Mr. Chairman, Senator Baucus, and Members of the Committee, the Congressional Budget Office (CBO) appreciates the opportunity to appear before you to discuss the budgetary and distributional implications of various options for slowing the growth of Social Security benefits.

As you know, Social Security is the single largest federal program. In 2004, the Social Security system took in $569 billion in tax revenue and paid out $493 billion in benefits. The program provided benefits to more than 47 million people--about two-thirds of them retired workers and the rest disabled workers, survivors of deceased workers, workers' spouses, and minor children.

Although today the program takes in more revenue than it spends, that situation will not continue once large numbers of baby boomers begin claiming retirement benefits. In coming years, the Social Security system will face mounting financial pressures as its outlays start to grow much faster than its revenue. CBO projects that scheduled Social Security outlays (those implied by the current benefit formula) will rise from 4.3 percent of gross domestic product (GDP) this year to 6.5 percent in 2050. Revenue, however, is scheduled to remain at 4.9 percent of GDP.

That financial outlook has prompted discussion of various ways to make the Social Security system solvent. My testimony today focuses on the spending side of the program, as requested by the Chairman. I will discuss several options for curtailing the growth of outlays and compare their effects on the system's finances and on different types of beneficiaries. CBO has also prepared a more comprehensive menu of options for changing scheduled benefits or revenue, which is included as an attachment at the end of this statement.
 

The Financial Outlook for Social Security

The next decade will see the beginning of a significant, long-lasting shift in the age profile of the U.S. population. Over the next 50 years, the number of people ages 65 and older will more than double, while the number of adults under age 65 will grow by less than 20 percent. That shift reflects demographic trends that have been evident for many years and that are expected to continue, such as the aging of the baby-boom generation, increases in life spans, and a fertility rate below that needed to replace the population.

Those trends imply that the number of workers per Social Security beneficiary will drop significantly, from 3.3 this year to 2.0 in 2050. Because Social Security depends on revenue from current workers to finance benefits, that demographic shift will have a profound impact on the system's finances. Without changes in tax or spending policies, expenditures will start to rise faster than revenue, pushing up federal debt and slowing the growth of the economy.

As requested by the Chairman, my testimony examines the outlook for Social Security using the same long-term economic and demographic assumptions used in the March 2005 report of the Social Security trustees. The differences between the projections of annual Social Security spending and revenue presented here and the ones that CBO released in March 2005 are small and occur largely because CBO's assumptions about future wage growth and interest rates are slightly higher than the trustees'.

The Financing Perspective

In 2009, the Social Security surplus--the amount by which the program's dedicated revenue in a year exceeds the benefits paid in that year--will start to diminish. In 2019, that surplus will disappear, and outlays for benefits will begin to exceed the system's annual revenue, CBO projects using the trustees' long-term economic assumptions (see Figure 1). To pay full benefits, the Social Security system will eventually have to rely on interest on government bonds held in its trust funds--and ultimately, on the redemption of those bonds. In the absence of other changes, bonds can continue to be redeemed until the trust funds are exhausted, which will occur in 2044, CBO projects. But where will the Treasury find the money to pay for the bonds? Will policymakers cut back other spending in the budget? Will they raise taxes? Or will they borrow more?
 

Figure 1.

Social Security Revenue and Outlays as a Share of GDP

(Percentage of GDP)

Source: Congressional Budget Office.

Notes: This figure is based on a simulation from CBO's long-term model using the Social Security trustees' 2005 intermediate demographic assumptions and long-term economic assumptions and CBO's January 2005 10-year economic assumptions.

Revenue includes payroll taxes and income taxes on benefits but not interest credited to the Social Security trust funds; outlays include Social Security benefits and administrative costs. Under current law, outlays will begin to exceed revenue in 2019; starting in 2045, the program will no longer be able to pay the full amount of scheduled benefits.

Once the trust funds are exhausted, the Social Security Administration will no longer have the legal authority to pay full benefits. As a result, it will have to reduce payments to beneficiaries to match the amount of revenue coming into the system each year. Although the exact size of that reduction is uncertain, benefits would probably have to be cut--both for current recipients and for new beneficiaries--by about 25 percent to match the system's available revenue.

The key message from those numbers is that with benefits reduced annually to equal revenue, as they will be under current law when the trust funds run out, some form of the Social Security program can be sustained forever. Of course, many people would not consider a sudden 25 percent cut in benefits to be desirable policy. In addition, the budgetary demands of bridging the gap between spending and revenue in the years before that cut could prove onerous. But Social Security is sustainable from a narrow programmatic perspective. What is not sustainable is continuing to provide the present level of scheduled benefits given the present financing.

The Budgetary and Economic Perspective

CBO's projections offer some guidance about the potential impact of those developments on the budget. Under the trustees' long-term assumptions, the Social Security surplus (excluding interest on bonds in the trust funds) will reach about $100 billion in 2007. By 2025, however, the surplus will have turned into a deficit of roughly $100 billion (in 2005 dollars). That $200 billion swing will represent a significant challenge for the budget as a whole, especially in light of the current budget deficit.

The demand on the budget from Social Security will take place at the same time as--but be eclipsed by--the demand from Medicare and Medicaid. Currently, outlays for Social Security benefits are slightly more than 4 percent of GDP, as is federal spending on Medicare and Medicaid combined. But whereas Social Security outlays are projected to grow to 6.5 percent of GDP by 2050, spending on the two health programs could reach a total of 20 percent of GDP if current trends in health care costs continue.

Without changes in policy, therefore, federal spending is likely to increase sharply in coming decades, widening the gap between outlays and revenues and expanding the amount of federal borrowing. The resulting rise in government debt could seriously harm the economy. It could crowd out private capital formation, and although its impact on capital accumulation could be muted by borrowing from abroad, foreign borrowing is no panacea. The debt owed to foreigners would still have to be serviced. In the end, federal debt would reduce the disposable income of U.S. residents and erode future living standards.
 

The Structure of the Current Social Security System

Social Security benefits are based on earnings during a person's working years. Workers with higher lifetime earnings receive higher benefits, as do their dependents and survivors. However, the benefit formula is structured to redistribute income: benefits replace a smaller portion of earnings for higher earners and a larger portion for lower earners.

The Benefit Formula

Benefits for retired or disabled workers are based on the average level of workers' taxable earnings over their working lifetime (their average indexed monthly earnings, or AIME). For retired workers, the AIME is based on the highest 35 years of earnings on which they paid Social Security taxes (up to the taxable maximum, $90,000 in 2005), with some adjustments. Earnings before age 60 are indexed to compensate both for past inflation and for real (after-inflation) growth in wages. For disabled workers and the survivors of deceased workers, the AIME can be based on a shorter period.

A progressive formula is applied to a worker's average indexed monthly earnings to calculate his or her primary insurance amount (PIA). The PIA is the monthly amount payable either to a worker who begins receiving Social Security retirement benefits at the age at which he or she is eligible for full benefits or to a disabled worker. The formula is designed to ensure that initial Social Security benefits replace a larger proportion of preretirement earnings for people with low average earnings than for those with higher earnings. For workers who turn 65 this year, the formula is:

The dollar thresholds at which changes occur in the percentage of the AIME replaced by the PIA are known as "bend points" (see Figure 2). The percentages themselves are known as "replacement factors."
 

Figure 2.

Primary Insurance Amount Under Current Law (For workers who turn 65 in 2005)

(Primary insurance amount in dollars)

Source: Congressional Budget Office based on data from the Social Security Administration.

Note: The bend points shown here are those in 2002, the first year in which workers turning 65 in 2005 were eligible to collect retirement benefits.

Each year, the bend points are increased to match growth in average annual earnings for the labor force as a whole. If earnings growth is roughly constant, benefits for new recipients rise at approximately the same rate as average earnings. So long as the system pays scheduled benefits, Social Security benefits will replace the same portion of earnings for future generations (at the normal retirement age) as they do for today's beneficiaries. But because average earnings typically grow faster than prices do, the purchasing power of those benefits will be higher than that of benefits paid today, allowing beneficiaries to share in future increases in workers' living standards. Once the trust funds are exhausted, however, those replacement rates will fall, under current law (see the lower lines of Figure 3).
 

Figure 3.

Projected Replacement Rate for Retired Workers at Age 65

(Benefits as a percentage of average indexed monthly earnings)

Source: Congressional Budget Office.

Notes: This figure is based on a simulation from CBO's long-term model using the Social Security trustees' 2005 intermediate demographic assumptions and long-term economic assumptions and CBO's January 2005 10-year economic assumptions.

The replacement rate is the ratio of the benefits that a worker would receive upon claiming them at age 65 to the worker's average indexed monthly earnings (AIME), both computed using earnings through age 61. Under current law, scheduled benefits cannot be paid starting in 2045.

Another perspective on trends in replacement rates comes from considering how benefits change over time for workers with the same level of real earnings. To illustrate that perspective, consider someone making $2,500 a month. That level of income is currently around the middle of the earnings distribution. But in 2050, someone earning $2,500 a month (adjusted for inflation) will earn less than two-thirds of workers, even though he or she will have the same purchasing power as a median worker today. Because the Social Security benefit formula is progressive and indexed to wages--through both the indexation of earnings before age 60 in calculating the AIME and the indexation of the bend points in the PIA formula--benefits will replace a larger portion of earnings for future workers at that earnings level (see the top line in Figure 3). Again, exhaustion of the trust funds would lead to lower replacement rates.

Retirement Age

Under current law, the age at which a worker becomes eligible for full Social Security retirement benefits--the normal retirement age (NRA)--depends on the worker's year of birth. For people born before 1938, the NRA is 65. For slightly younger workers, it increases by two months per birth year, reaching 66 for people born in 1943. The NRA remains at 66 for workers born between 1944 and 1954 and then increases in two-month increments again, reaching 67 for people born in 1960 or later.

Workers can begin receiving retirement benefits before their NRA--as early as age 62--but their monthly benefits will be permanently lower than if they had waited until the NRA to claim benefits. Likewise, if workers delay receipt until they are older than the NRA, their monthly benefits will be higher. Those adjustments are intended to be "actuarially fair," so that the total value of benefits received over a lifetime will be approximately equal regardless of when a worker first claims benefits.

Cost-of-Living Adjustment

At the end of each year, the Social Security Administration adjusts existing benefits by the amount of any increase in the consumer price index. For example, the cost-of-living adjustment of 2.7 percent that took effect in December 2004 reflected the increase in the consumer price index for urban wage earners and clerical workers that occurred between the third quarter of 2003 and the third quarter of 2004.
 

Policy Options for Slowing the Growth of Outlays

As discussed earlier, in the absence of policy changes, CBO expects the Social Security trust funds to be depleted in 2044, under the trustees' long-term assumptions. After that, the program would no longer have the legal authority to pay full benefits. Spending would have to be reduced to match available revenue, which could require across-the-board cuts of 25 percent in benefits. Those reductions would affect not only newly eligible beneficiaries but also existing Social Security recipients of all ages.

Providing the Authority for Full Scheduled Benefits

Those benefit cuts could be avoided by giving the Social Security program the legal authority to borrow money in the event of trust-fund exhaustion. That option, however, would not address the broader budgetary and economic issues stemming from the fiscal imbalances in the Social Security system. Borrowing money to pay benefits would not be a sustainable option in the long run. By contributing to the growth of federal debt, it could have a corrosive effect on economic growth and could eventually lead to a sustained economic contraction. Repaying that debt would ultimately require cuts in spending or higher taxes somewhere in the budget.

Cuts in benefits could also be avoided by increasing taxes or reducing other federal spending and directing the savings to Social Security. Although such approaches would address Social Security's fiscal imbalances, some types of tax increases could risk slowing economic growth by discouraging work and saving, and reducing other spending could be difficult in light of the projected rise in federal outlays for health care.

Improving Social Security's Financial Balance

A variety of proposals have been advanced for restoring balance to the Social Security system. As noted above, CBO has prepared a menu of illustrative options for altering scheduled benefits or revenue. That menu--which is attached to this statement--includes the effects of the options on Social Security's finances as well as on the taxes paid and benefits received by people in different income groups and birth cohorts. The menu is intended to be representative of the types of changes that could be made to Social Security, but it is far from exhaustive. For example, it does not include options to introduce individual accounts, because the effects of such options are too complex to be shown clearly in the limited space available in the menu. (CBO has analyzed proposals for individual accounts in other publications.)⁽¹⁾ Moreover, it must be emphasized that various changes would be likely to interact with each other, so the net effect of multiple changes would be different from the sum of the individual effects.

This testimony examines the budgetary and distributional implications of three options to slow the growth of benefits: the first is taken directly from the attached menu, the second is a variation on a menu option, and the third is a combination of two menu options. All three would reduce scheduled benefits for people who first become eligible for benefits in 2012, including retired and disabled workers and their dependents and survivors. All of the options would keep the Social Security system solvent for at least the next 100 years.

Descriptions of the Options. The first approach considered here is the provision for price indexing of initial benefit awards advanced by the President's Commission to Strengthen Social Security (option 1.1 in the attached menu). Under that option, the three replacement factors in the current PIA formula would be lowered each year to offset the effects of real wage growth (see Figure 4, which shows the effects of the options in 2035). The AIME and the bend points would continue to be indexed to wages. As a result, benefits would generally grow with inflation, so future beneficiaries would have the same purchasing power as today's beneficiaries, on average. Relative to scheduled benefits, payments to new beneficiaries would decline by one-quarter over 26 years and by one-half over 63 years, assuming that real wages grew by 1.1 percent a year, on average. Initially, Social Security outlays would increase relative to GDP, but in later years, they would decline as a share of GDP and fall substantially below the program's dedicated revenue (see Figure 5).
 

Figure 4.

Primary Insurance Amount Under Various Options (For workers who turn 65 in 2035)

(Primary insurance amount in 2005 dollars)

Source: Congressional Budget Office.

Notes: This figure is based on a simulation from CBO's long-term model using the Social Security trustees' 2005 intermediate demographic assumptions and long-term economic assumptions and CBO's January 2005 10-year economic assumptions.

The bend points shown here are those in 2032, the first year in which workers turning 65 in 2035 will be eligible to collect retirement benefits under current law.

Figure 5.

Social Security Revenue and Outlays as a Share of GDP Under Various Options

(Percentage of GDP)

Source: Congressional Budget Office.

Notes: This figure is based on a simulation from CBO's long-term model using the Social Security trustees' 2005 intermediate demographic assumptions and long-term economic assumptions and CBO's January 2005 10-year economic assumptions.

Revenue includes payroll taxes and income taxes on benefits but not interest credited to the Social Security trust funds; outlays include Social Security benefits and administrative costs. Under current law, outlays will begin to exceed revenue in 2019; starting in 2045, the program will no longer be able to pay the full amount of scheduled benefits. Under the alternative options, outlays will start to exceed revenue in 2020. All three of the options begin in 2012, and under each, scheduled benefits are always payable.

A variant of that type of price indexing is known as progressive price indexing. In the version of progressive price indexing that CBO analyzed (a variation of menu option 1.2), the replacement factors for workers with the highest earnings--those who earned the taxable maximum or more for at least 35 years--would be reduced to the same extent as under the previous option. For most workers below that earnings level, however, the reductions in replacement factors would be smaller, with the extent of the reduction correlated with earnings, so that workers with higher earnings would have their replacement factors reduced the most. Beneficiaries in the lowest 25 percent of the earnings distribution would not be directly affected by this policy change (see Figure 4). After 95 years, new beneficiaries with AIMEs above $3,150 (in 2005 dollars) would all receive the same benefit. Because fewer beneficiaries would be affected under this option and because their benefit reductions would be smaller, total outlays would be higher than under the previous price-indexing option, but they would fall below revenue around 2090 (see Figure 5).

The third option that CBO examined (a combination of menu options 1.3 and 1.6) would change the indexing of bend points and of the AIME and would adjust benefits for increases in longevity. Under this approach, bend points would grow with prices instead of with average wages, as they do under current law. Over time, the bend points would shift to lower levels of earnings, and average replacement rates would decline relative to those specified by current law (see Figure 4). In addition, in the calculation of the AIME, earnings would be indexed to prices instead of to wages. Finally, this option would adjust the benefit formula to offset increases in life expectancy in order to ensure that total lifetime benefits did not grow as life spans increased. (The longevity adjustments would apply only to retirement benefits.) All three of those changes would reduce scheduled benefits. Outlays would be higher than under price indexing of initial benefits but would fall below dedicated revenue after 2075 (see Figure 5).

Under all three options, the PIA formula would change annually. Before 2035, the proposed formulas would be closer to current law than shown in Figure 4, whereas in later years they would be lower than under price indexing of initial benefits.

Distributional Effects. The effects of those options on different groups of workers--younger and older, lower-earning and higher-earning--can be examined by estimating how much of a group's earnings the proposed benefits would replace. Under all of the options, as under current law, higher earnings would result in higher benefits in dollar terms, but the percentage of earnings replaced would be greater for lower earners. The three options differ in the degree to which they would affect replacement rates.

As discussed earlier, workers can be classified by earnings levels in various ways. One way is to group people with a specific real earnings level, such as $1,500 a month. Someone at that earnings level always has the same purchasing power but will fall lower in the earnings distribution over time. Alternatively, workers can be grouped by relative earnings--for example, the top 20 percent or bottom 20 percent of earners in each cohort. (For projections of replacement rates by birth cohort using those two classifications, see Figures 6 and 7.)
 

Figure 6.

Projected Replacement Rate for Retired Low and High Earners at Age 65 Under Various Options

(Benefits as a percentage of average indexed monthly earnings)

Source: Congressional Budget Office.

Notes: This figure is based on a simulation from CBO's long-term model using the Social Security trustees' 2005 intermediate demographic assumptions and long-term economic assumptions and CBO's January 2005 10-year economic assumptions.

The replacement rate is the ratio of the benefits that a worker would receive upon claiming them at age 65 to the worker's average indexed monthly earnings (AIME), both computed using earnings through age 61. Under current law, scheduled benefits cannot be paid starting in 2045. Under the alternative options, scheduled benefits are always payable.

Figure 7.

Projected Replacement Rate for Retired Workers at Age 65, by Earnings Quintile, Under Various Options

(Benefits as a percentage of average indexed monthly earnings)

Source: Congressional Budget Office.

Notes: This figure is based on a simulation from CBO's long-term model using the Social Security trustees' 2005 intermediate demographic assumptions and long-term economic assumptions and CBO's January 2005 10-year economic assumptions.

The replacement rate is the ratio of the benefits that a worker would receive upon claiming them at age 65 to the worker's average indexed monthly earnings (AIME), both computed using earnings through age 61. Under current law, scheduled benefits cannot be paid starting in 2045. Under the alternative options, scheduled benefits are always payable.

Under current law, people who died before 2044 would not be affected by the automatic benefit reductions that would occur upon trust-fund exhaustion. For the most part, their benefits would be lower under all three options, although benefits would be unchanged for lower earners in those cohorts under progressive price indexing.

Of the three options, price indexing of initial benefits would produce the largest change for future beneficiaries, especially later cohorts. Moreover, because that policy would involve an across-the-board cut in initial benefits, it would affect the benefits of all earnings groups by the same percentage.

Under progressive price indexing, benefits for high earners would be lower than under current law. But unlike under current law, those benefit reductions would allow the trust funds to remain solvent. As a result, workers in later cohorts would be spared the across-the-board benefit cuts that would occur when the trust funds were exhausted. For lower earners in those cohorts, benefits would be higher than under current law.

Under the third option, price indexing of the AIME and bend points plus longevity adjustments, replacement rates would be lower than under current law for all income groups. However, those rates would be slightly higher than under price indexing of initial benefits.

The replacement rates presented here consider only retired-worker benefits. Moreover, they do not account for expected increases in longevity (see Figure 8), which will allow future cohorts to claim benefits for a longer period of time. To address those issues, CBO estimated how the policy options discussed here would affect the lifetime Social Security benefits of people in different earnings levels and birth cohorts (see Figure 9). On average, real scheduled lifetime benefits for the cohort born from 2000 to 2009 will be more than twice as high as those for the 1940s cohort, CBO projects. Although lifetime benefits and replacement rates are different measures, both convey the same basic message about how these policy changes would affect various beneficiaries.
 

Figure 8.

Life Expectancy at Age 65

(Remaining years of life)

Source: Congressional Budget Office.

Notes: This figure is based on a simulation from CBO's long-term model using the Social Security trustees' 2005 intermediate demographic assumptions and long-term economic assumptions and CBO's January 2005 10-year economic assumptions.

Cohort life expectancies are calculated using death rates from the series of years in which a person will reach each succeeding age if he or she survives.

Figure 9.

Percentage Change in Lifetime Benefits Relative to Scheduled Benefits, by Earnings Quintile, Under Various Options

(Percent)

Source: Congressional Budget Office.

Notes: This figure is based on a simulation from CBO's long-term model using the Social Security trustees' 2005 intermediate demographic assumption sand long-term economic assumptions and CBO's January 2005 10-year economic assumptions, including only people who live to at least age 45.

Lifetime benefits are the present value of benefits received by an individual over his or her lifetime, including Old-Age and Disability worker benefits and Old-Age Spouse and Survivor benefits financed by dedicated payroll taxes, net of income taxes on benefits credited to the Social Security trust funds. Under current law, scheduled benefits cannot be paid starting in 2045;under the alternative options, scheduled benefits are always payable.