids Specified in the rates private insurers paid to providers in their Medicare Earlier Analysis. CBO has changed its estimates of two Advantage plans were, on average, similar to Medicare’s of the pressures on bids that were incorporated into FFS rates and much lower than the rates they paid to the agency’s earlier analysis, resulting, on net, in lower providers in their commercial plans. Because there was projected bids and greater federal savings. In addition, considerable uncertainty regarding insurers’ payment CBO has incorporated a new upward pressure on bids, rates to providers, CBO had anticipated that those rates described in the next section. could reasonably be connected to the proportion of Medicare beneficiaries enrolled in private plans. As a Reduction of Upward Pressure. In its 2013 analysis, CBO result, CBO expected that a decline in the market share anticipated that a reduction in the share of beneficia- of the FFS program under the premium support options ries enrolled in Medicare’s FFS program would tend to would reduce the importance of FFS payment rates in increase the prices that private insurers paid providers determining private insurers’ rates for Medicare enrollees, and thereby lead those insurers to increase their bids.25 causing those rates to increase. 25. For the rationale, see Congressional Budget Office, A Premium Support System for Medicare: Analysis of Illustrative Options (September 2013), pp. 39–40, www.cbo.gov/publication/44581. 14 A Premium Support System for Medicare: Updated Analysis of Illustrative Options October 2017 Box 1. The Role of the Medicare Fee-for-Service Program and Its Provider Payment Rates The Congressional Budget Office assumed that the provi- factor in enhancing insurers’ negotiating power with provid- sion of the Social Security Act that prohibits out-of-network ers and enabling them to pay Medicare’s FFS rates for their providers from charging more than Medicare’s fee-for-service Medicare Advantage plans.1 CBO expects that if such a provi- (FFS) rates to treat Medicare beneficiaries in private plans sion was excluded from the premium support options, private would be retained under the premium support options. plans would pay providers higher rates than they would under CBO also assumed that the Medicare FFS program would current law. be offered as a competing plan within the premium support In CBO’s assessment, eliminating the FFS program entirely options analyzed in this report. If either feature was removed, could cause a substantial increase in the rates that private CBO anticipates, private insurers’ payment rates to providers insurers pay providers and could cause a concomitant increase would be higher than those projected in this analysis and the in the costs of providing Medicare coverage. In the agency’s savings from the premium support options would be smaller assessment, the presence of Medicare’s FFS program as (or federal spending could be more than it would be under an alternative constrains the rates that private insurers pay current law). providers in Medicare, and eliminating the FFS program would The prices that private insurers pay providers under their cause those rates to rise toward commercial rates. Medicare Advantage plans are generally similar to Medicare’s FFS prices; insurers pay much higher prices under their com- 1. See Robert A. Berenson and others, “Why Medicare Advantage Plans mercial plans. Industry sources have identified the provision Pay Hospitals Traditional Medicare Prices,” Health Affairs, vol. 34, no. 8 of the Social Security Act mentioned above as an important (August 2015), pp. 1289–1295, http://dx.doi.org/10.1377/hlthaff.2014.1427. On the basis of more recent evidence that the provider would cause private insurers to reduce their bids relative payment rates of Medicare Advantage plans are not to the bids that would be submitted under the Medicare related to the FFS program’s share of beneficiaries in a Advantage program, with a resulting increase in federal plan’s service area, CBO eliminated that upward pressure savings. That downward pressure on bids is 25 percent from its analysis.26 The agency now expects that insur- weaker in CBO’s current analysis than it was in the ers’ payment rates under the premium support options earlier work, resulting in smaller federal savings, partially would not increase if there was a decline in the share of offsetting the effect discussed above. beneficiaries enrolled in the FFS program in a particu- lar market. That change lowered CBO’s projections of For this analysis, CBO reduced its estimate of the private insurers’ bids under a premium support system, amount of downward pressure that would result from relative to its earlier analysis, and increased the amount it increased competition, for two reasons. First, in 2013, projected in federal savings. (Box 1 describes some of the CBO expected that increased competition would result ways that changes in the options would increase upward in insurers’ reducing costs, by, for instance, lowering pressure.) administrative costs or profit, improving care manage- ment, restricting provider networks, or adopting new Reduction in Downward Pressure. In its earlier analysis, technologies more slowly. Since 2013, insurers have CBO estimated that the increased competitive pressure reduced their bids relative to Medicare FFS spending per created by the premium support options it analyzed capita, but how they did so is unclear. If, for instance, insurers reduced bids primarily by cutting profit mar- gins, it is unlikely that they would be able to cut margins 26. That change reflects the findings of CBO’s analysis of private insurers’ hospital payment rates in their Medicare Advantage by a similar magnitude again. Conversely, if they used plans. See Jared Lane Maeda and Lyle Nelson, An Analysis of different coding practices to increase the risk scores of Private-Sector Prices for Hospital Admissions, Working Paper 2017- Medicare Advantage enrollees relative to those of simi- 02 (Congressional Budget Office, April 2017), www.cbo.gov/ lar FFS beneficiaries—and thereby reduced bids while publication/52567. October 2017 A Premium Support System for Medicare: Updated Analysis of Illustrative Options 15 keeping payments from the federal government the support options.28 Specifically, CBO now estimates same—they might be able to achieve further reductions that the federal government would reduce private-plan through the mechanisms that CBO described in 2013. enrollees’ risk scores such that they would exceed those Given the uncertainty about how insurers reduced their of FFS beneficiaries in similar health by an average of bids, CBO projected that plans could still achieve most, only 5 percent. Without that adjustment, CBO expects, but not all, of the reductions that the agency considered the average difference in risk scores between FFS and possible in its 2013 report. private-plan enrollees would be substantially greater than 5 percent (assuming that a risk-adjustment system com- CBO’s consultation with outside experts also led it to parable to that used in the Medicare Advantage program reduce its estimates of downward pressure from com- would be used under premium support). petition. The agency now places greater weight on the possibility that the incentives insurers face to reduce bids The reduction in the risk scores would tend to reduce would be countered to some extent by their understand- federal payments to the plans (and thus increase federal ing that the federal contribution would be increased if savings) because those payments would be risk adjusted. they raised their bids. The strength of the incentive to However, CBO expects that private insurers would raise bids (or to limit their reduction) could depend on respond by increasing their bids, which would tend to the number of competitors in a given market and on increase federal payments to the plans, partially offsetting such other factors as the market share of the FFS pro- the other effects of reducing the risk scores. gram and the differences between private insurers’ bids and the FFS bid. How Much Did CBO’s Estimates of Effects With Grandfathering Change and Why? Change in the Specification of Risk Adjustment Under CBO’s current estimates of federal savings in the first the Premium Support Options. To adjust for differences few years under the illustrative premium support options in coding, federal law currently requires Medicare to with grandfathering are similar to its earlier estimates. apply an across-the-board reduction to the risk scores of In 2013, CBO estimated that in the third year of such Medicare Advantage enrollees. A minimum reduction is a system, the second-lowest-bid option would reduce specified each year, and although larger reductions are net federal spending for Medicare by $8 billion and permitted, to date Medicare has applied only the mini- the average-bid option would reduce net federal spend- mum. Despite that, the evidence suggests that the reduc- ing by $3 billion.29 CBO currently estimates that the tions have not fully compensated for Medicare Advantage second-lowest-bid option with grandfathering would plans’ more intensive coding.27 reduce net federal spending by $9 billion in the third year of operation and that the average-bid option would In its earlier analysis, CBO projected that the federal reduce net federal spending in that year by $4 billion. government would reduce the bids of private-plan enrollees by just the minimum amount required by law CBO’s current and earlier estimates are similar because under the Medicare Advantage program. In this analy- of two offsetting factors. On the one hand, the reduction sis, CBO anticipates that the federal government would in CBO’s projection of Medicare Advantage bids relative apply a greater downward adjustment to the risk scores of private-plan enrollees under the premium support 28. If the use of more intensive diagnostic coding led to substantially options than it has under the Medicare Advantage pro- higher risk scores for private-plan enrollees, private plans could reduce their bids, which would lower the federal contribution gram because the consequences of a divergence between and increase the premiums for beneficiaries who chose to remain the risk scores would be greater under the premium in the FFS program. CBO did not specify a larger reduction in its earlier analysis because, at that time, it expected that the risk 27. See Medicare Payment Advisory Commission, “MA Risk scores of private-plan enrollees would exceed those of similar FFS Adjustment and Coding Intensity Adjustment,” in Report to the beneficiaries by a much smaller amount than the current analysis Congress: Medicare Payment Policy (March 2016), pp. 344–346, indicates. www.medpac.gov/-documents-/reports; and Government 29. See Congressional Budget Office, Options for Reducing the Deficit: Accountability Office, Medicare Advantage: Substantial Excess 2014 to 2023 (November 2013), pp. 204–210, www.cbo.gov/ Payments Underscore Need for CMS to Improve Accuracy of Risk content/options-reducing-deficit-2014-2023. Score Adjustments, GAO-13-206 (January 2013), www.gao.gov/ products/GAO-13-206. 16 A Premium Support System for Medicare: Updated Analysis of Illustrative Options October 2017 to FFS spending per capita under current law increased under current law). Because health plans cannot increase the agency’s estimates of federal savings, compared with demographically based risk scores, there is limited scope its earlier estimates. On the other hand, as a result of for coding differences between private plans and FFS technical improvements to its modeling, CBO now Medicare for those beneficiaries. Under grandfathering, estimates that a smaller share of Medicare spending new beneficiaries would make up a larger share of people would be included in a premium support system with a who were affected by premium support in the early grandfathering provision, which reduced the estimate of years. As a result, CBO projects, the limits on coding savings. differences imposed on private plans would have a much smaller effect on plan payments, and thus those plans Some factors identified above as contributing to changes would need to increase their bids by correspondingly in CBO’s estimates resulted in much smaller changes for smaller amounts. options implemented with grandfathering. Two—the downward pressure on bids related to increased compe- The second reason that CBO expects a smaller gap tition and the upward pressure related to changes in the between the risk scores of private-plan and FFS enroll- market share of the FFS program—would have small ees in the early years under grandfathering is that the effects on federal savings in the early years because only private-plan enrollees would be in those plans for a small proportion of the Medicare population would be relatively short periods. CBO’s internal analysis suggests included in the new system. Changes in those factors had that differences between the risk scores of private-plan little effect on CBO’s estimates. enrollees and similar FFS beneficiaries increase with the length of time enrollees are in a given private plan. In an approach similar to that described above, CBO Because insurers have more opportunity to identify diag- changed the specification of the premium support noses for long-time enrollees, those insurers will be less options to include an assumption that the federal govern- able to increase risk scores for beneficiaries who are new ment would limit the risk scores of private-plan enrollees to Medicare. to be, on average, no more than 5 percent higher than the risk scores of FFS enrollees in similar health. CBO CBO therefore anticipates that the reduction in private expects that the limits on coding differences would plans’ risk scores to account for coding differences would have a smaller effect on private plans’ bids in the early have a smaller effect on revenue for those plans in a system years under grandfathering, for two reasons. First, in with a grandfathering provision than in a system without the initial year, all enrollees in the premium support one. Consequently, CBO anticipates that private insur- system would be new Medicare beneficiaries. Because ers would increase their bids by smaller amounts under a history of Medicare claims would not be available for grandfathering; as a result, that new specification had a those enrollees, their risk scores would be computed only minimal effect on CBO’s estimates of federal savings. on the basis of demographic characteristics (as is done About This Document This Congressional Budget Office report was prepared in response to interest expressed by Members of Congress. It is a supplement to the 18 budget options related to health that CBO published in December 2016, and it updates A Premium Support System for Medicare: Analysis of Illustrative Options, which CBO published in 2013. In keeping with CBO’s mandate to provide objective, impartial analysis, this report makes no recommendations. Daria Pelech and Lyle Nelson prepared the report with contributions from Tamara Hayford and Paul Masi (formerly of CBO). Elizabeth Bass, Tom Bradley, Holly Harvey, and David Weaver provided comments. Ru Ding and Romain Parsad assisted with programming, and Paul Jacobs (formerly of CBO) and Eamon Molloy provided technical assistance. Jeffrey Kling and Robert Sunshine reviewed the report, Kate Kelly edited it, and Jorge Salazar prepared it for publication. An electronic version and supplemental information are available on CBO’s website (www.cbo.gov/publication/53077). Keith Hall Director October 2017