Statement of
Douglas Holtz-Eakin
Director

Prescription Drug Coverage and
Medicare's Fiscal Challenges

before the
Committee on Ways and Means
U.S. House of Representatives

April 9, 2003

Chairman Thomas, Congressman Rangel, and Members of the Committee, I am pleased to be here with you today. I understand that the focus of this hearing is on expanding Medicare's coverage of prescription drugs, and I am prepared to discuss that topic in some detail. But I would first like to frame that discussion by looking at Medicare's overall financial picture, both in the near term and the long run. As this Committee well knows, Medicare is projected to consume an ever-larger piece of our national income just in delivering its current set of benefits. In determining whether and how to add prescription drug coverage to its benefit package--and the desirability of adopting other reforms to the program at the same time--lawmakers will face the challenge of balancing the needs of beneficiaries against the resulting pressures on the economy. To assist in that effort, I will describe the Congressional Budget Office's (CBO's) latest projections of prescription drug coverage and of drug spending for the Medicare population. I will then conclude my testimony by outlining some of the key issues that arise in designing a prescription drug benefit for Medicare.
 

Factors Driving Medicare Spending

Under current law, Medicare spending--measured as a share of the economy--is projected to nearly quadruple by 2075, growing to more than 9 percent of gross domestic product (GDP) from its current level of 2.5 percent. As a consequence, Medicare will necessarily compete with other spending priorities for a much greater share of the federal budget or with private-sector spending for a bigger share of the national economy--or with both. In thinking about how to address the substantial challenges that the Medicare program faces, however, it is important to recognize that they are not unique to Medicare; rather, they reflect the broader forces of an aging society, the rising costs of health care generally, and the looming long-range financial strains that will affect the federal government and the economy as a whole.

Clearly, part of the challenge facing Medicare stems from the demographic trends that are making the country as a whole older. From 1970 to 2010, the number of Americans ages 20 to 64 is projected to increase by nearly 80 million; the elderly population by 2010 will have grown by about 20 million, or roughly one-fourth as much. In contrast, for the period 2010 to 2030--when the baby-boom generation will retire--the number of working-age individuals is projected to grow by about 10 million, whereas the population ages 65 and older will increase by 30 million, or three times as much. The consequence of those diverging patterns is that the ratio of the elderly population to the population in its prime working years--which stood at 19 percent in 1970--is projected to grow from 21 percent today to 35 percent by 2030. The ratio is then expected to continue to climb (albeit at a slower rate) and could reach 42 percent in 2075. In other words, the shift to an older society will accelerate as the baby-boom generation retires, and it will persist afterwards, making the changes that the nation faces--and their implications for the spectrum of federal tax and spending policies--more than just temporary.

Compounding those demographic pressures are the seemingly inexorable increases in health costs per person--but that issue, too, is not limited to Medicare. Nationally, health care expenditures as a percentage of GDP have more than doubled over the past several decades, growing from 7.0 percent in 1970 to 14.8 percent in 2002. On a per capita basis, national spending for health care (in 2002 dollars) increased from $1,321 in 1970 to $5,366 in 2002, or at an average rate of about 4.5 percent per year--which is about 2.4 percentage points faster than the growth of the underlying economy. The factors contributing to the trend in real (inflation-adjusted) per capita health care spending include expansions in insurance coverage, rising income, medical price inflation in excess of general inflation, and the aging of the population--but the major impetus has been the development and diffusion of new medical technology. At the same time, it should be noted that improvements in that technology--while costly--have increased the health care system's potential to deliver high-quality care. If the adoption of new technology is driven by the needs of patients, the value of those improvements may well exceed their cost.

Over the 1970-2002 period, Medicare spending has risen even more rapidly than national health expenditures, growing eightfold even after adjusting for inflation. As a share of GDP, Medicare costs rose from 0.7 percent in 1970 to their current level of 2.5 percent. Although cost growth on a per-enrollee basis has been volatile, it has also tended to rise at a much faster pace than the economy has grown. Over the period, real costs per enrollee grew more than twice as fast as the economy--specifically, at the rate of per capita GDP plus 2.8 percentage points. One reason that total Medicare costs have grown more quickly than overall health costs is that the number of beneficiaries has grown more quickly than the U.S. population as a whole, owing both to program expansions and to the increase in the share of Americans who are elderly. In terms of costs per beneficiary, the growth of Medicare spending is due in part to the same factors that have driven increases in health care spending nationally, but it also reflects legislative and administrative expansions of the program's benefit package.

In general, precisely determining each factor's effect on overall program spending is difficult. As an illustration, however, consider spending for services provided to fee-for-service program enrollees during acute care hospital stays (which now account for about one-third of Medicare's total costs). The program's total spending for those services grew by 261 percent between 1972 and 1998, after adjusting for general inflation (see Table 1). That growth in total spending is the product of three factors: increases in the number of Medicare beneficiaries; increases in the number of hospital admissions per beneficiary; and--the most important factor--increases in the real cost per admission. That cost nearly doubled over the period in real terms and accounted for 57.4 percent of the overall growth. Over the same period, the number of enrollees in the fee-for-service program increased by about 50 percent, contributing 30.3 percent of the rise in spending. The number of hospital admissions per beneficiary grew more slowly and accounted for only 12.3 percent of the increase in total costs.

Table 1.
Sources of Fee-for-Service Medicare Cost Growth for Acute Care Hospital Services
	1972	1998	Percentage Increase, 1972-1998		Percentage Share of Total Increase
Total Costs (Millions of dollars)	21,744	78,522	261.1		100.0
Number of Beneficiaries (Millions)	21.1	32.0	51.3		30.3
Admissions per Beneficiary	0.302	0.365	20.9		12.3
Cost per Admission (Dollars)	3,408	6,724	97.3		57.4
Source: Congressional Budget Office based on Department of Health and Human Services, Health Care Financing Administration, Health Care Financing Review: Medicare and Medicaid Statistical Supplement, 2000.
Note: The costs noted in the table (which are in 1998 dollars) reflect inpatient costs for fee-for-service enrollees at acute care hospitals.

One valuable feature of such a breakdown is that it highlights the factors driving Medicare spending that lawmakers can influence and those that they cannot. In this case, costs per admission reflect the mix of, and prices for, therapies or services provided in an average admission. Today, policymakers can directly control only one of those two components: the price paid for a given service, which is updated annually as specified by statute. Thus, for example, lawmakers can seek to change the increase in payments for procedures such as a coronary artery bypass graft, but they do not control the share of total admissions accounted for by each procedure--which results from decisions made by doctors and their patients. The payment systems that are established in law do influence how doctors and other health care providers make treatment decisions. Similarly, features such as the cost sharing for those services can affect what beneficiaries choose to do. But the impact of changes in policy on those individual decisions is complicated and far from direct.
 

CBO's Projections of Medicare Spending Under Current Law

With that historical view in mind, let me turn now to CBO's projections of Medicare spending for the next 10 years, which were updated in March. CBO projects that gross outlays for Medicare benefits will total $271 billion in 2003 and $3.9 trillion over the 2004-2013 period (see Table 2). As a share of the economy, those Medicare outlays are projected to rise from 2.5 percent in 2003 to 2.9 percent in 2013, on average constituting 2.7 percent of GDP over the 2004-2013 period. After deducting projected premium payments by beneficiaries--which amount to $28 billion in 2003 and $461 billion over the 10-year period--CBO estimates that net spending for Medicare benefits will total $243 billion in 2003 and $3.4 trillion from 2004 through 2013. All of CBO's projections reflect the assumption that current law remains unchanged, thereby establishing the "baseline" for legislative proposals.

Table 2.
Summary of CBO's March 2003 Baseline Projections of Medicare Benefit Outlays
(By fiscal year)
	Billions of Dollars				Average Annual Rate of Growth, 2004-2013 (Percent)
	2003		2004-2013
Gross Benefit Outlays	271		3,880		6.9
Premiums	-28		-461		8.2
Net Benefit Outlays	243		3,419		6.8
Source: Congressional Budget Office.
a. Outlays exclude spending by Medicare for quality improvement organizations, health care fraud and abuse control, and other administrative costs. Total spending on those activities is projected to be $5.4 billion in 2003. Of that amount, $3.8 billion is subject to appropriation.

Focusing on the program's growth rates, CBO projects that net spending for Medicare benefits will increase by 5.9 percent in 2003 and will grow at an average annual rate of 6.8 percent over the 2004-2013 period. In recent years, the annual rate of growth of Medicare spending has varied considerably. Growth averaged 1.2 percent annually during the 1997-2000 period but has averaged more than 8 percent since then. Spending for benefits provided under Part B of Medicare (Supplementary Medical Insurance) grew particularly rapidly in 2002, driven by a significant rise in the volume and intensity of physician services and by increases of about 20 percent in spending for durable medical equipment and physician-administered pharmaceuticals. Costs for Part A of Medicare (Hospital Insurance) also rose sharply, including a 10 percent increase in spending for inpatient hospital services.

The projected growth rates of Medicare's payments vary by service type. Total payments to hospitals for inpatient services and payments to physicians, which together account for two-thirds of the program's outlays, are the slowest-growing components of spending for fee-for-service enrollees, respectively averaging 6.4 percent and 5.9 percent annually in CBO's baseline projections through 2013. By contrast, rates of growth for the costs of other services--for example, those provided by home health agencies and nonphysician professionals--are projected to average 10 percent to 13 percent annually (but will still constitute a relatively small share of total Medicare spending).

Over the next decade, CBO expects several factors to play a major role in the program's cost growth. Those factors include rising levels of enrollment in Medicare and automatic increases in payment rates for many services in the fee-for-service program (to adjust rates for rising input costs). CBO also projects changes in the use of Medicare's services, reflecting an increase in the number of services furnished per enrollee as well as a shift in the mix of services toward those that are higher priced and (often) more technologically advanced. In part offsetting the effects of those spending components on total costs will be small or negative updates (adjustments) to payment rates for physician services and smaller updates (relative to cost increases in the fee-for-service program) to the rates paid to Medicare+Choice plans.

Specifically, increases in payment rates account for about 45 percent of the projected rise in Medicare spending over the next decade; the other 55 percent is equally divided between increases in enrollment and changes in the quantity and mix of services delivered per beneficiary. As noted above, payment rates are the easiest factor for policymakers to control. Rates for many services are automatically adjusted for rising input costs. In the past, legislation has frequently limited those increases to less than the full change estimated for those costs. Since 1990, for example, updates to payment rates for hospital admissions have averaged about 1 percentage point less than the increase in the market-basket index used to measure increases in the cost of hospital inputs. Under current law, however, payment rates for services furnished by hospitals and many other providers will automatically rise by the full amount of the increase in estimated input costs, as a result of the expiration of many of the provisions contained in the 1997 Balanced Budget Act.

Medicare's payment rates for physician services are subject to a very different update formula. Most recently, the Balanced Budget Act established an ongoing target for cumulative spending for physician services (and services that accompany physician visits). By statute, that target is automatically adjusted each year for changes in physicians' input costs and in the program's enrollment--plus the change in GDP per capita. (Future effects of enacted legislation and of regulation are also taken into account.) In the absence of per capita GDP growth, the real (inflation-adjusted) target for spending per enrollee remains unchanged. Increases in GDP per capita thus act as an allowance to cover increases in the number and average cost of services being furnished per enrollee as technology and medical practices evolve over time. If total spending for physicians deviates from that allowance--in either direction--then the annual updates to payment rates are adjusted over a period of several years to bring cumulative spending back in line with the target.

By the time payment rates were set for 2002, expenditures for physician services had exceeded the cumulative target, so rates for those services were reduced by about 5 percent, and a further reduction of 4.4 percent was originally scheduled for 2003. However, the Department of Health and Human Services invoked a provision of the 2003 Consolidated Appropriations Resolution to increase the cumulative target for 2002. As a result, payment rates for physician services in 2003 were increased by 1.6 percent.

Nevertheless, CBO projects that spending for physician services will again exceed the target in 2003 and remain above it on a cumulative basis through 2013. Therefore, in the absence of further legislative action, payment rates for those services are likely to decline (in absolute terms) for the next several years. (For example, last month the Centers for Medicare and Medicaid Services released a very preliminary estimate of the physician fee schedule update for calendar year 2004 indicating that payment rates could be cut by 4.2 percent.) At the same time, the total volume of services provided will continue to rise as the number of beneficiaries increases and the number of services provided per beneficiary grows. As a result, CBO projects that total Medicare spending for physician services--which is the product of the prices paid and quantities used for the mix of services provided--will rise each year through 2013, on both an aggregate and a per capita basis. I should reiterate here that those projections reflect CBO's best estimate of what will occur under the assumption that no changes are made in current law; in the past, lawmakers have often acted to modify those payments, whether to correct discrepancies between payment rates and the costs providers incur or for other purposes.
 

Medicare's Long-Term Financing Challenges

Although the 10-year budget window for Medicare now includes enrollment of the first wave of the baby-boom generation--those individuals born between 1946 and 1948, who will turn 65 by 2013--a complete picture of the program's fiscal outlook requires an even longer view. Toward that end, CBO projected the cost of Medicare as a share of GDP out to 2075 to show how much of the country's production of goods and services would be needed to pay for the program as it is currently structured. Although we are continuing to refine our projection models, CBO currently estimates that Medicare's costs as a percentage of GDP will rise from 2.5 percent in 2003 to 9.2 percent in 2075. Approximately 30 percent of that growth is due to society's aging and the resulting increase in the number of Medicare beneficiaries; the remaining 70 percent is attributable to the growth of health care costs per enrollee in excess of the rate of growth of GDP per capita (see Figure 1).
 

Figure 1.

Projected Long-Term Growth of Medicare Spending

(Percentage of GDP)

Source: Congressional Budget Office.

For a sense of the magnitudes involved, if the Medicare program's costs accounted for 9.2 percent of GDP today, they would equal half of what is now spent by the entire federal government. If the program's higher costs were simply added to current federal spending, total federal receipts (which currently absorb about 18 percent of GDP) would have to be one-third larger to balance the budget. And if those increased costs were paid for entirely through a payroll-based tax, the rate for Social Security and Medicare, now set at 15.3 percent on the earnings of most workers, would have to more than double--a rise equal to roughly $6,000 per worker (that is, $3,000 each for the worker and his or her employer).

Of course, the fiscal challenges facing Medicare will occur in parallel with those for Social Security and Medicaid. Those three programs now absorb 8 percent of GDP, but if CBO's projections hold, that figure will rise to 14 percent by 2030. Beyond that point, spending pressures will only intensify, with life expectancy continuing to increase and health costs continuing to grow. CBO projects that by 2075, the cost of the three programs could climb to 21 percent of GDP, the largest portion of which would be attributable to Medicare. To accommodate the increase in spending, either taxes would need to be raised dramatically or spending on other federal programs would have to be curtailed severely--or federal borrowing would soar.

For Medicare, the most significant factor affecting those projections is that annual growth of spending per beneficiary is expected to increase faster than per capita GDP growth--but much less quickly than in the past. CBO's current projection assumes that per capita Medicare spending will eventually grow 1 percentage point faster than per capita GDP, a rate that is substantially slower than the 2.8 percentage-point "excess cost" rate that the program has experienced over the past 32 years (part of which has been due to program expansions). CBO's assumption of an eventual deceleration in the relative rise of health care costs is consistent with that of the Medicare trustees (as well as others) and reflects the view that forces within the health care sector will operate to slow the rate of growth somewhat.

But that assumption might be too optimistic, and even seemingly small deviations from it could have significant economic implications when costs are projected over long periods. For example, if the growth of per capita Medicare costs slowed only to the rate of per capita GDP growth plus 1.5 percentage points, then program outlays would equal 5.4 percent of GDP in 2030 and 13.2 percent in 2075 (and if the health sector as a whole grew at that rate, it would account for more than half of the economy's output by 2075). Adding to the uncertainty is the potential for program expansions, because enacting a new prescription drug benefit or easing existing limits on payments to providers could exacerbate the rising long-term spending trajectory.
 

Prescription Drug Coverage and Spending

I would now like to describe CBO's latest projections of prescription drug coverage and spending for the Medicare population under current law. I offer them not just because they serve as the basis for our estimates of legislative proposals to add a drug benefit to Medicare but also because they may provide useful insights for the design of such proposals.

Most Medicare beneficiaries now have coverage for prescription drugs at some point in the year, but the extent of that coverage varies widely. CBO's analysis of the Medicare Current Beneficiary Survey indicates that in 2000 (the most recent year for which data are available), 75 percent of the Medicare population--or roughly 30 million individuals--had some form of insurance coverage for the costs of prescription drugs for at least part of the year; 25 percent--or roughly 10 million beneficiaries--had no drug coverage. Beneficiaries who have coverage for their drug costs obtain it from a variety of sources. For example, nearly 30 percent of Medicare beneficiaries obtained coverage through employer-sponsored retiree benefits, and another 16 percent had coverage through the Medicaid program. About 12 percent of beneficiaries are estimated to have had drug coverage through individually purchased medigap policies, while the remainder obtained coverage through a Medicare+Choice plan or from another state or federal program.

CBO's estimates of the total number of Medicare beneficiaries grouped by income and the share of them who lacked drug coverage throughout 2000 appear in Figure 2. Although the fraction of beneficiaries without coverage varied from 32 percent (for those with income between 100 percent and 150 percent of the federal poverty level) to 22 percent (for those with income exceeding 400 percent of poverty), CBO's main finding is that the differences across the income spectrum are not dramatic. The varying degrees of coverage are likely to reflect both difficulties in obtaining private drug coverage as well as rational "nonpurchase" of such coverage by beneficiaries with low levels of drug spending.
 

Figure 2.

Medicare Beneficiaries in 2000, by Income Level and Drug Coverage

(Millions of beneficiaries)

Source: Congressional Budget Office.

Clearly, the extent of the drug coverage that Medicare beneficiaries have today--and whether and how that coverage should be added to Medicare--is of central interest to policymakers, for two reasons: the elderly and disabled as a group use substantial amounts of prescription drugs, and their spending for such drugs has been rising rapidly in recent years. CBO's analysis indicates that Medicare beneficiaries bought about $1,500 worth of drugs, on average, in 2000 and that more than 90 percent of beneficiaries filled at least one prescription that year. Overall, about three-eighths of those costs were paid out of pocket, a figure that combines the payments of those without coverage (who pay the full cost of their drugs) and those with coverage (who incur copayments and deductibles). When average drug spending and out-of-pocket costs for Medicare beneficiaries are broken down by beneficiaries' level of income, again, the main finding is that average spending--both total and out-of-pocket--is remarkably similar for all income groups (see Figure 3).
 

Figure 3.

Average Prescription Drug Spending in 2000 by and for Medicare Beneficiaries

(Dollars)

Source: Congressional Budget Office.

As Figure 4 indicates, an important consideration in designing any Medicare drug benefit is how it will affect the out-of-pocket costs of enrollees as well as the large amount of payments currently made by third parties (including other federal programs). For example, in 2000, the 8.5 million Medicare beneficiaries with income between 200 percent and 300 percent of the poverty level used about $13 billion worth of drugs. Beneficiaries paid about $5 billion of that cost directly, and $8 billion was paid on their behalf. (Beneficiaries ultimately pay part of those covered costs if they pay a premium for their coverage.)
 

Figure 4.

Total Prescription Drug Spending in 2000 by and for Medicare Beneficiaries

(Billions of dollars)

Source: Congressional Budget Office.

 

CBO's Projections of Future Drug Spending

As the above data illustrate, elderly and disabled Medicare beneficiaries now consume substantial amounts of drugs. In addition, their spending is projected to continue growing at a rapid pace (as is drug spending for the country as a whole). For the period 2004 through 2013, CBO estimates that spending for prescription drugs by and on behalf of the Medicare population will total roughly $1.8 trillion, or nearly 50 percent of the projected $3.9 trillion in Medicare outlays over that same period. Over that period, CBO expects Medicare beneficiaries' average spending for prescription drugs to climb quickly--at an average annual rate of about 9 percent--even in the absence of a Medicare drug benefit.

CBO's current estimate of total drug spending is about 4 percent higher than its projection last year for the 2003-2012 period. Typically, shifting the projection period forward by one year adds a relatively expensive year and drops a relatively inexpensive one, leading to a larger increase. This year's estimate, however, reflects two offsetting factors: new information about the degree to which drug spending is underreported in current surveys (which slightly lowered the starting point for the projections); and somewhat lower projections of the rate of growth of drug spending (the result, in part, of slower-than-expected economic growth in the near term).
 

Issues in Designing a Drug Benefit

The financial challenges already facing the Medicare program and the significant sums that projections indicate its beneficiaries will spend on drugs combine to make designing a drug benefit for that program a formidable task. In considering how to design such a benefit, it is useful to begin with some key principles of insurance design that--as an economist--help me think through the complex issues involved and are related to some of the options with which the Congress is now grappling.

The first and foremost issue to confront is the structure of the benefit that is provided--that is, the deductible and cost sharing it will require. In general, well-designed insurance should reduce the risk of catastrophic financial losses yet leave individuals to cover their routine, expected expenditures with their own resources. Such a design would also reflect concern about the phenomenon known as "moral hazard"--in which further coverage would induce additional and perhaps excessive demand for services.

Applying that principle would suggest that Medicare's drug benefit should focus on protecting beneficiaries against very high drug costs. If Medicare adopted some kind of catastrophic approach, most enrollees would receive no payments in any given year, but they would nonetheless benefit from being protected against the possibility of catastrophic expenses. Several factors related to the nature of drug spending, however, complicate the application of a "pure insurance" approach. The two most important factors are the degree to which the distribution of drug spending is skewed and the degree to which it is persistent.
 

Concentration and Persistence of Drug Spending

Although most Medicare enrollees use some prescription drugs, the bulk of such spending is concentrated among a much smaller group. In 2000, about 26 percent of enrollees had expenditures of $2,000 or more, and together they accounted for 65 percent of total drug spending by the Medicare population. At the same time, 32 percent of beneficiaries had expenditures of $500 or less, making up about 4 percent of total spending.

Of course, skewed annual expenses by themselves are actually typical of insurance markets, since insurance is usually purchased to protect against a small but relatively random risk of a large loss. What makes insurance for drug coverage difficult to provide is that prescription drug costs persist over time for the same enrollees. In particular, a large share of drug spending is associated with treatment of chronic conditions--such as hypertension, cardiovascular disease, and diabetes--which are often evident by the time individuals become eligible for Medicare. The result is that potential enrollees have important "private information" about their future drug costs. That fact makes stand-alone drug coverage particularly susceptible to adverse selection, in which enrollment is concentrated among those who expect to receive the most in benefits.

Indeed, those same facts help explain why beneficiaries may find it difficult today to purchase private coverage for prescription drugs--or why catastrophic protection is virtually unavailable except through subsidized retiree coverage or Medicaid. If beneficiaries were given a choice about whether and when to purchase individual prescription drug coverage, people with high drug costs would be most likely to participate. That would drive premiums up, which in turn would reduce enrollment as enrollees with below-average drug costs dropped out. In the extreme, that spiral could lead to a market failure in which no insurance was sold, even if most people would be willing to pay more than the average cost of a policy that had broad enrollment. Those theoretical pressures are well illustrated in practice by today's market for new medigap policies that include a drug benefit (which cover as much as half of an enrollee's drug costs but cap the benefit at $1,250 or $3,000 per year). Insurers that offer such policies often charge a premium that represents a very large share of the maximum potential drug benefit--to reflect the average cost of their enrollees. Similarly, the drug coverage available through Medicare+Choice plans is generally subject to caps.

Most proposals for a Medicare drug benefit have sought to correct for such market failures by including coverage for catastrophic drug costs but, accordingly, must also include measures designed to avoid an adverse selection spiral. One potential approach would be to make enrollment mandatory. A related option would integrate drug coverage into the benefit package for Part B of Medicare (and charge a correspondingly higher premium), so that beneficiaries could not separate their choice of whether to obtain drug coverage from their decision to purchase coverage for less predictable health costs.

But most of the drug benefit proposals developed in recent years have sought to keep enrollment in the benefit as a separate option for the elderly and disabled. To mitigate the potential for adverse selection, they would use some or all of the following three methods:

• Restrict Participation. Most proposals have either given enrollees only one opportunity to choose the drug benefit--at the time they first become eligible for it--or imposed a substantial premium surcharge on those who delay enrollment. (Otherwise, beneficiaries with low drug costs would simply wait until they needed coverage to enroll.)



• Provide Up-Front Coverage. Many proposals have sought to make enrollment more attractive for beneficiaries with low drug costs by providing some coverage for their initial drug expenditures--for example, covering a substantial share of costs after beneficiaries meet a deductible that can be as low as $100.



• Offer High Premium Subsidy Rates. The extent of federal subsidization of premiums for a drug benefit is a key determinant of total federal costs for such a program both because of the direct costs and because the availability of subsidies would lead employers and state Medicaid programs to encourage or require full participation. However, such subsidies would also serve to encourage other beneficiaries with relatively low drug costs to enroll in the benefit. Most recent proposals have contained relatively high subsidy rates--67 percent or higher--which mean that enrollees would pay one-third or less of the average covered costs through their monthly premiums.

The Administration of a Medicare Drug Benefit

The way in which a drug benefit is administered also affects its costs, and the options for administration involve many of the same trade-offs between insurance and incentives that arise in designing the benefit itself. Most recent proposals have envisioned adopting the common private-sector approach of using pharmacy benefit managers (PBMs) to process drug claims. Those proposals would also give beneficiaries a range of drug plans from which to choose, either in conjunction with their choice of medical coverage or as a stand-alone benefit. The extent to which the organizations that administered a Medicare drug benefit could effectively constrain its costs would depend on the organizations' having both the authority and the incentive to use the various cost-control mechanisms at their disposal. Proposals have differed, however, in the nature and extent of the risk that the entities responsible for administering the benefit would assume, the kind of restrictions that would be placed on them in managing drug costs, and the structure of the competition among those entities to enroll and serve beneficiaries.

Private health plans use PBMs to process claims and negotiate price discounts with drug manufacturers and dispensing pharmacies. PBMs also try to encourage the use of certain drugs, such as generic, preferred-formulary, or mail-order pharmaceuticals--in part so that they can obtain lower prices for those preferred drugs that have competitors. In addition, because of their centralized records for each enrollee's prescriptions, they may help prevent adverse drug interactions and take other steps to help beneficiaries manage their own drug use.

In the private sector, PBMs often have considerable leeway in the tools they can use, but they do not assume any insurance risk for the drug benefit (although they may be guided or selected by an employer or insurer who does bear the residual risk). At most, they may be subject to a bonus or a penalty added to their administrative fee, which is based on how well they meet prespecified goals for their performance. Some proposals have envisioned having PBMs or similar entities administer a Medicare drug benefit in that way--accepting "performance risk" but not "insurance risk." In such models, all costs for benefit claims would be paid by the federal government as they were incurred.

Other proposals have adopted a different model, more akin to the risk-based competitive model characteristic of Medicare+Choice plans. Those proposals envision multiple risk-bearing entities (such as partnerships between PBMs and insurers) that would compete to serve enrollees. Enrollees would have some choice among providers, so that beneficiaries who were willing to accept more-restrictive rules (such as a closed formulary) in return for lower premium costs could do so, whereas others could select a more expensive provider with fewer restrictions. If the entities bore all of the insurance risk for the drug benefit--that is, if they received a fixed per capita payment for each enrollee--they would have strong incentives to use whatever cost-control tools were permitted. However, such tools might be unattractive to many beneficiaries, and the plans' administrators would also have strong incentives to try to achieve favorable selection by avoiding enrollees with the highest spending.

An additional concern about this model has been that entities might be unwilling to participate if they had to assume the full insurance risk for a stand-alone drug benefit. To mitigate that concern, proposals have included federally provided reinsurance for high-cost enrollees as well as so-called risk-adjustment mechanisms that would vary the per capita payments on the basis of enrollees' characteristics, such as their age or previous disease diagnoses. (Reinsurance means that the federal government shares part or all of the claims costs of high-cost enrollees.) Although reinsurance would reduce the incentives to avoid the highest-cost enrollees that risk-bearing plans face, it would also tend to weaken the plans' incentives to control costs commensurately.

Complicating matters further, the incentives to control drug costs faced by entities administering a Medicare drug benefit would not depend solely on how they were paid; the financial incentives that beneficiaries faced would also be a key consideration. Such incentives might include lower beneficiary premiums for joining plans that could deliver the required benefits for a lower overall cost, as well as smaller out-of-pocket payments in plans that were able to negotiate lower prices for the drugs they covered. If plans competed primarily on the basis of the comprehensiveness of the coverage they provided, however, federal expenditures would probably be higher than if plans competed on cost factors. Moreover, to devise a proposal that would require plans to bear insurance risk but not allow beneficiaries' premiums to vary with their choice of plan appears to be difficult.

Although much depends on a proposal's specific design and details, a drug benefit could be structured so that entities bearing some insurance risk would choose to provide it; further, such coverage would probably be available across the country. That conclusion, which stands in contrast to the experience of the Medicare+Choice program, is based in part on the fact that the kind of competing pharmacy networks needed to provide such a drug benefit are already well established nationwide. At the same time, CBO concludes that plans bearing insurance risk would incur additional costs that would not be borne by PBMs that are subject only to performance risk. Whether and to what extent those added costs might offset any reductions in federal costs that accrued from having plan administrators face insurance risk would also depend on the specific provisions of the proposal.

Finally, recent discussions have included the notion of linking drug coverage with reforms of the delivery mechanism for Medicare's benefits. For example, the Bush Administration has put forward a set of principles for Medicare reform that suggests an "integrated" approach combining drug benefits and enrollment in private health plans. The budgetary implications of such an approach are, however, unclear--the Administration estimated that its initiative would cost a total of $400 billion through 2013 but did not submit sufficient details for CBO to make its own estimate. CBO is preparing to estimate the effects of any such proposals and looks forward to working with the Congress if and when such initiatives are introduced as legislation.
 

Conclusion

In conclusion, I would be remiss if I did not emphasize the important trade-offs involved in all of the policies now under consideration. Even when considered in isolation, a Medicare drug benefit might address a number of objectives--but objectives that might be thought desirable in the abstract are often mutually incompatible, necessitating difficult choices. For example, providing extensive drug coverage to all Medicare beneficiaries at a low cost to all parties is not possible; either enrollees' premiums or the government's subsidy costs would be high. If most of the costs were paid through enrollees' premiums to keep federal spending low, some Medicare beneficiaries would be unwilling or unable to participate in the program, particularly if coverage was limited to catastrophic expenses. If, instead, costs were limited by capping the annual benefits paid to each enrollee, the program would fail to protect participants from the impact of catastrophic drug costs. Proposals have taken various approaches to balance those competing objectives.

Looking at the Medicare program as a whole, the choices may be even more stark. If the program continues to operate as it is currently structured, its costs will rise significantly--even in the absence of program expansions such as a prescription drug benefit. In light of that outlook, policymakers may wish to incorporate two features in their approach to Medicare policy: a recognition of the larger economic and budgetary trade-offs, and consideration of the program structure that would best support Medicare's overall objective of providing financing for high-quality medical care for the elderly and disabled.

With regard to economic and budgetary trade-offs, two issues stand out. First, to the extent that the U.S. economy grows at a healthy pace, it will be better able to meet the Medicare population's demands for health care. Put differently, the overall level of national income available in the future constitutes the reservoir from which the resources for both private needs and public programs will be drawn, and the nation must endeavor, in making public policy, to enlarge that reservoir to the greatest degree possible. Second, the potential pressures on the federal budget from Medicare and other sources will necessitate trade-offs with other spending priorities if federal programs are to remain close to their historical fraction of national income.

Alternatively, public policy may steer a course toward devoting a larger fraction of the federal budget and the economy as a whole to Medicare. Even if that occurs, it will be desirable to use those Medicare funds as efficiently as possible--to purchase the highest-value care with each dollar. Medicare beneficiaries (or their families), together with their health care providers, are best positioned to guide the use of additional dollars and to choose services that meet therapeutic demands and match individual tastes. Providing those parties with a broader range of choices and improved information, and ensuring their sensitivity to the cost of those services, should facilitate better decisionmaking. At the same time, an appropriate balance must be struck between providing stronger financial signals to beneficiaries about the cost of their care and providing protection against greater financial exposure--in the program as a whole and in any drug benefit that is added to it.

This concludes my testimony, and I look forward to answering any questions that the Committee may have.