Skip Navigation

Long-Term Budgetary Pressures and Policy Options March 1997

The long-term deficit problem facing the United States could be resolved by a combination of approaches involving reductions in future spending commitments for Social Security, Medicare, and other programs, together with increases in revenues. Options for slowing the growth in future Social Security and Medicare spending are important because those programs are so large and clearly affected by the aging of the U.S. population. Spending for Medicaid has also been growing rapidly and could escalate with the aging of the baby boomers (see Box 5).

The illustrative goal that the Congressional Budget Office used was to prevent spending for Social Security and Medicare from growing more rapidly than the economy when the baby boomers become eligible for both programs, beginning around 2010. As discussed in the preceding chapter, for any path of total federal spending and revenue to be sustainable, the resulting debt must eventually grow no faster than the economy. Holding spending for Social Security and Medicare to a fixed percentage of gross domestic product would go a long way toward putting the federal budget on a sustainable path. If spending for those programs grew no more rapidly than GDP after 2010, the long-term outlook for the federal deficit and for the nation's economy would improve dramatically (see Chapter 3).

Stabilizing the ratio of spending to GDP provides a convenient yardstick, but it is not necessarily an appropriate goal in view of the magnitude of the demographic shift that will occur. People may reasonably differ about what proportion of GDP is appropriately spent on income support and health care for retired and disabled workers, their families, and their survivors. To prevent spending for those programs from exceeding their projected shares of GDP in 2010, spending for Social Security would need to be pared by about 25 percent below its projected level in 2030 and spending for Medicare would need to be cut by over 40 percent below its projected level in 2030. Changes of those magnitudes would not be easy to achieve. Smaller reductions in the growth of spending for those programs could also be used to reduce long-term budgetary pressures and could be combined with changes in other government programs or with tax increases to achieve similar economic benefits.

Through federal policies that have been in effect for many years, U.S. workers have come to expect that, when they retire or become disabled, Social Security will provide them with income that will replace a significant portion of their previous earnings, that Social Security benefits will be available for their survivors, and that Medicare will provide them with access to mainstream medical care. More than 43 million retired or disabled workers, their dependents, and survivors now receive monthly Social Security payments, and about 38 million people have Medicare coverage. Policy-makers will need to weigh the benefits of those programs against the need to make some policy changes-- if not in those programs, then in the rest of government spending or in the taxes needed to finance them.

Social Security and Medicare are generally credited with having substantially improved the lives of the el
 

Reliance on Social Security was especially high among those elderly whose cash income was relatively low. Families with at least one member collecting Social Security benefits who were in the lowest-income quintile of elderly families received almost 90 percent of their income from Social Security. Those in the highest-income quintile of elderly families received only 25 percent of their income from Social Security.

Options that would reduce the growth in spending for Medicare and Social Security can be thought of as interchangeable in the sense that a dollar saved in either program reduces the federal deficit by a dollar. Moreover, because most Medicare enrollees are also Social Security beneficiaries and vice versa, changes in either program generally affect the standard of living of the same people. That is an especially important point to keep in mind when considering a combination of options that would reduce Social Security benefits and increase Medicare premiums or cost sharing by enrollees.

The two programs differ, however, in an important way. Although federal savings resulting from a change in the Social Security program almost certainly translate into lower benefits paid to Social Security recipients, that is not necessarily the case for federal savings achieved by changes in the Medicare program. In particular, changes that would reduce payments to health care providers would reduce providers' income but would not necessarily diminish the standard of living of the enrollees if those payments were used to deliver health care services more efficiently.

Left untreated, the budgetary problem posed by Social Security and Medicare--and the difficulty of resolving it--will become formidable. In 2030, Social Security outlays will equal 6.4 percent of GDP, an increase of 1.7 percentage points over its share in 1996, according to the intermediate projections of the program's trustees in their 1996 annual report. Spending on Medicare, less premiums paid by enrollees, is projected to increase by about 4.7 percentage points to 7.1 percent of GDP over that period, based on the intermediate projections of the Medicare trustees in their 1996 annual report. Under those combined projections, spending for Social Security and Medicare would account for almost 14 percent of GDP in 2030, about double its current share of GDP.

The case for addressing the growth in spending for Social Security and Medicare before the boomers retire rests on at least two grounds. First, delay will only make the necessary actions more severe because the size of the accumulated federal debt will be that much larger. Second, concerns for both equity and efficiency suggest that the commitment to changes in those pro-grams be made well before they are carried out. Entitlement programs for the elderly and the disabled are generally viewed as long-term commitments between the government and the citizenry, and people have based their behavior on current provisions. Deciding soon on any future changes in such programs and making gradual changes in spending and tax policies would give people more time to plan and adjust.

The precedent set by the Congress when it amended the Social Security system in 1983 is instructive. The changes included a substantial cutback in the growth of benefits by raising the normal retirement age. The first workers affected by that change were then only 45 years old--17 years away from eligibility for retirement benefits. By announcing the change so far in advance, the government gave workers the opportunity to take it into account when planning for their retirement.
 
 

Social Security

To curtail the growth in spending for Social Security benefits, a proposal must either reduce the number of beneficiaries or reduce the benefits for which they are eligible. The last can be done by changing the method by which initial benefits are calculated or by reducing the rate at which benefits are subsequently increased.

Most of the discussion in this section focuses on Old-Age and Survivors Insurance (OASI), the part of the Social Security system that provides benefits to retired workers, members of their families, and their survivors. The other part, Disability Insurance (DI), provides benefits to disabled workers under age 65 and their dependents. OASI is by far the larger program: last year, it accounted for almost 90 percent of spend-ing for the two combined (referred to as OASDI). Ben-

Figure 6.
Growth in Social Security Outlays and Number of Beneficiaries, 1975-2070
 

SOURCE: Congressional Budget Office based on intermediate assumptions from the 1996 report of the Board of Trustees of the Federal Old-Age and Survivors Insurance and Disability Insurance trust funds.
 
NOTES: Social Security outlays as a percentage of gross domestic product are presented on a fiscal year basis for 1995 and earlier years; projections to 2070 are presented on a calendar year basis. Data are plotted at five-year intervals.
 
GDP = gross domestic product.

efits for both parts are financed primarily from payroll taxes paid by workers and their employers on earnings covered by the OASDI program. The combined tax rate for 1997 is 12.4 percent of up to $65,400 in covered earnings.

Source and Magnitude of the Problem

The Social Security eligibility and benefit rules have produced a stable spending pattern in recent years in which total spending has grown at about the same pace as the economy. But that relationship will change once
 

The source of the problem is absolutely clear: since the mid-1970s, the ratio of beneficiaries to workers covered by the Social Security system has been about 30 to 100. That ratio is projected to rise to about 50 beneficiaries for every 100 workers by 2030, with the retirement of most baby boomers, and the combination of a relatively low birthrate and longer life expectancy will keep increasing the ratio thereafter. Given the commitments to provide benefits under current law, the increases in the ratio of beneficiaries to workers directly translate into higher outlays as a percentage of GDP.

Major Issues

The Congress will need to plan for the retirement of the baby boomers by deciding what the Social Security system should attempt to accomplish and what legislative changes will be needed to ensure that the system achieves its goals.

The current design of the Social Security system represents a balance between the goal of ensuring an adequate level of benefits to even the poorest beneficiaries and the goal of equitably distributing benefits in the sense that workers who have paid more taxes for Social Security should receive more in benefits, providing a reasonable return on their tax payments. The progressive benefit structure reflects those dual goals. Retired workers with histories of low wages receive benefits that replace a higher percentage of their preretirement earnings than do the benefits of other retired workers. Nonetheless, workers who earned higher wages receive higher benefits. Achieving both goals will become much more difficult when there are fewer workers per beneficiary.
 
Policymakers will need to consider changes in the design of the Social Security system in the light of their potential effects on people's incentives to work and save. For example, lower benefits for retired workers could encourage them to remain in the labor force longer, particularly if the age of earliest eligibility was raised. Reductions in benefits could also encourage workers to save more.⁽¹⁾
 
The 1994-1996 Advisory Council on Social Security, appointed by the Secretary of Health and Human Services, struggled with the issue of how to improve the long-range financial status of the Social Security program for more than two years and failed to reach a consensus among its members. Part of the reason for disagreement was that they held divergent views about how large a role Social Security should play in the future (see Box 6 on page 28).⁽²⁾
 
Much of the debate within the council reflected competing views about the extent to which the government should be responsible for the well-being of workers and their families once they have retired or become disabled. At least two competing views emerged. One envisions keeping the Social Security benefit structure essentially as it is, continuing to provide the largest component of retirees' incomes. The other view envisions a smaller public system in which future workers would rely more heavily on other sources of income when they stopped working, such as private pensions, individual retirement accounts, and other savings.⁽³⁾
 
Specific Benefit Options
 
To keep outlays for Social Security from exceeding their projected rate in 2010 of about 5 percent of GDP, spending must be held to about three-fourths of its projected level under current law in 2030. The specific options considered in the pages that follow were patterned after several that have been proposed in recent years and were selected to illustrate both the strengths and weaknesses of the major approaches as well as trade-offs that the Congress would face in designing a specific policy. The options could be combined with one another or with revenue options.
 
The savings estimates reported are provided by the Social Security Administration's Office of the Actuary and are intended to indicate relative magnitudes of change. They are based on the intermediate economic and demographic assumptions used in the 1996 annual report of the trustees.
 
Reduce Initial Benefits. The most straightforward method of reducing the growth in Social Security spending is to lower the replacement rates in the benefit formula. The immediate effect of that approach would be to reduce benefits going to newly eligible beneficiaries. The full savings of a specified reduction would not be achieved until all of the beneficiaries whose initial benefits had been determined under the previous formula were no longer receiving benefits.
 
Under current law, benefits of retired (and disabled) workers are based on their past earnings, expressed as an average level of earnings over their working lifetime, known as the average indexed monthly earnings (AIME). From that average, a formula is used to calculate a worker's primary insurance amount (PIA), which is then adjusted for a number of factors, such as reductions for early retirement, credits for later retirement, and increases for inflation.
 
A worker's AIME is based on wages in covered employment (up to the taxable maximum), with some adjustments. Earnings on which retired workers and their employers paid Social Security taxes are indexed to compensate for past inflation and real wage growth. To convert the AIME to the PIA, a formula is applied that is progressive in that the PIA is a higher proportion of preretirement earnings for people with low average earnings than for those with higher earnings.
 
Under the formula, Social Security benefits replace 90 percent of the first part of a worker's AIME. How-ever, for subsequent portions of the AIME, the proportion falls--first to 32 percent and finally to 15 percent (see Figure 7). For workers who reached age 62 in 1996, the formula is as follows: a worker's PIA equals 90 percent of the first $437 of the AIME, plus 32 percent of the AIME between $437 and $2,635, plus 15 percent of the AIME over $2,635. The points at which the percentage of the AIME that is replaced by the PIA changes (known as "bend points") are indexed to average annual earnings for the labor force as a whole. Consequently, as wages rise over time, average replacement rates are maintained.
 
In general, workers receive 100 percent of their own PIA in benefits if they first receive benefits at the normal retirement age, which is currently 65. The benefit is reduced if they retire earlier. For example, a worker who retires at age 62 receives a permanent 20 percent reduction. The size of that reduction is intended to be actuarially fair in that the present value of the reduced monthly benefits that average workers could expect to receive at age 62 is similar to the present value of the full monthly benefits they could expect to receive by delaying initial benefits until the normal retirement age. Similarly, workers who delay collecting benefits beyond their normal retirement age receive a delayed retirement credit to compensate them for the reduction in the length of time that they will receive
 

Figure 7.
Primary Insurance Amounts in Relation to Average Indexed Monthly Earnings Under Current Law for Workers Who Turned Age 62 in 1996

SOURCE: Congressional Budget Office.
 
NOTE: For workers in this cohort who retired at age 65 (in 1999), the primary insurance amount would be based on the formula illustrated in this figure, with the amounts increased by the cost-of-living adjustments effective in 1996, 1997, and 1998.

benefits, although that credit will not reach its actu-arially fair level of 8 percent a year for another decade.⁽⁴⁾
 
Workers who had average earnings throughout their careers and retired at age 65 in 1996 were eligible for an annual "retired-worker benefit" of about $10,700, which replaced 43.2 percent of their previous annual earnings. Because the benefit structure is progressive, the replacement rate is inversely related to past earnings. For example, workers who earned 45 percent of average earnings each year would receive about $6,500, replacing about 58 percent of their past earnings. Workers who always earned the maximum taxable amount ($62,700 in 1996) would receive about $15,000, replacing about 24 percent of their past covered earnings.
 
Under current law, workers with average earnings who retire at age 65 after the turn of the century will be eligible for higher (inflation-adjusted) benefits than those paid to today's average earner, but the benefits will replace a smaller percentage of their past earnings. For example, the Social Security Administration projects that workers with average earnings who retire in 2030 will receive about $12,000 (in 1996 dollars), which will replace 36.4 percent of their earnings during the preceding year.⁽⁵⁾ Although that replacement rate is well below the average in recent years, it is similar to the percentage of earnings that was replaced for workers who retired at age 65 in the late 1960s.⁽⁶⁾
 
Most of the projected decline in the replacement rate is caused by the scheduled increase in the normal retirement age, which is to become age 67 for workers born in 1960 or later. Thus, workers who retire in 2030 at age 65 will receive a permanent reduction in their benefits of about 13 percent because of the actuarial reduction for early retirement. If they wait until 67 to retire, their replacement rate will be 41.8 percent, not far below the current rate for workers retiring at age 65.
 
The major advantage of using across-the-board re-ductions in replacement rates as a means of achieving savings is that they would do so in a way that would otherwise preserve the existing benefit structure. If the change in the formula was announced well in advance of the date when it would take effect, workers could try to adjust their retirement and savings plans accordingly. The major disadvantage of that approach is that some people, such as workers who become disabled and eligible for DI, would not be able to change their behavior and would therefore get substantially lower benefits after they stopped working than they would under current law.
 
By way of illustration, consider a specific option that would reduce the benefits of each successive cohort of workers who became eligible for Social Security disability or retired-worker benefits by 1 percent a year, starting in 1998 and ending in 2032. Under that option, workers becoming eligible in 2010 would receive about 88 percent of their benefits under current law, and those becoming eligible in 2032 and thereafter would receive about 70 percent. Workers who had average earnings, became eligible for benefits in 2030, and retired at age 65 would receive annual benefits of roughly $8,600 (in 1996 dollars)--about $2,000 below the amount that similar workers retiring at age 65 receive today.
 
The savings that would be achieved in a specific year would depend on the composition of beneficiaries by year of eligibility. The Social Security actuaries estimate that this option would ultimately achieve a 30 percent reduction in Social Security expenditures, once all beneficiaries were subject to the full reduction in replacement rates. It would achieve a 19 percent reduction in 2030 and a 25 percent reduction in 2040. Larger savings in future years would be achieved, of course, if the replacement rates of newly eligible beneficiaries were reduced further after 2032.
 
A variation of that option (included in one of three sets of options presented by the advisory council) would reduce the replacement rates in only the second and third brackets of the benefit formula. That is, beneficiaries would continue to receive 90 percent of their average earnings up to the first bend point. That variation, designed to help shield workers with histories of relatively low earnings, would save less money unless larger reductions were made in the second and third brackets. The actuaries estimate that such a modifica-

tion would eliminate nearly half of the savings that would be achieved by an across-the-board cut.
 
Raise the Retirement Age. Under current law, the age at which a worker becomes eligible for full retirement benefits is 65, and will gradually increase to 67. Members of Congress and others have recommended that the change to a normal retirement age (NRA) of 67 be accelerated and that the NRA be further increased thereafter. Proponents point out that people at age 65 today live longer than was the case in the early days of the Social Security system, that life expectancy is projected to continue to increase, and that this otherwise favorable development would raise the cost of the program.⁽⁷⁾ Two specific options to raise the retirement age illustrate that approach (see Table 10). The first would speed up the transition to age 67 and then further increase it to keep up with future increases in life expectancy. The NRA of workers who turn age 62 in 2011 would be age 67. Thereafter, the NRA would increase by one month every two years, reflecting projected growth in the ratio of life expectancy at the NRA to potential work years. For example, the NRA would be 68 for workers turning age 62 in 2035 and 69 for workers turning age 62 in 2059. Workers would still be able to begin receiving benefits at age 62, and the amounts would be reduced accordingly. That option is patterned after a proposal included in one of the three sets of options presented by the advisory council.
 
The second option would also accelerate the transition to age 67, but would continue increasing the NRA by two months a year until it reached 70 in 2029. Thereafter, it would raise the NRA from 70 by one month every other year. As with the first option, workers would still be able to begin receiving reduced benefits at age 62.
 
Each option would produce substantial savings, although not nearly enough by itself to achieve the spending targets presented above. In relation to projected spending levels under current law, the first option would reduce outlays by about 3 percent in 2030 and 8 percent in 2070. The second option would reduce outlays by about 8 percent in 2030 and 16 percent in 2070.
 
For most purposes, such an approach to cutting the growth in benefits is equivalent to cutting replacement rates. To arrive at that equivalence, compare the reductions from PIAs that workers who began receiving retired-worker benefits at age 65 would get under current law and under the two options. For example, workers retiring at age 65 in 2038 would have their benefits reduced by about 13 percent under current law, 20 percent under the first option, and more than 30 percent under the second option.

However, the options differ from the approach of directly reducing replacement rates in that the benefits of workers who qualified for Disability Insurance would not be reduced. Thus, workers would have a somewhat stronger incentive to apply for DI benefits in order to receive higher monthly benefits. Under current law, for instance, workers retiring at age 62 in 2011 would receive 75 percent of their PIA; if, instead, they qualified for DI benefits, they would receive 100 percent. Under both of the options for increasing the normal retirement age discussed above, workers retiring at age 62 in 2011 would only receive 70 percent of their PIA, but would still receive 100 percent if they qualified for DI benefits.
 
Finally, some proposals for increasing the normal retirement age would raise the earliest age of eligibility for retired-worker benefits as well. Currently, more than two-thirds of retired-worker beneficiaries choose to begin receiving benefits before age 65. Increasing the earliest age of eligibility would most likely increase the size of the workforce as some workers delayed retirement, thereby adding to the nation's economic output. Moreover, it would help to ensure that once they did retire, workers would have higher benefits because they would not have incurred the actuarial reduction.
 
Opponents of raising the earliest age of eligibility argue that some of the workers who begin receiving benefits at age 62 have little if any choice--for example, because the jobs they held were especially physically demanding or they have become incapacitated. Opponents also contend that many of those early retirees have no pensions or other sources of income.
 
Reduce the Cost-of-Living Adjustments. Each year, monthly benefits are adjusted by the increase in the consumer price index (CPI). To give an example, the 2.9 percent cost-of-living adjustment (COLA) effective for December 1996 was based on the increase in the CPI between the third quarter of 1995 and the third quarter of 1996. The CPI for urban wage earners and clerical workers is used for that calculation. The basic benefit amount is indexed by the increase in the CPI, beginning when a worker becomes eligible for Social Security benefits. For retired-worker benefits, indexing starts at age 62.
 
An additional or alternative way of reducing the growth in Social Security benefits is to reduce the automatic COLA. Instead of providing an annual COLA equal to the increase in the CPI, the law could be changed to provide a COLA equal to the CPI minus a specified number of percentage points. To illustrate that approach, Social Security actuaries estimated two specific options. The first would determine the COLA based on the increase in the CPI less 2.5 percentage points for 1998 and thereafter. The second would base the COLA on the increase in the CPI less 1 percentage point. To reduce outlays by 25 percent in 2030 (and beyond) solely by means of an across-the-board permanent reduction in the COLA would require that the steeper cut in the COLA be made. The second option would achieve less than half of the savings.
 
Reducing the automatic COLA for Social Security benefits has been widely discussed as a way of achieving considerable savings. Eliminating the COLA for one year or limiting it to less than the CPI could quickly produce large savings by exacting small reductions in benefits from a large number of people.⁽⁸⁾
 
Many analysts feel that the CPI overstates increases in the cost of living, although the magnitude of the overstatement and what should be done about it are subject to much debate. The Advisory Commission to Study the Consumer Price Index (also known as the Boskin Commission) recently estimated the size of the upward bias to be about 1 percentage point a year.⁽⁹⁾ If that is the case, then Social Security beneficiaries have been receiving increases in benefits beyond what was necessary to keep up with inflation. If the CPI overstates increases in the cost of living for beneficiaries, then the COLA could be reduced by a commensurate amount without lowering real benefits to beneficiaries below what they received when they became eligible for the program.
 
Compared with an equivalent across-the-board reduction in replacement rates (or an equivalent increase in the normal retirement age), the people whose benefits would be lowered most by reducing COLAs would be the oldest beneficiaries and those who initially became eligible for Social Security on the basis of disability. The option could be modified to reduce the COLAs only of beneficiaries whose benefits or incomes were above specified levels, but doing so would reduce the savings. (Some beneficiaries with low incomes and few assets would receive Supplemental Security Income (SSI) benefits, which would offset some or all of the reduction in their Social Security benefits; the increased spending for SSI would help those beneficiaries, but it would also reduce the budgetary savings from this option by a small amount.)
 
The impact of even a relatively small reduction in COLAs would be quite large for future older beneficiaries whose benefits would reflect the cumulative effects of a series of smaller COLAs. For example, if benefits were adjusted by 1 percentage point less than the CPI each year, retired workers (or their survivors) at age 74 would incur an 11 percent reduction in benefits, compared with the amount they would have received under current law; workers at age 84 would get a 19 percent reduction; and workers at age 94 would get a 27 percent reduction.
 
Whether or not the real value of the Social Security benefits received by older beneficiaries would then be below what it was when they first became eligible for benefits, their benefits would fall relative to those of new beneficiaries. That decline would occur because initial benefits would continue to be based on a formula in which past earnings are indexed to compensate for growth in nominal wages, which is the sum of inflation and real wage growth. Under current law, each new group of beneficiaries who begins receiving benefits at the normal retirement age receives a slightly higher av-erage benefit than the group who became eligible the previous year, reflecting the increase in real wages. If COLAs were reduced by 1 percentage point, then the gap between consecutive age groups would widen accordingly.
 
Other Options. Carrying out any of the options presented above would eventually reduce the amount of Social Security benefits (in relation to current law) for the majority of beneficiaries. Other approaches that have received attention in recent years would achieve savings by reducing or eliminating benefits for specific groups of beneficiaries. Much deeper reductions for those beneficiaries who were affected, would, of course, be required to achieve comparable savings.
 
In some cases, the number of beneficiaries affected would be too small to have much impact on total spending, even if their benefits were eliminated. For example, lowering the benefit to spouses from one-half to one-third of the retired worker's PIA would reduce Social Security outlays by less than 2 percent because most spouses would be eligible for benefits as retired workers anyway. Combining several options affecting specific groups could produce more significant savings.
 
Another approach to reducing expenditures for Social Security (as well as for other programs) is to reduce or eliminate benefits going to people in middle- and upper-income families, although such an approach could create a disincentive for families to save or to earn other income. In principle, Social Security benefits could be cut by any desired percentage by reducing benefits as beneficiaries' incomes rose, denying benefits to people with incomes above specified thresholds, or increasing the taxes on benefits. Specific options for doing so are presented in a separate CBO report.⁽¹⁰⁾ One option described in that report would pare Social Security and other entitlement benefits as the total family income of the beneficiaries rose above $40,000. That option, proposed by the Concord Coalition, would reduce projected spending for Social Security benefits by about 7 percent in 2002. Making Social Security benefits fully subject to individual income taxes would increase revenues by a similar amount.
 
Conclusions About Social Security

Preventing outlays for Social Security from becoming a larger share of national income in the face of an aging population would require substantial cutbacks in the commitments that have been made under current law. Two options discussed in this chapter and illustrated in Figure 8 suggest how large the reductions would need to be. An across-the-board cut in replacement rates would ultimately require reducing benefits by nearly 30 percent from the amounts provided under current law. To achieve similar savings through a cut in cost-of- living adjustments would require that benefits be increased each year by about 2.5 percentage points less than the increase in the CPI. Each of those options would leave beneficiaries, as a group, much worse off. The last option would leave initial benefits untouched, but would have enormous effects on the benefits of very elderly beneficiaries and those who began receiving benefits at an early age because of disabilities.
 
Smaller reductions in COLAs and gradual increases in the normal retirement age could be used separately to reduce the growth in benefits by smaller amounts or as a part of a larger package. One of the options described in this chapter--setting annual cost-of-living adjustments 1 percentage point below the increase in the CPI --would maintain Social Security outlays at about 5.8 percent of GDP, well below the 6.6 percent projected by 2070 under current law in the 1996 trustees' report. The option to raise the normal retirement age to 70 by 2029 and to 71 by 2053 would reduce projected outlays to about 5.5 percent.
 
Combining the smaller COLA cut with an increase in the normal retirement age would keep spending for Social Security from increasing much above its current percentage of GDP. But the combined effect on people affected by both cuts could be quite large. For those beneficiaries, the two seemingly modest reductions would add up to a large reduction in their benefits.

Figure 8.
Illustrative Options for Reducing Growth in Social Security Outlays

SOURCE: Congressional Budget Office based on estimates provided by the Social Security Administration, Office of the Actuary, December 5, 1996.
 
NOTES: These estimates are based on the intermediate assumptions used in the 1996 report of the Board of Trustees of the Federal Old-Age and Survivors Insurance and Disability Insurance Trust Funds. Data are plotted at five-year intervals.
 
CPI = consumer price index; GDP = gross domestic product.

Medicare
 
Medicare provides federal health insurance for 38 million people who are aged, disabled, or have end-stage renal disease. Part A of Medicare, or Hospital Insurance, covers inpatient services provided by hospitals, as well as skilled nursing, home health, and hospice care. Part B, or Supplementary Medical Insurance, covers services provided by physicians, limited-license practitioners (such as chiropractors and podiatrists), hospital outpatient departments, and suppliers of medical equipment.
 
Everyone who is eligible for Social Security benefits on the basis of age or disability is ultimately eligible for Medicare as well, although Medicare eligibility is delayed until age 65 for early retirees and by two years for disability beneficiaries. In addition, people who are 65 or older and not eligible for Medicare on the basis of their (or their spouse's) previous work history may enroll by paying the HI and SMI premiums.
 
Hospital Insurance benefits are financed primarily from current workers' payroll taxes, which are deposited in the HI trust fund. The actuarially fair HI premiums paid by the small proportion of aged beneficiaries who are not eligible on the basis of work history compose less than 1 percent of HI trust fund receipts. Since 1994, a portion of income taxes paid on Social Security benefits have also been credited to the HI trust fund, accounting for less than 4 percent of trust fund receipts. HI trust fund receipts were less than benefits paid in 1995 and 1996, and that imbalance will increase in later years under current law.
 
Supplementary Medical Insurance benefits are financed primarily from general revenues, although beneficiaries pay a premium to cover some of the costs. Under current law, the SMI premium is set to cover 25 percent of the expected average cost of benefits for aged enrollees each year. For 1999 and later years, current law will limit annual percentage increases in SMI premiums to no more than the cost-of-living adjustment made to Social Security benefits each year. Because health care costs per enrollee are expected to grow more rapidly than the cost of living, the share of SMI costs financed by the premiums of beneficiaries will start to fall after 1998, and an increasing portion of SMI trust fund receipts will come from general revenues.
 
Rapid increases in Medicare spending have been a concern almost from the program's inception (see Table 11). Spending has grown rapidly from the beginning--as a share of both gross domestic product and the federal budget--but the baby-boomers' retirement, beginning early in the next century, will greatly accelerate that trend unless substantial changes are made in the program.

In their 1996 report, Medicare's trustees indicated that the HI trust fund is not adequately funded even for the short term before the effects of the baby boom will be felt. In fact, according to the latest projections, the HI trust fund will be exhausted in 2001 under current law.⁽¹¹⁾ Depletion of the HI trust fund could be avoided, however, by transferring general revenues to it as necessary, just as is now done for the SMI trust fund. The more fundamental problem is that the expected rate of growth in Medicare spending is unsustainable over the long term, given the slower rate of growth expected for GDP.
 
This analysis uses the long-term projections presented by Medicare's board of trustees in their 1996 report. Those projections are not identical to the ones by CBO presented in Chapter I because CBO uses different economic assumptions. However, when expressed as a share of GDP, CBO's projections are similar to those of the trustees. Throughout this analysis, a distinction is made between federal spending for Medicare and total Medicare costs, which includes cost-sharing expenses of beneficiaries as well.
 
Sources and Magnitude of the Problem
 
Rapid growth in Medicare spending in relation to GDP is the result of two main factors. One is growth in the number of beneficiaries, which currently accounts for about one-sixth of the growth in spending. That growth will become more important after 2010, when the first of the baby-boom population will be eligible on the basis of age. Between 2010 and 2030, the rate of growth in enrollment is expected to average about 2.4 percent a year, whereas average growth from 1995 to 2010 will be about 1.5 percent a year. Medicare enrollment is expected to increase from about 14 percent of the population in 1996 to 22 percent in 2030 and to 25 percent by 2070. The second and more important factor is growth in costs per beneficiary, which has been substantially higher than growth in per capita income in the past and which is expected to continue at rapid rates. The first factor--growth in the number of benefi-ciaries--affects both the Social Security and Medicare programs, but only Medicare is affected by the second. For that reason, fiscal problems are more severe for the Medicare program than for the Social Security program.
 
In 1996, Medicare spending was about 2.7 percent of GDP, and spending net of premiums paid by enrollees was 2.4 percent of GDP (see Table 12). HI trust
 

retirement under Social Security (at age 62) might have difficulty getting insurance coverage. One effect of such an approach would be to shift costs now paid by Medicare to employers who offer health insurance to their retirees. Another effect might be to increase the number of applications for disability from the affected population, thereby reducing the savings that Medicare might otherwise realize. The last effect would probably be small and is not estimated here.
 
The option would also affect federal and state spending for Medicaid because about 15 percent of Medicare enrollees are eligible for Medicaid benefits as well. If Medicaid's age of eligibility for the aged category was increased in tandem with Medicare's, then spending for Medicaid would fall because the affected age group would lose eligibility for Medicaid at the same time that it lost eligibility for Medicare--although some of the affected people might regain eligibility by qualifying as disabled or medically needy. If Medicaid's age of eligibility was unchanged, there would be two offsetting effects on Medicaid spending. For those Medicare beneficiaries who are dually eligible for full Medicaid benefits, Medicaid spending would increase as Medicare withdrew its support for the affected age group. But Medicare beneficiaries who are eligible only for qualified Medicaid benefits (payment of Medicare's cost-sharing and premium requirements) would lose their eligibility for Medicaid along with their eligibility for Medicare, thereby reducing Medicaid spend-ing. Since the direction and the magnitude of the change in spending for Medicaid is uncertain, that effect is not estimated.
 
Collect More in Premiums or Taxes from Medicare Enrollees. Premiums paid by Medicare enrollees now cover only about 10 percent of the average benefit paid by Medicare through the HI and SMI programs--a share that is expected to drop after 1998 under current law. If, instead, collections from beneficiaries were gradually increased to cover 50 percent of Medicare's HI and SMI costs by 2010, net spending would not exceed 4.1 percent of GDP until 2060 (see Table 14).
 
Higher collections could be achieved by raising premiums for all enrollees, regardless of their circumstances. But such a rise could impose financial hardship on lower-income enrollees who are not eligible for Medicaid, and it would increase Medicaid costs for Medicare enrollees who were also receiving Medicaid benefits.
 
One alternative would vary the amounts collected from enrollees on the basis of their financial resources. For example, the current flat premium might be replaced with a sliding-scale premium that would collect an average of 50 percent of Medicare's costs, but the value of which would vary directly with enrollees' in

come. It might be set at zero or nominal amounts for enrollees with the lowest income, at 100 percent of Medicare's insurance value for those with income above a certain high threshold, and at intermediate amounts for middle-income enrollees. That approach would collect larger amounts from enrollees who could afford to pay more and could eliminate premium costs for the enrollees with the lowest income. Hence, it would incorporate into Medicare's structure part of the subsidy for low-income enrollees that Medicaid now provides. However, it would also increase implicit marginal tax rates for Medicare enrollees.
 
Such an approach would keep net spending within the limits specified, but only by shifting more costs to enrollees and only if growth in health care costs slowed after 2005, as assumed by the trustees. It would do little or nothing to induce slower growth, however. The premiums that Medicare enrollees now pay are an average of about 3 percent of their per capita income. Under this approach, Medicare's premiums would consume 25 percent to 30 percent of enrollees' income each year after 2015. Those costs for Medicare enrollees could be reduced only by spreading them over a larger (non-Medicare) population or slowing the growth in health care costs by more than the trustees assumed.
 
The option would increase spending for Medicaid because Medicaid pays the Medicare premium for some low-income Medicare beneficiaries. The resulting increase in federal spending for Medicaid (which is about 57 percent of total Medicaid spending) would raise net federal spending by only 0.04 percent of GDP in 2000; that share would climb to 0.36 percent of GDP by 2070.
 
Slow the Growth in Medicare's Spending per Enrollee. The growth in Medicare spending might be slowed, at least temporarily, by any of three general approaches. One that has been used extensively in the past decade would reduce the rates paid to Medicare providers. Another--and one that has not been used much--would increase the cost-sharing amounts that beneficiaries must pay. A third, which was the focal point of some Medicare proposals in the last Congress, would restructure the Medicare market to give patients and providers greater incentives to make cost-effective health care choices.
 
Reducing payment rates is the first approach. Rates for Medicare's fee-for-service providers normally increase each year in line with indexes of costs developed by the Health Care Financing Administration. If the Congress elects to update rates by less than increases in the relevant cost indexes, payment rates would be lower than those that Medicare would have paid if the Congress had not acted. Typically, however, not all of the potential savings to Medicare from lower payment rates are realized because providers are able to offset part of their potential loss in receipts from Medicare by increasing the volume of services for which they bill. Nevertheless, reducing payment rates can lower both federal and total health care costs for Medicare because providers are generally unable to offset all of their potential loss in receipts, at least from Medicare patients alone. If lower payment rates cut Medicare's fee-for-service costs, payment rates to HMOs would also be reduced under current law because those rates are based on Medicare spending per enrollee in the fee-for-service sector.
 
One undesirable aspect of cutting payment rates is that some providers may try to maintain revenues by shifting costs to other payers, although their ability to do so is lessening as private insurers adopt more aggressive rate-setting policies of their own. The access to care for Medicare enrollees could be threatened if the program's rates fell too far below those paid by other insurers. However, few people seem to have had trouble obtaining care so far, even though current estimates indicate that Medicare pays only 70 percent to 80 percent of the average rates that private insurers pay to hospitals and physicians.
 
Another undesirable aspect is that regulatory price setting often results in inappropriate, and therefore inefficient, prices--either lower or higher than necessary to generate adequate response from providers. Problems with access to care for beneficiaries would soon alert Medicare if its payment rates were too low, but there is no comparable mechanism to alert Medicare when its payment rates are higher than necessary. In some geographic areas and for some services (durable medical equipment, for example), Medicare's current payment rates may be higher than market-based rates. Demonstration studies are planned to assess the feasibility of and potential savings from using competitive bidding to set some of Medicare's payment rates.
 
Increasing cost-sharing requirements would reduce federal spending for Medicare, but the reduction would be achieved by shifting costs to enrollees without necessarily affecting total costs. Although cost-sharing requirements can, in principle, make enrollees more prudent consumers of health care, that effect is in fact weak in the Medicare program because most enrollees have supplementary coverage.
 
About 15 percent of Medicare beneficiaries also receive Medicaid benefits, which pay all of their cost-sharing liabilities under Medicare. Another 70 percent have medigap, an HMO supplement, or non-HMO employment-based coverage. Medigap plans and HMOs typically cover all or most of Medicare's cost-sharing requirements. The only common exclusion (affecting about 40 percent of people with medigap coverage) is the $100 deductible for Supplementary Medical Insurance. Those people who have employment-based plans generally pay the cost-sharing requirements of their private plan or Medicare, whichever is lower. Except for the deductible amount, which is generally higher than $100, employment-based plans typically have lower cost-sharing requirements than does Medicare.
 
Thus, only an increase in the SMI deductible amount would be likely to reduce use of services by people who have private insurance supplements. No change in Medicare's cost-sharing requirements would affect the use of services by those who also have Medicaid benefits. But any increase in cost-sharing requirements would reduce use of services by the 15 percent of enrollees who have no supplement.
 
To illustrate the way in which supplementary coverage negates the effects of Medicare's cost-sharing requirements on use of services, consider the following example: increasing the SMI deductible amount to $1,000 a year would reduce federal spending for Medicare by an estimated 9 percent for 1997, but total costs would drop by less than 1 percent given current patterns of supplementary coverage. Thus, most of the effect is a shift of costs from Medicare to enrollees, with very little reduction in use of services. By contrast, if current requirements for medigap plans were changed so that they could only cap the liabilities of enrollees for cost sharing under Medicare at $1,000 a year, rather than covering them all, both federal and total costs for Medicare would fall by about 3 percent, caused entirely by a reduction in the use of services.⁽¹³⁾
 
Restructuring the Medicare market would, in one approach, involve setting up a system of competing health care plans in which Medicare's fee-for-service sector might be just one of possibly several fee-for-service options. In that restructured market, all plans would offer at least a specified basic-benefit package. Plans could offer optional supplements to their basic package, but no plan could offer supplements to another plan's basic package. Without that restriction, plans could offer supplemental coverage only, as medigap plans do now. But medigap insurers do not bear the full costs of the coverage they offer. Most of the costs of the additional services that people with medigap coverage use are actually imposed on Medicare--the insurer providing coverage for the basic-benefit package. By permitting supplemental coverage only when it is linked to a basic-benefit package offered by the same insurer, all of the costs generated by medigap plans under current law would be internalized --that is, borne by the medigap insurer.
 
Thus, if insurers that were currently offering medigap plans wanted to continue to serve the Medicare market, they would have to offer full coverage for Medicare's basic package along with their supplemental benefits on the same basis as all other plans serving the Medicare market. Under current law, the constraints imposed on HMOs and medigap plans differ significantly, though both supplement the basic Medicare benefit package. For example, HMOs must offer community-rated premiums to all Medicare enrollees and may impose no exclusions on coverage for pre- existing conditions. Medigap plans may rate their premiums on the basis of age, base premiums on risk status for those who enroll after the first six months of Medicare eligibility, and impose a six-month exclusion on coverage for preexisting conditions.

Enrollees could choose the benefit and premium package they preferred from the menu of plans available in their area during an annual open-enrollment period. Medicare would contribute a fixed amount per enrollee toward the premiums charged by plans. Actual payments from Medicare to the plans would have to be adjusted for risk to discourage competition based on the characteristics of enrollees rather than price and quality. From the enrollees' perspective, however, Medicare's contribution toward their premiums would be uniform as long as plans were required to set community-rated premiums, as they are under current law.
 
Enrollees would be responsible for any excess premium amounts or would receive rebates for plans costing less than Medicare's contribution. Thus, Medicare's method of contribution to the costs of their health plan would give enrollees financial incentives to be prudent purchasers of plans. Moreover, the comparative information provided during the open-enrollment period would enable them to select the lowest-cost plan that would meet their needs. Because plans would be at risk for any costs above their predetermined premium collections, they would have financial incentives to limit unnecessary services, either through provider controls or cost-sharing requirements on beneficiaries.
 
Medicare's contribution could be set in one of two ways: to equal the premium charged by the lowest-cost basic-benefit plan in each area, or to equal some value set independently of the actual costs of the plans. In the former case, Medicare would continue to guarantee a defined benefit, and taxpayers would bear the financial risk if health care costs increased more rapidly than expected. In the latter case, Medicare would offer only a defined contribution, with no assurance that the contribution would be sufficient to purchase the basic-benefit package.
 
Medicare could be certain of controlling its costs only under the defined contribution approach, which would shift the financial risks from higher growth in health care costs to plans, and ultimately to enrollees through premiums. Either approach would make both enrollees and providers more prudent in their use of health care services. Supplemental premiums would be higher for Medicare beneficiaries who chose to remain in loosely managed plans compared with those in tightly managed plans, thereby accelerating the movement of enrollees to HMOs that is already occurring.
 
Medicare's fee-for-service plan would have to become more efficient to keep its supplemental premium at a competitive level. Furthermore, gains in efficiency would have to be large enough to offset any loss in the substantial leverage that Medicare currently has in setting providers' fees. Medicare's leverage would weaken as its fee-for-service enrollment fell as a share of the patient population in an area.
 
Given a coordinated open-enrollment period and the new pricing system, competition among plans for enrollment would intensify. If methods for risk- adjusting payments among plans were adequate, competition would be focused on providing services more efficiently rather than on enrolling low-cost beneficiaries. Consequently, growth in both federal and total costs per enrollee might be slowed compared with growth under current law.
 
For example, Medicare's defined contribution could be set to equal net spending per enrollee in 2000 (adjusted for geographic differences in costs), and increased by specified percentages in later years that might be lower than the growth in health care costs. A delay of a few years would probably be necessary to give Medicare time to transform its fee-for-service sector into a health care plan capable of competing with other risk-based plans serving Medicare enrollees. Some lead time would also be necessary before a coordinated open-enrollment period could be put into effect.
 
The savings potential of this approach could be increased gradually. In this illustrative option, federal savings through 2000 would be generated by keeping the SMI premium at 25 percent of SMI costs, rather than letting it drop after 1998 as under current law. The amount of Medicare's contribution in 2000 to the health plan premiums of enrollees would then be increased by 6.0 percent a year through 2005, 5.0 percent a year through 2010, and 4.2 percent a year thereafter.
 
Although the effects of this defined contribution approach on federal costs can be predicted with some certainty, its effects on total costs for the basic-benefit package--and therefore on the costs that enrollees would bear--is uncertain. If the average rate of growth in total costs per enrollee slowed only to the rate assumed by the trustees in their long-term projections, the premiums of enrollees as a percentage of their average income would increase from 3 percent in 1996 to about 37 percent in 2070 (see Table 15). However, some plans in each area would probably endeavor to offer the basic-benefit package for premiums equal to Medicare's defined contribution so that there would be no supplemental premium to collect. Enrollees in those plans would be liable only to Medicare for the basic premium equal to 25 percent of SMI costs, or less than 10 percent of total Medicare costs. In that case, the premiums of enrollees would fall over time as a share of income to about 2 percent.

If through increased efficiency some plans were able to reduce the rate of growth in total costs per enrollee to the 4.2 percent annual increase in Medicare's defined contribution, those plans would probably dominate the Medicare market. If improvements in efficiency did not cut costs sufficiently, so that low-cost plans had to restrict access or reduce the quality of their services, a two-tier Medicare market would probably develop. Lower-income enrollees would tend to choose the low-cost plans in which access and quality were poor, whereas higher-income enrollees would be more likely to opt for more expensive plans with less severe restrictions.
 
The effects of this option on spending for Medicaid would depend on the extent to which it slowed the growth in total Medicare costs and on whether Medicaid limited the choice of plans for dually eligible Medicare beneficiaries. If growth slowed to match the growth in the defined contribution, then spending for Medicaid would fall because the dollar value of Medicare's cost-sharing requirements would drop substan-

tially, while premiums would increase only slightly compared with current law. If the growth in total Medicare costs exceeded growth in the defined contribution, spending for Medicaid would probably increase as a result of higher premium costs--assuming that dually eligible beneficiaries were free to choose any plan they wanted. If, instead, Medicaid assigned dually eligible beneficiaries to the lowest-cost plans, then spending for Medicaid would probably fall. No estimate of the effects on Medicaid spending were made because of that uncertainty, although it appears that spending would be more likely to fall than to increase.
 
Conclusions About Medicare
 
The effects of the three general approaches discussed above are compared below under the assumption that average annual growth in Medicare spending per en-rollee will gradually slow between 2005 and 2020 as assumed by Medicare's trustees in their 1996 report. Keep in mind, however, that only the third approach would put into effect policies specifically intended to achieve slower growth in total costs per enrollee. The first approach would reduce federal spending by reducing enrollment, with no significant effect on growth in costs per enrollee. The second approach would reduce net federal spending, but not total costs, by increasing premiums paid by enrollees without fundamentally changing the Medicare market.
 
The first approach would reduce total enrollment in Medicare by delaying the age of eligibility to 70, phased in from 2003 through 2032. Compared with current law, that change would reduce Medicare net spending by about 13 percent in 2030 and by 15 percent in 2070 (see Figure 9). Nevertheless, net spending would exceed the target--4.1 percent of GDP--every year after 2010 by generally increasing amounts. The premiums of enrollees would be unaffected under current law because after 1998 they would be indexed to the cost-of-living adjustment for Social Security benefits (see Figure 10).
 
The second approach would increase enrollees' premiums to cover 50 percent of total Medicare spending by 2010, thereby reducing net Medicare spending by nearly 50 percent every year thereafter. There would be little or no effect, however, on growth in total costs for Medicare. Although enrollees' premiums are currently only 3 percent of their average income, under this plan premiums would rise to nearly 30 percent of the average income of enrollees by 2030, remaining around that level thereafter. Unless the premium was related to income, it would equal or exceed income for low-income enrollees not receiving Medicaid benefits.

SOURCE: Congressional Budget Office based on the Medicare trustees' reports for 1996.
 
NOTES: GDP = gross domestic product. Data are plotted at five-year intervals.

The third approach would restructure the Medicare market, making its fee-for-service sector one of a number of competing plans serving enrollees. Enrollees would receive a fixed federal contribution toward the premium costs of the plan they selected and would pay any excess premium costs out of pocket. Medicare's defined contribution would be set equal to net spending per enrollee in 2000, increased by 6.0 percent a year through 2005, 5.0 percent a year through 2010, and 4.2 percent a year thereafter. That plan would establish control over federal spending for Medicare on a per-enrollee basis and would keep net federal spending for Medicare at or below 4.1 percent of GDP. Compared with current law, net Medicare spending would be reduced by 42 percent in 2030 and by 62 percent in 2070. Although the federal subsidy per enrollee would be smaller than it would be under current law, competition among plans and providers could spur efficiency and increase real health benefits for each dollar spent.
 
The effect of the third approach on enrollees is uncertain, however. If the incentives that the approach would generate for more cost-conscious behavior reduced annual growth in total costs per enrollee only to the rate assumed by Medicare's trustees for their long-term projections, premiums paid by enrollees would steadily increase, reaching 24 percent of their average income by 2030 and 37 percent by 2070. If, instead, growth in costs per enrollee slowed to match annual growth in the federal defined contribution (4.2 percent), premiums would be only 2.2 percent of average income in 2070.
 
In practice, the effects of the third approach may differ among various groups of enrollees. Some basic plans might keep their costs low enough to avoid having to charge a supplemental premium, but the access and quality of services available in those plans might limit their appeal primarily to low-income enrollees. Higher-income enrollees might gravitate instead to plans that charged supplemental premiums and provided better access and quality.
 
The approaches discussed above are not necessarily mutually exclusive. For example, by both delaying the age of eligibility and introducing a defined federal contribution, growth in the federal contribution might be set somewhat higher than 4.2 percent a year after 2010, while still keeping net Medicare spending at or below 4.1 percent of GDP. The one certainty is that Medicare will come to consume a significant share of GDP unless major changes are made in the program.

Figure 10.
Premiums as a Percentage of Enrollee Income Under Alternative Options
 

SOURCE: Congressional Budget Office based on the Medicare trustees' reports for 1996.

NOTE: Data are plotted at five-year intervals.

1. Much has been written about the effect of Social Security on labor supply and private savings and on how much changes in Social Security provisions might alter people's decisions about when to retire and how much to save. This literature is reviewed in Michael D. Hurd, "Research on the Elderly: Economic Status, Retirement, and Consumption Saving," Journal of Economic Literature, vol. 28 (June 1990), pp. 565-637; see also 1994-1996 Advisory Council on Social Security, "Report of the Technical Panel on Trends and Issues in Retirement Savings," Report of the 1994-1996 Advisory Council on Social Security, vol. II (January 1997).

2. Until recently, the Social Security Act required that an advisory council be established every four years to review the status of the Social Security and Medicare trust funds and their relationship to their long-term commitments. That requirement ended when the Social Security Administration became an independent agency.

3. For a fuller discussion of the council's separate views, as well as a comprehensive survey of options for reducing the actuarial imbalance in the Social Security system and the presentation of a framework for assessment, see Report of the 1994-1996 Advisory Council on Social Security, vols. I and II (January 1997).

4. Starting with beneficiaries born in 1943, each year delayed beyond the normal retirement age (which will be age 66 for that cohort) will add 8 percent to their retired-worker benefits. The delayed retirement credit for workers reaching the normal retirement age in 1997 (age 65) is only 5 percent.

5. Board of Trustees, Federal Old-Age and Survivors and Disability Insurance Trust Funds, 1996 Annual Report (June 5, 1996), p. 184.

6. Robert J. Myers, Social Security, 4th ed. (Philadelphia: Pension Research Council and University of Pennsylvania Press, 1993), p. 363.

7. Social Security Board of Trustees, 1996 Annual Report (1996). The intermediate assumptions in the report are that in 2030 men who reach age 65 will live an additional 16.9 years and that women will live an additional 20.5 years. In 1996, the life expectancy of men age 65 was 15.4 years and that of women was 19.2 years. In 1940, the life expectancies of men and women age 65 were only 11.9 years and 13.4 years.

8. Congressional Budget Office, Reducing the Deficit: Spending and Revenue Options (March 1997), pp. 284-287.

9. Advisory Commission to Study the Consumer Price Index, Final Report to the Senate Finance Committee, Toward a More Accurate Measure of the Cost of Living (December 4, 1996).

10. Congressional Budget Office, Reducing the Deficit: Spending and Revenue Options, pp. 288-291.

11. Social Security and Medicare Boards of Trustees, Status of the Social Security and Medicare Programs: A Summary of the 1996 Annual Reports (June 1996), p. 6 (using the intermediate assumptions).

12. Although Medicare's trust funds also generate interest receipts, those are not included because they are intragovernmental transfers that do not affect the deficit.

13. Medigap coverage increases enrollees' use of services by an estimated 24 percent. See S. Christensen and others, "Acute Health Care Costs for the Aged Medicare Population: Overview and Policy Options," The Milbank Quarterly, vol. 65, no. 3 (1987). See also Physician Payment Review Commission, Annual Report to Congress (1996), Chapter 16.

---